Resveratrol analogues as selective cyclooxygenase-2 inhibitors: Synthesis and structure-activity relationship

Article abstract of DOI:10.1016/j.bmc.2004.08.008

A series of hydroxylated and methoxylated trans-stilbenes were synthesized and evaluated for their ability to inhibit COX-1 and COX-2. Some of the hydroxylated derivatives are highly selective COX-2 inhibitor with potency comparable or better than clinically established drugs. Resveratrol (3,5,4′-trihydroxy-trans-stilbene) is found in grapes and various medical plants. Among cytotoxic, antifungal, antibacterial cardioprotective activity resveratrol also demonstrates non-selective cyclooxygenase-1 (COX-1) and cyclooxygenase-2 (COX-2) inhibition. In order to find more selective COX-2 inhibitors a series of methoxylated and hydroxylated resveratrol derivatives were synthesized and evaluated for their ability to inhibit both enzymes using in vitro inhibition assays for COX-1 and COX-2 by measuring PGE₂ production. Hydroxylated but not methoxylated resveratrol derivatives showed a high rate of inhibition. The most potent resveratrol compounds were 3,3′,4′,5-tetra-trans-hydroxystilbene (COX-1: IC₅₀ = 4.713, COX-2: IC₅₀ = 0.0113 μM, selectivity index = 417.08) and 3,3′,4,4′,5,5′-hexa-hydroxy-trans-stilbene (COX-1: IC ₅₀ = 0.748, COX-2: IC₅₀ = 0.00104 μM, selectivity index = 719.23). Their selectivity index was in part higher than celecoxib, a selective COX-2 inhibitor already established on the market (COX-1: IC ₅₀ = 19.026, COX-2: IC₅₀ = 0.03482 μM, selectivity index = 546.41). Effect of structural parameters on COX-2 inhibition was evaluated by quantitative structure-activity relationship (QSAR) analysis and a high correlation was found with the topological surface area TPSA (r = 0.93). Docking studies on both COX-1 and COX-2 protein structures also revealed that hydroxylated but not methoxylated resveratrol analogues are able to bind to the previously identified binding sites of the enzymes. Hydroxylated resveratrol analogues therefore represent a novel class of highly selective COX-2 inhibitors and promising candidates for in vivo studies.

Full text of DOI:10.1016/j.bmc.2004.08.008

Bioorganic & Medicinal Chemistry 12 (2004) 5571–5578
Resveratrol analogues as selective cyclooxygenase-2
inhibitors: synthesis and structure–activity relationship
Marek Murias,a Norbert Handler,a Thomas Erker,a Karin Pleban,a Gerhard Ecker,a
Philipp Saiko,b Thomas Szekeresb and Walter Jagera,*
aInstitute of Pharmaceutical Chemistry, University of Vienna, Althanstrasse 14, A-1090 Vienna, Austria
bClinical Institute for Medical and Chemical Laboratory Diagnostics, Medical University of Vienna, Wa¨hringer, Gu¨rtel 18–20,
A-1090 Vienna, Austria
Received 21 May 2004; accepted 6 August 2004
Available online 11 September 2004
Abstract—Resveratrol (3,5,40-trihydroxy-trans-stilbene) is found in grapes and various medical plants. Among cytotoxic, antifungal,
antibacterial cardioprotective activity resveratrol also demonstrates non-selective cyclooxygenase-1 (COX-1) and cyclooxygenase-2
(COX-2) inhibition. In order to find more selective COX-2 inhibitors a series of methoxylated and hydroxylated resveratrol deriv-
atives were synthesized and evaluated for their ability to inhibit both enzymes using in vitro inhibition assays for COX-1 and COX-2
by measuring PGE2 production. Hydroxylated but not methoxylated resveratrol derivatives showed a high rate of inhibition. The
most potent resveratrol compounds were 3,30,40,5-tetra-trans-hydroxystilbene (COX-1: IC50 = 4.713, COX-2: IC50 = 0.0113lM,
selectivity index = 417.08) and 3,30,4,40,5,50-hexa-hydroxy-trans-stilbene (COX-1: IC50 = 0.748, COX-2: IC50 = 0.00104lM, selectiv-
ity index = 719.23). Their selectivity index was in part higher than celecoxib, a selective COX-2 inhibitor already established on the
market (COX-1: IC50 = 19.026, COX-2: IC50 = 0.03482lM, selectivity index = 546.41). Effect of structural parameters on COX-2
inhibition was evaluated by quantitative structure–activity relationship (QSAR) analysis and a high correlation was found with
the topological surface area TPSA (r = 0.93). Docking studies on both COX-1 and COX-2 protein structures also revealed that
hydroxylated but not methoxylated resveratrol analogues are able to bind to the previously identified binding sites of the enzymes.
Hydroxylated resveratrol analogues therefore represent a novel class of highly selective COX-2 inhibitors and promising candidates
for in vivo studies.
ꢀ 2004 Elsevier Ltd. All rights reserved.
1. Introduction
COX-2 is the principal isoform that participates in
inflammation, and induction of COX-2 is responsible
for the production of PGs at the site of inflammation.
Consequently, selective inhibition of COX-2 should
have therapeutic actions similar to those of non-steroi-
dal anti-inflammatory drugs (NSAIDs), but without
gastrointestinal side effects, which are being caused as
a consequence of COX-1 inhibition. Several selective
COX-2 inhibitors are currently used in the clinics (e.g.
celecoxib, valdecoxib, parecoxib, rofecoxib, etoricoxib,
and lumiracoxib) which provide effective treatment of
inflammatory disease states such as rheumatoid arthritis
and osteoarthritis.3 Several lines of evidence suggest that
selective COX-2 inhibitors may also provide an oppor-
tunity for both cancer prevention and therapy.4 Further-
more, promising in vitro data also indicate that
treatment with selective COX-2 inhibitors may also re-
duce the risk of AlzheimerÕs5,6 and ParkinsonÕs disease7,8
and may also be effective in the treatment of asthma.9
The enzyme cyclooxygenase (COX) catalyzes the first
two steps in the biosynthesis of prostaglandins (PGs)
from the substrate arachidonic acid. At least two forms
of this enzyme exist.1 One of these forms, COX-1, is
constitutively expressed and is responsible for maintain-
ing normal physiologic function and the PGs produced
by this enzyme play a protective role. The other known
form of the enzyme, cyclooxygenase-2 (COX-2), is an
inducible form and its expression is affected by various
stimuli such as mitogens, oncogenes, tumor promoters,
and growth factors.2
Keywords: Resveratrol; Nonsteroidal anti-inflammatory drugs; Cyclo-
oxygenase-1; Cyclooxygenase-2; QSAR.
*
Corresponding author. Tel.: +43 1 4277 55176; fax: +43 1 4277
9551; e-mail: walter.jaeger@univie.ac.at
0968-0896/$ - see front matter ꢀ 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmc.2004.08.008

5572
M. Murias et al. / Bioorg. Med. Chem. 12 (2004) 5571–5578
Resveratrol (3,40,5-trihydroxy-trans-stilbene) is a phyto-
alexin found mainly in the skin of grapes and red wine
and demonstrates anti-inflammatory, cardiovascular
protective, and cancer chemopreventive properties.10 It
has also been shown to be a non-selective inhibitor of
COX-1 and COX-2.11 In vitro and in vivo, resveratrol
is extensively metabolized to several glucuronides and
sulfates.12,13 Moreover, resveratrol undergoes cyto-
chrome P450 catalyzed hydroxylation to piceantannol,
3,4,40,5-tetrahydroxy-trans-stilbene, and to two other
unidentified mono- and dihydroxyresveratrol ana-
logues.14 As piceatannol15 demonstrates a several fold
higher antileukaemic activity than resveratrol and as
COX-2 is a known target for anticancer activity we
investigated whether piceatannol and other hydroxyl-
ated resveratrol analogues might achieve a better and
more selective COX-2 inhibition than resveratrol. Five
hydroxy and six methoxy analogues of resveratrol were
therefore synthesized using standard chemical methods.
Each analogue was then tested for COX-1 and COX-2
inhibition in an in vitro model and the resulting inhibi-
tion values compared with that of resveratrol and the
clinically established selective COX-2 inhibitor celec-
oxib. In addition, a quantitative structure–activity rela-
tionship (QSAR) study was being conducted to
evaluate the effects of various structural parameters of
the molecules on COX-1 and COX-2 inhibition.
O
4
5
R
R
H
R1
O
P
R1
R2
X
R6
OEt
OEt
(a)
R2
R3
R3
R1, R2, R3, R4, R5, R6 = OCH3 or H
X = Cl or Br
R4
R5
(b)
R1
R6
R2
R3
R1, R2, R3, R4, R5, R6 = OCH3 or H
R4
R5
R6
(c)
R1
R2
R3
R1, R2, R3, R4, R5, R6 = OH or H
Scheme 1. General synthetic route for synthesis of the resveratrol
analogues 1–6 and 8–12. Reagents and conditions: (a) P(OEt)3, 130ꢁC;
(b) NaOMe, DMF, 100ꢁC; (c) BBr3, CH2Cl2, rt.
2. Results and discussion
2.1. Chemistry
were weak inhibitors of COX-1 with 2–55 fold lower
IC50-values than resveratrol (7, IC50: 0.53lM). On the
other hand, hydroxylated resveratrol analogues,
especially 9 and 11 were more potent inhibitors with
IC50-values of 0.00998 and 0.01027lM, respectively.
Celecoxib showed only a moderate inhibition (IC50:
19.026lM).
Methoxylated (1–6) and hydroxylated (8–12) resveratrol
analogues were synthesized using standard chemical
methodologies (resveratrol (7) was obtained from a
commercial supplier and was of analytical grade). The
corresponding benzylphosphonates were obtained from
benzylhalides via Michaelis–Arbuzov rearrangement
with triethyl phosphite at 130ꢁC. The following Hor-
ner–Wadsworth–Emmons reactions were carried out at
100ꢁC with the phosphonate, the corresponding meth-
oxybenzaldehyde and sodiummethoxide in DMF to
yield the desired methoxystilbenes. Finally, demethyl-
ation of the phenolic ethers was successfully established
with boron tribromide at room temperature (Scheme 1).
Those procedures afforded all compounds in a good
yield (Table 1). Melting points of compounds 1–6 and
8–12 were consistent with those found in the litera-
ture.16–23 However, physicochemical characterization
of compounds 1–6 and 8–12 in the literature was
incomplete. Missing data of mass spectroscopy, 1H
NMR or 13C NMR were now included in the chemistry
part.
The present study revealed that all hydroxylated ana-
logues were also more potent against COX-2 than resve-
ratrol (7), (Fig. 1). Compounds 9–12 showed even lower
IC50-values (IC50s: of 0.00171, 0.0113, 0.00138, 0.001
and 0.035lM, respectively) which is up to 35-fold lower
than for the selective COX-2 inhibitor celecoxib (Table
2).
Based on the IC50 values for COX-1 and COX-2, the rel-
ative ratios of IC50 COX-1/IC50 COX-2 were calculated
and the data are presented in Table 2. Resveratrol (7)
showed a weak COX-2 inhibition with a selectivity index
(COX-1/COX-2) of 0.5. The COX-2/COX-1 ratios ob-
tained for compounds 10 and 12 (417 and 719), how-
ever, are comparable or significantly better than that
obtained for the clinically used selective COX-2 inhibi-
tor celecoxib (selectivity index: 546) (Table 2). Com-
pounds 2–6, 8, 9, and 11 were also more selective
against COX-2 than to COX-1 with COX-1/COX-2
selectivity index values up to 45. The selectivity index
of 0.54 for resveratrol (7) is comparable with data found
by others.11 Piceatannol (10), a natural compound
found in grapes and red wine, showed the second best
selectivity index from all compound tested. This data,
however, strongly differs from investigations performed
2.2. Inhibition of COX-1 and COX-2
Inhibition of COX-1 and COX-2 by resveratrol ana-
logues was analyzed in a cell-free immunoassay system.
Purified ovine enzyme served as the source of COX-1,
while the human recombinant enzyme formed the source
of COX-2. The inhibition of COX-1 by resveratrol ana-
logues is shown in Table 2. The methoxy derivatives 1–6

M. Murias et al. / Bioorg. Med. Chem. 12 (2004) 5571–5578
Table 1. Structures of the resveratrol analogues 1–12, DL1, and DL2
5573
R4
R5
R6
R1
R2
R3
Compd.
Pos. 3 (= R1)
Pos. 4 (= R2)
Pos. 5 (= R3)
Pos. 30 (= R4)
Pos. 40 (= R5)
Pos. 50 (= R6)
Yield (%)
1
–OCH3
–OCH3
–OCH3
–OCH3
–OCH3
–OCH3
–OH
–H
–OCH3
–OCH3
–OCH3
–OCH3
–OCH3
–OCH3
–OH
–H
–H
–OCH3
–OCH3
–H
–H
–H
59
55
52
55
59
49
Ns
55
60
51
57
45
Ns
Ns
2
–OCH3
–H
–H
3
–OCH3
–OCH3
–OCH3
–OCH3
–H
–OCH3
–H
4
–OCH3
–H
5
–OCH3
–OCH3
–H
–OCH3
–OCH3
–H
6
–OCH3
–OH
–OH
–H
7
8
–OH
–OH
–OH
–H
–OH
–OH
–H
–H
9
10
–OH
–OH
–OH
–OH
–H
–OH
–H
–OH
–OH
–H
–OH
–OH
–OH
–H
11
–OH
–OH
–H
–OH
–OH
–H
12
–OH
–OH
–OH
–OCH3
–OCH3
–OH
–OH
–OH
DL1a
DL2a
–OCH3
–H
–H
–H
Ns: not synthesized.
a Structures were taken from Ref. 11.
Table 2. Inhibitory effect of 1–12, DL1, DL2, and the reference
compound celecoxib on COX-1 and COX-2 activity
4.0
3.5
3.0
2.5
2.0
1.5
1.0
0.5
0.0
-0.5
-1.0
11
Compound
IC50 [lM]
Selectivity index
COX-1/COX-2
9
12
COX-1
COX-2
10
8
1
1.228
9.099
1.667
7.797
0.74
1.17
2
5
6
3
27.783
2.834
7.247
1.575
0.796
17.64
3.56
4
1
4
7
DL1
DL2
3
5
0.514
0.355
0.996
14.11
31.97
0.54
6
11.348
0.535
2
7
-1.5
8
2.072
0.00998
4.713
0.04537
0.00171
0.0113
0.00138
0.00104
0.03482
2.21a
45.67
5.83
50
75
100
125
25
9
10
TPSA
417.08
7.44
11
12
0.01027
0.748
Figure 1. Plot of TPSA-values of compounds 1–12, DL1 and DL2
versus log(l/EC50)-values for COX-2 activity.
719.23
546.41
2.23a
4.07a
Celecoxib
DL1
DL2
19.026
4.92a
4.84a
1.19a
a Values were taken from Ref. 11.
celecoxib measured by an ELISA assay was 404, which
is close to our result of 546.
2.3. QSAR-study
by Lee et al.24 who described piceatannol as inactive
(IC50 values >100lg/mL) in both the COX-1 and the
COX-2 inhibition assay. This discrepancy might be ex-
plained by the fact that the authors used extracted (from
the seed of a Peruvian plant) and not unambiguously
identified piceatannol. Furthermore, assay conditions
might also strongly effect COX-1 and COX-2 inhibition
as indicated for the COX-2 selectivity index of celecoxib.
While assays based on human whole blood,25,26 oxida-
tion of TMPD27 or thin layer chromatography28 showed
values of 7.3, 26.9, and 67, respectively, the selectivity of
To address the question, whether there exists a quantita-
tive relationships between chemical structure and
biological activity, a small set of physicochemical
descriptors (logP, lipophilicity index; TPSA, topological
surface area; MR, molar refractivity; APOL, atom
polarizability) were calculated on basis of 2D-structures.
Then multiple linear regression analyses were per-
formed. An intercorrelation matrix was calculated and
shown in Table 2. A close correlation between MR
and APOL was found. This strong relationship was
also observed for other data sets analyzed (G. Ecker,

5574
M. Murias et al. / Bioorg. Med. Chem. 12 (2004) 5571–5578
unpublished). However, regarding the biological activ-
ity, only for COX-2 a clear relationship to one of the
descriptors was found (TPSA; r = 0.93). Multiple linear
regression analysis with subsequent reduction of the
dimensionality of the model via stepwise elimination of
statistically insignificant descriptors (P >0.05) con-
firmed this relationship (Eq. 1).
3
2
1
0
12
11
8
9
10
logð1=IC50Þ
¼ 0:044ðÆ0:0056ÞTPSA
À 1:827ðÆ0:392Þ
7
DL1
COX-2
6
5
DL2
2
3
1
4
n ¼ 12; r ¼ 0:93; F ¼ 60:4
ð1Þ
-1
0
1
2
3
As indicated in Eq. 1 and shown in Figure 1, high TPSA
values are favorable for high COX-2 activity. In case of
COX-1, either APOL or MR are present in the final
equation. The MR-based model leads to slightly higher
correlation coefficients (0.650 vs 0.649; Eq. 2). In both
cases, higher values of the descriptor leads to lower
pharmacological activity.
log(selectivity) observed
Figure 3. Plot of calculated versus observed log(selectivity) as obtained
by Eq. 3.
Thus, for the present data set lipophilicity does not show
a significant contribution to either activity or selectivity
of the compounds.
logð1=IC50Þ
¼ À0:586ðÆ0:216ÞMR
þ 4:449ðÆ1:706Þ
COX-1
In order to further validate our models, two resveratrol
analogues (DL1 and DL2), carrying both hydroxy- and
methoxy-groups were taken from the literature.11 These
compounds, previously isolated from the stem wood of
Dracaena Loureiri, also demonstrate COX-1 and
COX-2 inhibition. Predictions of biological activities
are in good agreement with measured values which gives
additional support for the general validity of the model
(Figs. 1–3 and Table 3).
n ¼ 12; r ¼ 0:65; F ¼ 7:34
ð2Þ
Figure 2 shows the corresponding plot of MR versus
pharmacological activity. Obviously, the relationship is
by far less significant than in the case of COX-2. Addi-
tionally, the plot of observed versus calculated log(1/
EC50) values shows a slope of only 0.44 (Fig. not
shown). Interestingly, for the calculation of log(selectiv-
ity), TPSA showed to be the most predictive parameter
(Eq. 3; Fig. 3).
2.4. Docking studies
logðIC50COX-1=IC50COX-2Þ
In order to identify possible binding sites of reservatrol
analogues, the ligands were docked into the predefined
binding sites of COX-1 and COX-2. Several residues
have previously been shown to be involved in binding
of selective COX-2 inhibitors, including His90,
Arg120, Phe518, Tyr385, and Ser530.29 For none of
the methoxyl compounds, satisfactory solutions into
the putative binding sites of either COX-1 or COX-2
were found Table 4.
¼ 0:022ðÆ0:008ÞTPSA À 0:314ðÆ0:571Þ
n ¼ 12; r ¼ 0:64; F ¼ 7:05
ð3Þ
Generally, all significant descriptors, which explain the
variance in the data sets describe preferentially polar
type of receptor interactions. It has to be noted, that
none of the final equations contained logP as descriptor.
Table 3. Physicochemical parameters for compounds 1–12, DL1, and
DL2
2.5
9
11
Compound
MR
APOL
TPSA
logP
1
8.048
8.688
8.681
8.685
9.319
9.955
6.404
6.665
6.658
6.715
6.784
6.906
7.040
7.544
44.328
48.224
48.224
48.224
52.119
56.015
34.246
35.850
35.850
37.183
36.652
37.454
38.141
41.235
27.690
36.920
36.920
36.920
46.150
55.380
40.460
80.920
80.920
80.920
101.150
121.380
49.690
38.690
4.489
3.907
4.556
4.232
3.974
3.393
4.005
3.697
3.500
3.424
3.151
2.802
3.961
4.225
2
3
7
12
0.0
1
4
8
4
5
5
DL2
10
6
DL1
2
6
3
7
8
-2.5
9
10
11
12
DL1
DL2
6
7
8
9
10
11
12
MR
Figure 2. Plot of calculated MR-values of compounds 1–12, DL1 and
DL2 versus log(1/EC50)-values for COX-1 activity.

M. Murias et al. / Bioorg. Med. Chem. 12 (2004) 5571–5578
5575
Table 4. Intercorrelation matrix for physicochemical parameters, log(potency) and log(selectivity) values
APOL
MR
TPSA
logP
logCOX-1
logCOX-2
logSel
APOL
MR
TPSA
1
0.999
1
À0.675
0.499
À0.678
0.502
1
logP
À0.920
0.566
0.926
0.643
1
logCOX-1
logCOX-2
logSel
À0.649
À0.642
À0.163
À0.650
À0.645
À0.165
À0.450
À0.793
À0.588
1
0.700
À0.152
1
0.598
1
Figure 4. Ribbon drawing of the structure of COX-2 with docked ligand 11. The helices are colored in red, strands in yellow, loops in blue, and turns
in green. Amino acid residues interacting with the ligand are shown. Graph was generated using the MOE software package (Chemical Computing
Group, Montreal).
In contrast, all for hydroxylated analogues docked into
the binding site of both enzymes. Figure 4 shows a rep-
resentative ribbon drawing of the crystal structure of
COX-2 with compound 11. The binding mode of
hydroxylated resveratrol analogues involves 1–3 hydro-
gen bonds to the amino acid residues Arg120, Ser530,
and Tyr385, which is in accordance to results observed
for other NSAIDs.30 Contrary to the selective COX-2
inhibitor SC-558, the p-bromo analogue of celecoxib,31
compounds 8–12 are not able to have access to the addi-
tional subpocket (substitution of a valine for an isoleu-
cine in the active site) which is reported to be
responsible for COX-2 selectivity. The observed pro-
nounced COX-2 selectivity of compounds 10 and 12
may therefore result either from a different binding
mode to COX-2 and/or inhibition of COX-2 related en-
zymes, for example, peroxidases.32
4. Experimental
4.1. Cyclooxygenase assay
The effect of the test compound on COX-1 and COX-2
were determined by measuring prostaglandin E2
(PGE2) using a COX Inhibitor Screening Kit (Catalog
No 560131) from Cayman Chemicals, Ann Arbor Michi-
gan USA. Reactions mixtures were prepared in 100mM
Tris–HCl buffer, pH8.0 containing 1lM heme and
COX-1 (ovine) or COX-2 (human recombinant) and pre-
incubated for 10min in a waterbath (37ꢁC). The reaction
was initiated by the addition of 10lL arachidonic acid
(final concentration in reaction mixture 100lM). After
2min the reaction was terminated by adding 1M HCl
and PGE2 was quantitated by an ELISA method. The
test compounds were dissolved in DMSO and diluted
to the desired concentration with 100mM potassium
phosphate buffer (pH7.4). Following transfer to a 96-
well plate coated with a mouse anti-rabbit IgG, the tracer
prostaglandin acetylcholine esterase and primary anti-
body (mouse anti PGE2) were added. Plates were then
incubated at room temperature overnight, reaction mix-
tures were removed, and wells were washed with 10mM
potassium phosphate buffer containing 0.05% Tween 20.
EllmanÕs reagent (200lL) was added to each well and the
plate was incubated at room temperature (exclusion of
light) for 60min, until the control wells yielded an
OD = 0.3–0.8 at 412nm. A standard curve with PGE2
was generated from the same plate, which was used to
quantify the PGE2 levels produced in the presence of test
3. Conclusions
In summary, we have demonstrated that hydroxylated
resveratrol analogues especially compound 10 and 12
are selective COX-2 inhibitors with potency comparable
or better than the clinically established celecoxib. More-
over, hydroxylated resveratrol derivatives also show sig-
nificantly lower IC50 values against COX-2 than
celecoxib. This should result in lower doses necessary
to achieve the same efficacy in clinical studies. The meth-
oxylated analogues are poor inhibitors of COX-2 activ-
ity and do not exhibit COX-2 specificity.

5576
M. Murias et al. / Bioorg. Med. Chem. 12 (2004) 5571–5578
samples. Results were expressed as a percentage relative
to a control (solvent-treated samples). All determina-
tions were performed in duplicate, and values generally
agreed within 10%. Dose–response curves were gener-
ated for the calculation of IC50 values.
(JC,P = 6.5Hz), 123.2 (JC,P = 9.6Hz), 113.9 (JC,P =
2.7Hz), 62.1 (JC,P = 6.9Hz), 55.1, 32.6 (JC,P = 139.1Hz),
16.2 (JC,P = 6.1Hz); MS m/z 258 [M+, 10%], 121 [100%].
4.3.3. Diethyl (3,5-dimethoxybenzyl)phosphonate. Yield:
1
13.25g (92%); H NMR (CDCl3): d 6.47–6.45 (m, 2H),
6.36–6.33 (m, 1H), 4.13–3.96 (m, 4H), 3.77 (s, 6H),
3.09 (d, JH,P = 21.6Hz, 2H), 1.26 (t, J = 7.1Hz, 6H);
13C NMR (CDCl3): 160.6 (JC,P = 3.0Hz), 133.5
(JC,P = 3.0, 8.8Hz), 107.7 (JC,P = 6.5Hz), 99.0
(JC,P = 3.4Hz), 62.1 (JC,P = 6.5Hz), 55.2, 33.8
(JC,P = 138.0Hz), 16.3 (JC,P = 6.1Hz); MS m/z 288
[M+, 40%], 151 [100%].
4.2. Molecular modeling studies
Compounds were draw with ISIS Draw and an Accord
for Excel database was built. The database was exported
as sdf-file and imported into MOE (version 2003.02,
Chemical Computing Group). Multiple linear correla-
tion analysis was performed using Microsoft Excel
2000. Descriptors contributing significantly to the equa-
tion were retrieved via stepwise selection on the basis of
t-values (Student t-test). As dependent variable log(1/
IC50) values were used. In case of selectivity indices,
the log[IC50(COX-1)/IC50(COX-2)] was used. Thus,
higher values indicate higher selectivity for COX-2.
4.3.4. Diethyl (3,4,5-trimethoxybenzyl)phosphonate.
Yield: 14.94g (94%); 1H NMR (CDCl3): d 6.54 (d,
JH,P = 2.5Hz, 2H), 4.11–3.97 (m, 4H), 3.85 (s, 6H),
3.82 (s, 3H), 3.09 (d, JH,P = 21.5Hz, 2H), 1.27 (t,
J = 7.1Hz, 6H); 13C NMR (CDCl3): 153.0
(JC,P = 3.0Hz), 126.9 (JC,P = 9.2Hz), 106.7 (JC,P
=
6.5Hz), 62.1 (JC,P = 6.5Hz), 60.8, 55.9, 33.8
(JC,P = 138.7Hz), 16.3 (JC,P = 6.1Hz); MS m/z 318
[M+, 35%], 181 [100%].
Docking studies were performed using FlexX33 as imple-
mented in Sybyl 6.9.34 The X-ray crystal structures of
SC-558 bound to the active site of COX-2 and ibuprofen
bound to COX-1 were available in the RCSB Protein
Data Bank35 and used as templates to construct the
three-dimensional models of all test compounds. Water
molecules and ligands were removed and hydrogen
atoms were added to the enzymes. The active sites of
4.3.5. General procedure for synthesis of compounds 1–6.
In a flame-dried three-necked-flask the corresponding
methoxylated diethyl benzylphosphonate (10mmol)
was cooled to 0ꢁC under argon. Then 10mL dry
DMF, sodiummethoxide (1.12g, 20mmol) and the cor-
responding methoxylated benzaldehyde (10mmol) were
added. The mixture was stirred at room temperature
for 1h, then heated to 100ꢁC under argon for 1.5h.
The solution was poured into 250mL ice-water, the pre-
cipitate was filtered off and recrystallized from diluted or
pure ethanol to yield white crystals.
˚
COX-1 and COX-2 were defined by a sphere of 7.4A
around the amino acid residues His90, Leu352,
Leu359, and Ala527. Experiments were performed using
formal charges whereby the best 30 solutions were
inspected.
4.3. Chemistry
4.3.6. 3,40,5-Trimethoxystilbene (1). Yield: 1.59g (59%),
mp 55–57ꢁC; H NMR (CDCl3): d 7.43 (d, J = 8.5Hz,
2H), 7.04 (d, J = 16.3Hz, 1H), 6.92–6.85 (m, 3H),
6.65–6.64 (m, 2H), 6.38–6.36 (m, 1H), 3.81 (s, 9H); 13C
NMR (CDCl3): d 160.9, 159.3, 139.6, 129.8, 128.6,
127.7, 126.5, 114.0, 104.2, 99.5, 55.3; MS m/z 270 [M+,
100%].
1
All chemicals obtained from commercial suppliers were
used as received and were of analytical grade. Melting
points were determined on a Kofler hot stage apparatus
and are uncorrected. The 1H and 13C NMR spectra were
recorded on a Varian UnityPlus 200 (200 and 50MHz).
Chemical shifts are reported in d values (ppm) relative to
Me4Si line as internal standard and J values are reported
in Hertz. Mass spectra were obtained by a Shimadzu
GC/MSQP 1000 EX or Hewlett Packard (GC: 5890;
MS: 5970) spectrometer. Solutions in organic solvents
were dried over anhydrous sodium sulfate.
4.3.7. 3,4,40,5-Tetramethoxystilbene (2). Yield: 1.65g
(55%), mp 157ꢁC; 1H NMR (CDCl3): d 7.44 (d,
J = 8.8Hz, 2H), 6.98 (d, J = 16.3Hz, 1H), 6.91–6.83
(m, 3H), 6.71 (s, 2H), 3.90 (s, 6H), 3.86 (s, 3H), 3.82
(s, 3H); 13C NMR (CDCl3): d 159.2, 153.3, 137.7,
133.3, 129.9, 127.6, 127.5, 126.4, 114.1, 103.2, 60.9,
56.0, 55.2; MS m/z 300 [M+, 100%].
4.3.1. General procedure for synthesis of methoxylated
diethyl benzylphosphonates. The corresponding meth-
oxybenzylhalide (50mmol) was heated in a flame-dried
three-necked-flask with an excess of triethyl phosphite
(9.5mL, 55mmol) at 130ꢁC with a CaCl2 tube. After
24h the mixture was cooled to room temperature and
purified by distillation at 4 · 10À3 bar and 110ꢁC to yield
pale oils.
4.3.8. 3,30,5,50-Tetramethoxystilbene (3). Yield: 1.56g
1
(52%), mp 130–132ꢁC; H NMR (CDCl3): d 7.00 (s,
2H), 6.66 (d, J = 2.2Hz, 4H), 6.40–6.38 (m, 2H), 3.81
(s, 12H); 13C NMR (CDCl3): d 160.9, 139.0, 129.1,
104.5, 100.0, 55.3; MS m/z 300 [M+, 100%].
4.3.2. Diethyl (4-methoxybenzyl)phosphonate. Yield:
12.5g (96%); H NMR (CDCl3): d 7.23–7.18 (m, 2H),
6.84 (d, J = 8.6Hz, 2H), 4.15–3.93 (m, 4H), 3.78 (s,
3H), 3.09 (d, JH,P = 21.0Hz, 2H), 1.24 (t, J = 7.1Hz,
3H); 13C NMR (CDCl3): d 158.5 (JC,P = 3.4Hz), 130.6
4.3.9. 3,30,40,5-Tetramethoxystilbene (4). Yield: 1.65g
(55%), mp 67ꢁC; 1H NMR (CDCl3): 7.08–7.00 (m,
3H), 6.93–6.83 (m, 2H), 6.66 (d, J = 2.2Hz, 2H), 6.39–
6.37 (m, 1H), 3.94 (s, 3H), 3.90 (s, 3H), 3.82 (s, 6H);
13C NMR (CDCl3): d 160.9, 149.0, 148.9, 139.5, 130.1,
1

M. Murias et al. / Bioorg. Med. Chem. 12 (2004) 5571–5578
5577
128.9, 126.7, 119.9, 111.1, 108.7, 104.2, 99.6, 55.9, 55.8,
(br s, 6H), 6.57 (s, 2H), 6.44 (s, 4H); 13C NMR (d6-
DMSO): d 146.1, 132.9, 128.1, 125.8, 105.2; MS m/z
276 [M+, 100%].
55.3; MS m/z 300 [M+, 100%].
4.3.10. 3,30,4,5,50-Pentamethoxystilbene (5). Yield: 1.94g
1
4.3.18. Purity of compounds 1–12. The purity of com-
pounds 1–6 and 8–12 obtained by elemental analysis
was within 0.4% of the theoretical values for the for-
mulas given. Compound 7 (resveratrol) was purchased
from Sigma (Munich, Germany) with a purity of
approximately 99%.
(59%), mp 136–138ꢁC; H NMR (CDCl3): d 7.02 (d,
J = 16.1Hz, 1H), 6.91 (d, J = 16.1Hz, 1H), 6.73 (s,
2H), 6.66 (d, J = 2.2Hz, 2H), 6.40–6.38 (m, 1H), 3.91
(s, 6H), 3.87 (s, 3H), 3.82 (s, 6H); 13C NMR (CDCl3):
d 160.9, 153.3, 139.1, 137.9, 132.7, 129.0, 128.0, 104.4,
103.5, 99.8, 60.9, 56.0, 55.2; MS m/z 330 [M+, 100%].
4.3.11. 3,30,4,40,5,50-Hexamethoxystilbene (6). Yield:
1
Acknowledgements
1.76g (49%), mp 215ꢁC; H NMR (CDCl3): d 6.94 (s,
2H), 6.74 (s, 4H), 3.92 (s, 12H), 3.87 (s, 6H); 13C
NMR (CDCl3): d 153.3, 137.8, 132.8, 128.0, 103.3,
60.9, 56.0; MS m/z 360 [M+, 100%].
This study was supported by a grant of the Jubila¨ums-
¨
fonds der Osterreichischen Nationalbank 9894 (W.J.).
M. Murias thanks the European Commission for a
Marie Curie Fellowship.
4.3.12. General procedure for synthesis of compounds 8–
12. In a flame-dried three-necked-flask the corresponding
methoxystilbene (2.5mmol) was solved in dry methylen-
chloride under argon and cooled to À30ꢁC with cardice.
Then boron tribromide (1M-solution in methylenchlo-
ride, 1.5mmol per methoxy-group) was added via syringe.
The solution was warmed to room temperature and stir-
red for 2h. Then the reaction was quenched by adding
20mL water slowly. After stirring for another 30min
the methylenchloride was evaporated and the water-phase
was extracted three times with ethyl acetate. The organic
layers were dried over Na2SO4 and the solvent was re-
moved in vacuo. The resulting crystals were recrystallized
from ethanol/water or pure water to yield brown crystals.
References and notes
1. Xie, W. L.; Chipman, J. G.; Robertson, D. L.; Erikson, R.
L.; Simmons, D. L. Proc. Natl. Acad. Sci. U.S.A. 1991, 88,
2692.
2. Fournier, D. B.; Gordon, G. B. J. Cell Biochem. Suppl.
2000, 34, 97.
3. Hawkey, C.; Laine, L.; Simon, T.; Beaulieu, A.; Maldo-
nado, C. J.; Acevedo, E.; Shahane, A.; Quan, H.;
Bolognese, J.; Mortensen, E. Arthritis Rheum. 2000, 43,
370.
4. Anderson, W. F.; Umar, A.; Hawk, E. T. Expert Opin.
Pharmacother. 2003, 4, 2193.
5. Quadros, A.; Patel, N.; Crescentini, R.; Crawford, F.;
Paris, D.; Mullan, M. Neurosci. Lett. 2003, 15,
353.
6. Giovannini, M. G.; Scali, C.; Prosperi, C.; Bellocci, A.;
Pepeu, G.; Casamenti, F. Int. J. Immunopathol. Pharma-
col. 2003, 16, 31.
7. Hunot, S.; Vila, M.; Teismann, P.; Davis, R. J.; Hirsch, E.
C.; Przedborski, S.; Rakic, P.; Flavell, R. A. Proc. Natl.
Acad. Sci. U.S.A. 2004, 13, 665.
8. Teismann, P.; Vila, M.; Choi, D. K.; Tieu, K.; Wu, D. C.;
Jackson-Lewis, V.; Przedborski, S. Ann. N.Y. Acad. Sci.
2003, 99, 272.
9. Profita, M.; Sala, A.; Bonanno, A.; Riccobono, L.; Siena,
L.; Melis, M. R.; DiGiorni, R.; Mirabella, F.; Gjomarkaj,
M.; Monsignore, G.; Vignola, A. M. J. Allergy Clin.
Immunol. 2003, 112, 709.
10. Jang, M.; Cai, L.; Udeani, G. O.; Slowing, K. V.; Thomas,
C. F.; Beecher, C. W.; Fong, H. H.; Farnswoth, N. R.;
Kinghorn, A. D.; Mehta, R. G.; Moon, R. C.; Pezzuto,
J. M. Science 1997, 275, 218.
4.3.13. 3,4,40,5-Tetrahydroxystilbene (8). Yield: 0.33g
(55%), mp 240ꢁC; H NMR (d6-DMSO): d 8.78 (br s,
4H), 7.35 (d, J = 8.6Hz, 2H), 6.76–6.72 (m, 4H), 6.47
(s, 2H); 13C NMR (d6-DMSO): d 156.7, 146.1, 132.9,
128.5, 128.2, 127.4, 125.9, 125.1, 115.5, 105.2; MS m/z
244 [M+, 100%].
1
4.3.14. 3,30,5,50-Tetrahydroxystilbene (9). Yield: 0.37g
1
(60%), mp >320ꢁC; H NMR (d6-DMSO): d 9.24 (br
s, 4H), 6.84 (s, 2H), 6.41 (d, J = 2.0Hz, 4H), 6.16 (t,
J = 1.9Hz, 2H); 13C NMR (d6-DMSO): d 158.5, 138.7,
128.4, 128.4, 104.6, 102.2; MS m/z 244 [M+, 100%].
4.3.15. 3,30,40,5-Tetrahydroxystilbene (10). Yield: 0.31g
1
(51%), mp 236ꢁC; H NMR (d6-DMSO): d 8.70 (br s,
4H), 6.97 (s, 1H), 6.87–6.63 (m, 4H), 6.40 (d, 1.9Hz,
2H), 6.20–6.19 (m, 1H); 13C NMR (d6-DMSO): d
158.0, 144.9, 144.8, 138.9, 128.6, 127.8, 125.4, 118.3,
115.3, 112.7, 104.1,101.6; MS m/z 244 [M+, 100%].
11. Likhitwitayawuid, K.; Sawasdee, K.; Kirtikara, K. Planta
Med. 2002, 68, 841.
12. Yu, C.; Shin, Y. G.; Chow, A.; Li, Y.; Kosmeder, J. W.;
Lee, Y. S.; Hirschelman, W. H.; Pezzuto, J. M.; Mehta, R.
G.; vanBreemen, R. B. Pharm. Res. 2002, 19,
1907.
13. Goldberg, D. M.; Yan, J.; Soleas, G. J. Clin. Biochem.
2003, 36, 79.
14. Potter, G. A.; Patterson, L. H.; Wanogho, E.; Perry, P. J.;
Butler, P. C.; Ijaz, T.; Ruparelia, K. C.; Lamb, J. H.;
Farmer, P. B.; Stanley, L. A.; Burke, M. D. Br. J. Cancer
2002, 86, 774.
15. Geahlen, R. L. O.; McLaughlin, J. L. Biochem. Biophys.
Res. Commun. 1989, 165, 214.
4.3.16. 3,30,4,5,50-Pentahydroxystilbene (11). Yield:
0.370g (57%), mp 252ꢁC; 1H NMR (d6-acetone): d
8.00 (br s, 5H), 6.89 (d, J = 16.3Hz, 1H), 6.76 (d,
J = 16.3Hz, 1H), 6.64 (s, 2H), 6.51 (d, J = 2.1Hz, 2H),
6.28–6.25 (m, 1H); 13C NMR (d6-acetone): d 160.2,
147.4, 141.4, 134.7, 130.5, 130.3, 127.7, 107.3, 106.3,
103.3; MS m/z 260 [M+, 100%].
4.3.17. 3,30,4,40,5,50-Hexahydroxystilbene (12). Yield:
1
0.31g (45%), mp 270ꢁC; H NMR (d6-DMSO): d 8.66

5578
M. Murias et al. / Bioorg. Med. Chem. 12 (2004) 5571–5578
16. Solladie, G.; Pasturel-Jacope, Y.; Maigan, J. Tetrahedron
2003, 59, 3315.
17. Cushman, M.; Nagarathnam, D.; Gopal, D.; Chakraborti,
A. K.; Lin, C. M.; Hamel, E. J. Med. Chem. 1991, 34,
2579.
26. Tacconelli, S.; Capone, M. L.; Sciulli, M. G.; Ricciotti, E.;
Patrignani, P. Curr. Med. Res. Opin. 2002, 18, 503.
27. Reddy, C. M.; Bhat, V. B.; Kiranmai, G.; Reddy, M. N.;
Reddanna, P.; Madyastha, K. M. Biochem. Biophys. Res.
Commun. 2000, 277, 599.
18. Kim, S.; Ko, H.; Park, J. E.; Jung, S.; Lee, S. K.; Chun, Y.
J. Med. Chem. 2002, 45, 160.
28. Habeeb, A. G.; PraveenRao, P. N.; Knaus, E. E. J. Med.
Chem. 2001, 44, 3039.
19. Wang, M.; Jin, Y.; Ho, C. J. Agric. Food Chem. 1999, 47,
3974.
29. Akaho, E.; Fujikawa, C.; Runion, H. I.; Hill, C. R.;
Nakano, H. J. Chem. Software 1999, 5, 3.
20. Thakkar, K.; Geahlen, R.; Cushman, M. J. Med. Chem.
1993, 36, 2950.
30. Chavette, P.; Yous, S.; Marot, C.; Baurin, N.; Lesieur, D.
J. Med. Chem. 2001, 44, 3223.
21. Steynberg, J. P.; Ferreira, D.; Roux, D. G. J. Chem. Soc.,
Perkin Trans. 1 1987, 8, 1705.
22. Drewes, S. E.; Fletcher, I. P. J. Chem. Soc., Perkin Trans.
1 1974, 9, 961.
23. Gokaraju Ganga, R.; Gokaraju Rama, R.; Gottumuk-
kala, S. V.; Somepalli, V. International Patent WO 2004/
000302 A1.
24. Lee, D.; Cuendet, M.; Vigo, J. S.; Graham, J. G.;
Cabieses, F.; Fong, H. H. S.; Pezzuto, J. M.; Kinghorn,
A. D. Org. Lett. 2001, 3, 2171.
31. Kurumbail, R. G.; Stevens, A. M.; Gierse, J. K.;
McDonald, J. J.; Stegeman, R. A.; Pak, J. Y.; Gildehaus,
D.; Miyashiro, J. M.; Penning, T. D.; Seibert, K.; Isakson,
P. C.; Stallings, W. C. Nature 1996, 384, 644.
32. Szewczuk, L. M.; Forti, L.; Stivala, L. A.; Penning, T. M.
J. Biol. Chem. 2004, 279, 22727.
33. Kramer, B.; Rarey, M.; Lengauer, T. PROTEINS 1999,
37, 228–241.
34. SYBYLꢂ 6.9.2 Tripos Inc., 1699 South Hanley Rd., St.
Louis, Missouri, 63144, USA.
25. Palomer, A.; Cabre, F.; Pascual, J.; Campos, J.; Trujillo,
M. A.; Entrena, A.; Gallo, M. A.; Garcia, L.; Mauleon,
D.; Espinosa, A. J. Med. Chem. 2002, 45, 1402.
35. Berman, H. M.; Westbrook, J.; Feng, Z.; Gilliland, G.;
Bhat, T. N.; Weissig, H.; Shindyalov, I. N.; Bourne, P. E.
Nucleic Acids Res. 2000, 28, 235–242.