Synthesis of enantiomerically pure (S)-mandelic acid using an oxynitrilase-nitrilase bienzymatic cascade: A nitrilase surprisingly shows nitrile hydratase activity

Article abstract of DOI:10.1016/j.tetasy.2006.01.020

Benzaldehyde was converted into enantiomerically pure (S)-mandelic acid by sequential HCN addition and hydrolysis in the presence of a cross-linked enzyme aggregate composed of the (S)-selective oxynitrilase from Manihot esculenta and the non-selective recombinant nitrilase from Pseudomonas fluorescens EBC 191. Surprisingly, (S)-mandelic amide was formed in large amounts. It was shown, in separate experiments, that the nitrilase hydrolyses (S)-mandelonitrile into an approx. equimolar mixture of acid and amide, whereas with the (R)-enantiomer only 10% of amide was formed.

Full text of DOI:10.1016/j.tetasy.2006.01.020

Tetrahedron: Asymmetry 17 (2006) 320–323
Synthesis of enantiomerically pure (S)-mandelic acid using an
oxynitrilase–nitrilase bienzymatic cascade: a nitrilase
surprisingly shows nitrile hydratase activity
a
a
b
a
Cesar Mateo, Andrzej Chmura, Sven Rustler, Fred van Rantwijk,
b
a,
*
Andreas Stolz and Roger A. Sheldon
a
Laboratory of Biocatalysis and Organic Chemistry, Delft University of Technology, Julianalaan 136, 2628 BL Delft, The Netherlands
b
Institut f u¨ r Mikrobiologie, Universit a¨ t Stuttgart, Allmandring 31, D-70550 Stuttgart, Germany
Received 7 December 2005; accepted 24 January 2006
Abstract—Benzaldehyde was converted into enantiomerically pure (S)-mandelic acid by sequential HCN addition and hydrolysis in
the presence of a cross-linked enzyme aggregate composed of the (S)-selective oxynitrilase from Manihot esculenta and the non-selec-
tive recombinant nitrilase from Pseudomonas fluorescens EBC 191. Surprisingly, (S)-mandelic amide was formed in large amounts. It
was shown, in separate experiments, that the nitrilase hydrolyses (S)-mandelonitrile into an approx. equimolar mixture of acid and
amide, whereas with the (R)-enantiomer only 10% of amide was formed.
ꢀ
2006 Elsevier Ltd. All rights reserved.
1
. Introduction
route A
HCN
HnL
pH 4-5
OH
CN
OH
COOH
conc. HCl,
reflux
R
R
In recent years, there has been increasing interest in
the synthesis of enantiopure 2-hydroxy acids via
asymmetric enzymatic transformations; two pathways
have risen into prominence. The first is based on the
enantioselective hydrocyanation of an appropriate
aldehyde in the presence of an oxynitrilase (hydroxy-
nitrile lyase, HnL, E.C. 4.1.2.10), which gives rise to
the corresponding enantiomerically pure cyanohydrin,
followed by chemical hydrolysis in the presence of a
strong acid (Fig. 1A). This latter step generates copious
quantities of salt and is not compatible with sensitive
functional groups, which is a serious drawback. Alter-
natively, enantiopure 2-hydroxy acids can be obtained
via a dynamic kinetic resolution of the (chemically
synthesised) cyanohydrin in the presence of an enantio-
selective nitrilase (NLase, E.C. 3.5.5.1, see Fig. 1B).
This latter methodology, which is industrially applied
H
(
both enantiomers)
OH
R
O
OH
H2O
R
R
CN
HCN
pH 7
+
R
COOH
route B
OH
(R)-NLase
CN
Figure 1. Synthetic routes to enantiomerically pure 2-hydroxy acids,
via oxynitrilase (hydroxynitrile lyase, HnL) catalysed enantioselective
hydrocyanation (route A) and (R)-nitrilase (NLase) mediated dynamic
kinetic resolution (route B).
We surmised that a fully enzymatic route to the (S)-
acids could be possible by combining an (S)-selective
oxynitrilase and a non-selective nitrilase in a bienzy-
matic cascade (see Fig. 2). In addition to being more
environmentally acceptable, the mild reaction condi-
tions of the combined enzymatic reaction are compatible
with a wide range of hydrolisable groups.
1
for the multiton-scale synthesis of (R)-mandelic acid,
is restricted to (R)-2-hydroxy acids, because no (S)-
selective nitrilases for cyanohydrin substrates have yet
been identified.
Potential incompatibilities that must be resolved are the
pH and reaction medium. Oxynitrilase mediated hydro-
cyanations are preferably carried out at pH < 5 to
*
Corresponding author. Tel.: +31 15 278 2683; fax: +31 15 278
415; e-mail: r.a.sheldon@udelft.nl
1
0957-4166/$ - see front matter ꢀ 2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.tetasy.2006.01.020

C. Mateo et al. / Tetrahedron: Asymmetry 17 (2006) 320–323
321
Table 1. The effect of pH on the hydrocyanation of 1a in the presence
of MeHNL
H
OH
CN
OH
R COOH
HCN
S)-HnL
H2O
a
R
O
R
(
NLase
pH
5.5
Time (min)
120
Conv (%)
ee (% S)
1
(S)-2
(S)-3
90
94
95
95.4
94.7
94.3
1
2
80
40
a: R = C H5
6
6
6
.0
.5
60
20
80
90
94
94
92.9
90.9
88.4
Figure 2. Bienzymatic procedure for the synthesis of (S)-2-hydroxy-
acids, using an (S)-specific HnL and a nonspecific NLase in tandem.
1
1
60
91
93
94
89.5
87.8
85.8
9
0
suppress the competing uncatalysed hydrocyanation.2
The background reaction can be further reduced by
120
a
Reaction conditions: 1a (0.1 mmol), HCN (0.5 mmol), MeHnL
CLEA (0.25 mg, 19.6 U) in 0.1 mL, 0.1 M citrate buffer and 0.9 mL
DIPE at rt.
2
the use of a biphasic aqueous–organic or micro-aque-
3
ous reaction medium. Such reaction conditions are
tolerated rather well by the readily available (S)-selective
2
,4
HnL from Manihot esculenta (MeHnL). Nitrilases, in
contrast, are preferentially used at a pH close to 7 and
are readily deactivated by organic solvents. Rendering
the nitrilase compatible with the conditions of enzymatic
hydrocyanation is obviously an issue on which the
success of the methodology depends. We reasoned that
tion, at the cost of a significant loss in enantioselectivity
(Table 1).
The enzymatic hydrolysis of mandelonitrile 2a was
investigated in some detail, as this is evidently the criti-
cal step. It should be fast, moreover, to avoid accumula-
tion—and possible racemisation—of (S)-2a. Few
nitrilases displayed a useful activity at pH 5.5–6; most
of these were strongly (>90% ee) biased towards the
(R)-enantiomer and only converted (S)-mandelonitrile
immobilisation as a cross-linked enzyme aggregate
5
(
CLEA) would present a solution to this latter
problem. We herein report the application of our bien-
zymatic methodology to the synthesis of enantiomeri-
cally pure (S)-mandelic acid.
6
sluggishly, if at all. A recombinantly expressed nitrilase
7
from P. fluorescens EBC 191 (PfNLase), was an excep-
2
. Bienzymatic synthesis of (S)-mandelic acid
tion as it converted (S)- and (R)-2a at comparable rates
pH 5.5, which we adopted as a compromise pH for the
6
As noted above, enzymatic hydrocyanations are prefer-
ably performed at pH < 5, to suppress the non-enzy-
matic background reaction, whereas the optimum pH
of the common NLases is 7. A compromise in pH is
obviously required and we accordingly assessed the
effects of the pH on the MeHnL-mediated hydrocyana-
tion of benzaldehyde 1a (Fig. 2) in a biphasic aqueous-
diisopropyl ether (DIPE) medium. We found that
enantioselectivity was maintained at pH 5.5, provided
that the aqueous buffer phase accounted for <10% of
bienzymatic reactions.
Accordingly, we conducted experiments with CLEAs of
MeHnL and PfNLase in tandem in a 90:10 DIPE buffer
pH 5.5 medium. The reaction proceeded to nearly full
conversion (Fig. 3a) and the product ee was 94%. Com-
bining both enzymes in a bienzymatic catalyst (combi-
CLEA, Fig. 3b) resulted in further improvement and
8
98% enantiomerically pure (S)-3a was obtained. It
would seem that the nitrile intermediate is immediately
hydrolysed in the combi-CLEA particles, which sup-
presses diffusion into the water phase and possible
4
the reaction volume. In experiments at pH 6.0 and
6
.5, we surprisingly found a progressively faster reac-
1
0
8
6
4
2
0
10
8
6
4
2
0
0
5
10
15
20
0
5
10
15
20
(
a)
(b)
Time (h)
Time (h)
Figure 3. Time course of the bienzymatic synthesis of (S)-mandelic acid from benzaldehyde and HCN in the presence of a MeHnL and PfNLase in
8
9
0:10 DIPE buffer pH 5.5, rt, (a) two CLEAs, (b) combi-CLEA; legend: 1a: (d), 2a: (m), 3a: (j), 4a: (r).

322
C. Mateo et al. / Tetrahedron: Asymmetry 17 (2006) 320–323
OH
CN
OH
COOH +
OH
CONH2
H O
2
PfNLase
buffer pH 6,
0
˚C
2a
3a
4a
Figure 4. Hydrolysis of 2a in the presence of P. fluorescens NLase.
1
0
8
6
4
2
0
10
8
R
S
6
4
2
0
0
50
100
150
200
250
300
350
0
50
100
150
200
250
300
350
(
a)
(b)
Time (min)
Time (min)
Figure 5. Time course of the hydrolysis of (R)- and (S)-mandelonitrile in the presence of P. fluorescens NLase at pH 6, 0 ꢁC; legend: 2a: (m), 3a: (j),
a: (r).
4
racemisation. A full analysis of the reaction products
showed, however, that very appreciable amounts of
5). To avoid racemisation of the nitrile, the reactions
were performed at 0 ꢁC. It became clear that only a
minor amount of amide was formed from the (R)-enan-
tiomer (acid/amide ꢀ8), whereas the amide was a major
product formed from the (S)-enantiomer under other-
wise identical conditions (acid/amide <1, see Fig. 5).
Stereochemical integrity was fully maintained under
the reaction conditions and 3a and 4a were formed with
complete retention of configuration, as would be
expected. It should be noted that (S)-4a is a reaction
product, and not an intermediate in the formation of
(S)-3a, as the acid/amide ratio stays constant through-
out the reaction. Neither does (S)-4a result from a
reverse reaction of (S)-3a and ammonia, since such a
reaction did not take place in a separate experiment.
(
S)-mandelic amide ((S)-4a, see Fig. 3, approx. 40%)
accompanied the formation of (S)-3a (see later).
We finally attempted to improve the reaction rate by
increasing the pH to 6, expecting that the in situ hydrol-
ysis of (S)-2a would prevent its racemisation. The
increase in rate was quite modest, however, and came
at the expense of a slight decrease in product ee (96%).
In conclusion, the formation of nearly enantiomerically
pure (S)-3a proves that the bienzymatic methodology is
sound and that racemisation of the intermediate can be
avoided. The unexpected formation of large amounts of
(
S)-mandelic amide somewhat reduces the immediate
practical value of our procedure, but we are confident
that amide formation can be obviated by proper genetic
engineering of the NLase.
We wished to exclude any possibility that the near-
absence of the amide in the hydrolysis product of (R)-2a
was caused by spurious amidase activity of PfNLase,
rather than by the reactant stereochemistry. Accord-
ingly, when we incubated (R)-4a with PfNLase at either
pH 5.5 or 7, no reaction was observed. One more ques-
tion remained as to whether the formation of (S)-amide
could be ascribed to the presence of a contaminating
(S)-selective nitrile hydratase, but SDS-PAGE analysis
3. Amide formation
The formation of minor amounts of amides over the
course of NLase-catalyses nitrile hydrolysis has been
9
6
noted and the nitrilase from Arabidopsis thaliana pro-
showed that no contaminating proteins were present.
miscuously produced amides as the major product in
1
0
the hydrolysis of a-fluoronitriles as well as other elec-
1
1
tron-deficient nitriles. One would suppose that minor
amounts of amide formation have often gone unnoticed
in the past, due to the presence of contaminating ami-
dases (Fig. 4).
4. Conclusion
We can conclude that the amides are products of the
NLase and that the stereochemical configuration of
We investigated the amide formation from 2a more clo-
sely, by subjecting enantiomerically pure (R)- and (S)-2a
to hydrolysis in the presence of PfNLase (see Figs. 4 and
These experiments were carried out at somewhat different conditions
as they are part of a larger investigation.

C. Mateo et al. / Tetrahedron: Asymmetry 17 (2006) 320–323
323
the nitrile exerts a major influence on their formation, in
addition to the nature of the NLase and the electron
density at the a-C atom in the nitrile.
6. Kiziak, C.; Conradt, D.; Stolz, A.; Mattes, R.; Klein, J.
Microbiology 2005, 151, 3639–3648.
7
8
. Layh, N.; Parratt, J.; Willets, A. J. Mol. Catal. B: Enzym.
998, 5, 467–474.
1
. Combi-CLEA preparation: MeHnL (3 kU, from J u¨ lich
Fine Chemicals) and semi-purified PfNLase (85 U, see
below) were mixed with dextranpolyaldehyde solution
Acknowledgements
5
b
(
4 mL, cross-linking agent), 0.5 M phosphate buffer
This research was supported by the Netherlands
Research Council NWO and the Deutsche Fors-
chungsgemeinschaft in the framework of the CERC-3
programme. A grant to C.M. by the Spanish Ministry
of Education under the postdoctoral fellowships
programme is gratefully acknowledged. This work was
supported by the COST Action D25.
pH 7.5 (2 mL) and 1,2-dimethoxyethane (8 mL, aggrega-
tion agent). The mixture was stirred overnight at 40 ꢁC
and centrifuged. The pellet was resuspended in 0.1 M
sodium bicarbonate solution containing 1 g L of sodium
borohydride (40 mL) to reduce the Schiff’s bases, stirred
À1
for 45 min at 40 ꢁC, washed twice with Milli-Q water
(
40 mL), resuspended in 5 mL 0.5 M citrate buffer pH 5.5
and stored at 4 ꢁC.
One unit (U) of PfNLase will convert 1 lmol of racemic
mandelonitrile per min at rt and pH 7.
References
One pot synthesis of (S)-mandelic acid: Combi-CLEA
suspension (0.5 mL), DIPE, HCN solution in DIPE
(0.25 mmol)—CAUTION: HCN is highly poisonous; use
a well-ventilated hood and keep a HCN detector switched
on—were combined and DIPE was added to a total
volume of 5 mL. Benzaldehyde (0.05 mmol) was added
and the mixture stirred vigorously. Samples were with-
drawn periodically.
1
. (a) Ress-L o¨ schke, M.; Friedrich, T.; Hauer, B.; Mattes,
R.; Engels, D. (BASF AG), DE 19848129, 2000 [Chem.
Abstr. 2000, 132, 292813]; (b) Yamaguchi, Y.; Ushigome,
M.; Kato, T. (Nitto Chem. Ind.), EP 773297, 1997 [Chem.
Abstr. 1997, 127, 4190]; (c) Endo, T.; Tamura, K. (Nitto
Chem. Ind.), EP 449648, 1991 [Chem. Abstr. 1992, 116,
5
338].
. Schmidt, M.; Griengl, H. Top. Curr. Chem. 1999, 200,
93–226.
. (a) Han, S.; Lin, G.; Li, Z. Tetrahedron: Asymmetry 1998,
, 1835–1838; (b) Van Langen, L. M.; van Rantwijk, F.;
Analysis: the reaction progress was monitored by reversed-
phase HPLC (Merck 4.6 · 50 mm Chromolith Speed-
TM
2
3
1
ROD RP18e, ACN–H
2
O (5:95), 0.03% heptaflurobutyric
À1
acid, 30 mM ammonium formate pH 2.3 at 1 mL min ).
The enantiomeric purity was measured using chiral HPLC
(Daicel 4.6 · 250 mm 10l Chiralcel OD, hexane–isopropyl
9
Sheldon, R. A. Org. Proc. Res. Dev. 2003, 7, 828–831.
. Chmura, A.; Mateo, C.; Van der Kraan, G. M.; Celik, A.;
Van Langen, L. M.; Van Rantwijk, F.; Sheldon, R. A., in
preparation.
À1
4
5
alcohol (80:20), 0.1% TFA at 0.5 mL min ).
9. (a) Hook, R. H.; Robinson, W. G. J. Biol. Chem. 1964,
239, 4263–4267; (b) Stevenson, D. E.; Dumas, F.;
Groleau, D.; Mihoc, A.; Storer, A. C. Biotechnol. Appl.
Biochem. 1992, 15, 283–302.
. (a) Cao, L.; van Rantwijk, F.; Sheldon, R. A. Org. Lett.
2
000, 2, 1361–1364; (b) Mateo, C.; Palomo, J. M.; van
Langen, L. M.; van Rantwijk, F.; Sheldon, R. A.
Biotechnol. Bioeng. 2004, 86, 273–276; (c) Van Langen,
L. M.; Selassa, R. P.; Van Rantwijk, F.; Sheldon, R. A.
Org. Lett. 2005, 7, 327–329.
10. Effenberger, F.; Oßwald, S. Tetrahedron: Asymmetry 2001,
12, 279–285.
11. Osswald, S.; Wajant, H.; Effenberger, F. Eur. J. Biochem.
2002, 269, 680–687.