Construction of Recombinant Escherichia coli catalysts which simultaneously express an (S)-oxynitrilase and different nitrilase variants for the synthesis of (S)-mandelic acid and (S)-mandelic amide from benzaldehyde and cyanide

Article abstract of DOI:10.1002/adsc.200900087

Recombinant Escherichia coli strains were constructed which simultaneously expressed the genes encoding the (S)-oxynitrilase from cassava (Manihot esculenta) together with the wild-type or a mutant variant of the arylacetonitrilase from Pseudomonas fluorescens EBC191 in a single organism under the control of a rhamnose-inducible promoter. The whole cell catalysts obtained converted benzaldehyde and potassium cyanide in aqueous media at pH 5.2 mainly to (S)-mandelic acid and/or (S)-mandelic amide and synthesized only low amounts of the corresponding (R)-enantiomers. The conversion of benzaldehyde and potassium cyanide (KCN) by a whole-cell catalyst simultaneously expressing the (5)-oxynitrilase and the wild-type nitrilase resulted in a ratio of (S)-mandelic acid to (S)-mandelic amide of about 4:3. This could be explained by the strong nitrile hydratase activity of the wild-type nitrilase with (S)-mandelonitrile as substrate. The relative proportion of (S)-mandelic amide formed in this system was significantly increased by coexpressing the (S)-oxynitrilase with a carboxy-terminally truncated variant of the nitrilase. This whole-cell catalyst converted benzaldehyde and KCN to mandelic amide and mandelic acid in a ratio of about 9:1. The ee of the (S)-mandelic amide formed was calculated to be < 95%.

Full text of DOI:10.1002/adsc.200900087

FULL PAPERS
DOI: 10.1002/adsc.200900087
Construction of Recombinant Escherichia coli Catalysts which
Simultaneously Express an (S)-Oxynitrilase and Different
Nitrilase Variants for the Synthesis of (S)-Mandelic Acid and
(S)-Mandelic Amide from Benzaldehyde and Cyanide
Olga Sosedov,a Kathrin Matzer,a Sibylle Bꢀrger,a Christoph Kiziak,a,b
Stefanie Baum,a Josef Altenbuchner,c Andrzej Chmura,d Fred van Rantwijk,d
and Andreas Stolza,*
a
Institut fꢀr Mikrobiologie, Universitꢁt Stuttgart, Allmandring 31, 70569 Stuttgart, Germany
Fax : (+49)-711-685-65725; e-mail: Andreas.Stolz@imb.Uni-Stuttgart.de
Present address: Lonza AG, Biotechnology Research and Development, 3930 Visp, Switzerland
Institut fꢀr Industrielle Genetik, Universitꢁt Stuttgart, Allmandring 31, 70569 Stuttgart, Germany
Laboratory of Biocatalysis and Organic Chemistry, Department of Biotechnology, Delft University of Technology,
b
c
d
Julianalaan 136, 2628 BL Delft, The Netherlands
Received: February 10, 2009; Revised: May 15, 2009; Published online: June 16, 2009
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/adsc.200900087.
Abstract: Recombinant Escherichia coli strains were ratio of (S)-mandelic acid to (S)-mandelic amide of
constructed which simultaneously expressed the about 4:3. This could be explained by the strong ni-
genes encoding the (S)-oxynitrilase from cassava trile hydratase activity of the wild-type nitrilase with
(Manihot esculenta) together with the wild-type or a (S)-mandelonitrile as substrate. The relative propor-
mutant variant of the arylacetonitrilase from Pseudo- tion of (S)-mandelic amide formed in this system
monas fluorescens EBC191 in a single organism was significantly increased by coexpressing the (S)-
under the control of a rhamnose-inducible promoter. oxynitrilase with a carboxy-terminally truncated var-
The whole cell catalysts obtained converted benzal- iant of the nitrilase. This whole-cell catalyst convert-
dehyde and potassium cyanide in aqueous media at ed benzaldehyde and KCN to mandelic amide and
pH 5.2 mainly to (S)-mandelic acid and/or (S)-man- mandelic acid in a ratio of about 9:1. The ee of the
delic amide and synthesized only low amounts of the (S)-mandelic amide formed was calculated to be
corresponding (R)-enantiomers. The conversion of >95%.
benzaldehyde and potassium cyanide (KCN) by a
whole-cell catalyst simultaneously expressing the (S)- Keywords: amide formation; cascade reactions;
oxynitrilase and the wild-type nitrilase resulted in a enantioselectivity; hydroxynitrile lyase; nitrilase
Introduction
and further converted to a-hydroxy amides and a-hy-
droxy carboxylic acids which are important products
Nitriles are used as intermediates in a wide range of in the chemical industry.[2] There are two main enzy-
synthetic reactions in the chemical industry and there matic pathways known which result in the formation
is a considerable interest to use microorganisms or of optical active 2-hydroxy acids from a-hydroxy ni-
enzymes to synthesize or convert organic nitriles. triles (Figure 1).
These biotransformations allow chemo-, stereo- or
The most versatile type of reactions uses oxynitri-
enantioselective reactions that are often difficult to lases from plants which catalyse the enantioselective
perform in synthetic chemistry and are usually more addition of HCN to aldehydes or ketones.[3,4] Differ-
benign to the environment than traditional chemical ent oxynitrilases produce either (S)- or (R)-a-hydroxy
processes.[1] An interesting group of nitriles for chiral nitriles often with a high degree of enantioselectivity;
synthetic processes are a-hydroxy nitriles, because these subsequently can be chemically or enzymatically
these compounds can be easily chemically prepared converted in enantioconservative reaction steps to
Adv. Synth. Catal. 2009, 351, 1531 – 1538
ꢂ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1531

Olga Sosedov et al.
FULL PAPERS
Figure 1. Schematic representation of the possible pathways for the synthesis of optical active mandelic acid using nitrilases
and oxynitrilases
chiral a-hydroxy acids.[5,6] The second established bio- ylotrophic yeast Pichia pastoris, which simultaneously
transformation process applies the (reversible) chemi- expressed the genes encoding the (S)-specific oxyni-
cal addition of HCN to aldehydes and ketones (result- trilase from cassava and the arylacetonitrilase from P.
ing in racemic hydroxy nitriles) combined with enan- fluorescens EBC191 under the control of individual
tioselective nitrilases. Unfortunately, only very few AOX1 promotors. This whole-cell catalyst converted
truly enantiospecific nitrilases are currently known benzaldehyde plus cyanide in an aqueous medium at
and these nitrilases have only been shown to convert a pH of 3.8 to mandelic acid and mandelic amide with
(substituted) mandelonitrile(s) to (R)-mandelic a large surplus of the corresponding (S)-enantio-
acid(s).[7–11]
mers.[17] Unfortunalely, only rather low enzyme activi-
We are currently attempting to combine enantio- ties were observed and still some of the correspond-
specific oxynitrilases and bacterial nitrilases in a ing (R)-enantiomers were formed. In order to solve
“one-pot” reaction in order to synthesize enantiopure these problems, it was attempted in the present study
a-hydroxy carboxylic acids. In practice, this process is to combine both enzyme activities in a recombinant
hampered because both reactions must be performed E. coli strain in order to generate “bienzymatic
using rather different conditions. Thus, synthetic oxy- whole-cell catalysts” with higher specific activities for
nitrilase reactions are usually performed at pH <5 in the desired reaction.
order to suppress the uncatalyzed (and therefore non-
enantiospecific) hydrocyanation reactions. Further-
more, the synthetic oxynitrilase reactions are usually Results
performed in the presence of large amounts of organ-
ic solvents in order to further suppress the reversible Construction of a Recombinant E. coli Strain
chemical reaction and to provide a reservoir of Simultaneously Carrying an Oxynitrilase and a
HCN.[12] In contrast, the currently known nitrilases Nitrilase Gene
possess their pH optima close to pH 7 and are usually
inactivated at higher concentrations of organic sol- Plasmid pIK9 was used to express the nitrilase gene
vents.[13,14]
from Pseudomonas fluorescens EBC191. (For the
We have used two different strategies in order to plasmids used and constructed in this study see
combine the two enzymes for synthetic purposes. Table 1.) This expression plasmid is a pBR322-deriva-
First, methods were established to combine the en- tive carrying an ampicillin resistance gene and allows
zymes under in vitro conditions as (immobilized) bio- the highly efficient expression of the cloned nitrilase
catalysts. Thus, purified preparations of an (S)-specific under the control of a rhamnose-inducible promot-
oxynitrilase from cassava (Manihot esculenta) were er.[15]
co-immobilized with an arylacetonitrilase from Pseu-
Plasmid pQE4-MeHNL was used as source of the
domonas fluorescens EBC191, which converts a broad hydroxynitrile lyase gene from cassava (Manihot escu-
range of a-substituted nitriles and it was demonstrat- lenta).[18] The hydroxynitrile lyase gene was amplified
ed that bienzymatic cross-linked enzyme aggregates by PCR and subsequently transferred into the expres-
(CLEAs) could be generated which converted benzal- sion vector pAW229 giving plasmid pJOE 5361.1 as
dehyde and HCN to (S)-mandelic acid.[16] The second described in the Experimental Section. Plasmid
strategy focuses on the construction of acid-tolerant pAW229 is a pACYC184 derivative carrying a chlor-
whole-cell catalysts. We have previously described the amphenicol resistance gene and the rhaPBAD promot-
construction of a recombinant derivative of the meth- er.[19] Therefore, in plasmid pJOE 5361.1 the hydroxy-
1532
asc.wiley-vch.de
ꢂ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2009, 351, 1531 – 1538

Construction of Recombinant Escherichia coli Catalysts
Table 1. Plasmids used or constructed in the present study.
FULL PAPERS
soluble proteins (see Figure S1 in the Supporting In-
formation).
Plasmid
Relevant characteristics
Source
or ref.
pAW229
pIK9
Camr
[19]
Detection of Oxynitrilase and Nitrilase Activities in
E. coli JM109ACTHNUTRGNEU(GN pIK9)(pJOE5361.1)
Plasmid pJOE2775 with the nitrilase [15]
gene from P. fluorescens EBC 191
carrying a C-terminal (His)6-tag
C-terminally deleted nitrilase
(DelC-60) in pJOE2702
The enzymatic formation and turn-over of mandeloni-
trile had to be analyzed at pH ꢀ 5.5 in order to sup-
press the spontaneous (non-enantioselective) chemi-
cal formation or decomposition of mandelonitrile.[6]
pCK105
pQE4-
[23]
Gene encoding the oxynitrilase from [18]
MeHNLwt Manihot esculenta cloned into plas-
mid pQE4
pJOE5356.1 Part of the gene encoding the oxyni- [17]
trilase from Manihot esculenta
E. coli JM109ACHTUNTRGNEUNG(pIK9)(pJOE5361.1) was grown over-
night in LB-medium plus ampicillin (100 mgmLÀ1),
chloramphenicol (20 mgmLÀ1) and rhamnose (0.2%
w/v). The cells were harvested by centrifugation, re-
suspended in sodium citrate buffer (25 mM, pH 5.2)
and the nitrilase activity determined with resting cells
and mandelonitrile (10 mM) as substrate using HPLC.
These cells converted mandelonitrile to mandelic acid
and mandelic amide in a ratio of approximately 4:1 as
cloned into plasmid pAW229
pJOE5361.1 Complete gene encoding the oxyni- [17]
trilase from Manihot esculenta
cloned into plasmid pAW229
nitrile lyase was under the control of the same rham- previously also shown for E. coli JM109
nose-inducible promoter as present in plasmid pIK9. E. coli JM109(pIK9)(pJOE5361.1) a nitrilase specific
(pIK9).[15] For
ACHTUNGTRENNUNG
AHCTUNGTRENNUNG
Thus it was possible to simultaneously express both activity of 2.5 U/mg of protein was calculated. (In
genes in a single strain by the addition of rhamnose. contrast, the wild-type cells of Pseudomonas fluores-
cens EBC191 demonstrated about 0.7 U/mg with
mandelonitrile as substrate.[24]
)
Simultaneous Expression of the Cassava Oxynitrilase
and the Pseudomonas fluorescens EBC191 Nitrilase
in Escherichia coli JM109ACTHNUGTRNEUNG(pIK9)(pJOE5361.1)
The oxynitrilase activity of the whole cells was esti-
mated by incubating the cells in 50 mM sodium citrate
buffer (pH 5) with benzaldehyde (5 mM) and KCN
(5 mM) and analyzing the reactions by reversed-phase
E. coli JM109
pJOE5361.1 and putative transformants carrying both JM109
plasmids were selected on LB-medium with ampicillin dehyde faster than observed in control experiments
(100 mgmLÀ1) plus chloramphenicol (20 mgmLÀ1). The with cells of E. coli JM109
(pIK9) which only synthe-
obtained clones of E. coli JM109(pIK9)(pJOE5361.1) sized the nitrilase activity. For the oxynitrilase a spe-
ACHTUNGTRENNUNG
AHCTUNGTRENNUNG
AHCTUNGTRENNUNG
AHCTUNGTRENNUNG
were grown in liquid culture on the selective medium cific activity of about 0.4 U/mg of protein was esti-
and the nitrilase and oxynitrilase genes induced by mated from the decrease in the benzaldehyde concen-
the addition of rhamnose (0.2% w/v). For comparison, tration (corrected for the rate of the spontaneous de-
also
JM109(pJOE5361.1) were analyzed in the same way. of E. coli JM109
The cells were harvested after 1–18 h incubation with this experiment mandelic acid and mandelic amide in
rhamnose and protein synthesis analyzed by SDS- a ratio of 1.2:1. In contrast, E. coli JM109(pIK9) pro-
PAGE. These experiments demonstrated that E. coli duced from racemic mandelonitrile only about 19%
JM109(pJOE5361.1) and E. coli JM109
(pIK9) amide.[15] This indicated that this strain indeed formed
E.
coli
JM109
(pIK9)
and
E.
coli crease in the benzaldehyde concentration). The cells
AHCTUNGTRENNUNG
AHCTUNGTRENNUNG
AHCTUNGTRENNUNG
(pJOE5361.1) indeed synthesized a protein with the a large proportion of (S)-mandelonitrile because it
size of the cassava oxynitrilase. Furthermore, the ex- was earlier found that the nitrilase from P. fluorescens
periments suggested that the oxynitrilase expression EBC191 formed from (S)-mandelonitrile approxi-
was more pronounced in E. coli JM109(pJOE5361.1) mately the same amounts of (S)-mandelic amide and
than in E. coli JM109
ACHTUNGTRNE(NUNG pIK9)(pJOE5361.1) and that in (S)-mandelic acid, whereas (R)-mandelonitrile was
E. coli JM109(pIK9)(pJOE5361.1) the nitrilase was converted to (R)-mandelic acid and (R)-mandelic
ACHTUNGTRENNUNG
much more efficiently expressed than the oxynitrilase. amide in a ratio of about 9:1.[15,16]
Furthermore, it was found that both proteins were op-
timally synthesized after prolonged incubation times
(>12 h) of the recombinant strains with rhamnose.
SDS-PAGE gel electrophoresis suggested that the
oxynitrilase and nitrilase proteins represented about
2–3% and 5%, respectively, of the totally formed
Adv. Synth. Catal. 2009, 351, 1531 – 1538
ꢂ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
1533

Olga Sosedov et al.
FULL PAPERS
Figure 2. Conversion of benzaldehyde and KCN by resting cells of E. coli JM109ACTHNUGTRNEUNG(pIK9)(pJOE5361.1) (left) and E. coli
JM109ACHTUNGTRENNUNG(pIK9) (right). The reaction mixtures contained in a total volume of 2 mL 100 mM sodium citrate buffer (pH 5.0),
15 mM benzaldehyde and resting cells (OD546nm =20). The reactions were started by the addition of KCN (15 mM). Aliquots
(200 mL each) were taken at the indicated time intervals and the reactions terminated by the addition of 20 mL 1M HCl. The
*
cells were removed by centrifugation in an Eppendorf centrifuge (5 min, 14,000 g). The concentrations of benzaldehyde ( ),
~
&
*
(R,S)-mandelonitrile ( ), (R)-mandelic amide (&), (S)-mandelic amide ( ), (R)-mandelic acid ( ), and (S)-mandelic acid
(
!
) were determined in the supernatants by reversed phase and chiral HPLC.
Enantioselective Formation of (S)-Mandelic Acid
from Benzaldehyde and KCN using E. coli
JM109ACHTUNGTRENNUNG(pIK9)(pJOE5361.1)
cells was increased and the reaction times decreased
compared to the previously described experiment.
The reactions were analyzed by reversed-phase and
chiral HPLC and it was found that it was indeed pos-
The simultaneous production of the cassava oxynitri- sible to obtain only the (S)-enantiomers of the amide
lase and the Pseudomonas nitrilase in one recombi- and the acid (Figure 2, left; Table 2 entry A). In con-
nant organism should allow the enantioselective syn- trast, in the control experiment with E. coli JM109-
thesis of (S)-mandelic acid from benzaldehyde and cy-
ACHTUNGTREN(NUNG pIK9) more (R)-mandelic acid than (S)-mandelic
anide, because the cassava oxynitrilase is able to cata- acid was formed (Figure 2, right).
lyze the highly enantiospecific conversion of benzal-
dehyde and cyanide to (S)-mandelonitrile[20] and the
nitrilase from P. fluorescens EBC191 is able to con- Optimization of the Reaction Conditions
vert (S)-mandelonitrile to (S)-mandelic acid.[6,15] In
the following experiment it was tested if it was possi- The previously described reactions were performed
ble to completely suppress the formation of (R)-man- with rather low equimolar concentrations of benzal-
delic amide and (R)-mandelic acid using the recombi- dehyde and KCN (5–15 mM). In the following experi-
nant E. coli strain. Therefore, the amount of resting ments increasing initial benzaldehyde concentrations
Table 2. Whole cell conversions of benzaldehyde and cyanide using E. coli JM109ACTHNUTRGENUGN(pIK9) (pJOE5361.1).
Entry
Benzaldehyde
[mM]
Cyanide
[mM]
Phases[a]
Buffer
Time
[h]
Conversion
[%]
Acid/Amide
ee [%]
(S)-acid
A
B
C
D
E
F
15
10
25
10
25
50
100
15
30
75
30
75
150
300
1
1
1
2
2
2
2
Citrate (pH 5, 100 mM)
Citrate (pH 5, 200 mM)
Citrate (pH 5, 200 mM)
Citrate (pH 5, 50 mM)
Citrate (pH 5, 50 mM)
Citrate (pH 5, 50 mM)
Citrate (pH 5, 50 mM)
0.33
0.5
0.5
7.5
7.5
7.5
7.5
98.0
96.1
95.0
84.2
96.6
99.0
96.4
1.4
1.3
1.2
1.7
1.6
1.4
1.4
>99
>99
>99
>99
>99
>99
>99
G
[a]
The two-phase systems consisted of 70% (v/v) DIPE and 30% (v/v) 50 mM soduum citrate buffer (pH 5).
1534
asc.wiley-vch.de
ꢂ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2009, 351, 1531 – 1538

Construction of Recombinant Escherichia coli Catalysts
FULL PAPERS
of 10–200 mM combined with the respective three-
fold concentrations of KCN were tested (in 250 mM
sodium citrate buffer, pH 5.0). The reaction mixtures
contained cells of E. coli JM109ACHTUNTRGNEUNG(pIK9)(pJOE5361.1)
with an optical density (OD546nm) of 2.6 and the reac-
tions were started by the addition of benzaldehyde.
The addition of increasing concentrations of KCN re-
sulted in a significant alkalization of the media and
the media containing 50–200 mM KCN had initial pH
values of pH 9–13. Nevertheless, it was found that the
whole cell catalysts were able to convert up to 50 mM
benzaldehyde within 10 min and that even an initial
benzaldehyde concentration of 100 mM was almost
completely converted within 30 min. In contrast, the
reaction mixtures containing 200 mM benzaldehyde
were active only for about 10 min and only about
25% of the initially added benzaldehyde was convert-
ed. The termination of the reaction was correlated to
Figure 3. Conversion of benzaldehyde and KCN by resting
cells of E. coli JM109ACHTUNRGTNE(NUG pCK105)(pJOE5361.1). The reaction
an increased viscosity of the reaction mixtures which mixtures contained in a total volume of 1 mL 100 mM
sodium citrate buffer (pH 5.0), 10 mM benzaldehyde,
10 mM KCN and resting cells (OD600nm =22). Aliquots
(90 mL each) were taken at the indicated time intervals and
the reactions terminated by the addition of 10 mL 1M HCl.
The cells were removed by centrifugation in an Eppendorf
centrifuge (5 min, 14,000 g) and the concentrations of ben-
was probably caused by a rapid cell lysis at the alka-
line pH value of the medium.
The ee values of the products formed [(S)-mandelic
acid and (S)-mandelic amide] were >99% in the reac-
tions with initial benzaldehyde concentrations of
<50 mM (Table 2, entries B, C). The ee values signifi-
cantly dropped (<50%) in those experiments with in-
creasing substrate concentrations – very probably due
to the increased pH values of the reaction media.
Oxynitrilases are usually applied in vitro in the
~
&
*
zaldehyde ( ), mandelic amide ( ), mandelic acid ( ), and
*
mandelonitrile ( ) in the supernatants were analyzed by re-
versed phase HPLC.
“traditional” chemical synthesis in the presence of of the nitrilase formed up to 50% of mandelic amide
large amounts of organic solvents.[21,22] Therefore, the from racemic mandelonitrile.[23] It was therefore at-
activity of the whole cell catalyst was also tested in tempted to increase the amount of (S)-mandelic
aqueous-organic solvent two-phase systems consisting amide formed from benzaldehyde and KCN by co-ex-
of 50 mM sodium citrate buffer (pH 5) and diisopro- pressing the (S)-oxynitrilase with the C-terminally de-
pylether (DIPE) (70%). The cells were suspended in leted nitrilase variant (Del-C60) encoded on plasmid
the aqueous phase to an optical density (OD546nm) of pCK105.[23] The corresponding whole-cell catalyst E.
1.8 and cyanide was added to the organic phase from coli JM109ACTHNUTRGNEU(GN pCK105)(pJOE5361.1) indeed converted
an HCN stock solution in DIPE. The reactions were benzaldehyde plus cyanide to mandelic amide and
started by the addition of 25–200 mM benzaldehyde mandelic acid in a ratio of about 9:1 (Figure 3). The
(from a 2M stock solution in DIPE). The observed analysis of a sample taken after 45 min by chiral
reaction rates were significantly slower than those ob- HPLC suggested that only (S)-mandelic amide was
served in the purely aqueous systems. Nevertheless, present and no indications for (R)-mandelic amide
all reactions were completed in less than 7.5 h. The were found. In order to analyze the accuracy of mea-
enantiopurity of the (S)-mandelic acid was in all reac- surement, various defined amounts of (R)-mandelic
tions >99% (Table 2, entries D–F).
amide were added to the samples. These experiments
demonstrated that (S)-mandelic amide was formed
with >95% ee.
Increased Formation of Mandelic Amide by using a
C-Terminally Truncated Nitrilase Variant
Discussion
The nitrilase from P. fluorescens EBC191 formed
from (S)-mandelonitrile approximately the same The results of the present study demonstrated that it
amounts of (S)-mandelic acid and (S)-mandelic is possible to simultaneously express a plant-derived
amide.[6,16] Recently, several mutants of the nitrilase oxynitrilase and a bacterial nitrilase in E. coli and
from P. fluorescens EBC191 have been generated and that these recombinant “bienzymatic whole-cell cata-
analyzed for amide formation and enantioselectivity lysts” are efficient for the synthesis of (S)-mandelic
and it was found that C-terminally truncated mutants acid and (S)-mandelic amide. The enantioselectivity
Adv. Synth. Catal. 2009, 351, 1531 – 1538
ꢂ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
1535

Olga Sosedov et al.
FULL PAPERS
of the studied reactions presumably strongly depends
The bienzymatic process demonstrated in the pres-
on the relative proportions of the oxynitrilase and ni- ent study should have a wide range of possible appli-
trilase (or nitrile hydratase) activities which are pres- cations because both enzymes, the (S)-oxynitrilase
ent in the respective biocatalysts. It is expected that from cassava as well as the nitrilase from P. fluores-
the highly enantioselective formation of (S)-mandelic cens EBC191, convert a wide range of aliphatic and
acid and/or (S)-mandelic amide will require a surplus aromatic substrates. Furthermore, for both enzymes
of nitrilase activity in comparison to the oxynitrilase already strategies for genetic modifications have been
activities. This should enable the immediate and described which, in principle, allow further improve-
quantitative conversion of the enzymatically formed ments with respect to substrate specifity, activity, and
(S)-mandelonitrile by the nitrilase and result in the ir- enantioselectivity.[23,27] We are currently attempting to
reversible fixation of the stereochemistry at the rele- obtain nitrilase mutants which either form quantita-
vant carbon atom, because the chemical racemization tively (S)-mandelic amide or (S)-mandelic acid from
of mandelonitrile is suppressed. Obviously, this was (S)-mandelonitrile in order to construct “bienzymatic
achieved in the present study, because in all experi- catalysts” which convert benzaldehyde and cyanide to
ments only marginal amounts of mandelonitrile were stoichiometric amounts of (S)-mandelic amide or (S)-
observed. This indicated that in these cells the nitri- mandelic acid.
lase activity was much higher than the oxynitrilase ac-
tivity. This was confirmed by the comparison of the
reaction rates of the whole cell catalysts with benzal- Conclusions
dehyde plus cyanide or mandelonitrile, which also
demonstrated significantly higher nitrilase activities We have succeeded in developing an efficient re-
than oxynitrilase activities. The comparably lower ex- combinant whole-cell catalyst in which a plant-de-
pression of the oxynitrilase activity could have been rived and a bacterial enzyme were combined in order
caused either by problems in the transcription/transla- to perform a commercially relevant biotransformation
tion of the plant-derived oxynitrilase gene or by the reaction. We have shown that the double clone cells
fact that the oxynitrilase gene was cloned on a are efficient catalysts for synthesizing (S)-mandelic
pACYC184-derived plasmid which presumably has a acid and (S)-mandelic amide. The catalytic efficiency
slightly lower copy number in LB-grown cells of E. of the double clone is better in aqueous buffer than in
coli compared to the pBR322-derived plasmid carry- an aqueous-organic biphasic system but, in the
ing the nitrilase gene.[25]
former, pH control is rapidly lost and the resulting
The recombinant E. coli clones that were generated pH shift towards alkaline conditions reduces the ap-
in the present study appear to be better suited for the plicability of the bienzymatic procedure. The biphasic
intended biotransformation than the previously con- system, in contrast, although slower, is more suitable
structed recombinant clones of Pichia pastoris al- in reactions using high concentrations of reactants.
though they are less acid-tolerant.[17] This was mainly The utilization of nitrilase variants which produce
due to the higher specific activities that were obtained large amounts of amides might open up a more gener-
for the E. coli strains compared to the derivatives of al way for the enantioselective synthesis of a-substi-
P. pastoris. Thus, it was found that the nitrilase activi- tuted amides, which is currently hampered by the gen-
ties were more than 7-times higher and the oxynitri- erally observed limited enantioselectivity of nitrile hy-
lase activities more than 30-times higher in the re- dratases. We expect that this and similar bienzymatic
combinant E. coli strains than in the Pichia clones. cascades will be promising catalysts for the synthesis
This is presumably caused by the much higher copy of enantiopure a-hydroxy carboxylic acids and a-hy-
number of the plasmid encoded nitrilase and oxynitri- droxy carboxylic amides.
lase genes in the E. coli strains.
The wild-type nitrilase converted (S)-mandeloni-
trile to almost equimolar concentrations of (S)-man-
delic acid and (S)-mandelic amide. This resulted in
combination with the availability of mutant variants
with an increased tendency to convert mandelonitrile
Experimental Section
to mandelic amide in the surprising observation that a
combination of an oxynitrilase and a nitrilase could
be used for the enantioselective synthesis of (S)-man-
delic amide. This might open up a more general way
for the enantioselective synthesis of alpha-substituted
amides, which is currently hampered by the generally
observed limited enantioselectivity of nitrile hydrat-
Chemicals
The chemicals used were obtained from Fluka (Neu-Ulm,
Germany), Lancaster (Mꢀhlheim/Main, Germany), Merck
(Darmstadt, Germany), Serva (Heidelberg), Bayer (Lever-
kusen, Germany), and Sigma–Aldrich (Mꢀnchen, Germa-
ny). Restriction enzymes and other reagents for molecular
biology were supplied by Roche Diagnostics (Mannheim,
Germany) and NEB (Frankfurt/Main, Germany).
ACHTUNGTRENNUNG
ases.[25]
1536
asc.wiley-vch.de
ꢂ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2009, 351, 1531 – 1538

Construction of Recombinant Escherichia coli Catalysts
FULL PAPERS
60 min at 48C. Protein was determined by the method of
Bacterial Strains, Plasmids, and Culture Conditions
Bradford[30] using bovine serum albumin as a standard.
Escherichia coli JM 109 was used for all cloning experi-
ments. The strain was routinely cultured in Luria-Bertani
(LB) medium (plus 100 mgmLÀ1 ampicillin and/or
20 mgmLÀ1 chloramphenicol). Plasmid pIK9 has been de-
scribed before.[15] E. coli M15(pQE4-MeHNLwt)[18] was
kindly provided by Prof. H. Wajant (Wꢀrzburg, Germany).
Polyacrylamide Gel Electrophoresis
Sodium dodecyl sulfate-polyacrylamide gel electrophoresis
(SDS-PAGE) was performed by the method of Laemmli[31]
Gels were routinely stained with Coomassie blue. A molecu-
lar weight standard (Kaleidoskop, Bio-Rad, Richmond, CA,
USA) and a commercially available sample of the oxynitri-
lase from M. esculenta (Fluka, Buchs, Switzerland) were
used as references.
DNA Manipulation Techniques
Digestion of DNA with restriction endonucleases (MBI Fer-
mentas, St. Leon-Rot, Germany), electrophoresis, and liga-
tion with T4 DNA ligase (MBI Fermentas) were performed
according to standard procedures.[28] Transformation of E.
coli was done by the method of Chung et al.[29]
Enzyme Assays
One unit of enzyme activity was defined as the amount of
enzyme that converts 1 mmol of substrate per min.
The nitrilase activity of resting cells was determined in re-
action mixtures (1.5–2 mL each) containing 50 mM sodium
citrate buffer (pH 5.0), 10 mM mandelonitrile and an appro-
priate amount of cells. All nitrile stock solutions (200 mM–
1 M) were prepared in methanol. The reaction mixtures
were incubated at 308C. After different time intervals, sam-
ples (50–100 mL each) were taken and the reactions were
stopped by the addition of 1 M HCl (5–10 mL). The samples
were centrifuged in an Eppendorf centrifuge at 15,000 g for
10 min and the supernatants were analyzed using high-pres-
sure liquid chromatography (HPLC).
Cloning of the Oxynitrilase Gene from Manihot
esculenta in E. coli
The gene encoding the oxynitrilase from Manihot esculenta
(MeHNL) was amplified by PCR from plasmid pQE4-
MeHNLwt[18] using the primer pair s4299 (AAAAAACATATGG
TAACTGCACATTTTGTT) and s4300 (AAAAAAAGCTTAAGCA
TATGCATCAGCCAC). The PCR product was subsequently cut
with NdeI and HindIII and the DNA fragment obtained li-
gated into the vector plasmid pAW229[19] which was previ-
ously cut with the same restriction enzymes. This resulted in
the formation of plasmid pJOE5356.1 which contained the
5’-part of the gene encoding the MeHNL (because there
was a HindIII-cleavage site within the PCR product). Subse-
quently, the PCR product obtained using the primer pair
s4299 and s4300 and plasmid pJOE5356.1 were cleaved with
HindIII. The insertion of the HindIII-fragment obtained
from the PCR-product into the HindIII-cut plasmid
pJOE5356.1 resulted in the formation of plasmid pJOE5361
which contained the complete MeHNL gene under the con-
trol of a rhamnose-inducible promoter.
Analytical Methods
Mandelonitrile, mandelic amide, and mandelic acid were an-
alyzed by HPLC (HPLC Millenium Chromatography Man-
ager 3.2, equipped with a diode array detector 996 and
HPLC pump M510; Waters Associates, Milford, MA, USA).
For the achiral analysis, a reversed-phase column (125 by
4 mm [internal diameter]; Trentec, Gerlingen, Germany)
filled with 5-mm diameter particles of Lichrospher RP8 (E.
Merck AG, Darmstadt, Germany) was used to identify indi-
vidual compounds which were detected spectrophotometri-
cally at 210 nm. The solvent system consisted of methanol
(40% v/v) and H3PO4 (0.3% v/v) in H2O. Separation of the
enantiomers of mandelic acid and mandelic amide was ach-
ieved on a Chiral-HSA column (ChromTech AB, Hꢁgersten,
Sweden). The mobile phases consisted of sodium phosphate
buffer (10 mM, pH 7.0) plus 4.5% (v/v) acetonitrile, respec-
tively.
Expression of the Oxynitrilase and Nitrilase in E. coli
For the induction of the enzymes in the recombinant E. coli
strains, precultures (10 mL each) were grown in LB-medium
with ampicillin (100 mgmLÀ1) or ampicillin (100 mgmLÀ1)
plus chloramphenicol (20 mgmLÀ1). From these precultures
Erlenmeyer flasks (100–500 mL) with the same composition
were inoculated to an initial optical density (OD546nm) of
about 0.02. The cultures were cultivated at 308C on a orbital
shaker (100 rpm) until they reached an optical density of
about 0.3. Then, 0.2% (w/v) l-rhamnose was added and the
bacterial cultures further incubated under the same condi-
tions as before. The cells were usually harvested about 12 h
later.
Acknowledgements
The presented work was supported by the DFG (partially in
the framework of the CERC3 program) and the COST action
CM0701.
Preparation of Cell-Free Extracts
The bacterial cultures were harvested by centrifugation
(9500 g, 15 min, 48C), washed and resuspended in Na/K References
phosphate buffer (50 mM, pH 7.5) or Tris/HCl (50 mM,
ˇ
pH 7.5), respectively. The cells were disrupted by using a
French Press (Aminco, Silver Springs, Md. USA) at 80 MPa.
Cell debris was removed by centrifugation at 100,000 g for
[1] L. Martinkovꢃ, V. Kren, Biocatal. Biotransform. 2002,
20, 73–93.
[2] H. Grçger, Adv. Synth. Catal. 2001, 343, 547–558.
Adv. Synth. Catal. 2009, 351, 1531 – 1538
ꢂ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
1537

Olga Sosedov et al.
FULL PAPERS
[3] F. Effenberger, S. Fçrster, H. Wajant, Curr. Opin. Bio-
technol. 2000, 11, 532–539.
[4] D. V. Johnson, A. A. Zabelinskaja-Mackova, H.
Griengl, Curr. Opin. Chem. Biol. 2000, 4, 103–109.
[5] T. Ziegler, B. Hçrsch, F. Effenberger, Synthesis 1990,
575–578.
[6] S. Rustler, A. Mꢀller, V. Windeisen, A. Chmura, B. Fer-
nandes, C. Kiziak, A. Stolz, Enzyme Microb. Technol.
2007, 40, 598–606.
[16] C. Mateo, A. Chmura, S. Rustler, F. van Rantwijk, A.
Stolz, R. A. Sheldon, Tetrahedron: Asymmetry 2006,
17, 320–323.
[17] S. Rustler, H. Motejadded, J. Altenbuchner, A. Stolz,
Appl. Microbiol. Biotechnol. 2008, 80, 87–97.
[18] H. Wajant, K. Pfizenmaier, J. Biol. Chem. 1996, 271,
25830–25834.
[19] B. Wilms, A. Wiese, C. Syldatk, R. Mattes, J. Alten-
buchner, J. Biotechnol. 2001, 86, 19–30.
[20] H. Wajant, S. Fçrster, H. Bçttinger, F. Effenberger, K.
[7] A. Banerjee, P. Kaul, U. C. Banerjee, Arch. Microbiol.
Pfizenmaier, Plant Sci. 1995, 108, 1–11.
[21] F. Effenberger, T. Ziegler, S. Fçrster, Angew. Chem.
1987, 99, 491–492.
2006, 184, 407–418.
[8] P. Kaul, A. Banerjee, S. Mayilraj, U. C. Banerjee, Tetra-
hedron: Asymmetry 2004, 15, 207–211.
[22] D. Costes, E. Wehtje, P. Adlercreutz, Enzyme Microb.
[9] P. Kaul, A. Stolz, U. C. Banerjee, Adv. Synth. Catal.
Technol. 1999, 25, 384–391.
[23] C. Kiziak, J. Klein, A. Stolz, Protein Eng. Des. Sel.
2007, 20, 385–396.
2007, 349, 2167–2176.
[10] D. E. Robertson, J. A. Chaplin, G. DeSantis, M. Podar,
M. Madden, E. Chi, T. Richardson, A. Milan, M.
Miller, D. P. Weiner, K. Wong, J. McQuaid, B. Farwell,
L. A. Preston, X. Tan, M. A. Snead, M. Keller, E.
Mathur, P. L. Kretz, M. J. Burk, J. M. Short, Appl. En-
viron. Microbiol. 2004, 70, 2429–2436.
[11] a) K. Yamamoto, K. Oishi, I. Fujimatsu, K.-I. Komatsu,
Appl. Environ. Microbiol. 1991, 57, 3028–3032; b) K.
Yamamoto, I. Fujimatsu, K.-I. Komatsu, J. Ferment.
Bioeng. 1992, 73, 425–430.
[24] R. Moser, Diplomarbeit, Universitꢁt Stuttgart, 1996.
[25] T. Atlung, B. B. Christensen, F. G. Hansen, Plasmid
1999, 41, 110–119.
[26] a) M. Payne, S. Wu, R. Fallon, G. Tudor, B. Stieglitz, I.
Turner, M. Nelson, Biochemistry 1997, 36, 5447–5454;
b) R. Bauer, H.-J. Knackmuss, A. Stolz, Appl. Micro-
biol. Biotechnol. 1998, 49, 89–95; c) M. Hensel, S.
Lutz-Wahl, L. Fischer, Tetrahedron: Asymmetry 2002,
13, 2629–2633.
[27] H. Bꢀhler, F. Effenberger, S. Fçrster, J. Roos, H.
Wajant, ChemBioChem 2003, 4, 211–216.
[28] J. Sambrook, E. F. Fritsch, T. Maniatis, Molecular clon-
ing: a laboratory manual, 2nd edn., Cold Spring Harbor
Press, Cold Spring Harbor, N.Y., 1989.
[12] M. Bauer, H. Griengl, W. Steiner, Enzyme Microb.
Technol. 1999, 24, 514–522.
[13] A. Banerjee, R. Sharma, U. C. Banerjee, Appl. Micro-
biol. Biotechnol. 2002, 60, 33–44.
[14] U. Heinemann, D. Engels, S. Bꢀrger, C. Kiziak, R.
Mattes, A. Stolz, Appl. Environ. Microbiol. 2003, 69,
4359–4366.
[15] C. Kiziak, D. Conradt, A. Stolz, R. Mattes, J. Klein, Mi-
crobiology 2005, 151, 3639–3648.
[29] C. T. Chung, S. L. Niemela, R. H. Miller, Proc. Natl.
Acad. Sci. USA 1989, 86, 2172–2175.
[30] M. M. Bradford, Anal. Biochem. 1976, 72, 248–254.
[31] U. K. Laemmli, Nature 1970, 227, 680–685.
1538
asc.wiley-vch.de
ꢂ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2009, 351, 1531 – 1538