Enzymatic Hydrocyanation of a Sterically Hindered Aldehyde. Optimization of a Chemoenzymatic Procedure for (R)-2-Chloromandelic Acid

Article abstract of DOI:10.1021/op0340964

The asymmetric synthesis of (R)-o-chloromandelic acid, a key intermediate for the anti-thrombotic agent clopidogrel, via the almond meal catalyzed hydrocyanation of 2-chlorobenzaldehyde and subsequent acidic hydrolysis was developed into an industrially viable procedure. The use of a minimum amount of water consistent with enzyme activity and a slow feed of the reactants were the keys to obtaining (R)-2-chloromandelonitrile in a high (98%) yield and satisfactory (90%) enantiomeric excess. Acidic hydrolysis of the nitrile followed by crystallization from toluene afforded enantiopure (ee > 99%) (R)-2-chloromandelic acid.

Full text of DOI:10.1021/op0340964

Organic Process Research & Development 2003, 7, 828−831
Enzymatic Hydrocyanation of a Sterically Hindered Aldehyde. Optimization of a
Chemoenzymatic Procedure for (R)-2-Chloromandelic Acid
Luuk M. van Langen, Fred van Rantwijk, and Roger A. Sheldon*
Laboratory of Biocatalysis and Organic Chemistry, Delft UniVersity of Technology, Julianalaan 136,
2628 BL Delft, The Netherlands
Abstract:
considerable scope in the synthesis of enantiopure 2-hydroxy-
8
-11
The asymmetric synthesis of (R)-o-chloromandelic acid, a key
intermediate for the anti-thrombotic agent clopidogrel, via the
almond meal catalyzed hydrocyanation of 2-chlorobenzaldehyde
and subsequent acidic hydrolysis was developed into an
industrially viable procedure. The use of a minimum amount
of water consistent with enzyme activity and a slow feed of the
reactants were the keys to obtaining (R)-2-chloromandelonitrile
in a high (98%) yield and satisfactory (90%) enantiomeric
excess. Acidic hydrolysis of the nitrile followed by crystallization
from toluene afforded enantiopure (ee > 99%) (R)-2-chloro-
mandelic acid.
nitriles,
acted with only a modest enantioselectivity (ee
12
83%) in the hydrocyanation of 2-chlorobenzaldehyde.
This latter reaction was carried out in an aqueous-organic
two-phase solvent system, which is a proven concept for the
enzymatic synthesis of cyanohydrins. In such a reaction
system, the enzyme resides in the aqueous (working) phase
and the reactants and products reside in the organic (extrac-
tive) phase. The resulting low reactant concentration in the
aqueous phase suppresses the competing uncatalyzed back-
1
3
ground reaction, which otherwise would erode the enan-
tioselectivity. Additionally, the pH is maintained in the acidic
14
range to reduce the rate of the nonenzymatic reaction. Such
systems perform quite satisfactorily with a wide range of
aldehydes, and a computer model that predicts the final
conversion and enantiomeric excess has been constructed.1
Alternatively, the reaction can be conducted in a mi-
croaqueous organic phase, in combination with almond
Introduction
Enantiopure 2-hydroxycarboxyxlic acids are useful chiral
intermediates and building blocks. Thus, (R)-2-(2-chlorophen-
yl)-hydroxyacetic acid ((R)-o-chloromandelic acid) is emerg-
ing as the preferred chiral intermediate for the industrial
synthesis of the anti-thrombotic agent clopidogrel. Resolu-
tion procedures for (R)-o-chloromandelic acid include dia-
stereomeric crystallization2 and enantioselective enzymatic
5,16
1
7-19
meal,
which is the cheapest possible immobilisate of
1
almond oxynitrilase. We decided to investigate the scope of
such systems in the enzymatic hydrocyanation of 2-chlo-
robenzaldehyde. We now report the development of an
efficient chemoenzymatic route to (R)-o-chloromandelic acid.
-4
5
hydrolyis of the corresponding nitrile. Such resolution
procedures generally require the racemization of the un-
desired enantiomer. In the case of o-chloromandelic acid,
this can be accomplished in aqueous alkaline DMSO at 100
Results and Discussion
A baseline case was established by comparing the
enzymatic hydrocyanation of benzaldehyde (1) and 2-chlo-
robenzaldehyde (2) in a diisopropyl ether-water (50:50, v/v)
biphasic medium. Whereas 1 was rapidly converted into (R)-
mandelonitrile (3, ee > 99%), 2 reacted quite sluggishly at
°C. (R)-o-Chloromandelic acid has also been synthesized via
asymmetric transformations, which do not suffer from this
latter drawback. These include a microbial reduction of 2-(2-
chlorophenyl)-glyoxylic acid and the asymmetric addition
of trimethylsilyl cyanide to 2-chlorobenzaldehyde in the
presence of a chiral transition metal complex.7
6
(
8) Brussee, J.; Van der Gen, A. Biocatalysis in the Enantioselective Formation
of Chiral Cyanohydrins, Valuble Building Blocks in Organic Synthesis. In
StereoselectiVe Biocatalysis; Patel, R. N., Ed.; Dekker: New York, 2000;
pp 289-320.
The (R)-specific oxynitrilase from almonds (hydroxynitrile
lyase from Prunus amygdalus, E.C. 4.1.2.10), which has
*
Corresponding author. Roger A. Sheldon, Laboratory of Organic Chemistry
(9) Effenberger, F. Hydroxynitrile Lyases in Stereoselective Synthesis. In
StereoselectiVe Biocatalysis; Patel, R. N., Ed.; Dekker: New York, 2000;
pp 321-342.
(10) Schmidt, M.; Griengl, H. Top. Curr. Chem. 1999, 200, 193-226.
(11) North, M. Tetrahedron: Asymmetry 2003, 14, 147-176.
(12) Kirschbaum, B.; Wilbert, G.; Effenberger, F. U.S. Patent 2,002,052,523,
2002; Chem. Abstr. 2002, 136, 355075.
(13) Loos, W. T.; Geluk, H. W.; Ruijken, M. M. A.; Kruse, G. C.; Brussee, J.;
Van der Gen, A. Biocatal. Biotransform. 1995, 12, 255-266.
(14) Niedermeyer, U.; Kula, M. R. Angew. Chem., Int. Ed. Engl. 1990, 29, 386-
387.
(15) Willeman, W. K.; Straathof, A. J. J.; Heijnen, J. J. Enzyme Microb. Technol.
2002, 30, 200-208.
(16) Willeman, W. F.; Gerrits. P. J.; Hanefeld, U.; Brussee, J.; Straathof, A. J.
J.; Van der Gen, A.; Heijnen, J. J. Biotechnol. Bioeng. 2002, 77, 239-247.
(17) Zandbergen, P.; Van der Linden, J.; Brussee, J.; Van der Gen, A. Synth.
Commun. 1991, 21, 1387-1391.
(18) Kiljunen, E.; Kanerva, L. T. Tetrahedron: Asymmetry 1996, 7, 1105-1116.
(19) Lin, G.; Han, S.; Li, Z. Tetrahedron: Asymmetry 1998, 9, 1835-1838.
and Catalysis, Delft University of Technology, Julianalaan 136, 2628 BL Delft,
The Netherlands. Telephone: + 31 15 278 2683. Fax: + 31 15 278 1415.
E-mail: R. A.Sheldon@tnw.tudelft.nl.
(1) Bousquet, A.; Musolino, A. PCT Int. Appl. WO 9918110, 1999; Chem.
Abstr. 1999, 130, 296510e.
(
(
(
(
2) Noda, H.; Sakai, K.; Murakami, N. Jpn. Kokai Tokkyo Koho JP 2001072664,
2
001; Chem. Abstr. 2001, 134, 237304a.
3) Hyoda, T.; Nawata, H. Jpn. Kokai Tokkyo Koho JP 2002114737 2002;
Chem. Abstr. 2002, 136, 309759.
4) Balint, J.; Csatarine Nagy, M.; Dombrady, Z.; Fogassy, E.; Gajari, A.; Suba,
C. PCT Int. Appl. WO 0300636, 2003; Chem. Abstr. 2003, 136, 309759.
5) De Santis, G.; Zhu, Z.; Greenberg, W. A.; Wong, K.; Chaplin, J.; Hanson,
S. R.; Fanwell, B.; Nicholson, L. W.; Rand, C. L.; Weiner, D. P.; Robertson
D. E.; Burk, M. J. J. Am. Chem. Soc. 2002, 124, 9024-9025.
6) Mitsuhashi, K.; Yamamoto, H. Eur. Pat. Appl. EP 1316613, 2003; Chem.
Abstr. 2003, 139, 5772.
(
(
7) Blacker, A. J.; Houson, I. N. PCT Int. Appl. WO 0266410, 2002; Chem.
Abstr. 2002, 137, 201079.
8
28
•
Vol. 7, No. 6, 2003 / Organic Process Research & Development
10.1021/op0340964 CCC: $25.00 © 2003 American Chemical Society
Published on Web 09/12/2003

express the enzyme at the same high level, although it has
no biological function. Meals prepared from blanched
almonds (i.e., almonds from which the skin had been
removed) generally performed somewhat poorer. The blanch-
ing process involves a brief submersion in boiling water,
which presumably causes partial inactivation of the R-
oxynitrilase, resulting, in turn, in a reduced ee.
We surmised that the diverging activities and enantio-
selectivities observed with the various almond meals could
be caused by uncontrolled heating during the grinding of
the almonds. Hence, we decided to prepare almond meal
under carefully controlled conditions. A preparation with a
very high activity was obtained, but the ee of the product
did not increase beyond 91%. We note that 2-fluorobenz-
aldehyde also reacted with reduced enantioselectivity (84%
Figure 1. Enzymatic hydrocyanation of benzaldehyde and
2-chlorobenzaldehyde. (R)-Mandelonitrile (ee > 99%): [. (R)-
2-Chloromandelonitrile (ee 75%): 2.
2
0
ee); possibly the effect is inherent in a 2-halo substituent.
The described concept was subsequently expanded to a
0.1-1 L scale (see Table 2). A good result as regards reaction
time and ee was obtained at a relatively low (0.2 M) aldehyde
concentration, using only 10-20% excess HCN. When the
concentration of 2 was increased to 0.4 M, however, a
product with only 82% ee was obtained, even though the
reactant/catalyst ratio was kept at the same value. An
experiment at 5 °C gave a very similar result. We ascribe
this result to increased competition by the second-order-
uncatalysed background reaction. A 4-fold rate increase of
the latter is to be expected upon a doubling of the reactant
concentration, whereas the rate of the enzymatic reaction
increased only 1.75 times.
Figure 2. Enzymatic and uncatalyzed hydrocyanation of
benzaldehyde and 2-chlorobenzaldehyde.
a rate that was approximately 4 times slower than that of 1
(
Figure 1). The ee of the 2-(2-chlorophenyl)-2-hydroxy-
acetonitrile (4) formed also was disappointingly low (ee 52%
R)) and only modestly improved to 77% upon a reduction
(
Competition by the nonenzymatic background reaction
has in the past been addressed through the optimization of
of the temperature to 0 °C, which is close to the reported
value of 83%. These results explain why the enzymatic
the aqueous-organic phase ratio and the catalyst loading
hydrocyanation of 2-substitued benzaldehydes is conspicuous
15,16,21
and by introducing reactant transport limitations.
We
by its absence in the literature.8
-10
We reasoned that the
surmised that a slow feed of the reactants, to keep their
concentrations low, could adequately suppress the back-
ground reaction. This indeed proved to be the case; we even
could increase the reaction temperature to 5-15 °C with a
considerable reduction in the reaction time while maintaining
the ee at approximately 90%. Only at a 20 °C reaction
temperature the ee dropped to 86%.
enzymatic reaction of the severely hindered 2 is too slow to
compete adequately with the uncatalyzed background reac-
tion (see Figure 2). The obvious remedy was a drastic
reduction of the aqueous phase.
Exploratory experiments with defatted almond meal from
Sigma (see Table 1) revealed that an aqueous phase of ∼0.4
mL/g of biocatalyst, which corresponds with 4% (v/v) of
the reaction volume, was required for optimum activity. This
amount of buffer was easily absorbed by the biocatalyst,
which remained freely dispersed in the organic solvent.
Larger amounts of buffer increased the tendency of the
biocatalyst to coagulate, at a considerable penalty in rate and
ee. In contrast with the enzymatic hydrocyanation of 1, which
results in enantiopure (R)-3, 4 was formed with a much more
modest ee. Lowering the reaction temperature to 0 °C
improved the ee to 91%, at the expense of a 5-fold reduction
in reaction rate.
Similar experiments were carried out with almond meals
from various commerical sources. The one from California
Flavors also required 0.4 mL of buffer per g of biocatalyst;
hence, this amount was adopted as a standard. Other almond
meal preparations performed similarly to the Sigma prepara-
tion. We noted with some surprise that there was no
systematic difference between biocatalysts prepared from
bitter and sweet almonds. Hence, it would seem that the latter
A further improvement could be achieved by performing
the reaction at 10 °C while feeding either the aldehyde or
HCN. The latter procedure, which was much faster, resulted
in a 96% yield and 91% ee of 4 at a space-time yield of 14
-1 -1
-1 -1
g L
h
(336 g L d ). The productivity of the biocatalyst
-1
was 1.3 g g .
Further experiments revealed that a major fraction of the
biocatalyst rapidly settled when the stirrer was stopped at
the end of the reaction. A fraction of fine particles remained
suspended in the reaction medium, however, and had to be
separated by centrifugation. Alternatively, these could be
removed from the supernatant by stirring the latter with a
small amount (5%, V/V) of acidified water containing sodium
dodecyl sulfate. All fines migrated to the aqueous phase and
could easily be discarded.
(20) Han, S.; Chen, P.; Lin, G.; Huang, H.; I, Z. Tetrahedron: Asymmetry 2001,
12, 843-846.
(21) Gerrits, P. J.; Willeman, W. F.; Straathof, A. J. J.; Heijnen, J. J.; Brussee,
J.; Van der Gen, A. J. Mol. Catal. B: Enzymol. 2002, 15, 111-121.
Vol. 7, No. 6, 2003 / Organic Process Research & Development
•
829

Table 1. Performance of Various Almond Meal Preparations in the Hydrocyanation of 2-Chlorobenzaldehyde
catalyst
buffer
mL (g cat)
temp
(°C)
initial rate
time
(h)
conversion
(%)
ee
(% R)
-
1
-1
(mmol min (gcat)-1
-1
source
pretreatment
(g L
)
)
Sigma
Sigma
Sigma
Sigma
CF
none
100
100
100
100
200
200
200
200
200
200
200
200
200
200
200
0.1
0.2
0.4
0.4
0.2
0.2
0.4
0.4
0.4
0.4
0.4
0.4
0.4
0.4
0.4
20
20
20
0
0.014
0.014
0.08
1.5
3
1.5
5
3
50
97
90
94
98
56
98
88
89
93
76
98
50
96
58
53
78
82
91
86
80
90
90
90
91
86
77
73
91
88
none
none
none
0.07
none
0
0.008
0.008
0.020
0.012
0.012
0.010
0.014
0.010
0.008
0.036
0.009
CF
defatted
defatted
none
0
1.5
1.8
4
CF
0
Alldrin
Alldrin
Nuts4U
Nuts4U
Sictia
Sictia
Local
Local
0
blanched
none
blanched
none
defatted
defatted
blanched, defatted
0
4
0
2
0
3
0
4
0
1.5
1
0
0
2
Table 2. Preparative-Scale Hydrocyanation of
-Chlorobenzaldehyde (2)a
Conclusion
An industrially viable synthesis of (R)-2-chloromandelic
2
2
catalyst temp
time conversion ee
STY
acid of high enantiomeric purity was developed, based on
the almond meal catalyzed asymmetric hydrocyanation of
-
1
(%) g L h-1
-1
(
M) (g L
)
(°C) procedure (h)
(%)
2
-chlorobenzaldehyde and subsequent chemical hydrolysis
0
.2
.4
.4
.4
.4
.4
.6
.4
.4
.4
50
100
50
0
0
5
0
5
batch
batch
batch
fed batch 12
fed batch 10
14
8
97
98
97
94
98
97
98
98
96
95
90
82
84
88
90
91
89
86
91
90
2.4
8.2
of the obtained cyanohydrin. The major improvement
compared with existing procedures using almond meal was
the gradual addition of the substrates to the reaction mixture.
In the case of the hydrocyanation of 2-chlorobenzaldehyde,
this improved the space-time yield of the process while the
enantiopurity of the product did not suffer.
0
0
0
0
0
0
0
0
0
8
8.1
100
50
5.2
6.5
50
10 fed batch
15 fed batch
20 fed batch
10 HCN fed 4.5
10 2 fed
9
8
5
7.2
50
12.2
13.1
14.2
10.5
50
50
50
6
Experimental Section
a
Reaction conditions: 1.1-1.2-fold excess of HCN (dissolved in diisopropyl
Chemicals. 2-Chlorobenzaldehyde was obtained from
Tessenderlo Chemie, Belgium as a gift. Diisopropyl ether
and sodium cyanide were from Acros; citric acid mono-
hydrate and hydrochloric acid were procured from Baker.
Whole sweet almonds were purchased at a local grocery
shop; food grade almond meals were obtained from Cali-
fornia Flavors (Yuba City, California, USA), Alldrin Brothers
Almond Company (Turlock, California, USA), Nuts4U
ether), biocatalyst pretreated with 0.5 mL g-1 of 20 mM citrate buffer pH 5.5;
reactant feeding time (where appropriate) 5-6 h.
(Sugar Land, Texas, USA), and Sictia (France); and defatted
almond meal was obtained from Sigma.
Preparation of Almond Meal. Almonds (230 g) were
cooled to 4 °C, milled using a kitchen device, and extracted
6 times with 160 mL of diisopropyl ether. The solids were
separated by centrifugation, resulting in 300 g of “wet” meal
Figure 3. Acid-catalyzed stereoretentive hydrolysis of (R)-2-
chloromandelonitrile into (R)-2-chloromandelic acid.
(
as a slurry of meal in diisopropyl ether), which was stored
After the separation of the biocatalyst, the diisopropyl
ether was removed under reduced pressure, yielding (R)-4
as a yellow oil that rapidly crystallized. Treatment with
concentrated hydrochloric acid afforded (R)-2-chloroman-
delic acid without any racemization (Figure 3). Under the
applied conditions, a red color developed which was not
observed when hydrolyzing the chemically prepared cyano-
hydrin. Most likely this contaminant originates from the
almond meal. Extracting the acid after the hydrolysis
procedure with an organic solvent, for example, diisopropyl
ether, effectively separated the contaminant from the acid
product. The resulting crude acid obtained was enantiomeri-
cally enriched to 99.9% ee by recrystallization from toluene,
as already has been described in the literature.22
at 4 °C. This wet meal contained approximately 38% (w/v)
of dry almond meal. The commonly used drying step was
found to be superfluous and was, hence, omitted.
Preparation of the Reactant Solutions. WARNING.
Sodium cyanide and hydrocyanic acid are highly poisonous.
They should only be handled in a fume cupboard with a good
draught. It is strongly advised to keep an HCN alarm
switched on.
Sodium cyanide (20 g, 0.4 mol) and citric acid (0.2 g)
were dissolved in 200 mL of water. The solution was cooled
in an ice-water bath and stirred with 100 mL of diisopropyl
ether, while the pH was brought to 5.5 by the slow addition
(22) Jenkins, S. S. J. Am. Chem. Soc. 1931, 53, 2341-2343.
830
•
Vol. 7, No. 6, 2003 / Organic Process Research & Development

of 33% (w/v) hydrochloric acid. The aqueous layer with the
suspended sodium chloride was extracted 2 times with 50
mL of diisopropyl ether. The combined organic layers were
stored on a small volume of citrate buffer pH 5.5 in a dark
bottle.
The HCN concentration in the resulting solution was 2
M as established using the following procedure: exactly 5
mL of HCN solution was added to 25 mL of 2 M NaOH,
and the mixture was stirred for 2 min. A small amount of
potassium chromate was added as indicator, and the solution
was titrated with 0.1 M silver nitrate.
The resulting solid yellowish cyanohydrin was added to
50 mL of hydrochloric acid (33%, w/v), and the suspension
was stirred at room temperature until dissolution was
complete (approximately 1 h). The solution was refluxed for
3 h until all amide was converted into acid. The progress of
the reaction was monitored by HPLC. The ammonium
chloride precipitate was filtered off, and the solution was
extracted 6 times with 50 mL of diisopropyl ether. The
combined extracts were concentrated in vacuo; the resulting
crude product was recrystallized from 100 mL of toluene to
yield 25 g (134 mmol, 67%) of (R)-2-chloromandelic acid
A 2 M solution of 2-chlorobenzaldehyde in diisopropyl
ether was prepared and stored on a saturated sodium
bicarbonate solution. Diisopropyl ether used in the enzymatic
hydrocyanation reactions was presaturated with 20 mM
citrate buffer pH 5.5.
with ee 99.9%; mp 119 °C, R
D
-78.4° (c 0.5, ethanol)
-78.0° (c 0.5, ethanol)).
23
(
literature: mp 119-120 °C, R
D
1
H NMR (DMSO-d
6
) δ 5.34 (s, 1H), 7.32-7.44 (m, 4H).
HPLC Analysis. Samples were dissolved in hexane-
isopropyl alcohol (90:10, v/v), containing 1% trifluoroacetic
acid, and dried over sodium sulfate. The enantiomeric excess
of 2-chloromandelonitrile was measured after acetylation
with acetic anhydride. The hydrocyanation products were
analyzed by HPLC using a Chiralcel OB-H column, eluant
hexane-2-propanol (90:10, v/v). Retention times (in min):
Enzymatic Hydrocyanation. To a suspension of the
appropriate amount of almond meal in diisopropyl ether in
a polypropylene reaction vessel equipped with a magnetic
stirrer or turbine stirrer was added the appropriate volume
of 20 mM citrate buffer pH 5.5. The suspension was stirred
while cooling to the preferred temperature. Depending on
the procedure, the substrate solutions were added directly
to the reaction mixture or fed over a 5-6 h period.
Samples were taken from the reaction mixture and
centrifuged. The supernatant was diluted with hexane-
trifluoroacetic acid (95:5, v/v) and analyzed by HPLC.
Samples for enantiomeric analysis were, after evaporation
of the solvent, derivatized in dioxane-acetic anhydride-
pyridine (25:3:0.5, v/v).
2-chlorobenzaldehyde, 4; 2-chloromandelonitrile, 5; O-acetyl-
(S)-2-chloromandelonitrile, 7; O-acetyl-(R)-2-chloromande-
lonitrile, 10.
The hydrolysis products of 2-chloromandelonitrile were
analyzed using a Nucleosil C-18 column, eluant methanol-
water (40:60, v/v) acidified with phosphoric acid to pH 3.5.
The optical purity of 2-chloromandelic acid was measured
as follows: samples were esterified using boron trifluoride
in methanol and subjected to chiral HPLC using a Chiralcel
OD column, eluant hexane-2-propanol (80:20, v/v). Reten-
tion times (in min): (S)-2-chloromandelic acid methyl ester,
The enzymatic hydrocyanation rates of 1 and 2 were
compared as follows: a 0.5 g amount of almond meal
(Sigma) was dispersed in 5.5 mL of a 20 mM citrate buffer
9
.6; (R)-2-chloromandelic acid methyl ester, 10.8.
pH 5.5 and 3 mL of diisopropyl ether. A 0.5 mL aliquot of
a 2 M solution of aldehyde and 2 mL of a 2 M HCN solution
Acknowledgment
(both in diisopropyl ether) were added. Samples (10 µL) were
withdrawn at regular intervals and analyzed by HPLC.
Downstream Processing of the Hydrocyanation Prod-
uct. At the end point of a typical enzymatic hydrocyanation
reaction (1 L reaction volume, 200 mM aldehyde), the stirrer
was stopped and the almond meal was allowed to settle for
This work was financially supported by Calaire Chimie,
Tessenderlo Group (Calais, France). The authors wish to
thank Dr. J. F. Cavalier, Mr. A. Lavieville (Calaire Chimie),
Dr. M. Belmans, and Dr. L. Kemps (Tessenderlo Chemie)
for their valuable discussions and assistance.
5
min. The supernatant was discharged, and the remaining
almond meal was washed with two portions of 250 mL of
diisopropyl ether at 0 °C. The combined organic layers were
centrifuged for 10 min at 1000 rpm and 0 °C; the resulting
clear solution was concentrated using a rotary evaporator
under reduced pressure.
Received for review July 11, 2003.
OP0340964
(23) Acros catalogue.
Vol. 7, No. 6, 2003 / Organic Process Research & Development
•
831