A novel catalytic ability of γ-glutamylcysteine synthetase of Escherichia coli and its application in theanine production

Article abstract of DOI:10.1271/bbb.90538

γ-Glutamylcysteine synthetase (γGCS, EC 6.3.2.2) catalyzes the formation of γ-glutamylcysteine from L-glutamic acid (Glu) and L-cysteine (Cys) in an ATP-dependent manner. While γGCS can use various amino acids as substrate, little is known about whether it can use non-amino acid compounds in place of Cys. We determined that γGCS from Escherichia coli has the ability to combine Glu and amines to form γ-glutamylamides. The reaction rate depended on the length of the methylene chain of the amines in the following order: n-propylamine > butylamine > ethylamine methylamine. The optimal pH for the reaction was narrower and more alkaline than for the reaction with an amino acid. The newly found catalytic ability of γGCS was used in the production of theanine (γ-glutamylethylamine). The resting cells of E. coli expressing γGCS, in which ATP was regenerated through glycolysis, synthesized 12.1 mm theanine (18 h) from 429 mm ethylamine.

Full text of DOI:10.1271/bbb.90538

Biosci. Biotechnol. Biochem., 73 (12), 2677–2683, 2009
A Novel Catalytic Ability of ꢀ-Glutamylcysteine Synthetase
of Escherichia coli and Its Application in Theanine Production
y
1;
2
Koichiro MIYAKE and Shingo KAKITA
1Technical Research Laboratories, Kyowa Hakko Bio Co., Ltd., 1-1 Kyowa-cho, Hofu, Yamaguchi 747-8522, Japan
2Tokyo Research Park, Kyowa Hakko Kirin Co., Ltd., 3-6-6 Asahi-cho, Machida, Tokyo 194-8533, Japan
Received July 23, 2009; Accepted September 27, 2009; Online Publication, December 7, 2009
[doi:10.1271/bbb.90538]
ꢀ-Glutamylcysteine synthetase (ꢀGCS, EC 6.3.2.2)
catalyzes the formation of ꢀ-glutamylcysteine from
L-glutamic acid (Glu) and L-cysteine (Cys) in an ATP-
dependent manner. While ꢀGCS can use various amino
acids as substrate, little is known about whether it can
use non-amino acid compounds in place of Cys. We
determined that ꢀGCS from Escherichia coli has the
ability to combine Glu and amines to form ꢀ-glutamy-
lamides. The reaction rate depended on the length of the
methylene chain of the amines in the following order:
n-propylamine > butylamine > ethylamine ꢀ methyl-
amine. The optimal pH for the reaction was narrower
and more alkaline than for the reaction with an amino
acid. The newly found catalytic ability of ꢀGCS was
used in the production of theanine (ꢀ-glutamylethyl-
amine). The resting cells of E. coli expressing ꢀGCS, in
which ATP was regenerated through glycolysis, synthe-
sized 12.1 mM theanine (18 h) from 429 mM ethylamine.
in the reaction conditions have been reported.13) ꢀ-
Glutamyltranspeptidase can be also used in theanine
synthesis. Suzuki et al.14,15) explained that the ꢀ-
glutamyltranspeptidase from E. coli synthesized thea-
nine from Gln and ethylamine at pH 10. ꢀ-Amidation of
Glu can be achieved with glutamine synthetase (GS) and
with related enzymes (Fig. 1b). ꢀ-Amidation of Glu
needs ATP for the reaction, but Glu is stable and lower-
priced than Gln. GS originating from Micrococcus
glutamicus has been found to synthesize theanine from
Glu and ethylamine in the presence of ATP.16) GS from
Pseudomonas taetrolens Y-30 has also been reported to
have the same activity.17) Susana et al.18) explained that
IpuC (ꢀ-glutamylamide synthetase) of Pseudomonas sp.
shared homology with GS and retained theanine synthe-
sizing activity through ꢀ-amidation of Glu. Recently,
ꢀ-glutamylmethylamide synthetases (EC 6.3.4.12) from
Methylophaga sp. AA-30 and from Methylovorus mays
No. 9 were found to be efficient theanine-producing
enzymes.19,20) The amino acid sequence of the latter
enzyme showed significant identity to GS.21)
Key words: ꢀ-glutamylcysteine synthetase; theanine;
ꢀ-glutamylamide; amine
ꢀ-Glutamylcysteine synthetase (ꢀGCS, EC 6.3.2.2)
catalyzes the ATP-dependent synthesis of ꢀ-glutamyl-
cysteine from Glu and Cys, the first step in the reaction
of glutathione formation.22,23) The gshA gene encoding
ꢀGCS has also been identified in a variety of organism,
including E. coli.24) Biochemical studies have revealed
that ꢀ-GCS can ꢀ-glutamylate Cys and also many other
kinds of amino acids, such as alanine and ꢁ-amino-
butyrate,25) but little is known about whether the enzyme
can accept other non-amino acid compounds.
Theanine (ꢀ-glutamylethylamide) is an amino acid
found in green tea that enhances the umami taste.1)
Besides its graceful flavor, many physiological functions
of it have been reported, including mind relaxation,2)
improvement in sleep disturbances,3) suppression of
blood pressure elevation,4) anti-oxidation of LDL cho-
lesterol,5) modulation of neurotransmission,6) neuropro-
tection,7) anti-hypertension,8) enhancement of antitumor
activity,9) and improved recognition.10)
In the tea plant, theanine is synthesized by theanine
synthetase (EC 6.3.1.6), with Glu, ethylamine, and ATP
as substrates,11) but it is very labile, so it cannot be used
in industrial theanine production. Various methods have
been employed for theanine production, including
extraction from tea leaves, chemical synthesis, and
enzymatic synthesis. The enzymatic methods can be
classified into two groups. One starts from L-glutamine
(Gln) using ꢀ-glutamyltransferase activity, and the other
starts from Glu using ꢀ-amidation activity. Yamada
et al.12) have reported theanine synthesis from Gln and
ethylamine (Fig. 1a) using the ꢀ-glutamyl transfer
activity of the glutaminase of Pseudomonas nitroredu-
cens. Since glutaminase also catalyzes the hydrolysis of
the substrate, Gln, a high concentration of ethylamine
and highly alkaline pHs are necessary. Improvements
The fact that both GS and ꢀGCS form amide bonds
using the ꢀ-carboxy group of Glu prompted us to try to
determine whether ꢀGCS can react with ethylamine to
yield theanine. In this study, we found that ꢀGCS can
synthesis theanine with ethylamine. Our results also
indicate that certain amines are preferred substrates
for ꢀGCS.
Materials and Methods
Materials. General biochemical reagents were purchased from
Wako Pure Chemical Industries (Osaka, Japan). The amines were from
Tokyo Chemical Industry (Tokyo). E. coli K12 W3110 strain
(GenBank Accession no. AP009048) was from NITE Biological
Resource Center (NBRC, Tokyo). E. coli K12 DH5ꢁ strain was from
Toyobo (Osaka). Various restriction enzymes and Taq polymerases for
PCR and E. coli BL21 strain were from Takara Bio (Ohtsu, Japan).
y
To whom correspondence should be addressed. Tel: +81-835-22-2518; Fax: +81-835-22-2466; E-mail: koichiro.miyake@kyowa-kirin.co.jp
Abbreviations: ꢀGCS, ꢀ-glutamylcysteine synthetase; Glu, L-glutamic acid; Cys, L-cysteine; Gln, L-glutamine; GS, glutamine synthetase

2678
K. MIYAKE and S. KAKITA
a
O
O
O
NH3
R-NH2
HOOC
HOOC
R
HOOC
NH
2
O
Enz
NH
NH
2
NH
2
NH2
Enz -OH
Enz -OH
+
γ
-Glutamyl-enzyme intermediate
Gln
b
O
O
ε-
O
ATP
R-NH2
HOOC
HOOC
HOOC
OH
O-PO
-
O-PO3-
3
NH2-R
NH
2
NH
3
NH3
ε+
Glu
transition state
R-NH2: NH3 (GS)
γ
L-Cys ( GCS)
O
HOOC
R
NH
NH2
+ADP,Pi
Fig. 1. Reaction Mechanisms of ꢀ-Glutamylamide Bond-Producing Enzymes.
a, ꢀ-glutamyl transfer reaction of glutaminase or ꢀ-glutamyltranspeptidase. b, ꢀ-amidation of Glu by GS and by ꢀGCS.
Expression vector pTrS33 was previously reported in our patent.26) The
plasmid purification kit and pQE80L vector were from Qiagen (Hilden,
Germany).
altered to ATG. Primer D was added to the BamHI recognition
sequence. The lac promoter region was amplified from expression
vector pUC19 by PCR using primer E, 50-GGAAGAGAATTCAA-
TACGCAAACCGCCTCT-30, which was added to the EcoRI recog-
nition sequence, and Primer F, 50-GGTCATAAGCTTTTCCTGTGT-
GAAATTGTTAT-30, which was added to the HindIII recognition
sequence. Each of the PCR products was digested with restriction
enzymes, and the 1.6-kb gshA fragment and the 0.3-kb lac promoter
fragment were respectively recovered using the Geneclean Kit
(Funakoshi, Tokyo). Expression vector pTrS33 was double digested
with BamHI and EcoRI. GshA, the lac promoter fragment, and the
3.16-kb vector fragment were subjected to ligation using a Ligation kit
(Takara Bio) at 16 ꢀC for 16 h. E. coli DH5ꢁ was transformed using the
ligation reaction mixture by standard methods, and the resulting
transformant was spread on LB agar medium containing 50 mg/ml of
ampicillin and cultured overnight at 30 ꢀC. The recombinant plasmid
was extracted from a colony of the transformant, and it was confirmed
that the plasmid contained the inserted lac promoter region and gshA
by sequence analysis. The expression vector was designated pGSK1.
The structure of pGSK1 is shown in Fig. 2. The E. coli DH5ꢁ carrying
the plasmid was designated E. coli DH5ꢁ/pGSK1.
Expression and purification of E. coli (His)6-tagged ꢀGCS. The
gshA gene (KEGG Accession no. JW2663 http://www.genome.jp/
kegg/) was amplified from E. coli W3110 genome by PCR using
primer A, 50-CGCGGATCCATGATCCCGGACGTATCACAGGCG-
CTG-30, and primer B, 50-GCACCCAAGCTTTGCAGGCGTGTTTTT-
CCAGCCACACCGCA-30. Primer A was added to the BamHI recog-
nition sequence, and the initiation codon for the gene, which begins
with TTG, was altered to ATG in expectation of improved translation
efficiency. Primer B was added to the HindIII recognition sequence.
Expression vector pQE80L and the PCR product were digested with
BamHI and HindIII. Both DNA fragments were ligated, and the
resulting plasmid was introduced into E. coli DH5ꢁ. Recombinant
E. coli DH5ꢁ was cultivated in LB medium (10 g/l Bacto-tryptone,
BD (New Jersey, USA), 5 g/l yeast extract, BD, and 5 g/l NaCl)
containing 100 mg/l of ampicillin at 30 ꢀC overnight with shaking. The
culture was transferred to fresh LB medium and incubated at 25 ꢀC.
Four h after transfer, 0.2 mM IPTG (isopropyl-ꢂ-D-thiogalactopyrano-
side) was added, and cultivation was continued for 16 h. Cells were
collected by centrifugation, and the cell pellet was suspended in 50 mM
phosphate buffer (pH 7.8) and disrupted by sonication at 4 ꢀC. The
cellular debris was removed by centrifugation and the supernatant was
subjected to further purification using HiTrapꢀHP (GE Healthcare,
Sweden). The protein concentration was determined using the Quick
Startꢀ Bradford Dye Reagent (Bio-Rad, California, USA) with bovine
serum albumin as the standard.
Production of theanine using a treated culture as the enzyme
source. Recombinant E. coli DH5ꢁ/pGSK1 was inoculated into 40 ml
of LB medium containing 100 mg/l ampicillin and subjected to
shaking culture in a 300-ml Erlenmeyer flask at 30 ꢀC overnight. After
culture completion, the cells recovered by centrifugation of the culture
were suspended in 100 mM/l Tris–HCl (pH 8.0), and the cell
concentration was adjusted so that the OD660 was 70 as measured
with a spectrophotometer, and this was designated the cell-containing
solution. To the cell-containing solution was added xylene at a
concentration of 10 ml/l, and the mixture was vortexed for 15 min to
obtain treated cells. To 100 ml of the treated cells, 5 g/l MgSO4, 3.5 g/l
K2SO4, 143 mg/l NADþ, 66 mg/l FMN, 12.1 g/l ATP, 2.5 g/l
monosodium L-glutamate monohydrate (MSG), and 20 g/l ethylamine
hydrochloride were added, and the resulting reaction mixture (total
volume, 200 ml) was reacted at 37 ꢀC for 60 min. After completion of
Construction of a strain expressing ꢀGCS for theanine synthesis. A
strain expressing ꢀGCS was constructed by cloning gshA by the
following method: First, gshA was amplified from the E. coli W3110
genome by PCR using primer C, 50-CGGGAGAAGCTTATGATCCC-
GGACGTATCACAG-30, and primer D, 50-TTTTCTGGATCCTTA-
GGCGTGTTTTTCCAGCC-30. Primer C was added to the HindIII
recognition sequence, and the initiation codon in this primer, TTG, was

Novel Catalytic Ability of ꢀ-Glutamylcysteine Synthetase
2679
boiling at 100 ꢀC for 5 min. The reaction mixture was analyzed by
HPLC under the conditions described above, except that the mobile
phase was 8–18% acetonitrile aqueous solution containing 2 g/l sodium
1-heptanesulfonate (adjusted to pH 2.0 with phosphoric acid).
The phosphate released from ATP was monitored by spectrophoto-
metric assay (P-Test Wako, Wako Pure Chemical Industries, Tokyo) as
an indicator of ATP-dependent synthetic activity. Background rates
were determined from samples containing everything but ꢀGCS.
EcoRI
0.0
HindIII0.2
EcoRI 0.4
Plac
Pst
I 3.4
Ap
pGSK1
4.16kb
gshA
Bgl II 1.4
Tlpp
Identification of ꢀ-glutamylamide structure by MS and NMR. The
reaction products from ethylamine, propylamine, and allylamine were
purified as follows: The 1 ml reaction mixtures contained 60 mM ATP,
100 mM MgSO4, 5 g/l MSG, 10 g/l amines and 100 mg/l (His)6-tagged
ꢀ-GCS in 0.1 M Tris–HCl buffer (pH 8.0). The reaction was carried out
at 37 ꢀC for 60 min, and was stopped by boiling at 100 ꢀC for 5 min.
After completion of the reaction, the contents of the reaction mixture
were detected by HPLC analysis under the following conditions: The
mobile phase was 3–5% acetonitrile aqueous solution containing 1 g/l
sodium heptafluorobutyric acid (adjusted to pH 2.0 with phosphoric
acid). A Develosil ODS-HG5 (Nomura Chemical, Tokyo, 4:6 ꢁ 250
mm) column was used. The temperature was 40 ꢀC and the flow rate
was 0.9 ml/min. Products were detected by absorption at 210 nm. The
products were purified by HPLC and collected from the fractions of
interest, and then the solution was evaporated using a Centrifugal
Concentrator VC-360 (Taitec, Koshigaya, Japan). The samples ob-
tained were analyzed by MS and NMR. Mass spectrometry analysis was
performed on two types of equipment, a ZMD 2000 (Waters, MA) and a
LTQ Orbitrap (Thermo Fisher Scientific, MA), operated in positive
electrospray ionization mode. On the former single-quadrupole mass
spectrometer, each sample was dissolved in water and analyzed by flow
injection analysis-mass spectrometry (FIA-MS) to confirm the molecu-
lar weight. The pump was set to a flow rate of 0.25 ml/min, and the
mobile phase was 50% acetonitrile containing 0.1% v/v formic acid.
The latter spectrometer was used to obtain high-resolution mass spectra.
Each sample solution was diluted with 50% acetonitrile containing
0.1% v/v formic acid, and was injected with an automatic flow syringe
at a flow rate of 10 ml/min. The NMR spectra were recorded at
500.13 MHz for 1H and at 125.76 MHz for 13C on a Bruker DMX 500
spectrometer (Bruker Biospin, Karlsruhe, Germany) in D2O at 30 ꢀC.
The structures of the ꢀ-glutamylamides were identified by analysis
of the NMR spectra (1H, 13C, 1H–1H COSY, NOESY, HSQC,
and HMBC). Data processing was carried out using the program
XWINNMR (Bruker Biospin).
BamHI 1.8
BglII 1.9
Fig. 2. Structure of ꢀGCS Expression Vector pGSK1.
gshA, ꢀGCS gene origin of E. coli W3110; Plac, lac operon
promoter region; Tlpp, terminator region of lipoprotein gene from
E. coli W3110; Ap, ampicillin resistance gene.
the reaction, theanine was detected and quantitatively determined by
HPLC under the following conditions: The mobile phase was 10%
acetonitrile aqueous solution containing 2 g/l sodium 1-heptanesulfo-
nate (adjusted to pH 2.0 with phosphoric acid). The column was a
Develosil ODS-HG5 (Nomura Chemical, Tokyo, 4:6 ꢁ 250 mm), the
temperature was 40 ꢀC, and the flow rate was 0.9 ml/min. Products
were detected by absorption at 210 nm.
Next, recombinant plasmid pGSK1 was introduced into E. coli
BL21 by a standard method. BL21/pGSK1 was spread on LB agar
medium containing 100 mg/l ampicillin and this was subjected to static
culture at 30 ꢀC overnight. The cells that grew on the medium were
inoculated into 300 ml of a pre-culture medium (6% corn steep liquor,
1.15% MSG, 0.2% lactic acid, 200 mg/l casamino acid, 5 mg/l vitamin
B1, pH 7.2) in a 2 liters Erlenmeyer flask, and cultured at 28 ꢀC at
220 rpm for 20 h. The culture obtained (2.25 ml) was inoculated into
one liter of a seed medium (2% corn steep liquor, 0.5% soybean
peptide, (SMS, Fuji Oil, Osaka), 1.5% dipotassium hydrogen phos-
phate, 0.1% sodium chloride, 0.6% ammonium sulfate, 0.1% glycine,
0.06% arginine hydrochloride, 4.95 mg/l ferrous sulfate, 4.4 mg/l zinc
sulfate, 1.97 mg/l copper sulfate, 360 mg/l manganese chloride,
440 mg/l sodium borate, 185 mg/l ammonium molybdate, 5 mg/l
vitamin B1, 5 mg/l nicotinic acid, 20 mg/l leucine, 20 mg/l threonine,
20 mg/l tryptophan, 0.01% LG109 (Asahi Denka, Tokyo), 1% glucose,
0.05% magnesium sulfate, and 100 mg/l ampicillin (pH 6.5) in a 2
liters jar, and cultured with aeration at a rate of 1 l/min and agitation at
a speed of 800 rpm at 30 ꢀC for 8 h.
Results
The culture obtained (28 ml) was inoculated into 1 liter of a
production medium (2.25% corn steep liquor, 0.55% SMS, 1.68%
dipotassium hydrogen phosphate, 0.115% sodium chloride, 0.68%
ammonium sulfate, 5.57 mg/l ferrous sulfate, 4.95 mg/l zinc sulfate,
2.21 mg/l copper sulfate, 405 mg/l manganese chloride, 495 mg/l
sodium borate, 208 mg/l ammonium molybdate, 5.6 mg/l vitamin B1,
5.6 mg/l nicotinic acid, 22 mg/l leucine, 22 mg/l threonine, 22 mg/l
tryptophan, 0.018% LG109, 1.26% glucose, 0.08% magnesium sulfate,
and 100 mg/l ampicillin (pH 6.5)) in a 2 liters jar and this was cultured
with aeration at a rate of 1 l/min and agitation at a speed of 800 rpm at
30 ꢀC controlling the pH to 6.5 with 28% aqueous ammonia. During
culture, 340 ml of a sugar solution (57.7% glucose and 0.188 g/l
calcium chloride) for feeding was added at a fixed flow rate. Culturing
was terminated after 30 h, and 7.5 ml/l xylene was added, followed by
agitation for 10 min to obtain treated matter of the culture. To 700 ml
of the treated matter were added 4.4 g/l of magnesium sulfate, 3 g/l of
potassium sulfate, 438 mg/l of ATP, 62.5 mg/l of NADþ, 27.5 mg/l of
FMN, 81 g/l of glucose, 35 g/l of ethylamine hydrochloride, and 70 g/l
of MSG, and the mixture was aerated at a rate of 0.7 ml/min with
agitation at 950 rpm at 34 ꢀC for 18 h controlling the pH to 7.2 with a
sodium hydroxide solution. The reaction product was analyzed by
HPLC under the conditions described above.
Formation of theanine by ꢀGCS
The gshA gene encoding ꢀGCS was amplified from
the E. coli W3110 genome by PCR and was ligated with
expression vector pQE80L. The resulting plasmid was
introduced into E. coli DH5ꢁ. Recombinant E. coli
DH5ꢁ was cultivated in LB medium and disrupted by
sonication. Cellular debris was removed by centrifuga-
tion, and the supernatant was subjected to further
purification using HiTrapꢀHP. His-tagged ꢀGCS pro-
tein was eluted using 500 mM imidazole buffer, and was
confirmed to be a single band on SDS–PAGE analysis
(Fig. 3). ꢀGCS activity was confirmed by checking ꢀ-
glutamylcysteine and ꢀ-glutamylaminobutyrate-forming
ability using Glu and Cys and Glu and ꢁ-aminobutyrate
respectively (Fig. 4a, b).
The His-tagged ꢀGCS was incubated with Glu,
ethylamine, and ATP. HPLC analysis revealed a new
peak that had the same retention time as theanine
(Fig. 4g). Reactions without the enzyme and without
ATP did not yield peak (data not shown). The peak was
fractionated and subjected to MS and NMR analysis.
From a high-resolution ESI mass spectrum, ½M þ Hꢂþ
was measured at m=z 175.1086, corresponding to the
Substrate specificities of ꢀGCS for amines. The substrate specificity
of ꢀGCS was assayed as follows: The 0.2 ml reaction mixtures
contained 15 mM ATP, 100 mM MgSO4, 5 g/l MSG, 2 or 10 g/l amines,
and 100 mg/l (His)6-tagged ꢀGCS in 0.1 M Tris–HCl buffer (pH 8.5).
The reaction was carried out at 37 ꢀC for 60 min, and was stopped by

2680
K. MIYAKE and S. KAKITA
O
a
1
5
HO C
2
4
2
1'
3
N
H
1
1
H- H COSY
HMBC
2'
NH
2
O
b
c
1
5
HO C
2
4
2
1'
3'
3
N
H
2'
NH
2
O
1
5
HO C
2
2
4
3
1'
3'
N
H
1 2 3 4 5 6 7
2'
NH
2
Fig. 3. SDS-Polyacrylamide Gel Electrophoresis Analysis of Cell
Lysate and Purified Enzyme Fractions.
Lane 1, marker; lane 2, IPTG induced cell extracts; lanes 3, 4, 5,
6, and 7, fractions eluted from the nickel column with 50, 100, 200,
350, and 500 mM of imidazole respectively.
Fig. 5. Summary of NMR Spectral, COSY, and HMBC Data.
Reaction products from ethylamine (a), n-propylamine (b), and
allylamine (c).
f
a
g
b
h
i
c
d
e
j
k
Fig. 4. HPLC Analysis of the Reaction Samples of ꢀGCS with Glu and Various Amines.
a, cysteine; b, ꢁ-aminobutyrate; c, allylamine; d, allylamine; same as (c) but no added ꢀGCS. e, n-propylamine; f, n-butylamine; g,
ethylamine; h, 2-amino-1-butanol; i, ethanolamine; j, isopropylamine; k, methylamine. Arrows indicate reaction products that cannot be found in
the absence of ꢀGCS.
molecular formula C7H14N2O3. NMR spectra (1H, 13C,
1H–1H COSY, NOESY, HSQC, and HMBC) showed
that ethylamine was connected to the ꢀ-carboxyl unit of
Glu (Fig. 5a, Table 1). These results clearly demonstrate
that the compound formed by the reaction was ꢀ-
glutamylethylamide (theanine), indicating that ꢀGCS
from E. coli can synthesize theanine from Glu, ethyl-
amine, and ATP.

Novel Catalytic Ability of ꢀ-Glutamylcysteine Synthetase
2681
Table 1. 1H and 13C-NMR Data for the ꢀGCS Product Using Amines as Substrates
1H-NMR
Ethylamine
(ppm)
Propylamine
(ppm)
Allylamine
(ppm)
J (Hz)
6.4
J (Hz)
6.5
J (Hz)
6.4
2
4.09
2.24
2.48
3.21
1.12
—
1H (t)
2H (m)
2H (m)
2H (q)
3H (t)
4.10
2.23
2.48
3.14
1.51
0.88
1H (t)
4.01
2.23
2.54
3.83
5.89
5.21
5.18
1H (t)
3
2H (m)
2H (m)
2H (t)
2H (m)
2H (m)
2H (dt)
1H (ddt)
1H (ddt)
1H (ddt)
4
10
20
30
7.3
7.3
6.9
7.4
5.1, 1.7
2H (m)
3H (t)
17.5, 10.4, 5.1
17.5, 1.6, 1.7
10.4, 1.6, 1.7
13C-NMR
a
b
L-Glu+ethylamine
L-Glu+L-Ala
Ethylamine
(ppm)
Propylamine
(ppm)
Allylamine
(ppm)
40
2.5
2.0
1.5
1.0
0.5
0.0
1
172.5
53.2
26.6
32.1
174.5
35.4
14.2
—
s
d
t
172.3
53.1
26.6
32.1
174.8
42.1
22.5
11.4
s
d
t
173.2
53.8
26.8
s
d
t
30
20
2
3
4
t
t
32.1
175.0
42.4
t
γ
5
s
t
s
t
s
t
10
10
20
30
q
t
134.4
116.4
d
t
0
0
q
0
5
10
15
20
5
10
15
20
Time (h)
Time (h)
Table 2. Production of Theanine in E. coli DH5ꢁ/pGSK1
Fig. 6. Production of Theanine (panel a) and ꢀ-Glutamylalanine
(panel b) Using E. coli BL21/pGSK1 in a Jar Fermenter.
.
Ethylamine HCl
(g/l)
MSG
(g/l)
Theanine
(mg/l)
Strain
DH5ꢁ/pGSK1
DH5ꢁ/pGSK1
DH5ꢁ/pGSK1
DH5ꢁ
0
20
20
20
0
—
54
theanine broth (data not shown), and it was identified as
11.5 mM (2.5 g/l) ꢀ-glutamylalanine. When 197 mM of
Ala was used instead of ethylamine, 165 mM of ꢀ-
glutamylalanine was produced within 10 h (Fig. 6b).
Production of 12.1 mM of theanine requires the same
amount of ATP, 15 times as much as the amount of ATP
added at the beginning of the reaction. These results
indicate that theanine was produced by resting cells of
E. coli overexpressing ꢀGCS from Glu and ethylamine
using the ATP-regeneration activity of the cell itself.
0
2.5
2.5
520
—
—, not detected <5 mg/l
Theanine production by resting cells of E. coli over-
expressing ꢀGCS
Next we examined to determine whether E. coli cells
overexpressing ꢀGCS might be useful for theanine
synthesis. E. coli DH5ꢁ/pGSK1 was cultivated and
permeabilized by xylene treatment, as described in
‘‘Materials and Methods.’’ The treated cells were mixed
with ethylamine, Glu, and ATP, and incubated at 37 ꢀC.
After a 60-min reaction, the reaction mixture was
analyzed by HPLC. As shown in Table 2, 3 mM of
theanine was synthesized from 245 mM of ethylamine
and 14.8 mM of Glu. Without pGSK1, no theanine was
detected (see Table 2, column for DH5ꢁ alone), show-
ing the necessity of overexpression of ꢀGCS. DH5ꢁ/
pGSK1 cells formed theanine without the addition of
Glu (Table 2), probably due to the available intracellular
pool of Glu. Since glucose was consumed by the xylene-
treated resting cells (data not shown), we surmiseed that
ATP might be regenerated through the sugar metabolism
of the cells. To explore this possibility, we conducted a
resting-cell reaction in the presence of glucose in a pH-
controlled manner. E. coli BL21/pGSK1 was cultured
in a 2-liter jar fermenter. After cultivation, xylene was
added to the cultivation fluid for permeabilization, and
429 mM of ethylamine, 414 mM of Glu, 0.8 mM of ATP,
and glucose were added. The pH condition was kept at
7.2 during the reaction. As shown in Fig. 3a, 12.1 mM
(2.1 g/l) theanine was detected in the 18 h broth
(Fig. 6a). Another HPLC peak was detected in the
ꢀ-Glutamylamide synthesizing ability of E. coli ꢀGCS
The substrate specificity of E. coli ꢀGCS for other
amines was examined. The reaction mixtures of His-
tagged ꢀGCS, Glu, amines, and ATP were analyzed by
HPLC. Significant peaks were detected when several
kinds of amines were used as substrates (Fig. 4). These
peaks were absent in the reactions that omitted ꢀGCS, as
exemplified in Fig. 4d, indicating that they were unique
reaction products. sec-Butylamine, 2-heptylamine, and
heptylamine did not yield new products. It should be
noted that the practical insolubility of 2-heptylamine and
heptylamine might have hampered the reaction.
The products from n-propylamine and allylamine
were purified and analyzed by MS and NMR. On MS
analysis, peaks at m=z 189 (½M þ Hꢂþ), 211 (½M þ
Naꢂþ), 377 (½2M þ Hꢂþ), and 399 (½2M þ Naꢂþ) were
detected in the n-propylamine sample, and peaks at m=z
187 (½M þ Hꢂþ), 209 (½M þ Naꢂþ), 373 (½2M þ Hꢂþ),
and 395 (½2M þ Naꢂþ) were detected from the al-
lylamine sample. The high-resolution ESI mass spec-
trum of the n-propylamine sample gave a ½M þ Hꢂþ at
m=z 189.1244, which corresponds to the molecular
formula of ꢀ-glutamylpropylamide, C8H16N2O3. The
high-resolution ESI mass spectrum of the allylamine

2682
K. MIYAKE and S. KAKITA
HS
120
100
NH2
Cysteine
n-Propylamine
Allylamine
HOOC
2g/l substrate
80
60
40
20
0
α
-Aminobutylate
NH2
NH2
Ethylamine
n-Propylamine
NH2
NH2
Methylamine
Ethylamine
10g/l substrate
n-Propylamine
Isopropylamine
Allylamine
5
6
7
8
9
10
11
NH2
(pH)
NH2
NH2
NH2
Butylamine
Fig. 8. Effects of pH on ꢀ-Glutamylamide Forming Reaction by
E. coli ꢀGCS.
HO
HO
3-Amino-1-propanol
Ethanolamine
ꢀ-Glutamyl compounds were determined by HPLC peak area, and
activity was expressed relative to the various maximum activities, as
100%. One hundred mg/l of purified His-tagged ꢀGCS protein was
incubated with 15 mM of ATP, 100 mM of MgSO4, 5 g/l of MSG,
and 2 g/l of amines in 0.1 M Tris–HCl buffer (pH 7 to 9.5) or 0.1 M
CAPS buffer (pH 9.2 to 10.3). The reaction was carried out at 37 ꢀC
for 60 min, and was stopped by boiling (100 ꢀC for 5 min). The
reaction mixture was analyzed by HPLC. Symbols: , ꢁ-amino-
butylate; , ethylamine; , propylamine.
0
20
40
60
80 100 120
(%: Relative amount of released phosphate with cysteine as 100%)
Fig. 7. Substrate Specificity of E. coli ꢀGCS at pH 8.5.
Each bar indicates phosphate concentration released in the
reaction mixtures with the substrate shown on the left. The
phosphate concentration is expressed relative to that of the reaction,
with cysteine as 100%. The substrate concentrations of cysteine,
n-propylamine, and allylamine were 2 g/l, and of the others, 10 g/l.
the ability to synthesize ꢀ-glutamylamide compounds
from Glu and amines. The reason our results differ from
Rathbun et al. is thought to be differences in the origin
of ꢀGCS or the reaction conditions. This is the first
report showing that ꢀGCS has ꢀ-amidation activity with
amines. Our finding should expand the possibility of
producing useful compounds using ꢀGCS and should
provide new clues towards elucidating the substrate
recognition mechanism of ꢀGCS.
sample gave ½M þ Hꢂþ at m=z 187.1086, which corre-
sponds to the molecular formula of ꢀ-glutamylallyla-
mide, C8H14N2O3. The NMR spectra (1H, 13C, H–1H
COSY, NOESY, HSQC, and HMBC) revealed that the
amines were connected to the ꢀ-carboxyl unit of Glu
in both samples (Fig. 5b, c, Table 1). Based on these
results, the products from the reaction of n-propylamine
and allylamine were concluded to be ꢀ-glutamylpropy-
lamide and ꢀ-glutamylallylamide respectively, indicat-
ing that ꢀGCS of E. coli can synthesize various ꢀ-
glutamylamide compounds from amines and Glu.
To quantify the reaction rates, the amounts of
phosphate released from ATP through the ꢀGCS
catalyzed reactions with Glu and amino acids and with
amines were monitored (Fig. 7). The amount of phos-
phate was highest when Cys was the substrate. The
amount of phosphate from allylamine and from n-
propylamine was 40–50% of Cys. The amount of
phosphate released was elevated by increasing the
concentrations of the amines (Fig. 7). The amount of
phosphate obtained from the reactions with the amines
depended on the length of the carbon chain of the
amines: n-propylamine (carbon chain length, 3) >
butylamine (4) > ethylamine (2) > methylamine (1).
The influence of pH on the activity was examined using
ꢁ-aminobutyrate, ethylamine, and propylamine as sub-
strates. It was determined that the optimum pH varied
with the substrate (Fig. 8). While the optimum pH for
ꢁ-aminobutyrate was broad and around neutral, those for
ethylamine and propylamine were narrower and shifted
to alkaline conditions (pH 9.5–10).
1
Tokutake et al.28) concluded that the carboxy group in
Cys is the major recognition element of E. coli ꢀGCS in
the transition state. On the other hand, Hiratake et al.29)
suggested the importance of the length of the methylene
chain of the substrate amino acid, which has a large
influence on the hydrophobic interactions between the
substrate and the enzyme. Kelly et al.25) reported that
ꢀGCS from E. coli took 17 amino acids instead of Cys,
and showed that the relative activity (Vmax=Km values)
for amino acids was ꢁ-aminobutylate (carbon chain
length, 3) > alanine (2) > glycine (1) > norvaline (4).
Based on our experiments, the relative activities of the
amines depended on the length of the carbon chain
as follows (Fig. 7): n-propylamine (3) > butylamine
(4) > ethylamine (2) ꢃ methylamine (1). The order of
amine reactivities was similar to the order of amino acid
reactivity except for the position of the C4 compounds.
In both amino acids and amines, C3 compounds (ꢁ-
aminobutyrate and n-propylamine) were the most highly
reactive compounds. This appears to confirm the
importance of the hydrophobic interaction. Amines with
longer side chains are correspondingly more reactive
than the comparable amino acids. This suggests that
the carboxyl group of the amino acids contributes to
substrate recognition at the Cys binding site of the
enzyme. The reason the optimum pH for the reaction of
the amines was narrower and shifted to alkaline pHs is
difficult to determine but it might be related to the
difference in acid dissociation constants between the
amino acid and the amine.
Discussion
More than 40 years ago, Rathbun27) tested 71
compounds, including some amines, as substrates for
ꢀGCS from bovine lens, and reported no detectable
activities for ethylamine or isopropylamine. In this
study, however, we found that ꢀGCS from E. coli had
In this study, we also found that theanine can be
synthesized from Glu and ethylamine by a resting cell
reaction of E. coli overexpressing ꢀGCS. Despite the

Novel Catalytic Ability of ꢀ-Glutamylcysteine Synthetase
2683
Kakuda T, and Takeuchi N, Biosci. Biotechnol. Biochem., 59,
615–618 (1995).
fact that the optimal pH shifted to alkali conditions, it is
remarkable that theanine was synthesized in gram
quantities with coupling to E. coli’s own ATP regener-
ation system (Fig. 3). The amount of ATP initially
added was 0.8 mM, whereas 12.1 mM of theanine and
11.5 mM of ꢀ-glutamylalanine were formed, indicating
that most of the ATP for the synthesis reactions was
regenerated by the glycolytic activity of the cells.
Yamamoto et al.17) reported that theanine could be
synthesized from GS coupled with ATP regeneration
through the glycolytic activity of dry yeast. It would be
interesting to determine whether the theanine synthesis
found here can be achieved only in E. coli cells, because
we found that E. coli cells contained sufficient ATP
regeneration activity to drive the reaction.
5) Yokozawa T and Dong E, Exp. Toxicol. Pathol., 49, 329–335
(1997).
6) Yamada T, Terashima T, Okubo T, Juneja LR, and Yokogoshi
H, Nutr. Neurosci., 8, 219–226 (2005).
7) Kakuda T, Yanase H, Utsunomiya K, Nozawa A, Unno T, and
Kataoka K, Neurosci. Lett., 289, 189–192 (2000).
8) Yokogoshi H and Kobayashi M, Life Sci., 62, 1065–1068
(1998).
9) Sadzuka Y, Yamashita Y, Kishimoto S, Fukushima S, Takeuchi
Y, and Sonobe T, Yakugaku Zasshi (in Japanese), 122, 995–999
(2002).
10) Yamada T, Terashima T, Honma H, Nagata S, Okubo T, Juneja
LR, and Yokogoshi H, Biosci. Biotechnol. Biochem., 72, 1356–
1359 (2008).
11) Sasaoka K, Kito M, and Onishi Y, Agric. Biol. Chem., 29, 984–
988 (1965).
The formation of ꢀ-glutamylalanine is a drawback of
our theanine production system. The substrate Ala is
formed from pyruvate, the glycolytic product. As Fig. 3
suggests, the reactivity of ꢀGCS for Ala is higher than
that for ethylamine. Thus to reduce the ꢀ-glutamylala-
nine byproduct, alternating the host sugar metabolic
pathway or changing the substrate specificity of ꢀGCS
is necessary. The structure of ꢀGCS from E. coli was
recently published and the Cys binding sites were
determined.30) Perhaps by gaining additional insight
from the crystal structure into ways to alter the substrate
specificity of ꢀGCS, opportunities might be revealed to
engineer the system to avoid unwanted byproducts
further.
12) Yamada T, Shiode J-I, and Tachiki T, Ann. Rep. IC. Biotechnol.,
13, 351–353 (1990).
13) Abelian VH, Okubo R, Mutoh K, Chu DC, Kim M, and
Yamamoto T, J. Ferment. Bioeng., 73, 195–198 (1993).
14) Suzuki H, Izuka S, Miyakawa N, and Kumagai H, Enzyme
Microb. Technol., 31, 884–889 (2002).
15) Suzuki H, Izuka S, Minami H, Miyakawa N, Ishihara S, and
Kumagai H, Appl. Environ. Microbiol., 69, 6399–6404 (2003).
16) Tachiki T, Wakisaka S, Suzuki H, and Kumagai H, Agric. Biol.
Chem., 47, 287–292 (1983).
17) Yamamoto S, Uchimura K, Wakayama M, and Tachiki T,
Biosci. Biotechnol. Biochem., 68, 1888–1897 (2004).
18) Susana I, Wasch A, Ploeg JR, Maire T, Lebreton A, Kiener A,
and Leisinger T, Appl. Environ. Microbiol., 68, 2368–2375
(2002).
19) Kimura T, Sugawara I, Hanai K, and Tonomura Y, Biosci.
Biotechnol. Biochem., 56, 708–711 (1992).
Acknowledgment
20) Yamamoto S, Wakayama M, and Tachiki T, Biosci. Biotechnol.
Biochem., 71, 545–552 (2007).
We wish to thank Ms. Hiromi Murakami for expert
technical assistance, and Ms. Yuko Uosaki and Ms.
Yoshiko Matsumoto-Ishikawa for help in performing
MS and NMR experiments. We thank Dr. Shinichi
Hashimoto and Dr. Makoto Yagasaki for helpful
discussion and support during the writing of this paper.
21) Yamamoto S, Wakayama M, and Tachiki T, Biosci. Biotechnol.
Biochem., 72, 101–109 (2008).
22) Mandeles S and Bloch K, J. Biol. Chem., 214, 639–646 (1955).
23) Orlowski M and Meister A, Biochemistry, 10, 372–380 (1971).
24) Watanabe K, Yamano Y, Murata K, and Kimura A, Nucleic
Acids Res., 14, 4393–4400 (1986).
25) Kelly BS, Antholine WE, and Griffith OW, J. Biol. Chem., 277,
50–58 (2002).
26) Sasaki K, Nishi T, Yasumura S, Sato M, and Ito S, JN Patent,
2928287 (May 14, 1999).
References
1) Sakato Y, Nippon No¯geikagaku Kaishi (in Japanese), 23, 262–
267 (1949).
27) Rathbun WB, Arch. Biochem. Biophys., 122, 73–84 (1967).
28) Tokutake N, Hiratake J, Katoh M, Irie T, Kato H, and Oda J,
Bioorg. Med. Chem., 6, 1935–1953 (1998).
2) Kimura K, Ozeki M, Juneja LR, and Ohira H, Biol. Psychol., 74,
39–45 (2006).
29) Hiratake J, Irie T, Tokutake N, and Oda J, Biosci. Biotechnol.
Biochem., 66, 1500–1514 (2002).
3) Ozeki M, Juneja LR, and Shirakawa S, Nihon Seirijinruigaku
Kaishi (in Japanese), 9, 13–20 (2004).
4) Yokogoshi H, Kato Y, Sagesaka YM, Takihara-Matsuura T,
30) Hibi T, Nii H, Nakatsu T, Kimura A, Kato H, Hiratake J, and
Oda J, Proc. Natl. Acad. Sci. USA, 101, 15052–15057 (2004).