The combi-CLEA approach: Enzymatic cascade synthesis of enantiomerically pure (S)-mandelic acid

Article abstract of DOI:10.1016/j.tetasy.2013.08.013

Enantiomerically pure (S)-mandelic acid was synthesised from benzaldehyde by sequential hydrocyanation and hydrolysis in a bienzymatic cascade at starting concentrations up to 0.25 M. A cross-linked enzyme aggregate (CLEA) composed of the (S)-selective oxynitrilase from Manihot esculenta and the non-selective nitrilase from Pseudomonas fluorescens EBC 191 was employed as the biocatalyst. The nitrilase produces approx. equal amounts of (S)-mandelic acid and (S)-mandelic amide from (S)-mandelonitrile under standard conditions, but we surprisingly found that high (up to 0.5 M) concentrations of HCN induced a marked drift towards amide production. By including the amidase from Rhodococcus erythopolis in the CLEA we obtained (S)-mandelic acid as the sole product in 90% yield and >99% enantiomeric purity.

Full text of DOI:10.1016/j.tetasy.2013.08.013

Tetrahedron: Asymmetry 24 (2013) 1225–1232
Contents lists available at ScienceDirect
Tetrahedron: Asymmetry
journal homepage: www.elsevier.com/locate/tetasy
The combi-CLEA approach: enzymatic cascade synthesis
of enantiomerically pure (S)-mandelic acid
Andrzej Chmura a, Sven Rustler b, Monica Paravidino a, Fred van Rantwijk a, Andreas Stolz b,
Roger A. Sheldon a,
⇑
a Laboratory of Biocatalysis and Organic Chemistry, Department of Biotechnology, Delft University of Technology, 2628 BL Delft, The Netherlands
b Institut für Mikrobiologie, Universität Stuttgart, Allmandring 31, D-70550 Stuttgart, Germany
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 15 July 2013
Accepted 27 August 2013
Enantiomerically pure (S)-mandelic acid was synthesised from benzaldehyde by sequential hydrocyana-
tion and hydrolysis in a bienzymatic cascade at starting concentrations up to 0.25 M. A cross-linked
enzyme aggregate (CLEA) composed of the (S)-selective oxynitrilase from Manihot esculenta and the
non-selective nitrilase from Pseudomonas fluorescens EBC 191 was employed as the biocatalyst. The nitri-
lase produces approx. equal amounts of (S)-mandelic acid and (S)-mandelic amide from (S)-mandelonit-
rile under standard conditions, but we surprisingly found that high (up to 0.5 M) concentrations of HCN
induced a marked drift towards amide production. By including the amidase from Rhodococcus erythopolis
in the CLEA we obtained (S)-mandelic acid as the sole product in 90% yield and >99% enantiomeric purity.
Ó 2013 Elsevier Ltd. All rights reserved.
1. Introduction
interesting example within this context. Route A involves enzy-
matic synthesis of the enantiomerically pure (R)- or (S)-cyanohy-
Cascade conversion—combined (catalytic) procedures without
recovery of intermediate products—is considered an important
future methodology for reducing the environmental footprint of
organic synthesis.1 Industrial chemoenzymatic synthesis of enanti-
omerically pure 2-hydroxycarboxylic acids2 (see Fig. 1) presents an
drin, employing the appropriate hydroxynitrile lyase (HnL,
oxynitrilase, E.C. 4.1.2.10), followed by hydrolysis with strong
acid.3 Alternatively (route B), a dynamic kinetic resolution of the
chemically synthesised cyanohydrin in the presence of an enantio-
selective nitrilase (NLase, E.C. 3.5.5.1) can be employed.
Both routes suffer from limitations and disadvantages. In view
of the sustainability issues it is worth noting that route A generates
copious quantities of salt. It is not compatible, moreover, with sen-
sitive functional groups, due to the harsh hydrolysis conditions.
Route B, which is quite efficient and is industrially employed in
the multiton-scale industrial synthesis of (R)-mandelic acid,4 is re-
stricted to (R)-2-hydroxyacids since, so far, no NLase with prefer-
ence for (S)-cyanohydrins has been identified.5
OH
OH
+ NH4Cl
R
COOH
R
CN
HnL
2 H2O
conc. HCl,
reflux
both enantiomers
H
HCN, pH 4-5
R
O
Combining the enzymatic steps—hydrocyanation and nitrile
hydrolysis—into a bienzymatic cascade is particularly advanta-
geous in the synthesis of (S)-2-hydroxyacids and would avoid the
weaknesses of route A—salt production and incompatibility with
hydrolysable groups. As an additional bonus, the cascade would
lessen (but not entirely obviate) concerns with regard to racemisa-
tion of the cyanohydrin and would pull an incomplete hydrocyana-
tion equilibrium towards complete conversion.
An efficient cascade requires any potential incompatibilities
with regard to pH and reaction medium to be resolved; as will be-
come clear, such issues are particularly relevant here. HnL mediated
hydrocyanation, on the one hand, is preferably carried out at pH <5
to suppress the competing spontaneous hydrocyanation.6,7 It is
common practice, moreover, to conduct enzymatic hydrocyanation
OH
OH
R
CN
(R)-NLase
+
OH
+
HCN, pH 7
R
COO-NH4
only (R)
2 H2O
R
CN
Figure 1. Synthetic routes to enantiomerically pure 2-hydroxyacids, via hydroxy-
nitrile lyase (HnL, oxynitrilase) catalysed enantioselective hydrocyanation (route A)
or, alternatively, dynamic kinetic resolution mediated by an (R)-specific nitrilase
(NLase, route B).
⇑
Corresponding author. Tel.: +31 15 278 2683; fax: +31 15 278 1415.
E-mail address: r.a.sheldon@tnw.tudelft.nl (R.A. Sheldon).
0957-4166/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.tetasy.2013.08.013

1226
A. Chmura et al. / Tetrahedron: Asymmetry 24 (2013) 1225–1232
in biphasic aqueous-organic6 or micro-aqueous medium8 to reduce
the background reaction even further. Most NLases, on the other
hand, have an optimum pH of 7–9 and do not tolerate organic sol-
vents well.9
NLase.10c Indeed, 90% of the NLase starting activity was recovered
in the combi-CLEA, which is important to avoid accumulation of
cyanohydrin in the reaction mixture.
These incompatibilities can be solved by immobilising a suit-
able HnL and NLase as cross-linked enzyme aggregates (CLEAs),10
preferably in a combi-CLEA, as we have demonstrated in our
proof-of-principle paper.11 Alternatively, a whole-cell biocatalyst
expressing both enzymes may be employed.12 High conversions
and near-quantitative enantioselectivities were demonstrated
with the combi-CLEA but from the synthetic viewpoint the reac-
tant concentrations were rather low. Increasing these to >0.1 M
would be desirable but may not be trivial, since NLase inhibition
by aldehydes and HCN has been reported.12b,13 A further shortcom-
ing of the cascade procedure, as originally published, is the forma-
tion of amide as a major side product,11,14 which detracts from the
green as well as the practical value of the procedure.
2.2. Cascade synthesis of (S)-mandelic acid
Preliminary experiments have been performed with 10 mM
benzaldehyde (1, see Fig. 2) starting concentration and a fourfold
excess of HCN11 to reduce enzymatic dehydrocyanation to a mini-
mum.8c These reactions were performed at pH 5.5 as a compromise
between maintaining the enantiomeric purity of the HCN adduct
(2) and NLase activity.11,16 (S)-Mandelic acid ((S)-3) was produced
with high enantiomeric purity (cf. Table 1). Significant amounts of
(S)-mandelic amide ((S)-4) were also formed and we have previ-
ously shown that amide formation from (S)-2 is a major side-reac-
tion in the presence of PfNLase, whereas it is barely significant with
(R)-2.14 This amide side-product formation, which severely de-
tracts from the green as well as the practical value of the proce-
dure, will be addressed below.
In the present Letter we will report the application of the com-
bi-CLEA approach to the cascade synthesis of (S)-mandelic acid at
synthetically relevant concentrations and address the formation
of amide side product as well.
Table 1
Cascade synthesis of (S)-mandelic acida
2. Results and discussion
1 (mM)
HCN (mM)
Time (h)
2 (mM)
3 (mM)
ee (%)
4 (mM)
Our research was focused on the cascade synthesis of (S)-man-
delic acid, as it is a more challenging target than the (R)-enantio-
mer, which is already produced in an efficient chemoenzymatic
procedure.4 The HnL from Manihot esculenta (MeHnl) was the obvi-
ous choice as (S)-selective hydrocyanation biocatalyst, on account
of its operational stability8c and easy availability. A survey of
readily available NLases revealed that only the NLase from Pseudo-
monas fluorescens EBC19115 (PfNLase) converted (R)- and (S)-
mandelonitrile at comparable rates16,17 whereas other NLases were
(R)-selective to a varying extent.5,17 PfNLase was also exceptional
in maintaining activity down to pH 5.516,17 and is obviously the en-
zyme of choice for our purpose. Since proper immobilisation is a
crucial success factor, we set out to investigate the immobilisation
of MeHnL and PfNLase in a combi-CLEA.
10
50
125
210
415
750
1.6
1.6
1.6
1.6
2.0
<0.2
<1
<1
8
6.0
14
23
36
111
99
99
96
96
96
4.0
10
18
40
155
25
42
83
250
<1
a
Reaction conditions: combi-CLEA prepared from MeHnL (14.1 mg) and semi-
purified PfNLase (4.8 mg) in 20 mM citrate buffer pH 5.5 (2 mL); 1 and HCN as
appropriate in diisopropyl ether (4.7 mL) at 25 °C. Product concentrations are from
HPLC analysis.
The operational stability of the combi-CLEA was assessed in a
wash-and-reuse cycle. The recovered enzymatic activities in the
CLEA (see Fig. 3) were calculated from the reaction progress. An
activity loss was observed after the fourth cycle, which is tenta-
tively ascribed to mechanical losses as the biocatalyst particles
became difficult to spin down at this point.
2.1. Preparing a combi-CLEA
Fast and nearly complete conversion of 1 and 2 was still
observed when the starting concentration of 1 was increased to
the synthetically more relevant value of 0.25 M and the enantio-
meric purity of the products was P96% (see Table 1). As before,
the reactions were performed in a biphasic aqueous-organic med-
ium to suppress the uncatalysed hydrocyanation while preventing
precipitation of the products.
It became apparent from the results in Table 1 that the yield of
(S)-4 increased at higher starting concentrations, from 40% at
10 mM 1 to 58% at 0.25 M, which came to us as a surprise but
was also observed with a whole cell biocatalyst.12b One could sur-
mise, by analogy with the pH effects on the hydrolysis of phenyl-
acetonitrile,14 that the high final concentration of 3 induces an
upward pH shift.19 If this were the case, the enantiomeric purity
of the products would be expected to suffer, as it is highly sensitive
to small pH changes.11 Since this was not observed, it would rather
A cross-linked enzyme aggregate of MeHnL and PfNLase was
prepared using the protocol developed for the NLase,10c which in-
volves precipitation with 1,2-dimethoxyethane and cross-linking
with dextran polyaldehyde, followed by reduction with sodium
borohydride. The relative amounts of MeHnL and PfNLase in the
CLEA were based on experiments with separate biocatalysts.11
The activity recoveries of the enzymes were measured separately
(see Table 2). We used synthetic rather than—more convenient—
photometric assays because the outcome of these may be too
low, due to diffusion limitation in the CLEA at photometric reactant
concentrations.8c 57% Of the HnL starting activity was recovered in
the combi-CLEA, using a hydrocyanation assay of trans-cinnamic
aldehyde.8c,18 This outcome is much lower than the 93% recovery
that we achieved using an optimised CLEA protocol,8c but here
our objective was to maintain the activity of the more sensitive
H
OH
OH
OH
MeHnL
HCN
COO-NH4
CONH2
+
PfNLase
2 H2O
O
CN
+
3
4
(S)-
1
2
(S)-
(S)-
Figure 2. Bienzymatic procedure for the synthesis of (S)-2-mandelic acid (3), using the (S)-specific MeHnL and the non-specific PfNLase in tandem.

A. Chmura et al. / Tetrahedron: Asymmetry 24 (2013) 1225–1232
1227
hence, that (S)-2 partially protects the enzyme against HCN-in-
duced deactivation.
140
120
100
80
We also observed (Fig. 5b) that amide production became more
prominent at increasing HCN concentrations, in agreement with
our results presented in Table 1 and Ref. 12b. Hence, we confi-
dently conclude that the selectivity drift in the cascade synthesis
is due to interference of HCN with the NLase. Whether the phe-
nomenon is general with NLases and a wider range of nitrile sub-
strates will be the subject of future research.
60
40
20
2.4. In situ amide hydrolysis
0
0
1
2
3
4
5
6
7
The trend, noted above, towards increased amide production at
synthetically relevant concentrations prompted us to design a
solution. A fairly obvious approach is to conduct the reactions in
the presence of an amidase and convert (S)-4 into the desired prod-
uct (S)-3. Preliminary tests showed that racemic 4 was completely
hydrolysed in the presence of penicillin acylase as well as the ami-
dase from Rhodococcus erythopolis MP5020,21 (RheAMase). Both en-
zymes, when added to the cascade reaction mixture, accordingly
mediated the in situ hydrolysis of all 4 into 3, demonstrating their
compatibility with the reaction conditions. The lack of enantiose-
lectivity, implicated by the test results, is an important aspect
when considering the possible application of the cascade method
to (R)- as well as (S)-hydroxy acids.17 A similar reasoning prompted
us to focus the further development on RheAMase on account of its
relaxed substrate specificity,20 in contrast to penicillin acylase,
which is restricted to phenylacetic acid derivatives.
Cycle nr.
Figure 3. Activity recovery of MeHnL (N) and PfNLase (d) in the combi-CLEA upon
recycling and reuse. The CLEA particles resisted spinning down at Cycle nr. >4.
seem that an inhibitory effect of the reactants is involved (see
later).
2.3. Effects of high reactant concentrations on PfNLase
The shifting acid/amide ratio mentioned above prompted us to
investigate the possible inhibition or deactivation of PfNLase by
high reactant concentrations in some detail. It is worth noting here
that inhibition by aldehydes and HCN of PfNLase12b as well as the
NLase from Alcaligenes faecalis13 has been reported.
Exposing the combi-CLEA to 10 mM 1 for 1 h resulted in a slight
(10%) loss of NLase activity, increasing to 20% activity loss at
250–500 mM 1, as measured in the hydrolysis of phenylacetoni-
trile (data not shown). Hence, it would seem that 1 has little intrin-
sic effect on PfNLase if protected in a combi-CLEA.
HCN presented a more complicated picture as a low (25 mM)
concentration had little effect, whereas 125 mM HCN after 1 h of
preincubation caused 45% activity loss in the NLase (see Fig. 4a).
At 450 mM HCN deactivation was nearly complete after 1 h of pre-
incubation and could not be recovered by washing with aqueous
buffer. When the residual activity was measured over time, it
became clear that free PfNLase was deactivated faster than the
combi-CLEA (Fig. 4b).
The effects of HCN on the activity and selectivity of PfNLase
were investigated in some more detail in the hydrolysis of (S)-2.
Reactions were performed in the presence of a PfNLase CLEA with
0–400 mM added HCN, using the same biphasic system that was
employed in the cascade synthesis. An effect on the reaction rate
became apparent (Fig. 5a), but complete deactivation was not ob-
served, even after >2 h exposure to 400 mM HCN. It would seem,
2.5. Optimising precipitation and cross-linking
Precipitation of RheAMase with a range of common agents,8c
followed by redissolution in aqueous buffer, revealed that the
activity recovery was acceptable only with acetonitrile, 1,2-dime-
thoxyethane, tert-butyl alcohol and saturated aqueous ammonium
sulfate. Subsequently, the newly formed aggregates were incu-
bated in the precipitant for up to 20 h and sampled at appropriate
intervals. With ammonium sulfate the activity recovery upon
redissolution of the samples was a constant 45% of the starting
activity, whereas in the other media a nearly complete loss of
activity after 2 h of exposure was observed. Hence, ammonium sul-
fate was the precipitant of choice for the preparation of the ami-
dase CLEA.
These results made it necessary to adapt the procedure for pre-
paring a CLEA of MeHnL and PfNLase. An incubation/redissolution
test as described above showed excellent activity retention of
MeHnL, whereas aggregates of PfNLase rapidly lost activity upon
>2 h incubation in saturated ammonium sulfate. These data,
(a)
(b)
100
120
100
80
60
40
20
0
80
60
40
20
0
0
200
400
600
0
2
4
6
HCN (mM)
Time (h)
Figure 4. Residual activity of PfNLase upon exposure to HCN in the hydrolysis of phenylacetonitrile; (a) CLEA, 1 h incubation; (b) effects of 125 mM HCN versus time on the
residual activity; CLEA (ꢀ), free enzyme (h).

1228
A. Chmura et al. / Tetrahedron: Asymmetry 24 (2013) 1225–1232
(a)
(b)
100
2.00
80
60
40
20
0
1.50
1.00
0.50
0.00
0
1
2
3
0
100
200
300
400
Time (h)
HCN (mM)
Figure 5. Effects of HCN on the hydrolysis of 10 mM (S)-2 in the presence of PfNLase CLEA in 70:30 DIPE-buffer pH 5.5 at rt; (a) time-course, HCN 0 mM (ꢀ), 200 mM (4),
300 mM (s), 400 mM (h); (b) acid/amide ratio versus HCN concentration.
although somewhat disappointing, served as a basis for the cross-
linking experiments.
product resulting from the HnL assay should not be consumed by
the NLase, and the NLase assay substrate should not be prone to
dehydrocyanation; neither should it produce an amide.
Single enzymes (MeHnL, PfNLases and RheAMase) were aggre-
gated with saturated ammonium sulfate solution and subse-
quently cross-linked with dextran polyaldehyde solution at
various concentrations for 2–23 h. Schiff’s base bonds in the
cross-links were made permanent by reduction with cyanoboro
hydride, which is required by the NLase.10c We found that cross-
linking times should be limited to 2 h, since the activity recovery
(see Fig. 6a–c) of all three enzymes suffered when it was pro-
longed. MeHnL and PfNLase showed little tendency to leach, as
shown by the minute activity recovery from the supernatants.
RheAMase, in contrast, required a dextran polyaldehyde concen-
tration of 3 mg mLꢀ1 to suppress leaching (see Fig. 6d). At this con-
centration the recovery of NLase and amidase activity (>70%) was
good while the activity recovery of MeHnL (40%) was acceptable.
The MeHnL in the triple CLEA was assayed in a synthetic hydro-
cyanation of trans-cinnamic aldehyde, as before.18 Phenylacetoni-
trile was adopted as the NLase assay substrate since it is not
sensitive to dehydrocyanation, neither is amide produced in the
presence of PfNLase. The AMase was assayed in the hydrolysis of
4. A triple CLEA was prepared by successively precipitating MeHnL,
PfNLase and RheAMase in saturated ammonium sulfate, followed
by cross-linking with 2.5 mg mLꢀ1 of dextran polyaldehyde and
subsequent cyanoborohydride reduction. The activity recovery
was 45% (MeHnL), 58% (PfNLase) and 57% (RheAMase).
Our common practice, up to here, was to store the CLEAs in the
reaction buffer (20 mM citrate buffer pH 5.5) ready for immediate
use. Monitoring the activity of the enzymes in the CLEA revealed
this practice to be far from optimal, although acceptable for storage
over a few days (see Fig. 7a). MeHnL in the CLEA rapidly lost 35% of
the starting activity but the deactivation of PfNLase upon storage at
pH 5.5 for >10 d was particularly serious. We fortunately found
that the precipitation medium, saturated ammonium sulfate, is
also the optimum storage medium in which the activity retention
2.6. A triple CLEA
Rational optimisation of the protocol for a triple CLEA of MeHnL,
PfNLase and RheAMase requires enzyme-specific non-spectropho-
tometric assay techniques. More specifically, the HCN addition
(a)
(b)
100
100
80
60
40
20
0
80
60
40
20
0
0
2
4
6
8
0
1
2
3
4
X-Linker (mg/mL)
X-Linker (mg/mL)
(c)
(d)
80
60
40
20
0
30
20
10
0
0
1
2
3
4
0
1
2
3
4
X-Linker (mg/mL)
X-Linker (mg/mL)
Figure 6. Effects of the dextran polyaldehyde concentration and cross-linking time on the activity recovery in the CLEAs; (a) MeHnL, (b) PfNLase, (c) RheAMase, (d) leaching
of RheAMase activity from the CLEA. Cross-linking time: 2 h (}), 4 h (h), 6 h (4), 23 h (s).

A. Chmura et al. / Tetrahedron: Asymmetry 24 (2013) 1225–1232
1229
bienzymatic cascade employing a combi-CLEA of Manihot esculenta
hydroxynitrile lyase and the nitrilase from Pseudomonas fluorescens
EBC191. The conversion was fast and complete but the production
of large amounts of (S)-mandelic amide is a major drawback. This
latter problem was convincingly solved by including an amidase in
the CLEA, which proved that employing a triple CLEA in synthetic
biocatalysis is a viable option. We surprisingly found that the
selectivity of the nitrilase drifts towards amide at high HCN
concentrations.
(a) 140
120
100
80
60
40
20
0
4. Experimental section
0
10
20
30
4.1. Materials and methods
Time (d)
(b) 160
Semi-purified (S)-hydroxynitrile lyase from M. esculenta
(MeHnL, E.C. 4.1.2.10, protein content 88 mg mLꢀ1) was obtained
from Jülich Fine Chemicals (Jülich, Germany), which has termi-
nated delivery. Soluble MeHnL is available from Sigma-Aldrich
and ASA Spezialenzyme GmbH (URL: http://www.asa-enzyme.de).
MeHnL CLEA was received from CLEA Technologies B.V. as a gift.
Nitrilase from P. fluorescens EBC191 (PfNLase, E.C. 3.5.5.1), heterol-
ogously expressed in Escherichia coli JM109, was obtained as
described.16 The cell-free extract contained 34 mg mLꢀ1 of protein.
Amidase from Rhodococcus erthropolis MP50 (E.C. 3.5.1.4) was ob-
tained from heterologous expression in E. coli as described.21
Phenylacetonitrile and phenylacetic acid were obtained from
Merck-Schuchardt, ( )-mandelic amide 97% from Alfa Aesar;
(R,S)-mandelic acid and trans-cinnamic aldehyde +99% were from
Acros Organics; benzoic acid, 1,2-dimethoxybenzene +99% and
di-isopropylether (DIPE) were purchased from Fluka; these com-
pounds were used as received.
140
120
100
80
60
40
20
0
0
10
20
30
Time (d)
Figure 7. Storage stability of MeHnL (}), PfNLase (h), RheAMase (4) in the triple-
CLEA at 0 °C; (a) citrate buffer pH 5.5; (b) saturated (NH4)2SO4.
is satisfactory over one month (Fig. 7b). Remarkably, CLEA-
stabilised PfNLase was stable in saturated ammonium sulfate,
whereas non-crosslinked aggregates rapidly lost activity (see
above).
Dextran polyaldehyde solution (3.2%, 2.5 mg mLꢀ1
)
was
obtained by oxidation of Dextran 100–200 kDa (Serva Feinbioche-
mica) as described.10c
Hydrocyanation of 1 in the presence of the triple CLEA resulted
in a near-quantitative conversion into (S)-mandelic acid. Some
transient accumulation of 4 was observed (see Fig. 8 for compari-
son with the two-enzyme-CLEA), which required a somewhat pro-
longed reaction time but >99% enantiomeric purity of the product
was maintained nevertheless.
A stock solution of HCN in DIPE was prepared as previously
described.8c WARNING: sodium cyanide and HCN are highly poi-
sonous. They should be handled in a fume cupboard with a good
draught. It is strongly advised to keep an HCN alarm switched on.
(S)-Mandelonitrile ((S)-2): MeHNL CLEA (CLEA Technologies
B.V., 100 mg) and 20 mM citrate buffer pH 5.5 (1 mL) were added
to DIPE (100 mL) in a round-bottomed flask. The mixture was vig-
orously stirred at 0 °C until a fine suspension was obtained. HCN
(80 mL, 1.98 M solution in DIPE) was added, followed by a solution
of benzaldehyde (10 g, 0.09 mol) in DIPE (100 mL). The mixture
was stirred at room temperature.
3. Conclusion
Enantiomerically pure (S)-mandelic acid was produced from
benzaldehyde at 0.25 M starting concentration and HCN, using a
(a)
(b)100
100
100
80
60
40
20
0
100
80
60
40
20
0
80
60
40
20
0
80
60
40
20
0
0
5
10
15
0
0.5
1
1.5
2
2.5
Time (h)
Time (h)
Figure 8. (a) Bienzymatic synthesis of (S)-mandelic acid from benzaldehyde (10 mM) and HCN (50 mM) in the presence of a MeHnL and PfNLase combi-CLEA biocatalyst; (b)
with a triple CLEA of MeHnL, PfNLase and RheAMase. Reaction in 90:10 DIPE-buffer pH 5.5 at rt. Conversion: benzaldehyde (1 , }), (S)-mandelonitrile (2 , h), (S)-mandelic acid
(3, s), (S)-mandelic amide (4 , 4); ee: 3 (d).

1230
A. Chmura et al. / Tetrahedron: Asymmetry 24 (2013) 1225–1232
The progress of the reaction and the ee of the product were
measured by HPLC using a Waters 510 pump and a Waters 468
variable wavelength detector at 215 nm, equipped with a Daicel
PfNLase: To a 1.5 mL Eppendorf tube containing 20 mM citrate
buffer pH 5.5 (0.98 mL), NLase solution (10
0.003 mg protein), 1,2-dimethoxybenzene (IS) and 2-phenylaceto-
l
L, 100 ꢁ prediluted;
5
l
4.6 ꢁ 250 mm Chiralcel OB-H column, eluent heptane-
nitrile (10 mM) were added. Samples were taken and analysed by
isopropyl alcohol (95:5, v/v) with TFA (1%, v/v) at 1 mL/min at
40 °C. Retention times: benzaldehyde 7.32 min, (R)-(2) 15.72 min,
(S)-2 16.77 min. Complete conversion was obtained after 8 h. The
reaction mixture was filtered to remove the biocatalyst, dried over
Na2SO4 and concentrated in vacuo at 50 °C until the last traces of
HCN had been removed. (S)-2 (11.81 g, 94%, 98.0% ee) was obtained
as a colourless oil.
HPLC (procedure B). One unit (U) of PfNLase will hydrolyse 1
of 2-phenylacetonitrile per min.
lmol
Amidase: To a 1.5 ml Eppendorf tube containing 20 mM citrate
buffer (0.98 mL), amidase solution (0.195 mg protein) and 1,2-
dimethoxybenzene (IS) was added (R,S)-mandelic amide ((R,S)-4,
10 mM). Samples were taken and analysed by HPLC (procedure B).
2-Hydroxy-4-phenyl-trans-3-butenenitrile (modified proce-
dure22): to freshly distilled trans-cinnamaldehyde (0.02 mol) and
0.1 M phosphate buffer pH 7.5 in a magnetically stirred reactor a
solution of HCN in DIPE (300 mL, 2 M) was added dropwise. After
standing overnight equilibrium had been reached (>90% conver-
sion); the cyanohydrin was extracted into CH2Cl2–hexane (50:50)
and dried over MgSO4. The solvent was evaporated in vacuo until
crystallisation occurred spontaneously. The crystals were collected
and dried. The structure was confirmed by 1H NMR.
4.3. Combi-CLEA preparation and activity assay
4.3.1. Combi-CLEA preparation and assay
MeHnL (8.8 mg) and semi-purified PfNLase (3 mg) were added
to a mixture of dextran polyaldehyde solution (0.5 mL), 0.5 M
phosphate buffer pH 7.5 (0.25 mL) and 1,2-dimethoxyethane
(1 mL, aggregation agent). The resulting mixture was stirred over-
night at 4 °C and centrifuged. The pellet was resuspended in 0.1 M
sodium bicarbonate solution (40 mL) containing sodium borohy-
dride (1 g Lꢀ1) to reduce Schiff’s bases and stirred for 45 min at
40 °C. The CLEA was washed three times with Milli-Q water
(6 mL), resuspended in 20 mM citrate buffer pH 5.5 and stored at
4 °C.
4.2. Analysis
4.2.1. Compound characterisation
The kinetic assays were carried as described above with regard
to substrates and reaction conditions. The results are compiled in
Table 2.
Intermediates and products were characterised by comparison
with an authentic sample. Full spectral data are available in the lit-
erature: mandelonitrile (2),23 mandelic acid (3) 1H and 13C NMR,24
MS,25 mandelic amide (4),26 2-hydroxy-4-phenyl-trans-3-
butenenitrile.27
Table 2
The activity recovered in the combi-CLEA
mol minꢀ1 (mg protein)ꢀ1
Combi-CLEA
)
4.2.2. HPLC (general procedures)
Reaction rate (
l
The progress of all of the reactions was monitored by HPLC,
using either a Waters 590 pump and a Waters 486 Tunable Absor-
bance Detector at 215 nm or a Waters Alliance 2695 Separation
Module and a Waters 2487 Dual Wavelength Absorbance Detector
at 215 nm. The reaction products were identified by comparison
with an authentic sample.
Free enzyme
MeHnL
PfNLase
0.41
13.5
0.23
12.2
Procedure A: 4.6 ꢁ 50 mm Merck Chromolith™SpeedROD RP-
18e column, eluent acetonitrile–H2O (5:95, v/v) with heptafluo-
robutyric acid (0.03%, v/v) and ammonium formate buffer pH 2.3
4.4. Cascade synthesis of (S)-mandelic acid ((S)-2) in the
presence of the combi-CLEA
(30 mM) at 1 mL minꢀ1
.
4.4.1. Combi-CLEA preparation
Procedure B: 4.6 ꢁ 50 mm Merck Chromolith™SpeedROD RP-
MeHnL (70.4 mg) and semi-purified PfNLase (24 mg) were
added to a mixture of dextran polyaldehyde solution (4 mL),
0.5 M phosphate buffer pH 7.5 (2 mL) and 1,2-dimethoxyethane
(8 mL, precipitant). The resulting mixture was stirred overnight
at 4 °C and centrifuged. The pellet was resuspended in 0.1 M
sodium bicarbonate solution (40 mL) containing sodium borohy-
dride (40 mg) to reduce Schiff’s bases and stirred for 45 min at
40 °C. The CLEA was washed three times with Milli-Q water
(40 mL), resuspended in 20 mM citrate buffer pH 5.5 (5 mL) and
stored at 4 °C.
18e column, eluent acetonitrile–H2O (10:90, v/v) with trifluoroace-
tic acid (0.1%, v/v) at 1 mL minꢀ1
.
The enantiomeric purity of mandelonitrile (2) and its hydrolysis
products was measured by HPLC using a Waters 515 pump and a
Waters 486 Tunable Absorbance Detector at 215 nm and
a
4.6 ꢁ 250 mm Chiralpak AD-H column; eluent hexane–isopropyl
alcohol (80:20, v/v) with TFA (0.1%, v/v) at 0.6 mL minꢀ1
.
4.2.3. Protein contents
The total protein concentration of cell-free extracts was assayed
according to the standard Bradford procedure.28 The samples were
measured on a Shimadzu UV-240IPC spectrophotometer.
4.4.2. Cascade reaction
The reactions were carried out in a 10 mL thermostatted glass
reactor equipped with a magnetic stirrer at 25 °C. Freshly prepared
combi-CLEA in 20 mM citrate buffer pH 5.5 (1.0 mL) was diluted
with the same buffer (1 mL). Then, appropriate amounts of DIPE
and benzoic acid (IS) were added. Benzaldehyde (1) was added to
a concentration of 10, 25, 42, 83 or 250 mM, and stock solution
of HCN to a concentration of 50, 125, 210, 415 or 750 mM; concen-
trations of 1 and HCN are with respect to the total volume. The
total reaction volumes were 6 mL and consisted of 70% organic
phase. Samples were withdrawn periodically to monitor the pro-
4.2.4. Activity assays (general procedures)
MeHnL: To a 1.5 mL Eppendorf tube containing 20 mM citrate
buffer pH 5.5 (0.5 mL), enzyme solution (0.3 mg), DIPE (0.4 mL),
1,2-dimethoxybenzene (internal standard, IS) and HCN (100 mM)
were added. Finally, trans-cinnamic aldehyde (10 mM) was added.
The reaction was shaken (QInstruments ThermoTWISTER) at 25 °C
while kept closed to prevent the escape of HCN. Samples were ta-
ken from each phase and analysed by HPLC (procedure B). One unit
(U) of MeHnL will produce 1
butenenitrile per min.
lmol of 2-hydroxy-4-phenyl-trans-3-
gress of the reactions. For quantitative measurements 20
lL of
the lower (aqueous) phase and 100 L of organic phase were
l

A. Chmura et al. / Tetrahedron: Asymmetry 24 (2013) 1225–1232
1231
mixed with acidified mobile phase. DIPE was evaporated under
vacuum, and the residue was analysed by reverse-phase HPLC as
described above. The samples for enantiomeric purity measure-
ments were extracted from the aqueous phase into ethyl acetate
and then analysed as described above.
sulfate solution at 0 °C. The activity recovery (see procedure below)
was 35.8%.
Activity assay: to 0.5 M phosphate buffer pH 8 at 25 °C (1 mL)
were added 2-phenylacetonitrile (10
and 20
L of the 100ꢁ diluted CLEA suspension. The reaction was
monitored by reversed phase HPLC as described above.
lmol), veratrole (IS, 10 lmol)
l
Hydrolysis of (S)-mandelonitrile: a suspension of PfNLase CLEA
(74 mg in 1 mL saturated ammonium sulfate solution) was centri-
fuged and the solid was resuspended in 20 mM citrate buffer pH
5.5 (0.5 mL).
4.5. Operational stability of the combi-CLEA: catalyst recycling
The reactions were carried out in 1.5 mL Eppendorf tubes on a
QInstruments ThermoTWISTER comfort shaker at 25 °C. To
(S)-2 (6
were dissolved in 3.5 mL of DIPE, then 1.3 mL of citrate buffer was
added. 70 L of organic phase and 30 L of aqueous phase were
withdrawn with a syringe to determine the initial conditions.
The reaction was initiated by adding 200 l of the CLEA suspension
and the mixture then stirred at 25 °C. Samples were withdrawn
periodically to check the progress of the reaction: 30 L of the low-
er (aqueous) phase and 70 L of organic phase were mixed with
lL, 0,05 mmol) and benzoic acid (IS, 5 mg, 0.043 mmol)
20 mM citrate buffer pH 5.5 (450
was added (50 L, 0.24 U MeHnL and 10.1 U PfNLase). DIPE
(465 L), veratrole (IS), HCN (50 mM) and 1 (10 mM) were added.
lL), combi-CLEA suspension
l
l
l
l
The total reaction volume was 1 mL and consisted of 50% organic
phase. After 30 min, the reaction was spun down at 10 ꢁ 103 minꢀ1
for 3 min; the pellet was washed twice with citrate buffer and used
in the next reaction cycle.
l
l
l
In each reaction run a sample was withdrawn after 5 min to
acidified mobile phase. DIPE was evaporated under vacuum, the
sample spun down, and analysed by reversed phase HPLC as de-
scribed above.
The hydrolysis of (S)-2 in the presence of HCN was carried out in
the same way, except that an appropriate amount of a stock solu-
tion of HCN in DIPE (1.98 M) was added, while reducing the vol-
ume of additional DIPE to maintain a constant 30:70 aqueous–
organic ratio.
determine the enzyme activities;
30 min to check the progress of the reaction. For each sample
10 L aliquots of each phase were mixed with acidified mobile
phase (980 L) and analysed by reverse-phase HPLC as described
a sample was taken after
l
l
above, except that ACN–H2O: (10:90, v/v) with TFA (0.1%, v/v)
was used as the mobile phase.
4.6. Effects of reagents at high concentrations
4.7. Triple CLEA preparation and activity assay
4.6.1. Inhibition experiments
Benzaldehyde: The reactions were carried out in 1.5 mL Eppen-
dorf tubes on a QInstruments ThermoTWISTER comfort shaker at
4.7.1. Aggregation study
Samples of MeHNL (0.88 mg protein, 20
l
L), PfNLase (0.23 mg
25 °C. To 20 mM citrate buffer pH 5.5 (280
lL), combi-CLEA sus-
protein, 7.5 L and MP50 amidase (0.44 mg protein, 7.5
l
lL) were
pension (20 L, 2.8 U PfNLase) was added. Subsequently, DIPE
l
separately aggregated in 90% (v/v) aqueous aggregation reagent
in 0.5 mL Eppendorf tubes. After an appropriate interval the aggre-
gate was spun down (11 ꢁ 103 minꢀ1), the pellet was redissolved
in 20 mM citrate buffer pH 5.5 or 0.1 M phosphate buffer pH 7 as
appropriate and assayed for activity as described above. The effects
of the aggregation reagent on the activity recovery with RheAMase
are shown in Figure 9.
and 1 (10–500 mM) were added; the organic phase was 70% of
the total volume. After 1 h, veratrole (IS) and 2-phenylacetonitrile
were added to start the NLase activity assay. Samples were with-
drawn periodically from both phases and analysed on reverse-
phase HPLC as described above.
Effects of HCN concentration: The reactions were carried out in
1.5 mL Eppendorf tubes on a QInstruments ThermoTWISTER com-
fort shaker at 25 °C. To 20 mM citrate buffer pH 5.5 (250
lL), com-
0.150
0.100
0.050
0.000
bi-CLEA suspension (50 L, 7.0 U NLase) was added. Subsequently
l
DIPE and HCN (25–575 mM) were added. The organic phase was
70% of the total volume. After 1 h veratrole (IS) and 2-phenylaceto-
nitrile were added to start the NLase activity assay. Samples were
withdrawn periodically from both phases and analysed on reverse-
phase HPLC as described above.
Effects of 125 mM HCN over time: Combi-CLEA was suspended in
20 mM citrate buffer pH 5.5, HCN (125 mM) and DIPE (70% v/v)
were added; the organic phase was 70% (v/v) of the total volume.
Samples from the aqueous phase (containing the CLEA) were with-
drawn at appropriate intervals. These were diluted with 20 mM
0.00
2.00
4.00
18.00 20.00
citrate buffer pH 5.5 (920 lL), veratrole (IS) and 2-phenylacetoni-
trile (10 mM) were added and the reaction was monitored over
Time (h)
Figure 9. Activity recovery of RheAMase upon precipitation, incubation of the
aggregates in the precipitant and redissolution in buffer. Ammonium sulfate (h),
acetonitrile (s), 1,2-dimethoxyethane (4), tert-butyl alcohol (}).
time, using reversed-phase HPLC as described above.
4.6.2. Effects of HCN on PfNLase selectivity
PfNLase CLEA: A solution of semi-purified PfNLase (15
0.63 mg protein) was added to a cold, saturated ammonium sulfate
solution (135 L, pH 8.7). The aggregates were cross-linked for 2 h
with dextran polyaldehyde solution (3.2%, 15 L) and subsequently
lL,
4.7.2. Single enzyme CLEAs
l
MeHnL: Enzyme samples (1.34 mg protein, 15
gated in cold, saturated ammonium sulfate solution (pH 8.7,
135 L) in 1.5 mL Eppendorf tubes. The aggregates were cross-
lL) were aggre-
l
reduced with NaCNBH3 (1.5 mg, dissolved in a minimum volume of
l
water) for 1 h.
linked by the addition of appropriate amounts of dextran polyalde-
hyde solution and shaken (QInstruments ThermoTWISTER). After
the appropriate interval sodium cyanoborohydride (5.2 mg dis-
solved in a minimum amount of water) was added and shaking
The CLEA was isolated by centrifugation at 13 ꢁ 103 minꢀ1
(74 mg), washed three times with cold 10 mM phosphate buffer
pH 8.7 and stored as a suspension in 1 ml of saturated ammonium

1232
A. Chmura et al. / Tetrahedron: Asymmetry 24 (2013) 1225–1232
was continued for 1 h. The CLEA was spun down and three times
resuspended and centrifuged. The CLEA as well as the cross-linking
supernatant were assayed for activity as described above, except
that the assay substrate was benzaldehyde.
PfNLase: CLEAs were prepared from 0.45 mg (15 lL) enzyme
samples as described above (1.5 mg sodium cyanoborohydride).
References
1. (a) Mayer, S. F.; Kroutil, W.; Faber, K. Chem. Soc. Rev. 2001, 30, 332–339; (b)
Bruggink, A.; Schoevaart, R.; Kieboom, T. Org. Process Res. Dev. 2003, 7, 622–
640.
2. Breuer, M.; Ditrich, K.; Habicher, T.; Hauer, B.; Keßeler, M.; Stürmer, R.;
Zelinski, T. Angew. Chem., Int. Ed. 2004, 43, 788–824 [799-800, 801].
3. (a) Effenberger, F.; Ziegler, T.; Forster, S. Angew. Chem., Int. Ed. Engl. 1987, 26,
458–460; (b) Effenberger, F. Angew. Chem., Int. Ed. Engl. 1994, 33, 1555–1564.
4. (a) Yamamoto, K.; Oishi, K.; Fujimatsu, I.; Komatsu, K.-I. Appl. Environ. Microbiol.
1991, 57, 3028–3032; (b) Endo, T.; Tamura, K. (Nitto Chem. Ind.) EP 449 648,
1991; Chem. Abstr. 1992, 116, 5338.; (c) Yamaguchi, Y.; Ushigome, M.; Kato, T.
(Nitto Chem. Ind.) EP 773 297, 1997; Chem. Abstr. 1997, 127, 4190.; (d) Ress-
Löschke, M.; Friedrich, F.; Hauer, B.; Mattes, R.; Engels, D. (BASF AG), DE 1 9848
129, 2000; Chem. Abstr. 2000, 132, 292813.
The activity assay was as described above for soluble PfNLase.
MP50 amidase: CLEAs were prepared from 0.88 mg (15 lL) en-
zyme samples as described above (1.5 mg sodium cyanoborohy-
dride). The activity assay was carried out as described for the
soluble amidase.
5. Robertson, D. E.; Chaplin, J. A.; DeSantis, G.; Podar, M.; Madden, M.; Chi, E.;
Richardson, T.; Milan, A.; Miller, M.; Weiner, D. P.; Wong, K.; McQuaid, J.;
Farwell, B.; Preston, L. A.; Tan, X.; Snead, M. A.; Keller, M.; Mathur, E.; Kretz, P.
L.; Burk, M. J.; Short, J. M. Appl. Environ. Microbiol. 2004, 70, 2429–2436.
6. Niedermeyer, U.; Kula, M.-R. Angew. Chem., Int. Ed. Engl. 1990, 29, 386–387.
7. For a review see: Schmidt, M.; Griengl, H. Top. Curr. Chem. 1999, 200, 193–226.
8. (a) Han, S.; Lin, G.; Li, Z. Tetrahedron: Asymmetry 1998, 9, 1835–1838; (b) Van
Langen, L. M.; Van Rantwijk, F.; Sheldon, R. A. Org. Process Res. Dev. 2003, 7,
828–831; (c) Chmura, A.; Van der Kraan, G. M.; Kielar, F.; Van Langen, L. M.;
Van Rantwijk, F.; Sheldon, R. A. Adv. Synth. Catal. 2006, 348, 1655–1661.
9. See for example: Layh, N.; Willetts, A. Biotechnol. Lett. 1998, 20, 329–331.
10. (a) Cao, L.; Van Rantwijk, F.; Sheldon, R. A. Org. Lett. 2000, 2, 1361–1364; (b)
Van Langen, L. M.; Selassa, R. P.; Van Rantwijk, F.; Sheldon, R. A. Org. Lett. 2005,
7, 327–329; (c) Mateo, C.; Palomo, J. M.; Van Langen, L. M.; Van Rantwijk, F.;
Sheldon, R. A. Biotechnol. Bioeng. 2004, 86, 273–276.
4.7.3. Semi-optimised triple CLEA
MeHnL (22 mg protein, 0.25 mL), PfNLase (17 mg protein,
0.5 mL) and MP50 amidase (49 mg protein, 2.5 mL) were aggre-
gated in cold, saturated ammonium sulfate solution pH 8.7
(24.5 mL). The aggregate was cross-linked by the addition of 3.2%
dextran polyaldehyde solution (2.2 mL) followed by magnetic stir-
ring for 2 h at rt. Sodium cyanoborohydride (412 mg dissolved in a
minimum amount of water) was added; the suspension was sha-
ken for another 60 min and centrifuged at 4 ꢁ 103 minꢀ1. The CLEA
was washed in three resuspension–centrifugation cycles, using in
25 mL 20 mM citrate buffer pH 5.5 (25 mL) at 4 °C. The triple-CLEA
was stored on ice as a suspension in citrate buffer.
Storage stability: Samples of the triple CLEA as prepared above
were washed three times with 20 mM citrate buffer pH 5.5 or sat-
urated ammonium sulfate solution (pH 8.7) and suspended in
these media at 0 °C. The residual enzyme activities were measured
as described above.
11. Mateo, C.; Chmura, A.; Rustler, S.; Van Rantwijk, F.; Stolz, A.; Sheldon, R. A.
Tetrahedron: Asymmetry 2006, 17, 320–323.
12. (a) Sosedov, O.; Matzer, K.; Bürger, S.; Kiziak, C.; Baum, S.; Altenbuchner, J.;
Chmura, A.; Van Rantwijk, F.; Stolz, A. Adv. Synth. Catal. 2009, 351, 1531–1538;
(b) Baum, S.; Van Rantwijk, F.; Stolz, A. Adv. Synth. Catal. 2012, 345, 113–122.
13. (a) Yamamoto, K.; Fujimatsu, I.; Komatsu, K.-I. J. Ferment. Bioeng. 1992, 73, 425–
430; (b) Rey, P.; Rossi, J.-C.; Taillades, J.; Gros, G.; Nore, O. J. Agric. Food Chem.
2004, 52, 8155–8162.
14. Fernandes, B. C. M.; Mateo, C.; Kiziak, C.; Chmura, A.; Wacker, J.; Van Rantwijk,
F.; Stolz, A.; Sheldon, R. A. Adv. Synth. Catal. 2006, 348, 2597–2603.
15. Layh, N.; Parratt, J.; Willetts, A. J. Mol. Catal. B Enzym. 1998, 5, 467–474.
16. Kiziak, C.; Conradt, D.; Stolz, A.; Mattes, R.; Klein, J. Microbiology 2005, 151,
3639–3648.
17. Van Rantwijk, F.; Mateo, C.; Chmura, A.; Fernandes, B. C. M.; Sheldon, R. A. In
Modern Biocatalysis: Stereoselective and Environmentally Friendly Reactions;
Fessner, W.-D., Anthonsen, T., Eds.; Nitrilases in the Enantioselective Synthesis
4.7.4. Cascade synthesis of (S)-mandelic acid ((S)-2) in the
presence of the triple CLEA
Triple CLEA as prepared above (30 lL) was added to 20 mM cit-
rate buffer pH 5.5 (0.47 mL). To the CLEA suspension were added
successively DIPE (0.39 mL), HCN (0.2 M), veratrole (10 mM) and
1 (10 mM); concentrations are with respect to the total volume.
of a-Hydroxycarboxylic Acids; Wiley-VCH: Weinheim, 2008; pp 261–272.
18. The hydrocyanation product, (S)-2-hydroxy-4-phenyl-trans-3-butenenitrile
hardly reacts in the presence of PfNLase (Mateo, C.; Chmura, A. unpublished
results).
19. Ammonium mandelate is self-buffering at approx pH 6.4 and could exhaust the
buffer at high concentrations.
20. Hirrlinger, B.; Stolz, A.; Knackmuss, H. J. J. Bacteriol. 1996, 178, 3501–3507.
21. Trott, S.; Burger, S.; Calaminus, C.; Stolz, A. Appl. Environ. Microbiol. 2002, 68,
3279–3286.
22. Gerrits, P. J.; Willeman, W. F.; Straathof, A. J. J.; Heijnen, J. J.; Brussee, J.; Van der
Gen, A. J. Mol. Catal. B Enzym. 2001, 15, 111–121.
Samples (10 lL) were withdrawn periodically and mixed with
acidified mobile phase. DIPE was evaporated under vacuum and
the residue was analysed by reverse-phase HPLC as described
above. The samples for enantiomeric purity measurements were
extracted from the aqueous phase into ethyl acetate and then ana-
lysed as described above.
23. Lee, P.-T.; Chen, C. Tetrahedron: Asymmetry 2005, 16, 2704–2710.
24. Di Gennaro, P.; Bernasconi, S.; Orsini, F.; Corretto, E.; Sello, G. Tetrahedron:
Asymmetry 2010, 21, 1885–1889.
25. He, Y. C.; Xu, J. H.; Xu, Y.; Ouyang, L. M.; Pan, J. Chin. Chem. Lett. 2007, 18, 677–
680.
26. Kambe, H.; Inoue, T.; Takeda, T.; Fujiwara, S.-I.; Sonoda, N. J. Am. Chem. Soc.
2006, 128, 12650–12651.
27. Casas, J.; Nájera, C.; Sansanoa, J. M.; Saa, J. M. Tetrahedron 2004, 60, 10487–
10496.
Acknowledgements
Support by the COST Actions D25 and CASCAT is gratefully
acknowledged. This research was financially supported by the
Netherlands Research Council NWO and the Deutsche Forschungs-
gemeinschaft within the framework of the CERC-3 programme. The
authors thank CLEA Technologies B.V. (Delft, The Netherlands) for a
donation of Manihot esculenta hydroxynitrile lyase CLEA.
28. Bradford, M. M. Anal. Biochem. 1976, 72, 248–254.