An efficient route to both enantiomers of allo-threonine by simultaneous amino acid racemase-catalyzed isomerization of threonine and crystallization

Article abstract of DOI:10.1002/adsc.201100051

We present an efficient method for the production of D- and L-allo-threonine (allo-Thr) with very high purity by enzymatic isomerization of L- or D-threonine (Thr) and simultaneous crystallization. Isomerization of Thr to allo-Thr is catalyzed by a purified amino acid racemase (AArac12996) from Pseudomonas putida NBRC12996, which can easily be obtained from a recombinant E. coli strain by secretion into the medium and subsequent anion exchange chromatography. Crystallization of D- and L-allo-Thr was performed in a repetitive batch mode over a period of up to 55 days at 30 °C. Total amounts of 30.8 g D-allo-Thr and 32.4 g L-allo-Thr were obtained with a very good diastereomeric excess of de_D-allo>99.2% and de _L-allo>98.4%, respectively, and in good yields (D-allo-Thr: 83%, L-allo-Thr: 79%). The enzyme's remarkable high stability under process conditions resulted in enzyme specific yields of 2.56 g D-allo-Thr per mg AArac12996 and 1.62 g L-allo-Thr per mg AArac12996. In contrast to chemical multi-step syntheses of allo-Thr, our process consists of only one enzyme-catalyzed reaction step with simultaneous product crystallization. The process is performed under low energy consumption (30 °C, atmospheric pressure) in water and avoids the use and production of any toxic or harmful compounds. Recovery of the enantiomerically pure products is performed by simple filtration which reduces downstream processing significantly compared to chromatographic methods which are usually applied. Copyright

Full text of DOI:10.1002/adsc.201100051

FULL PAPERS
DOI: 10.1002/adsc.201100051
An Efficient Route to Both Enantiomers of allo-Threonine by
Simultaneous Amino Acid Racemase-Catalyzed Isomerization of
Threonine and Crystallization
Kerstin Wꢀrges,a Ursula Mackfeld,a Martina Pohl,a,* Stephan Lꢀtz,a,b
Susanne Wilhelm,c Wolfgang Wiechert,a and Tina Kubitzkia
a
IBG-1: Biotechnology, Forschungszentrum Jꢀlich GmbH, 52425 Jꢀlich, Germany
Fax : (+49)-2461-61-3870; phone: (+49)-2461-61-4388; e-mail: ma.pohl@fz-juelich.de
Present address: Novartis Institutes for BioMedical Research, Basel, Switzerland
Institute of Molecular Enzyme Technology, University of Dꢀsseldorf, Dꢀsseldorf, Germany
b
c
Received: January 21, 2011; Published online: May 23, 2011
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/adsc.201100051.
Abstract: We present an efficient method for the under process conditions resulted in enzyme specific
production of d- and l-allo-threonine (allo-Thr) with yields of 2.56 g d-allo-Thr per mg AArac12996 and
very high purity by enzymatic isomerization of l- or 1.62 g l-allo-Thr per mg AArac12996. In contrast to
d-threonine (Thr) and simultaneous crystallization. chemical multi-step syntheses of allo-Thr, our pro-
Isomerization of Thr to allo-Thr is catalyzed by a pu- cess consists of only one enzyme-catalyzed reaction
rified amino acid racemase (AArac12996) from step with simultaneous product crystallization. The
Pseudomonas putida NBRC12996, which can easily process is performed under low energy consumption
be obtained from a recombinant E. coli strain by se- (308C, atmospheric pressure) in water and avoids the
cretion into the medium and subsequent anion ex- use and production of any toxic or harmful com-
change chromatography. Crystallization of d- and pounds. Recovery of the enantiomerically pure prod-
l-allo-Thr was performed in a repetitive batch mode ucts is performed by simple filtration which reduces
over a period of up to 55 days at 308C. Total downstream processing significantly compared to
amounts of 30.8 g d-allo-Thr and 32.4 g l-allo-Thr chromatographic methods which are usually applied.
were obtained with a very good diastereomeric
excess of ded-allo >99.2% and del-allo >98.4%, respec-
tively, and in good yields (d-allo-Thr: 83%, l-allo- Keywords: amino acid racemases; crystallization; iso-
Thr: 79%). The enzymeꢁs remarkable high stability merization; allo-threonine
Introduction
l-allo-Thr aldolase from Aeromonas jandaei DK-39.
The same reaction can be performed using the serine
allo-Threonine (allo-Thr) is part of many biologically hydroxymethyltransferase GlyA from E. coli.[16] Lim
active peptides such as naturally occurring antibiotics et al.[17] reported on an amino acid racemase from
as well as other therapeutic chiral molecules.[1–7] For Pseudomonas putida ATCC 17642 with a threonine-a-
example, d-allo-Thr is used for chemical syntheses of epimerase activity (EC 5.1.1.6), which, to the best of
key intermediates for the production of thienamycin, our knowledge, is the first and only reference about
which is a potent antibiotic used against Gram-posi- an enzymatic isomerization of threonine (Thr) in the
tive and Gram-negative bacteria.[8,9] Furthermore, literature. Although several amino acid racemases
allo-Thr is used as starting material in the chemical show broad substrate specificity such as arginine race-
synthesis of lysobactin, an antibiotic used against me- mase from P. graveolens[18] and amino acid racemases
thicillin-resistant Staphylococcus aureus (MRSA).[10]
from A. punctata[19] and P. putida[20], they usually do
Several methods for the chemical as well as enzy- not act on amino acids with substituted b-methylene
matic synthesis of allo-Thr have been described.[11–15] groups such as threonine, isoleucine and valine. We
Kataoka et al.[2] reported a stereoselective C C bond could demonstrate that the amino acid racemase from
ꢀ
formation between glycine and acetaldehyde using Pseudomonas putida NBRC12996 (AArac12996) also
Adv. Synth. Catal. 2011, 353, 2431 – 2438
ꢂ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
2431

Kerstin Wꢀrges et al.
FULL PAPERS
catalyzes the isomerization of these amino acids. Nev- and d-forms, whereas diastereomers with two chiral
ertheless, chemical methods[11–15] clearly outnumber centers in a- and b-positions, such as Thr, are convert-
the described enzymatic approaches. However, they ed to their diastereomeric forms (Figure 1).
consist of multi-step reaction cascades with only low
yields. For example, Blaskowich et al.[12] described an
8-step synthesis of d- or l-allo-Thr starting from the Results and Discussion
respective threo-isomer with a total yield of 38%. Al-
though both enantiomers of allo-Thr are commercial- Cloning, Overexpression and Purification of
ly available, they are extremely expensive and, hence, AArac12996
not useful as starting material for large-scale synthe-
ses.[21]
The coding gene of AArac12996 from P. putida
Here we describe an easy and inexpensive method NBRC12996 was cloned into pET22b+ and overex-
for the production of high-priced allo-Thr starting pressed in E. coli BL21
AHCTUNGTRENNUNG
from cheap Thr using AArac12996. This enzyme was
Surprisingly, our studies on protein expression re-
chosen since Kino et al.[22] reported an activity on l- sulted in a progressive accumulation of racemase ac-
Thr, although it is very low compared to the preferred tivity in the extracellular medium over time (Table 1),
substrate l-lysine (0.1%). Separation of allo-Thr from
Thr in our approach is performed by simultaneous
Table 1. Development of the intracellular and extracellular
activity of AArac12996 during expression in E. coli
BL21ACTHUNRTGNEUNG(DE3) cells in a 2-L cultivation. Activity was assayed
crystallization in a simple batch reactor. In contrast to
replacing crystallization[23,24] or dynamic kinetic reso-
lution processes, where enzymatic racemization is
combined with preferential crystallization of single
enantiomers from a saturated racemic solution,[25] the
crystallization of allo-Thr is based on its lower solubil-
ity limit compared to Thr in water.
using the standard assay as described in the Experimental
Section.
Expression time Activity in medium
Activity in cells
[U]
[h]
[U]
AArac12996 has a broad substrate specificity which
is comparable to those of other pyridoxal-5’-phos-
phate (PLP)-dependent amino acid racemases.[26] This
group of enzymes catalyzes the isomerization of the
20.0
42.5
67.0
300
419
3504
1225
2361
527
a-C atom. For amino acids with the a-C atom as the indicating an intrinsic signal sequence in the coding
only chiral center this means interconversion of the l- gene. The full amino acid sequence (see Supporting
Information) of AArac12996 was analyzed for a cor-
responding signal sequence using the program SignalP
3.0[27] and a specific cleavage site was found between
Ala24 and Ala25. This signal sequence shows the typ-
ical tripartite structure of signal peptides, consisting
of a hydrophobic core region (h-region) flanked by a
positively charged n-region and a polar c-region that
contains the signal peptidase cleavage site.[28] The sub-
sequent release of periplasmic racemase through the
outer cell membrane is supposed to be an autolytic
response to the stress caused by overexpression and
secretion of a foreign protein.[29] Matsui et al.[30] pro-
posed a translocation mechanism for an amino acid
racemase ArgR from P. taetrolens, which has an
amino acid sequence identity of 73% compared to
AArac12996. ArgR is thought to be translocated via
the Sec pathway by cleavage of an N-terminal signal
peptide. According to the physiological role of ala-
nine racemase[31] it is assumed that also amino acid
racemase plays an important role for the synthesis of
d-alanine and d-glutamate in the murein layer.
Cultivation of the recombinant E. coli in 2 L LB-
medium yielded a total activity of 5900 U in the culti-
vation broth (medium and cells before cell separa-
tion). Purification of the secreted enzyme from the
medium by ultrafiltration and subsequent anion ex-
Figure 1. Isomerization of a-C atom catalyzed by amino
acid racemase. Top : amino acids with only one chiral center
are converted to their counterenantiomer. Middle and
bottom: isomerization of a-C atom so that l-Thr is convert-
ed to its diastereomer d-allo-Thr and d-Thr to its diastereo-
mer l-allo-Thr.
2432
asc.wiley-vch.de
ꢂ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2011, 353, 2431 – 2438

An Efficient Route to Both Enantiomers of allo-Threonine
change chromatography resulted in pure enzyme (ac-
cording to SDS-PAGE, see Supporting Information)
with a specific activity of 8.6 U mgꢀ1. The freeze-dried
protein (320 mg, protein concentration of ~73%) was
stored at ꢀ208C.
Characterization of AArac12996
AArac12996 is a PLP-dependent dimeric enzyme
with a molecular weight of 44.2 kDa per monomer.
While AArac12996 shows maximum activity for l-
lysine (100% relative activity), the activity for l-Thr
isomerization is rather low (0.1%). The enzyme was
initially characterized by following the isomerization
of l-asparagine (Asn), which was chosen because (i)
the isomerization of Thr cannot be followed polari-
metrically for easy activity measurements; (ii) the ac-
Figure 4. Stability of AArac12996 in water. Enzyme solu-
tions (1 mgmLꢀ1) were stored at 208C to 478C. Aliquots
were taken for polarimetric determination of residual activi-
ty [isomerization of l-Asn (90 mM) at 308C for 5 min].
tivity for l-Asn is ca. 20-fold higher than for l-Thr. 318C (Figure 3). For isomerization of Asn the activa-
The pH-optimum was found in the range of pH 8.6– tion energy Ea was calculated as 48.4 kJmolꢀ1, which
9.7 (Figure 2). Investigation of the temperature de- is comparatively high compared to related racemases,
pendency yielded a maximum initial rate activity at such as alanine racemases from P. fluorescens[32]
(29.2 kJmolꢀ1 for l-alanine) and B. stearothermophi-
lus[33] (7.54 kJmolꢀ1 for d-alanine). The stability of
AArac12996 was studied over 7 days by following the
residual activity of liquid enzyme samples between
208C and 478C (Figure 4). The enzyme is stable for 7
days at 20 and 308C, whereas a rapid decay of activity
was observed at 408C (6% residual activity in initial
rate studies, Figure 3). Nevertheless, stability investi-
gations at 408C (which assayed the residual activity at
308C) still showed a residual activity of about 16%
after 3 days. These results indicate a reversible inacti-
vation of AArac12996 at temperatures between 318C
and 408C. At 478C AArac12996 was inactivated irre-
versibly and showed no residual activity after an incu-
bation of 2 h. The influence of supplemented PLP,
which is covalently bound as cofactor to the active
site of AArac12996, was investigated in the concen-
Figure 2. Determination of pH-optimum for AArac12996
(0.1 mgmLꢀ1) in glycine/NaCl buffer (90 mM) at 308C. Rel-
ative activities were determined polarimetrically by follow-
ing the isomerization of l-Asn (25 mM) for 5 min.
tration range between 0 and 50 mm. Within this con-
centration range PLP had no influence on the enzyme
activity; thus, it was omitted during biotransforma-
tions.
Kinetic parameters were determined for the iso-
merization of Thr. While the maximal velocities for
the isomerization of both Thr-enantiomers are identi-
cal, their Km values differ by a factor of 2 with l-Thr
being the preferred enantiomer (Table 2). Based on
these parameter studies, AArac12996 was character-
ized in detail for its application in the production of
allo-Thr.
Repetitive Batch Crystallization of allo-Threonine
Figure 3. Determination of the temperature dependency for
AArac12996 (0.1 mgmLꢀ1) in water. Residual activity was
determined polarimetrically at 308C for the isomerization of
l-Asn (90 mM) for 5 min.
Both enantiomers of allo-Thr were produced and
crystallized with very high purity starting from either
Adv. Synth. Catal. 2011, 353, 2431 – 2438
ꢂ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
2433

Kerstin Wꢀrges et al.
FULL PAPERS
Table 2. Kinetic parameters for the isomerization of Thr cat-
alyzed by AArac12996 at 308C in water. Initial rate activi-
ties were determined by chiral HPLC. Vmax values were cal-
culated per mg of protein in the lyophilisate. kcat corre-
sponds to the turnover number per monomer.
ization of Thr to allo-Thr, which takes about 1.5 days
per gram Thr (thus the time per batch depends on the
amount of added excess Thr). The overall Thr/allo-
Thr concentration in the solution remains constant
throughout phases II and III and is fixed by the solu-
bility limits of both Thr species. allo-Thr crystals are
separated by filtration and dried at 608C. The next
batch can be started by the addition of fresh Thr.
Steps II and III are repeated for a repetitive batch
crystallization.
Reaction
Vmax [U mgꢀ1]
AArac12996
Km
[mM]
kcat
[sꢀ1]
ACHTUNGTRENNUNG
l-Thr!d-allo-Thr
d-Thr!l-allo-Thr
1.865ꢁ0.154
1.866ꢁ0.069
92.5ꢁ18.4
1.37
1.37
185.5ꢁ11.9
Repetitive batch processes (20-mL scale) were per-
formed for the production of d- and l-allo-Thr, re-
d- or l-Thr. The process was performed in a repetitive spectively. Process data and yields for both enantio-
batch mode, in which enzymatic isomerization of Thr mers are summarized in Table 3 and Table 4. Produc-
to allo-Thr and crystallization of allo-Thr took place tion of d-allo-Thr was performed over 55 days and
simultaneously in one single reaction vessel at 308C consisted of 12 batches with yields of 1.6 to 4.4 g d-
(Figure 5). Depending on the desired allo-Thr enan- allo-Thr per batch and a diastereomeric excess (deallo
tiomer, either l- or d-Thr was used as the substrate. between 96.9 and 100%. A space-time yield (STY) of
The process consists of three main steps:
28 gLꢀ1 dꢀ1 and a final yield of 30.8 g d-allo-Thr
)
I) Start of the first batch by addition of AArac1299 (deallo >99.2%) were achieved. An initial amount of
to a saturated Thr solution (119 gLꢀ1 at 308C in water, 10 mg AArac12996 was applied and additional 2 mg
the pH was titrated to 8.5 using 10N NaOH) contain- of AArac12996 were added after 13 days.
ing undissolved excess Thr: Thr is converted to allo-
Production of l-allo-Thr was performed over 56
Thr until the saturation limit of allo-Thr (83 gLꢀ1 at days and consisted of 12 batches with yields of 1.5 to
308C in water, pH 8.5) is reached. The concentration 4.0 g l-allo-Thr per batch and a deallo between 94.6
of Thr remains constant (due to progressive solubili- and 100%. An STY of 28.8 gLꢀ1 dꢀ1 and a final yield
zation of excess Thr).
II) Crystallization of allo-Thr: allo-Thr starts to using a total amount of 20 mg AArac12996.
crystallize spontaneously upon exceeding the solubili- The enzyme showed a remarkable high stability
of 32.2 g l-allo-Thr (deallo > 98.4%) were obtained
ty limit. The composition of the solution is 42% under process conditions yielding 2.56 g d-allo-Thr
d-allo-Thr and 58% l-Thr due to different solubilities. and 1.62 g l-allo-Thr per mg AArac12996, respective-
This process is driven by the conversion of Thr to ly.
allo-Thr until depletion of excess Thr.
The activity of AArac12996 in the crystallization
III) Complete conversion of excess Thr to allo-Thr: solution was monitored throughout the process.
The process stops upon complete solution and isomer-
Table 3. Production of d-allo-Thr by repetitive batch crystal-
lization and enzymatic isomerization of l-Thr (V=20 mL,
308C, pH 8.5, initial amount of AArac12996: 10 mg, addi-
tion of 2 mg AArac12996 after 13 days).
[b]
Batch Process
Added l- Batch
Yield[a] deallo
no.
time [d]
Thr [g]
time [h] [g]
[%]
1
2
3
4
5
6
7
8
5.9
8.8
7
2
4
2
3
3
2.5
3
2
2
4
2.5
37.0
141
69
120
53
94
93
74
94
77
73
3.05
1.68
3.55
1.97
3.08
2.84
2.08
2.6
1.83
1.63
4.41
2.10
30.82
100.0
100.0
98.0
96.9
98.3
100.0
99.7
99.3
100.0
99.4
99.9
99.7
99.2
13.8
16.0
19.9
23.8
26.8
30.8
34.0
37.0
47.8
54.8
9
Figure 5. Repetitive batch crystallization (20 mL) of d-allo-
Thr in one single reaction vessel. I: AArac12996 is added to
a saturated l-Thr solution (starting condition). II: dynamic
process of crystallization (d-allo-Thr), solubilization (l-Thr)
and isomerization (l-Thr!d-allo-Thr). III: addition of fresh
l-Thr when excess has depleted. Steps II and III are repeat-
ed for repetitive batch crystallization.
10
11
12
258
168
1314
sums 54.8
[a]
Yield of crystallized d-allo-Thr.
Diastereomeric excess of d-allo-Thr in crystals.
[b]
2434
asc.wiley-vch.de
ꢂ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2011, 353, 2431 – 2438

An Efficient Route to Both Enantiomers of allo-Threonine
Table 4. Production of l-allo-Thr by repetitive batch crystal-
lization and enzymatic isomerization of d-Thr (V=20 mL,
308C, pH 8.5, AArac12996: 20 mg).
Conclusion
Enantiopure l- and d-allo-Thr were produced by en-
zymatic isomerization of either d- or l-Thr and simul-
taneous crystallization. Recovery of the enantiomeri-
cally pure products was performed by simple filtration
which reduced downstream processing significantly
compared to, for example, chromatographic meth-
ods.[34]
The reaction was catalyzed by AArac12996 from P.
putida, which was cloned and overexpressed in E.
coli. As the enzyme was secreted into the medium, its
purification is easy. Vmax and Km values were deter-
mined for both Thr enantiomers. Thereby Vmax was
found to be similar for both reactions (1.87 U m-
gꢀ1 AArac12996), whereas the Km value, which is a
measure for the affinity for both enantiomers, differs
by a factor of 2 (Table 2), indicating a higher affinity
[b]
Batch Process
Added d- Batch
Yield[a] deallo
no.
time [d]
Thr [g]
time [h] [g]
[%]
1
2
3
4
5
6
7
8
9.8
9
2
3
3
3
4
3
2
2
4
2.5
3
40.5
236
70
74
99
70
99
92
77
73
258
75
117
1339
3.94
1.98
2.98
3.10
2.55
3.32
2.80
2.06
1.79
4.04
1.54
2.32
32.42
99.6
99.0
98.6
99.6
98.4
94.6
95.8
100.0
99.7
99.8
99.7
96.5
98.4
12.7
15.8
19.9
22.8
27.0
30.8
34.0
37.1
47.8
50.9
55.8
9
10
11
12
sums 55.8
[a]
for l-Thr (Km,l-Thr =92.5 mm,
Km,d-Thr =185.5 mM).
Yield of crystallized l-allo-Thr.
Diastereomeric excess of l-allo-Thr in crystals.
[b]
AArac12996 shows highest initial rate activity at a pH
of about 9 and a temperature of 318C. Surprisingly,
the enzyme was completely stable at 308C for at least
7 days. Thus, this temperature was chosen for the pro-
duction of d- and l-allo Thr.
d- and l-allo-Thr were crystallized with high puri-
ties in repetitive batch processes starting from saturat-
ed Thr solutions. Total amounts of 30.8 g d-allo-Thr
and 32.4 g l-allo-Thr were obtained with a very good
diastereomeric excess of ded-allo >99.2% and del-allo
>
98.4%, respectively, and in good yield (d-allo-Thr:
83%, l-allo-Thr: 79%). Final space-time yields of
28.0 gLꢀ1 dꢀ1 for d-allo-Thr and 28.8 gLꢀ1 dꢀ1 for
l-allo-Thr were reached. A conversion of almost
100% could easily be achieved by additional batches.
The enzymeꢁs remarkable high stability under process
conditions resulted in enzyme specific yields of 2.56 g
d-allo-Thr per mg AArac12996 and 1.62 gL-allo-Thr
per mg AArac12996.
Figure 6. Deactivation and half life (t1/2) of AArac12996
during repetitive batch crystallization of d-allo-Thr. Activi-
ties were determined using the standard assay. kd =0.0218
dꢀ1 (ꢁ0.0035 dꢀ1).
Compared to chemical synthesis methods for allo-
Thr, which contain several reaction steps each fol-
lowed by chromatographic purification of the inter-
Figure 6 shows the relative activities of AArac12996 mediate,[13] our process consists of only one enzyme-
during d-allo-Thr crystallization over a period of catalyzed reaction (isomerization of bulk Thr) and si-
55 days. The deactivation constant kd was calculated multaneous crystallization in one single reaction
from
vessel. The process was carried out at moderate tem-
perature (308C) in water and avoids any toxic or
harmful compounds. Product recovery was performed
by simple filtration of the crystallization solution and
subsequent crystal drying, which made downstream
activity. processing very comfortable. In contrast to other bio-
catalytic production and separation methods for allo-
where
v0
is
the
initial
enzyme
Thr[34] we could achieve very high purities with deallo
>
99% and theoretically approach complete conversion
of the substrate.
Half-life t1/2 of the enzyme was derived from
Adv. Synth. Catal. 2011, 353, 2431 – 2438
ꢂ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
2435

Kerstin Wꢀrges et al.
FULL PAPERS
PLP, pH 7) as described above, concentrated to a final
volume of 120 mL, frozen in liquid nitrogen and lyophilized
for 24 h (freeze-dryer: Christ Alpha 2-4, Germany). The
freeze-dried enzyme (320 mg, 73% protein content) was
stored at ꢀ208C.
Experimental Section
Cloning of AArac12996
The AArac12996 gene was amplified by standard PCR from
the genomic DNA of Pseudomonas putida NBRC12996
which was kindly provided by the Department of Biotech-
nology, Kansai University (Japan). Genomic DNA was iso-
lated from an overnight culture grown at 308C in LB-
Medium using the DNeasy Blood & Tissue Kit (Qiagen).
Following primers (Eurofins MWG Operon, Ebersberg,
Germany) have been used for cloning:
Activity Assay
The specific activity of AArac12996 was investigated by de-
termination of the initial reaction rate for the isomerization
of l-Asn. Therefore, the linear slope of product (d-Asn) for-
mation was determined within a conversion range <10%.
Specific activities were determined using a standard activity
assay and following the increase of d-Asn by chiral HPLC.
One unit of activity was defined as the amount of
AArac12996 which catalyzes the isomerization of 1 mmol l-
Asn to d-Asn per minute under standard conditions
(190 mM l-Asn in water, 358C). Relative activities for in-
vestigation of pH-optimum and temperature dependency as
well as for stability investigations were determined by fol-
lowing the decrease of optical rotation with a polarimeter.
In order to allow optimal solubility and crystallization in the
allo-Thr production process, all measurements (except for
the determination of the pH-optimum) were performed in
water without buffer salts.
The PCR-fragment was purified on an agarose gel (1%)
and digested with NdeI and XhoI (Fermentas, St. Leon-Rot,
Germany). Subsequently the AArac12996 gene was ligated
into the NdeI/XhoI digested vector pET22b+ (Novagenꢃ,
Merck KGaA, Darmstadt, Germany) yielding the final con-
struct pET22b-AArac12996. The expression host E. coli
Assay conditions HPLC: 950 mL l-Asn in water
(200 mM), 50 mL AArac12996 containing solution, To stop
the reaction 10 mL of assay solution were diluted with
990 mL HClO4 (pH 1) prior to injection. Column: Daicel
Crownpak CR(+) 0.4ꢄ15 cm; mobile phase: HClO4 (pH 1);
BL21ACHTUNGTRENNUNG(DE3) was transformed with pET22b-AArac12996 by
electroporation. Sequence identity was verified by DNA se-
quencing (Agowa, Berlin, Germany). For protein and DNA-
sequences see Supporting Information.
flow: 0.6 mLminꢀ1
; column temperature: 08C. HPLC-
system: Agilent Series 1100.
Assay conditions polarimeter: 1 mL l-Asn solution in
water (90 mM if not indicated otherwise), 110 mL
AArac12996 solution (1 mgmLꢀ1). Polarimeter: Jasco P-
2000; cell length=100 mm, l=365 nm, t=5 min, 308C
(varying temperatures for determination of temperature de-
pendency).
Expression and Purification
Cultivation of the recombinant E. coli BL21ACHTNUTRGNE(NUG DE3) strain
was performed in 5ꢄ400 mL of LB medium (10 gLꢀ1 pep-
tone, 5 gLꢀ1 yeast extract, 10 gLꢀ1 NaCl) at 378C and
100 rpm in 2-L shaking flasks. Each flask was inoculated
with 4 mL of an overnight pre-culture. At OD600 ~0.6 pro-
tein expression was induced by addition of IPTG (100 mM)
and the temperature was decreased to 278C. 67 h after in-
duction the cells were separated from the racemase contain-
ing medium by centrifugation (15000ꢄg, 20 min, 48C). Sepa-
ration of AArac12996 from low molecular weight medium
components was performed by ultrafiltration (ultrafiltration
device manufactured by mechanical workshop of IBT, For-
schungszentrum Jꢀlich GmbH, Germany; membrane cutoff:
10 kDa) by concentrating the medium to 10% of its initial
volume three times and refilling with potassium phosphate
buffer (10 mM KPi, 10 mM PLP, pH 7). Occurring precipi-
tates were separated by centrifugation and the supernatant
was filtered (syringe filter, 0.4 mm) before applying onto an
anion-exchange column [DEAE-Sepharose FF, 5ꢄ25 cm,
column volume (CV): 500 mL, equilibrated with 10 mM KPi
(10 mM PLP, pH 7)]. Unbound proteins were washed from
the column with 2 CV of 10 mM KPi (10 mM PLP, pH 7).
AArac12996 was eluted with 10 mM KPi (10 mM PLP, pH 7)
containing 100 mM NaCl after about 3.8 CV at a flow rate
of 10 mLminꢀ1. Enzyme containing fractions were detected
at 280 nm (absorption of aromatic side chains of proteins)
and 420 nm (absorption of PLP). A total volume of 850 mL
was desalted by ultrafiltration using 10 mM KPi (10 mM
Kinetic Parameters
Vmax and Km for the isomerization reaction of Thr to allo-
Thr were calculated by non-linear regression using Origin
7G (OriginLab Corporation, Northampton, USA). The turn-
over number kcat was derived from Vmax for a single mono-
mer. All kinetic parameters are related to the relative pro-
tein mass determined (according to Bradford) in the
enzyme preparation. The identity of allo-Thr was verified by
1H NMR spectroscopy (Bruker, Avance DR X 600) in D2O
(600 MHz): d=1.14–1.15 (d, 3H, J=6.42 Hz, CH3), 3.77–
3.78 (d, 1H, J=3.77 Hz, CH), 4.28–4.32 (m, 1H, CH).
Determination of Protein Concentration
Protein concentrations were determined according to Brad-
ford[35] using BSA as standard.
Investigation of Temperature Dependency and pH-
Optimum
The temperature dependency of the initial rate activity of
AArac12996 (0.1 mgmLꢀ1) was determined for l-Asn
2436
asc.wiley-vch.de
ꢂ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2011, 353, 2431 – 2438

An Efficient Route to Both Enantiomers of allo-Threonine
(90 mM) in water between 208C and 408C. Initial reaction
rates were measured polarimetrically.
Thr crystallization was determined as described above with
50 mL of enzyme containing crystallization solution.
The pH-optimum of AArac12996 (0.1 mgmLꢀ1) was de-
termined polarimetrically by initial rate measurements in
glycine/NaCl buffer (90 mM each, titrated with NaOH) in
the range of pH 8.1 to 12.6 using l-Asn (25 mM) as a sub-
strate. It has to be noted that the substrate concentration
(25 mM) was significantly lower than the Km value for l-
Asn (>100 mM). However, higher concentrations of Asn
could not be applied, because the capacity of the buffer was
exceeded otherwise. Reaction rates were determined only
from the linear part of the conversion curve. Since the activ-
ities were not obtained under Vmax conditions, initial reac-
tion rates at different pH values are given as relative activi-
ties. All measurements were performed twice and relative
activities were calculated as mean values.
Acknowledgements
We kindly thank Prof. Tadao Oikawa from the Department
of Biotechnology of Kansai University (Japan) for providing
the Pseudomonas putida strain NBRC12996.
References
[1] L. A. Flippin, K. Jalali-Araghi, P. A. Brown, H. R. Bur-
meister, R. F. Vesonder, J. Org. Chem. 1988, 54, 3006–
3007.
[2] M. Kataoka, M. Wada, M. Ikemi, T. Morikawa, T.
Miyoshi, S. Shimizu, Ann. N. Y. Acad. Sci. 1998, 864,
318–322.
[3] M. J. Miller, Acc. Chem. Res. 1986, 19, 49–56.
[4] T. Nagahara, T. Kametani, Heterocycles 1987, 25, 729–
806.
[5] J. Burke, Terrence R. , M. Knight, B. Chandrasekhar,
Tetrahedron Lett. 1989, 30, 519–522.
The activation energy Ea of AArac12996 was derived
from the Arrhenius plot [ln(v) vs. Tꢀ1] as the product of the
slope m and the negative gas constant R (8.314 JKꢀ1 molꢀ1):
Ea =m·(ꢀR).
Stability Investigations
Enzyme stabilities were investigated for AArac12996
(1 mgmLꢀ1) in water at temperatures between 20 and 478C.
Aliquots were taken over up to 7 days and residual activities
were determined polarimetrically at 308C for 5 min using l-
Asn (90 mM) as a substrate.
[6] P. J. Maurer, C. G. Knudsen, A. D. Palkowitz, H. Rapo-
port, J. Org. Chem. 1985, 50, 325–332.
[7] J. L. Morell, P. Fleckenstein, E. Gross, J. Org. Chem.
1977, 42, 355–356.
Repetitive Batch Crystallization
[8] M. Shiozaki, N. Ishida, H. Maruyama, T. Hiraoka, Tet-
Repetitive batch crystallization was performed for l-and d-
allo-Thr. Here we describe the process for the crystallization
of d-allo-Thr, which is similar to the one for l-allo-Thr but
starting from the other Thr enantiomer. Experiments were
performed in a double jacketed 50-mL glass reactor which
was tempered by a water thermostat to 308C. A saturated l-
Thr solution was prepared by the addition of 7 g l-Thr to
20 mL water and the pH was adjusted to 8.5 with 10N
NaOH. Excess l-Thr remained as sediment. The conversion
of l-Thr to d-allo-Thr was performed by 12 mg AArac12996
(for l-allo-Thr crystallization 20 mg AArac12996 were
used). Crystallization of d-allo-Thr started upon exceeding
the solubility limit. The crystallization process was moni-
tored by HPLC as described below. After complete solubili-
zation of excess l-Thr, d-allo-Thr crystals were harvested by
filtration, washed with 2 mL of ice cold water and dried for
1 h at 608C. The next batch was started by addition of fresh
l-Thr to the remaining solution (Figure 5).
rahedron 1983, 39, 2399–2407.
[9] M. Shiozaki, N. Ishida, T. Hiraoka, H. Maruyama, Tet-
rahedron 1984, 40, 1795–1802.
[10] A. Guzman-Martinez, R. Lamer, M. S. VanNieu-
wenhze, J. Am. Chem. Soc. 2007, 129, 6017–6021.
[11] P. L. Beaulieu, Tetrahedron Lett. 1991, 32, 1031–1034.
[12] M. A. Blaskovich, G. A. Lajoie, Tetrahedron Lett. 1993,
34, 3837–3840.
[13] G. Cardillo, L. Gentilucci, A. Tolomelli, C. Tomasini, J.
Org. Chem. 1998, 63, 3458–3462.
[14] P. Lloyd-Williams, N. Carulla, E. Giralt, Tetrahedron
Lett. 1997, 38, 299–302.
[15] D. Pons, M. Savignac, J.-P. GenÞt, Tetrahedron Lett.
1990, 31, 5023–5026.
[16] S. Makart, M. Bechtold, S. Panke, J. Biotechnol. 2007,
130, 402–410.
[17] Y.-H. Lim, K. Yokoigawa, N. Esaki, K. Soda, J. Bacter-
iol. 1993, 175, 4213–4217.
[18] T. Yorifuji, K. Ogata, K. Soda, J. Biol. Chem. 1971, 246,
5085-5092.
Monitoring of the Crystallization Process
The composition of solution during allo-Thr crystallization
and the diastereomeric excess (deallo) of crystals were deter-
mined by HPLC employing an HP 1100 series equipped
with a chiral column (Daicel Crownpak CR(+) 0.4ꢄ15 cm;
mobile phase: HClO4 (pH 1); flow: 0.8 mLminꢀ1; column
temperature: 58C). deallo was calculated from deallo =([allo-
[19] K. Inagaki, K. Tanizawa, H. Tanaka, K. Soda, Agric.
Biol. Chem. 1987, 51, 173–180.
[20] K. Soda, S. Osumi, Methods Enzymol. 1971, ##17B,
629–636.
[21] G. M. Coppola, H. F. Schuster, Asymmetric Syntheses,
Construction of Chiral Molecules Using Amino Acids,
John Wiley & Sons, New York, 1987.
[22] K. Kino, M. Sato, M. Yoneyama, K. Kirimura, Appl.
Microbiol. Biotechnol. 2007, 73, 1299–1309.
[23] H. Miyazaki, H. Morita, T. Shiraiwa, H. Kurokawa,
Bull. Chem. Soc. Jpn. 1994, 67, 1899–1903.
Thr]ꢀ
are the crystal fractions of the respective diastereomer.
Space time yields (STY) were calculated from STY=yield/
(t·V) where t is the process time and V is the reaction
volume. The remaining activity of AArac12996 during allo-
ACHTUNGTRENNUNG[Thr])/ACHTUNGTRENNUNG([allo-Thr]+[Thr]) where [allo-Thr] and [Thr]
AHCTUNG-
TRENNUNG
Adv. Synth. Catal. 2011, 353, 2431 – 2438
ꢂ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
2437

Kerstin Wꢀrges et al.
FULL PAPERS
[24] T. Shiraiwa, K. Fukuda, M. Kubo, Chem. Pharm. Bull.
2002, 50, 287–291.
[25] K. Wꢀrges, K. Petrusevska-Seebach, M. P. Elsner, S.
Lꢀtz, Biotechnol. Bioeng. 2009, 104, 1235–1239.
[26] E. J. Ebbers, G. J. A. Ariaans, J. P. M. Houbiers, A.
Bruggink, B. Zwanenburg, Tetrahedron 1997, 53, 9417–
9476.
[27] J. D. Bendtsen, H. Nielsen, G. von Heijne, S. Brunak, J.
Mol. Biol. 2004, 340, 783–795.
[28] H. Nielsen, A. Krogh, Proceedings of the Sixth Interna-
tional Conference on Intelligent Systems for Molecular
Biology, 1998, 6, 122–130.
[30] D. Matsui, T. Oikawa, N. Arakawa, S. Osumi, F. Laus-
berg, N. Stꢅbler, R. Freudl, L. Eggeling, Appl. Micro-
biol. Biotechnol. 2009, 83, 1045–1054.
[31] T. Uo, T. Yoshimura, N. Tanaka, K. Takegawa, N.
Esaki, J. Bacteriol. 2001, 183, 2226–2233.
[32] K. Yokoigawa, Y. Okubo, H. Kawai, N. Esaki, K. Soda,
J. Mol. Catal. B: Enzym. 2001, 12, 27–35.
[33] K. Inagaki, K. Tanizawa, B. Badet, C. T. Walsh, H.
Tanaka, K. Soda, Biochemistry 1986, 25, 3268–3274.
[34] S. Makart, M. Bechthold, S. Panke, Chem. Eng. Sci.
2008, 63, 5347–5355.
[35] M. M. Bradford, Anal. Biochem. 1976, 72, 248–254.
[29] I. Suominen, M. Karp, M. Lahde, A. Kopio, T. Glum-
off, P. Meyer, P. Mantsala, Gene 1987, 61, 165–176.
2438
asc.wiley-vch.de
ꢂ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2011, 353, 2431 – 2438