Synthesis and biological evaluation of a library of resveratrol analogues as inhibitors of COX-1, COX-2 and NF-κB

Article abstract of DOI:10.1016/j.bmc.2008.04.031

Resveratrol (4,3′,5′-trihydroxystilbene) is a naturally occurring antioxidant that inhibits cyclooxygenase-1 (COX-1), cyclooxygenase-2 (COX-2) and the transcription factor NF-κB. A 78-membered library of resveratrol analogues in which the substituents on

Full text of DOI:10.1016/j.bmc.2008.04.031

Available online at www.sciencedirect.com
Bioorganic & Medicinal Chemistry 17 (2009) 1044–1054
Synthesis and biological evaluation of a library of resveratrol
analogues as inhibitors of COX-1, COX-2 and NF-jB
Soo Sung Kang,a Muriel Cuendet,b Denise C. Endringer,b Vicki L. Croy,b
John M. Pezzutob and Mark A. Liptona,*
aDepartment of Chemistry, The Purdue Cancer Center, Purdue University, 560 Oval Drive, West Lafayette,
IN 47907-2084, United States
bDepartment of Medicinal Chemistry and Molecular Pharmacology, The Purdue Cancer Center, Purdue University,
560 Oval Drive, West Lafayette, IN 47907-2084, United States
Received 12 December 2007; revised 10 April 2008; accepted 11 April 2008
Available online 16 May 2008
Abstract—Resveratrol (4,30,50-trihydroxystilbene) is a naturally occurring antioxidant that inhibits cyclooxygenase-1 (COX-1),
cyclooxygenase-2 (COX-2) and the transcription factor NF-jB. A 78-membered library of resveratrol analogues in which the sub-
stituents on the two aryl rings and alkene were varied was synthesized using a solid-phase Wittig olefination reaction. The library
contains inhibitors against all three proteins that were more potent than resveratrol itself. Preliminary structure–activity relation-
ships were also obtained from these data that permitted the derivation of pharmacophore models for each of the three targets.
ꢀ 2008 Elsevier Ltd. All rights reserved.
1. Introduction
are consumed in even larger quantities than these are in
the US, but where red wine is consumed in far larger
quantity.16 Additionally, recent evidence has suggested
that resveratrol can extend the lifespan of many organ-
isms, possibly even humans, and also improve the health
of older individuals, through a mechanism related to that
exhibited by caloric restriction.17
Resveratrol (3,5,40-trihydroxystilbene), which occurs in
nature as both cis- and trans-isomers, is a naturally occur-
ring phytoalexin first isolated from the roots of white hel-
lebore in 1940.1 Since its initial discovery, resveratrol has
also been isolated from roughly 70 different plant species,
most important of which is in the skin and seeds of red
grapes2 but also in blueberries,3 peanuts, 4 and rhubarb.5
Resveratrol is believed to be made by plants to protect
OH
OH
7
OH
them against fungal pathogens,6 oxidative stress, and/
or UV radiation.8 Within the past decade, however, it
has become increasingly clear that resveratrol conveys a
number of health benefits to humans and may very well
be an important chemopreventive agent for several differ-
ent disease states, including several cancers,9–12 cardio-
vascular disease,13 and ischemia.14 Because red wine has
been identified as the most important dietary source of
resveratrol,15 it has been suggested that resveratrol is at
least partly responsible for the so-called ‘French para-
dox,’ which refers to the reduced incidence of cardiovas-
cular disease in regions of France where saturated fats
OH
OH
HO
trans-resveratrol
cis-resveratrol
Resveratrol was found to inhibit COX-1 with an ED50
of 15 lM during a bioassay-guided fractionation of
plant extracts to search for new cancer chemopreventive
agents.18 Subsequent investigations have determined
that resveratrol inhibits cyclooxygenase-2 (COX-2) as
well, both directly with an IC50 of 32 lM and at the
transcriptional level.19 COX-2 is induced during the
inflammatory response and has been implicated in
the etiology of cancer.20 Additionally, resveratrol has
been shown to inhibit the activation of the oncogenic
transcription factor NF-jB by various inflammatory
agents with an observed IC50 of 30 lM.21 It is important
Keywords: Resveratrol analogues; Parallel synthesis; Cyclooxygenase;
Inhibition; NF-jB .
*
Corresponding author. Tel.: +1 765 494 0132; fax: +1 765 494 0239;
e-mail: lipton@purdue.edu
0968-0896/$ - see front matter ꢀ 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmc.2008.04.031

S. S. Kang et al. / Bioorg. Med. Chem. 17 (2009) 1044–1054
1045
to note, however, some COX-2 inhibitors were recently
reported to increase the risk of serious cardiovascular
events.22
with respect to the inhibition of COX-1, COX-2 and
NF-jB. Another goal of these experiments was the
discovery of resveratrol analogues with activity compa-
rable to that of the natural product but which lacked
the hydroxyl groups known to lead to decreased
bioavailability.
Prompted by the promising in vitro data, Jang et al.
showed that topical application of resveratrol reduced
the incidence of skin cancer in a mouse model by
98%.17 Systemic administration of resveratrol inhibited
tumor growth and initiation in various mouse models.23
For instance, daily administration of 20 mg/kg of resve-
ratrol to mice with subcutaneous neuroblastomas re-
duced mortality rates from 100% to 30%.24 Currently,
there are several ongoing Phase I clinical trials of resve-
ratrol as a cancer chemopreventative or chemotherapeu-
tic agent.25
2. Library design
The library (Fig. 1) was designed to probe three different
structural features of resveratrol: substitution on each of
the two aryl rings (Xn and Yn) and substitution on the
alkene (R). To probe the electronic and steric demands
on each of the aryl rings, electron-donating (OH,
OMe, and NMe2) and -withdrawing (F, CF3, and
NO2) substituents were chosen as well as naphthyl sub-
stituents; to probe the required disposition of the two
rings and the steric requirements around the central al-
kene, four small substituents were placed on the alkene.
The alkene substituents were kept very minimal to avoid
producing molecules with estrogenic activity by avoid-
ing structural similarities to known estrogens such as
diethylstilbestrol (Fig. 2).
Although resveratrol shows tremendous promise as a
preventative lead, it is not without some complicating
problems. For instance, because of the many targets
known to interact with resveratrol, it is very difficult
to pinpoint which target is most important for the treat-
ment of a given disease state. The anticancer activity
could result from the inhibition of COX-1, COX-2,
NF-jB, other targets or any combination thereof. Such
an ambiguity makes lead optimization studies difficult,
and the broad-spectrum of activities exhibited by resve-
ratrol can lead to side effects. One possible solution to
these various problems is to develop resveratrol ana-
logues that exhibit selectivity for only one target.
The substituted aryl groups used in the library are
shown in Figure 3. Using the criteria outlined above,
these were chosen for their commercial availability as
benzaldehyde derivatives to accommodate our synthetic
plans (vide infra). Although an exhaustive combination
of the 16 aryl rings and four alkene substituents would
result in a library with 1008 unique members, a library
of 78 members was initially constructed to test the syn-
thetic methodology and to minimally explore the resve-
ratrol structure.
A second problem with resveratrol is its relative lack of
potency. Resveratrol is a micromolar inhibitor of both
COX-1 and COX-2, whereas the diarylpyrazole SC-
560 inhibits COX-1 with an IC50 of 9 nM 26 and the cor-
ticosteroid dexamethasone inhibits COX-2 with an IC50
of 7 nM.27 Resveratrol can be administered in high dos-
age without noticeable problems, but clearly it would be
advantageous to find resveratrol analogues with signifi-
cantly greater potency, thereby reducing the need for
large dosages and making them competitive with state
of the art cyclooxygenase inhibitors.
3. Chemistry
The general approach31 taken to library generation in-
volved the use of a Wittig olefination reaction employ-
ing a resin-bound phosphonium ylide to establish the
central alkene of resveratrol analogues. In the event, a
phosphine-substituted polystyrene resin (1) was used to
form a phosphonium ion with various substituted ben-
Yet another documented problem with resveratrol is its
limited bioavailability owing to its metabolism in the li-
ver. Studies have shown that circulating resveratrol has
a serum half-life of 8–14 min because it is rapidly metab-
olized by sulfation28 and glucuronation.29 Resveratrol 3-
sulfate and resveratrol 3-glucuronide, the two primary
metabolites of resveratrol, have both been shown to ex-
hibit far lower affinity for COX-1 and COX-2,30 thereby
suggesting that the observed in vitro activities of resve-
ratrol may not be realized in vivo. To be useful as a
chemotherapeutic, resveratrol must have greater bio-
availability. A straightforward approach to increasing
bioavailability involves finding analogues with compara-
ble activity that lack the hydroxyl groups of resveratrol
and consequently cannot be sulfated or glucuronated.
R
Yn
Xn
-,
X, Y =
m
p- OH, OMe, NMe2, F, CF3, NO2
R = H, Me, Et, CF3
Figure 1. Generic structure of resveratrol library.
OH
To begin to address these problems with resveratrol, the
synthesis and screening of a small library of analogues
were envisaged to find lead compounds displaying in-
creased selectivity and/or potency and to begin to eluci-
date the structure–activity relationships of resveratrol
HO
Figure 2. Structure of diethylstibestrol.

1046
S. S. Kang et al. / Bioorg. Med. Chem. 17 (2009) 1044–1054
HO
MeO
HO
HO
MeO
MeO
OMe
OH
OMe
OH
OH
OMe
A1
A2
A3
A4
A5
A6
A7
A8
F
Cl
Me2N
F
O2N
F3C
CF3
A9
A15
A16
A10
A11
A12
A13
A14
Figure 3. Substituted aryl groups used in the construction of resveratrol library.
zylic alcohols (2, Scheme 1). Wittig olefination was
accomplished by deprotonation of 3 and reaction with
various substituted benzaldehydes (4) to form the resve-
ratrol analogues 5. It was found that a sufficient sample
purity was achieved through filtration of the crude prod-
uct through a silica gel plug. This procedure was used to
make the majority of the library, but it was found to af-
ford unsatisfactory results in two classes of compounds:
those analogues requiring a 4-hydroxy substituent on
alcohol 2, in which case the phosphonium ion 3 could
not be formed; and those analogues bearing a trifluoro-
methyl substituent on the alkene.
Those library members containing a 4-hydroxyphenyl
unit, including resveratrol itself, were synthesized using
the procedure shown in Scheme 2. A Merrifield chlo-
romethylstyrene resin (6) was used to covalently tether
4-hydroxybenzaldehyde (7), which was then reacted
with a Grignard to form a secondary alcohol (9). Alco-
hol 9 was then converted to a substituted phosphonium
ion reagent (10) for use in a Wittig reaction to couple to
OH
R
R
1. NaH
2.
+
PPh2•HBr
P
CHCl3
X
X
Ph2
65oC, 8h
Y
(4)
OHC
1
3
2
R
E:Z = 1:1 ~ 87:13
Y
~ 21% Yield
Purity • 90%
54 library members made
X
5
Scheme 1. Solid-phase synthesis of resveratrol library (Method A).
CHO
CHO
RMgBr
Cl
NaH, DMF
rt, 20h
+
THF, rt, 1h
HO
O
6
7
8
OH
PPh3
R
Br
.
PPh3 HBr
1. LDA
R
CHCl3, 65oC, 8h
O
2.
OHC
O
X
(11)
10
9
R
R
X
E:Z = 1:1 ~ 85:15
~20% Yield (3-10 mg)
Purity ³ 92%
X
BCl3
7 library members made
HO
O
5
12
Scheme 2. Synthesis of library members containing a 4-hydroxyphenyl unit (Method B).

S. S. Kang et al. / Bioorg. Med. Chem. 17 (2009) 1044–1054
1047
a second aryl aldehyde (11). The completed resveratrol
4. Biological activity
analogue 5 was then cleaved from the resin with BCl3
and passed through a small plug of silica gel to afford
material of >90% purity.
The COX-1 and COX-2 assays were performed by first
screening all 78 library members for COX inhibition at
a concentration of 10 lg/ml. The assays followed estab-
lished protocols,32 the details of which can be found in
the Experimental section. Those library members that
showed at least 50% inhibition at that concentration
were then tested in triplicate in a dose–response to deter-
mine the IC50 value. Resveratrol was also tested because
it is a part of the synthetic library. Inhibition of NF-jB
was determined using a luciferase reporter gene assay.33
An initial screen for activity was performed at 20 lg/ml,
following which the library members that displayed
more than 50% inhibition at that concentration were
then tested in duplicate in a dose–response to determine
the IC50 value.
In those cases from Scheme 1 where the Wittig reaction
failed, presumably a result of steric hindrance of the ylide
intermediate, a primary alcohol corresponding to the less
substituted end of the alkene (13) was attached to resin 1
(Scheme 3). Deprotonation of the less hindered phospho-
nium ion 14 proceeded smoothly, permitting clean reac-
tion with a substituted aryl ketone (15), thereby
affording analogue 5 with concomitant cleavage from
the resin. Purification again was achieved to >90% purity
by filtration through silica gel.
The use of these three synthetic approaches permitted
the construction of a small, 78-membered library in suf-
ficient quantities and purity to assay for the inhibition of
COX-1, COX-2, and NF-jB. Because these approaches
all use substituted benzaldehyde derivatives as their key
starting material, all 78 library members could be made
from 16 different aldehydes. Additionally, the solid-
phase approach chosen for the production of this library
proved to be amenable to automation, so most of the li-
brary was made using an Argonaut Trident parallel syn-
thesis workstation.
In the COX-1 inhibition assay, eight of the 35 resvera-
trol analogues (Fig. 4) for which the IC50 was deter-
mined were found to have inhibitory potency
comparable to or better than that of resveratrol
(IC50 = 0.83 0.44 lM). In the COX-2 assay, 16 ana-
logues were tested in a dose–response, of which 6 proved
to inhibit COX-2 with a comparable or superior potency
to that of resveratrol (IC50 = 0.99 0.40 lM). In the
NF-jB assay, 33 analogues showed sufficient activity
OH
R
X
Br
1
1. NaH
2.
Y
P
Y
Ph2
X
HO
R
13
14
5
(15)
O
E:Z = 1:1
~ 10% Yield ( 3-5 mg)
Purity ³ 90%
6 library members made
Scheme 3. Alternative method for library generation via Wittig reaction (Method C).
OMe
OH
OH
OH
CH3
MeO
OMe
OH
OH
OH
CH3
HO
HO
HO
MeO
1H6
(0.70)
1M2
(1.90)
2M1
(1.90)
2H7
(0.29)
OH
OH
OH
OH
CH3
OH
OH
OH
OH
Me2N
Me2N
O2N
2E13
(0.99)
13M2
(0.97)
2H14
(2.18)
15E2
(0.17)
Figure 4. Structures of library members with greatest inhibition of COX-1. IC50 values are given in parentheses and the most potent library members
are highlighted.

1048
Table 1. COX-1, COX-2, and NF-jB inhibition data for resveratrol library
R2
S. S. Kang et al. / Bioorg. Med. Chem. 17 (2009) 1044–1054
R3
R1
R3
a
Code
R1
R2
IC50 (lM)
COX-1
COX-2
NF-jB
Resveratrol
1H3
1H6
4-HOPh
4-HOPh
4-HOPh
H
3,5-(HO)2Ph
3,4-(HO)2Ph
3,5-(MeO)2Ph
4-MeOPh
0.83 0.44
—
0.99 0.40
—
16.1 8.6
12.7
19.5
—
H
H
0.70
25.8
0.29
16.5
51.3
29.7
2.2
18.4
17.8
—
0.82
19.6
21.3
—
2H5
2H7
3,5-(HO)2Ph
3,5-(HO)2Ph
3,5-(HO)2Ph
3,5-(HO)2Ph
3,5-(HO)2Ph
3,5-(HO)2Ph
3,5-(HO)2Ph
3,5-(HO)2Ph
3,4-(HO)2Ph
3,4-(HO)2Ph
3,4-(HO)2Ph
3,4-(HO)2Ph
3,4-(HO)2Ph
3,4-(HO)2Ph
4-MeOPh
H
H
3,4-(MeO)2Ph
3-FPh
4-FPh
—
—
2H10
2H11
2H13
2H14
2H15
2H16
3H5
H
H
—
—
52.1
12.9
—
H
4-CF3Ph
4-NO2Ph
H
31.6
—
H
2-Naphthyl
4-Me2NPh
4-MeOPh
—
—
H
—
—
H
12.2
10.3
17.9
14.8
6.91
—
3H10
3H11
3H13
3H15
3H16
5H6
H
3-FPh
4-FPh
4-CF3Ph
11.5
—
—
—
H
H
—
—
—
—
H
2-Naphthyl
4-Me2NPh
3,5-(MeO)2Ph
3,4-(MeO)2Ph
3,4-(MeO)2Ph
3-FPh
H
—
—
—
—
H
28.3
52.0
—
5H7
6H7
4-MeOPh
H
—
—
—
—
3,5-(MeO)2Ph
3,5-(MeO)2Ph
3,5-(MeO)2Ph
3,5-(MeO)2Ph
3,5-(MeO)2Ph
3,5-(MeO)2Ph
3,5-(MeO)2Ph
3,4-(MeO)2Ph
3,4-(MeO)2Ph
3,4-(MeO)2Ph
3,4-(MeO)2Ph
3,4-(MeO)2Ph
3,4-(MeO)2Ph
4-HOPh
H
6H10
6H11
6H13
6H14
6H15
6H16
7H8
H
—
—
—
—
—
H
4-FPh
4-CF3Ph
4-NO2Ph
37.5
—
H
—
—
H
10.0
—
5.94
—
—
H
2-Naphthyl
4-Me2NPh
3-MeOPh
18.8
—
H
—
—
—
—
H
17.3
38.8
—
7H10
7H13
7H14
7H15
7H16
1M2
H
3-FPh
4-CF3Ph
4-NO2Ph
3.2
—
—
—
H
H
—
—
—
—
—
—
H
2-Naphthyl
4-Me2NPh
3,5-(HO)2Ph
3,4-(HO)2Ph
3,5-(MeO)2Ph
4-CF3Ph
H
—
—
34.7
—
Me
Me
Me
Me
Me
Me
Me
Me
Me
Me
Me
Me
Me
Me
Me
Me
Me
Me
Me
Me
Me
Me
Me
Me
CF3
CF3
CF3
CF3
1.9
—
1.57
—
1M3
1M6
4-HOPh
4-HOPh
21.3
—
—
—
1M13
1M16
2M1
4-HOPh
4-HOPh
36.3
—
0.47
—
—
—
4-Me2NPh
4-HOPh
3,5-(HO)2Ph
3,5-(HO)2Ph
3,5-(HO)2Ph
3,5-(HO)2Ph
3,5-(HO)2Ph
3,5-(HO)2Ph
3,5-(HO)2Ph
4-MeOPh
1.9
10.6
10.5
—
1.78
12.8
—
26.2
—
2M7
3,4-(MeO)2Ph
3-FPh
4-FPh
2M10
2M11
2M13
2M15
2M16
5M2
—
—
—
—
4-CF3Ph
10.9
—
—
2-Naphthyl
4-Me2NPh
3,5-(HO)2Ph
3,5-(MeO)2Ph
3,4-(MeO)2Ph
2-Naphthyl
3,5-(HO)2Ph
3,5-(MeO)2Ph
3,5-(HO)2Ph
3,4-(HO)2Ph
3,5-(HO)2Ph
3,5-(HO)2Ph
3,4-(HO)2Ph
3,5-(MeO)2Ph
3,5-(HO)2Ph
3,5-(HO)2Ph
3,5-(MeO)2Ph
4-MeOPh
—
42.3
—
10.7
12.4
—
1.74
16.0
—
35.2
27.1
—
5M6
5M7
4-MeOPh
4-MeOPh
4-MeOPh
—
—
—
—
5M15
7M2
7M6
48.2
13.1
—
3,4-(MeO)2Ph
3,4-(MeO)2Ph
4-ClPh
11.4
—
—
—
9M2
9M3
5.3
—
—
—
39.2
37.5
—
4-ClPh
3-FPh
10M2
13M2
13M3
13M6
1F2
4.9
0.97
—
2.54
—
4-CF3Ph
4-CF3Ph
34.2
—
—
—
4-CF3Ph
4-HOPh
32.6
—
—
—
—
—
5F2
5F6
4-MeOPh
4-MeOPh
3,5-(MeO)2Ph
—
—
—
—
—
37.8
—
6F5
7.8

S. S. Kang et al. / Bioorg. Med. Chem. 17 (2009) 1044–1054
1049
Table 1 (continued)
a
Code
R1
R2
R3
IC50 (lM)
COX-1
COX-2
NF-jB
11F6
12F6
12F7
1E2
4-FPh
3-CF3Ph
CF3
CF3
CF3
Et
3,5-(MeO)2Ph
3,5-(MeO)2Ph
3,4-(MeO)2Ph
3,5-(HO)2Ph
3,5-(MeO)2Ph
3,4-(MeO)2Ph
3-FPh
3.0
—
—
—
39.0
31.9
25.5
—
3-CF3Ph
4-HOPh
4-HOPh
—
—
—
—
1E6
2E7
Et
Et
—
—
—
—
—
3,5-(HO)2Ph
3,5-(HO)2Ph
3,5-(HO)2Ph
3,5-(HO)2Ph
3,5-(HO)2Ph
3,5-(HO)2Ph
3,4-(MeO)2Ph
4-ClPh
9.7
5.4
—
2E10
2E11
2E13
2E15
2E16
7E6
Et
Et
—
—
—
—
4-FPh
4-CF3Ph
Et
Et
0.99
—
—
—
—
—
2-Naphthyl
4-Me2NPh
Et
Et
7.4
—
2.2
—
—
3,5-(MeO)2Ph
3,5-(HO)2Ph
3,4-(HO)2Ph
3,5-(HO)2Ph
3,5-(HO)2Ph
3,5-(HO)2Ph
3,4-(HO)2Ph
27.2
38.3
—
9E2
9E3
Et
Et
11.6
—
—
—
4-ClPh
3-FPh
10E2
15E2
16E2
16E3
Et
Et
3.5
0.17
30.5
—
17.7
3.3
14.8
—
25.7
—
2-Naphthyl
4-Me2NPh
4-Me2NPh
Et
Et
—
—
a The most potent IC50 values are highlighted in bold, where no IC50 is shown, the IC50 > 10 lg/ml in COX-1 or -2 assays or >20 lg/ml in the NF-jB
assay.
for a dose–response curve, of which 11 showed activity
comparable to or better than that of resveratrol
(IC50 = 16.1 8.6 lM). All the analogues assayed, and
their associated activities, are shown in Table 1.
reasonable that either the phenol of resveratrol occu-
pies a large, hydrophobic cavity in the COX-1 active
site or that that region of resveratrol is solvent exposed
in the complex and that the increased potency of 2H7
and 15E2 owes somewhat to the hydrophobic effect.
Several trends emerge from an examination of these
data. Firstly, a resorcinol (A2) ring is present in all
the potent inhibitors of COX-1, although in 1H6 it is
methylated. This suggests that the diol of the resorcinol
ring is either involved in specific hydrogen bonding
interactions with residues in the COX-1 active site or
the active site of COX-1 recognizes a highly electron-
rich aryl ring present on the resveratrol skeleton. A sec-
ond trend is that most of the potent inhibitors feature a
substituent on either alkene carbon. This trend, cou-
pled with the observation that the two resveratrol
derivatives 1M2 and 2M1 are equipotent, suggests that
both rings of resveratrol must rotate out of the plane
of the alkene to bind COX-1. The logic behind this
conclusion is that molecular mechanics calculations
using the MM3 forcefield34 as implemented in the Mac-
roModel software package35 show that placing a sub-
stituent on the alkene sterically destabilizes the planar
conformation of stilbenes such as 1M2 by >20 kcal/
mol, forcing the rings out of plane. Since resveratrol it-
self preferentially adopts a planar conformation to
maximize conjugation, it is reasonable to presume that
the COX-1 pharmacophore features both rings rotated
out of plane. Such a hypothesis is also supported by
the SAR of 15E2, the most potent COX-1 inhibitor
in our library: 2H15, which lacks an ethyl substituent
on the alkene, is 100-fold less active in this assay.
The third and last trend that emerges from our
COX-1 data is that the phenol ring of resveratrol toler-
ates a wide variation of substitution, though most of
the active members of our library have another elec-
tron-rich ring in the R1 position. More importantly,
the two most active members (2H7 and 15E2) are those
with the largest R1 substituent. It therefore appears
Four analogues that showed good COX-1 inhibition
(1H6, 1M2, 2M1 and 2E7) also displayed potent inhibi-
tion of COX-2 (Fig. 5). Also, as with COX-1, all but one
inhibitor bore an alkene substituent, again suggesting
that COX-2 may recognize a conformation of resvera-
trol in which both rings are twisted out of the plane of
the alkene. However, unlike COX-1, COX-2 recognizes
a variety of substitution patterns on R3, all of which are
very electron-rich, suggesting that the active site of
COX-2 does not make any specific hydrogen bonding
contacts with the right-hand ring. Instead, it would ap-
pear that the COX-2 active site preferentially recognizes
electron-rich R3 substituents, possibly through a cation–
pi interaction.36 As with COX-1, R1 of the potent COX-
2 inhibitors displays no particular electronic preference
and can accommodate the steric bulk of a naphthalene
ring, as is shown by 2E7. Also of note is that, of the
COX-2 inhibitors found in our assay, only 1M13,
3M16 and 10M2 showed high (>15:1) selectivity for
COX-2 over COX-1.
In the inhibition studies of NF-jB, the standard devia-
tions associated with the luciferase gene reporter assay
had standard deviations of 20–50% associated with
them, so that drawing conclusions from the data was
less certain than in the other cases. Nonetheless, it is
instructive to consider those compounds that exhibited
an IC50 < 20 lM as ‘actives’ (Fig. 6) and the most potent
NF-jB inhibitor (3H15) as significantly more potent
than resveratrol. Viewing the actives collectively, three
general conclusions can be drawn. Firstly, all but one
active (7M2) have unsubstituted alkenes, implying that
a coplanar conformation may be needed for an effective

1050
S. S. Kang et al. / Bioorg. Med. Chem. 17 (2009) 1044–1054
OH OH
OMe
NMe2
CH3
CH3
OMe
OH
OH
CH3
HO
HO
HO
HO
1H6
(0.82)
1M2
(1.57)
2M1
(1.78)
1M13
(0.47)
OH
OH
OH
OH
CH3
F
OH
OH
CH3
F3C
3M16
(1.74)
10M2
(2.54)
15E2
(3.26)
Figure 5. Structures of library members with the greatest inhibition of COX-2. IC50 values are given in parentheses and the most potent library
member is highlighted.
OMe
OH
OH
OH
OH
OH
OMe
OH
HO
HO
Me2N
HO
1H3
(12.7)
1H6
(19.5)
2H13
(12.9)
3H13
(14.8)
OH
OH
OH
OH
OH
OH
OH
OH
F
MeO
F
3H5
(12.2)
3H15
(6.91)
3H10
(10.3)
3H11
(17.9)
OMe
OH
OH
OH
CH3
MeO
MeO
MeO
MeO
OMe
OH
7H8
(17.3)
7M2
(13.1)
6H15
(18.8)
Figure 6. Structures of library members with greatest inhibition of COX-2. IC50 values are given in parentheses and the most potent library member
is highlighted.
inhibition of NF-jB. Secondly, six of the 11 actives
(3H15, 3H10, 3H5, 1H3, 3H13, and 3H11) contain a
catechol subunit (A3), including the four most potent
inhibitors (3H15, 3H10, 3H5, and 1H3), compared to
only two, less potent inhibitors that preserve resvera-
trol’s resorcinol (A2) ring. Although the pseudo-symme-
try of the resveratrol analogues lacking an alkene
substituent complicates drawing firm conclusions, it is
fairly clear that the catechol occupies the R3 (resorcinol)
site. That being the case, it again appears that there is
little selectivity for the R1 ring, beyond a general prefer-
ence for large and electron-rich rings, again suggesting
either a hydrophobic or a solvent-exposed binding site
for R1.
5. Conclusions
The data obtained in these studies provide some initial
SAR for resveratrol’s interactions with COX-1, COX-

S. S. Kang et al. / Bioorg. Med. Chem. 17 (2009) 1044–1054
1051
2, and NF-jB. In each instance, at least one analogue
was identified with activity superior to that of resvera-
trol, and in each case a hypothetical pharmacophore
model was developed. Interestingly, both COX-1 and
COX-2 appear to bind a conformation of resveratrol
in which both aryl rings are rotated out of plane with re-
spect to the alkene. COX-1 would appear to require a
resorcinol ring in the R3 site (where resveratrol places
a resorcinol) but tolerate various aryl rings in the R1 site
(occupied by the phenol of resveratrol), including large
and hydrophobic aryl rings; COX-2, however, permits
greater variation in the R3 site, accepting various elec-
tron-rich aryl rings there and also tolerates large and
hydrophobic aryl rings in the R3 site. NF-jB inhibition
appears to require a planar conformation of resveratrol,
preferring large, hydrophobic R1 rings and a catechol at
the R3 site. These observations suggest possible second-
generation structures to either increase potency or pro-
vide greater selectivity. Studies are ongoing in our group
to explore these possibilities.
EtOAc (2· 3 ml). The combined filtrates were washed
with water (6 ml) and the organic layer was dried with
MgSO4, and passed through a 2 · 0.5 cm silica gel plug.
The solvent was evaporated to afford the desired
product.
Those products protected with TBS or i-Pr ethers were
deprotected; the organic residue was dissolved in
CH2Cl2 (5 ml) and BCl3 (2.0 M solution in CH2Cl2,
0.45 ml, 0.9 mmol) or TBAF (1.0 M solution in THF,
0.6 ml, 0.6 mmol) was added to the solution. The reac-
tion mixture was stirred for 2 h at room temperature,
and washed with water (5 ml). The organic layer was
dried with MgSO4, and passed through a 3 · 0.5 cm sil-
ica gel plug. The solvent was evaporated under reduced
pressure to afford the desired product.
6.1.3. Method B. Merrifield resin (200 mg, 0.18 mmol)
placed in a TridentTM reaction vessel was washed with
anhydrous DMF (3 ml) and treated with 4-hydroxy-
benzaldehyde (0.27 mmol) and NaH (15 mg, 0.36 mmol)
in DMF (2.5 ml). The resin was shaken for 9 h and
sequentially washed with MeOH (2· 3 ml), DMF (2·
6. Experimental
6.1. Chemistry
3
ml), CH2Cl2 (2· 3 ml), and Et2O (3 ml).
The resin was swelled in THF (2.5 ml), cooled to 0 ꢁC,
and the alkyl magnesium bromide (0.2 ml, 1.4 M in
THF, 0.27 mmol) was added to the resin. The resin
was shaken for 2 h and sequentially washed with MeOH
(2· 3 ml), DMF (2· 3 ml), CH2Cl2 (2· 3 ml), and
Et2O (3 ml).
6.1.1. General experimental—synthesis. Merrifield resin
(1% DVB crosslinked, 100–200 mesh, 0.9 mmol/g)
was purchased from Advanced Chemtech. Wang Bro-
mo resin (polymer bound 4-benzyloxybenzyl bromide,
1% DVB crosslinked, 100–200 mesh, 0.92 mmol/g)
was purchased from Sigma–Aldrich. Polystyrene resin
(1% DVB crosslinked, 100–200 mesh) was purchased
from Sigma–Aldrich and modified to triphenylphos-
phineÆHBr resin (1.8 mmol/g) using a literature meth-
od.37 All other reagents were purchased from
Aldrich, Alfa Aesar, or TCI and used without further
purification. The solvents were purified by passage
through an activated column prior to use.38 tert -Butyl-
dimethylsilyl (TBS) or isopropyl (i-Pr) protected hydro-
xyl benzaldehydes were prepared by literature methods
using TBSCl39 or i-PrBr.40 a-Alkyl benzyl alcohols
were prepared from benzaldehyde by literature meth-
ods.41 Solid-phase syntheses were preformed on an
PPh3ÆHBr (74 mg, 0.216 mmol) in CHCl3 (4 ml) was
added to the resin and the resin was heated to 65 ꢁC,
and shaken for 8 h. The resin was cooled to room tem-
perature and sequentially washed with MeOH (2· 3
ml), CH2Cl2 (2· 3 ml), and Et2O (3 ml).
The resin was swelled in THF (2.5 ml), cooled to 0 ꢁC,
and n-BuLi (0.54 ml, 0.5 M solution in THF,
0.27 mmol) was added to the resin, followed by the addi-
tion of a substituted benzaldehyde (0.18 mmol) in THF
(0.2 ml). The resin was shaken for 2 h and sequentially
washed with MeOH (2· 3 ml), DMF (2· 3 ml), CH2Cl2
(2· 3 ml) and Et2O (3 ml).
1
Argonaut TridentTM Synthesizer. H NMR spectra were
recorded at 300 MHz.
The resin was swelled in CH2Cl2 (2.5 ml) and BCl3
(0.45 ml, 2.0 M solution in CH2Cl2, 0.9 mmol) was
added. After shaking for 2 h, the reaction mixture was
removed by filtration and the resin was washed with
ethyl acetate (2· 3 ml), and quenched by the addition
of water (6 ml). The organic layer was separated, dried
with MgSO4, and passed through a 2 · 0.5 cm silica
gel plug. The solvent was evaporated to afford the de-
sired product.
6.1.2. Method A. TriphenylphosphineÆHBr resin
(100 mg, 0.18 mmol) placed in an Argonaut TridentTM
reaction vessel was swelled with CHCl3 and treated with
a solution of an a-alkyl benzyl alcohol (0.26 mmol) in
CHCl3 (3 ml). The resin was shaken for 8 h at 65 ꢁC,
cooled to room temperature, and sequentially washed
with CHCl3 (2· 3 ml), MeOH (2· 3 ml), DMF (2·
3
ml), and Et2O (3 ml).
The resin was swelled in DMF, and benzaldehyde
(0.05 mmol) in DMF (2.5 ml) was added to the resin,
followed by the addition of NaH (10 mg, 0.23 mmol).
The resin was shaken for 10 h at room temperature
and the reaction was quenched by the addition of
MeOH (0.5 ml). The solution was removed from the
reaction vessel and the remaining resin was washed with
6.1.4. Method C. TriphenylphosphineÆHBr resin
(100 mg, 0.18 mmol) in a TridentTM reaction vessel was
swelled with CH3CN and treated with a substituted ben-
zyl alcohol (0.26 mmol) in CH3CN (3 ml). The resin was
shaken for 15 h at 85 ꢁC, cooled to room temperature
and sequentially washed with CHCl3 (2· 3 ml), MeOH
(2· 3 ml), DMF (2· 3 ml), and Et2O (3 ml).

1052
S. S. Kang et al. / Bioorg. Med. Chem. 17 (2009) 1044–1054
1
hyde (24%, isomer ratio: 50/50). H NMR (300 MHz,
CDCl3) d 7.16 (m, 2H), 6.91 (m, 4H), 6.70 (m, 2H),
6.45 (m, 1H).
The resin was swelled in DMF and benzaldehyde or ace-
tophenone (0.05 mmol) in DMF (2.5 ml) was added to
the resin, followed by the addition of NaH (10 mg,
0.23 mmol). The resin was shaken for 40 h at room tem-
perature after which the reaction was quenched by the
addition of MeOH (0.5 ml). The reaction mixture was re-
moved by filtration and the resin was washed with ethyl
acetate (2· 3 ml), and the combined organic layers were
washed with water (6 ml). The organic layer was dried
with MgSO4, and passed through a 3 · 0.5 cm silica gel
plug. The solvent was evaporated to afford the desired
product. If the product was protected with TBS or i-Pr
group, the deprotection procedure in Method A was
performed.
Compound 3H11. Method B was used in combination
with 3-isopropoxybenzyl alcohol, and 4-fluorobenzalde-
1
hyde (24%, isomer ratio: 50/50). H NMR (300 MHz,
CDCl3) d 7.10 (m, 2H), 7.06 (m, 3H), 6.98 (m, 2H),
6.93 (m, 1H), 6.80 (m, 1H).
Compound 3H13. Method B was used in combination
with 3,4-diisopropoxybenzyl alcohol, and 4-(N,N-
dimethylamino)benzaldehyde (5%, isomer ratio: 95 >
1
5). H NMR (300 MHz, CDCl3) d 8.02 (s, 1H), 7.93
(s, 1H), 7.50 (d, 2H, 8.7 Hz), 7.14 (d, 1H, 1.8 Hz), 6.98
(s, 2H), 6.92 (dd, 2H, 8.1, 27.0 Hz), 3.06 (s, 6H).
6.1.5. Synthesis and characterization of selected resvera-
trol analogues. Compound 1H3. Method A was used in
combination with Wang Bromo resin and 3,4-diisoprop-
Compound 3H15. Method B was used in combination
with 3-isopropoxybenzyl alcohol, and 2-naphthylalde-
1
oxybenzaldehyde (7%, isomer ratio: 50/50). H NMR
1
hyde (21%, isomer ratio: 50/50). H NMR (300 MHz,
(300 MHz, CD3OD) d 7.43 (d, 2H, 8.7 Hz), 7.19 (m,
1H), 7.10 (m, 1H), 6.99 (m, 5H), 6.82 (m, 3H), 6.70
(m, 5H), 6.47 (m, 1H).
CDCl3) d 7.73 (m, 6H), 7.67 (dd, 2H, 10.8, 19.2 Hz),
7.42 (m, 6H), 7.12 (m, 1H), 7.10 (s, 2H), 7.02 (m, 1H),
6.87 (m, 1H), 6.78 (s, 1H), 6.74 (s, 2H), 6.60 (dd, 2H,
12.3, 33.5 Hz).
Compound 1H6. Method A was used in combination
with Wang Bromo resin and 3,5-dimethoxybenzalde-
1
hyde (20%, isomer ratio: 68/32). H NMR (300 MHz,
Compound 6H15. Method B was used in combination
with 3,5-dimethoxybenzylalcohol, and 2-naphthylalde-
CDCl3) d 7.42 (d, 2H, 8.7 Hz), 6.96 (dd, 2H, 15.9,
41.1 Hz), 6.84 (d, 2H, 8.7 Hz), 6.66 (d, 2H, 2.4 Hz),
6.44 (d, 1H, 2.4 Hz), 3.95 (s, 6H).
1
hyde (18%, isomer ratio: 59/41). H NMR (300 MHz,
CDCl3) d 7.87–7.40 (m, 14H), 7.23 (dd, 2H, 16.2,
30.3 Hz), *7.22 (dd, 2H, 12.0, 42.6 Hz), 6.74 (d, 2H,
2.1 Hz), 6.46 (1H, d, 2.1 Hz), *6.44 (d, 2H, 2.1 Hz),
Compound 2H4. Method C was used in combination
with 4-nitrobenzylalcohol, and 3,5-bis(tert-butyldimeth-
ylsilyloxy)benzaldehyde (23%, isomer ratio: 67/33). H
*
*6.35 (1H, d, 2.1 Hz), 3.86 (s, 6H), 3.61 (s, 6H) de-
*
1
notes signals arising from a minor isomer.
*
NMR (300 MHz, (CD3)2CO) d 8.30 (d, 2H, 8.7 Hz),
8.20 (d, 2H, 8.7 Hz), *7.91 (d, 2H, 8.7 Hz), 7.60 (d,
2H, 8.7 Hz), *7.38 (dd, 2H, 16.2, 37.8 Hz), 6.76 (dd,
Compound 7H8. Method B was used in combination
with 3-methoxybenzyl alcohol, and 3,4-dimethoxybenz-
aldehyde (21%, isomer ratio: 62/38). 1H NMR
(300 MHz, CDCl3) d 7.32–7.06 (m, 3H), 7.00–6.76 (m,
4H), 6.55 (s, 2H), 3.89 (s, 3H), 3.73 (s, 3H), 3.65 (s, 3H).
*
*
2H, 12.3, 24.6 Hz), 6.72 (d, 2H, 2.1 Hz), 6.45 (t, 1H,
2.1 Hz), 6.34 (d, 1H, 2.1 Hz), 6.31 (d, 2H, 2.1 Hz) de-
notes signals arising from a minor isomer.
*
Compound 2H7. Method B was used in combination
with 3,4-dimethoxybenzyl alcohol and 4-(tert-butyldi-
methylsilyloxy)benzaldehyde (20%, isomer ratio: 67/
33). H NMR (300 MHz, CDCl3) d 6.99–6.69 (m, 3H),
6.53 (d, 1H, 1.5 Hz), 6.42 (dd, 2H, 12.3, 28.5 Hz), 6.35
(s, 2H), 3.79 (s,3H), 3.62 (s, 3H).
Compound 1M2. Method A was used in combination
with Merrifield resin, 4-hydroxybenzaldehyde, methyl
magnesium bromide, and 3,5-diisopropoxylbenzalde-
1
1
hyde (21%, isomer ratio: 63/37). H NMR (300 MHz,
CD3OD) d 7.42 (d, 2H, 6.9 Hz), 6.84 (d, 2H, 6.9 Hz),
6.66 (s, 1H), 6.36 (d, 2H, 2.4 Hz), 6.22 (d, 1H, 2.4 Hz),
2.81 (s, 3H).
Compound 2H13. Method C was used in combination
with 4-(N,N-dimethylamino)benzyl alcohol, and 3,5-
diisopropoxybenzaldehyde (20%, isomer ratio: 63/37).
1H NMR (300 MHz, CDCl3) d 7.48 (d, 2H, 8.4 Hz),
7.23 (d, 2H, 8.4 Hz), 6.95 (dd, 2H, 16.5, 52.1 Hz), 6.67
(s, 2 H), 6.50 (s, 1H), 3.00 (s, 6H).
Compound 1M13. Method A was used in combination
with Merrifield resin, 4-hydroxybenzaldehyde, methyl
magnesium bromide, and 4-N,N-dimethylaminobenzal-
dehyde (10%, isomer ratio: 86/14). 1H NMR
(300 MHz, CDCl3) d 7.42 (d, 2H, 8.7 Hz), 6.83 (d, 2H,
8.7 Hz), 6.77 (s, 2H), 6.74 (s, 1H), 6.70 (s, 1H), 2.98 (s,
6H), 2.27 (s, 3H).
Compound 3H5. Method B was used in combination
with 4-isopropoxybenzyl alcohol, and 3,4-diisopropoxy-
benzaldehyde (23%, isomer ratio: 95 > 5). 1H NMR
(300 MHz, CD3OD) d 7.47 (d, 2H, 8.7 Hz) 7.03 (d,
1H, 1.8 Hz), 6.94 (d, 2H, 8.7 Hz), 6.92 (m, 2H), 6.88
(d, 1H, 2.4 Hz), 6.80 (s, 1H), 3.86 (s, 3H).
Compound 2M1. Method B was used in combination
with 3,5-diisopropoxybenzyl alcohol, and 4-isopropoxy-
benzaldehyde (5%, isomer ratio: 53/47). 1H NMR
(300 MHz, CD3OD) d *7.27(d, 2H, 8.7 Hz), 6.91 (d,
*
*
2H, 8.7 Hz), 6.84 (d, 2H, 8.7 Hz), 6.77 (s, 1H), 6.59
(d, 2H, 8.7 Hz), 6.52 (d, 1H, 2.4 Hz), 6.36 (s, 1H), 6.26
(t, 1H, 2.4 Hz), *6.22 (d, 1H, 2.4 Hz), *6.19 (d, 2H,
Compound 3H10. Method B was used in combination
with 3-isopropoxybenzyl alcohol, and 3-fluorobenzalde-

S. S. Kang et al. / Bioorg. Med. Chem. 17 (2009) 1044–1054
1053
2.4 Hz), *2.24 (s, 3H), 2.16 (s, 3H) denotes signals aris-
*
ing from a minor isomer.
TA. Following transfer to a 96-well plate (Nunc-
Immuno Plate Maxisorp, Fisher Scientific, Pittsburgh,
PA) coated with a goat anti-mouse IgG (Jackson Immu-
no Research Laboratories, West Grove, PA), the tracer
(PGE2-acetylcholinesterase, Cayman Chemical, Ann
Arbor, MI) and primary antibody (mouse anti-PGE2,
Monsanto, St. Louis, MO) were added. Plates were then
incubated at room temperature overnight, reaction mix-
tures were removed, and wells were washed with a solu-
tion of 10 mM potassium phosphate buffer (pH 7.4)
containing 0.01% sodium azide and 0.05% Tween 20.
Ellman’s reagent (200 ll) was added to each well and
the plate was incubated at 37 ꢁC for 3–5 h, until the con-
trol wells yielded an OD = 0.5–1.0 at 412 nm. A stan-
dard curve with PGE2 (Cayman Chemical, Ann
Arbor, MI) was generated on the same plate, which
was used to quantify the PGE2 levels produced in the
presence of test samples. Results were expressed as a
percentage, relative to control (solvent-treated) samples,
and dose–response curves were constructed for the
determination of IC50 values. IC50 values were generated
from the results of four serial dilutions of test com-
pounds and are the mean of two different experiments.
Compound 2M16. Method C was used in combination
with 4-(trifluoromethyl)benzyl alcohol, and 1-[3,5-
bis(tert-butyldimethylsilyloxy)phenyl]ethanone
(10%,
isomer ratio: 55/45). 1H NMR (300 MHz, CDCl3) d
7.62 (d, 2H, 7.5 Hz), 7.44 (d, 2H, 7.5 Hz), *7.37 (d,
*
2H, 8.1 Hz), 7.10 (d, 2H, 8.1 Hz), 6.99 (s, 1H), 6.82
*
*
*
(s, 1H), 6.58 (s, 2H), 6.43 (s, 1H), 6.29 (s, 1H), 6.21
*
(s, 2H), 2.22 (s, 3H), *2.06 (s, 3H) denotes signals aris-
ing from a minor isomer.
Compound 7M2. Method B was used in combination
with 3-methoxybenzyl alcohol, and 3,5-diisopropoxy-
benzaldehyde (6%, isomer ratio: 65/35). 1H NMR
(300 MHz, CDCl3) d 7.00 (m, 2H), 6.83 (m, 1H), 6.60
(s, 1H), 6.41 (s, 2H), 6.35 (s, 1H), 3.90 (s, 3H), 3.88 (s,
3H), 2.03 (s, 3H).
Compound 13M2. Method B was used in combination
with 1-[4-(N,N-dimethylamino)phenyl]ethanol, and 3,5-
bis(tert-butyldimethylsilyloxy)benzaldehyde (10%, iso-
1
mer ratio: 95 > 5). H NMR (300 MHz, CDCl3) d 7.19
(d, 2H, 8.7 Hz), 6.60 (d, 2H, 8.7 Hz), 6.40 (m, 1H),
6.37 (s, 2H), 6.21 (s, 1H), 2.95 (s, 6H), 2.49 (m, 3H).
6.2.2. Determination of TNFa-induced NF-jB activity in
293 cells. Two hundred and ninety-three cells stably
transfected with NF-jB-luciferase plasmid (Panomics,
Fremont, CA) were treated with test compounds and
the determination of luciferase activity was performed
as described previously.32 In brief, transfected cells were
incubated for 24 h in 96-well plates. After 24 h incuba-
tion with TNFa (20 ng/ml) and test compounds, cells
were analyzed for luciferase activity. Cells were washed
with PBS, lysed using 50 ll 1X Reporter Lysis Buffer
(Promega, Madison, WI) for 10 min, and the luciferase
determination was performed according to the manufac-
turer’s protocol. Results were expressed as a percentage,
relative to control (TNFa-treated) samples, and dose–
response curves were constructed for the determination
of IC50 values. IC50 values were generated from the re-
sults of five serial dilutions of test compounds and are
the mean of two different experiments. With the experi-
mental conditions used, no signs of overt cellular toxic-
ity were observed.
Compound 2E13. Method C was used in combination
with 4-(N,N-dimethylamino)benzyl alcohol, and 1-[3,5-
bis(tert-butyldimethylsilyloxy)phenyl]propanone (6%,
isomer ratio: 56/44). 1H NMR (300 MHz, CDCl3) d
6.89 (d, 2H, 8.7 Hz), 6.61 (s, 1H), 6.52 (s. 1H), 6.51 (d,
2H, 8.7 Hz), 6.25 (s, 2H), 2.89 (s, 6H), 2.41 (q, 2H,
7.8 Hz), 1.02 (t, 3H, 7.8 Hz).
Compound 15E2. Method B was used in combination
with 1-(2-naphthalenyl)propanol and 3,5-bis(tert-butyl-
dimethylsilyloxy)benzaldehyde (6%, isomer ratio: 95 >
1
5). H NMR (300 MHz, CDCl3) d 7.86 (m, 4H), 7.61
(d, 1H, 8.4 Hz), 7.48 (m, 2H), 6.69 (s, 1H), 6.44 (s,
2H), 6.30 (s, 1H), 2.86 (q, 2H, 7.5, 7.8 Hz), 1.25 (t,
3H, 7.5 Hz).
6.2. Biological protocols
6.2.1. Evaluation of COX-1 and COX-2 activity by
quantitation of PGE2. The effect of test compounds on
COX activity was determined by measuring PGE2 pro-
duction as described previously.30 Reaction mixtures
were prepared in 100 mM Tris–HCl buffer (pH 8.0) con-
taining 1 lM heme, 500 lM phenol, 300 lM epineph-
rine, sufficient amounts of COX-1 or COX-2 to
generate 150 ng of PGE2/ml, and various concentrations
of test samples. The reaction was initiated by the addi-
tion of arachidonic acid (final concentration, 10 lM)
and incubated for 10 min at room temperature (final
volume, 200 ll). Then, the reaction was terminated by
adding 20 ll of the reaction mixture to 180 ll of
27.8 lM indomethacin, and PGE2 was quantitated by
an ELISA method. The samples were diluted to the de-
sired concentration with 100 mM potassium phosphate
buffer (pH 7.4) containing 2.34% NaCl, 0.1% bovine ser-
um albumin, 0.01% sodium azide and 0.9 mM Na4ED-
References and notes
1. Takaoka, M. J. J. Faculty Sci. Hokkaido Imperial
University 1940, 3, 1.
2. Burns, J.; Yokota, T.; Ashihara, H.; Lean, M. E.; Crozier,
A. J. Agric. Food. Chem. 2002, 50, 3337.
3. Rimando, A. M.; Kalt, W.; Magee, J. B.; Dewey, J.;
Ballington, J. R. J. Agric. Food. Chem. 2004, 52, 4713.
4. Sanders, T. H.; McMichael, R. W., Jr.; Hendrix, K. W.
J. Agric. Food. Chem. 2000, 48, 1243.
5. Aaviksaar, A.; Haga, M.; Kuzina, K.; Pussa, T.; Raal, A.;
¨
Tsoupras, A. Proc. Estonian Acad. Sci. Chem. 2003, 52, 99.
6. Langcake, P.; Pryce, R. J. Physiol. Plant Pathol. 1976, 9,
77.
7. Rosemann, D.; Heller, W.; Sandermann, H. Plant Physiol.
1991, 97, 1280.
8. Creasy, L. L.; Coffee, M. J. Am. Soc. Hortic. Sci. 1988,
113, 230.

1054
S. S. Kang et al. / Bioorg. Med. Chem. 17 (2009) 1044–1054
9. Kamei, H.; Hashimoto, Y.; Koide, T.; Kojima, T.;
27. Masferrer, J. L.; Zweifel, B. S.; Manning, P. T.; Hauser, S.
D.; Leahy, K. M.; Smith, W. G.; Isakson, P. C.; Seibert,
K. Proc. Natl. Acad. Sci. 1994, 91, 3228.
28. Marier, J.-F.; Vachon, P.; Gritsas, A.; Zhang, J.; Moreau,
J.-P.; Ducharme, M. P. J. Pharmacol. Exp. Ther. 2002,
302, 369.
29. Yu, C.; Shin, Y. G.; Chow, A.; Li, Y.; Kosmeder, J.
W.; Lee, Y. S.; Hirschelman, W. H.; Pezzuto, J. M.;
Mehta, R. G.; van Breemen, R. V. Pharm. Res. 2002,
19, 1907.
30. Walle, T.; Hsieh, F.; DeLegge, M. H.; Oatis, J. E., Jr.;
Walle, U. K. Drug Metab. Dispos. 2004, 32, 1377.
31. (a) Zhang, W.; Go, M. L. Bioorg. Med. Chem. Lett. 2007,
42, 841; (b) Lee, H. J.; Seo, J. W.; Lee, B. H.; Chung, K.-H.;
Chi, D. Y. Bioorg. Med. Chem. Lett. 2004, 14, 463; (c)
Mylona, A.; Nikokavouras, J.; Takakis, I. M. J. Org.
Chem. 1988, 53, 3838; (d) Hadfield, J. A.; Gaukroger, K.;
Hirst, N.; Weston, A. P.; Lawrence, N. J.; McGown, A. T.
Eur. J. Med. Chem. 2005, 40, 529.
32. Cuendet, M.; Pezzuto, J. M. Drug Metabol. Drug Interact.
2000, 17, 109.
33. Homhual, S.; Zhang, H. J.; Bunyapraphatsara, N.; Kon-
dratyuk, T. P.; Santarsiero, B. D.; Mesecar, A. D.;
Herunsalee, A.; Chaukul, W.; Pezzuto, J. M.; Fong, H.
H. S. Planta Med. 2006, 72, 255.
34. Allinger, N. L.; Yuh, Y. H.; Lii, J. H. J. Am. Chem. Soc.
1989, 111, 8551.
35. Mohamadi, F.; Richard, N. G. J.; Guida, W. C.;
Liskamp, R.; Lipton, M. A.; Caufield, C.; Chang,
G.; Hendrickson, T.; Still, W. C. J. Comput. Chem.
1990, 11, 440.
Hasegawa, M. Cancer Biother. Radiopharm. 1998, 13, 447.
10. Soleas, G. J.; Grass, L.; Josephy, P. D.; Goldberg, D. M.;
Diamandis, E. P. Clin. Biochem. 2002, 35, 119.
11. Damianaki, A.; Bakogeorgou, E.; Kampa, M.; Notas, G.;
Hatzoglou, A.; Panagiotou, S.; Gemetzi, C.; Kouroumalis,
E.; Martin, P. M.; Castanas, E. J. Cell. Biochem. 2000, 78,
429.
12. Schoonen, W. M.; Salinas, C. A.; Kiemeney, A. L. M. L.;
Stanford, J. L. Int. J. Cancer 2005, 113, 133.
13. Bohm, M.; Rosenkranz, S.; Laufs, U. Nephrol. Dial.
Transplant. 2004, 19, 11.
14. Bertelli, A. A.; Giovannini, L.; Giannessi, D.; Migliori,
M.; Bernini, W.; Fregoni, M.; Bertelli, A. Int. J. Tissue
React. 1995, 17, 1.
15. Careri, M.; Corradini, C.; Elviri, L.; Nicoletti, I.; Zagnoni,
I. J. Agric. Food. Chem. 2004, 52, 6868.
16. Renaud, S.; de Lorgeril, M. Lancet 1992, 339, 1523.
17. Valenzano, D. R.; Terzibasi, E.; Genade, T.; Cattaneo, A.;
Domenici, L.; Cellerino, A. Curr. Biol. 2006, 16, 296.
18. Jang, M.; Cai, L.; Udeani, G. O.; Slowing, K. V.; Thomas,
C. F.; Beecher, C. W. W.; Fong, H. S.; Farnsworth, N. R.;
Kinghorn, D. A.; Mehta, R. G.; Moon, R. C.; Pezzuto, J.
M. Science 1997, 275, 218.
19. Subbaramaiah, K.; Chung, W. J.; Michaluart, P.; Telang,
N.; Tanabe, T.; Inoue, H.; Jang, M.; Pezzuto, J. M.;
Dannenberg, A. J. J. Biol. Chem. 1998, 273, 21875.
20. Tsujii, M.; Kawano, S.; DuBois, R. N. Proc. Natl. Acad.
Sci. 1997, 94, 3336.
21. Holmes-McNary, M.; Baldwin, A. S., Jr. Cancer Res.
2000, 60, 3477.
22. Mukherjee, D.; Nissen, S. E.; Topol, E. J. JAMA 2001,
286, 954.
36. Gallivan, J. P.; Dougherty, D. A. Proc. Natl. Acad. Sci.
1999, 96, 9459.
23. Kimura, Y.; Okuda, H. J. Nutr. 2001, 131, 1844.
24. Chen, Y.; Tseng, S.-H.; Lai, H.-S.; Chen, W.-J. Surgery
2004, 136, 57.
37. Thomas, G. L.; Bohner, C.; Ladlowb, M.; Spring, D. R.
Tetrahedron 2005, 61, 12153.
38. Solvent columns are composed of activated alumina and
supported copper redox catalyst reactant. See: Organome-
tallics 1996, 15, 1518–1520.
39. Grealish, M. P.; Jung, M. K.; Hamel, E.; Pettit, R. K.;
Chapuis, J.; Schmidt, J. M.; Pettit, J. R. J. Med. Chem.
2002, 45, 2534.
40. Pasturel-Jacope, Y.; Maignan, J.; Solladie, G. Tetrahedron
2003, 59, 3315.
25. Phase I trials at NCI: http://www.cancer.gov/clinicaltrials/
CCUM-2004-0535; Phase I trials Universities of Michigan
and Leicester http://www.personal-umich.edu/~monk/
protocols.html; Colon cancer trials at UC Irvine: http://
www.clinicaltrials.gov/ct/show/NCT00256334.
26. Smith, C. J.; Zhang, Y.; Koboldt, C. M.; Muhammad, J.;
Zweifel, B. S.; Shaffer, A.; Talley, J. J.; Masferrer, J. L.;
Seibert, K.; Isakson, P. C. Proc. Natl. Acad. Sci. 1998, 95,
13313.
41. Hartmann, R. W.; Buchborn, H.; Kranzfelder, G.; Scho-
nenberger, H. J. Med. Chem. 1981, 24, 1192.