Immobilized hydroxynitrile lyases for enantioselective synthesis of cyanohydrins: Sol-gels and cross-linked enzyme aggregates

Article abstract of DOI:10.1002/adsc.200606139

The hydroxynitrile lyases (HNLs) from Prunus amygdalus (paHNL), Manihot esculenta (MeHNL), and Hevea brasiliensis (HbHNL) were successfully immobilized in sol-gels. The cross-linked enzyme aggregate (CLEA) of HbHNL was also prepared. These immobilized enzymes and the commercial PaHNL- and MeHNL-CLEAs were employed for the enantioselective synthesis of cyanohydrins. The sol-gels were highly efficient at low catalyst loading and particularly stable towards the organic solvent (diisopropyl ether) and substrate/product deactivation. The stabilization effect was inconsistent for CLEAs of different HNLs and significant deactivation of PaHNL- and HbHNL-CLEAs in diisopropyl ether was observed. In contrast commercial MeHNL-CLEA proved to be a remarkably robust and efficient biocatalyst in diisopropyl ether.

Full text of DOI:10.1002/adsc.200606139

FULL PAPERS
DOI: 10.1002/adsc.200606139
Immobilized Hydroxynitrile Lyases for Enantioselective Synthesis
of Cyanohydrins: Sol-Gels and Cross-Linked Enzyme Aggregates
Fabien L. Cabirol,a, b Ulf Hanefeld,a and Roger A. Sheldona,*
a
Laboratory of Biocatalysis and Organic Chemistry, Delft University of Technology, Julianalaan 136, 2628 BL Delft,
The Netherlands
Fax : (+31)-15-278-1415; e-mail: r.a.sheldon@tudelft.nl
Institute of Chemical and Engineering Sciences, 1 Pesek Road, Jurong Island, S(627833), Singapore
b
A
Fax : (+65)-6316-6182
Received: March 24, 2006; Accepted: May 29, 2006
Abstract: The hydroxynitrile lyases (HNLs) from solvent (diisopropyl ether) and substrate/product de-
Prunus amygdalus (PaHNL), Manihot esculenta activation. The stabilization effect was inconsistent
(MeHNL), and Hevea brasiliensis (HbHNL) were for CLEAs of different HNLs and significant deacti-
successfully immobilized in sol-gels. The cross-linked vation of PaHNL- and HbHNL-CLEAs in diisoprop-
enzyme aggregate (CLEA) of HbHNL was also pre- yl ether was observed. In contrast commercial
pared. These immobilized enzymes and the commer- MeHNL-CLEA proved to be a remarkably robust
cial PaHNL- and MeHNL-CLEAs were employed and efficient biocatalyst in diisopropyl ether.
for the enantioselective synthesis of cyanohydrins.
The sol-gels were highly efficient at low catalyst Keywords: asymmetric synthesis; CLEA; cyanohy-
loading and particularly stable towards the organic drins; hydroxynitrile lyase; oxynitrilase; sol-gel
Introduction
the pH of the water phase in the reaction
system.[10,11] By immobilizing HNLs it might become
The addition of hydrogen cyanide to carbonyl com- possible to avoid an organic/water biphasic reaction
pounds is an important carbon-carbon bond forming mixture and the racemic background reaction should
reaction. The cyanohydrins generated are versatile in- ideally be completely suppressed. Although several
termediates in organic synthesis and different catalyt- immobilization methods have been employed for
ic approaches toward enantioenriched cyanohydrins HNLs, each having their particular advantages, the re-
from prochiral aldehydes and ketones have been de- action conditions under which they have been tested
scribed.[1–5] Hydroxynitrile lyases (HNLs) can be iso- were not always identical. Here, we compare different
lated from many different plants and they eliminate forms of immobilized HNLs in diisopropyl ether
HCN from cyanohydrins leaving the corresponding (DIPE), a solvent used industrially for an HbHNL-
carbonyl compound. This natural defense mechanism catalyzed reaction.[5]
occurs when the plant is attacked by herbivores. In re-
In order to improve the HNLs robustness, immobi-
verse, this reaction serves for the enantioselective syn- lization onto supports such as celite,[10] cellu-
thesis of cyanohydrins. Indeed many successful studies lose[10,12–14] or nitrocellulose[13,14] has been studied.
have been described and HNLs are today used on an Our attention focuses on immobilization strategies
industrial scale for the synthesis of enantiopure cya- where the HNL can be recycled efficiently. Cross-
nohydrins.[1,5–9]
linked enzyme crystals (CLECs) of the HNL from
The stability of enzymes in organic solvents can be Manihot esculenta (MeHNL) were reported earlier as
greatly enhanced by immobilization. Moreover the stable and recyclable biocatalysts for the addition of
catalyst can be filtered off easily and no extraction hydrogen cyanide to carbonyl compounds.[15] Howev-
step is required to recover the product of the reaction er, challenges inherent to the preparation of crystals
from the liquid phase. It is also particularly beneficial from proteins have limited the range of applica-
in the HNL-catalyzed enantioselective cyanohydrin tions.[16] Attempts to overcome these limitations on a
synthesis since the racemic, non-catalyzed, addition of large scale have been reported[17] but a recently es-
HCN to carbonyl compounds can reduce the enantio- tablished technology based on the cross-linking of
purity of the product significantly. The rate of this enzyme aggregates rather than crystals is now consid-
competing process depends on the water content and ered
a more viable alternative for immobiliza-
Adv. Synth. Catal. 2006, 348, 1645 – 1654
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1645

Fabien L. Cabirol et al.
FULL PAPERS
tion.[18,19] Cross-linked enzyme aggregates (CLEAs)
of the HNLs from Prunus amygdalus (PaHNL) and
Manihot esculenta (MeHNL) are even commercially
available biocatalysts for the enantioselective synthe-
sis of (R)- or (S)-cyanohydrins, respectively. Poly-
(vinyl alcohol) hydrogels (Leutikatsꢁ) of PaHNL
have been reported as efficient and robust catalysts
for the synthesis of (R)-mandelonitrile.[20] Further-
more, the PaHNL-CLEA-catalyzed synthesis of cya-
nohydrins derived from substituted aromatic alde-
hydes was recently described in microaqueous (2% v/
vaqueous buffer in organic solvent) and biphasic sys-
tems. Under these conditions the biocatalyst could be
recycled up to 10 times.[21]
Another successful immobilization technique is the
encapsulation of enzymes in sol-gels. Encapsulation in
a sol-gel matrix of the HNL from Hevea brasiliensis
(HbHNL) was reported recently and the system
proved to be efficient for the synthesis of a range of
cyanohydrins.[22] Capsules of the sol-gel matrix were
Figure 1. Substrates investigated.
filled with buffer solution, thereby forming an “aqua capsulated using the same amount of precursor as in
gel” with the enzyme maintained in an aqueous envi- the original procedure, the recovered activity was
ronment. Although some loss of activity was observed found to be significantly lower (Table 1). However, it
upon recycling, the “aqua gel” was still catalytically has earlier been observed that the supposedly low ac-
active in buffer-saturated DIPE.
tivity of an immobilized enzyme in aqueous buffers
Here we compare the catalytic performance of does not effectively represent its catalytic activity in
“aqua gels” and cross-linked enzyme aggregates organic solvents.[21] Indeed, this drop in activity when
(CLEAs) of (S)-selective HbHNL, (S)-selective encapsulating approximately 13 times less HbHNL in
MeHNL and (R)-selective PaHNL for HCN addition the aqua gel is consistent with a diffusion-limited
to a range of carbonyl compounds, including the in- system where the catalyst is immobilized but not ac-
dustrially relevant m-phenoxybenzaldehyde 1d cessible enough for a fast activity test such as the de-
(Figure 1).
composition of dilute rac-mandelonitrile 2a in aque-
ous buffers. Aqua gels of MeHNL and PaHNL were
prepared according to the procedure reported for
HbHNL, and the recovered activities were again low
(Table 1).
Results and Discussion
Immobilized Enzymes
When the CLEA of HbHNL (2.61 kU/g) was pre-
pared according to the procedure developed for
When immobilizing a homogeneous catalyst two tar- PaHNL[21] the activity recovered after immobilization
gets have to be achieved: A high percentage of the was 16%, which is somewhat higher than the recov-
catalyst has to be immobilized in its active form, and ery value reported for PaHNL-CLEA (9.6%).[21] As
the immobilized catalyst, here the enzyme, has to be described for the PaHNL-CLEA and proposed for
stable. Aqua gels of HbHNL were first prepared ac- the aqua gels above, the low activity is most likely
cording to the reported procedure[22] and 58% of the due to diffusion limitations. To compare the different
activity was recovered when starting from a 3.6 kU/ enzymes and the effects of the carriers on them, the
mL stock solution, in line with the literature value. aqua gels of HbHNL, PaHNL and MeHNL, as well
The residual methanol in the gel precursor and diffu- as the CLEA of HbHNL, were used at loadings of 6
sion limitations are believed to be responsible for the U/mmol substrate in the catalytic experiments. Com-
decrease in activity.[22] When diluted HbHNL was en- mercially available CLEAs from PaHNL (4.61 kU/g)
Table 1. Enzyme activity recovered upon encapsulation in aqua gels according to activity test described in the Experimental
Section.
HbHNL
HbHNL
MeHNL
PaHNL
Enzyme Stock Solution [U/mL]
Activity Recovery in Aqua Gel [%]
3,600
58
274
22
230
8.5
300
15
1646
www.asc.wiley-vch.de
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2006, 348, 1645 – 1654

Immobilized Hydroxynitrile Lyases for Enantioselective Synthesis of Cyanohydrins
FULL PAPERS
and MeHNL (1.02 kU/g) were also included into the a much shorter period of time with the same enantio-
study and standardized at 6 U/mmol of substrate. This selectivity (Table 2). Under these standard reaction
amount of catalyst was relatively low when compared conditions, with relatively low catalyst loading (6 U/
to loadings typically used for these enzymes.
mmol), both the free enzymes and the HNL aqua gels
The aqua gel-catalyzed reactions were carried out operated well. However, initial rates were higher for
in diisopropyl ether (DIPE) saturated with 50 mM cit- HNL aqua gels than for the corresponding free
rate/phosphate buffer (pH 5.0 to suppress the unde- enzyme. Clearly the diffusion limitation that indicated
sired racemic background reaction) to prevent a pos- a low enzyme loading in the aqueous activity test
sible drying effect of the solvent on the catalyst capsu- (Table 1) was giving misleading results and much less
les. For the same reason the hydrogen cyanide stock enzyme was deactivated during the immobilization
solution in DIPE was also saturated with aqueous than indicated by this test. Furthermore, the increase
buffer. This buffer (pH 5.5) was selected in order to in surface area of the aqueous phase for encapsulated
stabilize the hydrogen cyanide solution upon storage HNLs, in comparison with the aqueous environment
and again to avoid the background reaction (see Ex- of homogeneous enzymes, is also believed to be re-
perimental Section). The PaHNL-CLEA-catalyzed sponsible for this effect.
addition of HCN to a range of aldehydes has been de-
When the reaction was allowed to proceed over an
scribed in microaqueous media (2% v/v buffer in an extended period of time, racemization of mandeloni-
organic solvent) but under those conditions[21] a water trile 2a was observed in all cases. This might be due
layer can still be observed. In the study reported here to the chemical background reaction. However, these
the CLEAs were suspended in commercial DIPE in enzymes do not only catalyze the formation of one
order to avoid the extraction step during the reaction enantiomer, they also catalyze its degradation and
work-up. Under these conditions the organic solvent thus speed up racemization significantly. Rates of rac-
still contained trace amounts of buffer from the HCN emization for the free enzyme-catalyzed reactions
stock solution but the reaction medium was a single were comparable to the rates observed in a biphasic
liquid phase. Thus, in both the aqua gel and the system for the non-catalyzed racemization of (S)-
CLEA system a single organic phase was present. In mandelonitrile (Figure 2), demonstrating a deactiva-
the case of the CLEAs the enzyme is surrounded by tion of the enzymes. In the presence of HNL aqua
DIPE providing a true one-phase system, in the case gels, racemization rates were significantly higher than
of the aqua gel the enzyme is in the aqueous buffer for the free enzymes indicating the high stability of
inside the aqua gel and the system is comparable to a the aqua gel immobilized enzymes. The racemization
biphasic medium. For comparison the reaction was of (S)-mandelonitrile in the presence of enzyme-free
also studied with the free enzymes (6 U/mmol sub- aqua gels proceeded at rates comparable to the non-
strate); in this case a biphasic buffer-DIPE system catalyzed reaction in a biphasic system. Thus the race-
had to be used (see Experimental Section).
mization was enzyme induced.
Overall the aqua gels of HNLs investigated here
had very similar catalytic properties for the synthesis
of 2a although these enzymes come from different
plants and are not always structurally related.[23] The
Benzaldehyde
As a first model reaction the enantioselective addition performance of cross-linked enzyme aggregates of
of HCN to benzaldehyde was studied in DIPE. these HNLs, however, varied greatly. The synthetic
Almost enantiopure mandelonitrile (2a) was obtained activity of MeHNL-CLEA was comparable to the re-
at excellent conversions within 4 h when the reaction sults obtained for the free enzyme and MeHNL aqua
was catalyzed by free HNLs. Remarkably the aqua gel while the CLEAs from PaHNL and HbHNL suf-
gels of all three HNLs catalyzed the same reaction in fered from a noticeable loss of activity under the con-
Table 2. Conversion ratios and ees (parentheses) at optimum reaction times in the synthesis of mandelonitrile (2a) catalyzed
by free HNLs and the corresponding aqua gels.
(S)-HbHNL
(S)-MeHNL
(R)-PaHNL
Free[a]
4 h: 97(97)
Aqua Gel[b]
0.5 h: 97(99)
Free[a]
4 h: 97(98)
Aqua Gel[b]
0.5 h: 96(99)
Free[a]
4 h: 98(97)
Aqua Gel[b]
2 h: 97(97)
Reaction conditions: Benzaldehyde (0.5 mmol/mL DIPE), HCN (3 equivs.), and the catalyst (6 U/mmol) were shaken at
room temperature and the reaction was monitored by GC.
[a]
The catalyst stock solution was diluted with citrate/phosphate buffer (50 mM, pH 5.0) to a DIPE:aqueous media ratio of
5:1.
[b]
DIPE saturated with citrate/phosphate buffer (50 mM, pH 5.0).
Adv. Synth. Catal. 2006, 348, 1645 – 1654
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.asc.wiley-vch.de
1647

Fabien L. Cabirol et al.
FULL PAPERS
Figure 2. The ees in the HNL-catalyzed synthesis of mandelonitrile. Results in I: A: free PaHNL, B: free HbHNL, C:
PaHNL aqua gel and D: HbHNL aqua gel. II: Results of blank racemisations (no enzyme) in DIPE (a) and in a biphasic
system (b); and results from MeHNL. E: free MeHNL, F: MeHNL aqua gel and G: MeHNL-CLEA.
Table 3. Conversion ratios and ees (parentheses) in the syn-
thesis of mandelonitrile catalyzed by CLEAs of different
HNLs.
(S)-HbHNL
(S)-MeHNL
(R)-PaHNL
CLEA
CLEA
CLEA
72 h: 55(67)
2 h: 96(97)
72 h: 97(99)
Reaction conditions: Benzaldehyde (0.5 mmol/mL DIPE
containing traces of water from the HCN solution), HCN (3
equivs.), and the respective CLEA (6 U/mmol) were shaken
at room temperature and the reaction was monitored by
GC.
ditions used and both enzymes catalyzed the reaction
very slowly (Table 3). HbHNL and PaHNL are
known to be unstable in the absence of water[24,25]
and the decrease in activity might be due to the
“drying” effect of the reaction medium (DIPE with
traces of buffer from the HCN solution) on the cata-
lyst. The rate of racemization for MeHNL-CLEA was
even higher than observed for the aqua gel of the
Figure 3. Conversion ratios in the HNL-CLEA catalyzed
synthesis of mandelonitrile. A: MeHNL-CLEA, B: PaHNL-
CLEA, C: HbHNL-CLEA (2% aqueous suspension: “mi-
croaqueous”), D: HbHNL-CLEA .
The loss of activity observed for CLEAs of PaHNL
same enzyme (Figure 2), indicating that the MeHNL- and HbHNL in DIPE with traces of buffer from the
CLEA is particularly robust in this single-phase or- HCN solution was consistent for all the substrates in-
ganic reaction medium.
vestigated and these catalysts will not be included fur-
Full conversion was achieved with PaHNL-CLEA ther in this study. Cross-linked enzyme aggregates are
over an extended period of time (Figure 3). (R)-Man- carrier-free biocatalysts where the protein is directly
delonitrile, being the natural substrate of PaHNL, exposed to the reaction media. Relative robustness of
was obtained in high selectivity over the course of the HNLs from different sources is therefore revealed to
reaction. The CLEA of HbHNL was most dramatical- a greater extent in a CLEA than upon encapsulation
ly affected by the reaction conditions. The reaction in an aqua gel where the enzyme is maintained in an
rate and catalyst selectivity were significantly lower aqueous buffer environment. The catalytic perfor-
than for all other catalysts. When it was employed mance of MeHNL-CLEA is reported in the accompa-
under microaqueous conditions (Figure 3) its perfor- nying paper.[35] The remarkable performance of this
mance improved significantly, but HbHNL-CLEA re- immobilized form made it a good candidate for a
mained very sensitive toward deactivation by the or- comparison with the aqua gels of HNLs and the free
ganic solvent.
enzymes.
1648
www.asc.wiley-vch.de
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2006, 348, 1645 – 1654

Immobilized Hydroxynitrile Lyases for Enantioselective Synthesis of Cyanohydrins
FULL PAPERS
2-Furaldehyde
the solubility of 2-furaldehyde is significant[28] (ca. 8.2
wt%). The results obtained suggest that the substrate/
The synthesis of 2-furaldehyde cyanohydrin (2b) was product deactivation phenomenon is significant for
achieved within 30 min at high conversion ratios and PaHNL and MeHNL whereas free HbHNL remained
ees when catalyzed by HNL-aqua gels, free enzymes active over the course of the reaction (3 days). Fur-
and the CLEA of MeHNL (Table 4).
thermore, conversion ratios reported by Griengl et al.
Enantiomeric excesses at full conversion were for this substrate in biphasic[29] and aqueous[30] media
slightly lower for 2b than for 2a. Rapid racemization by HbHNL were significantly different (95% and
of 2b accounted for the lower ee values obtained. Sta- 55%, respectively). This trend seems to indicate that
bilization of free PaHNL and MeHNL upon encapsu- greater concentrations of 2b in the aqueous phase de-
lation in an aqua gel matrix was observed (Figure 4). activate HbHNL. Aqua gel encapsulation of PaHNL
This stabilization is much less pronounced for and MeHNL improved the stability of these enzymes.
HbHNL. While the free enzyme catalyzed the racemi- The stabilization effect upon immobilization observed
zation more efficiently than the other two free HNLs for MeHNL was also noticed for the CLEA of this
its aqua gel displays the lowest racemization rate.
enzyme. The CLEA immobilization strategy seems ef-
HNLs are susceptible to deactivation not only by ficacious in preventing substrate/product deactivation.
solvents, but also by the substrate[11,26] and by the
product formed. This effect could not be observed in
the case of benzaldehyde (1a) since the enzyme load-
ings were standardized according to their catalytic Hexanal
properties in aqueous media for the decomposition of
mandelonitrile (2a). The extent of substrate/product As a representative alkyl substrate the addition of hy-
deactivation was therefore leveled out and accounted drogen cyanide to hexanal was studied. The free
for in the reaction studied earlier. Using 2-furalde- enzyme, aqua gel and CLEA form of MeHNL ach-
hyde (1b), the relative stability of individual enzyme ieved full conversion about as fast as for benzalde-
toward a specific substrate became apparent. More- hyde (Table 5). Enantiomeric excesses obtained indi-
over the greater solubility of this aldehyde in aqueous cate a trend in selectivity for HNLs from different
media is believed to enhance this effect. Indeed, sources for this substrate. Selectivity in the HbHNL-
whilst mandelonitrile is sparingly soluble in water,[27] catalyzed reactions were as high as 94% whereas the
Table 4. Conversion ratios and ees (parentheses) after 30 min in the synthesis of 2-furaldehyde cyanohydrin.
(S)-HbHNL
(S)-MeHNL
(R)-PaHNL
Free
89 (94)
Aqua Gel
89 (94)
Free
90 (95)
Aqua Gel
95 (87)
CLEA
94 (94)
Free
91 (96)
Aqua Gel
95 (82)
Figure 4. The ees in the HNL-catalyzed synthesis of 2-furaldehyde cyanohydrin. Results in I are presented for PaHNL and
HbHNL. A: free PaHNL, B: free HbHNL, C: HbHNL aqua gel and D: PaHNL aqua gel. Results for MeHNL are presented
in II. E: free MeHNL, F: MeHNL-CLEA and G: MeHNL aqua gel.
Adv. Synth. Catal. 2006, 348, 1645 – 1654
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.asc.wiley-vch.de
1649

Fabien L. Cabirol et al.
FULL PAPERS
Table 5. Conversion ratios and ees (parentheses) at optimum reaction times in the synthesis of hexanal cyanohydrin.
(S)-HbHNL
(S)-MeHNL
(R)-PaHNL
Free
Aqua Gel
Free
Aqua Gel
CLEA
Free
Aqua Gel
4 h: 91 (94)
2 h: 92 (94)
3 h: 93 (84)
3 h: 96 (85)
3 h: 92 (81)
3 h: 91 (87)
3 h: 88 (85)
PaHNL- and MeHNL-catalyzed reactions reached ee the preparation of pyrethroid insecticides.[31,32] High
values of only 87% and 85%, respectively. conversions and ees have been reported for the addi-
The racemization of 2c was very limited, as illus- tion of hydrogen cyanide to 1d but the reaction had
trated for MeHNL (Figure 5), and could not account to be performed over 6 days[33] (50 U/mmol of HNL
for the lower ees observed. Catalyst selectivity for the from Sorghum bicolor) or at high catalyst loadings to
addition of hydrogen cyanide to hexanal is clearly shorten the reaction time[29] (15 min using 1000 U/
higher for HbHNL than PaHNL and MeHNL under mmol of HbHNL). Very good results have also been
the conditions used. The enantioenriched mixtures of reported for (S)-2d using mutant strains of
2c racemized at lower rates than those observed for MeHNL.[34] In the HbHNL aqua gel-catalyzed forma-
2a in the same conditions and only a slight influence tion of 2d, using acetone cyanohydrin as the cyanide
was observed for the immobilized enzymes vs. the source,[22] high conversion ratios were obtained but
free enzymes. The general trend of enzyme stabiliza- the ee of the final product was just over 40%, most
tion upon immobilization was consistent with the re- likely due to very long reaction times. The reaction
sults obtained for the other substrates.
time was also relatively long in our investigations (2–
3 days) but encapsulated HNLs afforded the corre-
sponding cyanohydrin in excellent conversion ratios
and ees (Table 6).
The use of hydrogen cyanide evidently improved
the performance of the encapsulated catalyst consid-
ering that the enzyme loading was much lower than
described earlier.[22] Free HNLs catalyzed the reac-
tion but the performance of the enzymes was strongly
dependent on the enzyme source (Figure 6). In paral-
lel with the results obtained for 2-furaldehyde (1b)
the stability of individual enzyme toward the sub-
strate/product of the reaction became apparent. High
initial rates were obtained using free HbHNL, but the
reaction became very sluggish upon accumulation of
the product in the reaction medium, suggesting a cya-
nohydrin-induced deactivation. Free MeHNL showed
very low activity under the conditions used for the
synthesis of 2d. It is noteworthy that the two oxygen-
containing substrates investigated (1b and 1d) gave
better results for free HbHNL than free-MeHNL. On
the other hand, the catalytic performances of encap-
sulated MeHNL were in both cases greater than for
HbHNL. The stabilization upon encapsulation toward
substrate/product deactivation for these compounds
was therefore more efficient for MeHNL than for
Figure 5. The ees in the MeHNL-catalyzed synthesis of hexa-
nal cyanohydrin. A: free MeHNL, B: MeHNL aqua gel, C:
MeHNL-CLEA. Racemisation rates upon the catalysis of
HbHNL and PaHNL (free enzyme and aqua gel) were in
the range presented here for MeHNL.
m-Phenoxybenzaldehyde
The (S)-enantiomer of cyanohydrin 2d, derived from HbHNL. CLEA immobilization also improved the
m-phenoxybenzaldehyde, is of commercial interest for stability of this enzyme but its overall performance
Table 6. Conversion ratios and ees (parentheses) at optimum reaction times in the synthesis of m-phenoxybenzaldehyde cya-
nohydrin.
(S)-HbHNL
(S)-MeHNL
(R)-PaHNL
Free
Aqua Gel
Free
Aqua Gel
CLEA
Free
Aqua Gel
72 h: 45 (82)
72 h: 92 (98)
72 h: 14 (75)
45 h: 98 (97)
72 h: 81 (83)
72 h: 68 (99)
72 h: 97 (99)
1650
www.asc.wiley-vch.de
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2006, 348, 1645 – 1654

Immobilized Hydroxynitrile Lyases for Enantioselective Synthesis of Cyanohydrins
FULL PAPERS
Figure 6. Conversion ratios in the HNL-catalyzed synthesis of m-phenoxybenzaldehyde cyanohydrin. Results in I are pre-
sented for free enzymes. A: PaHNL, B: HbHNL, C: MeHNL. The respective aqua gels and MeHNL-CLEA are presented
in II. D: MeHNL aqua gel, E: PaHNL aqua gel, F: HbHNL aqua gel and G: MeHNL CLEA.
was inferior to that of the aqua gel form. Further-
more, the solubility of m-phenoxybenzaldehyde in
aqueous media is very limited and the biphasic
system used for the free-enzyme-catalyzed reaction
did not favor the kinetics of the reaction when com-
pared to the organic media used for immobilized
forms. Similar results were obtained for the PaHNL-
catalyzed synthesis of (R)-2d.
3-Methyl-2-butanone
Conversion ratios greater than 90% were achieved in
the synthesis of 3-methyl-2-butanone cyanohydrin
(2e) but poor selectivity (ee<40%) was observed
under the conditions used. Literature results for the
Figure 7. Conversion ratios in the HNL-catalyzed synthesis
HbHNL aqua gel-catalyzed formation of 2e, using
of 3-methyl-2-butanone cyanohydrin. A: free MeHNL, B:
free HbHNL, C: PaHNL aqua gel, D: free PaHNL.
acetone cyanohydrin as a cyanide source,[22] were sig-
nificantly lower in terms of conversion (below 30%).
HCN as cyanide source shifted the equilibrium favor-
ably but catalyst selectivity in the relevant litera-
ture[27] was nonetheless higher (>70% ee) than in Conclusions
our results. The low catalyst loading was considered
unsuitable for this ketone and no acceptable enantio- The biocatalysts investigated herein were active for
purity could be achieved for 2e. Free MeHNL and the addition of hydrogen cyanide, in DIPE, to five
HbHNL were significantly more active than free structurally different carbonyl compounds. Catalytic
PaHNL for this substrate (Figure 7).
performance and robustness were strongly influenced
Aqua gel encapsulation of PaHNL improved its by the reaction media, the substrate/product, the
catalytic activity, but the reaction was still sluggish vs. enzyme source and consequently the structure of the
free MeHNL and HbHNL. The latter two (S)-selec- enzyme as well as the immobilization method. Encap-
tive enzymes have ketone-derived cyanohydrins as sulation of HNLs in an aqua gel matrix stabilized
natural substrates, while mandelonitrile is the natural them against deleterious effects of the reaction media
substrate of PaHNL. Hence, the catalytic activity of and substrate/product interference. Moreover, it could
HNLs from different sources toward 2e is in line with be demonstrated that a fast and highly enantioselec-
the natural substrate preferences of the individual en- tive formation of the desired cyanohydrins could be
zymes.
achieved with much lower enzyme loadings than re-
Adv. Synth. Catal. 2006, 348, 1645 – 1654
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.asc.wiley-vch.de
1651

Fabien L. Cabirol et al.
FULL PAPERS
ported earlier.[1,21,22] However, the reaction conditions
used here were not suitable for the preparation of 3-
methyl-2-butanone cyanohydrin in high enantiomeric
excess and efforts are currently being made to address
this limitation.
The stability of cross-linked enzyme aggregates of
HNLs in DIPE containing only traces of water was
strongly dependent on the enzyme structure. The
CLEA prepared from HbHNL as well as the commer-
cial CLEA from PaHNL did not allow an application
of these enzymes in a single organic phase. In contrast
quots) into cyanohydrin acetates, and GC analysis were per-
formed as reported in the literature.[22] Shorter retention
times are obtained with helium as carrier gas as compared
to nitrogen (Table 7). Depending on the reaction scale, 200
mL or 40 mL additions of n-dodecane were used as internal
standards to determine conversions and yields. Enantiomeric
excess values were calculated from the areas of the respec-
tive cyanohydrin acetate peaks. UV measurements were car-
ried out at 25.08C on a Varian CARY 3 spectrophotometer.
Table 7. GC retention times with helium as carrier gas.
the MeHNL CLEA showed outstanding stability in Substrate Rt
Dodecane
[min] Rt [min]
(R)-Acetate (S)-Acetate
Rt [min]
Rt [min]
DIPE. This opens up new opportunities for the appli-
cation of this enzyme, which are described in an ac-
companying article.[35]
1a
1.16 1.74
1.21 3.84
1.67 6.92
6.74 1.73
1.18 18.85
4.04
4.91
7.28
9.84
15.37
4.52
4.46
7.61
10.06
15.59
1b[a]
1c
1d
1e
Experimental Section
[a]
CIP rules invert the designation of the respective enantio-
mers for 2-furaldehyde cyanohydrin. Temperature as ref.[27]
General Remarks
Enzymes: The hydroxynitrile lyase (HNL) from Prunus
Enzyme Activity Measurements
˝
amygdalus (PaHNL, [EC 4.1.2.10], Julich Fine Chemicals,
300 U/mL) was commercially available in 50% glycerol.
More concentrated enzyme solutions from Hevea brasiliensis
(HbHNL, [EC 4.1.2.39], DSM, 3.6k U/mL), and Manihot es-
Enzymatic activity of free enzymes[26] and aqua gels[22] was
measured according to reported literature procedures.
CLEAs were suspended in 25 mM potassium phosphate
buffer (pH 6.5) and samples of this suspension were used to
calculate the activity according to the procedure reported
for the free enzyme.[26] UV measurements were performed
with continuous stirring in order to keep the CLEA sus-
pended.
˝
culenta (MeHNL, [EC 4.1.2.39], Julich Fine Chemicals, 2.3
kU/mL) were diluted to 274 U/mL and 230 U/mL, respec-
tively, using 25 mM potassium phosphate buffer (pH 6.5).
CLEAs of PaHNL (4.61 kU/g) and MeHNL (1.02 kU/g)
were prepared according to commercial procedures (CLEA
Technologies).
Chemicals: Diethyl ether solutions of benzaldehyde
(Fluka, 99%+), (Æ)-mandelonitrile (Acros Organics, tech-
nical grade, distilled before use), 2-furaldehyde (Acros Or-
ganics, 99%), hexanal (Aldrich, 98%), and m-phenoxyben-
zaldehyde (Acros Organics, 97%) were treated with saturat-
ed sodium bicarbonate solution prior to each use to remove
traces of acid. The organic phase was dried and the solvent
was removed under reduced pressure. (S)-Mandelonitrile
(96% purity, containing 4 mol% benzaldehyde; 99.3% ee)
was prepared according to a literature procedure.[35] Ethyl-
ene glycol dimethyl ether (glyme, Aldrich, 99.5%), dichloro-
methane (Aldrich, Anhydrous 99.8%), pyridine (Aldrich,
Anhydrous, 99.8%), acetic anhydride (Acros Organics,
99%+), dodecane (Acros Organics, 99%), methyltrime-
thoxysilane (MTMS, Aldrich, 98%), tetramethoxysilane
(TMOS, Fluka, 99%+), glutaraldehyde (Fluka, 25% in
water, ca. 2. 6M), and 3-methyl-2-butanone (MIPK, Aldrich,
99%) were used as supplied, without further purification.
Diisopropyl ether (DIPE, Acros Organics, 98%+, stabilized
with 2,6-di-tert-butyl-p-cresol) was used without further
treatment unless otherwise specified. Aqueous buffers were
prepared from analytical grade salts and stabilized with
0.09% sodium azide.
Preparation of HbHNL-CLEA
HbHNL-CLEA was prepared based on the reported proce-
dure for PaHNL.[21] HbHNL stock solution (frozen at
À208C) was diluted to 1.8 kU/mL in 25 mM potassium
phosphate buffer (pH 6.5). The dilute sample (4.5 mL) was
allowed to warm up slowly from À208C to 08C (ice bath)
and glyme (4.5 mL) was then added. Precipitation was al-
lowed to proceed for 15 min while stirring at 08C and gluta-
raldehyde (1.5 mL, 25% in water) was then added. The mix-
ture was stirred at 08C for 17 h. The CLEA was filtered and
rinsed thoroughly with acetonitrile and diethyl ether. After
vacuum drying for 4 h, HbHNL-CLEA (496 mg, 2.61 kU/g
as determined by the activity test described above) was
stored at À208C.
Hydrogen Cyanide (HCN): 2M Solution in DIPE
Caution: Due to the toxicity of hydrogen cyanide, all proce-
dures involving this stock solution were performed in a well-
ventilated fume hood equipped with an HCN detector.
HCN-containing wastes were neutralized using commercial
bleach and stored independently for disposal.
Analytical Methods: The course of the reaction was fol-
lowed by chiral gas chromatography on a Shimadzu Gas
Chromatograph GC-14B equipped with a FID detector and
a beta-cyclodextrin column (CP-Chirasil-Dex CB 25 m ”
0.25 mm). Derivatization of reaction samples (20 mL ali-
1652
www.asc.wiley-vch.de
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2006, 348, 1645 – 1654

Immobilized Hydroxynitrile Lyases for Enantioselective Synthesis of Cyanohydrins
FULL PAPERS
Sodium cyanide (49 g, 1.0 mol) was dissolved in a mixture
of water (100 mL) and diisopropyl ether (DIPE) (250 mL)
at 08C. The biphasic system was stirred vigorously for
15 min and 30% aqueous HCl (100 mL) was added slowly.
This mixture was allowed to warm slowly to room tempera-
ture (at least 25 min). The phases were separated and 150
mL of DIPE was added to the organic layer. The combined
organic phases were stirred and residual water was separat-
ed. This procedure was repeated with another 100 mL of
DIPE. The 2M standard HCN solution[36] was kept over
citric acid buffer (pH 5.5) in the dark.
chiral GC over three days while shaking the sealed flask at
room temperature.
HNL-CLEA-Catalyzed Synthesis of Cyanohydrins
The procedure given here for HbHNL-CLEA was scaled ac-
cordingly for individual CLEAs to suit 6 U/mmol loading of
catalyst. HbHNL-CLEA (30 units, 2610 U/g) was suspended
in commercial DIPE (2.5 mL). The starting material
(5 mmol) and dodecane (200 mL) were added to the mixture
and a GC sample was taken to determine initial conditions.
The reaction was initiated by addition of 2M HCN in DIPE
(7.5 mL, 3 equivs.) and monitored by chiral GC over 3 days
while shaking in a sealed flask at room temperature. The
HbHNL-CLEA-catalyzed synthesis of mandelonitrile in mi-
croaqueous media was performed by first suspending the
catalyst (30 units) in 200 mL of 50 mM citrate/potassium
phosphate buffer (pH 5.0) and the procedure was followed
accordingly.
Free HNL-Catalyzed Synthesis of Cyanohydrins
The HNL (30 units) from Prunus amygdalus (300 U/mL),
Manihot esculenta (230 U/mL) or Hevea brasiliensis (274 U/
mL) was diluted in 50 mM citrate/potassium phosphate
buffer (pH 5.0) to a total volume of 2 mL. Diisopropyl ether
(2.5 mL) was then added followed by the carbonyl com-
pound of interest (5 mmol) and n-dodecane (200 mL). An
analytical sample representative of initial conditions (5 mL)
was drawn from the organic layer and diluted in DIPE (1
mL) for GC analysis. The reaction was initiated by addition
of 2M HCN in DIPE (7.5 mL, 3 equivs.) and monitored by
chiral GC over three days while shaking the sealed flask at
room temperature.
Racemization of (S)-Mandelonitrile
The non-enzymatic racemization of (S)-mandelonitrile was
based on the conditions at full conversion of the synthetic
reaction. Conditions in the biphasic system used for the free
enzyme were reproduced by adding HCN (2 equivs., 5 mL)
to a mixture of DIPE (4 mL) and 2 mL of 50 mM citrate/po-
tassium phosphate buffer (pH 5.0). The reaction was initiat-
HNL Aqua Gel-Catalyzed Synthesis of Cyanohydrins
ed by addition of (S)-mandelonitrile (1 mL) from
a
5 mmolmLÀ1 stock solution in DIPE:dodecane (4:1) and
monitored by chiral GC. The racemization reaction de-
scribed above was repeated in commercial DIPE in order to
model the conditions of the CLEA-catalyzed reaction.
The influence of the carrier in aqua gels was evaluated
from the rate of racemization of (S)-mandelonitrile in the
presence of enzyme-free aqua gels prepared from 500 mL of
precursor and 500 mL of 50 mM citrate/potassium phosphate
(pH 5.0). Enzyme free gels were ground into a fine powder
and DIPE (4 mL) saturated with 50 mM citrate/potassium
phosphate buffer (pH 5.0) was added followed by HCN (5
mL, 2 equivs.). The reaction was initiated by addition of the
(S)-mandelonitrile stock solution described above (1 mL)
and monitored by chiral GC over three days while shaking
in a sealed flask at room temperature.
500 mL of PaHNL (300 U/mL), MeHNL (230 U/mL) and
HbHNL (274 U/mL) were encapsulated into aqua gels using
500 mL of precursor prepared according to the literature.[22]
The exchange of aqueous buffers in the gels was allowed to
proceed over two days. Gels were then ground into a fine
powder and used immediately. The scale of the reaction
(Table 8) was adapted from the activity recovery (Table 1)
to suit 6 U/mmol catalyst loading.
The HbHNL sol gel reaction is given as a representative
procedure. The aqua gels were ground into a fine powder
and DIPE (2.5 mL) saturated with 50 mM citrate/potassium
phosphate buffer (pH 5.0) was added. The substrate
(5 mmol) and n-dodecane (200 mL) were dissolved into the
mixture. A GC sample was taken to determine initial condi-
tions and the reaction was initiated by addition of 2M HCN
in DIPE (7.5 mL, 3 equivs.). The reaction was monitored by
Acknowledgements
Table 8. Reaction conditions for the synthesis of cyanohy-
drins catalyzed by aqua gels.
The authors wish to thank L. Veum (TUDelft), A. Chmura
(TUDelft), C. Mateo (TUDelft), J. Highfield (ICES) and L.
van Langen (CLEA Technologies) for helpful discussions.
The generous gift of CLEAs and (S)-mandelonitrile by A.
Chmura (TUDelft) and HbHNL from DSM (Marcel Wub-
bolts) are gratefully acknowledged. The authors also express
their thanks to ICES for financial support.
HbHNL MeHNL PaHNL
Residual activity in aqua gel[a] 30
[units]
9.8
1.6
6.1
22.5
3.75
6
Reaction Scale[b] [mmol sub-
strate]
Catalyst Loading [U/mmol]
5
6
[a]
Calculated for a gel prepared from 500 mL of precursor
and 500 mL of HNL dilute solution.
All other reaction parameters were scaled accordingly.
[b]
Adv. Synth. Catal. 2006, 348, 1645 – 1654
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.asc.wiley-vch.de
1653

Fabien L. Cabirol et al.
FULL PAPERS
References
[20] H. Grçger, E. Capan, A. Barthuber, K. D. Vorlop, Org.
Lett. 2001, 3, 1969–1972.
[1] R. J. H. Gregory, Chem. Rev. 1999, 99, 3649–3682.
[2] J.-M. Brunel, I. P. Holmes, Angew. Chem. Int. Ed. 2004,
43, 2752–2778.
[3] M. North, Tetrahedron:Asymmetry 2003, 14, 147- 176.
[4] J. Blacker, M. North, Chemistry & Industry 2005, 20
June, 22–25.
[5] J. Sukumaran, U. Hanefeld, Chem. Soc., Rev. 2005, 34,
530–542.
[6] M. Schmidt, H. Griengl, Top. Curr. Chem. 1999, 200,
193–226.
[21] L. M. van Langen, R. P. Selassa, F. van Rantwijk, R. A.
Sheldon, Org. Lett. 2005, 7, 327–329.
[22] L. Veum, U. Hanefeld, A. Pierre, Tetrahedron 2004, 60,
10419–10425.
[23] K. Gruber, C. Kratky, J. Pol. Sci. A:Polymer Chemistry
2004, 42, 479–486.
[24] U. Hanefeld, A. J. J. Straathof, J. J. Heijnen, J. Mol.
Catal. B:Enzym. 2001, 11, 213–219.
[25] S. Han, G. Lin, Z. Li, Tetrahedron:Asymmetry 1998, 9,
1835–1838.
[7] F. Effenberger, S. Fçrster, H. Wajant, Curr. Opin. Bio-
technol. 2000, 11, 532–539.
[8] H. Griengl, H. Schwab, M. Fechter, Trends in Biotech-
[26] U. Hanefeld, A. J. J. Straathof, J. J. Heijnen, Biochim.
Biophys. Acta 1999, 1432, 185–193.
[27] S. Budavari, The Merck Index, 12th edn., Merck & Co.,
Inc., Whitehouse Station, NJ, 1996, entry 5759.
[28] D. R. Lide, Handbook of Chemistry and Physics, CRC
Press LLC, Boca Raton, 2000, 8–94.
nol. 2000, 18, 252–256.
[9] D. V. Johnson, A. A. Zabelinskaja-Mackova, H.
Griengl, Curr. Opin. Chem. Bio. 2000, 4, 103–109.
[10] D. Costes, E. Wehtje, P. Adlercreutz, Enzyme Microb.
Technol. 1999, 25, 384–391.
[29] H. Griengl, N. Klempier, P. Pçchlauer, M. Schmidt, N.
Shi, A. A. Zabelinskaja-Mackova, Tetrahedron 1998,
54, 14477–14486.
[30] M. Schmidt, S. HervØ, N. Klempier, H. Griengl, Tetra-
hedron 1996, 52, 7833–7840.
[31] J. Martel, in: Chirality in Industry, (Eds.: A. N. Collins,
G. N. Sheldracke, J. Crosby), John Wiley & Sons Ltd,
Chichester, 1992, pp. 87–109.
[11] M. Bauer, H. Griengl, W. Steiner, Enzyme Microb.
Technol. 1999, 24, 514–522.
[12] F. Effenberger, J. Eichhorn, J. Roos, Tetrahedron:
Asymmetry 1995, 6, 271–282.
[13] C. Kolber, A. Bohrer, F. Effenberger, Tetrahedron
2004, 60, 10397–10410.
[14] C. Kolber, F. Effenberger, Tetrahedron:Asymmetry
2004, 15, 3731–3742.
ˇ
[32] G. M. Ramos Tombo, D. Bellus, Angew. Chem. Int. Ed.
1991, 30, 1193–1215.
[33] F. Effenberger, B. Hçrsch, S. Fçrster, T. Ziegler, Tetra-
[15] D. Costes, E. Wehtje, P. Adlercreutz, J. Mol. Cat. B:
Enzym. 2001, 11, 607–612.
hedron Lett. 1990, 31, 1249–1252.
[16] P. López-Serrano, L. Cao, F. van Rantwijk, R. A. Shel-
don, Biotechnol. Lett. 2002, 24, 1379–1383.
[17] T. S. Lee, J. D. Vaghjiani, G. J. Lye, M. K. Turner,
Enzyme Microb. Technol. 2000, 26, 582–592.
[18] L. Cao, L. M. van Langen, R. A. Sheldon, Curr. Opin.
Biotechnol. 2003, 14, 387–394.
[34] H. Bꢂhler, F. Effenberger, S. Fçrster, J. Roos, H.
Wajant, ChemBioChem. 2003, 4, 211–216.
[35] A. Chmura, G. M. van der Kraan, F. Kielar, L. M. van
Langen, F. van Rantwijk, R. A. Sheldon, Adv. Synth.
Catal. 2006, 348, 1655–1661.
[36] L. M. van Langen, F. van Rantwijk, R. A. Sheldon, Or-
ganic Process Research & Development 2003, 7, 828–
8831.
[19] R. Schoevaart, M. W. Wolbers, M. Golubovic, M.
Ottens, A. P. G. Kieboom, F. van Rantwijk, L. A. M.
van der Wielen, R. A. Sheldon, Biotechnol. Bioeng.
2004, 87, 754–762.
1654
www.asc.wiley-vch.de
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2006, 348, 1645 – 1654