Co-pigmentation and flavonoid glycosyltransferases in blue Veronica persica flowers

Article abstract of DOI:10.1016/j.phytochem.2010.02.008

Glycosylation is one of the key modification steps for plants to produce a broad spectrum of flavonoids with various structures and colors. A survey of flavonoids in the blue flowers of Veronica persica Poiret (Lamiales, Scrophulariaceae), which is native of Eurasia and now widespread worldwide, led to the identification of highly glycosylated flavonoids, namely delphinidin 3-O-(2-O-(6-O-p-coumaroyl-glucosyl)-6-O-p-coumaroyl-glucoside)-5-O-glucoside (1) and apigenin 7-O-(2-O-glucuronosyl)-glucuronide (2), as two of its main flavonoids. Interestingly, the latter flavone glucuronide (2) caused a bathochromic shift on the anthocyanin (1) toward a blue hue in a dose-dependent manner, showing an intermolecular co-pigment effect. In order to understand the molecular basis for the biosynthesis of this glucuronide, we isolated a cDNA encoding a UDP-dependent glycosyltransferase (UGT88D8), based on the structural similarity to flavonoid 7-O-glucuronosyltransferases (F7GAT) from Lamiales plants. Enzyme assays showed that the recombinant UGT88D8 protein catalyzes the 7-O-glucuronosylation of apigenin and its related flavonoids with preference to UDP-glucuronic acid as a sugar donor. Furthermore, we identified and functionally characterized a cDNA encoding another UGT, UGT94F1, as the anthocyanin 3-O-glucoside-2″-O-glucosyltransferase (A3Glc2″GlcT), according to the structural similarity to sugar-sugar glycosyltransferases classified to the cluster IV of flavonoid UGTs. Preferential expression of UGT88D8 and UGT94F1 genes in the petals supports the idea that these UGTs play an important role in the biosynthesis of key flavonoids responsible for the development of the blue color of V. persica flowers.

Full text of DOI:10.1016/j.phytochem.2010.02.008

Phytochemistry 71 (2010) 726–735
Contents lists available at ScienceDirect
Phytochemistry
journal homepage: www.elsevier.com/locate/phytochem
Co-pigmentation and flavonoid glycosyltransferases in blue
Veronica persica flowers q
Eiichiro Ono a, Miho Ruike b, Takashi Iwashita c, Kyosuke Nomoto b,1, Yuko Fukui a,
*
a Institute for Health Care Science, Suntory Ltd., 1-1-1 Wakayamadai, Shimamoto, Mishima, Osaka 618-8503, Japan
b Faculty of Life Science, Toyo University, 1-1-1 Izumino, Itakura, Ora, Gunma 374-0113, Japan
c Suntory, Institute for Bioorganic Research, 1-1-1 Wakayamadai, Shimamoto, Mishima, Osaka 618-8503, Japan
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 30 April 2009
Received in revised form 17 December 2009
Available online 10 March 2010
Glycosylation is one of the key modification steps for plants to produce a broad spectrum of flavonoids
with various structures and colors. A survey of flavonoids in the blue flowers of Veronica persica Poiret
(Lamiales, Scrophulariaceae), which is native of Eurasia and now widespread worldwide, led to the iden-
tification of highly glycosylated flavonoids, namely delphinidin 3-O-(2-O-(6-O-p-coumaroyl-glucosyl)-6-
O-p-coumaroyl-glucoside)-5-O-glucoside (1) and apigenin 7-O-(2-O-glucuronosyl)-glucuronide (2), as
two of its main flavonoids. Interestingly, the latter flavone glucuronide (2) caused a bathochromic shift
on the anthocyanin (1) toward a blue hue in a dose-dependent manner, showing an intermolecular co-
pigment effect. In order to understand the molecular basis for the biosynthesis of this glucuronide, we
isolated a cDNA encoding a UDP-dependent glycosyltransferase (UGT88D8), based on the structural sim-
ilarity to flavonoid 7-O-glucuronosyltransferases (F7GAT) from Lamiales plants. Enzyme assays showed
that the recombinant UGT88D8 protein catalyzes the 7-O-glucuronosylation of apigenin and its related
flavonoids with preference to UDP–glucuronic acid as a sugar donor. Furthermore, we identified and
functionally characterized a cDNA encoding another UGT, UGT94F1, as the anthocyanin 3-O-glucoside-
200-O-glucosyltransferase (A3Glc200GlcT), according to the structural similarity to sugar–sugar glycosyl-
transferases classified to the cluster IV of flavonoid UGTs. Preferential expression of UGT88D8 and
UGT94F1 genes in the petals supports the idea that these UGTs play an important role in the biosynthesis
of key flavonoids responsible for the development of the blue color of V. persica flowers.
Keywords:
Veronica persica
Lamiales
Flavonoid
Co-pigment
Glycosyltransferase
Glucuronosylation
Glucosylation
UGT
Ó 2010 Elsevier Ltd. All rights reserved.
1. Introduction
thocyanins from two Salvia species with blue petals showed that
the pigment complex consists of six molecules each of diglycosyl
Blue flowers are rare in nature. Limited examples includes day-
flower (Commelina communis) and cornflower (Centaurea cyanus),
both of which produce natural supramolecules known as metallo-
anthocyanins in their petals (Kondo et al., 1992; Shiono et al.,
2005). Those metalloanthocyanins are formed through stoichiom-
etric self-association of metal-pigment complexes, and metal com-
plexation of flavonoid glycosides and intermolecular hydrophobic
association confer pure blue coloration on flowers (Goto and Kon-
do, 1991; Yoshida et al., 2009). Recent studies on the metalloan-
anthocyanins and diglycosyl flavones, although the composition
of the flavonoids within the complex is different between the
two species (Kondo et al., 2001; Mori et al., 2008). In the case of
the Himalayan blue poppy (Meconopsis grandis) has a diglucosyl
flavonol as a component of the metal complex-pigment in the
sky-blue flower (Yoshida et al., 2006).
Co-pigmentation, in addition to the metal complex-pigment, is
also well known to cause bathochromic shift by an intermolecular
association that colored anthocyanin pigments and colorless/
transparent co-pigments, such as flavone glycosides stack
hydrophobically in an aqueous solution (Goto and Kondo, 1991).
Co-pigmentation occurs only in solution in a non-stoichiometrical
way, and confers a bluish color on flowers and fruits of plants, and
also during fermentation processes of wines (Boulton, 2001).
Flavones/flavonols per se exhibit pale yellow color; however, their
role as co-pigments and UV-absorbents seem to be more signifi-
cant on flower coloration, serving as key modulators of anthocya-
nin pigments for pollinator attraction (Asen et al., 1972;
Abbreviations: MeCN, acetonitrile; UDP, uridine diphosphate; GlcA, glucuronic
acid; Glc, glucose; 1H{13C}-HMBC, heteronuclear multiple bond correlation.
q
The DDBJ Accession Numbers for UGT88D8, UGT94C2, and UGT94F1 are
AB465708, AB514128, and AB514127, respectively.
* Corresponding author. Tel.: +81 75 962 6207; fax: +81 75 962 1690.
E-mail address: Yuko_Fukui@suntory.co.jp (Y. Fukui).
1
Present address: Nara Prefectural Small and Medium-sized Enterprises Support
Corporation, Nara Prefectural Agricultural Experiment Station, 88 Shijo, Kashihara,
Nara 634-0813, Japan.
0031-9422/$ - see front matter Ó 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.phytochem.2010.02.008

E. Ono et al. / Phytochemistry 71 (2010) 726–735
727
Thompson and Meinwald, 1972). For instance, the C-glucosyl fla-
A
vone isovitexin in the petals of the garden iris (Iris ensata Thunb.)
is known to be a strong co-pigment co-existing with the iris antho-
cyanin, delphinidin 3-O-p-coumaroyl-rutinoside-5-O-glucoside
(Yabuya et al., 1997). Moreover, apigenin 7-O-glucuronide and
luteolin 7-O-glycoside have been reported as a strong co-pigment
in snapdragon (Antirrhium majus) and wishbone flowers (Torenia
hybrida), respectively (Asen et al., 1972; Aida et al., 2000). It is
noteworthy that co-pigmentation efficiency depends on the struc-
tural affinity of modified pigments to co-pigments, as well as their
concentrations, and vacuolar pH of the solution (Fukui et al., 2003).
The fact that flavonoid composition in flowers with similar color
varies among different species highlights the importance of the
structural diversity of this class of pigments for developing orna-
mentals with novel flower colors (Katsumoto et al., 2007).
Plant secondary metabolites, or specialized metabolites, have
extraordinary diverse core structures that are often further elabo-
rated through modifications including glycosylation (Harborne and
Baxter, 1999). Specific plant lineage has specialized flavonoids with
unique glycosylation pattern, that are a consequence of the general
differentiation of regio-specificity of the UGT enzyme followed by
local differentiation of sugar donor specificity in a lineage-specific
manner (Vogt and Jones, 2000; Noguchi et al., 2008, 2009). For
example, flavone 7-O-glucuronide, a specialized metabolite of
Lamiales plants, is produced by a flavonoid 7-O-glucuronosyltrans-
ferase (F7GAT), which is considered to be locally differentiated
from flavonoid 7-O-glucosyltransferase (F7GlcT) by acquiring the
binding specificity to UDP–glucuronic acid (UDP–GlcA) instead of
UDP–glucose (Harborne, 1963; Yoshida et al., 1993; Hirotani
et al., 1998; Yamazaki et al., 2003; Noguchi et al., 2009).
Here, we describe identification and functional characterization
of two novel UGTs responsible for the blue flower color of Veronica
persica Poiret (Lamiales, Scrophulariaceae), which is native to Med-
iterranean area and domesticates throughout temperate zone
(Fig. 1A). Most recently, flavonoids that accumulated in the flowers
of this plant were also reported (Mori et al., 2009) after this sub-
mission was received. Here we show the cloning of the two V. per-
sica UGTs, one is involved in the biosynthesis of anthocyanin
glycosides, and the other in the biosynthesis of flavone glucuronide
co-pigments. Observation of bathochromic shift of the anthocyanin
1 by the addition of flavone glucuronide 2 in vitro further under-
scores the significance of the glycosylation of these major flavo-
noids for color development of the blue petals of V. persica.
B
Abs
580.4
1.0
0.8
0.6
628.8
350
400
500
600
700
OD
Wavelength (nm)
C
molar ratio
Petal
L *
a *
17.67
b *
-33.49
H
297.8
ax (nm)
580, 629
48.01
in vitro
1 :
1 :
1 :
1 :
1 :
1 :
1 :
0
1
2
3
4
5
6
43.52
41.94
40.65
39.96
39.46
39.59
39.84
42.28
38.64
36.79
34.98
33.41
31.65
29.92
-56.75
-59.36
-60.30
-60.34
-60.02
-59.18
-58.13
306.6
303.0
301.4
300.1
299.1
298.1
297.2
615
614
632
630
639
641
645
1
1.021
1.031
1.038
1.045
1.035
1.037
2. Results and discussion
Fig. 1. Veronica persica flowers and co-pigment effect. (A) Bluish flowers of Veronica
persica (B) absorption spectra of intact petals of V. persica (C) L a b color value and
* * *
* * *
2.1. Measurement of pH and the L a b color of Veronica petals
visible spectral data of co-pigment assay. Co-pigment analyses of compound 1 with
2 were measured in a McIlvaine buffer (pH 6.2). Compound 1 was dissolved in
1 mM concentration, and 2 was added in 1–6 equivalent to 1. The hue (H) was
shifted to blue in molar ratio dependent. The molar ratio is anthocyanin 1: flavone
2.
Vacuolar pH significantly affects the coloration of anthocyanin
pigments and varies in each species. As the vacuole is the largest
organella in petal cells, the value of pressed petal juice is often
measured as the approximate vacuolar pH (Fukada-Tanaka et al.,
2000; Verweij et al., 2008). The pH value of the petal juice of V. per-
sica was estimated to be 6.2, which is relatively high compared to
that of carnation and rose petals (Fukui et al., 2003; Katsumoto
et al., 2007). The intact petal showed characteristic spectra with
the two kmax peaks at 580.4 and 628.8 nm (Fig. 1B). The petal sur-
(calcd. 623.1248, err: +0.5 ppm), respectively. These values corre-
lated to the masses calculated using their molecular formulae:
C51H53O26 and C27H26O17, respectively. The kmax of 1 was 541 nm
in 10% MeCN in H2O with 0.1% TFA.
After acid hydrolysis, the HPLC analysis of 1 indicated the pres-
ence of delphinidin (ret. time: 4.0 min, kmax: 538 nm). We were
able to establish full assignment data for the 1H NMR and 13C
NMR spectra for 1 and 2 (see Supplementary Tables 1 and 2,
respectively), using 1H{13C}-HSQC, 1H{13C}-HMBC, TOCSY, DQF-
COSY, and ROESY. The 1H NMR and 13C NMR spectroscopic data
of 1 indicated the presence of three glucosyl residues and two p-
coumaroyl moieties. In the case of compound 1, the analysis of
the 1H{13C}-HMBC established that the hydroxyl groups at the
* * *
*
*
face colors in an L a b colorimetric value were L ; 48.01, a ; 17.67,
*
b ; ꢀ33.49 and the hue was 297.8° (Fig. 1C).
2.2. Identification of flavonoid glycosides
High-resolution time-of-flight MS (HR-TOF-MS) analysis of iso-
lated compounds 1 and 2 gave molecular ions at m/z 1081.2876,
[M]+ (calcd. 1081.2825, err: +4.7 ppm) and 623.1251, [M+H]+

728
E. Ono et al. / Phytochemistry 71 (2010) 726–735
C-3 in the C-ring and C-5 positions in the A-ring of the delphinidin
nucleus were bound with Glc-1 and Glc-2, respectively. In addition,
the resonances of methylene protons at the H-6 position in Glc-1 (d
4.38) correlated with the carbonyl carbon in the p-coumarate-1 (d
168.8), and those of methylene protons at the H-6 position in Glc-3
(d 4.17) correlated with the carbonyl carbon in the p-coumarate-2
(d 168.5), those of the anomeric proton in Glc-3 (d 4.75) correlated
with the 13C signal of the C-2 carbon in the Glc-1 (d 84.5), and this
carbon signal shifted to a lower magnetic field. Therefore, 1 was
deduced to be delphinidin 3-O-(2-O-(6-O-p-coumaroyl-glucosyl)-
6-O-p-coumaroyl-glucoside)-5-O-glucoside (Fig. 2).
Anthocyanin 1 is identical to one of the compounds mentioned
in the recent study on flavonoids in V. persica flowers (Mori et al.,
2009) being submitted after our own paper. The malonylated form
of anthocyanin 1 has also been reported in Ajuga reptans (Lamiales,
Lamiaceae) and V. persica (Terahara et al., 1996; Yoshida et al.,
2009). Analysis using high-resolution LC-TOF-MS established that
both anthocyanin 1 and its mono-malonylated form accumulate
as main pigments in the blue petals in the ratio 4:6 in the fresh
petals.
On the other hand, 2 had an apigenin nucleus and two glucur-
onosyl moieties in the molecule. In the 1H{13C}-HMBC analysis of
2, cross-peaks between H-1 of GlcA-2 (d 4.68) and C-2 of GlcA-1
(d 83.9) were observed, and a ROE was observed between the H-
1 of GlcA-1 (d 5.40) and both H-6 and H-8 protons in the A ring
of the apigenin nucleus (d 6.45 and 6.76). Therefore, 2 was deduced
to be apigenin 7-O-(2-O-b-glucuronosyl)-b-glucuronide (Fig. 2).
The structure of 2 was identical to a flavone glucuronide previously
isolated from Perilla ocimoides leaves (Lamiales, Lamiaceae), and
also mentioned in the recent study on V. persica flowers (Yoshida
et al., 1993; Mori et al., 2009).
our experimental conditions, the addition of six equivalents of 2–1
resulted in the maximum bathochromic shift to 645 nm. Moreover,
* * *
the hue of this solution was 297.2° in the L a b colorimetric value,
which is in the range of that of an intact petal (297.8°). These re-
sults indicate that 2 has a strong co-pigment effect on 1 under con-
ditions where the molar ratio of 1–2 is 1:6, respectively. Actually,
pigments 1 and co-pigment 2 were estimated to be present in a
molar ratio of 1:9 in the fresh petals, which roughly fits with the
result from the in vitro co-pigment assay (Fig. 1C). Therefore, these
data strongly suggest that this co-pigmentation occurs in planta
and supposedly influences the blue coloration of V. persica flowers.
These data are remarkable since only a small bathochromic shift
of kmax was observed when 5 equivalents of 2 were added to mal-
onylated 1 (Mori et al., 2009). These contrasting results suggest
that the malonyl moiety on 1 sterically exerts an inhibitory effect
on intermolecular stacking which, in turn, reduces the co-pigmen-
tation effect. This is then consistent with the previous notion that
structural affinity between anthocyanin pigment and flavone co-
pigment is critical for intermolecular stacking (Fukui et al., 2003).
The data also implicate that, among various flavonoids existing in
bluish flowers, only selected combinations of anthocyanins and
flavones with specific and proper modifications display a co-pig-
mentation effect.
Considering the color alteration of anthocyanins upon pH
changes, we therefore conclude that the underlying reason for
the bluish hue of the V. persica flowers is provided by (1) the accu-
mulation of highly aromatic acylated delphinidin-type anthocya-
nins, (2) co-pigmentation by the proper combination of
anthocyanin pigment 1 and flavone co-pigment 2, and (3) the rel-
atively high vacuolar pH in the petals (estimated to be 6.2).
2.4. Molecular cloning of UGT88D8
2.3. Co-pigment assay
The significant bathochromic effect of co-pigment 2 on antho-
cyanin pigment 1 prompted us to investigate genes responsible
for biosyntheses of those flavonoids. Flavone 7-O-glucuronide is
known to be a specialized metabolite of Lamiales plants; therefore,
Importance of flavone glycosides as co-pigments has been doc-
umented in several species (Asen et al., 1972; Yabuya et al., 1997;
Fukui et al., 2003). To investigate whether that is the case with the
flavone 2 on anthocyanin 1, the bathochromic shift of 2 on 1 was
measured in Mcllvaine buffer (pH 6.2) (Fig. 1C). When 2 was added
to 1 in various molar ratio, a remarkable bathochromic shift of the
absorbance maximum (Dkmax) was observed in a dose-dependent
manner; the original kmax of 1, which is 615 nm, was shifted to-
ward longer wavelength as the molar ratio of 2–1 increases. Under
it is expected that
a flavonoid 7-O-glucuronosyltransferase
(F7GAT) is also present in V. persica (Noguchi et al., 2009). To inves-
tigate whether 7-O-glucuronosylation activity on apigenin exists in
V. persica, we first conducted enzyme assays using crude extracts
prepared from V. persica petals. The results show that the extracts
contain a significant level of 7-O-glucuronosylation activity on
OH
OH
3’
B
O+
HO
H
OH
A
GlcA 1
OH
OH
O
3
H
H
5
O
O
H
O
B
O
H
H
HO
7
O
O
O
H
HO
Glc 3
HO
2’’
A
OH
2’’
O
O
H
O
OH
OH
OH
O
O
O
Glc 1
H
H
H
OH
HO
HO
Glc 2
O
OH
OH
H
O
OH
O
HO
H
GlcA 2
HO
OH
O
Coum 2
Coum 1
OH
OH
1. Delphinidin 3-O-(2-O-(6-O-p-coumaroyl-glucosyl)-6-O-p-coumaroyl-glucoside)-5-O-glucoside
2. Apigenin 7-O-(2-O-glucuronosyl)-glucuronide
Fig. 2. Flavonoid structures of Veronica persica. Major anthocyanin pigment 1 and flavone co-pigment 2 isolated from the blue petals. Solid lines indicate 1H{13C}-HMBC.
Arrows indicate ROE. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

E. Ono et al. / Phytochemistry 71 (2010) 726–735
729
His:UGT88D8
E200
apigenin, which was dependent on UDP–glucuronic acid (UDP–
A
C
M
P
kDa
GlcA) (Supplementary Fig. 1). The increased level of apigenin 7-
O-glucuronide was accompanied with a decrease in apigenin level,
thereby demonstrating that UGT enzymes should be involved in
the co-pigment formation. Moreover, crude enzymes prepared
from leaves also showed glucuronosylation activity on apigenin
(Supplementary Fig. 1).
%
100
100
100
75
80
60
40
20
50
37
Detection of the glucuronosylation activity on apigenin in planta
prompted us to clone UGT genes responsible for the enzyme activ-
ity. A cDNA fragment showing a significant sequence similarity to
the snapdragon (Lamiales, Scrophulariaceae, Antirrhinum majus)
F7GAT (UGT88D4) gene was amplified in cDNAs prepared from
the petals by reverse transcription-polymerase chain reaction
(RT-PCR) (Ono et al., 2006; Noguchi et al., 2009). Its full-length se-
quence was determined by the rapid amplification of cDNA ends
(RACE) method and was designated as UGT88D8 according to the
nomenclature of UGTs (Mackenzie et al., 1997). UGT88D8
(1362 bp) contains an open reading frame (ORF) corresponding
to a protein of 454 amino acids in length with a predicted molec-
ular mass of 50.1 kDa, and shows 61% amino acid sequence identity
to snapdragon F7GAT; therefore, it is most likely to be a Lamiales
F7GAT.
5.8
n.d.
B
D
mV
40
Abs 350nm
*
30
20
10
0
0.0
2.5
5.0
7.5
10.0
12.5
mi n
%
100
88.3
34.0
2.5. Biochemical characterization of UGT88D8
18.0
14.9
13.1
12.9
13.3
1.0
To determine the biochemical property of UGT88D8, the re-
combinant protein was heterologously expressed as a His(x6)-
tag fused chimeric protein in Escherichia coli and purified with a
nickel-affinity column (Fig. 3A). The purified UGT88D8 was sub-
jected to enzymatic assays. In the reaction with apigenin and
UDP–GlcA as substrates, UGT88D8 gave a product at the same
retention time of the authentic apigenin 7-O-glucuronide when
analyzed by HPLC (Fig. 3B). The product exhibited a molecular
ion at m/z 445.0759 [MꢀH]ꢀ, which was consistent with the mass
calculation of apigenin 7-O-glucuronide (C21H17O11, 445.0771, err:
+3.1 ppm). Furthermore, LC–MS/MS analysis confirmed that a
fragment ion of the product is consistent with apigenin (m/z
269.0450 [MꢀH]ꢀ). No significant product peak was observed in
the reaction mixture without UDP–GlcA. In sharp contrast to
UDP–GlcA, neither UDP–galactose nor UDP–glucose served as an
effective sugar donor for the catalysis (Fig. 3C), demonstrating
that UGT88D8 catalyzes glucuronosylation of apigenin at the 7-
hydroxy group.
R1
R1
OH
R2
R3
R2
R3
7
7
7
O
O
HO
O
HO
R4
O
HO
OH
OH
OH
O
OH
O
Kaempferol (R1=H, R2=OH, R3=H)
Quercetin (R1=OH, R2=OH, R3=H)
Naringenin
Apigenin (R1=H, R2=OH, R3=H, R4=H)
Luteolin (R1=OH, R2=OH, R3=H, R4=H)
Diosmetin (R1=OH, R2=OCH3, R3=H, R4=H)
Scutellarin (R1=H, R2=OH, R3=H, R4=OH)
Baicalein (R1=H, R2=H, R3=H, R4=OH)
Chrysoeriol (R1=OCH3, R2=OH, R3=H, R4=H)
E
%
KPB
Tris HCl
100
80
60
40
20
Next, we tested various flavonoids as the possible sugar accep-
tors for UGT88D8. The most favorable substrate among tested for
UGT88D8 in vitro was apigenin, followed by scutellarein and cry-
soeriol, all of which have an hydroxyl group at 40-position in the
B-ring of flavone backbone (Fig. 3D). This finding is in accordance
with the observation that apigenin 7-O-(200-O-glucuronosyl)-glu-
curonide is the most abundant flavone in the petals, suggesting
that UGT88D8 plays major role in the accumulation of this fla-
vone. The sugar acceptor preference of UGT88D8 was similar to
those of non-Scutellaria F7GATs, perilla UGT88D7, sesame
UGT88D6, and snapdragon UGT88D4, and apparently distinguish-
able at the enzyme activity level from Scutellaria F7GATs, S. baical-
ensis UGT88D1 and S. laeteviolaceae UGT88D5, which specifically
catalyze flavones with ortho-substituents at the 7-position, e.g.,
baicalein (Hirotani et al., 2000; Noguchi et al., 2009). The optimal
pH was tested in a range from 5.0 to 9.5, and estimated to be 8.0
for the catalysis of UGT88D8 on apigenin (Fig. 3E). The calculated
km values for apigenin and UDP–GlcA were 10.72 1.67 and
5
5.8
6
6.5
7
7.5
8
8.2
9
9.5
pH
Fig. 3. Biochemical characterization of UGT88D8. (A) The recombinant His(-
x6)::UGT88D8 fusion protein (arrowhead) was expressed in E. coli, purified by a
His Trap™ HP column, and detected using Coomassie Brilliant Blue (CBB) staining.
M: molecular size marker, P: pellet fraction, S: soluble fraction, F: flow throw, E50,
E200, and E500: eluted fraction with 50 mM, 200 mM, and 500 mM imidazole,
respectively. (B) Chromatograms at abs 350 nm of reaction mixtures of UDPGA and
apigenin in the absence (thin line) or presence (thick line) of the enzyme. The
asterisk indicates a new product given by UGT88D8. (C) Relative sugar donor
specificity of UGT88D8. Apigenin was used as the sugar acceptor. The highest
specific activity on UDP–GlcA is set as 100%. n.d. indicates ‘‘not detected”. (D) Sugar
acceptor specificity of UGT88D8. Relative activities toward 100 lM solutions of
flavonoids are shown. The highest glucuronosylating activity toward apigenin is
taken to be 100%. (E) Optimal pH of UGT88D8 for apigenin was tested in a range
from pH 5 to 9.5 using potassium phosphate buffer (KPB) for pH 5 to 8.2 (black) and
Tris–HCl buffer for pH 9 and 9.5 (gray). The highest activity was taken to be 100%.
2.6. Phylogenetic analysis of UGT88D8
36.55 8.66 lM, respectively. The kcat value for apigenin was
determined to be 7.42 0.36 sꢀ1. These kinetic parameters were
comparable to those of a perilla F7GAT, UGT88D7 (Noguchi
et al., 2009), demonstrating that UGT88D8 is an orthologous
F7GAT of V. persica.
It is generally considered that flavonoid UGTs with the same
regio-specificity toward flavonoids form distinct phylogenetic
clusters beyond species regardless of sugar donor specificity

730
E. Ono et al. / Phytochemistry 71 (2010) 726–735
(Noguchi et al., 2008). Based on this consideration, UGT88D8 was
successfully identified from V. persica. A neighbor-joining phytoge-
netic tree showed that UGT88D8 lies in the functional cluster IIIb
of flavonoid UGTs with the regio-specificity for 7-position of flavo-
noids (Ono et al., 2006; Noguchi et al., 2009) (Fig. 4A). Furthermore,
UGT88D8 is a new member of the Lamiales F7GAT subcluster, and
is structurally more similar to non-Scutellaria F7GATs such as snap-
dragon F7GAT (UGT88D4) than to Scutellaria F7GATs in this sub-
cluster. These data support the consistency between the
biochemical similarity and the taxonomic relationship among
F7GATs.
Recent study shows that a conserved Arg residue in the C-ter-
minal PSPG-box of Lamiales F7GATs is crucial for the recognition
of the anionic carboxylate of the glucuronic acid moiety of UDP–
GlcA (Noguchi et al., 2009). Sequence alignment analysis indi-
cates that Arg352 of UGT88D8 corresponds to the Arg residue
conserved in F7GATs (Supplementary Fig. 2), suggesting that this
Arg residue contributes to the exclusive specificity of UGT88D8
for UDP–GlcA as a sugar donor (Fig. 3C). On the other hand,
the amino acid residue indispensable for the sugar donor speci-
ficity of another glucuronosyltransferase, BpGAT (UGT94B1) from
red daisy (Bellis perennis), to UDP–GA is Arg25 located at the N-
terminus, as opposed to Arg352 of UGT88D8 (Fig. 4B, Supple-
mentary Fig. 2) (Sawada et al., 2005; Osmani et al., 2008). These
results suggest that the Arg residues that are important for the
sugar donor specificity of F7GATs and BpGAT to UDP–GA oc-
curred independently as a consequence of the convergent evolu-
tion (Noguchi et al., 2009). This flexibility of altering sugar donor
specificity by substitution of only a few amino acids in turn
might explain why UGTs with equivalent regio-specificities
grouped in a clade often exhibit various substrate specificities
for sugar donors.
A
B
Fig. 4. Phylogenetic trees of UGT88-related UGTs (upper) and sugar–sugar UGTs (lower). Unrooted trees were constructed using MEGA ver. 4.1 software (http://
www.megasoftware.net/) based on the CLUSTAL-W multiple alignment using neighbor-joining (NJ) (Thompson et al., 1994; Tamura et al., 2007). The solid lines indicate the
Lamiales F7GAT cluster in the upper tree. The percentage of replicate trees in which the associated taxa clustered together in the bootstrap test (1000 replicates) is shown
next to the branches. The accession numbers of these UGTs are available in previous reports (Noguchi et al., 2008, 2009; Masada et al., 2009). Vp: Veronica persica, Am:
Antirrhinum majus, Th: Torenia hybrida, Pf: Perilla frutescens, Si: Sesamum indicum, Sb: Scutellaria baicalensis, Sl: Scutellaria laeteviolacea, Lb: Linaria bipartite, Gm: Glycine max,
At: Arabidopsis thaliana, In: Ipomoea nil, Ph: Petunia hybrida, Sr: Stevia rebaudiana, Nb: Nierembergia, Bp: Bellis perennis, Cm: Citrus maxima, Ca: Catharanthus roseus.

E. Ono et al. / Phytochemistry 71 (2010) 726–735
731
His:UGT94F1
2.7. Identification and characterization of UGT94F1
A
C
M
P
E200
kDa
Identification of UGT88D8 led to further exploration of the sec-
ond enzyme catalyzing 200-O-glucuronosylation of sugar moiety of
apigenin 7-O-glucuronide, thereby completing the structure of
co-pigment 2. Considering that the functional cluster IV of flavo-
noid UGTs represented by BpGAT (UGT94B1) has been known to
catalyze glycosylation of sugar moiety in flavonoid glycosides (su-
gar–sugar glycosylation) with a regio-specificity to the 2 or 6 posi-
tion of sugar moiety but not of flavonoid aglycones, the second GAT
is likely to be a member of this cluster (Fig. 4B) (Frydman et al.,
2004; Sawada et al., 2005; Morita et al., 2005; Noguchi et al.,
2008; Masada et al., 2009). In addition to the previous observation
of 2 in Perilla (Yoshida et al., 1993), acacetin 7-O-(2-O00-glucurono-
syl)-glucuronide was reported in Clerodendrum trichotomum (Ver-
benaceae), which lies within the Lamiales order (Okigawa et al.,
1971; Harborne and Baxter, 1999). Therefore, as is the case of the
first GAT (F7GAT), the second GAT may also be widespread in Lam-
iales plants. Based on this rationale, we isolated two candidate
cluster IV UGT genes (UGT94C2 and UGT94F1) from V. persica pet-
als by RT-PCR (Fig. 4B). UGT94C2 (1380 bp) and UGT94F1
(1356 bp); these encode putative proteins of 460 and 452 amino
acids in a length with the predicted molecular mass of 52.4 and
51.0 kDa, respectively. They showed highest structural similarity
to snapdragon UGT94C1 of unknown function (Ono et al., 2006);
UGT94C2 and UGT94F1 exhibit 58% and 49% identities at amino
acid level to UGT94C1, respectively, and 45% sequence identity to
each other.
%
100
100
80
100
75
50
37
60
40
20
4.7
n.d.
B
mV
35
A520nm
*
25
15
5
0
0.0
5.0
10.0
15.0
20.0
25.0
min
To test biochemical activities of the two UGTs, their His-tag
fused proteins were heterologously expressed in E. coli and then
purified through a nickel-affinity column (Fig. 5A). Unexpectedly,
however, both of the recombinant proteins did not show a detect-
able level of glucuronosylation activity for apigenin 7-O-glucuro-
nide. For this reason, we further searched for flavonoids other
than apigenin 7-O-glucuronide as possible substrates for UGT94C2
and UGT94F1. Inspired by the fact that anthocyanin 1 has a del-
phinidin 3-O-sophoroside structure (b1 ? 2 di-glucoside), we then
tested whether these two UGTs would glucosylate the 200-hydroxy
group of sugar moiety of delphinidin 3-O-glucoside. The data
clearly show that UGT94F1, but not UGT94C1, accepts delphinidin
3-O-glucoside as a substrate. Therefore, UGT94F1 turned out to be
a UDP–glucose-dependent glucosyltransferase toward delphinidin
3-O-glucoside (Fig. 5B). The product exhibited a molecular ion at
m/z 627.1568 [M]+, which was consistent with the mass calcula-
R1
R2
R3
OH
+
7
HO
O
OH
O
7
O
O
O
O
3
HO
O
2’’ OH
5
HO
O
OH
OH
2’’
OH
GlcA
O
Glc
OH
OH
OH
Apigenin 7-O-glucuronide
Delphinidin 3-O-glucoside (R1=OH, R2=OH, R3=OH)
Cyanidin 3-O-glucoside (R1=OH, R2=OH, R3=H)
Pelargonidin 3-O-glucoside (R1=H, R2=OH, R3=H)
D
%
100
90
80
70
60
tion of delphinidin di-glucoside (C27H31O17
, 627.1556, err:
+1.9 ppm). The 1H{13C}-HMBC spectrum showed a cross-peak be-
tween the H-1 of glucose-3(Glc-3) (d 4.71, d, J = 7.8 Hz) and the
C-2 of glucose-1 (d 83.47), consistent with covalent linkage be-
tween two glucose moieties in a b-D-glucosyl-(1 ? 2)-b-D-gluco-
side structure (Supplementary Table 1). Thus, the product was
determined to be delphinidin 3-O-(2-O-glucosyl)-glucoside (del-
phinidin 3-sophoroside) (Harborne and Baxter, 1999). In addition,
UGT94F1 also glucosylated 3-O-glucosides of cyanidin and pelarg-
onidin although both anthocyanins have not been found in V. per-
sica petals. In sharp contrast, UGT94F1 did not show glucosylation
activity for either flavonol 3-O-glucosides (kaempferol and querce-
tin), anthocyanidin aglycones (cyanidin and delphinidin), or their
3,5-O-di-glucosides. This sugar acceptor specificity was similar to
that of Ipomoea nil dusky/3GGT/UGT79G16 enzyme (Morita et al.,
2005). Given that both V. persica and I. nil accumulate anthocyanin
3,5-O-glucoside derivatives as their primary pigments in petals,
glucosylation on the 200-hydroxy group of sugar moiety of anthocy-
anidin 3-O-glucosides could occur prior to 5-O-glucosylation in
planta. The km values for delphinidin 3-O-glucoside and UDP–glu-
50
6.5
7
7.5
8
8.5
pH
Fig. 5. Biochemical characterization of UGT94F1. (A) The recombinant His
(x6)::UGT94F1 fusion protein (arrowhead) was expressed in E. coli, purified by a
His Trap™ HP column, and detected using Coomassie Brilliant Blue (CBB) staining.
M: molecular size marker, P: pellet fraction, E200: eluted fraction with 200 mM
imidazole. (B) Chromatograms at abs 520 nm of reaction mixtures of UDP–Glc and
delphinidin 3-O-glucoside in the absence (thin line) or presence (thick line) of the
enzyme. The asterisk indicates a new product given by UGT94F1. (C) Relative sugar
donor specificity of UGT94F1. Delphinidin 3-O-glucoside was used as the sugar
acceptor. The highest specific activity on UDP–glucose is set as 100%. n.d. indicates
‘‘not detected”. (D) Optimal pH of UGT94F1 for delphinidin 3-O-glucoside was
tested in a range from pH 6.5 to 8.5 using potassium phosphate buffer (KPB). The
highest activity was taken to be 100%.

732
E. Ono et al. / Phytochemistry 71 (2010) 726–735
cose are 0.32 0.05 and 0.59 0.11 mM, respectively. The sugar
donor of UGT94F1 was preferably UDP–glucose (Fig. 5C), and the
optimal pH of this enzyme was estimated to be 7.5 for the catalysis
on delphinidin 3-O-glucoside in a range between pH 6.5 and
8.5 (Fig. 5D). Taken together, we conclude that UGT94F1 is a
V. persica anthocyanidin 3-O-glucoside-200-O-glucosyltransferase
(A3G200GlcT).
mation of anthocyanin pigment 1 by catalyzing glucosylation of
sugar moiety of delphinidin 3-O-glucoside at the 2-hydroxy group
(Fig. 6). The expression of UGT94F1 in other organs such as stem
and leaf suggest that UGT94F1 might play some biological role
by accepting unknown substrates.
3. Concluding remarks
In conclusion, we analyzed flavonoids of blue petals of V. persica
and found the anthocyanin 1 and the flavone 2 as two main flavo-
noids. We then show the evidence of the bathochromic effects of 2
on coloration toward 1, strongly suggesting that an intermolecular
co-pigmentation between these two flavonoids occurs in planta. In
addition, the relatively higher pH of the petals and extensive aro-
matic acylated delphinidin-type anthocyanins should contribute
to the blue coloration.
By reverse genetics, we identified V. persica flavonoid 7-O-glu-
curonosyltransferase (UGT88D8) and anthocyanidin 3-O-gluco-
side-200-O-glucosyltransferase (UGT94F1) as new members of
functional clusters IIIb and IV of flavonoid UGTs, respectively.
The identification of UGT88D8 confirmed that the F7GAT gene is
widespread in Lamiales plants. UGT88D8 participate in flavone
co-pigment 2 formation by providing a further glucuronosylation
site for unidentified second GAT. On the other hand, UGT94F1 par-
ticipates in the anthocyanin pigment 1 formation by providing fur-
ther acylation site. Considering the preferred expression of the
UGT88D8 and UGT94F1 genes in the petals as well as the batho-
chromic effects of 2 on 1, both UGTs are involved in the bluish col-
oration of V. persica flowers. Our findings herein not only establish
the major components of a naturally occurring blue flower and
their biosynthetic enzyme genes, but also provide promising tools
for the molecular breeding of flowers with novel colors via the
modification of pigments and co-pigments.
2.8. Expression analyses of UGT88D8 and UGT94F1 genes
To probe the organ distribution of UGT88D8 and UGT94F1 tran-
scripts, RT-quantitative PCR (RT-qPCR) analysis was performed on
separate organs from Veronica plants (Fig. 6). The UGT88D8 gene
was significantly expressed in petals, which is clearly in accor-
dance with the preferred accumulation of the 7-O-glucuronosyl
apigenin in this tissue. These data, together with the consistency
of apigenin being accepted by recombinant UGT88D8 as the best
substrate among tested (Fig. 3d), and being the core structure of
glucuronides that are accumulated as major flavones in the petals
of V. persica, suggest that UGT88D8 plays key role in blue flower
coloration of V. persica by participating in the formation of co-pig-
ment 2. On the other hand, while moderate expression of UGT88D8,
and comparable level of F7GAT activity was detected from the
crude extracts of the leaves (Supplementary Fig. 1), a very low le-
vel, if any, of 2 was identified in the leaves by LC–MS analysis.
Therefore, in the case of the leaf, the aglycone substrate for
UGT88D8 is not likely to be easily available as compared to the pe-
tal. Alternatively, UGT88D8 might catalyze 7-O-glucuronosylation
of other flavonoids in the leaf.
For UGT94F1, substantial level of expression in the petals was
observed, supporting the idea that UGT94F1 is involved in the for-
3.5
4. Experimental
UGT88D8
3
4.1. Plant material
2.5
2
V. persica plants grown at Toyo University (Gunma, Japana) was
used in this work. 530 g (fresh weight) of petals were collected on
spring.
1.5
1
4.2. Chemical
0.5
0
Delphinidin chloride, apigenin, luteolin, diosmetin, chrysoeriol,
kaempferol, quercetin, and naringenin were purchased from
Funakoshi (Tokyo, Japan). Scutellarin was prepared as previously
described (Noguchi et al., 2009). Baicalein, UDP–galactose, UDP–
glucose, and UDP–glucuronic acid were purchased from Sigma
(Germany).
Leaves
Petals
Fruits
Stems
Roots
5
4.5
4
UGT94F1
3.5
3
4.3. MS measurement
2.5
2
High-resolution TOF-MS analysis of 1 and 2 was performed
using a Q-TOF Premier mass spectrometer (MICROMASS, Manches-
ter, UK) fitted with a Z-spray electrospray ion source (ESI) operated
in a positive V mode at a 3-kV capillary voltage, and a 35.0-kV cone
voltage. In the reference channel of the Lockspray ion source, to
provide a lock mass ion at m/z 556.2771 [M+H]+, a 0.6 ng/mL solu-
tion of leucine enkephalin in 50% MeCN and 0.1% (v/v) HCO2H was
continually delivered using a Jasco PV-2085 Plus semi-micro-HPLC
pump system at a flow rate of 0.01 ml/min.
1.5
1
0.5
0
Leaves
Petals
Fruits
Stems
Roots
Fig. 6. Organ-specific distribution of UGT88D8 and UGT94F1 transcripts. Expression
analysis of the UGT88D8 gene was performed by RT-qPCR. These graphs show the
relative expression of UGT88D8 (AB465708) and UGT94F1 (AB514127) in compar-
ison with the reference gene, V. persica rRNA (AF509785).
LC-TOF-MS analysis of the petal extract was carried out
using the mass spectrometer in the positive V mode under the
conditions described above.
A Shodex-Asahipack ODP-40-3E

E. Ono et al. / Phytochemistry 71 (2010) 726–735
733
column (2.0 mm ꢁ 150 mm, Showa Denko, K. K. Japan) was used
for LC on a LC-20AD HPLC (Shimadzu Co., Kyoto, Japan) with a
ternary solvent system comprised of H2O containing 0.1% (v/v)
HCO2H (A), CH3CN containing 0.1% (v/v) HCO2H (B). Each sample
was eluted using a linear gradient of 10–60% of solvent B in sol-
vent A for 20 min at a flow rate of 0.2 ml/min followed by 60%
of solvent B for 5 min.
LC-TOF-MS analysis of the enzyme reaction mixtures was car-
ried out. Develosil C30-UG-3 column (2.0 mm ꢁ 150 mm, Nomura
Chemical, Aichi, Japan) was used for LC on a LC-20AD HPLC (Shima-
dzu Co., Kyoto, Japan) with a ternary solvent system comprised of
H2O containing 0.1% (v/v) HCO2H (A), CH3CN containing 0.1% (v/v)
HCO2H (B). Each sample was eluted using a linear gradient of 50–
90% of solvent B in solvent A for 20 min at a flow rate of 0.2 ml/min
and then with 90% of solvent B for 5 min. The column eluent was
applied to the Q-TOF Premier in the negative V mode at a 3-kV cap-
illary voltage and a 35.0-kV cone voltage. The reference, leucine
enkephalin indicates m/z 554.2615 [MꢀH]ꢀ.
ing (v/v) 0.1% TFA, H2O–CH3CN (400 mL, 1:1, v/v) containing (v/
v) 0.1% TFA, and H2O–acetone (1:1, v/v) in a stepwise manner.
The MeCN–H2O (1:9) fraction, which contained anthocyanins,
was further purified by prep HPLC. The HPLC was accomplished
with the use of ODS (Capcell Pak C-18-UG-80, 2 cm ꢁ 25 cm, Shise-
ido Ltd., Japan) with a flow rate of 6 ml/min and monitoring at
A270 nm. The solvent systems used included a linear gradient elu-
tion for 60 min using 20–46% of solvent B (CH3CN–H2O (1:1) con-
taining 0.1% (v/v) TFA) in solvent A (H2O containing 0.1% (v/v) TFA),
10 min using 20% of an isocratic elution, and further elution for
15 min with 46% of solvent B isocratically. This chromatogram
gave six fractions. The main pigment Fr.3 containing 1 was further
purified by prep HPLC. The HPLC was accomplished with the use of
YMC-PAK ODS-A (2 cm ꢁ 25 cm, YMC Co. Ltd., Japan) with a flow
rate of 6 ml/min and monitoring at A270 nm. The solvent systems
used included a linear gradient elution for 40 min using 0–100% of
solvent B (50 mM KH2PO4 in MeOH:H2O (1:4, v/v) containing 0.1%
(v/v) TFA) in solvent A (50 mM KH2PO4 containing 0.1% TFA).
Anthocyanin 1 (2.3 mg) was obtained in this way.
4.4. NMR measurement
The H2O–acetone (1:1, v/v) fraction which contained co-pig-
ment 2 was further purified by prep HPLC. The HPLC was accom-
plished with the use of ODS (Capcell Pak C-18-UG-80,
2 cm ꢁ 25 cm, Shiseido Ltd., Japan) as described above. Co-pigment
2 was obtained in 14 mg by repeating this chromatogram.
The 1H NMR, 13C NMR, 1H{13C}-HSQC, 1H{13C}-HMBC, TOCSY,
DQF-COSY and ROESY spectra of 1 and 2 were obtained on a
AVANCE-750 spectrometer (BRUKER BIOSPIN, Germany). The
observation frequency for 1H and 13C measurement was
750.13 MHz and 188.62 MHz, respectively. Compound 1 was dis-
solved in a 10% (v/v) TFA–d/CD3OD mixture. Compound 2 was dis-
solved in CD3OD. The residual proton peaks and 13C peaks of
CD3OD were used as the internal standard (d 3.30 for 1H and d
48.97 for 13C).
4.8. Co-pigment analysis and absorption spectra of petals
The co-pigment effects of 2 with 1 were examined as follows.
First, compound 1 and 2 were dissolved to 100 mM in dimethyl
sulfoxide (DMSO). Then, compound 1 in DMSO was diluted to
1 mM in a McIlvaine buffer (pH 6.2), and 1–6 mM of 2 in DMSO
was added to the solution. The addition of six equivalents of 2–1
is an upper limit for a clear solution in this condition. The McIlva-
ine buffer was made by mixing 0.1 M of citric acid and 0.2 M of a
Na2HPO4 solution to adjust to pH 6.2. Ten minutes after mixing,
their visible absorption spectra at 780–380 nm and the colorimet-
4.5. pH measurement and co-pigment assay
Fresh V. persica petals (2 g) was frozen at ꢀ80 °C and homoge-
nized. The pH of the supernatant of the pressed juice was mea-
sured with an F-22 pH meter with a 6069-10C electrode (HORIBA
Ltd., Japan).
* * *
ric values of a C.I.E. L a b were measured (C.I.E., http://www.cie.-
co.at/index_ie.html). The absorption spectra for the observation of
co-pigment effects were measured on a Shimadzu UV-2500PC (Shi-
madzu Co. Ltd., Japan) with a 10 mm crystal cell. The colorimetric
4.6. HCl hydrolysis of 1 and HPLC analysis of anthocyanidin
* * *
Compound 1 (0.1 mg) was dissolved in 6 N HCl (0.2 ml) and
kept at 100 °C for 20 min. The hydrolyzed anthocyanidin was ex-
tracted with 1-pentanol (0.2 ml).
HPLC conditions: an ODS-A312 (6 mm ꢁ 150 mm, YMC Co. Ltd.,
Japan) column, a solvent flow rate of 1 ml/min, and detection at an
absorbance of 600–400 nm on a photodiode array detector SPD-
M10A (Shimadzu Co. Ltd.). Under these HPLC conditions, the reten-
tion time and kmax of delphinidin were 4.0 min and 534 nm,
respectively, which was identified in comparison with authentic
delphinidin chloride.
values of a C.I.E. L a b system under a C light source were mea-
sured with color meter CR-200 for petals and CT-210 for the solu-
tion (Minolta, Japan). The visible spectra of fresh veronica petals
were measured using a UV-2500PC spectrophotometer (Shimadzu
Co., Japan).
4.9. Molecular cloning
Total RNA was extracted from the petals of V. persica using an
RNeasy Plant Mini Kit (QIAGEN, Germany). cDNAs were synthe-
sized from 1 lg of total RNA using a First-strand Synthesis System
4.7. Pigment extraction
for RT-PCR (Invitrogen, CA). To obtain partial cDNA of UGT88D8,
PCR with rTaq DNA polymerase (TaKaRa Bio Japan) was run at
94 °C for 3 min followed by 30 cycles at 94 °C for 1 min, 54 °C for
1 min, and 72 °C for 2 min using the AmF7GAT-Fw1 (50-GTGATAG
ATTTCTTTTGCAAT-30) and AmF7GAT-Rv3 (50-ACCCTATTCATCCTCT
GCTCC-30) primers for amplification of the fragment of UGT88D8
gene, and the AmUGT94C-Fw5 (50-CTTGTGTACCACAGCCAT-30)
and AmUGT94C-Rv3 (50-GCTTTCTGTGATAGAACTCCA-30) primers
for amplification of the fragment of the UGT94F1 gene. Amplified
fragments were subcloned into pCR-TOPO II (Invitrogen, CA).
Sequencing reactions were conducted using a BigDye-Terminator
ver.3.1 cycle sequencing kit (Applied Biosystems, CA) and then
analyzed using a 3100 Genetic analyzer (Applied Biosystems, CA).
Because UGT88D8 and UGT94F1 had an incomplete open reading
frame, the 50- and 30-ends were amplified with a GeneRacer kit
Petals of V. persica (530 g fresh weight) were lyophilized, pul-
verized, and extracted with H2O–MeOH (2 L, 3:7, v/v) of containing
0.1% (v/v) TFA under room temperature overnight. After filtration,
the extract was evaporated up to a quarter of its volume under vac-
uum. The solution was applied to HP-20 (200 ml) (Mitsubishi
Chemical Co. Ltd., Japan) and afterwards washed with H2O; the fla-
vonoid fraction was eluted with H2O–CH3CN (800 mL, 4:1, v/v) and
H2O–CH3CN (400 mL, 1:1, v/v) containing 0.1% (v/v) TFA. The H2O–
MeCN (4:1, v/v) eluate was concentrated and, after being washed
with H2O, was applied to a Sephadex LH-20 (Amarshem Pharmacia
Biotech, UK). The column was eluted with H2O–CH3CN (9:1, v/v,
1.3 L) containing (v/v) 0.1% TFA, H2O–CH3CN (600 mL, 4:1, v/v)
containing (v/v) 0.1% TFA, H2O–CH3CN (400 mL, 3:7, v/v) contain-

734
E. Ono et al. / Phytochemistry 71 (2010) 726–735
(Invitrogen, CA) and the following primer sets: GR-VpUGT88D8-Rv
(50-TTCCAGGAGGGTTTCGAACGGACCATA-30) and VpUGT88D8-
nest-Rv (50-CTAGAGGTGCAACGAATAAAACTT-30) for 50-RACE, GR-
Vp UGT88D8-Fw (50-TATGGTCCGTTCGAAACCCTCCTGGAA-30) and
Vp UGT88D8-nest-Fw (50-AGGATCCTGACCTGGAAACA-30) for 3-
RACE, GR-VpUGT94F-Rv (50-CAGAGTTCCAACCCTTTGGCCACTTCT-
30) and VpUGT94F-nest-Rv (50-CTGGCTCATCTTTCTCATCAGAA-30)
for 50-RACE, and GR-VpUGT94F-Fw (50-CCGTGGGCAGCCAAACAC
GCGTTATCTCA-30) and VpUGT94F-nest-Fw (50-ACTGGCTCGCTTCT
TCCTTATGAT-30) for 30-RACE.
with 63% at a flow rate of 0.6 ml/min. Reaction mixtures containing
anthocyanins were analyzed using reversed-phase HPLC on the
column Shodex RSPak DE-413L (4.6 mm i.d. ꢁ 250 mm, SHOWA
DENKO, Kanagawa, Japan) with a linear gradient of H2O–CH3CN
(1:9 to 1:1, v/v) containing 0.5% (v/v) TFA for 15 min and then held
at this composition for 10 min at a flow rate of 0.6 ml/min. Flav-
ones/flavonols was detected at A280 nm, and anthocyanins at
A520 nm using LC-2010HT HPLC system with a SPD-M20A photo-
diode array detector (Shimadzu, Kyoto, Japan).
Crude enzyme fractions were prepared from each sample
(80 mg) of V. persica petals and leaves 80 mg (fresh weight) using
a P-PER Plant Protein Extraction kit (Thermo Scientific, IL, USA)
according to standard procedures for using the standard assay sys-
tem (see above) in the presence or absence of UDP–GlcA (Supple-
mentary Fig. 1). To determine the initial velocity of recombinant
UGT88D8, the assays were carried out under steady-state condi-
tions using the standard assay system with various substrate con-
centrations. The apparent km and kcat values for the glucuronosyl
donor and acceptor substrate in the presence of a saturating con-
centration of the counter substrate were determined by fitting
the initial velocity data to the Michaelis–Menten equation using
non-linear regression analysis (Segel, 1975; Leatherbarrow, 1990).
4.10. Heterologous expression
The coding sequence of UGT88D8 and UGT94F1 was amplified
by PCR using the specific primers of NdeI-VpF7GAT-Fw (50-CACC
CATATGGAAGACACAATCATCCT-30) and XhoI-VpF7GAT-Rv (50-CTC
GAGTTTTTACCCAATAACCAACTTGAT-30) for UGT88D8, and primers
of NdeI-VpUGT1-Fw (50-CACCCATATGGAGAAAGAAGAAGCAAAAA
TG-30) and BamHI-VpUGT1-Rv (50-GGATCCTCAATCACATTTCTGCAA
CTT-30) for UGT94F1. The amplified fragment was cloned into a
pENTR-Directional-TOPO vector (Invitrogen, CA) and sequenced
to confirm the absence of PCR errors. The plasmid of UGT88D8
and UGT94F1 cDNAs was digested using NdeI and XhoI, and NdeI
and BamHI, respectively. The resulting DNA fragment was ligated
with a pET-15b vector (Novagen) that had previously been di-
gested with NdeI and XhoI, and NdeI and BamHI. The resultant
plasmid was transformed into E. coli BL21 (DE3). The transformant
cells were prepared as previously described in Noguchi et al.
(2008). The recombinant E. coli cells were harvested by centrifuga-
tion (7000g, 15 min), washed with distilled H2O, and resuspended
in buffer A (20 mM sodium Pi (pH 7.4), containing 14 mM 2-
mercaptoethanol and 0.5 M NaCl) containing 20 mM imidazole.
The cells were disrupted at 4 °C by five cycles of ultrasonication
(where one cycle corresponds to 10 kHz for 1 min followed by an
interval of 1 min). The cell debris was removed by centrifugation
(7000g, 15 min). To the supernatant solution, polyethyleneimine
was slowly added to a final concentration of 0.12% (v/v). The mix-
ture was allowed to stand at 4 °C for 30 min, followed by centrifu-
gation (7000g, 15 min). The supernatant was applied to a HisTrap™
HP column (1 ml, GE Healthcare Bio-Science) that had been equil-
ibrated with buffer A containing 20 mM imidazole. The column
was washed with buffer A containing 20 mM imidazole, and the
enzyme was eluted with buffer A containing 200 and 500 mM
imidazole. The active column-bound fractions were concentrated
and desalted using VIVASPIN 30,000 MWCO (VIVASCIENCE, Han-
nover, Germany), followed by substitution with buffer B (20 mM
potassium Pi (pH 7.5), containing 14 mM 2-mercaptoethanol).
The protein concentration was determined using the Bradford
method (Bradford, 1976) with bovine serum albumin as a stan-
dard. SDS–PAGE was carried out according to the method of Lae-
mmli (Laemmli, 1970), and the proteins in the gels were
visualized by Coomassie Brilliant Blue R250.
4.12. RT-qPCR
Total RNA was extracted from each V. persica organ using the
RNeasy Plant Mini Kit (QIAGEN, CA). cDNAs were reverse-tran-
scribed from 1 lg of each total RNA with SuperScript III (Invitro-
gen, CA). RT-qPCR was performed using a TaqMan probe in a
7500 Real-Time PCR System (Applied Biosystems, CA) as described
in Lanot et al. (2006). Gene-specific primer sets for UGT88D8
(qVpF7GAT-Fw: 50-GCGGTTTCGGCCTCTGT-30 and qVpF7GAT-Rv:
50-TCCGATATCTTGAGGGATGATTTC-30), for UGT94F1 (qVpUGT1-
Fw: 50-CCGTGGGCAGCCAAAC-30 and qVpUGT1-Rv: 50-GCACCTG
ATGCCATAAACCA-30), and for rRNA (Accession AF509785)
(VprRNAq-Fw: 50-GCGGAAGGATCATTGTCGAT-30 and VprRNA-
qRv: 50-CTAGCGGGCGGAGCTTATTA-30) were designed using Pri-
mer Express 3.0 software (Applied Biosystems, CA). The results
were normalized to the reference gene rRNA and always expressed
as expression ratios relative to a ‘control’ value using the DDCt
method (Applied Biosystems, CA). The standard deviations corre-
DDCt+s)
spond to half the range of the expression between 2ꢀ(
and
2ꢀ( Ctꢀs), where s is the SD of DDCt (n = 2).
DD
Acknowledgements
We thank Professor P.I. Mackenzie (Flinders University, Austra-
lia) for assigning UGT numbers, Dr. A. Noguchi (University of To-
kyo) for the productive discussions related to this study, and Dr.
J. Murata (SUNBOR) for critical reading of the manuscript. We also
thank Dr. Y. Katsumoto, Mrs. K. Iwasa, Mrs. A. Saito, Ms. N. Nakam-
ura, Ms. H. Toyonaga (Suntory), and Dr. M. Horikawa (SUNBOR) for
their excellent technical support with the experimental
procedures.
4.11. Enzyme assays and kinetics
The standard reaction mixture (50
ll) consisted of a 100 mM
Appendix A. Supplementary data
glycosyl acceptor, a 2 mM glycosyl donor, a 50 mM potassium
phosphate buffer (pH 7.5), and enzymes. After a 10-min pre-incu-
bation of the mixture without the enzyme at 30 °C, the reaction
was initiated by addition of enzyme. After incubation at 30 °C for
Supplementary data associated with this article can be found, in
the online version, at doi:10.1016/j.phytochem.2010.02.008.
30 min, the reaction was stopped by the addition of 50 ll of CH3CN
References
containing 0.5% (v/v) TFA. Reaction mixtures containing flavones/
flavonols were analyzed using reversed-phase HPLC on the column
Develosil C30-UG-5 (4.6 mm i.d. ꢁ 150 mm, Nomura Chemical, Ai-
chi, Japan) with a linear gradient of 18 to H2O–CH3CN (37:63, v/v)
containing 0.1% (v/v) TFA for 10 min and further elution for 5 min
Aida, R., Yoshida, K., Kondo, T., Kishimoto, S., Shibata, M., 2000. Copigmentation
gives bluer flowers on transgenic torenia plants with the antisense
dihydroflavonol-4-reductase gene. Plant Sci. 160, 49–56.
Asen, S., Norris, K.H., Stewart, R.N., 1972. Copigmentation of aurone and flavones
from petals of Antirrhinum majus. Phytochemistry 11, 2739–2741.

E. Ono et al. / Phytochemistry 71 (2010) 726–735
735
Boulton, R., 2001. The copigmentation of anthocyanins and its role in the color of
red wine. Am. J. Enol. Vitic. 52, 67–87.
Noguchi, A., Fukui, Y., Iuchi-Okada, A., Kakutani, S., Satake, H., Iwashita, T., Nakao,
M., Umezawa, T., Ono, E., 2008. Sequential glucosylation of a furofuran lignan,
Bradford, M.M., 1976.
A
rapid and sensitive method for the quantitation of
(+)-sesaminol
by
Sesamum
indicum
UGT71A9
and
UGT94D1
microgram quantities of protein utilizing the principle of protein–dye binding.
Anal. Biochem. 72, 248–254.
Frydman, A. et al., 2004. Citrus fruit bitter flavors: isolation and functional
glucosyltransferases. Plant J. 54, 415–427.
Noguchi, A., Fukui, Y., Horikawa, M., Fukuchi-Mizutani, M., Iuchi-Okada, A., Ishiguro,
M., Kiso, Y., Nakayama, T., Ono, E., 2009. Local differentiation of sugar donor
specificity of flavonoid 7-O-glycosyltransferase in Lamiales. Plant Cell 21, 1556–
1572.
characterization of the gene Cm1, 2RhaT encoding
a
1,
2
rhamnosyltransferase, key enzyme in the biosynthesis of the bitter
a
flavonoids of citrus. Plant J. 40, 88–100.
Fukada-Tanaka, S., Inagaki, Y., Yamaguchi, T., Saito, N., Iida, S., 2000. Colour-
enhancing protein in blue petals. Nature 407, 581.
Fukui, Y., Tanaka, Y., Kusumi, T., Iwashita, T., Nomoto, K., 2003. A rationale for the
shift in colour toward blue in transgenic carnation flowers expressing the
flavonoid 3’, 5’-hydroxylase gene. Phytochemistry 63, 15–23.
Goto, T., Kondo, T., 1991. Structure and molecular stacking of anthocyanins – flower
color variation. Angew. Chem. Int. Ed. Engl. 30, 17–33.
Harborne, J.B., 1963. Plant polyphenols. X. Flavone and aurone glycosides of
Antirrhinum. Phytochemistry 2, 327–334.
Harborne, J.B., Baxter, H., 1999. The Handbook of Natural Flavonoids, vol. 2. John
Wiley and Sons, New York, USA.
Okigawa, M., Hatanaka, H., Kawano, N., Matsunaga, I., Tamura, Z., 1971. Studies on
the components of Clerodendron trichotomum Thunb. III. A new glycoside, a
cacetin-7-glucurono-(1 ? 2)-glucuronide from the leaves. Chem. Pharm. Bull.
19, 148–152.
Ono, E. et al., 2006. Yellow flowers generated by expression of the aurone
biosynthetic pathway. Proc. Natl. Acad. Sci. USA 103, 11075–11080.
Osmani, S., Bak, S., Imberty, A., Olsen, C.E., Møller, B.L., 2008. Catalytic key amino
acids and UDP–sugar donor specificity of
UGT94B1. Plant Physiol. 148, 1295–1308.
a plant glucuronosyltransferase,
Sawada, S., Suzuki, H., Ichimaida, F., Yamaguchi, M.A., Iwashita, T., Fukui, Y., Hemmi,
H., Nishino, T., Nakayama, T., 2005. UDP–glucuronic acid:anthocyanin
glucuronosyltransferase from red daisy (Bellis perennis) flowers. Enzymology
Hirotani, M., Nagashima, S., Yoshikawa, T., 1998. Baicalin and baicalein productions
of cultured Scutellaria baicalensis cells. Nat. Med. 52, 440–443.
Hirotani, M., Kuroda, R., Suzuki, H., Yoshikawa, T., 2000. Cloning and expression of
UDP–glucose: flavonoid 7-O-glucosyltransferase from hairy root cultures of
Scutellaria baicalensis. Planta 210, 1006–1013.
Katsumoto, Y. et al., 2007. Engineering of the rose flavonoid biosynthetic pathway
successfully generated blue-hued flowers accumulating delphinidin. Plant Cell
Physiol. 48, 1589–1600.
and phylogenetics of a novel glucuronosyltransferase involved in flower
pigment biosynthesis. J. Biol. Chem. 280, 899–906.
Segel, I.H., 1975. Enzyme Kinetics. Behavior and Analysis of Rapid Equilibrium and
Steady-state Enzyme Systems. John Wiley and Sons, New York.
Shiono, M., Matsugaki, N., Takeda, K., 2005. Structure of the blue cornflower
pigment. Nature 436, 791.
Tamura, K., Dudley, J., Nei, M., Kumar, S., 2007. MEGA4: molecular evolutionary
genetics analysis (MEGA) software version 4.0. Mol. Biol. Evol. 24, 1596–1599.
Terahara, N., Callebaut, A., Ohba, R., Nagata, T., Ohnishi-Kameyama, M., Suzuki, M.,
1996. Triacylated anthocyanins from Ajuga reptans flowers and cell cultures.
Phytochemistry 42, 199–203.
Kondo, T., Yoshida, K., Nakagawa, A., Kawai, T., Tamura, H., Goto, T., 1992. Structural
basis of blue-color development in flower petals from Commelina communis.
Nature 358, 515–518.
Kondo, T., Oyama, K.-I., Yoshida, K., 2001. Chiral molecular recognition on formation
of a metalloanthocyanin: a supramolecular metal complex pigment from blue
flowers of Salvia patens. Angew. Chem. Int. Ed. Engl. 40, 894–897.
Laemmli, U.K., 1970. Cleavage of structural proteins during the assembly of the
head of bacteriophage T4. Nature 227, 680–685.
Lanot, A., Hodge, D., Jackson, R.G., George, G.L., Elias, L., Lim, E.K., Vaistij, F.E., Bowles,
D.J., 2006. The glucosyltransferase UGT72E2 is responsible for monolignol 4-O-
glucoside production in Arabidopsis thaliana. Plant J. 48, 286–295.
Leatherbarrow, R.J., 1990. Using linear and non-linear regression to fit biochemical
data. Trends Biochem. Sci. 15, 455–458.
Thompson, W.R., Meinwald, J., 1972. Flavonols: pigments responsible for ultraviolet
absorption in nectar guide of flower. Science 177, 528–530.
Thompson, J.D., Higgins, D.G., Gibson, T.J., 1994. CLUSTAL W: improving the
sensitivity of progressive multiple sequence alignment through sequence
weighting, position-specific gap penalties and weight matrix choice. Nucl.
Acids Res. 22, 4673–4680.
Verweij, W., Spelt, C., Di Sansebastiano, G.P., Vermeer, J., Reale, L., Ferranti, F., Koes,
R., Quattrocchio, F., 2008. An H+ P-ATPase on the tonoplast determines vacuolar
pH and flower colour. Nat. Cell Biol. 10, 1456–1462.
Vogt, T., Jones, P., 2000. Glycosyltransferases in plant natural product synthesis:
characterization of a supergene family. Trends Plant Sci. 5, 380–386.
Yabuya, T., Nakamura, M., Iwashita, T., Yamaguchi, M., Takehara, T., 1997.
Anthocyanin-flavone copigmentation in bluish purple flowers of Japanese
garden iris (Iris ensata Thunb.). Euphytica 98, 163–167.
Mackenzie, P.I. et al., 1997. The UDP glycosyltransferase gene superfamily:
recommended nomenclature update based on evolutionary divergence.
Pharmacogenetics 7, 255–269.
Masada, S., Terasaka, K., Oguchi, Y., Okazaki, S., Mizushima, T., Mizukami, H., 2009.
Functional and structural characterization of
glucosyltransferase from Catharanthus roseus. Plant Cell Physiol. 50, 1401–1415.
Mori, M., Kondo, T., Yoshida, K., 2008. Cyanosalvianin, supramolecular blue
metalloanthocyanin, from petals of Salvia uliginosa. Phytochemistry 69, 3151–
3158.
Mori, M., Kondo, T., Yoshida, K., 2009. Anothocynian components and mechanism
for color development in blue Veronica flowers. Biosci. Biotechnol. Biochem. 73,
2329–2331.
Morita, Y. et al., 2005. Japanese morning glory dusky mutants displaying reddish-
brown or purplish-gray flowers are deficient in a novel glycosylation enzyme
for anthocyanin biosynthesis, UDP–glucose: anthocyanidin 3-O-glucoside-200-O-
glucosyltransferase, due to 4-bp insertions in the gene. Plant J. 42, 353–363.
a
flavonoid glucoside 1, 6-
Yamazaki, M., Nakajima, J., Yamanashi, M., Sugiyama, M., Makita, Y., Springob, K.,
Awazuhara, M., Saito, K., 2003. Metabolomics and differential gene expression
in anthocyanin chemo-varietal forms of Perilla frutescens. Phytochemistry 62,
987–995.
Yoshida, K., Kameda, K., Kondo, T., 1993. Diglucuronoflavones from purple leaves of
Perilla ocimoides. Phytochemistry 33, 917–919.
Yoshida, K., Kitahara, S., Ito, D., Kondo, T., 2006. Ferric ions involved in the flower
color development of the Himalayan blue poppy, Meconopsis grandis.
Phytochemistry 67, 992–998.
Yoshida, K., Mori, M., Kondo, T., 2009. Blue flower color development by
anthocyanins: from chemical structure to cell physiology. Nat. Prod. Rep. 26,
884–915.
a