Enantioselective formation of mandelonitrile acetate - Investigation of a dynamic kinetic resolution

Article abstract of DOI:10.1016/S0957-4166(02)00174-X

Investigations into the separate reactions of a dynamic kinetic resolution and the combined reactions revealed that the overall sequence is highly susceptible to the water content of the reaction mixture. While the racemization and formation of mandelonitrile as well as its kinetic resolution proceeded rapidly when performed independently, the dynamic kinetic resolution was severely hampered by the undesired formation of acetic acid during the reaction. The utilization of drying reagents and neutralizing agents in order to suppress the formation of acetic acid or its consequences were investigated.

Full text of DOI:10.1016/S0957-4166(02)00174-X

TETRAHEDRON:
ASYMMETRY
Pergamon
Tetrahedron: Asymmetry 13 (2002) 739–743
Enantioselective formation of mandelonitrile
acetate—investigation of a dynamic kinetic resolution
Yu-Xin Li,a Adrie J. J. Straathofb and Ulf Hanefelda,*
aApplied Organic Chemistry and Catalysis, Delft University of Technology, Julianalaan 136, 2628 BL Delft, The Netherlands
bKluyverlaboratory for Biotechnology, Delft University of Technology, Julianalaan 67, 2628 BC Delft, The Netherlands
Received 21 March 2002; accepted 9 April 2002
Abstract—Investigations into the separate reactions of a dynamic kinetic resolution and the combined reactions revealed that the
overall sequence is highly susceptible to the water content of the reaction mixture. While the racemization and formation of
mandelonitrile as well as its kinetic resolution proceeded rapidly when performed independently, the dynamic kinetic resolution
was severely hampered by the undesired formation of acetic acid during the reaction. The utilization of drying reagents and
neutralizing agents in order to suppress the formation of acetic acid or its consequences were investigated. © 2002 Published by
Elsevier Science Ltd.
1. Introduction
19916 and in more detail in 1992.7,8 These papers have
been quoted frequently; however, only one report has
since described the application of the methodology.9
Since all the chemicals and the enzyme utilized in this
synthesis of the cyanohydrin acetates 6 are readily
available, this is somewhat surprising. All the more so,
since the method has even been included in a book on
preparative biotransformations.10 The main drawback
of the published synthesis is the relatively long reaction
time of 6–8 days. In order to obtain a better insight into
the possible reasons for the long reaction time we
investigated this multi component reaction and its sepa-
rate steps. Few modifications were introduced. As an
enzyme the highly selective Candida antarctica lipase B
(CAL-B, chirazyme L-2, c.-f., C2, Lyo)11,12 rather than
Pseudomonas cepacia lipase was utilized.7,10 The study
was performed in dry toluene (instead of diisopropyl
ether), a solvent that proved to be particularly suitable
for the kinetic resolution of cyanohydrin acetates with
CAL-B. Furthermore no zeolite molecular sieve was
added for water adsorption,13 since this might also act
as a Lewis or Brønsted acid,14 thereby disturbing the
hydrocyanation which is catalyzed by alkaline Amber-
lite (−OH form).7,10 Benzaldehyde 1 was used as a
model substrate, since it is known to be a particularly
good substrate for this reaction sequence.
Enantiomerically pure cyanohydrins and their esters are
important building blocks in organic chemistry.1–3
Their synthesis has therefore attracted considerable
attention.4,5 One particularly elegant approach is the
dynamic kinetic resolution of cyanohydrins formed in
situ from an aldehyde 1 and acetone cyanohydrin 2, the
cyanide source. The utilization of an enantioselective
lipase for catalyzing the esterification leads to high yield
and enantiomeric purity of the formed cyanohydrin
ester 6 (Scheme 1). This method was first published in
Scheme 1.
Three separate reactions constitute the overall synthetic
sequence from benzaldehyde 1 to the final product
(S)-mandelonitrile acetate 6. The elimination of HCN
* Corresponding author. Tel.: +31 (0)15 278 9304; fax: +31 (0)15 278
4289; e-mail: u.hanefeld@tnw.tudelft.nl
0957-4166/02/$ - see front matter © 2002 Published by Elsevier Science Ltd.
PII: S0957-4166(02)00174-X

740
Y.-X. Li et al. / Tetrahedron: Asymmetry 13 (2002) 739–743
from acetone cyanohydrin 2 as well as the addition of
HCN to benzaldehyde 1 are reversible, base catalyzed,
reactions. Under acidic conditions 2, 4 and 5 are stable,
and the reactions are extremely slow in either direction.
The two reversible reactions are essential for the forma-
tion and racemization of mandelonitrile (4+5), the sub-
strate of the enzyme reaction. They can be driven to
completion by the third, irreversible reaction.
tion, the molar ratio of 1 to acetone cyanohydrin 2 to
iso-propenyl acetate 7 was fixed at 1:2:3.
2. Results and discussion
Utilizing 0.05 mol equiv. of alkaline Amberlite, the
equilibrium constants and the equilibration times of the
combined first two reactions were determined. The
concentrations of 1 and 2 were quantified simulta-
neously by IR (Fig. 1a). Even at room temperature the
equilibrium of the combined reactions was reached
within 5 h (Fig. 1b).
This third step is the CAL-B catalyzed acylation of 4
with iso-propenyl acetate 7 to form (S)-mandelonitrile
acetate 6 and acetone 3. The enantiopurity of the
product depends on the enantioselectivity E of the
enzyme for the specific substrate under the specific
conditions. In a kinetic resolution the ee of the product
decreases with the degree of conversion, since the
availability of the favored substrate enantiomer (here 4)
decreases much more rapidly than that of the less
favored substrate enantiomer (here 5). In order to
achieve the greatest possible enantiomeric purity in an
enzymatic dynamic kinetic resolution, it is therefore
important not only to utilize an enzyme with a high E
value, but also to ensure that the substrate remains
(almost) racemic throughout its conversion. Thus, it is
essential that the racemization is faster than the
acylation. Therefore we first studied the reactions indi-
vidually and then in combination. In the overall reac-
The racemization of 5 under reaction conditions (40 or
60°C) proceeded completely within an hour or less (Fig.
2), which is in accordance with data reported earlier.9
Moreover, the racemization reaction was not influenced
by the presence of CAL-B. It was therefore anticipated
that the elimination and addition reactions of HCN
would not form the bottleneck of the overall procedure.
In the CAL-B catalyzed acylation of the racemic mix-
ture of 4 and 5 with 7, only 4 is converted. Several
conditions for the esterification of the racemate were
investigated (Fig. 3). At 25, 40 and 60°C in toluene with
different enzyme concentrations CAL-B always showed
−
Figure 1. (a) IR spectra of the measurements of the reaction of 1 and 2 in the presence of Amberlite (0.0625 equiv. OH) to form
3, 4 and 5. The concentration of 1 was measured at 2733 and 2815 cm−1 and the concentration of 2 at 972 cm−1. The equilibrium
constant for the racemic reaction was determined as Keq=6.53. (b) Reaction of 1 and 2 in the presence of Amberlite (0.0625 equiv.
−OH) to form 3, 4 and 5. The concentration of HCN is not shown.

Y.-X. Li et al. / Tetrahedron: Asymmetry 13 (2002) 739–743
741
inhibition of the enzyme. In the course of the reaction,
however, a change in rates was observed. Instead of
rapid racemization and acylation of mandelonitrile, the
concentration of 1 seems to almost stabilize on a
plateau. Moreover, the ee of 5 increases, a clear indica-
tion of a kinetic resolution without racemization of the
initially racemic starting material. This strongly sug-
gests that the Amberlite is deactivated during the
course of the reaction. Similar results were obtained at
40°C. The same observations were also made when
performing the overall reaction of 1 (0.1 M) with 2 and
7 in the presence of CAL-B (37 U/mL) and Amberlite
(0.05 equiv. −OH). The reaction was extremely slow and
even after 47 h the yield of 6 was only 9.5%.
Figure 2. Racemization of 5 at 40 and 60°C in the presence of
Recently, it has been demonstrated that the presence of
even very small quantities of water can lead to signifi-
cant enzyme-catalyzed hydrolysis of the acylating
reagent.16 The toluene utilized in the reactions
described above contained 120 ppm of water. In order
to investigate the consequences of this amount of water,
the CAL-B catalyzed hydrolysis of ( )-mandelonitrile
acetate and 7 in dry toluene (obtained from Aldrich)
was examined separately. Both compounds were ini-
tially rapidly hydrolyzed. At 25°C and in the presence
of 120 ppm of water, 18% of ( )-mandelonitrile acetate
and 21% of 7 were hydrolyzed within 5 h. The hydroly-
sis proceeded even after an amount of water equal to
the original amount of water in the dry toluene had
been consumed. This indicates that water adsorbed on
the enzyme and its carrier must also be utilized in this
hydrolysis. During the dynamic kinetic resolution of
4+5 some enzymatic hydrolysis of 6 and 7 can also be
expected, although 6 and 7 were stable in the presence
of Amberlite alone. If this is indeed the case, the acetic
acid that is formed will quickly and completely neutral-
ize the alkaline ion exchanger, which is present in only
0.05 mol equiv. The reaction mixture turns from alka-
line to acidic. This stabilizes 2, 4 and 5. Such neutral-
ization regenerates water that had been consumed by
hydrolysis, so the water concentration will remain con-
stant, while the alkaline Amberlite will be depleted
(Scheme 2).
Amberlite (0.05 equiv. −OH).
Figure 3. Kinetic resolution of ( )-mandelonitrile (4+5) with
CAL-B and 7 at different temperatures and enzyme concen-
trations. " 25°C and 37 U/mL; ꢀ 40°C and 37 U/mL; ꢁ
60°C and 37 U/mL; × 25°C and 74 U/mL; * 40°C and 74
U/mL; ꢂ 60°C and 74 U/mL.
excellent selectivity (E15 always larger than 400). In
every case 6 was formed with an ee ]99%. The reac-
tion rate was proportional to the amount of enzyme.
These results clearly indicate that the separate reactions
are fast and proceed with high selectivity. As is required
for a successful dynamic kinetic resolution, racemiza-
tion is significantly faster than acylation if either 370 or
740 units of CAL-B per mmol of substrate are utilized.
In order to achieve overall reaction times that are
significantly shorter than those described in the litera-
ture, the reaction temperature should be maintained at
40°C or above. No side products were detected in any
case. Moreover, Amberlite did not catalyze the
acylation step, ruling out any interference of the chem-
ical catalyst with the enzyme catalyzed reaction.
These results indicate that the combination of the
racemization of mandelonitrile (4+5) and its kinetic
resolution should lead to the fast and highly selective
formation of (S)-mandelonitrile acetate 6. This is
indeed the case during the initial period of the reaction
(Fig. 4). Comparison of Figs. 3 and 4 show that the
observed slowing down of the formation of 6 can be
attributed to the formation of 1 rather than to any
Figure 4. Dynamic kinetic resolution of ( )-mandelonitrile
(4+5) in the presence of 7, CAL-B and Amberlite (0.05 equiv.
−OH) at 60°C.

742
Y.-X. Li et al. / Tetrahedron: Asymmetry 13 (2002) 739–743
CAL-B (Chirazyme L-2, cf., C2, Lyo) was a gift from
Roche Diagnostics (Penzberg, Germany). Its activity
was determined according to the procedure developed
by Roche Diagnostics.18 The enzyme was stored in a
dessicator with silica gel as a drying agent at room
temperature. The activity of the stored enzyme (3.7
U/mg) remained unchanged over a period of half a
year. Amberlite IRA904 was purchased from Acros and
conditioned according to the literature.7 HPLC analy-
ses were conducted with a Waters 510 pump, a Chiracel
OB-H column (0.46×25 cm), a Waters 486 UV detector
at 215 nm and a mixture of hexane and iso-propanol
(92:8) with 0.1% acetic acid as eluent. The flow rate was
1.0 mL/min. 1,3,5-Triisopropylbenzene was used as an
internal standard. Retention times: 1,3,5-triisopropyl-
benzene 3.7 min, 7 5.7 min, 1 7.4 min, 5 12.4 min; 4
13.2 min; 6 15.1 min and (R)-mandelonitrile acetate
20.3 min. The water concentration in toluene was mea-
sured by Karl–Fischer Titration with a Mettler DL35.
Hydranal®–Titrant 2 (from Riedel–deHaen) was used
as titrant. IR analysis was performed with a Perkin–
Elmer FT-IR Spectrometer Spectrum 1000 in a KBr
cell with a path length of 0.115 mm. Absorptions were
measured at 2733 and 2815 cm−1 1 and 972 cm−1 2.
Scheme 2.
Upon deactivation of the racemization catalyst, the
dynamic kinetic resolution is turned into a conventional
kinetic resolution. In addition, not all of 1 has at this
stage been converted into 4 and 5 nor has all of 2 been
converted into HCN and acetone 3. Therefore, the
concentration of 4+5 is relatively low. Further drying of
the solvent with molecular sieve prior to the incubation
of ( )-mandelonitrile acetate with CAL-B lowered the
water content to 60 ppm. This did reduce the rate of the
hydrolysis, but the problem remained. In order to
overcome this problem and to ensure that the reaction
medium remained alkaline, 0.25 mol equiv. of base was
added in the form of Amberlite. This did indeed lead to
an improvement in the overall reaction. The conversion
of 1 was significantly improved and the yield of 6 after
45 h was higher than before (15.9 instead of 9.5%).
This clearly demonstrates that the Amberlite and CAL-
B catalyzed synthesis of cyanohydrin acetate 6 is very
sensitive to hydrolysis and the acids released thereby. In
order to overcome this problem and to improve the
3.2. Kinetic resolution of ( )-mandelonitrile (4+5)
−
In an 8 mL vessel ( )-mandelonitrile (4+5) (66.5 mg, 0.5
mmol) and iso-propenyl acetate 7 (125 mg, 1.25 mmol)
were dissolved in 5 mL dry toluene. A baseline sample
was taken with a microsyringe. The solution was then
transferred to a nitrogen gas flushed vessel with CAL-B
(50 mg, 185 U or 100 mg, 370 U) in it. The reaction
system was closed with a rubber septum and stirred at
the desired temperature (25, 40 or 60°C). Samples (15
mL) were taken with a microsyringe at the time intervals
given in Fig. 3 and diluted with 1 mL solvent (same as
the mobile phase for HPLC) for chiral HPLC analysis.
The enantiomeric excess of 6 was always >99%.
yield of 6 a large amount of Amberlite (1.0 mol OH
equiv.) was added. However, instead of improving the
yield, only 2.1% of 6 was formed after 45 h. Moreover,
this reaction mixture turned brownish, something not
observed in the other reactions. This points to a base
induced polymerization of the HCN, which is well
known to yield brown and even black products.17
Adding more base, in order to neutralize the released
acid does not circumvent the problem of hydrolysis. It
seems to be necessary to suppress hydrolysis or its
consequences more efficiently.
In summary, low concentrations of alkaline Amberlite
efficiently catalyze the formation and racemization of
mandelonitrile (4+5) and do not inhibit the enzyme-cat-
alyzed acylation. Similarly, CAL-B is an excellent cata-
lyst for the enantioselective formation of cyanohydrin
acetates 6. However, the combination of these reactions
to a dynamic kinetic resolution is severely hampered by
the acid formed by the hydrolysis of 6 and 7. Utiliza-
tion of larger quantities of base does not solve this
problem, possibly due to the polymerization of HCN.
3.3. Dynamic kinetic resolution of ( )-mandelonitrile
(4+5)
In an 8 mL vessel ( )-mandelonitrile (4+5) (66.5 mg, 0.5
mmol) and iso-propenyl acetate 7 (125 mg, 1.25 mmol)
were dissolved in 5 mL of dry toluene. A baseline
sample was taken with a microsyringe. The solution
was then transferred to a nitrogen gas flushed vessel
with CAL-B (50 mg, 185 U) and Amberlite IRA904
(−OH form, 19.2 mg, 0.025 mmol −OH) in it. The
reaction system was closed with a rubber septum and
stirred at 60°C. Samples (15 mL) were taken with a
microsyringe at the time intervals given in Fig. 4 and
diluted with 1 mL of solvent (the same as the mobile
phase for HPLC) for chiral HPLC analysis. The enan-
tiomeric excess of 6 was always >99%.
3. Experimental
3.1. Materials and methods
Benzaldehyde, acetone, acetone cyanohydrin and iso-
propenyl acetate were distilled prior to use and were
stored under nitrogen. Technical grade ( )-mandeloni-
trile from Acros was purified by silica gel column
chromatography and stored under nitrogen. Dry tolu-
ene was obtained from Aldrich. Molecular sieves (4 A)
were obtained from Aldrich and used as supplied.
3.4. Racemization of (R)-mandelonitrile 5
Amberlite IRA904 (−OH form, 19.1 mg, 0.025 mmol
−OH) was added to a solution of 5 (66.5 mg, 0.5 mmol)
in dry toluene (5 mL). The reaction mixture was stirred
at the desired temperature (40 or 60°C). Samples (15
,

Y.-X. Li et al. / Tetrahedron: Asymmetry 13 (2002) 739–743
743
mL) were taken with a microsyringe at the time intervals
given in Fig. 2 and analyzed by chiral HPLC to mea-
sure the ee values.
Acknowledgements
U.H. thanks the Royal Netherlands Academy of Arts
and Sciences (KNAW) for a fellowship. The authors
gratefully acknowledge Roche Diagnostics Penzberg
(Dr. W. Tischer) for the generous gift of CAL-B.
3.5. Chemical transcyanation between benzaldehyde 1
and acetone cyanohydrin 2 and vice versa
Amberlite IRA 904 (−OH form, 38.5 mg, 0.05 mmol
−OH) was added to a solution of 1 (84.8 mg, 0.8 mmol)
and 2 (70.0 mg, 0.82 mmol) in dry toluene (8 mL). The
mixture was stirred in a closed system at room temper-
ature (25°C). Samples were taken (at the times given in
Fig. 1b) with a syringe for infrared spectroscopic analy-
sis (see Section 3.1). After 5 h the equilibrium situation
was reached with 0.35 mmol 1, 0.16 mmol 2, 0.66 mmol
3 and 0.45 mmol ( )-mandelonitrile (4+5).
References
1. Gregory, R. J. H. Chem. Rev. 1999, 99, 3649–3682.
2. Johnson, D. V.; Zabelinskaja-Mackova, A. A.; Griengl,
H. Curr. Opin. Chem. Biol. 2000, 4, 103–109.
3. Effenberger, F.; Fo¨rster, S.; Wajant, H. Curr. Opin. Bio-
technol. 2000, 11, 532–539.
4. Gro¨ger, H. Chem. Eur. J. 2001, 7, 5247–5251.
5. Brussee, J.; van der Gen, A. In Stereoselective Biocataly-
sis; Patel, R. N., Ed.; Dekker: New York, 2000; pp.
289–320.
6. Inagaki, M.; Hiratake, J.; Nishioka, T.; Oda, J. J. Am.
Chem. Soc. 1991, 113, 9360–9361.
The reverse reaction was carried out similarly using
racemic mandelonitrile (109.3 mg, 0.82 mmol) and 3
(51.0 mg, 0.88 mmol) in dry toluene (8 mL). After 5 h
the equilibrium situation was reached with 0.38 mmol
1, 0.14 mmol 2, 0.74 mmol 3 and 0.44 mmol ( )-mande-
lonitrile (4+5).
7. Inagaki, M.; Hiratake, J.; Nishioka, T.; Oda, J. J. Org.
Chem. 1992, 57, 5643–5649.
8. Inagaki, M.; Hatanaka, A.; Mimura, M.; Hiratake, J.;
Nishioka, T.; Oda, J. Bull. Chem. Soc. Jpn. 1992, 65,
111–120.
3.6. Hydrolysis of ( )-mandelonitrile acetate or iso-pro-
penyl acetate 7
9. Kanerva, L. T.; Rahiala, K.; Sundholm, O. Biocatalysis
1994, 10, 169–180.
( )-Mandelonitrile acetate (or 7) (0.5 mmol) and CAL-
B (50 mg, 185 U) were added to dry toluene (5 mL)
that contained 120 ppm water or toluene (5 mL) that
contained 60 ppm water (obtained by drying the dry
toluene from Aldrich with molecular sieves). The reac-
tion mixture was stirred at room temperature (25°C).
The conversion was measured by chiral HPLC. After 5
h in the presence of 120 ppm H2O, 21.2% 7 and 18.6%
( )-mandelonitrile acetate were hydrolyzed. After 5 h in
the presence of 60 ppm H2O, 12.9% ( )-mandelonitrile
acetate were hydrolyzed.
10. Inagaki, M.; Hiratake, J.; Nishioka, T.; Oda, J. In Bio-
catalysts for Fine Chemicals; Roberts, S. M., Ed.; John
Wiley & Sons: Chichester, New York, Weinheim, Bris-
bane, Toronto, Singapore, 1999; Vol. 1:10, pp. 30–39.
11. Hanefeld, U.; Straathof, A. J. J.; Heijnen, J. J. J. Mol.
Catal. B Enzym. 2001, 11, 213–218.
12. Hanefeld, U.; Li, Y.-X.; Sheldon, R. A.; Maschmeyer, T.
Synlett 2000, 1775–1776.
13. Fontes, N.; Partridge, J.; Halling, P. J.; Barreiros, S.
Biotechnol. Bioeng. 2002, 77, 296–305.
14. Flanigen, E. M. In Introduction to Zeolite Science and
Practice; van Bekkum, H.; Flanigen, E. M.; Jacobs, P.
A.; Jansen, J. C., Eds.; Elsevier: Amsterdam, London,
New York, Oxford, Paris, Shannon, Tokyo, 2001; pp.
11–35.
15. Rakels, J. L. L.; Straathof, A. J. J.; Heijnen, J. J. Enzym.
Microb. Technol. 1993, 15, 1051–1056.
3.7. One-pot synthesis of (S)-mandelonitrile acetate 6
from benzaldehyde 1, acetone cyanohydrin 2 and iso-
propenyl acetate 7
1 (84.8 mg, 0.8 mmol), 2 (136 mg, 1.6 mmol) and 7 (240
mg, 2.4 mmol) were added to dry toluene (8 mL). To
this solution different −OH equivalents of Amberlite
IRA904 (−OH form, 30.8 mg/0.05 equiv., 154 mg/0.25
equiv., or 616 mg/1.0 equiv.) and CAL-B (80 mg, 296
U) were added. The reaction mixture was stirred at
40°C and 15 mL samples were taken for the analysis by
chiral HPLC. The enantiomeric excess of 6 was always
>99%. The results are shown in Table 1.
16. Weber, H. K.; Weber, H.; Kazlauskas, R. J. Tetrahedron:
Asymmetry 1999, 10, 2635–2638.
17. Wadsten, T.; Andersson, S. Acta Chem. Scand. 1959, 13,
1069–1074.
18. Veum, L.; Kuster, M.; Telalovic, S.; Hanefeld, U.;
Maschmeyer, T. Eur. J. Org. Chem. 2002, 1516–1522.
Table 1. One-pot synthesis of 6 from 1, 2 and 7 in the presence of different amounts of Amberlite (−OH form)
Amberlite (−OH equiv.)
Time (h)
Yield of 6 (%)
Conv. of 1 (%)
ee of 5 (%)
1.0
21
45
21
45
20
47
1.8
2.1
9.1
15.9
4.1
9.5
49.7
47.9
60.1
67.3
40.3
59.7
0.6
6.2
2.4
0.8
2.9
3.0
0.25
0.05