New dibenzotropolone derivatives characterized from black tea using LC/MS/MS

Article abstract of DOI:10.1016/j.bmc.2004.03.027

Theaflavins and thearubigins are major pigments in black tea, and it is generally accepted that they are produced by oxidation of flavan-3-ols (catechins) during tea fermentation. In the course of studies on the oxidation mechanism of tea polyphenols, especially the formation of thearubigins, a method combining the enzymatic synthesis and LC/ESI-MS/MS analysis was developed to search for new higher molecular weight polymers from black tea. Three new dibenzotropolones, theadibenzotropolone A, B, and C, together with one new tribenzotropolone, theatribenzotropolone A, were formed by the reaction of theaflavins and tea catechins with horseradish peroxidase in the presence of H₂O₂. The structures of these new benzotropolone derivatives were elucidated on the basis of MS and 2D NMR spectroscopic analyses. The existence of these compounds in black tea was characterized by LC/ESI-MS/MS. Theadibenzotropolone A and B were the first benzotropolone-type trimers of catechins found in the black tea extract. The observation that galloyl ester groups of theaflavins can be oxidized to form di- or tri-benzotropolone skeletons strongly implied that this type of oxidation is an important pathway to extend the molecular size of thearubigins.

Full text of DOI:10.1016/j.bmc.2004.03.027

Bioorganic & Medicinal Chemistry 12 (2004) 3009–3017
New dibenzotropolone derivatives characterized from black tea
using LC/MS/MS
Shengmin Sang,a,* Shiying Tian, Ruth E. Stark, Chung S. Yang and Chi-Tang Ho
b
b
a
c
a
Department of Chemical Biology, Ernest Mario School of Pharmacy, Rutgers University, 164 Frelinghuysen Road, Piscataway,
NJ 08854-8020, USA
Department of Chemistry, College of Staten Island and Institute for Macromolecular Assemblies, City University of New York,
b
2800 Victory Boulevard, Staten Island, NY 10314, USA
Department of Food Science, Rutgers University, 65 Dudley Road, New Brunswick, NJ 08901-8520, USA
c
Received 9 December 2003; revised 2 March 2004; accepted 14 March 2004
Available online 20 April 2004
Abstract—Theaflavins and thearubigins are major pigments in black tea, and it is generally accepted that they are produced by
oxidation of flavan-3-ols (catechins) during tea fermentation. In the course of studies on the oxidation mechanism of tea
polyphenols, especially the formation of thearubigins, a method combining the enzymatic synthesis and LC/ESI-MS/MS analysis
was developed to search for new higher molecular weight polymers from black tea. Three new dibenzotropolones, theadibenzo-
tropolone A, B, and C, together with one new tribenzotropolone, theatribenzotropolone A, were formed by the reaction of
2 2
theaflavins and tea catechins with horseradish peroxidase in the presence of H O . The structures of these new benzotropolone
derivatives were elucidated on the basis of MS and 2D NMR spectroscopic analyses. The existence of these compounds in black tea
was characterized by LC/ESI-MS/MS. Theadibenzotropolone A and B were the first benzotropolone-type trimers of catechins found
in the black tea extract. The observation that galloyl ester groups of theaflavins can be oxidized to form di- or tri-benzotropolone
skeletons strongly implied that this type of oxidation is an important pathway to extend the molecular size of thearubigins.
ꢀ
2004 Elsevier Ltd. All rights reserved.
1
. Introduction
orange-red in color, possess a benzotropolone skeleton
that is formed from co-oxidation of appropriate pairs of
catechins, one having a vic-trihydroxy structure, and the
The role of tea in health promotion and disease pre-
vention has been supported by results from large num-
ber of studies in cell culture and tumor bioassays in
4
;5
other having an ortho-dihydroxy group. It is known
that theaflavins contribute importantly to the properties
1
;2
6
7
animal models carried out over more than a decade.
of black tea, such as color, ‘mouthfeel’ and extent of
8
Three types of tea are produced from the leaves of
Camellia sinensis––green tea, oolong tea, and black tea.
Green tea (non-fermented) and oolong tea (semi-fer-
mented) are more popular in China, Japan, Korea, and
some African countries, whereas black tea (fermented),
which accounts for almost 80% of the world tea pro-
duction, is preferred in India and the Western countries.
tea cream formation. Their structures are well stud-
ied.
9–12
On the other hand, thearubigins, which are red-
13
brown or dark brown, are heterogeneous polymers. So
far information about their formation, structures, and
contribution to black tea quality is very limited. A
partial structure of thearubigins from black tea was
elucidated by using chemical degradation, which indi-
cated them to be heterogeneous polymers of flavan-3-ols
and flavan-3-ol gallates having bonds at C-4, C-6, C-8,
Theaflavins and thearubigins are major pigments in
black tea, and it is generally accepted that they are
produced by oxidation of flavan-3-ols (catechins) during
0
0
0
14
C-2 , C-5 , and C-6 in the flavan-3-ol units. In addi-
tion, the possible participation of theaflavins in the
3
3
tea fermentation. Theaflavins, which are orange or
formation of thearubigins has been suggested. It was
shown in a recent study that horseradish peroxidase
could oxidize theaflavins into thearubigins in the pres-
15
ence of H O by a model fermentation.
2
Keywords: Theadibenzotropolone A, B and C theatribenzotropolone
A; Peroxidase; LC/MS/MS; Thearubigin; Theaflavin; Black tea.
2
*
Corresponding author. Tel.: +1-732-445-3400x300; fax: +1-732-445-
687; e-mail: ssang@rci.rutgers.edu
However, it is very difficult to isolate thearubigins from
black tea extract since they are the complex polymers
0
0
968-0896/$ - see front matter ꢀ 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmc.2004.03.027

3010
S. Sang et al. / Bioorg. Med. Chem. 12 (2004) 3009–3017
produced from coupling reactions mainly between five
major tea catechins (())-epigallocatechin gallate
NMR of theadibenzotropolone B (10) and C (11) and
theatribenzotropolone A (13), and the attempt of the
characterization from black tea by LC/ESI-MS/MS of
these new tea catechin trimers and tetramer.
(
EGCG), ())-epigallocatechin (EGC), ())-epicatechin
gallate (ECG), ())-epicatechin (EC), and catechin (C)).
Thus, model fermentation experiments using pure cate-
chins are very useful to understand the formation of
black tea components, especially thearubigins. As an
important analytical tool, LC/MS has been widely used
to identify unknown peaks from biological samples. In
the course of our studies on the structures of thearubi-
gins, a method combining the enzymatic synthesis and
LC/ESI-MS/MS analysis was developed to search for
new higher molecular weight polymers from black tea.
Enzymatic synthesis can provide the structure informa-
tion of some new pigments formed by the oxidation of
tea catechins; whereas LC/MS/MS can establish the
existence of these new pigments in black tea extract.
2. Results and discussion
16
Our preliminary communication has reported that the
galloyl ester group of theaflavin 3-gallate is as reactive as
the B-ring (vic-trihydroxy) of EGCG or EGC (epigal-
locatechin) and the galloyl ester group of ECG (epi-
catechin gallate), and can further react with EC to form
the new theaflavin-type tea catechin trimers, thea-
dibenzotropolone A (9), which was characterized from
subfraction 12 of the ethyl acetate fraction of black tea
extract by LC/ESI-MS/MS. In addition, two additional
peaks corresponding to the same molecular weight as
that of theadibenzotropolone A were found in subfrac-
tion 13. To identify these unknown peaks, two new
dibenzotropolones, theadibenzotropolone B and C,
together with one new tribenzotropolone, theatri-
benzotropolone A, were synthesized by the reaction of
theaflavins and tea catechins with horseradish peroxi-
dase in the presence of H O .
16
Our preliminary communication has reported the
enzymatic synthesis, structural elucidation and charac-
terization of a new type of pigment, theadibenzotropo-
lone A (9), from black tea by LC/ESI-MS/MS. In
addition, two additional peaks corresponding to the
same molecular weight as that of theadibenzotropolone
A were found in black tea extract. To identify these
unknown peaks and find more new peaks, two new
dibenzotropones, theadibenzotropolone
B and C,
2
2
together with one new tribenzotropolone, theatri-
benzotropolone A were synthesized by the reaction of
theaflavins and tea catechins (Fig. 1) with horseradish
peroxidase in the presence of H O . This paper reports
2 2
the enzymatic synthesis and structural elucidation by 2D
Theaflavin 3-gallate (5) could not only react with EC (1)
to form theadibenzotropolone A, but also react with
catechin (4) to form the isomer of theadibenzotropolone
A, theadibenzotropolone B (10). The molecular formula
OH
OH
HO
O
HO
O
OH
OH
OR2
R1
OH
OH
OH
1
2
3
. R
. R
. R
1
1
1
=R =H
2
4
=H, R
2
=galloyl
=OH, R =galloyl
2
OH
OH
O
O
OH
OH
OR2
OH
OH
HO
HO
O
O
OH
O
HO
HO
O
O
OH
OH
OH
O
OH
OR1
OH
OH
OH
5
6
8
. R
. R
. R
1
1
1
=H, R
=galloyl, R
=R =galloyl
2
=galloyl
7
2
=H
2
Figure 1. Structures of compounds 1–8.

S. Sang et al. / Bioorg. Med. Chem. 12 (2004) 3009–3017
3011
of 10 was determined to be C50
ꢀ
H
38
O
21 by negative-ion
5 with the B-ring of catechin. Another three olefinic
0
13
0
0
APCI-MS ([M)H] at m=z 973) as well as from its
NMR data, which indicated that this compound con-
C
protons (d 7.29 br s H-c ; 7.64 br s H-e ; 7.81 s H-g )
showed similar correlation patterns to those of H-c, H-e,
1
13
0
0
0
sisted of three flavan-3-ol units. Its H and C NMR
spectra showed very similar patterns as those of com-
and H-g. H-c had cross peaks with C-a (d 186.2), C-b
0
0
0
(156.2), C-d (d 134.0), C-e (d 126.1), and C-2 (d 79.6);
1
0
0
0
pound 9. The H NMR spectrum exhibited three sets of
H-e showed correlation peaks with C-c (d 117.6), C-d
0
0
0
signals, due to protons, at the 2-, 3-, 4-, 6-, and 8-
positions of the flavan-3-ol nucleus. These observations
also indicate that the A and C rings of 9 did not undergo
any change during oxidation. In comparison with the H
NMR spectrum of 5, compound 10 is distinguished by
(d 134.0), C-f (d 132.0), C-j (d 122.6), and C-2 (d 79.6);
0
0
0
H-g had correlations with C-f (d 132.0), C-h (d 147.2),
C-i (d 152.0), C-k (d 129.1), and C-2 (d 76.9) (Fig. 3).
These through-bond connectivities proved that H-c , H-
e , and H-g belonged to the benzotropolone of the
theaflavin part of 10. Thus, the structure of 10 was
deduced as shown (Fig. 2) and named theadibenzotropo-
lone B. The complete interpretation of the NMR data
was based on the results of HMQC and HMBC experi-
ments (Tables 1 and 2).
0
0
00
1
0
0
0
the absence of galloyl ester signals, a large downfield
0
shift of H-3 , an additional set of A and C ring signals
from flavan-3-ol, and three more olefinic proton signals
13
(
d 7.51 br s H-c; 7.57 s H-g; 8.76 br s H-e). The
C
NMR spectrum of 10 displayed 50carbon signals, 27 of
which were assigned to the A and C rings of flavan-3-ols.
Besides the A and C ring signals, 23 carbon signals were
observed including two carbonyls (d 187.5 s C-a; 186.2 s
Whereas the isomer of theaflavin 3-gallate, neotheafla-
vin 3-gallate (7), could react with EC to form thea-
dibenzotroplone C (11), it could not react with catechin
to generate compound 12. The negative-ion APCI-MS
0
C-a ), one ester carbonyl (d 168.0s C-l) and 20olefinic
carbons (Table 1). All of these spectral features sup-
ported the presence of two benzotropolone groups in
compound 10. Thus, the galloyl ester group on 5 could
react with the B-ring of catechin to form another ben-
zotropolone. This assertion was supported by the
HMBC NMR spectrum. The HMBC spectral analysis
ꢀ
of 11 displayed a molecular ion peak at m=z [M)H]
973, supporting a molecular formula of C50
NMR spectra of 11 displayed signal patterns similar to
38
H O21. The
1
those of 10. The H NMR spectrum of 11 also showed
three sets of signals, due to protons, at the 2-, 3-, 4-, 6-,
and 8- positions of the flavan-3-ol nucleus, and two sets
(
Fig. 3) yielded correlation peaks between H-c (d 7.51)
and C-a (d 187.5), C-b (d 155.5), C-d (d 123.9), C-e (d
35.5), C-l (d 168.0); H-e (d 8.76) and C-c (d 115.2), C-d
characteristic signals for the benzotropolone group (d
0
1
7.61 br s H-c; 7.80s H-g; 8.37 br s H-e; 7.30br s H-c
;
0
0
13
(
d 123.9), C-f (d 135.8), C-l (d 168.0), C-j (d 123.1); H-g
7.49 s H-g ; 7.97 br s H-e ). The C NMR spectrum of
11 also displayed 50carbon signals. Besides the A and C
ring signals, 23 carbon signals were observed including
(
d 7.57) and C-2 (d 79.6), C-h (d 150.2), C-i (d 153.0), C-
k (d 129.3). Thus, the correlation peaks of H-c and H-e
with the ester carbonyl carbon C-l, and H-g with C-2 of
the flavan-3-ol unit all indicated the presence of a ben-
zotropolone group formed by the galloyl ester group on
0
two carbonyls (186.5 s C-a; 185.8 s C-a ), one ester
carbonyl (d 168.0s C-l), and 20olefinic carbons (Table
1). All of these spectral features support the presence of
two benzotropolone groups in compound 11. Therefore,
just as we expected that the galloyl ester group on 7
could react with the B-ring of EC to form another
benzotropolone. This was demonstrated by the HMBC
data. The cross peaks involved H-c (d 7.61) and C-a (d
Table 1. d
C
(150MHz) NMR spectra data of compounds 10 and 11
(
CD
3
OD) (d in ppm)
10
11
10
11
0
0
2
3
4
5
6
7
8
9
79.6 d
69.5 d
29.0t
75.9 d
65.7 d
28.4 t
9
10
156.9 s
99.9 s
156.5 s
100.8 s
186.5 s
155.0s
1
1
86.5), C-b (d 155.0), C-d (d 124.1), C-e (d 132.2), C-l (d
68.0); H-e (d 8.37) and C-c (d 115.4), C-d (d 124.1), C-f
00
a
187.5 s
155.5 s
115.2 d
123.9 s
135.5 d
135.8 s
122.7 d
150.2 s
153.0 s
123.1 s
129.3 s
168.0s
(
d 135.0), C-l (d 168.0), C-j (d 122.8); H-g (d 7.80), and
157.7 s
96.7 d
157.1 s
95.9 d
158.0s
100.7 s
79.6 d
70.2 d
25.4 t
157.0s
96.2 d
157.4 s
95.5 d
157.3 s
99.4 s
79.1 d
70.3 d
24.2 t
b
C-2 (d 75.9), C-h (d 150.0), C-i (d 152.2), C-k (d 126.4).
Thus, the correlation peaks of H-c and H-e with the
ester carbonyl carbon C-l, and H-g with C-2 of the fla-
van-3-ol unit all indicated the presence of a benzotropo-
lone group formed by the galloyl ester group on 7 with
the B-ring of EC. Another three olefinic protons (d 7.30
c
115.4 d
124.1 s
132.2 d
135.0s
d
e
f
1
2
3
4
5
6
7
8
9
1
2
3
4
5
6
7
8
0
g
h
i
123.6 d
150.0 s
152.2 s
122.8 s
126.4 s
168.0s
0
0
0
0
0
0
0
0
0
0
0
br s H-c ; 7.97 br s H-e ; 7.49 s H-g ) showed similar
0
j
correlation patterns to those of H-c, H-e, and H-g. H-c
0
156.5 s
97.1 d
156.4 s
95.7 d
158.9 s
99.1 s
76.9 d
64.4 d
29.2 t
154.9 s
96.4 d
157.8 s
95.2 d
158.0s
99.1 s
78.4 d
70.1 d
28.6 t
k
l
0
0
0
had cross peaks with C-a (d 185.8), C-b (d 155.3), C-d
0
0
0
(
132.9), C-e (d 128.3), and C-2 (d 79.1); H-e showed
correlation peaks with C-c (d 118.8), C-d (d 132.9), C-f
(d 132.6), C-j (d 121.8), and C-2 (d 79.1); H-g had
correlations with C-f (d 132.6), C-h (d 147.5), C-i (d
a
186.2 s
156.2 s
117.6 d
134.0s
185.8 s
155.3 s
118.8 d
132.9 s
128.3 d
132.6 s
122.7 d
147.5 s
152.4 s
121.8 s
130.4 s
0
0
0
0
b
0
0
0
c
0
0
0
0
0
0
e
0
d
0
00
0
0
0
0
0
0
0
0
0
0
0
0
0
0
152.4), C-k (d 130.4), and C-2 (d 78.4) (Fig. 3). These
126.1 d
132.0s
0
0
0
0
f
data indicated that H-c , H-e , and H-g belonged to the
benzotropolone of the theaflavin part of 11.Therefore,
the structure of 11 was deduced as shown (Fig. 2) and
named theadibenzotropolone C. The complete inter-
pretation of the NMR data was based on the results of
HMQC and HMBC experiments (Tables 1 and 2).
0
0
g
124.5 d
147.2 s
152.0s
156.6 s
96.8 d
157.9 s
95.8 d
155.0s
95.6 d
156.2 s
95.0d
h
0
0
i
j
122.6 s
129.1 s
0
k

3012
S. Sang et al. / Bioorg. Med. Chem. 12 (2004) 3009–3017
OH
OH
OH
5
4
OH
g
OH
OH
6
OH
OH
OH
1
0
3
2
h
HO
7
9
O
O
O
OH
O
HO
OH
O
HO
OH
O
8
f
i
k
j
e
a
d
OH
'
b
OH
OH
OH
OH
OH
c
5
4'
O
O
O
l
O
O
O
6
'
1
0' 3'
O
2
'
c'
OH
OH
OH
HO 7' 8'
9'
b'
HO
HO
O
O
HO
HO
O
O
d'
e'
a'
O
O
O
j'
k'
OH
OH
OH
8
''
i'
HO
9'' O
0'' 3''
f'
h'
7
6
''
''
2''
g'
OH
OH
OH
1
OH
5
''
4''
OH
OH
OH
OH
OH
9
10
11
OH
5
OH
1
0
4
OH
h
6
3
g
2
HO
7
9
O
8
i
OH
O
OH
f
j
OH
k
e
a
OH
d
b
OH
'
OH
HO
O
l
c
OH
5
O
1
0'
6
'
O
4
'
3
'
O
2
'
c' OH
b'
a'
7
HO
'
9'
O
8
'
OH
OH
d'
O
e'
O
j'
O
i' OH
k'
8
''
OH
HO 7''
6''
9'' O
HO
HO
f'
h'
O
O
2
''
g'
OH
O
3''
4
''
1
0''
O
5
''
O
OH
OH
c'' OH
l''
e''
b''
d''
OH
a''
O
j''
OH
k''
f''
OH
8
'''
OH
i''
HO 7'''
9''' O
2
'''
g'' h'' OH
3
'''
4
'''
6
'''
1
0'''
1
2
5'''
OH
OH
1
3
Figure 2. Structures of compounds 9–13.
Interestingly, theatribenzotropolone A (13), instead of
the two isomers of theadibenzotropolone that we
assigned to the A and C rings of flavan-3-ols. Besides the
A and C ring signals, 35 carbon signals were observed
0
expected, was obtained by the reaction between theaflavin
0
including three carbonyls (d 186.5 s C-a; 185.5 s C-a ;
00
0
3,3 -digallate (8) and EC. Compound 13 was assigned
the molecular formula C H O based on negative-ion
186.4 s C-a ), two ester carbonyls (d 167.4 s C-l and
0
166.2 s C-l ) and 30olefinic carbons (Table 3). All of
these spectral features supported the presence of three
benzotropolone groups in compound 13. Thus, the two
galloyl ester groups on 8 could react with the B-ring of
EC to form two additional benzotropolones. This
hypothesis was supported by the HMBC NMR spec-
trum. The HMBC spectrum (Fig. 3) indicated correla-
tion peaks between H-c (d 7.46) and C-a (d 186.2), C-b
(d 154.7), C-d (d 124.0), C-e (d 133.2), C-l (d 167.4); H-e
(d 8.32) and C-c (d 115.2), C-d (d 124.0), C-f (d 134.2),
C-l (d 167.4), C-j (d 123.2); H-g (d 7.82), and C-2 (d
7
1
52 30
ꢀ
13
APCI-MS ([M)H] at m=z 1383) as well as from its
NMR data, which indicated that this compound con-
sisted of four flavan-3-ol units. The H NMR spectrum
exhibited four sets of signals, due to protons, at the 2-,
C
1
3-, 4-, 6-, and 8- positions of the flavan-3-ol nucleus and
three sets of characteristic signals for the benzotropo-
lone group (d 7.46 br s H-c; 7.82 s H-g; 8.32 br s H-e;
0 0 0 00
.25 br s H-c ; 7.44 s H-g ; 7.73 br s H-e ; 7.15 br s H-c ;
.71 s H-g ; 7.78 br s H-e ). The C NMR spectrum of
7
7
1
00
00
13
3 displayed 71 carbon signals, 36 of which were

S. Sang et al. / Bioorg. Med. Chem. 12 (2004) 3009–3017
3013
OH
OH
OH
O
HO
OH
O
OH
OH
OH
OH
OH
OH
OH
OH
O
O
O
O
HO
OH
O
HO
OH
O
OH
O
O
HO
HO
O
OH
OH
OH
OH
OH
O
O
O
O
OH
OH
OH
O
O
O
O
HO
HO
HO
HO
O
O
O
O
OH
OH
HO
OH
OH
O
OH
OH
OH
O
OH
OH
OH
OH
OH
OH
OH
1
0
11
1
3
Figure 3. Significant HMBC (H fi C) correlations of compounds 10, 11, and 13.
Table 2. d
H
(600 MHz) NMR spectra data of compounds 10 and 11 (CD
3
OD) (d in ppm, J in Hz)
10
11
10
11
0
0
0
0
0
0
2
3
4
5.14 d 8.4
3.86 m
2.74 dd 16.2 4.8
5.14 br s
4.05 m
2.58 d 16.8
2
3
4
5.48 br s
4.21 m
2.89 dd 16.8, 3.6
5.11 d 8.4
3.80m
2.82 dd 16.2, 6.0
2
.35 dd 16.2, 9.02.52 dd 16.8, 3.6
2.78 d 16.8
2.44 dd 16.2, 9.6
5.83 d 1.8
0
0
0
0
6
8
5.82 d 1.8
5.93 d 1.8
5.18 br s
5.82 m
5.85 d 1.8
5.76 d 1.8
5.22 br s
5.78 m
6
8
c
e
g
5.77 d 1.8
5.92 d 1.8
7.51 br s
8.76 br s
7.57 s
5.25 d 1.8
7.61 br s
8.37 br s
7.80 s
0
2
0
3
0
4
3.08 dd 17.4, 4.2
2.96 dd 17.4, 3.6
3.07 m
0
0
0
g
c
7.29 br s
7.64 br s
7.81 s
7.30br s
7.97 br s
7.49 s
0
6
5.99 br s
5.99 br s
6.01 d 1.8
5.96 d 1.8
e
0
8
0
0
0
0
0
0
7
7
6.6), C-h (d 149.4), C-i (d 151.6), C-k (d 126.1); H-c (d
C-d (d 132.3), C-f (d 130.1), C-j (d 122.5) and C-2 (d
00
00
00
0
0
.15) and C-a (d 186.4), C-b (d 154.6), C-d (d 123.3),
79.4); H-g had correlations with C-f (d 130.1), C-h (d
0
0
00
00
00
0
0
00
C-e (132.1), C-l (d 166.2); H-e (d 7.78) and C-c (d
146.4), C-i (d 151.2), C-k (d 128.9), and C-2 (d 75.6)
(Fig. 3). These through-bond connectivities proved that
00
00
00
0
1
(
15.1), C-d (d 123.3), C-f (d 134.8), C-l (d 166.2), C-j
d 123.2); H-g (d 7.71) and C-2 (d 76.6), C-h (d 149.2),
C-i (d 151.5), C-k (d 127.0). Thus, the correlation
peaks of H-c and H-e with the ester carbonyl carbon C-l
and H-c and H-e with the ester carbonyl carbon C-l ,
and H-g with C-2 of the flavan-3-ol unit and H-g with
00
0
00
0
0
0
H-c , H-e , and H-g belonged to the benzotropolone of
the theaflavin part of 13. Thus, the structure of 13 was
deduced as shown (Fig. 2) and named theatribenzo-
tropolone A. The complete interpretation of the NMR
data was based on the results of HMQC and HMBC
experiments (Table 3).
0
0
00
00
00
00
00
0
C-2 indicated the presence of two benzotropolone
groups formed by the galloyl ester group on 8 with the
B-ring of EC. Another three olefinic protons (d 7.25 br s
H-c ; 7.73 br s H-e ; 7.44 s H-g ) showed similar corre-
1
6
Our previous communication reported characteriza-
tion of theadibenzotropolone A from subfraction 12 of
the ethyl acetate fraction of black tea extraction by using
selected-ion monitoring (SIM) chromatograms. In
addition, subfraction 13 showed three peaks corre-
sponding to the same molecular ion as that of thea-
dibenzotropolone A. In order to identify these peaks, we
0
0
0
lation patterns to those of the above two sets of ben-
zotropolone protons (H-c, H-e, and H-g, and H-c , H-
0
0
0
0
00
0
0
e , and H-g ). H-c had cross peaks with C-a (d 185.4),
C-b (d 155.7), C-d (132.3), C-e (d 124.9), and C-2 (d
9.4); H-e showed correlation peaks with C-c (d 117.4),
0
0
0
0
0
0
7

3014
S. Sang et al. / Bioorg. Med. Chem. 12 (2004) 3009–3017
Table 3. d
compound 13 (CD
H
(600 MHz) and d
C
(150MHz) NMR spectra data of
of our three standards, which indicated this peak rep-
3
OD) (d in ppm, J in Hz)
resented the isomer of these theadibenzotropolone-type
structures. We have tried to synthesize the isomers of
these dibenzotropolone-type structures (9–11) by react-
d
H
d
C
d
H
d
C
0
2
3
4
5.12 br s
4.08 br s
76.6 d
65.7 d
29.1 t
9
156.7 s
100.1 s
186.2 s
154.7 s
0
0
ing theaflavin 3 -gallate (6) with EC and catechin in the
peroxidase/H O system, but were not successful. Since
10
a
b
c
2.65 d 17.4
.39 dd 17.4, 3.6
2
2
2
no dibenzotropolone-type compounds were obtained
0
5
6
7
8
9
1
2
3
4
157.1 s
96.2 d
156.8 s
95.7 d
156.9 s
7.46 br s 115.2 d
124.0s
from the reaction between theaflavin 3,3 -digallate and
5.91 br s
5.94 br s
d
e
EC, and compound 12 was not obtained by reacting
neotheaflavin 3-gallate with catechin, we could not
identify the structure of the third peak in subfraction 13.
The reason that most oxidation reactions worked with
EC, but catechin did not work for some, maybe due to
the redox potential value of catechin (79 mV at pH 13.5)
8.32 br s 133.2 d
134.2 s
f
g
7.82 s
122.4 d
099.9 s
h
149.4 s
0
0
0
5.17 br s
5.98 br s
79.4 d
70.2 d
26.7 t
i
151.6 s
123.2 s
126.1 s
167.4 s
185.4 s
155.7 s
j
29
is higher than that of EC (48 mV at pH 13.5). Effort
was also made to identify theatribenzotropolone A from
black tea using selected-ion monitoring (SIM) chroma-
tograms. However, no peak showed up in all of the
fractions of the black tea extract.
3.16 dd 17.4, 4.2
3.05 d 17.4
k
l
0
0
0
0
0
0
0
0
00
0
0
0
0
5
6
7
8
9
1
2
3
4
156.8 s
97.4 d
156.3 s
95.4 d
157.8 s
99.1 s
75.6 d
69.8 d
26.9 t
a
5.93 d 1.8
6.05 d 1.8
b
c
7.25 br s 117.4 d
132.3 s
d
e
7.73 br s 124.9 d
130.1 s
There are two major enzymes involve in the fermenta-
17
tion process of making black tea. One is polyphenol
oxidase (PPO), which plays a key role in the oxidation of
flavanols to black tea components, such as theaflavins
(TF) and thearubigins (TR). Many studies have been
0
0
f
0
0
0
0
0
0
5.66 br s
5.64 m
g
7.44 s
121.9 d
146.4 s
151.2 s
122.5 s
128.9 s
186.4 s
154.6 s
h
0
0
3.09 18.0, 3.6
2.94 d 18.0j
i
0
0
0
0
0
0
0
0
0
0
0
00
00
00
00
00
00
5
6
7
8
9
1
2
3
4
157.4 s
97.0d
k
carried out on the PPO-catalyzed formation of black tea
18–23
5.89 d 1.2
5.52 br s
a
oxidation products.
The other is peroxidase (POD),
157.2 s
95.3 d
158.3 s
99.3 s
76.6 d
66.3 d
29.6 t
b
which also can catalyze the oxidation of o-diphenols to
their quinones by using a peroxide, such as hydrogen
peroxide, formed during the action of PPO on certain
flavanols. However, the contribution of peroxidase to
the formation of black tea pigment, especially thearub-
igins, in tea fermentation is still not clear. Our results
here clearly showed that some theaflavins can further
react with tea catechins to form di- or tri-benzotropo-
lone-type compounds in peroxidase/H O system.
c
7.15 br s 115.1 d
123.3 s
d
0
0
0
0
0
0
e
7.78 br s 132.1 d
134.8 s
4.96 br s
3.83 br s
2.70d 16.2
f
00
00
0
g
7.71 s
122.5 d
149.2 s
151.5 s
123.2 s
127.0s
h
0
2
.60d 16.2
i
j
0
0
0
0
00
5
6
7
8
157.0s
96.9 d
156.6 s
95.8 d
00
00
5.89 d 1.2
5.52 br s
k
2
2
l
166.2 s
It is generally accepted that theaflavins possess a ben-
zotropolone skeleton that is formed from co-oxidation
of appropriate pairs of catechins, one having a vic-tri-
hydroxyl structure, and the other having an ortho-
dihydroxy group. The formation of theaflavin-type
compounds involves the oxidation of the B rings to the
quinones followed by a Michael addition of the gallo-
catechin quinone to the catechin quinone, prior to car-
bonyl addition across the ring and subsequent
analyzed the LC/MS/MS spectra of these three peaks
and our three dibenzotropolone standards. Figures 4
and 5 showed the respective LC/MS/MS chromato-
grams and fragment ion mass spectra of compounds 9–
9
;24
1
1 and the peaks found in subfraction 13. As shown in
decarboxylation
di-gallate, such as theaflavin 3-gallate, neotheaflavin
(Fig. 6). Since theaflavin mono- or
these two figures, compound 9 and the peak at retention
time 26.86 in subfraction 13 exhibited not only the same
chromatographic retention time and molecular masses,
but also the same fragment ion mass spectra. Whereas
compound 10 showed the same chromatographic
retention time, molecular mass, and fragment ion mass
spectra as those of the peak at retention time 27.29.
These findings proved the presence of theadibenzo-
tropolone A and B in black tea. The amount of thea-
dibenzotropolone A in black tea (220ug/kg dry tea
leaves) is about three fold higher than that of thea-
dibenzotroplone B (70ug/kg). The third peak (retention
time 30.24) in subfraction 13 could not be identified as
our standard theadibenotropolone C since they had
different retention times. However, the MS/MS spec-
trum of this peak showed similar fragment ions as those
0
3-gallate, and theaflavin 3,3 -digallate, have vic-trihy-
droxy structure, and EC or catechin has an ortho-dihy-
droxy group, they can form a benzotropolone skeleton.
The formation of theadibenzotropolone A, B, and C,
and theatribenzotropolone A confirmed this hypothesis.
These compounds provided the first evidence to suggest
that the galloyl ester group of theaflavin was as reactive
as the B-ring (vic-trihydroxy) of EGCG or EGC and the
galloyl ester group of ECG in the same mechanism as
the formation of theaflavins and theaflavate A and
25;26
B.
The observation that the galloyl ester group of
theaflavins could be oxidized to form di- or tribenzo-
tropolone skeletons strongly implied that this type of
oxidation is an important pathway to extend the
molecular size of thearubigins. Theadibenzotropolone A

S. Sang et al. / Bioorg. Med. Chem. 12 (2004) 3009–3017
3015
28.22
10 0
26.85
100
Compound 11
Compound 9
8
0
0
8
6
4
2
0
0
0
0
6
40
20
0
0
5
15
25
35
45
5
15
25
35
45
Time (min)
Time (min)
30.24
27.45
10 0
26 .86
100
Compound 10
Black tea
8
0
0
27 .29
8
6
4
2
0
0
0
0
6
40
20
0
0
5
15
25
35
45
5
15
25
35
45
Time (min)
Time (min)
Figure 4. LC/MS chromatograms of compounds 9–11 and the peaks found in subfraction 13 of the ethyl acetate fractions of black tea extract.
m/z
m/z
m/z
m/z
m/z
m/z
Figure 5. Negative ion LC/ESI-MS/MS spectra of compounds 9–11 and the three peaks found in subfraction 13 of the ethyl acetate fractions of black
tea extract.
and B are the first theaflavin-type trimer of catechins in
black tea.
black tea. It is challenging to isolate and identify these
compounds, especially the minor constituents. Struc-
tural information of these minor constituents can help
us to fully understand the fermentation process of black
tea and to approach the formation of the so-called
polymer thearubigins, which count as 20% by weight of
black tea extract. In other words, this can help us further
Besides the four major tea catechins (EC, EGC, EGCG,
and ECG), their related epimers (C, GC, GCG, and CG)
also exist in tea leaves. Therefore, more stereo isomers
can be formed during fermentation process of making

3016
S. Sang et al. / Bioorg. Med. Chem. 12 (2004) 3009–3017
OH
(v/v) H
dase, epicatechin, catechin, H
2
SO
4
in ethanol solution. Horseradish peroxi-
, and CD OD were
OH
OH
2
O
purchased from Aldrich Chemical Co. (Milwaukee, WI).
2
3
O
O
O
0
Theaflavin 3-gallate, theaflavin 3 -gallate, theaflavin
0
R
3
previously in our lab.
,3 -digallate, neotheaflavin 3-gallate were synthesized
28
Peroxidase/H
2
O
2
R
H
O
OH
OH
3.2. Synthesis of theadibenzotropolone A, B, C, and
theatribenzotropolone A
R
O
R
3
.2.1. Theadibenzotropolone A (9). EC (1) (1.0g) and
O
EGCG (3) (1.0g) were dissolved in a mixture of acetone-
pH 5 buffer (1:10, v/v, 50 mL), which contained 5 mg
horseradish peroxidase. While being stirred, 2 mL of
O
OH
O
O
R
O
3.13% H O was added four times during 45 min. One
2 2
R
O
O
C
O
major reaction product, compound 5, was formed. After
addition of 0.5 mL of H , stirring was continued for
R
2
O
2
R
an additional 30min; the amount of compound 5
decreased, while another reaction product, compound 9
was formed. Compound 5 (120mg) and 9 (18 mg), the
two major reaction products, and the recovered EC
-
O
(
180mg) and EGCG (70mg) were isolated using a
OH
O
OH
O
OH
[
O]
combination of Sephadex LH-20(4 0% acetone/water)
column and RP C-18 (50% MeOH/water) column
chromatography. In addition, following the same pro-
cedure, 300 mg EC (1) can react with 100 mg theaflavin
3-gallate (5) to generate 10mg theadibenzotropolone A.
HO
OH
OH
C
O
CO
2
R
R
R
R
1
13
H and C NMR: see literature 16.
Figure 6. Proposed mechanism for the formation of benzotropolone
skeleton.
3.2.2. Theadibenzotropolone B (10). Following the pro-
cedure for the synthesis of 9, 500 mg catechin (4) and
understand the health benefits of drinking black tea.
Thus, the method that we develop here is important. It
can be applied to study the structures of other polymers
whose formation involve in a similar enzymatic process.
2
3
00 mg theaflavin 3-gallate (5) were used to synthesize
13
1
0mg theadibenzotropolone B. H and C NMR: see
ꢀ
Table 1. Negative APCI-MS m=z 973 [M)H] .
3
.2.3. Theadibenzotropolone C (11). Following the pro-
cedure for the synthesis of 9, 500 mg EC (1) and 200 mg
neotheaflavin 3-gallate (7) were used to synthesize 35 mg
3
. Experimental
1
13
theadibenzotropolone C. H and C NMR: see Table 1.
ꢀ
3
.1. General procedures
Negative APCI-MS m=z 973 [M)H] .
1
13
H (600 MHz), C (150MHz), and all 2D NMR spectra
Unity
were acquired on a Varian
INOVA 600 NMR
3.2.4. Theatribenzotropolone A (13). Following the pro-
cedure for the synthesis of 9, 500 mg EC (1) and 200 mg
spectrometer (Palo Alto, CA) equipped with a z-gradient
1
13
0
inverse-detection triple resonance probe. H– C
HMQC and HMBC experiments were performed as
theaflavin 3,3 -digallate (8) were used to synthesize
1
13
25 mg theatribenzotropolone A. H and C NMR: see
ꢀ
27
described previously. The general mass spectrometry
was performed on a VG Platform II single quadrupole
mass spectrometer (Micromass Co., MA) equipped with
an atmospheric pressure ion source and atmospheric
pressure chemical ionization (ApCI) interface. The
effluent from the LC column was delivered to the ion
source (150 °C) through a heated nebulizer probe
Table 2. Negative APCI-MS m=z 1383 [M)H] .
3.3. Preparation of black tea fractions
Yunnan black tea (908 g), which was purchased from the
local Chinese supermarket, was extracted with 80%
acetone (five times) at room temperature for two weeks.
The extract was concentrated to dryness under reduced
pressure, and the residue was dissolved in water and
partitioned with chloroform, ethyl acetate, and n-buta-
nol, respectively. The ethyl acetate fraction was sub-
jected to Sephadex LH-20eluted by acetone/water
solvent system (30–60%) to give 14 subfractions.
(
450 °C) using nitrogen as drying gas (300 L/h) and
sheath gas (150L/h). Negative ions were acquired in full
scan. RP C-18 silica gel and Sephadex LH-20gel were
purchased from Sigma Chemical Co. (St. Louis, MO).
Thin-layer chromatography was performed on Sigma-
Aldrich TLC plates (250 lm thickness, 2–25 lm particle
size), with compounds visualized by spraying with 5%

S. Sang et al. / Bioorg. Med. Chem. 12 (2004) 3009–3017
3017
3
.4. HPLC/ESI-MS Detection of theadibenzotropolone
3. Gross, G. G.; Hemingway, R. W.; Yoshida, T. Plant
Polyphenols 2: Chemistry, Biology, Pharmacology, Ecol-
ogy; Kluwer Academic/Plenum: New York, 1999; pp 697–
A, B, C, and theatribenzotropolone A in black tea
fractions
7
24.
4
5
. Geissman, T. A. Chemistry of Flavonoid compounds;
Pergamon: Oxford, UK, 1962; pp 468–512.
. Runeckles, V. C.; Tso, T. C. In Recent Advances in
Phytochemistry; Academic: New York, 1972; Vol. 5, pp
LC/MS analysis was carried out with a Finnigan Spectra
System, which consisted of a Finnigan model P4000
pump, a model AS3000 refrigerated autosampler, and a
Finnigan LCQ Deca mass detector (ThermoFinigan,
San Jose, CA) incorporated with electrospray ionization
2
47–316.
6. Roberts, E. A. H. Two A Bud 1962, 9, 3–8.
7. Millin, D. J.; Crispin, D. J.; Swaine, D. J. Agric. Food
Chem. 1969, 17, 717–722.
(
ESI) interface. A Supelco Discovery HS C18 column
(
75 · 2.1 mm i.d.; particle size, 3 lm) was used for sep-
8
. Powell, C.; Clifford, M. N.; Opie, S.; Robertson, A.;
Gibson, C. J. Sci. Food Agric. 1992, 63, 77–80.
. Takino, Y.; Imagawa, H.; Horikawa, H.; Tanaka, A.
Agric. Biol. Chem. 1964, 28, 64–71.
aration with a flow rate of 0.2 mL/min. The column
elution started with 100% solvent A (10% aqueous
methanol). The linear gradient was changed to 20% B
9
(
70% aqueous methanol) at 8 min, and 100% B at 34 min
1
0. Takino, Y.; Imagawa, H. Agric. Biol. Chem. 1964, 28,
25–130.
1. Nakagawa, M.; Torii, H. Agric. Biol. Chem. 1965, 29, 278–
84.
to clean the column and then reequilibrate to 100% A
from 37 to 52 min for the next run. The LC elute was
introduced into the ESI interface. The negative ion
polarity mode was set for ESI ion source with the
voltage on the ESI interface maintained at approxi-
mately )4 kV. Nitrogen gas was used as the sheath gas
at a flow rate of 80arb and the auxiliary gas at 10arb,
respectively. The heated capillary temperature and
voltage were maintained at 260 °C and )24 V, respec-
tively. The tube lens offset voltage was set at )55 V. The
structural information of the standard theadibenzo-
troplones and theatribenzotropole were obtained by
tandem mass spectrometry (MS/MS) through collision-
induced dissociation (CID) with a relative collision
energy setting of 25%.
1
1
2
12. Davis, A. L.; Cai, Y.; Davies, A. P. Magn. Reson. Chem.
1995, 33, 549–552.
13. Roberts, E. A. H. J. Sci. Food Agric. 1958, 9, 212–216.
1
1
1
4. Ozawa, T.; Kataoka, M.; Morikawa, K.; Negishi, C.
Biosci. Biotechnol. Biochem. 1996, 60, 2023–2027.
5. Subramanian, N.; Venkatesh, P.; Ganguli, S.; Sinkar,
V. P. J. Agric. Food Chem. 1999, 47, 2571–2578.
6. Sang, S. M.; Tian, S. Y.; Meng, X. F.; Stark, R. E.; Rosen,
R. T.; Yang, C. S.; Ho, C. T. Tetrahedron Lett. 2002, 43,
7
129–7133.
1
7. Dix, M. A.; Fairley, C. J.; Millin, D. J.; Swaine, D. J. Sci.
Food Agric. 1981, 32, 920–932.
18. Robertson, A. Phytochemistry 1983, 22, 883–887.
1
2
2
9. Robertson, A. Phytochemistry 1983, 22, 889–896.
0. Robertson, A. Phytochemistry 1983, 22, 897–903.
1. Opie, S. C.; Robertson, A.; Clifford, M. N. J. Sci. Food
Agric. 1990, 50, 547–561.
Acknowledgements
2
2
2. Opie, S. C.; Clifford, M. N.; Robertson, A. J. Sci. Food
Agric. 1993, 63, 435–438.
3. Opie, S. C.; Clifford, M. N.; Robertson, A. J. Sci. Food
Agric. 1995, 67, 501–505.
24. Tanaka, T.; Mine, C.; Inoue, K.; Matsuda, M.; Kouno, I.
This work was supported by NIH Grant PO1 CA88961.
Funds to purchase the 600 MHz NMR spectrometer at
CUNY College of Staten Island were provided by the
New York State Dormitory Authority and the New
York State Higher Education Applied Technology
Program.
J. Agric. Food Chem. 2002, 50, 2142–2148.
5. Wan, X. C.; Nursten, H. E.; Cai, Y.; Davis, A. L.; Wilkins,
J. P. G.; Davies, A. P. J. Sci. Food Agric. 1997, 74, 401–
2
4
6. Lewis, J. R.; Davis, A. L.; Cai, Y.; Davies, A. P.; Wilkins,
08.
2
J. P. G.; Pennington, M. Phytochemistry 1998, 49, 2511–
519.
2
References and notes
27. Fang, X.; Qiu, F.; Yan, B.; Wang, H.; Mort, A. J.; Stark,
R. E. Phytochemistry 2001, 57, 1035–1042.
1. Yang, C. S.; Maliakal, P.; Meng, X. F. Annu. Rev.
Pharmacol. Toxicol. 2002, 42, 25–54.
2. Higdon, J. V.; Frei, B. Crit. Rev. Food Sci. Nutr. 2003, 43,
28. Sang, S. M.; Lambert, J.; Tian, S. Y.; Hong, J.; Hou, Z.;
Stark, R. E.; Rosen, R. T.; Yang, C. S.; Ho, C. T. Bioorg.
Med. Chem. 2004, 12, 459–467.
8
9–143.
29. Steenken, S.; Neta, P. J. Phys. Chem. 1982, 86, 3661–3667.