Resolution of (RS)-phenylglycinonitrile by penicillin acylase-catalyzed acylation in aqueous medium

Article abstract of DOI:10.1016/S0957-4166(03)00523-8

A new strategy for the biocatalytic resolution of (R,S)-phenylglycinonitrile, a crucial intermediate in the antibiotic industry, has been developed. While former techniques exploit nitrilases or combinations of nitrile hydratases and amidases, manipulating with nitrile functionality, the current approach is based on a highly efficient and enantioselective acylation of the α-amino group with phenylacetic acid catalyzed by a well known enzyme, penicillin acylase from E. coli, in slightly acidic aqueous medium. It is shown that since the condensation product is poorly soluble, removal of (S)-phenylglycinonitrile from the reaction sphere is almost complete and irreversible, favoring kinetics of the process and making high conversion possible. The proposed approach is characterized by high space-time yield and extends the scope of enzymatic synthesis in aqueous medium.

Full text of DOI:10.1016/S0957-4166(03)00523-8

TETRAHEDRON:
ASYMMETRY
Pergamon
Tetrahedron: Asymmetry 14 (2003) 2613–2617
Resolution of (RS)-phenylglycinonitrile by penicillin
acylase-catalyzed acylation in aqueous medium
a
b
b
a,
Ghermes G. Chilov, Harold M. Moody, Wilhelmus H. J. Boesten and Vytas K. Svedas *
&
a
Department of Bioengineering and Bioinformatics, Belozersky Institute of Physicochemical Biology,
Lomonosov Moscow State University, Lenin Hills, 119992 Moscow, Russia
DSM Research, LS-ASC, PO Box 18, NL-6160 MD Geleen, The Netherlands
b
Received 28 May 2003; accepted 26 June 2003
Abstract—A new strategy for the biocatalytic resolution of (R,S)-phenylglycinonitrile, a crucial intermediate in the antibiotic
industry, has been developed. While former techniques exploit nitrilases or combinations of nitrile hydratases and amidases,
manipulating with nitrile functionality, the current approach is based on a highly efficient and enantioselective acylation of the
a-amino group with phenylacetic acid catalyzed by a well known enzyme, penicillin acylase from E. coli, in slightly acidic aqueous
medium. It is shown that since the condensation product is poorly soluble, removal of (S)-phenylglycinonitrile from the reaction
sphere is almost complete and irreversible, favoring kinetics of the process and making high conversion possible. The proposed
approach is characterized by high space-time yield and extends the scope of enzymatic synthesis in aqueous medium.
©
2003 Elsevier Ltd. All rights reserved.
6
1. Introduction
are prone to loose activity when recycling, and purified
7
enzymes are very capricious. In addition only low
a-Aminonitriles are crucial intermediates in the produc-
tion of the corresponding amides and acids. Within the
framework of numerous applications of these com-
pounds in fine chemistry, phenylglycinonitrile is of spe-
cial interest since its derivative, (R)-phenylglycine
amide, serves as an acyl chain donor in a biocatalytic
concentrations of substrates could be converted by
currently available techniques: the best examples show
productivity (the so called space-time yield) of 0.1–0.3
3
kg of product per m of reactor space per hour, which
4,6
is quite low for industrial purposes. With respect to
phenylglycinonitrile there have been a number of stud-
1
ampicillin and cephalexin syntheses. However, each
ies on nitrilase and hydratase/amidase catalyzed resolu-
particular application requires only one enantiomer and
hence a problem of chiral resolution arises since a-
aminonitriles are available via Strecker reaction only in
a racemic form. The current strategies tackle this prob-
lem by using nitrilases or combining nitrile hydratases
and amidases. The former, nitrilases, perform enan-
tioselective conversion of a nitrile directly into the
carboxylic acid while the latter imply (as a rule) non-
specific hydration of a nitrile to an amide with subse-
6,8,9
tions.
It has been shown, that the latter approach
works better with respect to yields and enantioselectiv-
ity. However, productivity of the process was still mod-
6
erate even at high catalyst loading.
Herein we report an alternative route of chiral phenyl-
glycinonitrile resolution based on the enzymatic recog-
nition of a-amino- rather than nitrile-functionality.
Namely, we show how the amino group of the nitrile
could be acylated enantioselectively and with nearly
quantitative yield. For this purpose penicillin acylase
from E. coli, an enzyme, well known for its affinity
towards (S)-phenylglycine derivatives as external nucleo-
quent amidase-catalyzed hydrolysis of
a
single
2
,3
enantiomer. Although a variety of strategies to culti-
vate strains, harboring activity towards a given nitrile,
4
,5
are available, industrial applications of biocatalytic
nitrile conversions are still handicapped due to the lack
of a properly formulated biocatalyst: immobilized cells
10–12
philes,
was used with phenylacetic acid as an
acylating agent. The proposed approach allows the
conversion of high substrate concentrations and enables
easy separation of non-reactive (R)-phenylglycinonitrile
from the acylated (S)-form by simple filtration due to
very low solubility of the latter.
*
0
Corresponding author. Tel./fax: +7 095 939 2355; e-mail: vytas@
belozersky.msu.ru
957-4166/$ - see front matter © 2003 Elsevier Ltd. All rights reserved.
doi:10.1016/S0957-4166(03)00523-8

2
614
G. G. Chilo6 et al. / Tetrahedron: Asymmetry 14 (2003) 2613–2617
Scheme 1. The general scheme of a-phenylglycinonitrile resolution by penicillin acylase-catalyzed direct condensation with
phenylacetic acid in aqueous medium.
2
. Results
via a retro-Strecker reaction at neutral and alkaline pH.
As lowering the temperature contributes to the stability
of a-aminonitriles, we have found that at 5°C and pH 5
decomposition of phenylglycinonitrile was negligible
2
.1. Effectiveness and enantioselectivity of acylation
(
1
monomolecular decomposition rate constant was 3.3×
We have found that direct condensation of racemic
phenylglycinonitrile (Scheme 1) with phenylacetic acid
catalyzed by penicillin acylase from E. coli in aqueous
solution was highly enantioselective and nearly stoi-
chiometric. Thus, condensation in the concentrated
solution of phenylacetic acid 0.2 mol/kg (mol per kg of
solution) and racemic 1 (0.4 mol/kg) at pH 5 gave
−
5
−1
0
s ).
Another trap to be aware of is the spontaneous race-
mization of the remaining (R)-form. However in an
independent experiment [with isolated (R)-form and
(
R)-phenylglycinonitrile tartarate salt] we have shown
that racemization was even slower than spontaneous
degradation: in 20 h at pH 5 and 5°C the value of e.e.
dropped from 0.99 to 0.98 which is insufficient for
acylation reaction having typical time 10 h.
(
R)-phenylglycinonitrile with 99% e.e. at a conversion
of the (S)-form higher than 49%. From the kinetic
progress curve we have estimated E to be >500. Finally,
the reaction mixture presented almost pure (R)-phenyl-
glycinonitrile solution: it contained ꢀ1 mM of phenyl-
acetic acid (conversion >99.5%), less than 1 mM of the
(
S)-enantiomer and ꢀ7 mM of phenylacetylated (S)-
3. Discussion
phenylglycinonitrile (equilibrium solubility of the com-
pound) while the rest of 3 was in the precipitate phase.
Condensation proceeded at a constant rate without
inhibition effects (Fig. 1) up to the conversion degree
dictated by thermodynamic equilibrium. No loss of
enzyme activity was detected in the course of the
reaction.
3.1. The principle of enzymatic resolution
Acylation of aminocompounds in aqueous medium
could imply two options: acylation by a substituted
2.2. pH and temperature optimisation of the reaction
As the reaction proceeds in an aqueous medium where
both components, the acyl donor and the aminocompo-
nent, are able to ionize, there is a pH optimum of
synthetic efficiency. In the case of phenylglycinonitrile
condensation with phenylacetic acid the thermody-
namic pH optimum is 4.45 (half sum of pK 4.3 of 2 and
pK 4.6 of 1 at 5°C). In practice, condensation was
nearly equally favorable over the pH range 4–5, while
at pH >5 and at pH <4 ionization of acid and amino-
nitrile correspondingly reduces the synthetic efficiency.
In our experiments we have chosen pH 5 in order to
gain in enzyme activity.
Figure 1. Kinetic progress curve of penicillin acylase-cata-
lyzed direct condensation: (RS)-phenylglycinonitrile (ꢀ),
phenylacetic acid (ꢁ), acylation product 3 (ꢂ) (left axis); e.e.
of (R)-phenylglycinonitrile (ꢃ) (right axis).
Another factor, which has influenced the choice of
appropriate reaction conditions was the stability of
reagents. Aminoacid nitriles are prone to decompose

G. G. Chilo6 et al. / Tetrahedron: Asymmetry 14 (2003) 2613–2617
2615
carboxyl derivative (an amide or an ester) known as
acyl transfer, or direct condensation with a corre-
sponding carboxylic acid. The principal difference
between these processes is the following: while in the
condensation reaction chemical equilibrium is attained
gradually (e.g. the product is accumulated
monotonously until the equilibrium is reached), in the
former product accumulation may exceed the final
equilibrium value if the acyl transfer itself is faster
than the hydrolysis of the acyl donor and the acyl
transfer product formed. The latter example describes
the so-called kinetically-controlled synthesis, which is
the common way to couple thermodynamically unfa-
vorable (synthetic) reactions to the consumption of
energetically-rich compounds (be it activated acyl
donor or ATP and acetyl-CoA as in biological sys-
tems). As for the enzymatic reactions in vitro, there
are examples of successful amine resolution by enan-
tioselective acyl transfer in aqueous medium, catalyzed
Experimental conversion degrees and the values of e.e.
perfectly correlated with thermodynamic predictions.
This meant that no kinetic obstacles (such as enzyme
non-specificity, enzyme inactivation, strong inhibition
effects, etc.) took place and true chemical equilibrium
was attained, providing the best possible characteris-
tics of enantiomer resolution. Note that if the product
of condensation was soluble and hence the extra driv-
ing force of the reaction disappeared, conversion of
the amino-compound would be sufficiently lower thus
leading to lower e.e.
3.3. Kinetic features
Some kinetic features of the reaction should be men-
tioned. First of all we should point out that penicillin
acylase from E. coli appeared to be stable and active
under the described conditions. Surprisingly high
activity of the enzyme in acidic medium has been
exploited earlier for effective hydrolysis of benzylpeni-
cillin and benzyldeacetoxycephalosporin hydrolysis in
13
by penicillin acylase from A. faecalis.
a
two-phase ‘aqueous1 solution–water immiscible
The choice of acylating agent for resolution purposes
4,20
organic solvent’ medium.
biocatalytic applications of penicillin acylase in acidic
media.
This new example widens
was predefined by the high affinity of penicillin acylase
1
1–13
towards phenylacetic acid derivatives.
Since we
considered reaction at pH 5 (enzyme activity permitted
this) we might hope that 2 (pK 4.3), namely, its proto-
nated form, could serve as an effective acylating agent.
These findings demonstrate the feasibility of resolution
of the aminoacid nitriles by enzymatic direct conden-
sation in aqueous medium.
As seen from the kinetic progress curve (Fig. 1), sub-
strate consumption evolves at a constant rate until the
equilibrium conversion is reached. This means that the
reaction proceeds under enzyme saturation by sub-
strate and is not complicated by product inhibition
(
since the product is poorly soluble). This feature
3.2. Thermodynamic considerations
favors scalability of the process towards higher con-
centrations of the substrates converted.
Coincidence of thermodynamic pH optimum with pKs
of 2 and 1 made condensation reaction not only kinet-
ically, but also thermodynamically feasible. In con-
trast, peptide bond synthesis in an aqueous medium
seems to work only via acyl transfer: due to the big
difference in pK values of carboxylic (ꢀ2) and amino
groups (ꢀ9–10) of amino acids direct condensation is
thermodynamically unfavorable. However we should
mention here that while favorable pKs of reagents
make the reaction feasible, precipitation of the con-
densation product makes it practically irreversible,
that is, stoichiometric. We have found that the solubil-
ity of 3 was 6.8 mM at 5°C, hence this example of
enantiomer resolution could nicely represent an ideal
Note also that as far as the condensation reaction is
seen at a pH close to the half-sum of reagents pKs,
the pH shift during the reaction is insignificant and
hence the reaction can be conducted per se in a simple
batch mode without pH control and buffering.
Finally, we should comment on the substrate specific-
ity. High enantioselectivity of penicillin acylase from
E. coli towards (S)-phenylglycine and other (S)-amino
acids as external nucleophiles in acyl transfer was
11,13
known from our previous experience.
Herein we
have shown that enzyme enantioselectivity towards
(S)-phenylglycinonitrile (E >500) is comparable with
selectivity to other amino acids and distinctly exceeds
that for amino acid esters and aliphatic amines.
1
5
of precipitation-driven catalysis.
The last notion on the thermodynamics of aminonitrile
resolution by direct condensation with carboxylic acid
3
.4. Prospects
14,16
concerns the ‘strength’ of amide bond formed.
We
have determined that apparent equilibrium constant
characterizing synthesis of amide bond in phenylacetyl-
ated (S)-phenylglycinonitrile (expressed as the concen-
The proposed approach of chiral resolution provides a
high space-time yield ꢀ3 kg/m /h at a moderate bio-
3
catalyst charging used in our experiments. This pro-
ductivity is at least an order of magnitude higher than
−
1
tration ratio of the 3 with 1 and 2) was 56 M . We
6,8
should mention that this value is characteristic for
using nitrile hydratases and amidases. It should be
emphasized that penicillin acylase is a very stable
enzyme and a number of its immobilized forms (using
different carriers, prepared by cross-linking enzyme
crystals or enzyme aggregates) are easily available.
14,17,18,21
amide bonds formed by amines with low pK,
and in general can be considered as a quite moderate:
for aliphatic amides and peptides the synthetic con-
16
stant can be an order of magnitude larger.

2
616
G. G. Chilo6 et al. / Tetrahedron: Asymmetry 14 (2003) 2613–2617
4. Conclusion
5.2. Enzymatic condensation of (RS)-aminonitrile with
phenylacetic acid
The proposed approach of a-aminonitrile resolution
principally differs from the existing techniques: while
the latter are targeted at selective hydrolysis of a
nitrile group, converting one enantiomer into an
amide or carboxylic acid, the former implies enzy-
matic acylation of a-amino group of the (S)-form.
We have shown that penicillin acylase from E. coli
was able to catalyze direct condensation of phenylgly-
cinonitrile with phenylacetic acid in acidic medium
a-Phenylglycinonitrile hydrochloride (2.65 mmol) and
phenylacetic acid (1.325 mmol) were dissolved in
water (6 mL) at 5°C and pH was adjusted to 5.0 with
1
0N NaOH. Reaction was carried out in the presence
of 10 mM PA-E. coli under permanent stirring in a
thermostatted cell of a pH-stat (Metrohm Titrino 719
S) charged with 2M aqueous HCl solution. Reaction
was accompanied by precipitation of the condensation
product.
(
pH 5) and the reaction was exclusively enantioselec-
tive and effective, which allowed nearly stoichiometric
acylation (with respect to both components). The
quantitative conversion of reagents was due to the
highly efficient precipitation of the acylation product,
making condensation essentially irreversible. In addi-
tion, precipitation of the product has eliminated inhi-
bition effects, so that the reaction proceeded at a
constant rate until the complete conversion of the
reagents. The proposed approach is scalable and
allows for the conversion of high substrate concentra-
tions. Further implications of precipitation-driven a-
Progress curves of the enzymatic acylation were fol-
lowed by HPLC analysis on a reversed phase column
(
phenylacetic acid, aminonitrile and their condensa-
tion product were detected) and separately on a chiral
column to check for e.e. of the remaining aminoni-
trile. For analysis on a reversed phase column 20 mL
of heterogeneous reaction mixture were added to 0.98
mL of eluent in order to dissolve reactants and stop
the reaction; 20 mL of obtained solution were added
to 0.98 mL of eluent and the sample was injected
into HPLC system. For chiral analysis 20 mL of het-
erogeneous reaction mixture were filtrated using
microfiltration kit (membrane pore size 0.2 mm by
Biochrom, Moscow, Russia) in combination with
Eppendorf centrifuge 5415 D (5 s at 13.200 rpm). 10
mL of the filtrate were dissolved in 0.5 mL of eluent
aminonitrile
resolution
by
enzymatic
direct
condensation are under investigation.
(
for chiral analysis); 20 mL of obtained solution were
5. Experimental
diluted with 0.38 mL of eluent and subjected to chiral
analysis.
Solutions of penicillin acylase from E. coli were
obtained from DSM Anti-Infectives, Delft, The
Netherlands. The concentration of penicillin acylase
Finally, the condensation product (N-phenylacetylated
phenylglycinonitrile) was filtrated on the pore glass
filter, washed for several times with water and dried
19
active sites was determined as described earlier.
RS)-Phenylglycinonitrile hydrochloride and (R)-
(
in an exsiccator over concentrated H SO overnight.
2 4
Enantiomeric purity of N-phenylacetyl-(S)-phenylgly-
cinonitrile has not been checked on a chiral column
but we could suppose it was high enough (>99%)
standing on mass balance data: conversion of acyl
donor was ꢀ99.5% and e.e. of residual (R)-phenyl-
glycinonitrile was ꢀ99% as if only the (S)-form was
acylated.
phenylglycinonitrile tartarate salt were products of
DSM. Phenylacetic acid was obtained from Aldrich.
Eluent components were purchased at Reakhim
(
Moscow, Russia) and at Kriochrom (St. Petersburg,
Russia). In all experiments MilliQ water was used.
5.1. HPLC analysis
The remaining (R)-phenylglycinonitrile was isolated as
follows: in the filtrated mother liquor pH was
brought to 7 and aminonitrile was extracted into 20
mL of ethyl acetate at 5°C. Then back extraction
with 10 mL of 0.3 M HCl followed. Hydrochloride of
Concentrations of the reactants were determined by
HPLC using a Waters M6000 pump, Chrompack
Nucleosil 100 C-18 column (150 mm×4.6 mm, 5 mm)
and LKB 2138 uvicord S detector at 208 nm with an
eluent containing 5 mM phosphate, acetonitrile 35%
v/v, 0.25 g/L of sodium dodecylsulfate at pH 3.0. The
flow rate was 0.8 mL/min. Retention times (in min):
phenylacetic acid (4.7), phenylglycinonitrile (6.5), N-
phenylacetyl-phenylglycinonitrile (18).
(
R)-phenylglycinonitrile was isolated from water
phase by vacuum evaporation at 50°C and dried in a
dessiccator over concentrated H SO overnight.
2
4
5
.2.1. (R)-Phenylglycinonitrile. (R)-Phenylglycinonitrile
hydrochloride was isolated from the reaction mixture as
described in 4.2. Yield 0.106 g (50%); 99% e.e.; mp
Chiral analysis was performed by HPLC using
25
D
1
1
45–146°C (dec.); [h] =+51.0 (c 1, 1N HCl); H NMR
®
Waters M6000 pump, Crownpack CR(+) column
(
3
300 MHz, DMSO+CCl ): l 5.87 (s, 1H, CH), l 7.45 (m,
4
(
150 mm×4 mm, 5 mm) and LKB 2138 uvicord S
H, Ph), l 7.75 (m, 2H, Ph), l 9.85 (br. s, 3H, NH ). MS
3
detector at 208 nm with an eluent containing 0.01 M
aqueous solution of perchloric acid. The flow rate
was 0.5 mL/min. Retention times, in min: (S)-phenyl-
glycinonitrile (13), (R)-phenylglycinonitrile (17).
m/z: 167 (5, M−H), 132 (85, M−HCl), 131 (100, M−
HCl−H), 116 (26, PhCHCN), 106 (28, M−HCl−CN), 105
(15, M−HCl−HCN), 104 (40, M−HCl−H−HCN), 77 (47,
Ph), 55 (20, M−HCl−Ph), 51 (30), 36 (25, HCl).

G. G. Chilo6 et al. / Tetrahedron: Asymmetry 14 (2003) 2613–2617
2617
5
.2.2.
N-Phenylacetyl-(S)-phenylglycinonitrile.
N-
Stolz, A.; Sheldon, R. A. Org. Proc. Res. Dev. 2000, 4,
318–322.
Phenylacetyl-(S)-phenylglycinonitrile was isolated from
the reaction mixture as described in 4.2. Yield 0.3 g
7. Hjort, C. M.; Godtfredsen, S. E.; Emborg, C. J. Chem.
Tech. Biotechnol. 1990, 48, 217–226.
(
90%); >99% e.e. (estimated from mass balance, see 4.2.
25
for details); mp 118–119°C (dec.); [h] =+17.2 (c 1,
8. Layh, N.; Hirrlinger, B.; Stolz, A.; Knackmuss, H.-J.
Appl. Microbiol. Technol. 1997, 47, 668–674.
9. Wegman, M. A.; Heinemann, U.; Van Rantwijk, F.;
Stolz, A.; Sheldon, R. A. J. Mol. Catal. B: Enzym. 2001,
11, 249–253.
D
1
methanol); H NMR (300 MHz, DMSO+CCl ): l
4
3
.52s, 2H, CH ), l 6.05 (d, 1H, CH), l 7.2–7.5 (m, 10H,
2
Ph), l 9.20 (d, 1H, NH). MS m/z: 250 (50, M), 116 (35,
PhCHCN), 91 (100, PhCH ), 77 (12, Ph), 65 (20).
2
1
1
1
1
0. Didziapetris, R.; Drabnig, B.; Schellenberger, V.;
Jakubke, H.-D.; S
1. Didziapetris, R. J.; S
991, 50, 237–242.
2. Svedas, V. K.; Savchenko, M. V.; Beltser, A. I.; Guranda,
&
vedas, V. FEBS Lett. 1991, 287, 31–33.
Acknowledgements
&
vedas, V. K. Biomed. Biochim. Acta
1
&
Financial support by the Russian Foundation for Basic
Research (03-04-48472; 03-04-06307) and DSM
Research is gratefully acknowledged.
D. F.; Annals, N. Y. Acad. Sci. 1996, 799, 659–669.
3. Guranda, D. T.; van Langen, L. M.; van Rantwijk, F.;
Sheldon, R. A.; S
001, 12, 1645–1650.
4. Chilov, G.; Svedas, V. K. Can. J. Chem. 2002, 80,
99–707.
5. Ulijn, R. V.; Janssen, A. E.; Moore, B. D.; Halling, P. J.
Chem. Eur. J. 2001, 7, 2089–2098.
&
vedas, V. K. Tetrahedron: Asymmetry
2
1
1
&
References
6
1
2
3
4
. Bruggink, A.; Roos, E. C.; de Vroom, E. Org. Proc. Res.
Dev. 1998, 2, 128–133.
. Crosby, J.; Moilliet, J.; Parrat, J. S.; Turner, N. J. Chem.
1
1
6. Carpenter, F. H. J. Am. Chem. Soc. 1960, 82, 1111–1122.
7. Tewari, Y. B.; Goldberg, R. N. Biophys. Chem. 1988, 29,
Soc., Perkin Trans. 1 1994, 1679–1687.
245–252.
. Sugai, T.; Yamazaki, T.; Yokoyama, M.; Ohta, H.
Biosci. Biotechnol. Biochem. 1997, 61, 1419–1427.
. Ospiran, I.; Jarret, C.; Strauss, U.; Kroutil, W.; Orru, R.
V. A.; Felfer, U.; Willetts, A. J.; Faber, K. J. Mol. Catal.
B: Enzym. 1999, 6, 555–560.
. Stolz, A.; Trott, S.; Binder, M.; Bauer, R.; Hirrlinger, B.;
Layh, N.; Knackmuss, H.-J. J. Mol. Catal. B: Enzym.
998, 5, 137–141.
. Wegman, M. A.; Heinemann, U.; Van Rantwijk, F.;
1
1
2
2
8. Djachenko, E. D.; Kozlov, L. V.; Antonov, V. K. Bioorg.
Khim. (Rus) 1977, 3, 99–104.
9. Svedas, V.; Guranda, D.; van Langen, L.; van Rantwijk,
&
F.; Sheldon, R. FEBS Lett. 1997, 417, 414–418.
0. Diender, M. B.; Straathof, A. J.; van der Does, T.; Ras,
C.; Heijnen, J. J. Biotechnol. Bioeng. 2002, 78, 395–402.
5
&
1. Svedas, V. K.; Margolin, A. L.; Berezin, I. V. Enzyme
Microb. Technol. 1980, 2, 138–144.
1
6