Structure-activity relationships and molecular modeling analysis of flavonoids binding to the benzodiazepine site of the rat brain GABA(A) receptor complex

Article abstract of DOI:10.1021/jm991010h

The affinities for the benzodiazepine binding site of the GABA(A) receptor of 21 flavonoids have been studied using [³H]flumazenil binding to rat cortical membranes in vitro. We show that flavonoids with high affinity for the benzodiazepine receptor in vitro spanning the whole efficacy range from agonists (1q) to inverse agonists (1l) can be synthesized. The receptor binding properties of the flavonoids studied can successfully be rationalized in terms of a comprehensive pharmacophore model recently developed by cook and co-workers (Drug Des. Dev. 1995, 12, 193-248), supporting the validity of this model. However, in contrast to the requirement by the model that an interaction with the hydrogen bond-accepting site A2 is necessary for compounds to display inverse agonistic activity, 6-methyl-3'-nitroflavone (1l), which cannot engage in such an interaction, nevertheless displays inverse agonism. The analysis of the binding affinities of 3'- and 4'- substituted flavones in terms of the pharmacophore model has yielded new information for the further development of the pharmacophore model.

Full text of DOI:10.1021/jm991010h

J . Med. Chem. 1999, 42, 4343-4350
4343
Str u ctu r e-Activity Rela tion sh ip s a n d Molecu la r Mod elin g An a lysis of
F la von oid s Bin d in g to th e Ben zod ia zep in e Site of th e Ra t Br a in GABAA
Recep tor Com p lex
Kim Dekermendjian,*,† Pia Kahnberg,‡ Michael-Robin Witt,† Olov Sterner,‡ Mogens Nielsen,† and
Tommy Liljefors§
Research Institute of Biological Psychiatry, Sct. Hans Hospital, DK-4000 Roskilde, Denmark, Department of
Organic Chemistry 2, University of Lund, P.O. Box 124, S-221 00 Lund, Sweden, and Department of Medicinal Chemistry,
Royal Danish School of Pharmacy, Universitetsparken 2, DK-2100 Copenhagen, Denmark
Received J anuary 28, 1999
The affinities for the benzodiazepine binding site of the GABAA receptor of 21 flavonoids have
been studied using [3H]flumazenil binding to rat cortical membranes in vitro. We show that
flavonoids with high affinity for the benzodiazepine receptor in vitro spanning the whole efficacy
range from agonists (1q) to inverse agonists (1l) can be synthesized. The receptor binding
properties of the flavonoids studied can successfully be rationalized in terms of a comprehensive
pharmacophore model recently developed by Cook and co-workers (Drug Des. Dev. 1995, 12,
193-248), supporting the validity of this model. However, in contrast to the requirement by
the model that an interaction with the hydrogen bond-accepting site A2 is necessary for
compounds to display inverse agonistic activity, 6-methyl-3′-nitroflavone (1l), which cannot
engage in such an interaction, nevertheless displays inverse agonism. The analysis of the
binding affinities of 3′- and 4′-substituted flavones in terms of the pharmacophore model has
yielded new information for the further development of the pharmacophore model.
In tr od u ction
to the BZD receptor (BzR) with high affinity, e.g.,
triazolopyridazines, cyclopyrrolones, quinolines, and
â-carbolines.9 Additionally, it has been found that some
naturally occurring flavones and synthetic derivatives
bind with high affinity to the BzR.10,11 More than 4000
chemically unique flavonoids have been isolated from
plants, making the flavonoids one of the most important
classes of metabolites.12
The major inhibitory neurotransmitter in the mam-
malian central nervous system (CNS), γ-aminobutyric
acid (GABA), exerts its physiological effects by binding
to three different receptor types in the neuronal mem-
brane: the GABAA, GABAB, and GABAC receptors. The
GABAB receptor belongs to the G-protein-coupled recep-
tor superfamily,2 while the GABAA and GABAC recep-
tors are ligand-gated chloride ion channel complexes.
The GABAA receptor is thought to form a pentameric
structure assembled from a gene family consisting of
six R, four â, three γ, one δ, one ꢀ, one π, and three F
subunits, each encoded by different genes.5 In addition
to GABA binding sites, the GABAA receptor chloride
channel complex possesses binding sites for compounds
that allosterically modify the chloride channel gating
of GABA, such as benzodiazepines (BZDs), â-carbolines,
barbiturates, and certain steroids.6,7 BZDs pharmaco-
logical effects (anxiolytic, anticonvulsant, muscle relax-
ant, and sedative-hypnotic) make them the most im-
portant GABAA receptor-modulating drugs in clinical
use.8 The interaction of BZD agonists with the GABAA
receptor complex increases the GABA-induced chloride
channel opening frequency, which results in membrane
hyperpolarization, thus reducing neuronal excitability.
The BZD binding sites in the brain were identified and
described by radioligand receptor binding assays using
[3H]BZDs as ligands, and originally it was found that
only 1,4-BZD derivatives bind to these receptors. It has
since been shown that many groups of compounds bind
3
4
A comprehensive pharmacophore model for the BzR
based on structure-activity relationship studies for 136
different ligands from 10 structurally different classes
of compounds has recently been developed by Cook and
co-workers.1 In this model the authors assume that BzR
agonists, antagonists, and inverse agonists share the
same binding pocket. The pharmacophore model has
successfully been employed in the design of novel BzR
ligands13,14 and has recently been used in the structure-
activity analyses of novel â-carboline ligands15 such as
N-(indol-3-ylglyoxylo)benzylamines16 and N′-(phenylin-
dol-3-yl)glyoxylohydrazides.17 Although the pharma-
cophore model is based on ligands from a large number
of different classes of compounds, flavonoids were not
included in the development of the model. We have used
radioligand receptor binding techniques to investigate
the affinities for the benzodiazepine receptor of 21
flavonoid derivatives (9 commercially available, 5 re-
ported in the literature, and 7 developed in our labora-
tory), providing an opportunity to test the validity and
usefulness of the model.
Resu lts a n d Discu ssion
* Corresponding author. Tel: (45) 4633 4971. Fax: (45) 4633 4367.
E-mail: kim.dekermendjian@shh.hosp.dk.
† Sct. Hans Hospital.
Ch em istr y. The structures of the flavonoids inves-
tigated in this study are shown in Chart 1. 6-Methyl-
2′-nitroflavone (1j), 3′,6-dimethylflavone (1k ), 6-methyl-
‡ University of Lund.
§ Royal Danish School of Pharmacy.
10.1021/jm991010h CCC: $18.00 © 1999 American Chemical Society
Published on Web 09/24/1999

4344 J ournal of Medicinal Chemistry, 1999, Vol. 42, No. 21
Dekermendjian et al.
Sch em e 1a
Sch em e 2a
a
(a) H2PO4, P2O5.
Sch em e 3a
a
(a) HNO3, P2O5.
a
(a) Ac2O, pyridine; (b) AlCl3; (c) o-, m-, p-nitrobenzoyl chloride
or m-toluoyl chloride, pyridine; (d) KOH, pyridine; (e) H2SO4.
Ta ble 1. Ki Values and GABA Ratio of Flavones Tested on
[3H]Flumazenila Binding in Vitro to Rat Cortical Membranes
Ch a r t 1
no.
name
Ki value (nM) GABA ratio
1a flavone
4200 ( 30
>75000
1.05 ( 0.03
NR
1b 3-bromoflavonea,28
1c 5-hydroxyflavone
1d 6-methylflavone
1e 6-hydroxyflavone
1900 ( 34
125 ( 8
400 ( 60
570 ( 30
70 ( 9
4200 ( 270
920 ( 80
1000 ( 40
29 ( 2
5.6 ( 0.7
16
1.03 ( 0.07
0.98 ( 0.08
1.06 ( 0.10
1.51 ( 0.16
1.80
1.14 ( 0.08
0.90 ( 0.11
1.49 ( 0.32
1.63 ( 0.27
0.72 ( 0.05
1.4
1f
6-methoxyflavone
1g 6-bromoflavonea,28
1h 7-hydroxyflavone
1i
1j
5,7-dihydroxyflavone
6-methyl-2′-nitroflavone
1k 3′,6-dimethylflavone
1l
6-methyl-3′-nitroflavone
1m 6-bromo-3′-nitroflavonea,28
1n 3′,6-dinitroflavonea,29
26 ( 3
1.3
1o 6-methyl-4′-nitroflavone
7300 ( 2300 ND
1p 4′,6-dinitroflavonea,30
17000 ( 5000 NR
1q 6-methyl-3′,5-dinitroflavone
1.9 ( 0.3
770 ( 40
310 ( 37
33 ( 2
1.41 ( 0.11
1r
4′,5,7-trihydroxyflavone
1.03 ( 0.12
1.35 ( 0.18
1.51 ( 0.19
1.28 ( 0.05
2.13 ( 0.35
1.05 ( 0.11
0.57 ( 0.14
2a 6-methylthioflavone
2b 6-methyl-3′-nitrothioflavone
3
5,6-benzoflavone
diazepam
1480 ( 15
8.5 ( 1.6
1.2 ( 0.4
7.6 ( 1.3
flumazenil
DMCM
a
Compounds referred from literature were tested on [3H]fluni-
trazepam binding. All values are the mean ( SEM of three to five
3′-nitroflavone (1l), and 6-methyl-4′-nitroflavone (1o)
were synthesized according to procedures that previ-
ously have been described for other flavones.18 The
synthesis involved an acetylation, a Fries rearrange-
ment, a modified Schotten-Baumann reaction, and a
Baker-Venkataraman rearrangement followed by a
cyclization with acid (see Scheme 1).
6-Methylthioflavone (2a ) was prepared as previously
published,19 and the same procedure was adapted for
the synthesis of 6-methyl-3′-nitrothioflavone (2b) ac-
cording to Scheme 2. 6-Methyl-3′,5-dinitroflavone (1q)
was synthesized by nonspecific nitration of 6-methylfla-
vone (1d ), according to Scheme 3.
separate experiments. NR, not reported; ND, not determined.
on [3H]flumazenil binding to rat cortical membranes;
this affinity is of similar potency as that of flumazenil
(Ki ) 1.2 nM). Compound 1q shows a competitive
inhibition (Figure 1) of [3H]flumazenil binding (with an
increase in the KD value while the number of binding
sites was unaffected). Several other flavone derivatives
investigated in this study showed competitive inhibition
of [3H]flumazenil binding (data not shown).
To evaluate the pharmacological profile of compounds
interacting with the BzR by receptor binding assays, the
GABA ratio (see Experimental Section) can be useful
to indicate the efficacy of BzR ligands.20 GABA ratios
> 1 indicate that compounds have agonistic profiles, e.g.
diazepam, while GABA ratios < 1 indicate compounds
with inverse agonistic profiles, e.g. DMCM, and GABA
ratios ∼ 1 indicate that compounds have antagonistic
Recep tor Bin d in g. As shown in Table 1 the tested
flavonoids inhibit [3H]flumazenil binding with a wide
range of affinities, from low nanomolar (1q and 1l) to
high micromolar (1b and 1p ) concentrations. Compound
1q shows the highest affinity with a Ki value of 1.9 nM

SAR and Molecular Modeling of Flavonoid Binding
J ournal of Medicinal Chemistry, 1999, Vol. 42, No. 21 4345
Ch a r t 2
at A2, H1, and L1 are required for potent inverse
agonism. Furthermore, it has been concluded that a
planar or close-to-planar geometry is required for ligands
with potent activity.1
Con for m a tion a l An a lysis of F la von es. Conforma-
tional analysis by using the MM3(92) and MM2(91)
programs (see Experimental Section) indicates that all
flavones which do not contain a 3- or 2′-substituent are
calculated to be almost planar molecules. In the global
energy minimum, these flavones display only a moder-
ate twist (ca. 20°) of the 2-phenyl ring with respect to
the bicyclic ring system. In addition, the energy required
for coplanarity of the two ring systems is calculated to
be small, less than 0.2 kcal/mol. In contrast, due to
strong steric repulsions between the bromo substituent
and the o-hydrogens of the 2-phenyl ring in 3-bromofla-
vone (1b), the phenyl ring in this compound is calculated
to be 63° twisted and the energy required for planarity
is calculated to be high, 6.4 kcal/mol. Thus, the inactiv-
ity of 1b (Ki > 75 000 nM, Table 1) may be attributed
to the high energy for the planar conformation in
accordance with the conclusion drawn by Zhang et al.1
that a planar or close-to-planar geometry is required for
ligand binding to the benzodiazepine receptor.
F igu r e 1. Representative Scatchard plots of [3H]flumazenil
binding to rat cortical GABAA receptors in vitro in the absence
(b) and presence (2) of 1q (6-methyl-3′,5-dinitroflavone) (2.5
nM): (b) KD ) 0.9 nM, Bmax ) 1.1 pmol/mg of protein; (2)
KD ) 1.8 nM, Bmax ) 1.2 pmol/mg of protein.
The 2-phenyl ring of the 2′-NO2-substituted flavone
1j is calculated to have a twist of 45°. However, the
energy required for coplanarity of the 2-phenyl ring and
the bicyclic ring system is quite small, 1.5 kcal/mol. The
coplanar 2-phenyl group rotamer with the NO2 group
pointing toward the ether oxygen is calculated to be 2.9
kcal/mol lower in energy than the alternative one. It
should be noted that in the planar conformation of 1j
the 2′-NO2 group is calculated to be essentially orthogo-
nal to the 2-phenyl ring.
Due to the larger size of the sulfur atom compared to
the oxygen atom, the thioflavones 2a and 2b display a
larger twist (30-32°) of the 2-phenyl ring than the
corresponding flavones 1d and 1l. The conformational
energy required for a planar geometry of 2a and 2b is
calculated to be 1.2-1.4 kcal/mol.
In conclusion, all flavones studied in the present work,
except 1b, may attain coplanarity or close-to-coplanarity
of the 2-phenyl ring and the bicyclic ring system with
only a small increase of the conformational energy.
F ittin g of th e F la von es to th e P h a r m a cop h or e
Mod el. To fit the flavones studied in the present work
to the pharmacophore model, the high-affinity partial
agonist CGS-9896 (4) (see Chart 2) was employed as a
template. The conformation of the flavones with co-
planar ring systems was used in the molecular super-
impositions. The fitting points used are (i) the centroids
of the fused benzene ring in 4 and of the 2-phenyl ring
in the flavones, (ii) the carbonyl oxygen, and (iii) the
N1 atom in 4 and the oxygen or sulfur atom in position
1 in the flavones. The fitting of the flavone derivative
1d to the pharmacophore model is shown in Figure 3.
F igu r e 2. Pharmacophore model developed by Cook and co-
workers.1 CGS-9896 (4; filled atoms) and dihydropyrido[3,4-
b:5,4-b′]diindole (5; unfilled carbon atoms) (see also Chart 2)
are shown in the model.
profiles, e.g. flumazenil (see Table 1). As shown in Table
1, 1q has a GABA ratio of 1.4 suggesting that this
compound is a partial agonist as compared to diazepam
(GABA ratio 2.1 in the assay conditions used here),
which is a full agonist. The GABA ratios of 1f, 1g, 1j,
1k , 1m , 1n , 2a , 2b, and 3 suggest that these compounds
have a partial agonistic profile, while for compounds 1a ,
1c, 1d , 1e, 1h , 1i, and 1r an antagonistic profile would
be predicted. Compound 1l (Ki ) 5.6 nM) has a GABA
ratio of 0.7 indicating an inverse agonistic activity. Thus
it seems that, as for other classes of BzR ligands,9
flavone derivatives spanning the whole efficacy range
can be developed.
P h a r m a cop h or e Mod el. The key elements of the
pharmacophore model developed by Cook and co-work-
ers1 are shown in Figure 2. H1 and A2 are hydrogen
bond donor and acceptor sites, respectively, whereas H2/
A3 is a bifunctional hydrogen bond donor/acceptor site.
L1, L2, and L3 are three lipophilic pockets, and S1, S2,
and S3 denote regions of steric repulsive ligand-
receptor interactions (receptor-essential volumes). Ac-
cording to the analysis of Zhang et al.,1 occupation of
areas L2 and/or L3 and interactions at H1, H2, and L1
are important for agonist activity, whereas interactions

4346 J ournal of Medicinal Chemistry, 1999, Vol. 42, No. 21
Dekermendjian et al.
F igu r e 4. Proposed binding mode of 6-methyl-3′,5-dinitrofla-
vone (1q) in the pharmacophore model.
F igu r e 3. 6-Methylflavone (1d ; unfilled carbon atoms) fitted
to the pharmacophore model by superimposition with CGS-
9896 (4; filled atoms).
Compound 1o with a 4′-NO2 substituent in the
2-phenyl ring displays a drastically decreased affinity
compared to its parent compound 1d . Similarly, the 4′-
NO2 group in 1p is responsible for the very low affinity
of this compound. On comparing the affinities of 1n and
1p it is clear that the 6-NO2 group in the 6,4′-di-NO2
compound 1p is compatible with a reasonably high
affinity. The area in the vicinity of the 4′-position in the
flavones (see Figure 3) does not seem to have been
mapped out in the pharmacophore model developed by
Cook and co-workers.1 According to the alignment rules
of the pharmacophore model, the 4′-position in the
flavones corresponds to the 6-position in â-carbolines,
and a methoxy substituent in this position in the
â-carbolines (DMCM) is compatible with high affinity
for the receptor (Table 1). In addition, from a comparison
of the affinities of 1j and 1r it is clear that the
4′-hydroxy group in 1r does not result in a decrease of
the affinity. Molecular electrostatic potentials calculated
by the AM1 method show very little change of the
electrostatic potentials in the vicinity of carbonyl and
ether oxygens due to the presence of a 4′-NO2 substitu-
ent. Thus, it may be concluded that the very low
affinities of the 4′-NO2 flavones 1o and 1p are most
probably due to electrostatic repulsive interactions
between the strongly electron-rich 4′-NO2 substituent
and the receptor.
The 2′- and 3′-NO2 groups in compounds 1j, 1l, 1m ,
1n , 1q, and 2b may be placed in two alternative
positions in the superimposition shown in Figure 3. This
substituent may be positioned “upwards” toward the A2
site or “downwards” toward the H2/A3 site. As described
above, a 2′-NO2 group strongly favors a “downwards”
orientation of the substituent, whereas the two phenyl
group rotamers have essentially identical energies for
the 3′-NO2-substituted compounds. However, consider-
ing that the A2 site is a hydrogen bond-accepting site
and, thus, is negatively charged or at least electron-rich,
a positioning of the NO2 groups toward this site in the
pharmacophore model should lead to prohibitive elec-
trostatic repulsions. Thus, 2′- and 3′-NO2 groups should
be directed “downwards” as exemplified by the 3′-NO2
group in compound 1q displayed in Figure 4. Da Settimo
et al.16,17 recently reported structure-activity relation-
ships for a series of N-(indol-3-ylglyoxylyl)benzylamines.
In ter a ction s of th e F la von es w ith th e Sites of
th e P h a r m a cop h or e Mod el. As shown in Figure 3,
the flavone carbonyl group and the carbonyl group in 4
(see Chart 2) interact with the hydrogen bond-donating
site H1 in essentially the same manner. Furthermore,
the benzene rings chosen for fitting superimpose almost
perfectly and the position of the ether oxygen in the
flavones is very close to that of N1 in 4, making very
similar hydrogen bond interactions with site H2 pos-
sible. The fused phenyl ring in the flavones occupies the
lipophilic region L1 as does the corresponding benzene
ring in 4. The 6-methyl substituent in 1d fills out the
lipophilic site L2, but not to the same extent as the
chlorine atom in 4 (Figure 3).
The 6-substituted flavones clearly fulfill the interac-
tion criteria for agonist activity described above. The
data in Table 1 show that the affinity of the flavone
ligand increases significantly when a 6-substituent, in
particular a less polar one as in 1d and 1g, is present.
In addition, the larger methoxy (1f) and bromo (1g)
substituents efficiently fill out the lipophilic pocket L2
which, in agreement with the proposed pharmacological
effects of such an interaction,1 enhances the agonist
properties of the compounds (Table 1). In contrast to
this, a substituent in the 7-position, as in 1h , does not
give a positive contribution to the affinity, presumably
due to repulsive interactions with the receptor essential
volume at the S2 site.
Interestingly, interaction with the hydrogen bond-
accepting site A2 is not possible for the flavones studied
in the present work. According to the pharmacophore
model such an interaction is required for potent inverse
agonism.1 Despite this, the 6-methyl-3′-nitro derivative
1l displays inverse agonist activity (GABA ratio 0.7, see
Table 1).
Replacement of the ether oxygen in 1d and 1l by
sulfur giving 2a and 2b results in a modest affinity
decrease by a factor of 3-6. This is most probably due
to the weaker hydrogen bond-accepting properties of
sulfur compared to oxygen and is consistent with the
result from a 3D-QSAR study that a less negative
electrostatic potential corresponding to a weaker hy-
drogen bond at the hydrogen bond-donating H2 site
decreases the affinity.21,22

SAR and Molecular Modeling of Flavonoid Binding
J ournal of Medicinal Chemistry, 1999, Vol. 42, No. 21 4347
Ch a r t 3
The compounds were analyzed in terms of the pharma-
cophore model developed by Cook and co-workers1.
Comparing the molecular alignments in ref 16 with the
alignment displayed in Figure 3, it is clear that the
position of the 5-NO2 group in 6 (Chart 3) corresponds
to the 3′-NO2 group in the flavones 1l, 1m , 1n , 1q, and
2b when the 3′-NO2 group is positioned as in Figure 4.
As shown in Table 1, a 3′-NO2 group in the flavones
leads to a significant increase of the affinity (cf. 1d vs
1l, 1g vs 1m , and 2a vs 2b). Similarly, a 5-NO2 in the
N-(indol-3-ylglyoxylyl)benzylamine series increases the
affinity for the receptor.16 This strongly supports the
proposed bioactive conformation of the 3′-NO2-phenyl
group in the flavone series shown in Figure 4. Other
types of substituents in this position in the pharma-
cophore model such as a 3′-methyl group in the flavone
series (1k ) and a 5-chloro group in the N-(indol-3-
ylglyoxylyl)benzylamine series16 also result in an affin-
ity increase, albeit somewhat smaller than shown by a
NO2 substituent. Thus, this substituent position is
clearly of great interest for further exploration.
F igu r e 5. Molecular electrostatic potentials (kcal/mol) cal-
culated by ab initio HF/6-31G* calculations. The electrostatic
potential in the plane of the carbonyl group and at a distance
2.0 Å from the carbonyl oxygen in the direction toward the
H1 site is denoted by an asterisk.
F igu r e 6. Superimposition of 5,6-benzoflavone (3) and the
inactive dihydropyrido[3,4-b:5,4-b′]diindole (7).
Adding a 5-nitro group to 1l giving 1q increases the
affinity by a factor of 3 and gives the most potent
compound in the present series. According to MM3(92)
supported by ab initio HF/6-31G* calculations, the
5-nitro group in 1q is, as shown in Figure 4, calculated
to be essentially orthogonal to the bicylic part of the
flavone skeleton. This is due to steric repulsions between
the oxygens of the nitro group and the flanking methyl
and carbonyl groups. As illustrated in Figure 4, the
5-NO2 group in 1q fits nicely between the steric repul-
sive region S1 and the hydrogen donor site H1 in the
pharmacophore model. A rationalization of the positive
effect of the 5-NO2 group on the affinity is shown in
Figure 5. Molecular electrostatic potentials calculated
by ab initio HF/6-31G* calculations show that the 5-NO2
group increases the negative electrostatic potential at
a position corresponding to a hydrogen-bonding hydro-
gen at site H1. According to a 3D-QSAR analysis by
Allen et al.,22 this is predicted to increase the affinity
of the ligand. On comparing the affinities of compounds
1a and 1h with those of 1c and 1i, respectively, it is
clear that a 5-hydroxy substituent gives a similar
increase in the affinity. Calculations of the electrostatic
potentials as above also in this case display an increased
electrostatic potential at site H1 when the hydroxy
group acts as a hydrogen bond acceptor in the direction
toward H1.
2′-NO2 flavone derivative most probably has its nitro
group directed toward the H2/A3 site in the pharma-
cophore model. To attain coplanarity of the ring systems
in 1j the NO2 group is calculated to be essentially
orthogonal to the 2-phenyl ring. This orthogonality of
the 2′-NO2 group may be detrimental to high-affinity
binding. Another reason for its low affinity may be a
steric conflict between the 2′-NO2 group and the H2/A3
site. Such a conflict may result in rotation about the
1′-2 bond and consequently to a noncoplanarity of the
flavone ring systems.
The low affinity of compound 3 compared to the
6-methyl derivative 1d may be rationalized by the
presence of steric repulsive interactions with the recep-
tor in region S1 (Figure 2) due to the fused benzene ring
in 3. Such steric repulsions are also displayed by the
inactive dihydropyrido[3,4-b:5,4-b′]diindole (7) shown in
Figure 6.1 Figure 6 also displays a superimposition of 3
and 7 demonstrating the similarities in the positions of
the corresponding fused benzene rings in the two
compounds.
Con clu sion s
We have shown that flavonoids with high affinity (e.g.
1q and 1l) for the BzR can be synthesized, spanning
the whole efficacy range of BzR ligands from agonistic
(1q) over antagonistic (1d ) to inverse agonistic (1l)
The 2′-NO2 group decreases the affinity of 1j com-
pared to that of its parent compound 1d by a factor of
8. As discussed above, the bioactive conformation of a

4348 J ournal of Medicinal Chemistry, 1999, Vol. 42, No. 21
Dekermendjian et al.
(CDCl3) 20.9, 21.0, 53.4, 96.7, 118.5, 118.8, 119.1, 119.3, 124.8,
125.0, 128.8, 128.9, 128.9, 129.2, 130.5, 130.5, 130.8, 131.4,
132.0, 133.3, 135.2, 137.6, 138.1, 139.0, 148.8, 161.1, 176.3,
196.2, 196.5, 199.7; MS m/z (% rel int) 299.0796 (M+, 12,
properties. The receptor binding properties of the fla-
vonoids studied in this work can successfully be ratio-
nalized in terms of the pharmacophore model developed
by Cook and co-workers1 supporting the validity of this
model. However, in contrast to the requirement by the
model of an interaction with the hydrogen bond-accept-
ing site A2 in order to obtain potent inverse agonism,
6-methyl-3′-nitroflavone (1l), which cannot engage in
such an interaction, is an inverse agonist. Analysis of
the affinities of 3′- and 4′-substituted flavones in terms
of the pharmacophore model has yielded new informa-
tion on the corresponding areas of the pharmacophore
model. This information may be employed in the further
development of the model.
C
16H13NO5 requires 299.0794), 165 (16%), 150 (22%), 135
(100%), 109 (11%), 77 (16%).
6-Meth yl-2′-n itr ofla von e (1j). Concentrated sulfuric acid
(95-97%) (1.32 g, 13.5 mmol) was added to a suspension of
the diketone II (4.0 g, 13.4 mmol) in 20 mL of glacial acetic
acid and refluxed for 2 days, whereafter the mixture was
poured onto 160 g of ice. After 30 min the white crystals formed
were collected by filtration, washed with 200 mL of water and
recrystallized from acetone to give white crystals (yield
1
100%): mp 178-181 °C; H NMR (CDCl3) 2.48 (s, 3H), 6.59
(s, 1H), 7.31 (d, 1H, J ) 8.6), 7.70 (ddd, 1H, J ) 7.3, 1.8, 0.5),
7.73 (ddd, 1H, J ) 8.1, 7.8, 1.8), 7.78 (ddd, 1H, J ) 7.8, 7.3,
1.5), 8.04 (s, 1H), 8.09 (d, 1H, J ) 8.1); 13C NMR (CDCl3) 21.4,
111.6,118.2, 123.9, 125.4, 125.5, 128.4, 131.6, 132.2, 133.8,
135.8, 136.1, 148.6, 155.1, 162.4, 178.6; MS m/z (% rel int)
281.0685 (M+, 33, C16H11NO4 requires 281.0688), 253 (12%),
225 (11%), 224 (9%), 196 (8%), 178 (11%), 152 (6%), 134 (100%),
106 (23%).
Exp er im en ta l Section
Ch em istr y. 1H and 13C NMR were recorded at room
temperature with a Bruker DRX400 or a Bruker ARX500
spectrometer. The spectra were recorded in CDCl3, and the
solvent signals (7.26 and 77.0 ppm, respectively) were used
as reference; data are reported in δ values and J in Hz. The
raw data were transformed and the spectra were evaluated
with the standard Bruker UXNMR software (rev. 941001).
Analytical thin layer chromatography (TLC) was performed
on Kiselgel 60 F254 plates (Merck), while preparative TLC was
performed on precoated PLC plates, silica gel 60F-254, 2 mm.
Column chromatography was performed on SiO2 (Matrex LC-
gel: 60A, 35-70 MY; Grace), using a mixture of ethyl acetate
and heptane as an eluent. 6-Methylthioflavone (2a ) was
prepared according to ref 19. 2-Acetyl-4-methylphenol was
prepared from p-cresol by esterification followed by a Fries
rearrangement according to ref 18. Compounds 1a , 1c, 1d , 1e,
1f, 1h , 1i, 1r , and 3 (>90% pure) were purchased from
Extrasynthese S.A. (Lyon, France).
2-Ace t yl-1-(4-m e t h ylb e n zoyloxy)-4-m e t h ylb e n ze n e
(Sch em e 1; I lea d in g to 1k ). The compound was prepared
according to the procedure described for I leading to 1j. The
crude product was recrystallized from methanol to give white
crystals and purified by flash chromatography (heptane:ethyl
acetate, 40:1). The yield was 90%: mp 70-73 °C; 1H NMR
(CDCl3) 2.44 (s, 3H), 2.46 (s, 3H), 2.54 (s, 3H), 7.12 (d, 1H, J
) 8.2), 7.39 (ddd, 1H, J ) 8.2, 2.3, 0.7), 7.42 (dd, 1H, J ) 7.9,
7.6), 7.47 (d, 1H, J ) 8.3), 7.67 (d, 1H, J ) 2.0), 8.02 (d, 1H, J
) 8.3), 8.04 (s, 1H); 13C NMR (CDCl3) 21.3, 21.8, 30.3, 124.0,
127.9, 129.0, 129.6, 131.0, 131.2, 131.4, 134.5, 135.0, 136.4,
139.0, 147.7, 165.9, 198.2; MS m/z (% rel int) 268.1091 (M+,
26, C17H16O3 requires 268.1099), 150 (26%), 135 (29%), 119
(100%), 91 (22%).
1-(3-Meth ylp h en yl)-3-[2-(1-h yd r oxy-4-m eth ylp h en yl)]-
p r op a n e-1,3-d ion e (Sch em e 1; II lea d in g to 1k ). The
compound was prepared according to the procedure described
for II leading to 1j. The crude product ware crystallized from
ethanol to give yellow crystals. The mother liquor was purified
with flash chromatography (heptane:ethyl acetate, 40:1). The
yield was 73%, mp 91-93 °C. Only the enol form was
obtained: 1H NMR (CDCl3) 2.37 (s, 3H), 2.47 (s, 3H), 6.83 (s,
1H), 6.93 (d, 1H, J ) 8.5), 7.30 (dd, 1H, J ) 8.5, 2.1), 7.38-
7.41 (m, 2H), 7.57 (d, 1H, J ) 1.5), 7.76 (d, 1H, J ) 7.8), 7.77
(s, 1H), 11.95 (s, 1H); 13C NMR (CDCl3) 21.1, 21.9, 92.7, 119.0,
119.0, 124.5, 127.8, 128.6, 128.7, 129.1, 133.6, 134.1, 137.3,
139.0, 160.8, 178.2, 196.0; MS m/z (% rel int) 268.1091 (M+,
67, C17H16O3 requires 268.1099), 135 (27%), 119 (100%), 91
(26%).
2-Ac e t y l-1-(2-n it r o b e n zo y lo x y )-4-m e t h y lb e n ze n e
(Sch em e 1; I lea d in g to 1j). 2-Nitrobenzoyl chloride (9.3
g, 50 mmol) was added dropwise to a stirred solution of
2-acetyl-4-methylphenol (5 g, 33 mmol) in 7 mL of pyridine
over a period of 10 min. After warming to 50 °C and stirring
for 20 min the solution was poured into a mixture of 100 g of
ice and 200 mL of 1 M hydrochloric acid. The crude product
was collected by filtration, washed with water, and recrystal-
lized from acetone to give white crystals (yield 71%): mp 139-
140 °C; 1H NMR (CDCl3) 2.45 (s, 3H), 2.58 (s, 3H), 7.29 (d,
1H, J ) 8.0), 7.45 (d, 1H, J ) 8.2), 7.67 (brs, 1H), 7.71 (dd,
1H, J ) 8.2, 7.5), 7.82 (dd, 1H, J ) 8), 8.09 (d, 1H, J ) 8.1),
8.12 (d, 1H, J ) 7.6); 13C NMR (CDCl3) 21.4, 29.6, 123.8, 124.5,
128.4, 130.5, 131.4, 132.2, 134.2, 134.9, 137.1, 146.7, 147.7,
164.7, 198.1; MS m/z (% rel int) 299.0798 (M+, 5, C16H13NO5
requires 299.0794), 150 (100%).
1-(2-Nitr oph en yl)-3-[2-(1-h ydr oxy-4-m eth ylph en yl)]pr o-
p a n e-1,3-d ion e (Sch em e 1; II lea d in g to 1j). 2-Acetyl-1-(2-
nitrobenzoyloxy)-4-methylbenzene (6 g, 20.1 mmol) was dis-
solved in 25 mL of pyridine and heated to 50 °C. Finely
powdered potassium hydroxide (1.6 g, 28.1 mmol) was added
in small portions over a period of 20 min, and stirring was
continued for an additional 15 min at 50 °C. After cooling the
mixture to room temperature and addition of 30 mL of 10%
acetic acid in water, the precipitate formed was collected by
filtration and recrystallized from acetone to give yellow crystals
(yield 89%), mp 133-135 °C. In chloroform, the product is
present as a mixture of the keto (K) and enol (E) forms, 1 to 4:
1H NMR (CDCl3) 2.31 (s, 3H, (E)), 2.33 (s, 3H, (K)), 4.57 (s,
2H, (K)), 6.55 (s, 1H, (E)), 6.90 (d, 1H, J ) 8.7, (K)), 6.92 (d,
1H, J ) 8.5, (E)), 7.31 (dd, 1H, J ) 8.5, 1.9 (E)), 7.32 (m, 1H,
(K)), 7.45 (d, 1H, J ) 1.4 (E)), 7.55 (d, 1H, J ) 1.3 (K)), 7.61
(dd, 1H, J ) 7.7, 1.3 (K)), 7.64 (ddd, 1H, J ) 8.2, 7.9, 1.3 (K)),
7.65 (ddd, 1H, J ) 8.1, 7.8, 1.2 (E)), 7.67 (dd, 1H, J ) 8.0, 1.1
(E)), 7.69 (ddd, 1H, J ) 8.1, 8.0, 0.8 (K)), 7.76 (ddd, 1H, J )
7.9, 7.7, 0.9 (K)), 7.96 (dd, 1H, J ) 7.8, 0.8 (E)), 8.18 (dd, 1H,
J ) 8.2, 0.9 (K)), 11.60 (s, 1H, (K)), 11.69 (s, 1H, (E)); 13C NMR
3′,6-Dim eth ylfla von e (1k ). The compound was prepared
according to the procedure described for 1j. The crude product
was recrystallized from acetone to give white crystals (yield
98%): mp 135-137 °C; 1H NMR (CDCl3) 2.47 (s, 3H), 2.48 (s,
3H), 6.81 (s, 1H), 7.36 (d, 1H, J ) 7.3), 7.42 (dd, 1H, J ) 8),
7.49 (d, 1H, J ) 8.4), 7.53 (dd, 1H, J ) 8.6, 2.0), 7.74 (d, 2H,
J ) 8.1), 8.03 (s, 1H); 13C NMR (CDCl3) 21.4, 22.0, 107.9, 118.3,
123.9, 124.1, 125.5, 127.3, 129.4, 132.3, 132.8, 135.4, 135.6,
139.2, 155.0, 163.9, 179.1; MS m/z (% rel int) 250.0996 (M+,
100, C17H14O2 requires 250.0994), 222 (17%), 134 (47%).
2-Ac e t y l-1-(3-n it r o b e n zo y lo x y )-4-m e t h y lb e n ze n e
(Sch em e 1; I lea d in g to 1l). The compound was prepared
according to the procedure described for I leading to 1j. The
crude product was recrystallized from acetone/H2O to give
white crystals (yield 93%): mp 108-110 °C; 1H NMR (CDCl3)
2.47 (s, 3H), 2.56 (s, 3H), 7.15 (d, 1H, J ) 8.2), 7.44 (dd, 1H,
J ) 8.2, 2.2), 7.71 (d, 1H, J ) 1.8), 7.75 (dd, 1H, J ) 8), 7.51
(ddd, 1H, J ) 8.3, 2.3, 1.6), 8.54 (ddd, 1H, J ) 7.8, 1.6, 1.5),
9.05 (dd, 1H, J ) 2); 13C NMR (CDCl3) 21.4, 29.6, 124.0, 125.7,
128.4, 130.3, 130.3, 131.6, 131.8, 134.8, 136.4, 137.0, 147.1,
148.8, 164.0, 197.8; MS m/z (% rel int) 299.0786 (M+, 38,
C16H13NO5 requires 299.0794), 150 (100%), 104 (16%).

SAR and Molecular Modeling of Flavonoid Binding
J ournal of Medicinal Chemistry, 1999, Vol. 42, No. 21 4349
1-(3-Nitr oph en yl)-3-[2-(1-h ydr oxy-4-m eth ylph en yl)]pr o-
p a n e-1,3-d ion e (Sch em e 1; II lea d in g to 1l). The compound
was prepared according to the procedure described for II
leading to 1j. The crude product was recrystallized from
acetone to give yellow crystals (yield 96%), mp 165-167 °C.
Only the keto form was obtained: 1H NMR (CDCl3) 2.39 (s,
3H), 6.91 (s, 1H), 6.95 (d, 1H, J ) 8.5), 7.34 (dd, 1H, J ) 8.5,
2.1), 7.59 (d, 1H, J ) 1.5), 7.72 (dd, 1H, J ) 8), 8.30 (d, 1H,
J ) 7.9), 8.41 (ddd, 1H, J ) 8.2, 2.3, 1.0), 8.78 (dd, 1H, J ) 2),
11.76 (S, 1H); 13C NMR (CDCl3) 21.0, 93.7, 118.7, 119.2, 122.1,
126.9, 128.7, 129.0, 130.4, 132.8, 136.0, 138.2, 149.0, 161.1,
requires 326.0539), 309 (45%), 263 (27%), 234 (16%), 149 (14%),
133 (25%), 84 (100%), 66 (100%).
6-Meth yl-3′-n itr oth iofla von e (2b). 1 g of phosphorus
pentoxide (7.05 mmol) and 0.5 mL of 100% phosphoric acid
were mixed in a 25-mL round-bottomed flask. The mixture was
heated under nitrogen to 80-90 °C for 15 min, whereafter
thiocresol (68 mg, 0.55 mmol) and ethyl m-nitrobenzoylacetate
(135 mg, 0.57 mmol) were added. After 90 min the reaction
mixture was poured onto ice, and the water phase was
extracted with methylene chloride. The compound was purified
by flash chromatography (heptane:ethyl acetate, 9:1) to give
white crystals (yield 24%): mp 192-194 °C; 1H NMR (CDCl3)
2.53 (s, 3H), 7.27 (s, 1H), 7.52 (ddd, 1H, J ) 8.2, 2.0, 0.4), 7.60
(d, 1H, J ) 8.2), 7.73 (dd, 1H, J ) 8), 8.02 (ddd, 1H, J ) 7.8,
1.8, 1.0), 8.37 (m, 1H), 8.38 (m, 1H), 8.57 (dd, 1H, J ) 2); 13C
NMR (CDCl3) 21.8, 122.5, 124.9, 125.6, 126.9, 128.9, 130.9,
131.0, 133.1, 133.1, 133.9, 134.3, 138.8, 139.2, 150.2, 181.0;
MS m/z (% rel int) 297.0462 (M+, 100, C16H11NO3S requires
297.0460), 269 (27%), 223 (13%), 208 (16%), 150 (15%), 121
(13%).
174.1, 196.7; MS m/z (% rel int) 299.0787 (M+, 100, C16H13
-
NO5 requires 299.0794), 282 (18%), 177 (9%), 150 (65%), 134
(61%), 104 (12%).
6-Meth yl-3′-n itr ofla von e (1l). The compound was pre-
pared according to the procedure described for 1j. The crude
product was recrystallized from acetone to give white crystals
1
(yield 42%): mp 221-223 °C; H NMR (CDCl3) 2.51 (s, 1H),
6.90 (s, 1H), 7.55 (d, 1H, J ) 8.5), 7.59 (dd, 1H, J ) 8.6, 2.1)
7.76 (dd, 1H, J ) 8), 8.05 (m, 1H), 8.23 (ddd, 1H, J ) 7.9, 1.7,
1.0), 8.41 (ddd, J ) 8.2, 2.2, 1.0), 8.83 (dd, 1H, J ) 2.1, 1.9);
13C NMR (CDCl3) 21.4, 109.1, 109.1, 118.4, 121.7, 124.0, 125.6,
125.6, 126.3, 130.7, 123.2, 134.2, 136.0, 136.3, 160.8, 178.6;
MS m/z (% rel int) 281.0692 (M+, 100, C16H11NO4 requires
281.0688), 253 (13%), 325 (8%), 134 (33%), 106 (8%).
Bin d in g Stu d ies. [3H]Flumazenil (specific activity 87.5 Ci/
mol) was obtained from DuPont, New England Nuclear.
Protein concentration was determined using the DC protein
assay from Bio-Rad Laboratories. All compounds tested were
dissolved in DMSO in a stock solution of 100 µM and diluted
into 50% ethanol prior to the binding assay.
2-Ac e t y l-1-(4-n it r o b e n zo y lo x y )-4-m e t h y lb e n ze n e
(Sch em e 1; I lea d in g to 1o). The compound was prepared
according to the procedure described for I leading to 1j. The
crude product was recrystallized from acetone/H2O to give
white crystals (yield 56%): mp 109-111 °C; 1H NMR (CDCl3)
2.46 (s, 3H), 2.56 (s, 3H), 7.15 (d, 1H, J ) 8.2), 7.44 (dd, 1H,
J ) 8.2, 2.2), 7.71 (d, 1H, J ) 2.0), 8.37 (d, 2H, J ) 8.0), 8.39
(d, 2H, J ) 8.0); 13C NMR (CDCl3) 21.4, 29.6, 124.0, 124.2,
130.3, 131.6, 131.8, 134.8, 135.4, 137.0, 147.1, 151.3, 164.2,
197.8; MS m/z (% rel int) 299.0795 (M+, 15, C16H13NO5 requires
299.0794), 150 (100%), 120 (9%), 104 (33%), 76 (20%).
Ra t Br a in Mem br a n e P r ep a r a tion for Bin d in g Stu d -
ies. The radioreceptor binding assay was carried out as
previously described.23 In brief, male Wistar rats (weighing
about 200 g) were decapitated and the brains excised. The
cerebral cortex (∼260 mg) was removed and homogenized in
10 mL of Tris-citrate buffer (50 mM, pH 7.1) using an Ultra-
Turrax homogenizer, and the homogenate was centrifuged
(20000g at 0-4 °C for 10 min). The pellet was resuspended in
15 mL of Tris-citrate buffer (50 mM, pH 7.1). This washing
procedure of the membranes was repeated twice, and the final
pellet was resuspended in Tris-citrate buffer (50 mM, pH 7.1)
to give a final concentration of 2 mg of original tissue/mL and
used for [3H]flumazenil binding assays.
[3H]F lu m a zen il Bin d in g Assa y. Each assay consists of
0.5 mL of rat cortical membrane suspensions (2 mg/mL), 25
µL containing test compound, and 25 µL of [3H]flumazenil
(final concentration 0.3-0.5 nM). For saturation binding
experiments, 6-8 different concentrations of [3H]flumazenil
(final concentration 0.05-10 nM) were used. Nonspecific
binding was obtained by adding diazepam (final concentration
10 µM) to separate samples. All samples was incubated on an
ice bath at 0-4 °C for 40 min, followed by filtration using a
semiautomatic cell harvester (Skatron Instruments) to sepa-
rate bound from free radioactive ligand. Filters were washed
with 7 mL of ice-cold Tris-citrate buffer, and the radioactivity
on the filters was determined by conventional liquid scintil-
lation counting. Nonspecific binding represented less than 10%
of the total binding. All assays were done in duplicate.
1-(4-Nitr oph en yl)-3-[2-(1-h ydr oxy-4-m eth ylph en yl)]pr o-
p a n e-1,3-d ion e (Sch em e 1; II lea d in g to 1o). The compound
was prepared according to the procedure described for II
leading to 1j. The crude product was recrystallized from
acetone to give orange crystals (yield 53%), mp 193-195 °C.
Only the enol form was obtained: 1H NMR (CDCl3) 2.37 (s,
3H), 6.91 (s, 1H), 6.96 (d, 1H, J ) 8.5), 7.35 (dd, 1H, J ) 8.5,
2.2), 7.57 (d, 1H, J ) 1.3), 8.13 (d, 2H, J ) 9.1), 8.36 (d, 2H,
J ) 9.1), 11.77 (s, 1H); 13C NMR (CDCl3) 21.0, 94.6, 118.8,
119.2, 124.4, 128.1, 128.7, 128.9, 138.2, 139.9, 161.1, 173.9,
196.7; MS m/z (% rel int) 299.0794 (M+, 85, C16H13NO5 requires
299.0794), 282 (16%), 177 (8%), 150 (100%), 134 (82%), 104
(20%).
6-Meth yl-4′-n itr ofla von e (1o). The compound was pre-
pared according to the procedure described for 1j. The crude
product was recrystallized from acetone to give white crystals
1
(yield 83%): mp 276-278 °C; H NMR (CDCl3) 2.43 (s, 1H),
7.21 (s, 1H), 7.68 (dd, 1H, J ) 8.9, 2.1), 7.72 (d, 1H, J ) 8.6),
7.85 (m, 1H), 8.37 (s, 4H); 13C NMR (CDCl3) 21.4, 109.9, 119.4,
125.0, 125.0, 128.6, 136.4, 136.6, 138.1, 149.9, 154.9, 160.9,
177.1; MS m/z (% rel int) 281.0685 (M+, 100, C16H11NO4
requires 281.0688), 253 (12%), 235 (8%), 134 (34%).
The GABA ratio was defined by dividing the Ki value of the
test compound without adding GABA to the binding assay by
the Ki value of the test compound in the presence of GABA
(100 µM) in the binding assay. The Ki value was calculated by
the equation: Ki ) IC50/(1 + [3H]flumazenil]/KD). The KD value
for [3H]flumazenil binding under the assay conditions used in
this study is 1.1 ( 0.2 nM (mean ( SEM of five separate
experiments).
6-Meth yl-3′,5-d in itr ofla von e (1q) (Sch em e 3). 100 mg
of 6-methylflavone dissolved in 3 mL of dry methylene chloride
was added to a mixture of 5 mL of 65% nitric acid with an
excess of phosphorus pentaoxide in an ice-cooled 50-mL round-
bottomed flask. The mixture was allowed to reach room
temperature and was quenched with 6 M KOH after 15 min.
After neutralization and extraction with methylene chloride,
1q was isolated from the crude product mixture by preparative
TLC (mobile phase toluene:methyl tert-butyl ether, 5:1): 1H
NMR (DMSO) 2.06 (s, 3H), 5.74 (s, 3H), 7.88 (dd, 1H, J ) 8),
7.94 (d, 1H, J ) 8.8), 8.08 (d, 1H, J ) 8.7), 8.44 (ddd, 1H, J )
8.2, 2.3, 0.9), 8.56 (ddd, 1H, J ) 7.9, 1.8, 1.0), 8.86 (dd, 1H,
J ) 2); 13C NMR (CDCl3) 16.2, 112.3, 114.9, 121.4, 126.0, 126.8,
127.8, 132.7, 133.8, 135.1, 138.3, 146.6, 148.2, 154.8, 162.9,
174.6; MS m/z (% rel int) 326.0547 (M+, 80, C16H10N2O6
Com p u ta tion a l Meth od s. Geometry optimizations and
conformational analyses of the flavones were performed by
using the molecular mechanics program MM3(92) developed
by Allinger and co-workers.24,25 In addition to the standard
MM3(92) parameter set, the following three sets of torsional
parameters were used: (Od)C-Csp2-Csp2-O V1 ) V3 ) 0.0,
V2 ) 10.0; (Od)C-Csp2-Csp2-N(O2) V1 ) V3 ) 0.0, V2 )
10.0; Csp3-Csp2-Csp2-N(O2) V1 ) V3 ) 0.0, V2 ) 10.0.
These parameters are required for the calculations to run.
However, the Csp2-Csp2 bonds belong to π-systems which
stay essentially planar in all calculations. Thus, the precise

4350 J ournal of Medicinal Chemistry, 1999, Vol. 42, No. 21
Dekermendjian et al.
(15) Cox, E. D.; Diaz-Arauzo, H.; Huang, Q.; Reddy, M. S.; Ma, C.;
Harris, B.; McKernan, R.; Skolnick, P.; Cook, J . M. Synthesis
and evaluation of analogues of the partial agonist 6-(propyloxy)-
4-(methoxymethyl)-â-carboline-3-carboxylic acid ethyl ester (6-
PBC) and the full agonist 6-(benzyloxy)-4-(methoxymethyl)-â-
(carboline-3-carboxylic acid ethyl ester (Zk 93423) at wild type
and recombinant GABAA receptors. J . Med. Chem. 1998, 41,
2537-2552.
(16) Da Settimo, A.; Primofiore, G.; Da Settimo, F.; Marini, A. M.;
Novellino, E.; Greco, G.; Martini, C.; Giannaccini, G.; Lucacchini,
A. Synthesis, structure-activity relationships, and molecular
modelling studies of N-(indol-3-ylglyoxylyl)-benzylamine deriva-
tives acting at the benzodiazepine receptor. J . Med. Chem. 1996,
39, 5083-5091.
(17) Da Settimo, A.; Primofiore, G.; Da Settimo, F.; Marini, A. M.;
Novellino, E.; Greco, G.; Gesi, M.; Martini, C.; Giannaccini, G.;
Lucacchini, A. N′-Phenyl-3-ylglyoxylohydrazide derivatives: Syn-
thesis, structure-activity relationships, molecular modelling
studies, and pharmacological action on brain benzodiazepine
receptors. J . Med. Chem. 1998, 41, 3821-3830.
values of the torsional constants do not significantly affect the
results. Since π-atom pair parameters for (Od)C‚‚‚S are not
included in the MM3(92) parameter set, geometry optimiza-
tions and conformational analyses of the thioflavones 2a and
2b were performed by using the MM2(91) program.26
For the calculations of molecular electrostatic potentials, the
HF/6-31G* basis set was employed with geometries calculated
by the same set. The calculations were performed by using
the SPARTAN program version 5.0.3.27
Ack n ow led gm en t. This work was supported by
grants from the Danish Medical Research Council,
Swedish Natural Science Research Council, and Lund-
beck Foundation.
Refer en ces
(1) Zhang, W.; Koehler, K. F.; Zhang, P.; Cook J . M. Development
of a comprehensive pharmacophore model for the benzodiazepine
receptor. Drug Des. Dev. 1995, 12, 193-248.
(2) Kaupmann, K.; Huggel, K.; Heid, J .; Flor, P. J .; Bischoff, S.;
Mickel, S. J .; McMaster, G.; Angst, C.; Bittiger, H.; Froestl, W.;
Bettler, B. Expression cloning of GABAB receptors uncovers
similarity to metabotropic glutamate receptors. Nature 1997,
386, 239-246.
(3) Rabow, L. E.; Russek, S. H.; Farb, D. H. From ion currents to
genomic analysis: Recent advances in GABAA receptor research.
Synapse 1995, 21, 189-274.
(4) J ohnston, G. A. R. GABAC receptors: relatively simple transmit-
ter-gated ion channels. TiPS 1996, 17, 319-323.
(5) Barnard, E. A.; Skolnick, P.; Olsen, R. W.; Mohler, H.; Sighart,
W.; Biggio, G.; Braestrup, C.; Bateson, A. N.; Langer, S. Z.
International union of pharmacology. XV. Subtypes of γ-ami-
nobutyric acidA receptors: Classification of the basis of subunit
structure and receptor function. Pharmacol. Rev. 1998, 50, 291-
313.
(18) Tietze, L. F.; Eicher, Th. Reactions and synthesis in the organic
chemistry laboratory; University Science Books Co., 1989; pp 122,
139-140.
(19) Bossert, F. U¨ ber eine neue thiochromon-synthese. Ann. Chem.
1964, 680, 41-51.
(20) Braestrup, C.; Honore´, T.; Nielsen, M.; Petersen, E. N.; J ensen,
L. H. Ligands for benzodiazepine receptors with positive and
negative efficacy. Biochem. Pharmacol. 1984, 33, 859-862.
(21) Allen, M. S.; Tan, Y. C.; Trudell, M. L.; Narayanan, K.; Schindler,
L. R.; Martin, M. J .; Schultz, C.; Hagen, T. J .; Koehler, K. F.;
Codding, P. W.; Skolnick, P.; Cook, J. M. Synthetic and computer-
assisted analyses of the pharmacophore for the benzodiazepine
receptor inverse agonists site. J . Med. Chem. 1990, 33, 2343-
2357.
(22) Allen, M. S.; LaLoggia, A. J .; Dorn, L. J .; Martin, M. J .;
Constantino, G.; Hagen, T. J .; Koehler, K. F.; Skolnick, P.; Cook,
J . M. Predicitive binding of â-carboline inverse agonists and
antagonists via the CoMFA/GOLPE approach. J . Med. Chem.
1992, 35, 4001-4010.
(6) Whiting, P. J .; McKernan, R. M.; Wafford, K. A. Structure and
pharmacology of vertebrate GABAA receptor subtypes. In Inter-
national Review of Neurobiology, 38th ed.; Bradley, R. J ., Harris,
R. A., Eds.; Academic Press: New York, 1995; pp 95-138.
(7) Sieghart, W. Structure and pharmacology of γ-aminobutyric
acidA receptor subtypes. Pharmacol. Rev. 1995, 47, 181-234.
(8) Doble, A.; Martin, I. L. The GABAA/ benzodiazepine receptor as
a target for psychoactive drugs; Springer-Verlag: Heidelberg,
Germany, 1996.
(9) Gardner, C. R.; Tully, W. R.; Hedgecock, J . R. The rapidly
expanding range of neuronal benzodiazepine receptor ligands.
Prog. Neurobiol. 1993, 40, 1-61.
(10) Ai, J .; Dekermendjian, K.; Sterner, O.; Nielsen, M.; Witt, M. R.
Compounds isolated from medicinal plants as ligands for ben-
zodiazepine receptors. Recent Res. Dev. Phytochem. 1997, 1,
365-385.
(11) Medina, J . H. M.; Viola, H.; Wolfman, C.; Marder, M.; Wasowskj,
C.; Calvo, D.; Paladini, A. C. Neuroactive flavonoids; New ligands
for the benzodiazepine receptors. Phytomed 1998, 5, 235-243.
(12) Harborne, J . B., Ed. The Flavonoids, advances in research since
1986; Chapman and Hall: London, 1993.
(13) Diaz-Arauzo, H.; Evoniuk, G. E.; Skolnick, P.; Cook, J . M. The
agonist pharmacophore of the benzodiazepine receptor. Synthe-
sis of a selective anticonvulsant/anxiolytic. J . Med. Chem. 1991,
34, 1754-1756.
(14) Diaz-Arauzo, H.; Koehler, K. F.; Hagen, T. J .; Cook, J . M.
Synthetic and computer assisted analysis of the pharmacophore
for agonists at benzodiazepine receptors. Life Sci. 1991, 49, 207-
216.
(23) Dekermendjian, K.; Ai, J .; Nielsen, M.; Sterner, O.; Shan, R.;
Witt, M. R. Characterisation of the furanocoumarin phellopterin
as a rat brain benzodiazepine receptor partial agonist in vitro.
Neurosci. Lett. 1996, 219, 151-154.
(24) Allinger, N. L.; Yuh, H. Y.; Lii, J . H. The MM3 force field for
hydrocarbons. J . Am. Chem. Soc. 1989, 111, 8551-8666.
(25) Allinger, N. L.; Li, F.; Yan, L.; Tai, J . C. Molecular Mechanics
(MM3) calculations on conjugated hydrocarbons J . Comput.
Chem. 1990, 11, 868-95.
(26) Burkert, U.; Allinger, N. L. Molecular Mechanics; ACS Mono-
graph 177; American Chemical Society: Washington, DC, 1982.
(27) SPARTAN SGI version 5.0.3, Wavefunction Inc., 18401 von
Karman, Suite 370, Irvine, CA 92715.
(28) Marder, M.; Viola, H.; Wasowskj, C.; Wolfman, C.; Waterman,
P. G.; Cassels, B. K.; Medina, J . H.; Paladini, A. C. 6-Bromofla-
vone, a high affinity ligand for the central benzodiazepine
receptors is a member of a family of active flavonoids. Biochem.
Biophys. Res. Commun. 1996, 223, 284-389.
(29) Wolfman, C.; Viola, H.; Marder, M.; Wasowskj, C.; Ardenghi,
P.; Izquierdo, I.; Paladini, A. C.; Medina, J . H. Anxioselective
properties of 6,3′-dinitroflavone, a high-affinity benzodiazepine
receptor ligand. Eur. J . Pharmacol. 1996, 318, 23-30.
(30) Marder, M.; Viola, H.; Wasowskj, C.; Wolfman, C.; Waterman,
P. G.; Medina, J . H.; Paladini, A. C. 6,3′-Dinitroflavone, a novel
high affinity ligand for the benzodiazepine receptor with potent
anxiolytic properties. Bioorg. Med. Chem. Lett. 1995, 5, 2717-
2720.
J M991010H