Purification and characterization of glutamate
N-acetyltransferase involved in citrulline accumulation
in wild watermelon
Kentaro Takahara, Kinya Akashi and Akiho Yokota
Graduate School of Biological Sciences, Nara Institute of Science and Technology, Japan
Drought in the presence of strong light is a major
environmental stress that reduces plant productivity
[1]. To adapt to this adverse condition, numerous bio-
chemical and physiological tolerance mechanisms are
expressed in plant cells [2]. One such response involves
accumulation of small organic metabolites, such as
mannitol, proline and glycine betaine, which are collec-
tively referred to as compatible solutes [3]. Compatible
solutes are thought to play important roles in drought
tolerance in plants, acting as mediators of osmotic
adjustment, stabilizers of subcellular structures, and
scavengers of active oxygen radicals [4]. The mecha-
nisms of proline, mannitol, and glycine betaine accu-
mulation are highly regulated through activation of
biosynthesis and ⁄ or suppression of catabolism [3–5].
Wild watermelon plants, which inhabit the Kalahari
Desert, Botswana, exhibit high drought ⁄ strong-light
stress tolerance [6]. They are able to maintain their
photosynthetic apparatus during prolonged periods of
drought in strong light, suggesting the presence of
Keywords
citrulline; drought/strong-light stress;
glutamate N-acetyltransferase;
thermostability; wild watermelon
Correspondence
A. Yokota, Nara Institute of Science and

watermelon leaves during drought ⁄ strong-light stress.
Abbreviations
AOD, acetylornithine deacetylase; CLGAT, Citrullus lanatus glutamate N-acetyltransferase; DRIP-1, drought-induced polypeptide 1;
DTT, dithiothreitol; GAT, glutamate N-acetyltransfease; GFP, green fluorescence protein.
FEBS Journal 272 (2005) 5353–5364 ª 2005 FEBS 5353
mechanisms that allow them to tolerate oxidative stress
arising from excess light energy absorbed by the leaves.
Drought ⁄ strong-light stresses result in an accumulation
of a novel compatible solute, citrulline, in the leaves
[6]. The concentration of citrulline in the stressed
leaves reaches up to 30 mm, compared to only 0.6 mm
in unstressed leaves [6]. Among known compatible sol-
utes, citrulline is one of the most efficient scavengers
for hydroxyl radicals [7]. These findings suggest that
citrulline functions as a hydroxyl radical scavenger in
the presence of strong light.
In addition to citruline, concentration of arginine
increases from 0.3 mm in unstressed conditions to
7mm under drought ⁄ strong-light stress in the leaves of
wild watermelon [6]. Although arginine is the final
product of the arginine biosynthetic pathway, wild
watermelon plants accumulate larger quantities of
citrulline, an intermediate in this pathway, during
drought ⁄ strong-light stress. Arginine is a key meta-
bolite in regulation of this pathway in plants [8]. How-
ever, the mechanism of accumulation for these
metabolites in wild watermelon remains unclear.
Regulation of citrulline and arginine synthesis has
been studied extensively in prokaryotes and Saccharo-
myces cerevisiae [9,10]. The pathway starts with acety-

inhibitions in the microorganisms examined so far.
How wild watermelon is able to accumulate high levels
of citrulline is therefore an intriguing question.
In plants, knowledge on the pathway of citrulline
and arginine biosynthesis is still fragmentary [8,18].
The genome project revealed that both GAT and
AOD-homologous genes exist in Arabidopsis thaliana
[19]. Our previous study showed that a novel protein,
drought-induced polypeptide 1 (DRIP-1), which shares
sequence homology with bacterial AOD, is strongly
induced by drought ⁄ strong-light stress in wild water-
melon [6]. However, the catalytic property of DRIP-1
remains to be determined, and it is not known whether
DRIP-1 contributes to massive accumulation of citrul-
line in wild watermelon.
As a first step to understand the mechanism of citrul-
line and arginine accumulation in wild watermelon, we
focused on the fifth step of citrulline biosynthesis, at
which point DRIP-1 was expected to function as
AOD. However, GAT activity, not AOD activity, was
detected in wild watermelon leaves in which DRIP-1
Fig. 1. The pathway of citrulline and arginine biosynthesis. AGS,
N-acetylglutamate synthase; AGK, N-acetylglutamate kinase; AGPR,
N-acetylglutamate 5-phosphate reductase; AOAT, N-acetylornithine
transaminase; GAT, glutamate N-acetyltransferase; AOD, N-acetyl-
ornithine deacetylase; OCT, ornithine carbamoyltransferase; ASS,
argininosuccinate synthase and ASL, argininosuccinate lyase.
Wild watermelon glutamate N-acetyltransferase K. Takahara et al.
5354 FEBS Journal 272 (2005) 5353–5364 ª 2005 FEBS
had been strongly induced. This paper reports the

ded in the reaction mixture (data not shown). In a pos-
itive control experiment, we could detect similar AOD
activity in the extract of E. coli to that reported in the
literature [11], demonstrating that the assay procedures
were valid for detecting AOD activity. The undetect-
able AOD activity in extracts from wild watermelon
leaves is in contrast to the strong expression of DRIP-
1 during stress (Fig. 2).
On the contrary, GAT activity was detected in
leaves of wild watermelon (Fig. 2A). The specific activ-
ity of GAT in unstressed leaves was approximately
3.2 nmolÆmin
)1
mg protein
)1
, and this did not change
significantly during stress. This constant GAT activity
did not correlate with the induction of DRIP-1 protein
during drought ⁄ strong-light stress (Fig. 2B).
Purification of GAT from wild watermelon leaves
To characterize the GAT in detail, it was purified from
wild watermelon leaves (Table 1). The purification
procedure was developed by taking advantage of the
thermal stability of the GAT activity in crude extracts.
After centrifugation of the total leaf extract to remove
cell debris, the supernatant was heated at 70 °C for
10 min and centrifuged. Negligible loss of enzymatic
activity and 24-fold purification were achieved. Heat
treatment was followed by six chromatography steps
comprising hydrophobic interaction, anion and cation

)
Purification
(fold)
Crude extract 5500 37 0.00067 1
Heat treatment 280 46 0.017 24
Butyl sepharose 15 35 0.23 340
Mono Q (pH 8.0) 0.94 11 1.2 1700
Sephadex 200 0.24 4.6 1.9 2800
Mono Q (pH 7.0) 0.13 3.4 2.6 3900
Mono S 0.07 3.0 4.3 6400
Hydroxyapatite < 0.02 0.94 > 4.7 > 7000
K. Takahara et al. Wild watermelon glutamate N-acetyltransferase
FEBS Journal 272 (2005) 5353–5364 ª 2005 FEBS 5355
acids. The N-terminal amino acid sequence of the a
peptide was identical to a portion of the amino acid
sequence predicted from a watermelon EST clone
(accession number AI563351).
Cloning of watermelon GAT cDNA
To isolate a full-length cDNA clone of the GAT, gene-
specific primers were designed from the sequence of
watermelon EST AI563351 and used for 5¢- and
3¢-RACE. The cloned cDNA encoded a protein
composed of 460 amino acids. The amino acid
sequence had homology to At2g37500 from A. thaliana
(70% identity), B. subtilis GAT (38%) and S. cerevisiae
GAT (26%). Two regions of the deduced sequence
(residues 27–42 and 239–273) were identical to the
N-terminal amino acid sequences of the a and b pep-
tides determined by Edman sequencing, respectively
(Fig. 4); the enzyme was designated CLGAT (Citrullus

mined by SDS ⁄ PAGE (Fig. 3A). However, MALDI-
MS analysis of CLGAT detected two major peaks at
m ⁄ z 21.3 and 23.5 kDa, which corresponded to the
masses deduced from the cDNA (data not shown).
The molecular mass of native GAT was determined by
gel filtration chromatography as  90 kDa (Fig. 3B),
suggesting that CLGAT is a heterotetramer composed
of two each of the a and b subunits.
Analysis of the N-terminal transit peptide of
CLGAT
Analysis of the CLGAT cDNA predicted a chloroplast
transit peptide upstream from the N-terminus of
CLGAT. To examine whether the precursor of
CLGAT is imported into the chloroplasts, we prepared
a plasmid in which the first 26 codons of the CLGAT
A
B
Fig. 3. Molecular mass of GAT purified from leaves of wild water-
melon. (A) SDS ⁄ PAGE of GAT. The peak fraction obtained after
hydroxylapatite chromatography was analysed by PAGE (12% w ⁄ v,
polyaclylamide) and detected by silver staining. (B) Molecular mass
of the native form of wild watermelon GAT was determined from
the plot between the e lution volumes of the enzyme and marker
proteins in Superdex 200 gel chromatography. Thy, thyroglobulin;
Fer, ferritin; Ald, aldolase; Ova, ovalbumin.
Wild watermelon glutamate N-acetyltransferase K. Takahara et al.
5356 FEBS Journal 272 (2005) 5353–5364 ª 2005 FEBS
cDNA were fused in-frame to the coding sequence of
green fluorescent protein (GFP). This fusion protein
was transiently expressed in tobacco leaves and the

ured at a physiologically relevant concentration of
citrulline or arginine. Citrullus lanatus glutamate acetyl-
transferase activities in the presence of 30 mm citrulline
or 7 mm arginine were 100 and 98% of the original
activities, respectively, showing that CLGAT is not
inhibited by these downstream products. In this study,
the effect of ornithine was not determined because the
concentration of ornithine in wild watermelon leaves is
very low and constant before or after drought ⁄ strong
light stress, whereas concentrations of citrulline and
arginine in the leaves increased greatly during the
stress [6].
Surprisingly, CLGAT activity was maximum at
70 °C (Fig. 6D). To determine its thermostability, the
enzyme was incubated for 30 min at various tempera-
tures between 30 and 90 °C and residual activity was
determined (Fig. 6E). Citrullus lanatus glutamate acetyl-
transferase retained 98% of the original activity after
incubation at 70 °C, and still showed about 15% of the
original activity after incubation at 80 °C.
The thermotolerance of CLGAT observed above
prompted us to examine the leaf temperature under
drought conditions in the presence of strong light at
an atmospheric temperature of 35 °C. Under
unstressed conditions, the rate of leaf transpiration
was about 450 mmol H
2
OÆm
)2
Æs

suggesting that the fifth step of the citrulline pathway
A
B
Fig. 5. CLGAT is a chloroplast protein the
import of which depends on its N-terminal
transit peptide. Fluorescence microscopy
pictures of tobacco leaf cell transiently
expressing a CLGAT-GFP fusion (A) or GFP
alone as a control (B).
Wild watermelon glutamate N-acetyltransferase K. Takahara et al.
5358 FEBS Journal 272 (2005) 5353–5364 ª 2005 FEBS
is mainly catalysed by GAT in wild watermelon.
CLGAT was subsequently purified and identified from
wild watermelon. This is the first report dealing with
GAT purified from plant sources.
Under drought ⁄ strong-light conditions, plants close
their stomata to avoid loss of water, thereby increasing
leaf temperature [26]. In this study, the temperature of
wild watermelon leaves increased from 30 to 44 °C
under experimental stress conditions (Fig. 7). Under
natural desert conditions, leaf temperature of stressed
wild watermelon plants rises up to 60 °C [27]. In
this study, the optimum temperature of CLGAT was
revealed to be 70 °C, which was comparable to that
reported from the thermophilic microorganism B. stearo-
thermophilus (Table 2). Moreover, CLGAT retained
about 15% of its original activity after incubation at
80 °C for 30 min, whereas GAT from B. stearothermo-
philus completely lost its activity after incubation at
75 °C for 30 min [28]. These results suggest that CLGAT

CLGAT for N-acetylornithine (B) and glutam-
ate (C). The concentration of N-acetyl-
ornithine was varied from 0.5 to 15 m
M
with glutamate fixed at 10 mM in (B), and in
(C) that of glutamate was changed between
0.5 and 30 m
M with N-acetylornithine fixed
at 10 m
M. Double reciprocal plot of CLGAT
activity against the concentration of sub-
strates are shown in the insets. The fitted
linear regression lines and parameters are
also presented. The K
m
was estimated
based on these parameters. (D) Tempera-
ture dependency of CLGAT activity. Enzyme
activity was measured by incubating the
reaction mixtures at 20–90 °C. Activity is
expressed as percentage of the maximal
activity. (E) Thermostability of CLGAT. The
enzyme was incubated for 30 min at the
indicated temperatures and residual enzyme
activity was measured at 30 °C. Values
represent the mean ± SE from three
independent measurements.
K. Takahara et al. Wild watermelon glutamate N-acetyltransferase
FEBS Journal 272 (2005) 5353–5364 ª 2005 FEBS 5359
these GATs is between 1 and 3 mm, much higher than

localization of citrulline and arginine biosynthetic
enzymes is scarce, but it has been reported that ornith-
ine carbamoyltransferase, an enzyme required for the
sixth step in the citrulline and arginine pathway, is
localized in chloroplasts in Canavalia lineate [33], rais-
ing the possibility that citrulline biosynthesis in plants
is catalysed in chloroplasts. In fact, cDNAs for all
six citrulline biosynthetic enzymes in Arabidopsis were
predicted as having chloroplast-targeting sequences
using the various programs [18].
A
B
Fig. 7. (A) Change in the stomatal conductance (h) and leaf tem-
perature (d) of wild watermelon plants during drought. Data
points represent the means from three independent experiments
and vertical bars are the SD. (B) Visualization of the thermal distri-
bution in wild watermelon plants. Images 1 and 2 represent
photographs of wild watermelon plants watered daily and subjec-
ted to drought for 5 days, respectively, and images 3 and 4 show
their thermal distributions analysed using an infrared thermal
camera, respectively.
Table 2. Comparison of the kinetic parameters for GAT from wild watermelon, S. cerevisiae and B. stearothermophilus. ND, not deter-
mined.
Wild watermelon S. cerevisiae
a
B. stearothermophilus
b
K
m glutamate
(mM) 17.8 7.2 19.2

The function of DRIP-1 remains to be determined.
In this study, no AOD activity was detected in wild
watermelon leaves in which abundant DRIP-1 accumu-
lation was seen (Fig. 2). Moreover, recombinant
DRIP-1 expressed in E. coli had no detectable AOD
activity (data not shown). One possible explanation for
this discrepancy is that DRIP-1 requires a cofactor or
activator for its catalysis, although we tested Co
2+
and Zn
2+
in the present enzymatic assays. The second
possibility is that DRIP-1 possesses a different func-
tion unrelated to AOD.
In this study, we demonstrated that CLGAT,
which catalyses the first and fifth steps of the citrul-
line biosynthetic pathway, can contribute effectively
to citrulline biosynthesis under drought conditions
owing to its high thermostability and insensitivity to
the inhibition induced by citrulline and arginine.
However, many different enzymes are involved in the
metabolism of citrulline (Fig. 1), and their kinetic
properties and ⁄ or expression pattern remain to be
examined. Comprehensive analysis of the citrulline
biosynthesis pathway is required to fully understand
the mechanism of citrulline accumulation in wild
watermelon.
Experimental procedures
Materials
N-Acetylornithine was purchased from Sigma (St. Louis,

(TVS-8500, Nippon Avionics Ltd, Tokyo, Japan). Data
were collected around 15:00 (7 h after the start of the light
period).
Enzyme assays
Acetylornithine deacetylase and GAT activity were
measured by quantifying the production of ornithine
from N-acetylornithine using the colorimetric ninhydrin
procedure as described previously [11,15]. The AOD
assay mixture (200 lL) contained 100 mm potassium
phosphate buffer pH 7.0, 6 mm N-acetylornithine,
0.5 mm metal salt (CoCl
2
or ZnCl
2
) and the enzyme.
The GAT assay mixture (200 lL) contained 100 mm
potassium phosphate buffer (pH 7.0), 6 mm N-acetylorni-
thine, 6 mm glutamate and the enzyme. The reaction
was started by adding N-acetylornithine and stopped by
adding 600 lL of ninhydrin reagent (0.4 m citric acid ⁄ 1%
ninhydrin in 2-methoxyethanol, 1 : 2 v ⁄ v). After boiling for
10 min, 200 lL4m NaOH was added to the mixture and
light absorbance at 470 nm was measured. One unit
is defined as the amount of enzyme that forms 1 lmol
ornithineÆmin
)1
.
Purification of GAT from wild watermelon leaves
All purification steps were carried out at 0–4 °C, and col-
umn chromatographies were performed with FPLC system

(NH
4
)
2
SO
4
. The enzyme was eluted with a linear gradient
of 30–0% saturation of (NH
4
)
2
SO
4
in buffer A. Active
fractions were collected and applied to a Sephadex G-25
column (5 cm i.d. · 20 cm) equilibrated with buffer B con-
taining 5 mm potassium phosphate buffer (pH 8.0), 1 mm
EDTA, 10 mm b-mercaptoethanol, 1 mm DTT, and 20%
(v ⁄ v) glycerol. Active fractions were pooled and applied
to a Mono Q HR column (0.5 cm i.d. · 5 cm; Amersham
Bioscience) equilibrated with buffer B. The column was
developed with a 0–200 mm linear gradient of KCl in buffer
B. Active fractions were pooled and concentrated with
Centriplus YM-10 (Amicon, Bevery, MA, USA). The
enzyme solution was applied to a Sephadex 200 column
(1.6 cm i.d. · 60 cm; Amersham Bioscience) equilibrated
with 20 mm potassium phosphate buffer pH 7.0, 1 mm
EDTA, 10 mm 2-mercaptoethanol, 1 mm DTT, 150 mm
KCl and 20% (v ⁄ v) glycerol and developed with the same
buffer. Active fractions were loaded onto a Mono Q HR

)1
a-cyano-4-hydroxycinnamic
acid and 50% (v ⁄ v) acetonitrile. The mixture was loaded
onto a sample plate and the solvent was removed by evapor-
ation. The molecular mass of CLGAT was determined by
MALDI-TOF MS (Autoflex II, Bruker Daltonics, Billerica,
MA, USA), calibrated with insulin (5734 Da), cyto-
chrome c (12361 Da), myoglobin (16952 and 8476 Da) and
ubiquitin (8566 Da; all Sigma) as standard proteins.
Western blotting
Proteins from SDS ⁄ PAGE were transferred to polyvinyl-
dene difluoride membranes (Sequi-Blot PVDF membrane;
Bio-Rad) using a semidry blotting apparatus (NA-1512, Ni-
hon-Eido, Tokyo, Japan). The specific antibody for DRIP-
1 [25] was used at a dilution of 1 : 500 in buffer containing
30 mm Tris ⁄ HCl buffer pH 7.5, 200 mm NaCl and 5%
(w ⁄ v) skim milk. Immunoreactive proteins were detected
with an Immunostaining HRP-1000 kit (Konica, Tokyo,
Japan) according to the manufacturer’s instructions.
cDNA cloning
Total RNA was prepared from 3 g wild watermelon leaves
subjected to drought stress for 1 day using TRIzol (Invitro-
gen, Carlsbad, CA, USA). Poly(A)
+
RNA was isolated
from the total RNA using a mRNA purification kit (Amer-
sham Bioscience) according to the manufacturer’s instruc-
tions. The sequence of the watermelon EST clone
(accession number AI563351) was used to design four
GAT-specific primers; CLGAT5a (5¢-GGCATCAACAT-

melon as described previously [36]. Aliquots of genomic
DNA (10 lg) were digested with the indicated restriction
enzymes. The digests were then subjected to electrophor-
esis on 0.8% agarose gel and transferred to Hybond N
+
membranes (Amersham Bioscience). For probe prepar-
ation, a 615 bp GAT fragment was amplified from the
CLGAT cDNA by PCR with the primers CLGAT3b and
CLGAT3c (5¢-GGTACTCGTATCCCCATCCACCG-3¢).
The amplified fragment was purified using the MinElute
Gel Extraction Kit (Qiagen, Valencia, CA, USA), and
used as a template for preparation of a radioactive probe
with the Prime-It II Random primer Labeling Kit (Strata-
gene). Hybridization was performed as described previ-
ously [37].
GFP transient assay
The sequence of the putative chloroplast transit peptide
(amino acids 1–26) was amplified by PCR and cloned into
the modified GFP vector pTH2XA [38]. The construct was
designed to fuse the DNA sequence of the putative chloro-
plast transit peptide to the 5¢-end of the GFP gene to be
expressed under control of the cauliflower mosaic virus 35S
promoter. The plasmid (1 lg) was introduced into tobacco
leaves by bombardment (PDS-1000 ⁄ He apparatus, Bio-
Rad) according to the manufacturer’s instructions. The
fluorescence was visualized with confocal microscopy
(LSM510, Carl Zeiss, Jena, Germany) after overnight incu-
bation.
Acknowledgements
We thank Junko Tsukamoto for technical assistance

8 Thompson JF (1980) Arginine synthesis, proline synthe-
sis, and related processes. In The Biochemistry of Plants,
Vol. 5, pp. 375–402. Academic Press, New York.
9 Cunin R, Gransdorff N, Pierard A & Stalon V (1986)
Biosynthesis and metabolism of arginine in bacteria.
Microbiol Rev 50, 314–352.
10 Davis RH (1986) Commpartmental and regulatory
mechanisms in the arginine pathways of Neurospora
crassa and Saccharomyces cerevisiae. Microbiol Rev 50,
280–313.
11 Vogel HJ & Bonner DM (1956) Acetylornithinase of
Escherichia coli: partial purification and some proper-
ties. J Biol Chem 218, 97–106.
12 Leisinger T & Haas D (1975) N-acetylglutamate
synthase of Escherichia coli regulation of synthesis
and activity by arginine. J Biol Chem 250, 1690–1693.
13 Glansdorff N (1964) Topography of cotransducible
arginine mutations in Escherichia coli K-12. Genetics
51, 167–179.
14 Sakanyan V, Kochikyan A, Mett I, Legrain C,
Charlier D, Pierard A & Glansdorff N (1992) A
re-examination of the pathway for ornithine biosynthesis
in a thermophilic and two mesophilic Bacillus specoes.
J Gen Microbiol 138, 125–130.
15 Liu Y, Van Heeswijck R, Hoj P & Hoogenraad N
(1995) Purification and characterization of ornithine
acetyltransferase from Saccharomyces cerevisiae. Eur J
Biochem 228, 291–296.
16 Pauwels K, Abadjieva A, Hilven P, Stankiewicz A &
Crabeel M (2003) The N-acetylglutamate synthase ⁄

(2000) The yeast ARG7 gene product is autoproteo-
lyzed to two subunit peptides, yielding active
ornithine acetyltransferase. J Biol Chem 275, 11361–
11367.
24 Mass WK (1994) The arginine repressor of Escherichia
coli. Microbiol Rev 58, 631–640.
25 Yokota A, Kawasaki S, Iwano M, Nakamura C,
Miyake C & Akashi K (2002) Citrulline and DRIP-1
protein (ArgE homologue) in drought tolerance of wild
watermelon. Ann Bot (Lond) 89, 825–832.
26 Radin JW, Lu Z, Percy RG & Zeiger E (1994) Genetic
variability for stomatal conductance in Pima cotton
and its relation to improvements of heat adaptation.
Proc Natl Acad Sci USA 91, 7217–7221.
27 Larcher W (1995) Plant under stress. In Physiological
Plant Ecology, 3rd edn, pp. 321–448. Springer-Verlag,
New York.
28 Marc F, Weigel P, Legrain C, Almeras Y, Santrot M,
Glansdorff N & Sakanyan V (2000) Characterization
and kinetic mechanism of mono- and bifunctional
ornithine acetyltransferases from thermophilic micro-
organisms. Eur J Biochem 267, 5217–5226.
29 Van de Casteele M, Demarez M, Legrain C, Glansdorff
N & Pierard A (1990) Pathways of arginine biosynthesis
in extreme thermophilic archeo- and eubacteria. J Gen
Microbiol 136, 1177–1183.
30 Crabeel M, Abadjieva A, Hilven P, Desimpelaere J &
Soetens O (1997) Characterization of the Saccharomyces
cerevisiae ARG7 gene encoding ornithine acetyltrans-
ferase, an enzyme also endowed with acetylglutamate

Yokota A & Lagarias JC (2001) The Arabidopsis HY2
gene encodes pytochromobilin synthase, a ferredoxin-
dependent biliverdin reductase. Plant Cell 13, 425–436.
Wild watermelon glutamate N-acetyltransferase K. Takahara et al.
5364 FEBS Journal 272 (2005) 5353–5364 ª 2005 FEBS