A new mechanism for oxidation of epigallocatechin and production of benzotropolone pigments

Article abstract of DOI:10.1016/j.tet.2006.03.021

Enzymatic oxidation of (-)-epigallocatechin gave two new quinone dimers, dehydrotheasinensin C and proepitheaflagallin. Dehydrotheasinensin C has a hydrated cyclohexenetrione structure and its oxidation-reduction dismutation reaction yielded black tea pol

Full text of DOI:10.1016/j.tet.2006.03.021

Tetrahedron 62 (2006) 4774–4783
A new mechanism for oxidation of epigallocatechin
and production of benzotropolone pigments
*
Yosuke Matsuo, Takashi Tanaka and Isao Kouno
Graduate School of Biomedical Sciences, Nagasaki University, 1-14 Bunkyo-machi, Nagasaki 852-8521, Japan
Received 9 February 2006; revised 5 March 2006; accepted 7 March 2006
Available online 29 March 2006
Abstract—Enzymatic oxidation of (ꢀ)-epigallocatechin gave two new quinone dimers, dehydrotheasinensin C and proepitheaflagallin.
Dehydrotheasinensin C has a hydrated cyclohexenetrione structure and its oxidation–reduction dismutation reaction yielded black tea poly-
phenols, theasinensins C and E, and desgalloyl oolongtheanin. The structure of proepitheaflagallin was determined based on spectroscopic
data of its quinoxaline derivatives prepared by condensation with o-phenylenediamine. Proepitheaflagallin was decomposed on heating to
give epitheaflagallin and hydroxytheaflavin. The former is a known black tea pigment and the latter is a new pigment with 10,20,30-
trihydroxy-3,4-benzotropolone moiety. The results revealed a new mechanism for the production of these pigments from epigallocatechin.
Ó 2006 Elsevier Ltd. All rights reserved.
1. Introduction
a mechanism similar to that of theaflavin synthesis. Epithea-
flagallin (14) and its 3-O-gallate, on the other hand, have
10,20,30-trihydroxy-3,4-benzotropolone moiety, the hydrox-
ylation pattern of which is different from those of theafla-
vins. These compounds were originally synthesized by in
vitro experiments,15 and their presence in commercial black
tea was later disclosed by Nonaka et al.16 It was presumed
that 14 is produced by oxidative coupling between the B-
ring of 1 and 1,2,3-trihydroxybenzene (pyrogallol) or gallic
acid based on the results of in vitro enzymatic and chemical
synthesis.2
Tea plant (Camellia sinensis) was originally used as a medi-
cine and beverage in East Asia thousands of years ago and is
currently of agricultural and commercial importance world-
wide.1 In addition to caffeine, the presence of four catechin
monomers (epigallocatechin (1), epicatechin, and their 3-O-
galloyl esters) in a high concentration (10–25% dry weight)
is what makes tea plant distinctive from other polyphenol-
rich plants.1,2 The catechin composition of green tea, which
is commonly consumed in China and Japan, is similar to that
of fresh tea leaves, because the polyphenoloxidase is inacti-
vated by steaming or roasting immediately after harvesting
of fresh leaves. On the other hand, black tea contains a com-
plex mixture of oxidation products produced by enzymatic
oxidation of the original catechins, because the fresh leaves
are rolled and crushed during black tea manufacturing, and
thus, the catechins are mixed with active enzymes and oxy-
gen molecules. The oxidation reaction yields numerous
products, only some of which have so far been chemically
clarified.3,4
Since the contents of 1 and its 3-O-galloyl ester account for
over 70% of total tea catechins in tea leaves, oxidation of
these pyrogallol-type catechins is most important in produc-
tion of black tea polyphenols. Previously, we disclosed that
dimers of the B-ring quinones of 1 and its gallate accumulate
at the initial stage of tea fermentation.17 Furthermore, a sub-
sequent study showed that dehydrotheasinensin A (2a), one
of the quinone dimers produced from (ꢀ)-epigallocatechin-
3-O-gallate, is easily converted to stable products such as
theasinensins18 and oolongtheanins19 by oxidation–reduc-
tion dismutation.20 Since theasinensins are major black tea
constituents,2 this oxidation pathway is important in the pro-
duction of black tea polyphenols. The results also indicated
that pyrogallol-type catechin B-rings are much more easily
oxidized compared to the galloyl groups. However, partici-
pation of the galloyl groups in catechin oxidation has also
been reported recently,7,11–14,21 and it was suggested that ox-
idation of galloyl groups generates minor products, resulting
in a complex reaction mixture. Therefore, to further under-
stand the oxidation of tea catechins with pyrogallol type
B-rings, 1 should be used as the substrate rather than its
Theaflavins are well known reddish-yellow pigments char-
acteristic of black tea and with a unique 10,20-dihydroxy-3,
4-benzotropolone moiety.5 The pigments are formed by
oxidative coupling between catechol-type catechins (epica-
techin and its gallate) and pyrogallol-type catechins (1 and
its gallate). The presence of minor pigments with related
benzotropolone structures has also been reported in black
tea,6–14 and these pigments were shown to be produced by
* Corresponding author. Tel.: +81 95 819 2433; fax: +81 95 819 2477;
e-mail: t-tanaka@net.nagasaki-u.ac.jp
0040–4020/$ - see front matter Ó 2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.tet.2006.03.021

Y. Matsuo et al. / Tetrahedron 62 (2006) 4774–4783
4775
3-O-gallate. In the present study, we examined enzymatic
oxidation of 1 and found a novel oxidation pathway produc-
ing 14 and a related new pigment.
hemiacetal carbons (d 91.6 and 95.5) indicated the presence
of a hydrated form of cyclohexenetrione moiety. In addition,
the appearance of a [M+H]+ ion peak at m/z 627 in the
FABMS and the lack of signals of galloyl groups in the
NMR spectra suggested that 2 is a desgalloyl analog of 2a.
On hydrogenation with dithiothreitol, 2 was converted to
theasinensin C (5),18 which has an R-biphenyl bond, show-
ing that configuration of the benzyl methine carbon (C-20)
was S.20 In addition, treatment of 2 with o-phenylenedi-
amine yielded a phenazine derivative 2b, which was previ-
ously obtained by similar treatment of crushed fresh tea
leaves.17 In neutral phosphate buffer, 2 was decomposed to
give 5, theasinensin E (6), which is an atropisomer of 5,19
desgalloyl oolongtheanin (7),19 and dehydrotheasinensin E
(8)22 (Scheme 1). The former three products were produced
by oxidation–reduction dismutation, which also occurs in
the case of 2a.20 Product 8 was formed by isomerization at
the C-20 position of 2 and subsequent intramolecular acetal
formation. These results allowed us to conclude the structure
of 2, which we named dehydrotheasinensin C. Products
2 and 3 were considered to be dimerization products of the
B-ring o-quinone 1a, which was initially generated by
enzymatic hydrogenation. When o-phenylenediamine was
directly added to the initial enzymatic oxidation mixture,
phenazine derivatives 2b and 2c17 were produced as the ma-
jor products, and the phenazine derivative (1b)24 derived
from the monomer quinone (1a) was only obtained as one
of the minor products. This result indicated that stereoselec-
tive dimerization of the quinone 1a predominantly occurred
and this reaction is the most important oxidation route of 1.
2. Results and discussion
2.1. Enzymatic oxidation of epigallocatechin and
isolation of quinone metabolites
Substrate 1, which was prepared from commercial green
tea,18 was oxidized by mixing with a Japanese pear homog-
enate until the substrate disappeared. Japanese pear was used
because it has strong catechin oxidation activity and does not
result in interfering side products derived from compounds
originally contained in the pear fruits.22 The reaction mix-
ture was cooled and acidified prior to separation, because
quinone dimers are relatively stable in weakly acidic condi-
tions.20 After filtration, the filtrate was separated by MCI-gel
CHP20P column chromatography to give compound 2, and
further chromatography of the remaining fractions yielded
compounds 3 and 4 (Scheme 1). Among these products,
compound 3 was identified as a symmetrical quinone dimer
formerly synthesized as the radical oxidation product of 1 in
an aprotic solvent.23
Major product 2 was obtained as a white amorphous powder
and its NMR data were closely related to those of dehydro-
theasinensin A (2a). In the 13C NMR spectrum, signals of
a conjugated ketone (d 191.2), a trisubstituted double bond
(d 161.5 and 122.5), a benzyl methine (d 45.3), and two
OH
OH
OH
3"
OH
B
OH
HO
HO
OH
OH
2"
1"'
2"'
HO
HO
HO
HO
OH
OH
HO
O
O
OH
O
OH
HO
O
C
OH
OH
A
2'
H
OH
OH
OH
OH
8
3"'
OH
OH
+
1'
OH
OH
HO
O
4
O
HO
OH
2
HO
O
HO
O
6
OH 5'
O
OH
4a
3
1
OH
OH
OH
OH
OH
2
5
6
OH
HO
OH
O
O
O
OH
OH
HO
HO
O
OH
OH
O
OH
OH
OH
OH
HO
O
x 2
HO
HO
+
O
HO
O
O
OH
OH
O
H
OH
O
OH
OH
OH
HO
O
HO
O
HO
O
O
1a
H
O
O
OH
OH
OH
OH
OH
OH
3
7
8
OH
3
OH
O
c
b
2
HO
HO
O
HO2Cl
a
d
OH
O
e
k
f
i
j
O
2'
h
O
H
O
OH
3'
OH
4
Scheme 1. Production of 2–4 from 1 and decomposition of 2.

4776
Y. Matsuo et al. / Tetrahedron 62 (2006) 4774–4783
OH
OH
N
OH
OH
HO
O
HO
O
N
OH
OH
OH
OH
1b
9
OH
OH
OH
O
OH
OH
OH
HO
O
HO
OH
OH
HO
O
HO
HO
O
OH
HO
HO
O
OH
H
OH
OH
HO
O
O
N
N
N
HO
O
OH
HO
N
OH
HO
O
O
OH
O
OH
OH
OH
O
OH
OH
OH
OH
OH
OH
2a
2b
2c
Product 4 was obtained as brown amorphous powder and
showed a [M+H]+ peak at m/z 625 in FABMS, indicating
that this compound is another oxidation product with a di-
meric structure. This was supported by the observation of
two sets of signals for the catechin A- and C-rings in the
1H and 13C NMR spectra. In the 13C NMR spectrum, the
remaining signals were attributed to an acetal (d 94.0,
C-g), a methylene (d 37.4, C-f), a methine (d 49.2, C-e), a
quaternary carbon (d 44.6, C-k), four olefinic carbons (d
152.3, C-a; 128.2, C-c; 156.2, C-d; 120.5, C-j), two conju-
gated carbonyl carbons (d 185.6, C-b; 191.6, C-h), and a car-
boxyl carbon (d 166.0, C-l). The presence of a carboxyl
group was supported by the appearance of a [MꢀCO2+H]+
peak at m/z 581 in FABMS. Since the molecular weight
was calculated to be 624 from the results of FABMS, the
molecular formula was suggested as C30H24O15, and this was
supported by elemental analysis. However, the 13C NMR
spectrum showed only 29 carbon signals, suggesting that
one carbonyl carbon signal (C-i) was not detected in the
spectrum, probably due to severe broadening caused by
keto–enol tautomerization or hydration.
the C-ring of C-2, and H-e and the aliphatic methylene H-f
were correlated with C-20. H-c was also coupled with C-d
and the conjugated carboxyl carbon C-l. In addition, correla-
tions of H-e with C-k, and C-j, H-f with C-g, C-h, C-k, and
C-j, and other correlations illustrated in Figure 1 suggested
connection between C-c–C-d (–C-2)–C-e (–C-l)–C-k (–C-
1
20, –C-j)–C-f–C-g–C-h. Observation of allyl H–1H cou-
pling of H-c with H-2 and H-e in the 1H–1H COSY spectrum
supported this structure. Formation of a hemiacetal ring be-
tween the acetal carbon C-g and C-30 hydroxyl group was
deduced from resonance of the C-ring of C-40 at a higher
field (d 25. 5) compared to usual catechin C-4 carbons (d
28), which was similarly observed in the case of 8 (d
25.0).22 C-b showed no correlation peaks with any protons;
however, its chemical shift (d 185.6) indicated that this
carbonyl carbon is located between two double bonds.
4
Furthermore, appearance of a J correlation between H-e
and C-a suggested that C-a is located between C-j and
C-b. At this stage, a novel structure represented by formula
4 or its tautomer is proposed for this unstable oxidation prod-
uct. However, we could not obtain further spectral evidence
to conclude the complete structure of 4 because of tautome-
rization and gradual decomposition of the compound during
NMR measurements.
In the HMBC spectrum (Fig. 1), proton signals of the olefinic
methine H-c and aliphatic methine H-e were correlated to
2.2. Phenazine derivatives of the quinone metabolites
OH
To obtain further evidence on the structure of the new metab-
olite 4, we next performed an experiment to trap the unstable
oxidation products as quinoxaline derivatives by condensa-
tion with o-phenylenediamine, because compound 4 appar-
ently has an a-diketone or equivalent structure.17,20 After 1
was oxidized enzymatically, o-phenylenediamine was di-
rectly added to the reaction mixture, resulting in isolation
of four new condensation products, 10 (yield 0.3%), 11
(0.6%), 12 (0.9%), and 13 (0.08%), along with 2b (yield
33%), 2c (6%), 1b (3%), 3 (0.8%), 5 (0.5%), and 9 (0.3%),
and recovery of 1 (7.7%). The results confirmed the predom-
inant formation of 2 from 1 in the reaction mixture.
OH
3
2
C
O
A
O
c
b
HO
a
d
e
HO
l
OH
O
j
i
k
O
f
h
g
2'
HO
O
O
A'
OH
C'
OH
O
4
Figure 1. Selected HMBC correlations for 4.

Y. Matsuo et al. / Tetrahedron 62 (2006) 4774–4783
4777
OH
OH
OH
OH
N
N
HO
HO
O
O
HO
HO
O
N
N
S
HO2C
O
OH
OH
R
O
H
H
N
O
O
N
O
OH
OH
OH
OH
10
11
OH
OH
OH
N
O
HO
HO
O
O
HO
HO
N
OH
N
R
H
H
H
N
O
O
S
N
N
OH
O
O
OH
OH
OH
12
13
Product 10 was obtained as a brown amorphous powder and
showed UV absorption at 354 nm. FABMS exhibited
a [M+H]+ peak at m/z 697 and [MꢀCO2+H]+ peak at 653.
Since 1H and 13C NMR spectra (Table 1) indicated the pres-
ence of one quinoxaline ring,17 the molecular weight of 696
coincided with that of the expected condensation product
lated with H-e, showed a small correlation peak with H-40.
Taking the large low field shift of H-30 into account, the C-
30 hydroxyl group was believed to be esterified by this car-
boxyl group. Thus, the plane structure of 11 was deduced
from these spectral data. As for the relative stereochemistry
at C-e and C-k, H-e showed strong NOESY correlations with
both methylene protons of C-f, indicating that H-e and H-f
were located on the same side of the molecule.
1
between 4 and o-phenylenediamine. The H and 13C NMR
spectral data and HMBC correlations (Fig. 2) were also re-
lated to those of 4, indicating a partially identical structure.
The location of the quinoxaline ring at C-a (d 152.7) and C-
b (d 152.4) was determined based on the HMBC correlation
between these carbons and H-c. A chemical shift in C-h (d
193.2) indicated that this carbonyl group is conjugated
with a double bond. In addition, correlations of H-f with
Product 12 exhibited a [M+H]+ peak at m/z 653 in FABMS,
indicating a molecular weight 44 mass units smaller than
that of 10. In addition, the absence of the carboxyl carbon
signal in the 13C NMR spectrum suggested that 12 is a qui-
noxaline derivative of the decarboxylated form of 4. The 1H
and 13C NMR spectra revealed the presence of two methyl-
enes (C-e and C-f), and HMBC correlations of these methyl-
enes showed that the structure of 12 was partially similar to
that of 10, except for the absence of the carboxyl group at
C-e (Fig. 2). The chemical shift of a carbonyl carbon (d
186.1) indicated its location between two double bonds, and
the HMBC correlations of this carbon with H-2 and H-e sug-
gested that it is attributable to C-b. Correlations of H-c with
C-a (d 163.6), and H-e with C-j (d 113.4) supported construc-
tion of a seven-membered ring structure. Although C-i (d
151.7) showed no correlation peaks, correlation of C-h
with H-f showed the location of the quinoxaline ring at
C-h and C-i. The UV absorption of this product at 405 nm
reflected the long conjugation of double bonds including the
quinoxaline ring. The formation of an ether linkage between
the C-30 hydroxyl group and acetal carbon C-g was indicated
4
C-e, C-k, C-j, C-g, and C-h, and J correlation of H-e to
an oxygen-bearing olefinic carbon (C-i) allowed construc-
tion of a cyclohexenone ring. Furthermore, the upfield shift
of C-40 suggested formation of a hemiacetal ring between C-
30 and C-g. Thus, the plane structure of this quinoxaline de-
rivative was concluded to be as shown in formula 10. The
NOESY correlations of H-e with H-f, H-2, H-3, and H-20,
and H-c with H-2 and H-3 supported this structure; however,
the absolute configuration at C-e and C-k could not be deter-
mined at this stage, but was presumed to be the same as that
of the quinoxaline derivatives 12 and 13 described below.
Product 11 was shown to have two quinoxaline units using
1
FABMS (m/z 751 [M+H]+) and according to the H and
13C NMR spectra (Table 1). Other NMR signals were related
to those of 10, except for the appearance of an sp2 carbon sig-
nal at d 152.6 (C-g) instead of an acetal carbon and a large
low field shift of the C-ring of H-30 (d 5.60) compared to
that of 10 (d 4.83). The HMBC correlations (Fig. 2) revealed
the presence of similar carbon strings from C-c to C-h
through C-k. Correlations of H-c with C-a and C-b, and H-f
with C-g and C-h confirmed the location of the quinoxaline
units. The carboxyl group (d 172.4, C-l), which was corre-
4
by observation of an upfield shift of C-40 and a J HMBC
correlation between H-4 and C-g. As for stereochemistry,
strong NOEs between H-e and H-f in the NOESY spectrum
indicated that these protons were located on the same side of
the molecule. In addition, an NOE cross peak between H-2
and H-80 was also observed (Fig. 3). Since the absolute con-
figuration at C-2 and C-20 was R, this observation indicated

4778
Y. Matsuo et al. / Tetrahedron 62 (2006) 4774–4783
Table 1. 1H (500 MHz), 13C (125 MHz) NMR data for phenazine derivatives 10–13 (d in ppm, J in Hz)
Position
10
11
12
13
1H
13C
81.4
67.3
28.8
1H
13C
80.7
62.4
27.2
1H
13C
80.5
62.8
29.7
1H
13C
80.1
63.0
29.4
2
3
4
4.65 (s)
4.40 (m)
2.80–2.76 (2H, m)
4.94 (s)
4.57 (br s)
2.94 (dd, 16.9, 2.1)
2.88 (dd, 16.9, 4.6)
4.88 (s)
4.61 (br s)
2.97 (dd, 16.8, 4.1)
2.90 (dd, 16.8, 2.4)
4.79 (s)
4.63 (br s)
2.93 (br d, 16.7)
2.85 (dd, 16.7, 4.6)
4a
5
6
7
8
8a
20
30
40
99.7
157.2c
96.2d
156.9c
96.0d
156.7c
70.9
99.4
155.9c
96.4d
157.5c
95.3d
157.4c
73.1
99.5
157.7b
96.6c
157.6b
95.6c
157.5b
72.4
99.3
157.5b
96.5c
157.5b
96.4c
157.0b
73.1
5.99 (d, 2.5)a
5.95 (d, 2.5)a
6.07 (d, 1.7)a
6.05 (d, 1.7)a
6.05 (d, 2.2)a
5.94 (d, 2.2)a
6.01 (br s)
6.01 (br s)
5.04 (s)
4.83 (m)
2.80–2.76 (2H, m)
4.33 (s)
5.60 (br s)
2.80 (2H, m)
4.13 (s)
4.03 (m)
2.70 (dd, 17.4, 2.4)
2.64 (dd, 17.4, 4.6)
4.24 (s)
3.66 (m)
2.60 (dd, 17.6, 5.3)
2.50 (dd, 17.6, 1.6)
66.3
25.6
72.4
25.5
66.0
25.7
66.2
25.1
4a0
50
60
70
80
8a0
a
b
c
d
e
98.7
156.6c
95.4d
156.1c
95.3d
56.0c
152.7e
152.4e
129.7
151.0
51.4
97.4
154.8e
95.4f
99.0
157.2b
96.5c
99.1
156.3b
95.5c
155.4b
95.0c
155.4b
158.4
155.0
127.1
145.4
34.3
5.86 (d, 2.5)b
5.89 (d, 2.5)b
5.97 (d, 2.1)b
5.98 (d, 2.1)b
5.96 (d, 2.4)
5.90 (d, 2.4)
6.04 (d, 2.3)a
5.96 (d, 2.3)a
157.8
155.8b
95.2c
96.9f
157.2e
150.5
152.6
130.4
143.9
44.9
155.6b
163.6
186.1
128.3
152.4
39.9d
6.91 (s)
4.20 (s)
7.46 (s)
4.13 (s)
6.77 (s)
7.42 (d, 1.6)
3.18 (d, 16.2)
2.96 (d, 16.2)
2.52 (d, 13.2)
2.15 (d, 13.2)
3.29 (br d, 14.9)
1.91 (br d, 14.9)
2.42 (d, 13.0)
f
2.63 (d, 13.5)
2.45 (d, 13.5)
40.7
4.04 (d, 16.0)
3.77 (d, 16.0)
38.0
39.9d
33.9
2.36 (d, 13.0)
g
h
i
j
93.9
193.2
150.2
130.6f
55.1
152.6
144.2
152.0
115.2
54.3
93.1
153.4
151.7
113.4
41.2
94.5
151.4
154.0
53.4
4.91 (s)
k
53.3
l
174.9
172.4
a-OH
g-OH
PHE
16.60 (s)
6.30 (s)
8.13 (1H, m)
8.10 (1H, m)
7.92 (1H, m)
7.85 (1H, m)
8.02–8.00 (1H, m)
7.95–7.93 (1H, m)
7.73–7.70 (2H, m)
141.2
140.4
130.6f
130.3
129.0
128.8
8.18–8.16 (1H, m)
8.09–8.04 (3H, m)
7.91–7.87 (2H, m)
7.83–7.77 (2H, m)
143.1, 142.2
141.8, 140.8
131.7, 131.4
131.3, 130.7
129.9, 129.7
129.7, 129.3
140.1
137.8
132.6
130.9
129.7
126.8
8.18 (1H, m)
8.00 (1H, m)
7.86–7.81 (2H, m)
7.78–7.74 (1H, m)
7.69 (1H, m)
143.5, 142.0
141.1, 140.2
131.2, 130.9
130.5, 130.1
129.8, 129.6
129.4, 129.3
7.54 (1H, m)
7.39 (1H, m)
a–f
Assignments may be interchanged in each column.
OH
OH
OH
OH
3
that the configurations at C-g and C-k were S and R, respec-
tively.
N
2
N
b
c
2
c
HO
HO
O
O
b
HO
HO
O
d
e
a
a
d
e
N
l
N
k
j
The 1H NMR spectra of product 13 were related to those of
12 (Table 1), showing signals of two sets of catechin A- and
C-rings, two methylenes (H-e and H-f), and an olefinic
methine (H-c). However, two sets of signals arising from
quinoxaline moieties and a singlet aliphatic methine signal at
d 4.91 also appeared. In the 13C NMR spectrum, a methine
carbon (C-j) and two sp2 carbons bearing nitrogen atoms
(C-a and C-b) were observed instead of the enone unit at
C-b, C-a, and C-j in 12. Remaining carbon signals and their
HMBC correlations (Fig. 2) were similar to those of 12, sup-
porting close structural relationships between 12 and 13. The
C-a and C-b of 13 were correlated with both the H-c and
newly appeared methine proton (H-j) in the HMBC spec-
trum, indicating that 13 was generated by condensation of
additional o-phenylenediamine to the C-a and C-b positions
of 12. This was supported by the observation of a [M+Na]+
peak at m/z 747 in the MALDI-TOF MS. In the NOESY
HO2C
O
2'
k
j
l
OH
OH
f
i
O
i
f
h
h
2'
g
g
N
O
N
O
O
3'
OH
4'
OH
OH
OH
10
11
OH
OH
OH
2
b O
a
N
2
c
k
c
k
b
O
O
HO
HO
HO
HO
a
d
e
d
e
OH
N
N
j
j
i
N
O
i
O
4'
2'
4'
f
h
f
2'
g
h
g
N
N
O
OH
OH
O
OH
OH
12
13
Figure 2. Selected HMBC correlations for 10–13.

Y. Matsuo et al. / Tetrahedron 62 (2006) 4774–4783
4779
c
c
a
2
2
e
e
a
i
j
j
e
e
i
2'
f
2'
f
h
f
3'
8'
f
g
h
g
12
13
Figure 3. Selected NOESY correlations for 12 and 13.
spectrum (Fig. 3), H-j showed NOE correlations with H-20,
H-30, and H-c. In addition, NOEs between H-e and H-f,
and between H-2 and H-20 indicated that the configurations
at C-g, C-j, and C-k were S, R, and R, respectively.
by comparisons of 1H and 13C NMR, UV and MS spectra.16
At this stage, the unstable intermediate 4 was named proepi-
theaflagallin because it was found to be a precursor of the
black tea pigment 14.
Production of the derivatives 10–13 not only supported the
proposed structure of 4, but also indicated the occurrence
of decarboxylation at the C-e in the molecule of 4 in the
following metabolism.
Pigment 15 showed characteristic UV absorptions at 286,
309, 378, and 428 nm, which closely resembled to those of
14. The 1H and 13C NMR spectra were also related to those
of 14, suggesting the presence of a trihydroxybenzotropolone
unit. However, two sets of signals arising from catechin A-
and C-rings were also observed. In addition, the absence
2.3. Degradation products of the quinone metabolites
1
of an aromatic singlet attributable to H-f of 14 in the H
The unstable product 4 was expected to be a precursor of
some black tea polyphenols, as we have shown that 2 was
a precursor of important black tea polyphenols 5, 6, and 7.
So, we examined decomposition of 4 on heating, which is
expected to occur at the final stage of black tea manufactur-
NMR spectrum and low field shift of the C-f signal (d
115.4, 14: d 111.8) in the 13C NMR spectrum indicated
that 15 has additional catechin A- and C-rings at the C-f po-
sition of compound 14. This structure was confirmed by the
HMBC correlations shown in Figure 4. Interestingly, the
chemical shift of the C-i hydroxyl proton (d 15.21) indicated
hydrogen bonding of this hydroxyl group with the adjacent
carbonyl group at C-a. This pigment was found to be
a new compound and named hydroxytheaflavin.
ꢁ
ing. An aqueous solution of 4 was heated at 80 C for 10 min
and then analyzed by HPLC. Photodiode array detection re-
vealed production of two pigments, 14 and 15; however, iso-
lation of these pigments directly from the reaction mixture
failed owing to the shortage of 4. Isolation of 14 and 15
was achieved by performing a large-scale experiment in
which the initial enzymatic oxidation products of 1 were
heated and the unstable products including 2 and 4 were con-
verted to more stable products (5, 6, 7, 8, 14, and 15). Here, it
might be of interest to note that MALDI-TOF MS analysis of
the total mixture of phenolic products suggested production
of a trimer and tetramer of 1, though the peak intensity was
much smaller than those of the dimeric products.
OH
δ
H 8.827
OH
OH
b
c
2
δ
H
H 15.213
a
d
O
HO
O
j
k
O
i
e
f
HO
O
h
OH
g
2'
OH
OH
OH
OH
HO
OH
15
OH
OH
OH
O
O
Figure 4. Selected HMBC correlations for 15.
HO
O
OH
OH
O
HO
HO
O
3. Conclusion
OH
OH
OH
OH
Many theaflavin related pigments, namely, theaflavins,
isotheaflavins,6,7 neotheaflavins,7,8 theaflavic acids,9,10 thea-
flavates,7,11,12 theadibenzotropolones,13,14 and theatribenzo-
tropolone A,14 have been isolated and synthesized. All
were produced by oxidative condensation between a catechol
ring and pyrogallol ring. Theaflagallins, including 14, are
exceptional because they were produced by condensation
OH
OH
15
14
Pigment 14 was identified as epitheaflagallin, having a
characteristic 10,20,30-trihydroxy-3,4-benzotropolone unit,

4780
Y. Matsuo et al. / Tetrahedron 62 (2006) 4774–4783
H+
O
1a
O
O
H
O
O
O
H+
R
OH
OH
R
O H+
R
R
O
O
R
HO-
OH
- H2
HO2C
OH
O
O
R
H
H O
R
R
H
O
O
R
H+
R
O
H
O
O
O
O
H+
O
H
O
O
HO
HO
O
4a
HO
O
O
1a
R =
12, 13
OH
OH
R
OH
O
HO
O
OH
OH
3
+
OH
OH
(a)
(b)
R
OH
R
OH
(a)
HO-
OH
16
HO
O
O
R
H
H
14
R
O
OH
R
OH
R
OH
O
O
O
HO
O
H+
H+
O
rearrangement
10a
R
OH
R
H
OH
OH
HO
OH
HO
10, 11
15
Scheme 2. Proposed mechanisms for production of 14 and 15 from 1.
between two pyrogallol rings.15,16 Since pyrogallol itself
does not occur in fresh tea leaf, 14 and its gallate found in
black tea were considered to have formed by condensation
with gallic acid via decarboxylation.2,16 However, our re-
sults revealed that 14 was produced from 1 alone even in
the absence of gallic acid, and we succeeded in confirming
the key intermediate of this mechanism. Details of the pro-
posed production of 14 and 15 are shown in Scheme 2. In
the last step, migration or elimination of the flavan C-ring
occurs; this was supported by the fact that compound 16,
which corresponds to the eliminated flavan A, C-rings
were actually isolated from black tea.20 Recently, we also
demonstrated that the 3-O-galloyl ester of 14 is produced
by enzymatic oxidation of the 3-O-gallate of 1,20 and the
reaction probably proceeds in a similar manner.
used as the matrix for the FABMS measurements.
MALDI-TOF MS was measured on a Voyager-DE PRO
(Applied Biosystems), and 2,5-dihydroxy benzoic acid
(10 mg/mL) was used as the matrix.
Column chromatography was performed with Diaion
HP20SS, MCI-gel CHP20P (75–150 mm, Mitsubishi Chem-
ical Co.), Sephadex LH-20 (25–100 mm, Pharmacia Fine
Chemical Co. Ltd), TSK-gel Toyopearl HW-40 (TOSO
Co. Ltd), and Chromatorex ODS (Fuji Silysia Chemical
Ltd). TLC was performed on precoated Kieselgel 60 F254
plates (0.2 mm thick, Merck) with benzene–ethyl formate–
formic acid (1:7:1, v/v) or chloroform–methanol–water
(14:6:1, v/v) and spots were detected by ultraviolet (UV) il-
lumination and by spraying with 2% ethanolic FeCl3 or 10%
sulfuric acid reagent followed by heating. Analytical HPLC
was performed on a Cosmosil 5C18-AR II (Nacalai Tesque
Inc.) column (250ꢂ4.6 mm i.d.) with gradient elution from
10 to 30% (30 min) and from 30 to 75% (15 min) of
CH3CN in 50 mM H3PO4 (flow rate, 0.8 mL/min; detection:
JASCO photodiode array detector MD-910). Epigallo-
catechin was isolated from commercial green tea and recrys-
tallized from water.
4. Experimental
4.1. General
UV spectra were obtained with a JASCO V-560 UV/VIS
spectrophotometer. Optical rotations were measured with
a JASCO DIP-370 digital polarimeter. 1H, 13C NMR,
1H–1H COSY, NOESY, HSQC and HMBC spectra were
recorded in a mixture of acetone-d6 and D2O (19:1, v/v) at
4.2. Enzymatic oxidation of epigallocatechin and
isolation of quinone metabolites
ꢁ
27 C with a Varian Unity plus 500 spectrometer operating
at 500 MHz for 1H NMR and 125 MHz for 13C NMR. Cou-
pling constants are expressed in Hertz, and chemical shifts
are given on a d (ppm) scale with tetramethylsilane as an in-
ternal standard. MS were recorded on a JEOL JMS DX-303
spectrometer, and glycerol or m-nitrobenzyl alcohol was
Japanese pear fruits (200 g) were homogenized with 200 mL
ꢁ
of H2O and filtered through four layers of gauze at 0 C. The
filtrate was mixed with an aqueous solution of 1 (1.5 g/
75 mL) and vigorously stirred for 90 min at room tempera-
ture. The mixture was acidified with 2 mL of trifluoroacetic

Y. Matsuo et al. / Tetrahedron 62 (2006) 4774–4783
4781
1
acid (TFA) and directly applied to a column of MCI-gel
CHP20P (3 cm i.d.ꢂ30 cm). After washing with 0.1%
TFA, the column was eluted with 0–50% MeOH in 0.1%
TFA (5% stepwise gradient elution, each 200 mL), and frac-
tions (each 15 mL) were analyzed by HPLC. Fractions con-
taining 2 were collected and lyophilized to give a white
powder (422 mg). Fractions containing 3 and 4 were sepa-
rately collected and purified by Chromatorex ODS (H2O–
MeOH) to give 3 (47 mg) and 4 (50 mg). Compound 3
was identified with a known oxidation product of 1 by com-
parison of the spectral data.23
1H–1H coupling observed in H–1H COSY, H-2/H-c, H-e/
H-c, H-20/H-f (d 2.08); 13C NMR (125 MHz, d6-acetone)
d 25.5 (C-40), 29.0 (C-4), 37.4 (C-f), 44.6 (C-k), 49.2 (C-e),
65.2 (C-3), 66.2 (C-30), 94.0 (C-g), 95.2, 95.4, 96.6, 96.7
(C-6, C-8, C-600, C-800), 98.3, 99.1 (C-4a, C-4a0), 120.5
(C-j), 128.2 (C-c), 152.3 (C-a), 155.2, 155.3, 157.0, 157.5,
157.6 (2C) (C-5, C-7, C-8a, C-50, C-70, C-8a0), 156.2
(C-d), 116.0 (C-l), 185.6 (C-b), 191.6 (C-h) C–I was not
detected; important HMBC correlations (H to C), H-c/C-2,
C-3, C-d, C-e, C-k, C-l, H-e/C-2, C-20, C-a, C-c, C-d, C-f,
C-k, C-j, C-l, H-2/C-d, C-c, C-e, H-20/C-e, C-f, C-k, H-f/
C-e, C-k, C-j, C-g, C-h, C-20; FABMS m/z 625 [M+H]+,
581 [MꢀCO2+H]+; Anal. Calcd for C30H24O15$3.5H2O: C,
52.41; H, 4.54. Found: C, 52.14; H, 4.58.
4.2.1. Dehydrotheasinensin C (2). White amorphous pow-
der, [a]2D0 ꢀ74.5 (c 0.1, MeOH); IR nmax 3365, 2920, 1696,
1
1627, 1606 cmꢀ1; UV (MeOH) lmax 269(3 3640) nm; H
NMR (500 MHz, d6-acetone) d 2.70 (1H, dd, J¼16.4,
4.4 Hz, H-400), 2.73 (2H, br s, H-4), 2.95 (1H, dd, J¼16.4,
4.4 Hz, H-400), 4.17 (1H, m, H-3), 4.34 (1H, m, H-300), 4.42
(1H, s, H-20), 4.73 (1H, br s, H-2), 5.62 (1H, br s, H-200),
5.94, 5.97, 5.97, 6.00 (each 1H, d, J¼2.2 Hz, H-6, 8, 600,
800), 6.43 (1H, s, H-60), 6.80 (1H, s, H-6000); important long-
4.3. Enzymatic oxidation of epigallocatechin and
treatment with o-phenylenediamine
Japanese pear fruits (4 kg) were homogenized with 1.6 L of
ꢁ
H2O and filtered through four layers of gauze at 0 C. The
filtrate (4.0 L) was mixed with an aqueous solution of 1
(30 g/1.0 L) and vigorously stirred for 2.5 h at room temper-
ature. The mixture was poured into EtOH (15 L) containing
o-phenylenediamine (11.5 g) and AcOH (750 mL), and
stirred gently for 30 min. After removal of insoluble precip-
itates by filtration, the filtrate was concentrated and applied
to Diaion HP20SS (5.5 cm i.d.ꢂ55 cm) column chromato-
graphy. After washing the column with H2O to remove
sugars and AcOH, the products were eluted out with H2O
containing 0–100% MeOH (10% stepwise elution, each
1.0 L) yielding nine fractions: fr. 1 (4.0 g), fr. 2 (0.89 g),
fr. 3 (16.5 g), fr. 4 (2.59 g), fr. 5 (1.81 g), fr. 6 (3.18 g), fr.
7 (1.07 g), fr. 8 (1.77 g), and fr. 9 (0.56 g). Fr. 1 was suc-
cessively subjected to Sephadex LH-20 (0–60% MeOH,
gradient elution), Diaion HP20SS (0–50% MeOH), and
Chromatorex ODS (0–50% MeOH) chromatography to
give 1 (2.3 g), 3 (265 mg), 5 (145 mg), and 9 (93 mg). Crys-
tallization of fr. 2 from water yielded 1 (0.5 g). The recovery
of starting material 1 was 9.3%. Separation of fr. 3 by succes-
sive chromatography over MCI-gel CHP20P (10–100%
MeOH) and Sephadex LH-20 (20–100% MeOH) yielded
2b (10.24 g) and 2c (1.84 g), which were identified as iso-
mers of phenazine derivatives prepared in our previous
work.17 Fr. 4 was applied to Sephadex LH-20 column
chromatography (30–90% MeOH) and then further sepa-
rated by Chromatorex ODS (20–80% MeOH) and MCI-gel
CHP20P (25–75% MeOH) chromatography to yield a phen-
azine derivative 10 (97.2 mg). Fr. 6 was separated into two
fractions, fr. 6-1 and fr. 6-2, by Sephadex LH-20 chromato-
graphy (50–90% MeOH). The concentration of aqueous
MeOH solution of fr. 6-1 yielded reddish-brown powder
(806 mg), which was identified as a phenazine derivative
of epigallocatechin (1b). Fr. 6-2 gave microcrystalline pow-
der of 12 (272 mg) from aqueous MeOH. Fr. 7 was separated
by Chromatorex ODS (40–90% MeOH) and Sephadex LH-
20 (30–90% MeOH) chromatography to give 1b (200 mg).
Fr. 8 was separated into two fractions, fr. 8-1 and fr. 8-2,
by Sephadex LH-20 chromatography (30–100% MeOH).
Fr. 8-1 was successively chromatographed over MCI-gel
CHP20P (30–80% MeOH), Sephadex LH-20 (60–80%
MeOH), Toyopearl HW-40 (40–80% MeOH), and Sephadex
LH-20 (EtOH) to give 13 (23.6 mg). Chromatography of fr.
range H–1H couplings observed in H–1H COSY, H-2/H-
20, H-2/H-60; 13C NMR (125 MHz, d6-acetone) d 27.5,
29.0 (C-4, C-400), 45.3 (C-20), 64.4 (C-3), 65.6 (C-300), 75.5
(C-200), 78.2 (C-2), 91.6 (C-30), 95.0, 95.2, 95.5, 95.9, 96.1
(C-6, C-8, 000C-40, C-600, C-800), 99.0, 99.1 (C-4a, C-4a0),
108.2 (C-6 ), 112.0 (C-2000), 122.5 (C-60), 127.3 (C-1000),
132.4 (C-4000), 142.3 (C-3000), 145.1 (C-5000), 155.4, 156.4,
156.7, 156.9 (2C), 156.9 (C-5, C-7, C-8a, C-500, C-700, C-
8a00), 161.5 (C-10), 191.2 (C-00); important HMBC correla-
tions (H to C), H-2/C-10, C-20, C-60, H-20/C-10, C-30, C-40,
C-60, C-1000, C-2000, C-3000, H-200/C-1000, C-2000, C-6000, H-6000/C-
1000, C-2000, C-4000, C-5000; FABMS m/z 627 [M+H]+, 609
[MꢀH2O+H]+; Anal. Calcd for C30H26O15$7H2O: C,
47.88; H, 5.36. Found: C, 47.89; H, 5.21.
1
1
4.2.1.1. Treatment of 2 with o-phenylenediamine.
Compound 2 (2 mg) was treated with a solution of o-phenyl-
enediamine (2 mg) in 5% AcOH in EtOH (1 mL) at room
temperature for 1 h. HPLC analysis of the reaction mixture
showed a product peak at tR 30.3 min corresponding to the
phenazine derivative 2b.
4.2.1.2. Degradation of 2 under neutral conditions.
Compound 2 (100 mg) was dissolved in 0.1 mM Na–K
phosphate buffer (pH 7.1) (100 mL) and stirred for 3 h at
room temperature. The mixture was acidified to pH 2 by ad-
dition of a few drops of diluted HCl and directly applied to
a Sephadex LH-20 column (2 cm i.d.ꢂ20 cm) with water
containing increasing proportions of MeOH (0–80%) to
give 5 (24.0 mg), 6 (3.3 mg), 7 (22.3 mg), and 8 (6.8 mg).
4.2.2. Proepitheaflagallin (4). Brown amorphous powder,
[a]2D0 92.5 (c 0.1, MeOH); IR nmax 3403, 1706, 1631, 1519,
1
1468 cmꢀ1; UV (MeOH) lmax 271(3 9490) nm; H NMR
(500 MHz, d6-acetone) d 2.09 (1H, d, J¼13.0 Hz, H-f’),
2.58 (1H, d, J¼13.0 Hz, H-f), 2.82 (2H, m, H-40), 2.84
(1H, dd, J¼16.8, 2.7 Hz, H-4), 2.91 (1H, dd, J¼16.8,
4.3 Hz, H-4), 3.88 (1H, br s, H-20), 4.47 (1H, br s, H-3),
4.52 (1H, m, H-300), 4.81 (1H, s, H-2), 4.81 (1H, s, H-e),
5.94, 5.97, 5.95, 5.97, 6.03, 6.07 (each 1H, d, J¼2.3 Hz,
H-6, 8, 600, 800), 6.67 (1H, s, H-c); important long-range

4782
Y. Matsuo et al. / Tetrahedron 62 (2006) 4774–4783
8-2 over Sephadex LH-20 (50–90% MeOH) yielded 11
(190.2 mg).
appeared. The mixture was poured into a stainless beaker
(10ꢁL) and heated until the temperature was elevated to
80 C in order to inactivate the enzymes and degrade the un-
stable quinone dimers. The stainless beaker was then placed
in ice water to cool down the reaction mixture. To the mix-
ture, acetone (15 L) was added and stirred gently for 30 min,
and then the precipitates were removed by filtration. A sam-
ple (50 mL) of the filtrate was concentrated and applied to
a Diaion HP20SS (2 cm i.d.ꢂ10 cm). After washing the col-
umn with water, the polyphenols were eluted out with 70%
MeOH and concentrated to dryness. The total mixture of
products thus obtained was subjected to MALDI-TOF MS
analysis. The spectrum showed large peaks at m/z 603 (rela-
tive intensity 37%), 633 (100%), and 649 (52%) correspond-
ing to the [M+Na]+ of 7, [M+Na]+ and [M+K]+ of 5 or 6,
respectively. The spectrum also showed peaks at 936
(14%), 952 (12%), 972 (7%), 1212 (18%), 1228 (12%),
and 1240 (10%). The remainder of filtrate was concentrated
until acetone was removed and subjected to Diaion HP20SS
(5.5 cm i.d.ꢂ55 cm), then the column was washed with
H2O. The products were eluted out with H2O–MeOH
(10% stepwise elution, each 1.0 L), yielding six fractions:
fr. 1 (13.2 g), fr. 2 (5.7 g), fr. 3 (9.5 g), fr. 4 (4.5 g), fr. 5
(2.0 g), and fr. 6 (1.4 g). HPLC and TLC analyses indicated
the presence of 5 and 6 in fr. 1, 1 in fr. 2, 1 and 7 in fr. 3, and 7
and 8 in fr. 4 as the major constituents of each fraction.
HPLC analysis showed that fr. 6 contained the pigments pro-
duced from 4. This fraction was separated by Sephadex LH-
20 (40–100% MeOH) and Chromatorex ODS (20–80%
MeOH) to give 14 (530.8 mg) and 15 (185.2 mg). Pigment
14 was identified as epitheaflagallin by comparison of spec-
tral data and co-HPLC with authentic sample.16
4.3.1. Phenazine derivative 10. Brown amorphous powder,
[a]2D6 104.5 (c 0.1, MeOH); IR (dry film) nmax 3383, 1696,
1627, 1609, 1519, 1469 cmꢀ1; UV (MeOH) lmax 354 (3
1
7520) nm; H and 13C NMR data, see Table 1; important
NOESY correlations: H-c/H-2 and H-3, H-e/H-2, H-3,
H-20, H-f (d 2.63), and H-f (d 2.452); FABMS m/z
697 [M+H]+, 653 [MꢀCO2+H]+; Anal. Calcd for
C36H28N2O13$4.5H2O: C, 55.60; H, 4.80; N, 3.60. Found:
C, 55.45; H, 4.82; N, 3.57.
4.3.2. Phenazine derivative 11. Brown amorphous powder,
[a]2D6 196.8 (c 0.1, MeOH); IR (dry film) nmax 3366, 1715,
1626, 1609, 1519, 1467 cmꢀ1; UV (MeOH) lmax 249 (3
30,900), 382 (16,300) nm; 1H and 13C NMR data, see Table
1; important NOESY correlations: H-c/H-2 and H-3, H-e/
H-2, H-3, H-f (d 4.039), and H-f (d 3.768), H-f (d 4.039)/
H-20, H-e, and H-f (d 3.768), H-f (d 3.768)/H-2, H-e, and
H-f (d 4.039); FABMS m/z 751 [M+H]+; Anal. Calcd for
C42H30N4O10$5.5H2O: C, 59.36; H, 4.86; N, 6.59. Found:
C, 59.49; H, 4.78; N, 6.25.
4.3.3. Phenazine derivative 12. Yellow powder (from H2O–
MeOH, 4:1), [a]2D6 ꢀ104.5 (c 0.1, acetone); IR (dry film) nmax
3394, 1697, 1631, 1521, 1469 cmꢀ1; UV (MeOH) lmax 252
1
(3 18,000), 405 (12,700) nm; H and 13C NMR data, see
Table 1; important NOESY correlations: H-2/H-80, H-c/H-2
and H-20, H-e (d 3.181)/H-f (d 2.153) and H-e (d 2.960), H-e
(d 2.960)/H-f (d 2.524) and H-e (d 3.181); FABMS m/z 653
[M+H]+; Anal. Calcd for C35H28N2O11$4.5H2O: C, 57.30;
H, 5.08; N, 3.82. Found: C, 57.01; H, 5.39; N, 3.76.
4.5.1. Hydroxytheaflavin. Red amorphous powder, [a]D20
ꢀ230.6 (c 0.1, MeOH); IR (dry film) nmax 3366, 1627,
1604, 1517, 1467, 1430 cmꢀ1; UV (EtOH) lmax 286 (3
21,500), 309 (25,300), 378 (4990), 428 (2910) nm; 1H
NMR (500 MHz, d6-acetone) d 2.80 (1H, dd, J¼16.8,
2.0 Hz, H-4), 2.93 (1H, dd, J¼16.8, 4.3 Hz, H-4), 2.95
(1H, br d, J¼17.0 Hz, H-40), 3.06 (1H, dd, J¼17.0, 4.3 Hz,
H-40), 4.36 (1H, br s, H-3), 4.72 (1H, br s, H-30), 4.95 (1H,
s, H-2), 5.94 (1H, s, H-20), 5.96, 6.04 (each 1H, d,
J¼2.3 Hz, H-8 and H-6), 5.97, 6.09 (each 1H, d,
J¼2.3 Hz, H-80 and H-60), 7.51 (1H, d, J¼0.9 Hz, H-c),
8.10 (1H, br s, H-e), 8.83 (1H, br s, OH at C-b), 15.21
(1H, s, OH at C-i); 13C NMR (125 MHz, d6-acetone)
d 29.4 (C-4), 29.9 (C-40), 66.4 (C-3), 66.8 (C-30), 78.6 (C-
20), 81.7 (C-2), 95.6, 96.9 (C-8 and C-6), 95.7, 96.5 (C-80
and C-60), 99.5 (C-4a), 99.6 (C-4a0), 115.4 (C-f), 117.3
(C-c), 117.4 (C-j), 127.8 (C-e), 132.0 (C-k), 135.8 (C-d),
136.0 (C-h), 151.8 (C-i), 152.2 (C-g), 154.9 (C-b), 156.6,
156.7, 157.6 (2C), 157.7, 157.9 (C-5, C-50, C-7, C-70,
C-8a, and C-8a0), 183.1 (C-a); HR-FABMS m/z 581.1294
[M+H]+ (Calcd for C29H25O13: 581.1295).
4.3.4. Phenazine derivative 13. Brown amorphous powder,
[a]2D6 234.1 (c 0.1, MeOH); IR (dry film) nmax 3369, 1715,
1626, 1489, 1467 cmꢀ1; UV (MeOH) lmax 235 (3 62,400),
324 (18,100) nm; 1H and 13C NMR data, see Table 1; impor-
tant NOESY correlations: H-2/H-20, H-j/H-20, H-30, and H-c,
H-e (d 3.287)/H-2, H-3 and H-e (d 1.914), H-e (d 1.914)/H-f
(d 2.361); TOF MS m/z 747 [M+Na]+, 763 [M+K]+; Anal.
Calcd for C41H32N4O9$6.5H2O: C, 58.50; H, 5.39; N,
6.66. Found: C, 58.59; H, 4.90; N, 6.54.
4.4. Degradation of 4
ꢁ
Aqueous solution of 4 (0.1 mg/0.1 mL) was heated at 80 C
for 10 min and then 5 mL of the solution was analyzed by
HPLC. HPLC showed peaks arising from two pigments at
38.3 and 42.1 min, respectively. The retention times and
UV absorptions of these peaks coincided with those of 14
and 15, respectively.
4.5. Oxidation of epigallocatechin and heating of the
reaction mixture
Japanese pear fruits (5.5 kg) were homogenized with 2.2 L
ꢁ
Acknowledgments
of H2O and filtered through four layers of gauze at 0 C.
The filtrate (6.7 L) was mixed with an aqueous solution of
1 (40 g/0.8 L) and vigorously stirred at room temperature,
and then the mixture was analyzed by HPLC at intervals
of 30 min. After 3 h, the starting material had almost dis-
appeared and a large peak of 2 and moderate peak of 4
The authors are grateful to Mr. K. Inada and Mr. N. Yamagu-
chi (Nagasaki University) for NMR and MS measurements.
This work was supported by a Grant-in-aid for Scientific
Research (No. 12680594) from the Japan Society for the
Promotion of Science.

Y. Matsuo et al. / Tetrahedron 62 (2006) 4774–4783
4783
12. Sang, S.; Lambert, J. D.; Tian, S.; Hong, J.; Hou, Z.; Ryu, J. H.;
Stark, R. E.; Rosen, R. T.; Huang, M. T.; Chung, S.; Yang, C. S.;
Ho, C. T. Bioorg. Med. Chem. 2004, 12, 459–467.
13. Sang, S.; Tian, S.; Meng, X.; Stark, R. E.; Rosen, R. T.; Yang,
C. S.; Ho, C. T. Tetrahedron Lett. 2002, 43, 7129–7133.
14. Sang, S.; Tian, S.; Stark, R. E.; Yang, C. S.; Ho, C. T. Bioorg.
Med. Chem. 2004, 12, 3009–3017.
15. (a) Takino, Y.; Ferretti, A.; Flanagan, V.; Gianturco, M. A.;
Vogel, M. Can. J. Chem. 1967, 45, 1949–1956; (b) Takino, Y.;
Imagawa, H. Agric. Biol. Chem. 1964, 28, 125–130; (c) Rob-
erts, E. A. H.; Myers, M. J. Sci. Food Agric. 1959, 10, 167–172.
16. Nonaka, G.; Hashimoto, F.; Nishioka, I. Chem. Pharm. Bull.
1986, 34, 61–65.
17. Tanaka, T.; Mine, C.; Watarumi, S.; Fujioka, T.; Mihashi, K.;
Zhang, Y.-J.; Kouno, I. J. Nat. Prod. 2002, 65, 1582–1587.
18. Nonaka, G.; Kawahara, O.; Nishioka, I. Chem. Pharm. Bull.
1983, 31, 3906–3914.
19. Hashimoto, F.; Nonaka, G.; Nishioka, I. Chem. Pharm. Bull.
1988, 36, 1676–1684.
References and notes
1. Harbowy, M. E.; Balentine, D. A. Crit. Rev. Plant Sci. 1997, 16,
415–480.
2. Hashimoto, F.; Nonaka, G.; Nishioka, I. Chem. Pharm. Bull.
1992, 36, 1383–1389.
3. Haslam, E. Practical Polyphenolics, from Structure to Mole-
cular Recognition and Physiological Action; Cambridge
University Press: Cambridge, 1998; pp 335–354.
4. Haslam, E. Phytochemistry 2003, 64, 61–73.
5. (a) Takino, Y.; Imagawa, H.; Horikawa, H.; Tanaka, A. Agric.
Biol. Chem. 1964, 28, 64–71; (b) Takino, Y.; Ferretti, A.;
Flanagan, M.; Gianturco, M.; Vogel, M. Tetrahedron Lett.
1965, 4019–4025; (c) Brown, A. G.; Falshaw, C. P.; Haslam,
E.; Holmes, A.; Ollis, W. D. Tetrahedron Lett. 1966, 1193–
1204.
6. Coxon, D. T.; Holmes, A.; Ollis, W. D. Tetrahedron Lett. 1970,
11, 5241–5246.
7. Lewis, J. R.; Davis, A. L.; Cai, Y.; Davies, A. P.; Wilkins, J. P.
G.; Pennington, M. Phytochemistry 1998, 49, 2511–2519.
8. Collier, P. D.; Bryce, T.; Mallows, R.; Thomas, P. E.; Frost,
D. J.; Korver, O.; Wilkins, C. K. Tetrahedron 1973, 29,
125–142.
20. Tanaka, T.; Watarumi, S.; Matsuo, Y.; Kamei, M.; Kouno, I.
Tetrahedron 2003, 59, 7939–7947.
21. Tanaka, T.; Matsuo, Y.; Kouno, I. J. Agric. Food Chem. 2005,
53, 7571–7578.
22. Tanaka, T.; Mine, C.; Inoue, K.; Matsuda, M.; Kouno, I.
J. Agric. Food Chem. 2002, 50, 2142–2148.
9. Coxon, D. T.; Holmes, A.; Ollis, W. D. Tetrahedron Lett. 1970,
11, 5247–5250.
23. Valcic, S.; Burr, J. A.; Timmermann, B. N.; Liebler, D. C.
Chem. Res. Toxicol. 2000, 13, 801–810.
24. Tanaka, T.; Kondou, K.; Kouno, I. Phytochemistry 2000, 53,
311–316.
10. Bryce, T.; Collier, P. D.; Mallows, R.; Thomas, P. E.; Frost, D.
J.; Wilkins, C. K. Tetrahedron Lett. 1972, 13, 463–466.
11. Wan, X.; Nursten, H. E.; Cai, Y.; Davis, A. L.; Wilkins, J. P. G.;
Davies, A. P. J. Sci. Food Agric. 1997, 74, 401–408.