Preparation of D-amino acids by enzymatic kinetic resolution using a mutant of penicillin-G acylase from E. coli

Article abstract of DOI:10.1016/j.tetasy.2005.12.023

We have demonstrated for the first time that d-glutamine (d-Gln) and d-glutamic acid (d-Glu) can be efficiently obtained in high ee (97% and 90%, respectively) by enzymatic kinetic resolution of d,l-Gln and d,l-Glu. This was achieved by enantioselective conversion of the l-enantiomers to their N-phenylacetyl derivatives in aqueous solution, using a mutant of penicillin-G acylase (PGA) from E. coli and phenylacetic acid methylester as the acyl donor. Kinetic modeling studies suggest that the high ee values obtained are both due to a strong enantiopreference for the l-amino acid in the deacylation step of the covalent enzyme intermediate, as well as to completeness of conversion that is transiently obtained as a result of the distinct preference of the mutant PGA for phenylacetic acid methylester over the N-phenylacetyl-l-amino acid product. For the other amino acids tested (Asn, Asp, and Ser), the highest ee values that were obtained for the remaining d-enantiomer are moderate (50-80%) because of lower enantioselectivity in the enzyme deacylation step and due to less complete conversion of the l-amino acid caused by competition for the active site between the acyl donor and the N-phenylacetyl-l-amino acid that is produced. The results demonstrate that the mutated PGA has great potential for the production of optically active D-amino acids by kinetic resolution.

Full text of DOI:10.1016/j.tetasy.2005.12.023

Tetrahedron:
Asymmetry
Tetrahedron: Asymmetry 17 (2006) 245–251
Preparation of D-amino acids by enzymatic kinetic resolution
using a mutant of penicillin-G acylase from E. coli
Chiara Carboni,a,b Hans G. T. Kierkels,a Lucia Gardossi,b Kamil Tamiola,c
Dick B. Janssenc and Peter J. L. M. Quaedfliega,*
aDSM Research, Life Sciences—Advanced Synthesis, Catalysis and Development, PO Box 18, 6160 MD Geleen, The Netherlands
b
`
Dipartimento di Scienze Farmaceutiche, Universita degli Studi di Trieste, Piazzale Europa 1, 34127 Trieste, Italy
cBiochemical Laboratory, GBB, University of Groningen, Nijenborgh 4, 9747 AG Groningen, The Netherlands
Received 21 November 2005; accepted 28 December 2005
Abstract—We have demonstrated for the first time that D-glutamine (D-Gln) and D-glutamic acid (D-Glu) can be efficiently obtained
in high ee (97% and 90%, respectively) by enzymatic kinetic resolution of D,L-Gln and D,L-Glu. This was achieved by enantioselec-
tive conversion of the L-enantiomers to their N-phenylacetyl derivatives in aqueous solution, using a mutant of penicillin-G acylase
(PGA) from E. coli and phenylacetic acid methylester as the acyl donor. Kinetic modeling studies suggest that the high ee values
obtained are both due to a strong enantiopreference for the L-amino acid in the deacylation step of the covalent enzyme inter-
mediate, as well as to completeness of conversion that is transiently obtained as a result of the distinct preference of the mutant
PGA for phenylacetic acid methylester over the N-phenylacetyl-L-amino acid product. For the other amino acids tested (Asn,
Asp, and Ser), the highest ee values that were obtained for the remaining D-enantiomer are moderate (50–80%) because of lower
enantioselectivity in the enzyme deacylation step and due to less complete conversion of the L-amino acid caused by competition
for the active site between the acyl donor and the N-phenylacetyl-L-amino acid that is produced. The results demonstrate that
the mutated PGA has great potential for the production of optically active D-amino acids by kinetic resolution.
ꢀ 2006 Published by Elsevier Ltd.
1. Introduction
enzymatic hydrolysis of the N-acylated racemates with
an L-amino acid acylase.9
Enantiopure D-amino acids are widely used as inter-
mediates in the production of semi-synthetic antibiotics,
agrochemicals, and novel pharmaceuticals,1–5 but their
industrial synthesis is not straightforward.6 Although
methods for the asymmetric synthesis of amino acids
have been described, these have never been turned into
economically viable processes. On the other hand, there
are several cost-efficient synthetic methods for the pro-
duction of racemic amino acids, which can be readily
combined with biocatalytic technologies to give the
enantiopure D-enantiomers. For instance, some D-amino
acids can be cost-effectively produced from the corre-
sponding D,L-hydantoins with the D-hydantoinase-
carbamoylase system7 or from D,L-amino acid amides
using an L-amidase,8 followed by chemical amide hydro-
lysis. D-Amino acids are also industrially accessible by
Penicillin-G acylase (PGA) is currently one of the acy-
lases most widely applied industrially, not only for the
production of semi-synthetic antibiotics7 but also for
the synthesis of enantiomerically pure L-amino acids
through enantioselective enzymatic hydrolysis of N-
phenylacetyl-D,L-amino acids in aqueous solution. If a
D-amino acid is the desired product, an additional chem-
ical hydrolysis step is required. Cole10,11 showed that
PGA can also be applied in the synthetic direction, while
Zmijewski et al.12 demonstrated that PGA is a suitable
catalyst for carrying out kinetically controlled acylations
in the resolution of racemates. This conversion is based
on the differences in efficiencies with which the covalent
acyl-enzyme intermediate, in which the nucleophilic ser-
ine is acylated, reacts with the enantiomers of an amino
acid. When such a kinetic phenylacetylation is per-
formed in aqueous solution with wild-type PGA from
E. coli (EC 3.5.1.11), the N-phenylacetyl-L-amino acids
can be obtained in high ee but the ee values obtained
for the remaining D-amino acids are low. This is mainly
*
Corresponding author. Tel.: +31 46 4761592; fax: +31 46 4767604;
e-mail: peter.quaedflieg@dsm.com
0957-4166/$ - see front matter ꢀ 2006 Published by Elsevier Ltd.
doi:10.1016/j.tetasy.2005.12.023

246
C. Carboni et al. / Tetrahedron: Asymmetry 17 (2006) 245–251
due to the fact that the acyl-enzyme intermediate may be
subject not only to deacylation by the nucleophilic
amino acid but also can be deacylated by water. Further-
more, hydrolysis of the produced N-phenylacetyl-L-
amino acids can also occur. Both these competing
processes can reduce the completeness of conversion of
the substrate enantiomer that must be acylated and
thereby reduce the ee that can be reached (see Fig. 1).
2. Results and discussion
The F24A mutant of PGA was applied to the resolution
of D,L-Gln, D,L-Glu, D,L-Asn, D,L-Asp, and D,L-Ser.
For these five amino acids no direct methods have been
published so far to obtain the D-enantiomers in one step
via L-selective acylation of the racemate or through
D-selective hydrolysis of the D,L-hydantoin. To test these
conversions, 1 M aqueous solutions of the racemates
were reacted with 1.3–1.8 equiv phenylacetic acid
methylester at pH = 9–9.6 in the presence of F24A
mutant PGA at 25 ꢁC. At regular time intervals, samples
were taken and the ee values of both the N-phenylacetyl-
L-amino acid products and of the remaining D-amino
acids, were determined by chiral HPLC. Conversions
and E values were subsequently calculated16 on the basis
of these experimentally determined ee values.
One solution to the hydrolysis problem is to perform the
enzymatic phenylacetylation in organic media, which we
demonstrated by the preparation of D-phenylglycine
and D-p-hydroxyphenylglycine methylesters.13 The
drawbacks of this method for industrial production
are the use of environmentally unfriendly organic sol-
vents on a large scale, the fact that the amino acids must
first be transformed into esters or amides to render them
soluble in organic solvent, and the generally lower
enantioselectivity of PGA in organic solvent than in
aqueous solution.14 A second possible solution is to
improve the kinetic parameters of the wild-type PGA by
site-directed mutagenesis. Recently, we studied a similar
PGA-catalyzed acylation reaction, namely the conden-
sation of D-phenylglycine amide with 6-aminopenicill-
anic acid to give the important semi-synthetic b-lactam
antibiotic ampicillin.15 When comparing the kinetic
properties of several mutants with those of wild-type
PGA, we found that the F24A mutant (with Phe changed
to Ala on position 24 of the b subunit) possesses a
threefold higher ratio between the rate of acyl transfer
to the nucleophile and the rate of acyl transfer to water,
as well as a twofold higher ratio between the maximum
concentration of ampicillin and the level of D-phenyl-
glycine at that point of conversion. Furthermore, this
mutant still has a sufficiently high activity for preparative
conversions.15 Thus, we reasoned that this F24A mutant
might also behave as an improved catalyst in PGA-
catalyzed kinetic resolutions of amino acids through
enantioselective acylation in aqueous solution using
phenylacetic acid methylester as the acyl donor.
The results show that when Gln was the amino acid sub-
strate used, the ee of the remaining D-enantiomer (eeD)
initially showed a rapid increase with time due to prefer-
ential enantioselective acylation of L-Gln (Fig. 2). A
maximum was reached after 3 h, after which the eeD
decreased slowly. This is due to the hydrolysis of the
N-phenylacetyl-L-amino acid, as evident from the con-
comitant reduction of the degree of conversion of total
free amino acid. The maximal eeD that was transiently
obtained during the resolution of Gln was surprisingly
high, peaking at a value of 97% for eeD (Table 1). This
is a tremendous improvement compared to the maximal
eeD obtained with wild-type PGA (13%). Under similar
conditions, a high maximal eeD (90%) was also obtained
for Glu (Table 1). For the other amino acids tested (Asn,
Asp, and Ser) the N-phenylacetyl-L-amino acids were
obtained in high ee (eeL >93% in all cases, Table 1),
but the maximal eeD values for the remaining free
D-amino acids were only moderate (50–80%). In the case
of Asn and Asp, the amount of acyl donor was increased
from 1.3 to 1.8 equiv, although the maximal eeD hardly
improved. Neither an increase of the initial amino acid
R
H2N
O-
R
E
CH3OH
R
O
HN
H2N
+
O-
OCH3
E
O-
O
O
O
O
O
Acyl-enzyme complex
Hydrolysis
PhAc-L-amino acid
D-amino acid
Product
hydrolysis
R
H2N
O-
O-
O
O
L-amino acid
Figure 1. Kinetic resolution of D,L-amino acids by enantioselective acylation of the L-enantiomers using PGA and phenylacetic acid methylester.
The vertical arrows indicate (undesirable) hydrolysis reactions. E = enzyme.

C. Carboni et al. / Tetrahedron: Asymmetry 17 (2006) 245–251
247
100
90
80
70
60
50
40
30
20
10
0
100
90
Ser
Gln
80
70
60
50
40
30
20
10
0
0
4
8
12
16
20
24
0
12
24
36
48
Time (h)
Time (h)
100
90
80
70
60
50
40
30
20
10
0
100
90
80
70
60
50
40
30
20
10
0
Asp
Glu
0
4
8
12
16
20
24
0
4
8
12
16
20
24
Time (h)
Time (h)
100
90
80
70
60
50
40
30
20
10
0
Asn
0
4
8
12
16
20
24
Time (h)
Figure 2. Plots of conversion and eeD versus time for the phenylacetylation for various amino acids with F24A mutant PGA. Symbols (j, eeD;
m, conversion) indicate experimental values, lines represent simulated data, using the lumped parameters given in Table 2.
concentration from 1 to 2 M, nor a variation of the tem-
perature between 5 and 30 ꢁC altered the maximum eeD
value that was reached.
lumped steady-state kinetic parameters, a, b0 and c
(Table 2), that govern the course of a kinetic resolution
in terms of ee and c in time and can be regarded as
relative preference for product as compared to acyl
donor (a), concentration-dependent nucleophile (ND,NL)
reactivity (b0), and maximum reactivity of the nucleo-
phile at saturating concentrations of nucleophile as com-
pared to water (c). The values that were obtained by
fitting (see Experimental) may not be unique solutions,
but do show how the lumped kinetic parameters are
related to the ee and yield that are obtained in PGA-
catalyzed kinetic resolutions.
In order to obtain a better understanding of the kinetics
of these F24A mutant PGA-catalyzed resolutions, we
developed a computer model that describes conversion
as a function of steady-state kinetic parameters. The
model consists of a set of ordinary differential equations
that describe the conversion of the synthetic coupling
products PL and PD, the hydrolysis product A, and the
free enzyme (see Fig. 3). The model contains three

248
C. Carboni et al. / Tetrahedron: Asymmetry 17 (2006) 245–251
Table 1. Enzymatic resolution of racemic D,L-amino acids using the F24A mutant of PGA from E. coli as well as wild-type PGA as the biocatalyst
and phenylacetic acid methylester as acyl donor in aqueous solution at 25 ꢁC (for further details, see Section 4)
R
R
R
F24A mutant
of PGA (H2O)
HN
H2N
+
H2N
O-
+
O-
OCH3
O-
O
O
O
O
O
PhAc-L-amino acid
D-amino acid
a
Amino acid
R
Acyl donor (equiv)
Enzyme
pH
eeD max (%)
eeL (%)
Conversiona,b (%)
E valuec
Gln
Gln
Glu
Asn
Asp
Ser
(CH2)2CONH2
(CH2)2CONH2
(CH2)2CO2H
CH2CONH2
CH2CO2H
1.3
1.3
1.3
1.3 (1.8)
1.3 (1.8)
1.8
F24A
Wild-type
F24A
F24A
F24A
9.0
9.4
9.0
9.4
9.6 (9.4)
9.4
97
13
90
76 (78)
48 (56)
73
>99
99
>99
97 (97)
95 (91)
93
49.5
12
48
44 (45)
34 (38)
44
810
48
250
140
46
CH2OH
F24A
60
a Measured at maximal eeD.
b Of the racemic amino acid.
c Calculated using the eeL and eeD values at maximal eeD value.
E +A+ NL
khL
ktL
KPL
E~A.NS
E.PS
E + PL
KNL
k-tL
NL
+
KS
k2
kh
E + AD
E.AD
E~A
E + A
+
ND
ktD
KPD
KND
E~A.ND
E.PD
E +PD
khD
k-tD
E + A +ND
Figure 3. Kinetic scheme for
a PGA-mediated resolution of amino acids in water. AD = acyl donor; N = nucleophilic amino acid;
P = phenylacetylated product; A = phenylacetic acid; E = enzyme.
(as a nucleophile) of the L-enantiomer as compared to
Table 2. Steady state parameters obtained by a Monte Carlo type
numerical fitting of the experimental data to the model described in the
text
the D-enantiomer. The same L-preference is caused
by the low ratio of the c values (0.017), which indicates
that the preference for L-Gln as the deacylating enantio-
mer prevails at high amino acid concentration. The low
value for aL reflects a low tendency of the F24A mutant
PGA to react with the N-phenylacetyl-L-Gln produced,
allowing conversion of L-Gln to proceed to completion
by accepting phenylacetic acid methylester even when
product starts to accumulate, and contributing to the
high eeD. With Gln, after 2.5 h the reverse reaction
became important. Hydrolytic degradation of N-phen-
ylacetyl-L-Gln was accompanied by the release of free
L-Gln, which caused the degree of conversion c and
eeD to drop drastically.
a
b0 (mMꢀ1
)
c
L-Gln
D-Gln
L-Ser
D-Ser
L-Glu
D-Glu
L-Asp
D-Asp
L-Asn
D-Asn
2.5
2.5
12.5
125
2.5
2.5
0.033
5
0.125
30
1
300
56
30
2
3.75
2
8.75
2
0.05
3
0.05
0.053
0.05
1.5
0.33
2
0.29
1
10
The definition of the parameters is given in the experimental.19,20
The high enantioselectivity for Gln conversion is in
agreement with a high ratio (30) between the b0 values
for L- and D-Gln, which reflects a much higher reactivity
The kinetic resolution experiment with the F24A mutant
PGA and D,L-Ser followed a similar trend, but the ee
and c obtained were significantly lower. The time course

C. Carboni et al. / Tetrahedron: Asymmetry 17 (2006) 245–251
249
of ee and c with D,L-Ser could also be fitted using the
model, and when comparing the lumped parameters
obtained for the enantiomers of Gln and Ser it appeared
that the lower quality of the kinetic resolution of Ser
can be explained by a lower L-selectivity of the enzyme
in the deacylation reaction, as reflected by b /b ratios
of 30 and 5 for Gln and Ser, respectively. The cL/cD
ratios predicted by the model were 0.017 and 1, respec-
tively, which also indicates a better enantioselectivity in
the deacylation of the enzyme coupled to the L-enan-
tiomer in the case of Gln than with Ser, since a low value
for c means a high tendency to react at saturating con-
centration. Besides the lower eeD obtained with Ser as
the nucleophile, it is also remarkable that there is no
drop in the ee during the course of the experiment,
unlike with Gln. This can be explained by higher b0 val-
ues and lower c values for Ser (c should be low for good
conversion), which indicate a low tendency of the acyl-
enzyme to undergo hydrolysis, implying that the acyl
donor is not rapidly depleted. The incompleteness of
conversion as in the case of Ser can be explained by
binding site of the acyl-enzyme intermediate. The model
may be used as a tool to direct experiments aimed at
improving conversions. For example, in the case of
D,L-Gln and D,L-Glu, engineering of PGA should be
aimed at extending the period of time at which a high
ee and c are maintained, for example, by means of
additional mutagenesis that leads to suppression of
hydrolysis of the acyl donor. For the current experi-
ments, the kinetic modeling studies confirm the more
distinct preference of the F24A mutant PGA for phenyl-
acetic acid methylester as compared to the N-phenylacet-
yl-L-amino acid in the case of conversion of Gln as
compared to the case with Ser as the nucleophile. The
present results are of significant industrial importance
especially for the cost-efficient production of D-Gln
and D-Glu. These D-amino acids can now be obtained
in good ee in one single step from the racemate with a
simple work-up.
L
D
0
0
4. Experimental
4.1. Materials
the relatively high value for a (aL /aL,Gln = 5), which
,Ser
reduces c even if the acyl donor is still present. This
phenomenon appears as product inhibition, and can
possibly be influenced by selecting another acyl donor
or by further mutagenesis. The progress of the kinetic
resolution with Glu was similar to that obtained with
Gln, although the maximal eeD obtained was signifi-
cantly lower, which is in agreement with the smaller
difference between the b0 values and the c values for
the enantiomers of this substrate, as compared to
Gln. The resolution with Asp was even worse, and the
differences between the b0 and c values, which mainly
determines enantioselectivity, was even smaller in this
case.
The F24A mutant of PGA was prepared as described
earlier.15 The wild-type PGA was obtained from DSM
Anti-infectives (Delft, The Netherlands). D,L-Gln, D,L-
Glu, D,L-Asn, D,L-Asp, D,L-Ser, and phenylacetic acid
methylester were obtained from Fluka and used without
further purification.
4.2. Enzymatic acylation reactions
To a mixture of 25 mL of a 1 M solution of the D,L-
amino acid in water, which had been adjusted to
pH = approximately 9 with 2 M aqueous NaOH, and
1.3–1.8 equiv of phenylacetic acid methylester, was
added 2.2 g of the PGA mutant (or of the wild-type
PGA). The mixture was stirred at 25 ꢁC and during
the reaction the pH was maintained at approximately
9 by titration with 2 M aqueous NaOH using a pH stat
apparatus. To monitor the reaction, a 1 mL aliquot was
withdrawn from the reaction mixture with regular time
intervals and diluted 10-fold with 50 mM phosphate
buffer (pH = 2.7). The resulting sample was used for
both HPLC analysis 1 (ee determination of the amino
acid substrates) and HPLC analysis 2 (ee determination
of the N-phenylacetylated products).
3. Conclusion
Herein, we have shown that D-Gln and D-Glu can be
obtained in one step in high ee (97% and 90%, respectively)
in an enzymatic kinetic resolution of their racemates.
This was achieved by enantioselective conversion of
the L-enantiomers to the N-phenylacetyl derivatives with
phenylacetic acid methylester as an acyl donor in aque-
ous solution by using an F24A mutant of PGA from
E. coli. The high eeD values are mainly due to the
significantly suppressed hydrolysis rate of N-phenylacet-
yl-L-Gln and N-phenylacetyl-L-Glu, respectively, com-
pared to wild-type PGA. Although, for the other
amino acids tested (Asn, Asp, and Ser), the intrinsic E
values of the F24A mutant are also high and the N-phen-
ylacetylated L-enantiomers can be obtained in high ee
(>93%), because the maximal eeD values are only
moderate (50–80%), the hydrolysis rate of the N-phenyl-
acetyl-L-amino acid products is still too high and the
conversion of the amino acid nucleophile is complete.
A kinetic model describing conversion in time appears
suitable for the numerical simulation of the outcome
of the kinetic resolutions of amino acids with PGA,
and can explain differences in enantioselectivity in terms
of variations in nucleophile reactivity and competition
between substrate and product at the nucleophile
4.3. Synthesis of the N-phenylacetyl-D,L-amino acid
reference compounds
To a 2.5 M aqueous solution of the D,L-amino acid at
pH = 10 was added, under vigorous stirring, phenyl-
acetyl chloride (1.1 equiv) during 15 min at 0–5 ꢁC,
maintaining the pH at approximately 10 by the
simultaneous addition of 50 wt.% aqueous NaOH solu-
tion. Subsequently, the solution was stirred for another
2 h at ambient temperature and acidified to pH = 1
using 32 wt.% aqueous HCl solution, which resulted in
precipitation of the N-phenylacetylated D,L-amino acids.
The precipitates were isolated by filtration, recrystallized

250
C. Carboni et al. / Tetrahedron: Asymmetry 17 (2006) 245–251
from water/ethanol and dried under reduced pressure.
N-Phenylacetyl-D,L-Gln, N-phenylacetyl-D,L-Glu, N-phen-
ylacetyl-D,L-Asn, N-phenylacetyl-D,L-Asp, and N-phen-
ylacetyl-D,L-Ser were obtained in 72%, 75%, 68%,
4.5. HPLC analysis 2: determination of the ee of
N-phenylacetylated amino acid products
For N-phenylacetylated Gln, Asn, Asp, and Ser, a Chiro-
biotic T HPLC column (Astec, size 250 mm · 4.0 mm)
was used in series with an Inertsil ODS-3 HPLC column
(150 mm · 4.6 mm, particle size 5 lm). Both columns
were operated at ambient temperature. The mobile
phase was a gradient of eluent A (80 vol % 15 mM
ammonium acetate in water at pH = 4.1 and 20 vol %
methanol) and eluent B (20 vol % 15 mM ammonium
acetate in water at pH = 4.1 and 80 vol % methanol)
with the following programme: 0–13 min (0% B), 13–
13.1 min (0!100% B), 13.1–18 min (100% B), 18–
18.1 min (100!0% B) and 18.1–25 min (0% B). The
injection volume was 10 lL and the flow 1.0 mL/min.
The compounds were detected by UV (215 nm). Under
these conditions the D- and L-enantiomers of the various
N-phenylacetylated amino acids had the retention times
as shown in Table 4.
1
61%, and 49% yield, respectively. H NMR (DMSO-
d6), d (ppm): N-phenylacetyl-D,L-Gln: 1.77 (1H, m,
–CH2–CHNH), 1.96 (1H, m, –CH2–CHNH), 2.13
(2H, t, –CH2–CONH2), 3.48 (2H, s, CH2Ph), 4.15 (1H,
dd, CH), 6.77 (1H, s, CONH2), 7.28 (5H, m, Ph), 7.37
(1H, s, CONH2), 8.39 (1H, s, NHCO). N-Phenylacetyl-
D,L-Glu: 1.60 (1H, m, –CH2–CHNH), 1.95 (1H, m,
–CH2–CHNH), 2.38 (2H, t, CH2COOH), 3.50 (2H, s,
CH2Ph), 4.15 (1H, dd, CH), 7.28 (5H, m, Ph), 7.70
(1H, s, NHCO). N-Phenylacetyl-D,L-Asn: 2.52 (2H, m,
–CH2–CON), 3.48 (2H, s, CH2Ph), 4.50 (1H, dd, CH),
6.90 (1H, s, CONH2), 7.27 (5H, m, Ph), 7.37 (1H, s,
CONH2), 8.30 (1H, s, NHCO). N-Phenylacetyl-D,L-
Asp: 2.72 (2H, m, CH2CO2H), 3.44 (2H, s, CH2Ph),
4.40 (1H, dd, CH), 7.27 (5H, m, Ph), 8.20 (1H, s,
NHCO), 11.0 (COOH). N-Phenylacetyl-D,L-Ser: 2.10
(1H, br s, OH), 3.44 (2H, s, CH2Ph), 4.00 (2H, d,
CH2OH), 4.50 (1H, dd, CH), 7.06–7.14 (5H, m, Ph),
8.20 (1H, s, NHCO).
For N-phenylacetyl-Glu, a Chiralcel OJ HPLC column
(Chiral Technologies Inc., size 250 mm · 4.6 mm, parti-
cle size 10 lm) was used in series with a Lichrospher diol
(Alltech) HPLC column (150 mm · 4.6 mm, 5 lm), both
operated at ambient temperature. The mobile phase was
n-hexane/t-butanol/trifluoroacetic acid (80/20/0.1 v/v/
v). The injection volume was 20 lL and the flow rate
1.0 mL/min. The compounds were detected by UV
(210 nm). Under these conditions, D- and L-N-phenyl-
acetyl-Glu had a retention time of 22.5 and 26.5 min,
respectively.
4.4. HPLC analysis 1: determination of the ee of amino
acid substrates
An HPLC column (size 150 mm · 4.0 mm) consisting of
Crownpak Cr(+) from Daicel Technologies was used,
which was operated at 5 ꢁC, except for the phenacetyl-
ation of D,L-Ser where it was operated at 0 ꢁC. Aqueous
HClO4 (with concentrations as indicated in Table 3) was
used as the eluent, the injection volume was 20 lL and
the flow 0.5 mL/min except for the phenylacetylation
of D,L-Ser where the flow was 0.4 mL/min. The
compounds were detected by fluorescence at 338 nm
after post-column derivatization of the amino group
consisting of a reaction with o-phthalaldehyde and
2-mercapto-ethanol at pH = 10 at ambient temperature.
Under these conditions, the D- and L-enantiomers of the
various amino acids had the retention times as shown in
Table 3.
4.6. Computer simulations
The kinetic model follows similar principles as the
model described by Youshko et al.17,18 for the acylation
of b-lactam compounds, but incorporates two enantio-
mers of the nucleophile (Fig. 3). The model consists of
differential equations (dA/dt, dPD/dt, dPR/dt, dE/dt),
which were derived from the scheme shown in Figure
3 and describe the change of the concentrations of all
substrates and enzyme in time. The microscopic rate
Table 3. HPLC eluents and retention times for the amino acid substrate enantiomers
Amino acid
Eluent
pH
Retention time of D-enantiomer (min)
Retention time of L-enantiomer (min)
Gln
Glu
Asn
Asp
Ser
10 mM HClO4
10 mM HClO4
100 mM HClO4
50 mM HClO4
100 mM HClO4
2
2
1
1.3
1
4.50
4.90
4.60
5.20
5.85
7.45
11.45
5.45
7.10
7.35
Table 4. Retention time of the enantiomers of the N-phenylacetylated amino acids
N-Phenylacetylated amino acid
Retention time of L-enantiomer (min)
Retention time of D-enantiomer (min)
N-PhAc-Gln
N-PhAc-Asn
N-PhAc-Asp
N-PhAc-Ser
8.40
6.50
9.40
7.90
11.50
7.65
11.10
9.10

C. Carboni et al. / Tetrahedron: Asymmetry 17 (2006) 245–251
251
constants and equilibrium constants in the equations for
References
substrate conversion were replaced by three lumped
steady-state kinetic parameters called a, b0 and c, for
each enantiomer. The parameters aL and aD describe
the competition between acyl donor (phenylacetic acid
methylester, AD in Fig. 3) and product (phenylacety-
lated amino acid, PL or PD) for the free enzyme and
equal the ratios between the kcat/Km values of the
1. Kim, M. G.; Lee, S. B. J. Mol. Catal. B: Enzym. 1996, 1,
71–80.
´
´
2. Rosell, C. M.; Fernandez-Lafuente, R.; Guisan, J. M. J.
Mol. Catal. 1993, 84, 365–371.
´
´
3. Rosell, C. M.; Fernandez-Lafuente, R.; Guisan, J. M.
Ann. N.Y. Acad. Sci. 1995, 750, 425–428.
4. van Langen, L. M.; de Vroom, E.; van Rantwijk, F.;
Sheldon, R. FEBS Lett. 1999, 456, 89–92.
5. Bossi, A.; Cretich, M.; Righetti, P. G. Biotechnol. Bioeng.
1998, 60, 454–461.
6. See, for instance, the following reviews: (a) Breuer, M.;
Ditrich, K.; Habicher, T.; Hauer, B.; Kesseler, M.;
enzyme for substrate and product. Parameter b0 equals
D
ktD/khÆK (and a similar relation exists for the L-enan-
ND
tiomer) and describes the concentration dependence of
the reactivity (relative to water, mMꢀ1) of the nucleo-
philic amino acid (ND,NL) that is converted by reacting
with the acylenzyme. Parameter c (ktD/khD) is the inverse
of the maximum nucleophile reactivity (relative to
Sturmer, R.; Zelinski, T. Angew. Chem., Int. Ed. 2004,
¨
43, 788–824; (b) Yagasaki, M.; Ozaki, A. J. Mol. Catal. B:
Enzym. 1998, 4, 1–11; (c) Bommarius, A. S.; Schwarm, M.;
Drauz, K. J. Mol. Catal. B: Enzym. 1998, 5, 1–11.
7. Wegman, M. A.; Janssen, M. H. A.; van Rantwijk, F.;
Sheldon, R. A. Adv. Synth. Catal. 2001, 343, 559–
576.
water). The value of b /b D corresponds to the apparent
L
0
0
E value of the enzyme for the chiral amino acids, defined
as VN /VN = E [NL]/[ND]. The lumped parameters can
*
L
D
be determined experimentally in separate steady-state
measurements, for example, by measuring the initial rate
of product formation with the pure enantiomers, as will
be described in detail elsewhere. After replacing the con-
stants given in Figure 3 by these lumped parameters a, b,
and c, the equations were numerically integrated using
the programme Mathcad 12 (Mathsoft) to obtain sub-
strate and product concentrations in time, from which
the ee and conversion in time were calculated. Further-
more, a Monte Carlo routine implemented in Matlab 7
(Mathworks) was used for fitting the differential equa-
tions to the experimental data (Fig. 2). This gave values
for the lumped kinetic parameters that provided a good
match of the simulations with the experimental data
(Fig. 2). The initial substrate concentrations that were
used in these simulations were in accordance with the
experiments (1.3 M phenylacetic acid methylester, 1 M
racemic amino acid). Starting values for microscopic
kinetic constants and equilibrium constants were
adopted from Youshko et al.19 and converted to lumped
parameters, that were used as starting values for the
simulations, which yielded the results shown in Table
2. All conversions were treated as closed systems, with-
out removal of products. The obtained parameters were
used in simulations, from which ee and c were calculated
and compared to the experimental data (Fig. 2).
8. See, for instance: (a) Schoemaker, H. E.; Boesten, W. H.
J.; Broxterman, Q. B.; Roos, E. C.; Kaptein, B.; van den
Tweel, W. J. J.; Kamphuis, J.; Meijer, E. M. Chimia 1997,
51, 308; (b) Koksch, B.; Quaedflieg, P. J. L. M.; Michel,
T.; Burger, K.; Broxterman, Q. B.; Schoemaker, H. E.
Tetrahedron: Asymmetry 2004, 15, 1401–1407.
9. See, for instance: Bommarius, A. S.; Drauz, K.; Groeger,
U.; Wandrey, C. In Chirality in Industry; John Wiley &
Sons Ltd., 1992; pp 371–397.
10. Cole, M. Biochem. J. 1969, 115, 747–756.
11. Cole, M. Biochem. J. 1969, 115, 757–764.
12. Zmijewski, M.; Briggs, B. S.; Thompson, A. R.; Wright,
I. G. Tetrahedron Lett. 1991, 32, 1621–1622.
13. Basso, A.; Braiuca, P.; De Martin, L.; Ebert, C.; Gardossi,
L.; Linda, P. Tetrahedron: Asymmetry 2000, 11, 1789–
1796.
14. Carboni, C.; Quaedflieg, P. J. L. M.; Broxterman, Q. B.;
Linda, P.; Gardossi, L. Tetrahedron Lett. 2004, 45, 9649–
9652.
15. Alkema, W. B. L.; Dijkhuis, A.-J.; de Vries, E.; Janssen,
D. B. Eur. J. Biochem. 2002, 269, 2093–2100.
16. E = (ln((1ꢀc/100) (1ꢀeeS/100))) / (ln((1ꢀc/100) (1 + eeS/
*
*
100))); see: Chen, C.-S.; Fujimoto, Y.; Girdaukas, G.; Sih,
C. J. J. Am. Chem. Soc. 1982, 104, 7294–7299.
ˇ
17. Youshko, M. I.; Svedas, V. K. Biochemistry (Mosc) 2000,
65, 1367–1375.
18. Youshko, M. I.; Chilov, G. G.; Shcherbakova, T. A.;
ˇ
Svedas, V. K. Biochim. Biophys. Acta 2002, 1599, 134–
140.
Acknowledgments
ˇ
19. Youshko, M. I.; Bukhanov, A. L.; Svedas, V. K. Biochem-
istry (Mosc) 2003, 68, 334–338.
20. Gabor, E. M.; Janssen, D. B. Protein Eng. Des. Sel. 2004,
17, 571–579.
The authors wish to gratefully acknowledge Ilse Maes
for the analytical support.