4-Quinolone derivatives: High-affinity ligands at the benzodiazepine site of brain GABAA receptors. Synthesis, pharmacology, and pharmacophore modeling

Article abstract of DOI:10.1021/jm058057p

The 3-ethoxycarbonyl-4-quinolone compound 1 has previously been identified via a database search as an interesting lead compound for ligand binding at the benzodiazepine site of GABA_A receptors (Kahnberg et al. J. Mol. Graphics Modelling 2004, 23, 253-261). Pharmacophore-guided optimization of this lead compound yielded a number of high-affinity ligands for the benzodiazepine site including compounds 20 and 23-25 displaying sub-nanomolar affinities. A few of the compounds have been tested on the α₁β ₂γ_2S and α₃β ₂γ_2S GABA_A receptor subtypes, and two of the compounds (5 and 19) display selectivity for α₁-versus α₃-containing receptors by a factor of 22 and 27, respectively. This selectivity for α₁β₂γ_2S is in the same range as that for the well-known α₁ subunit selective compound zolpidem.

Full text of DOI:10.1021/jm058057p

2526
J. Med. Chem. 2006, 49, 2526-2533
4-Quinolone Derivatives: High-Affinity Ligands at the Benzodiazepine Site of Brain GABAA
Receptors. Synthesis, Pharmacology, and Pharmacophore Modeling
Erik Lager,† Pierre Andersson,† Jakob Nilsson,† Ingrid Pettersson,| Elsebet Østergaard Nielsen,§ Mogens Nielsen,‡
Olov Sterner,† and Tommy Liljefors*,‡
Department of Organic Chemistry, Lund UniVersity, P.O.B. 124, SE-221 00 Lund, Sweden, Department of Medicinal Chemistry,
The Danish UniVersity of Pharmaceutical Sciences, UniVersitetsparken 2, DK-2100 Copenhagen, Denmark, NeuroSearch A/S,
DK-2750 Ballerup, Denmark, and NoVo Nordisk A/S, NoVo Nordisk Park, DK-2760 Ma˚løV, Denmark
ReceiVed December 12, 2005
The 3-ethoxycarbonyl-4-quinolone compound 1 has previously been identified via a database search as an
interesting lead compound for ligand binding at the benzodiazepine site of GABAA receptors (Kahnberg et
al. J. Mol. Graphics Modelling 2004, 23, 253-261). Pharmacophore-guided optimization of this lead
compound yielded a number of high-affinity ligands for the benzodiazepine site including compounds 20
and 23-25 displaying sub-nanomolar affinities. A few of the compounds have been tested on the R1â2γ2S
and R3â2γ2S GABAA receptor subtypes, and two of the compounds (5 and 19) display selectivity for R1-
versus R3-containing receptors by a factor of 22 and 27, respectively. This selectivity for R1â2γ2S is in the
same range as that for the well-known R1 subunit selective compound zolpidem.
Introduction
antagonists, and inverse agonists all share the same binding
pocket. Recent research with synthetic flavone derivatives led
to a further development of the pharmacophore model.11,12 The
model was subsequently used for identification of novel lead
structures by a 3D database search using the program Catalyst.14
One of the hits identified was the 4-quinolone 1 (Table 1) which
displayed the highest affinity of the purchased and tested
compounds (Ki ) 122 nM).14 In the present work we have
performed pharmacophore-guided optimization of this lead
structure. In addition to compounds derived from 1, a few
pyrazolopyrimidinones based on another hit obtained in the
database search mentioned above were also synthesized and
tested.
γ-Aminobutyric acid, GABA, is one of the major inhibitory
neurotransmitters in the central nervous system and exerts its
physiological effect by binding to three different receptor types
in the neuronal membrane: GABAA, GABAB, and GABAC
receptors.1 GABAB belongs to the family of G-protein-coupled
receptors and act via intracellular second messengers.2 GABAA
3
and GABAC4 are coupled directly to anion channels and cause
an increase in chloride permeability of the membrane. Mam-
malian brain GABAA receptors have been shown to be hetero-
pentameric assemblies of protein subunits from seven different
classes with multiple isoforms (R1-6, â1-4, γ1-4, δ, ꢀ, π, and
F1-F3), each encoded by different genes.5 Most GABAA receptors
are composed of two R-, two â-, and one γ-subunit. It is believed
that receptors with different subtype composition can give rise
to different physiological effects, e.g. R1-containing receptors
are implicated in sedation and anterograde amnesia, and R2-,
R3-, or possibly R5-containing receptors in anxiolytic activity.6,7
Benzodiazepines, â-carbolines, barbiturates, ethanol, and certain
steroids are the best known ligands that allosterically modify
the chloride channel gating effect of GABA on GABAA
receptors.1
Results and Discussion
Chemistry. The structures of the compounds investigated in
this study are shown in Table 1. The 4-quinolones were
synthesized according to procedures previously described15 by
a condensation of an appropriate aniline with either diethyl
ethoxymethylenemalonate or diethyl acetylmalonate followed
by cyclization of the intermediate upon reflux in diphenyl ether
(Scheme 1). The alkyl group (R1) was varied by transesterifi-
cation using different alcohols. Initially p-toluenesulfonic acid
was used as a catalyst for this reaction, but the corresponding
acid 13 was obtained as a side product which hampered the
purifying process. The use of titanium(IV) isopropoxide as a
catalyst16 gave cleaner reactions without the need to use dried
solvents. This procedure also worked well with secondary
alcohols and primary amines.
The most important drugs in clinical use for GABAA receptor
modulating purposes are the benzodiazepines, which mainly give
anxiolytic, anticonvulsant, muscle relaxant, and sedative-
hypnotic effects.8 Many classes of compounds bind to the same
binding site as the benzodiazepines, e.g. â-carbolines, triazol-
opyridazines, pyrazoloquinolinones, cyclopyrrolones, pyrido-
diindoles, quinolines, and flavones.9-12
Cook and co-workers13 have developed a comprehensive
pharmacophore model for the benzodiazepine receptor based
on structure-activity relationship studies of 136 different ligands
from 10 structurally different classes of compounds. The model
was developed assuming that benzodiazepine receptor agonists,
The pyrazolopyrimidinones were synthesized according to
previously described procedures,17 utilizing a one-pot procedure
which includes the condensation of 3-aminopyrazole and
ethoxymethylenemalonate followed by cyclization of the inter-
mediate. The alkyl chain was subsequently varied, by transes-
terification with n-propanol using titanium(IV) isopropoxide as
a catalyst (Scheme 2).
Receptor Binding. Table 1 shows that the tested 4-quinolones
inhibit the specific binding of 3H-flumazenil with a range of Ki
values from low or sub-nM (3, 19, 20, 23-25) to µM
concentrations. Compound 25 displays the highest affinity in
* To whom correspondence should be sent. Tel. +45-35 30 65 05.
E-mail: tl@dfuni.dk. Fax: +45-35 30 60 40.
† Lund University.
‡ The Danish University of Pharmaceutical Sciences.
§ NeuroSearch A/S.
| Novo Nordisk A/S.
10.1021/jm058057p CCC: $33.50 © 2006 American Chemical Society
Published on Web 03/14/2006

4-Quinolone Brain GABAA Receptor Ligands
Journal of Medicinal Chemistry, 2006, Vol. 49, No. 8 2527
3
Table 1. Ki Values of 4-Quinolones and Pyrazolopyrimidinones Tested on H-Flumazenil Binding in Vitro to Rat Cortical Membranes
compd
R1
R2
R3
X
Ki value (nM)a
1
2
3
4
5
6
7
8
CH2CH3
CH2CH3
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
CH3
CH3
H
H
H
CF3
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
NH
NH
NH
122 ( 46
20 ( 5
CH2CH3
CH2CH3
CH2CH3
CH2CH3
CH2CH3
CH2CH3
CH2CH3
CH2CH3
CH2CH3
CH2CH3
CH2CH3
CH2CH3
H
CH2CH2CH3
CH2CH2CH2CH3
CH2CH2CH(CH3)2
CH2CH(CH3)CH2CH3
CH2CH2CH2CH2CH3
CH2CH(CH2CH3)2
cyclopentyl
1.8 ( 0.3
13 ( 4
28 ( 8
28 ( 6
35 ( 8
92 ( 13
9
19 ( 4
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
CH(CH3)2
214 ( 10
295 ( 17
2600 ( 570
208 ( 53
78 ( 10
CH(CH3)CH2CH(CH3)2
CH(CH2CH3)2
H
CH2CH3
CH2CH2CH2CH3
CH2CH3
CH2CH3
CH2CH3
CH2CH3
CH2CH2CH3
CH2CH3
H
Br
54 ( 45
16 ( 5
17 ( 0.5
15 ( 4
1.4 ( 0.2
0.17 ( 0.01
6850 ( 1900
4200 ( 560
0.26 ( 0.03
0.54 ( 0.06
0.048 ( 0.006
>10000
CH2CH2CH3
CH(CH3)2
CH2C6H5
CH2C6H5
CH2CH3
Br
CH2CH3
CH2CH3
CH2C6H5
CH2CH3
CH2CH2CH3
CH2CH2CH2CH3
CH2CH2CH3
CH2CH3
CH2CH2CH3
CH2CH3
CH3
CH3
H
870 ( 285
>10000
7130 ( 4200
CH2CH2CH3
H
a Each Ki value is the mean ( SD of three determinations.
Scheme 1a
a Reagents and conditions: (a) neat, 130 °C, 2 h; (b) p-TsOH, pentane, reflux (Dean-Stark trap), 24 h; (c) diphenyl ether, reflux, 1 h; (d) p-TsOH or
Ti(O-i-Pr)4, alcohol or amine, reflux overnight.
the present series of compounds with a Ki value of 0.048 nM.
The results show that amide derivatives (23-25) display higher
affinities than the corresponding esters (3, 4, and 20). The
pyrazolopyrimidinones (26-29) inhibited the specific binding
workers13 and previously validated and further refined by us
using the results of structure-activity studies for a series of
flavonoids, 11,12 is shown in Figure 1a. The figure displays the
interactions between the high-affinity flavonoid 5′-bromo-2′-
hydroxy-6-methylflavone (Ki ) 0.9 nM12) and the sites of the
pharmacophore model. H1 and A2 are hydrogen bond donating
and hydrogen bond accepting sites, respectively. H2/A3 repre-
3
of H-flumazenil in µM concentrations.
Fitting to the Pharmacophore Model. The pharmacophore
model for the benzodiazepine site, developed by Cook and co-

2528 Journal of Medicinal Chemistry, 2006, Vol. 49, No. 8
Lager et al.
Scheme 2a
29, conformational analysis of the amides 23-25 indicates that
for these compounds the proposed binding conformation with
respect to carbonyl arrangement is the global energy minimum
conformation. This is due to stabilization of the proposed
bioactive conformation by an intramolecular NH- - -OdC
hydrogen bond. The amides 23-25 display significantly higher
affinities than the corresponding esters 3, 4, and 20 (Table 1).
The differences in conformational energy penalties between the
esters and the amides for adopting the bioactive conformation
is too small (0.1-0.3 kcal/mol) to fully explain these affinity
differences. The major reason for the higher affinities of the
amide compounds compared to the esters is most probably due
to the higher electron density of an amide carbonyl oxygen
compared to that of an ester carbonyl oxygen. This makes the
hydrogen bond interactions with the H1 site stronger for the
amide compounds. It is also possible that the NH group of the
amides have additional interactions with the A3 part of the H2/
A3 site.
a Reagents and conditions: (a) neat, 200 °C, 10 min; (b) Ti(O-i-Pr)4
and n-propanol, reflux, overnight.
sents a bifunctional site with the ability to act as a hydrogen
bond donor as well as an acceptor. L1-L3 are lipophilic sites
and S1-S5 denote regions of steric repulsive ligand-receptor
interactions (receptor essential volumes). The term “interface”
is used to denote a partly lipophilic region in the vicinity of the
5′-position in the flavone series. This region has been proposed
to represent the interface between an R- and a γ- subunit in the
GABAA receptor.12 Figure 1b shows the proposed binding mode
of the lead structure 1 in the pharmacophore model, and Figure
2 displays a superimposition of 5′-bromo-2′-hydroxy-6-meth-
ylflavone and 1 fitted to the pharmacophore model. As shown
in Figure 1b, the ester and the ring carbonyl groups of 1 interact
in a coplanar arrangement with H1 and H2, respectively, and
the NH group forms a hydrogen bond with the A2 site
corresponding to the hydrogen bond involving the 2′-OH group
in the flavones (Figure 1a). The 6-CF3 group of 1 coincides
with the 5′-bromo substituent in the flavone. Substitution in this
position in flavone derivatives by small substituents has
previously been shown to give a significant increase in
affinity.11,12
The proposed bioactive conformation of 1 in Figure 1b does
not correspond to the global energy minimum conformation of
the compound in aqueous solution. Conformational analysis
using the MMFF94s force field and the GB/SA hydration model
and performed as described in the Experimental Section
indicates that the lowest energy conformation of 1 has parallel
carbonyl groups with a 13° twist of the ester group with respect
to the bicyclic ring system. In contrast, the carbonyl groups in
the proposed bioactive conformation point at different directions
as shown in Figure 3. However, the proposed binding confor-
mation of 1 is calculated to be only 0.1 kcal/mol higher in energy
than that of the calculated global energy minimum. All
4-quinolone esters in Table 1 except compounds 21 and 22 can
adopt the proposed binding conformation with low conforma-
tional energy penalties, 0.1-0.3 kcal/mol. For the 2-methyl-
substituted compounds 21 and 22, the energy cost for adopting
the proposed bioactive arrangement of the carbonyl groups is
calculated to be significantly higher, 3.1 kcal/mol. The calculated
lowest energy conformation of these compounds displays a 63°
twist of the ester group with respect to the bicyclic ring. Steric
repulsive interactions between the 2-methyl group and the ester
carbonyl in the proposed bioactive conformation prevent the
carbonyl groups to adopt a coplanar conformation with a low
conformational energy penalty as required by the pharmacophore
model. The calculated conformational energy penalty rational-
izes the reduced affinities of 21 and 22 by a factor of 300
compared to 2 and 16, respectively (Table 1). This is in
accordance with previous conclusions that a planar or close to
planar geometry of the core part of the ligand is required for
high-affinity binding to the benzodiazepine receptor.11,13 For
the pyrazolo-pyrimidinones 26-29, the energy difference
between the lowest energy minimum and the proposed binding
conformation is calculated to be 0.1 kcal/mol. In contrast to
the conformational properties of compounds 1-22 and 26-
3-Carboxamide-4-quinolones have previously in a patent
application been disclosed as GABAA brain receptor ligands.18
Interactions in the Lipophilic L1 and L2 Regions. Figures
1 and 2 clearly demonstrate that the ethyl ester group of lead
structure 1 does not fill out the lipophilic regions L1 and in
particular L2. It should be noted that the 6-methyl group in the
flavone series (Figures 1a) increases the affinity by a factor of
23 and that the corresponding 6-ethyl, 6-propyl, and 6-isopropyl
compounds in the flavone series also have significantly higher
affinities than the 6-unsubstituted parent compound.11,12 There-
fore, a series of alkyl esters with different alkyl chain length
and alkyl branching (3-12) was synthesized and tested in order
to further explore the dimensions and other properties of the
L2 region. The parent compound for this series is the 6-ethyl
compound 2 which displays 6 times higher affinity than the
lead compound 1.
Extending the ester ethyl group in 2 to a propyl group in 3
increases the affinity by a factor of 11 (Table 1), only slightly
lower than the effect on the affinity by a 6-methyl group in the
flavone series mentioned above. As can be inferred from Figure
2, the methyl group of the ester propyl group in 3 coincides
with the 6-methyl group in the flavonoids. Further chain
elongations from propyl to butyl (4) and pentyl (7) result in
affinities similar to that of 2 but to a significant reduction of
affinity compared to the propylester 3. The N-propyl amide 23
also has a somewhat higher affinity than the N-butyl amide 24.
Thus, with respect to alkyl chain length the propyl group is
clearly optimal.
By analyzing the branched alkyl esters (5, 6, 8-12) as alkyl
(methyl or ethyl) substituted ethyl, propyl, or butyl esters, Table
1 shows that despite higher lipophilicities, the branched alkyl
esters display affinities which are lower than those of their parent
straight chain alkyl esters. This indicates the presence of steric
repulsions with the receptor in the L2 site and/or significant
differences in the conformational energies required to adopt the
bioactive conformations. The γ-branched alkyl ester 5 displays
only a slightly lower affinity than the straight chain butyl ester
4. This is also the case for the â-branched compound 6 if its
ester alkyl group is analyzed as a methyl-substitued butyl group
(6 vs 4). However, if it is analyzed as an ethyl-substituted propyl
group the affinity decrease is 16-fold (6 vs 3). The â-branched
alkyl group in 8 decreases the affinity by a factor of 7 compared
to that of 4.
The large variation in affinities for the compounds with
R-branched alkyl groups (9-12) cannot be understood in terms
of lipophilicity (11 and 12 have high lipophilicities but low

4-Quinolone Brain GABAA Receptor Ligands
Journal of Medicinal Chemistry, 2006, Vol. 49, No. 8 2529
Figure 1. Binding modes of (a) the high-affinity flavonoid 5′-bromo-2′-hydroxy-6-methylflavone and (b) the lead compound 1 in the pharmacophore
model.
tion, two conformational searches were performed both with
the carbonyl groups constrained to their proposed bioactive
arrangement. In the first conformational search, no additional
constraints were made. In the second search, one of the ester
alkyl branches was additionally constrained to adopt an all-
anti conformation whereas the other branch was allowed to find
its energetically preferred conformation. The energy of lowest
energy conformation found in the first search was then in each
case subtracted from the energy of lowest energy conformation
found in the second search. This gives the lowest possible
conformational energies required for the ester alkyl groups in
9, 10, and 12 to adopt the same bioactive conformation as the
ethyl group in 2 or the propyl group in 3 with one branch of
the chain. The energy penalties calculated in this way are 1.1,
1.9, and 3.1 kcal/mol for 9, 10, and 12, respectively. If the
relative affinities of these compounds are solely determined by
the conformational energy penalties, these calculations predict
that 9, 10, and 12 should have 6, 22, and 163 times lower
Figure 2. A molecular superimposition of the high-affinity flavonoid
5′-bromo-2′-hydroxy-6-methylflavone (orange carbon atoms) and the
lead compound 1 (green carbon atoms) in the pharmacophore model.
affinities than their parent compounds 3, 2, and 3, respectively.
The experimental relative affinities are 11, 11, and 1444. Thus,
the calculated conformational energy penalties for the assumed
bioactive conformation of these compounds capture a major part
of the observed relative affinities.
Properties of the Receptor Region in the Vicinity of the
6-Substituent (the “Interface” Region). According to the
superimposition in Figure 2, the 6-position in the 4-quinolones
corresponds to the 5′-position in the previously studied flavone
series. Substitution in this position in the flavones by small
substituents has, as mentioned above, been shown to result in
a significant affinity increase.11,12 As shown in Table 1, small
alkyl substituents in the 6-position increase the affinity also in
the 4-quinolone series. Ethyl (2), propyl (17), and isopropyl (18)
substitution all increases the affinity by a factor 5 compared to
that of the 6-unsubstituted compound 14. In the flavone series,
methyl substitution in the corresponding position increases the
affinity by a factor of 6. The 6-bromo substituent in 16 increases
the affinity by a factor of 5, which is a similar or slightly smaller
substituent effect than is observed in the flavone series (a factor
of 7-15).12
Figure 3. Conformational equilibrium with respect to arrangement of
the carbonyl groups in the lead compound 1.
affinities) or steric bulk (10 is smaller than 9 but has a lower
affinity). In an attempt to understand the origin of the significant
affinity differences for the R-branched compounds, we have
performed conformational analysis for 9, 10, and 12. According
to the pharmacophore model (Figure 1b), the ester ethyl in 2 as
well as the propyl group in 3 display an all-anti (extended)
bioactive conformation with the alkyl carbon atoms in the same
plane as the ester carbonyl and the ring system. We have
assumed that in the bioactive conformations of 9, 10, and 12,
one branch of the chain has an all-anti conformation superim-
posable on an all-anti ethyl or propyl group. To calculate the
energy required for adopting the assumed bioactive conforma-
A significant difference between the substituent effect on the
affinities in the flavone and the 4-quinolone series is observed
for the 6-CF3 substitutent. In the previously studied flavone
series12 a CF3 substituent in the 5′-position increases the affinity
by a factor of 18, whereas there is essentially no effect on the
affinity in the 4-quinolone series (1 vs 14 in Table 1). This

2530 Journal of Medicinal Chemistry, 2006, Vol. 49, No. 8
Lager et al.
3
Table 2. Affinity of Selected 4-Quinolones Tested on H-Flumazenil Binding to R1â2γ2S and R3â2γ2S GABAA Receptor Subtypes
compound
R1
R2
R3
X
Ki R1 (nM)a
Ki R3 (nM)a
Ki ratio: R3/R1
5
16
19
diazepam
zolpidem
CH2CH2CH(CH3)2
CH2CH3
CH2CH3
H
H
H
CH2CH3
Br
CH2C6H5
O
O
O
4.1 ( 2.6
14 ( 7
89 ( 18
26 ( 8
7.2 ( 0.3
14 ( 10
1850 ( 880
22
2
27
1
0.27 ( 0.06
12 ( 7
53 ( 50
35
a Each Ki value is the mean ( SD of three determinations. Ki for diazepam on R1 is the mean ( SD of seven determinations.
indicates that the electronic properties of the aromatic ring
induced by a polar substituent differ in the flavone and
4-quinoline systems and that the electronic properties of the
aromatic ring to which the substituent is attached is of
importance for the binding.
substituents in the imidazopyridopyrimidinone series also
displayed functional selectivity at receptor subtypes.19 This made
it of interest to test 19 on R1 and R3 containing receptors. Since
compound 16 only has a small alkoxy substituent and a small
substituent in the 6-position, it was selected as a reference
compound.
The results given in Table 2 interestingly show that 5 as well
as 19 display signicantly higher affinity for R1 subunit-
containing receptors compared to R3 subunit-containing recep-
tors. In contrast, the quinolone derivative 16 has essentially the
same affinity for R1- and R3-containing receptors (Table 2). The
ratio values (Ki R3/Ki R1) for 5 and 19 are in the same range as
that of the well-known R1 subunit selective compound zolpidem,
but the affinities of 5 and 19 for R1â2γ2S and R3â2γ2S are
significantly higher than those of zolpidem (Table 2). In terms
of the pharmacophore model, these results indicate that selectiv-
ity for R1â2γ2S over R3â2γ2S may be achieved with appropriate
substituents in the L2 region as well as in the receptor region
in the vicinity of the 6-position in 4-quinolones (the “interface”
region in Figure 1b).
Albaugh et al. have shown that arylalkyl substituents such
as a benzyl group could be introduced in a series of imida-
zopyridopyrimidinones resulting in compounds with high af-
finity for the benzodiazepine receptor.19 According to pharma-
cophore modeling by Albaugh et al.19 and the pharmacophore
model in Figure 1b, the position of these substitutions in the
imidazopyridopyrimidinones corresponds to the 6-position in
the 4-quinolone moiety. This prompted us to synthesize and
test the 6-benzyl compounds 19, 20, and 25. Interestingly, the
benzyl group in 19 increases the affinity by a factor of 56
compared to that of the 6-unsubstituted compound 14. Replacing
the ethyl ester group in 19 by a propyl ester giving 20 further
increases the affinity by a factor of 8 leading to subnanomolar
affinity (Ki ) 0.17 nM). The replacement of the ethyl group in
3 by a benzyl group (20) increases the affinity by a factor of
10. Similarily, the replacement of the 6-ethyl group in 23 by a
benzyl group giving 25 increases the affinity by a factor of 5,
yielding the highest affinity compound in the present series of
compounds (Ki ) 0.048 nM).
Pyrazolopyrimidinones. The pyrazolopyrimidinones inves-
tigated in this study all display low affinities for the benzodi-
azepine site compared to those of the corresponding 4-quino-
lones. The low affinities are most probably due to larger
desolvation energies for the pyrazolopyrimidones. The free
energy of hydration for the pyrazolopyrimidone 27 was calcu-
lated to be -22 kcal/mol using AM1/SM2.20 This should be
compared to the calculated free energy of hydration for the
4-quinolone 14, -17 kcal/mol. Thus, the desolvation energy of
28 is calculated to be 5 kcal/mol higher than that of 14. The
logP values for the two compounds as calculated by using the
ClogP algorithm are -0.17 and 0.93 for 28 and 14, respectively,
indicating a significantly higher lipophilicity for 14 than for
28.
Conclusions
The lead structure 1 has successfully been optimized with
respect to affinity for the benzodiazepine receptor by using a
pharmacophore-guided optimization approach. A number of high
affinity compounds including compounds 20 and 23-25
displaying subnanomolar affinities have been obtained. These
results validate the usefulness of the pharmacophore model. A
few compounds were selected for testing on the R1â2γ2S and
R3â2γ2S GABAA receptor subtypes. Two of the compounds (5
and 19) display selectivity for R1- versus R3-containing receptors
by a factor of 22 and 27, respectively. These results demonstrate
that selectivity for R1- over R3-containing receptors may be
obtained via substituents interacting with the receptor in the
L2 as well as in the “interface” region in terms of the
pharmacophore model.
Experimental Section
Subtype Selectivity. A few compounds, 5, 16, and 19,
representing different types of substitution patterns were selected
for testing on the GABAA receptor subtypes. The selection of
5 was based on the observation that the â-carboline BCCT which
contains a tert-butyl ester group displays selectivity for R1â2γ2
over R3â2γ2 by a factor of 26, whereas the corresponding ethyl
ester only displays low selectivity by a factor of 5.9 According
to pharmacophore modeling reported by He et al.,9 the bulky
tert-butyl group in BCCT interacts with the receptor in the L2
region. In terms of our pharmacophore model (Figure 1b), the
bulky part of the 3-methylbutoxy group in 5 is proposed to
interact with the receptor in the L2 region, making 5 an
interesting candidate for a subtype selective compound.
As discussed above, compound 19 displays high affinity for
the benzodiazepine receptor (Table 1). This in agreement with
our expectations based on the effect observed by Albaugh et
al. of benzyl substitution in imidazopyridopyrimidinones. In
addition to their affinity increasing effect, some alkylaryl
Chemistry. 1H and 13C NMR were recorded at room temperature
with a Bruker AR300 or a Bruker DR400 spectrometer. The spectra
were recorded in CDCl3, DMSO-d6, and CD3OD, and the solvent
signals (7.27 and 77.0, 2.50 and 39.5, or 3.31 and 49.0 ppm,
respectively) were used as reference. The raw data were trans-
formed, and the spectra were evaluated with the standard Bruker
UXNMR software (rev. 941001). Analytical thin-layer chromatog-
raphy (TLC) was performed on Kiselgel 60 F254 plates (Merck).
Column chromatography was performed on SiO2 (Matrex LC-gel:
60A, 35-70 MY, Grace). Melting points (uncorrected) were
determined with a Reichert microscope. EI mass spectra were
recorded at 70 eV with a JEOL S102 spectrometer, and ESI spectra
were recorded with Micromass Q-TOF Micro. Compounds 1 and
25 were purchased from Maybridge.
Diethyl 4-Ethylanilinomethylenemalonate (2a). 4-Ethylaniline
(0.145 g, 1.2 mmol) and diethyl ethoxymethylenemalonate (0.26
g, 1.2 mmol) were mixed and heated at 130 °C for 2 h. Low boiling
components were evaporated at low pressure with a cold trap
yielding 2a (0.35 g, 1.2 mmol, quantitative yield), as a yellow oil.

4-Quinolone Brain GABAA Receptor Ligands
Journal of Medicinal Chemistry, 2006, Vol. 49, No. 8 2531
3-Ethoxycarbonyl-6-ethyl-4-quinolone (2). 2a (0.756 g, 3.50
mmol) was added in portions to refluxing diphenyl ether (4 mL).
The mixture was refluxed for 1 h and then allowed to cool to room
temperature. Petroluem ether was added, and the resulting crystals
were collected and washed with a large amount of petroleum ether.
The crystals were purified by trituration with diethyl ether yielding
2 (0.320 g, 1.31 mmol, 37%), a white solid (mp: 271-273 °C).
Anal. (C14H15NO3) C, H, N.
6-Ethyl-3-propoxycarbonyl-4-quinolone (3). 2 (0.053 g, 0.22
mmol) and a catalytic amount of p-toluenesulfonic acid were
dissolved in 25 mL of n-propanol. The mixture was refluxed
overnight. A small amount of silica was added to the mixture
followed by evaporation at reduced pressure. The crude product
was purified by chromatography (CHCl3:EtOH 9:1). The fractions
containing the product (and the corresponding acid as a side
product) were washed with sat. NaHCO3(aq.) and dried with
MgSO4, yielding 3 (0.047 g, 0.18 mmol, 83%), a white solid (mp:
256-258 °C). Anal. (C15H17NO3) C, H, N.
3-Butoxycarbonyl-6-ethyl-4-quinolone (4) was prepared and
purified according to the procedure described for 3, using n-butanol
as solvent and reacting alcohol. The reaction yielded 4 (90%), a
white solid (mp: 225-227 °C). Anal. (C16H19NO3) C, H, N.
6-Ethyl-3-(3-methylbutoxycarbonyl)-4-quinolone (5) was pre-
pared and purified according to the procedure described for 3, using
3-methyl-1-butanol as solvent and reacting alcohol. The reaction
yielded 5 (46%), a white solid (mp: 252-254 °C). Anal. (C17H21-
NO3) C, H, N.
6-Ethyl-3-(2-methylbutoxycarbonyl)-4-quinolone (6) was pre-
pared and purified according to the procedure described for 3, using
2-methyl-1-butanol as solvent and reacting alcohol. The reaction
yielded 6 (73%), a white solid (mp: 239-241 °C). Anal. (C17H21-
NO3) C, H, N.
amine. The crude product was purified by recrystallization from
diethyl ether yielding 14a (82%), a white solid (mp: 49-51 °C).
3-Ethoxycarbonyl-4-quinolone (14) was prepared and purified
according to the procedure described for 2, starting from 14a. The
reaction yielded 14 (43%), a white solid (mp: 270-272 °C). Anal.
(C12H11NO3) C, H, N.
3-Butoxycarbonyl-4-quinolone (15) was prepared and purified
according to the procedure described for 3, starting from 14 and
using n-butanol as solvent and reacting alcohol. The reaction yielded
15 (89%), a white solid (mp: 252-254 °C). Anal. (C14H15NO3)
C, H, N.
Diethyl 4-bromoanilinomethylenemalonate (16a) was prepared
according to the procedure described for 2a, using 4-bromoaniline
as reacting amine. The crude product was purified by recrystalli-
zation from diethyl ether yielding 16a (73%), a white solid (mp:
93-95 °C).
6-Bromo-3-ethoxycarbonyl-4-quinolone (16) was prepared and
purified according to the procedure described for 2, starting from
16a. The reaction yielded 16 (57%), a white solid (mp: 320-322
°C). Anal. (C12H10NO3Br) C, H, N.
Diethyl 4-propylanilinomethylenemalonate (17a) was prepared
according to the procedure described for 2a, using 4-n-propylaniline
as reacting amine. Low boiling components were evaporated
yielding 17a (quantitative), a white solid (mp: 43-45 °C).
3-Ethoxycarbonyl-6-propyl-4-quinolone (17) was prepared and
purified according to the procedure described for 2, starting from
17a. The reaction yielded 17 (74%), a white solid (mp: 268-270
°C). Anal. (C15H17NO3) C, H, N.
Diethyl 4-i-propylanilinomethylenemalonate (18a) was pre-
pared according to the procedure described for 2a, using 4-isopro-
pylaniline as reacting amine. Low boiling components were
evaporated yielding 18a (quantitative), a brown oil.
6-Ethyl-3-pentoxycarbonyl-4-quinolone (7). 2 (51 mg, 0.21
mmol) and a catalytic amount of Ti(O-i-Pr)4 were dissolved in 10
mL of n-pentanol. The mixture was refluxed overnight. A small
amount of silica was added to the mixture followed by evaporation
at reduced pressure. The crude product was purified by chroma-
tography (CHCl3:EtOH 9:1) yielding 7 (49 mg, 0.172 mmol, 83%),
a white solid (mp: 231-233 °C). Anal. (C17H21NO3) C, H, N.
6-Ethyl-3-(2-ethylbutoxycarbonyl)-4-quinolone (8) was pre-
pared and purified according to the procedure described for 3, using
2-ethyl-1-butanol as solvent and reacting alcohol. The reaction
yielded 8 (51%), a white solid (mp: 216-218 °C). Anal. (C18H23-
NO3) C, H, N.
3-Cyclopentoxycarbonyl-6-ethyl-4-quinolone (9) was prepared
and purified according to the procedure described for 7, using
cyclopentanol as solvent and reacting alcohol. The reaction yielded
9 (80%), a white solid (mp: 270-272 °C). Anal. (C17H19NO3) C,
H, N.
6-Ethyl-3-i-propoxycarbonyl-4-quinolone (10) was prepared
and purified according to the procedure described for 7, using
i-propanol as solvent and reacting alcohol. The reaction yielded
10 (74%), a white solid (mp: 235-237 °C). Anal. (C15H17NO3)
C, H, N.
6-Ethyl-3-(4-methyl-2-pentoxycarbonyl)-4-quinolone (11) was
prepared and purified according to the procedure described for 7,
using 4-methyl-2-butanol as solvent and reacting alcohol. The
reaction yielded 11 (53%), a white solid (mp: 196-198 °C). Anal.
(C18H23NO3) C, H, N.
3-Ethoxycarbonyl-6-i-propyl-4-quinolone (18) was prepared
and purified according to the procedure described for 2, starting
from 18a. The reaction yielded 18 (43%), a white solid (mp: 271-
273 °C). Anal. (C15H17NO3) C, H, N.
Diethyl 4-benzylanilinomethylenemalonate (19a) was prepared
according to the procedure described for 2a, using 4-benzylaniline
as reacting amine. Low boiling components were evaporated
yielding 19a (quantitative) as a brown oil.
6-Benzyl-3-ethoxycarbonyl-4-quinolone (19) was prepared and
purified according to the procedure described for 2, starting from
19a. The reaction yielded 19 (87%), a white solid (mp: 278-280
°C). Anal. (C19H17NO3) C, H, N.
6-Benzyl-3-propoxycarbonyl-4-quinolone (20) was prepared
and purified according to the procedure described for 7, starting
from 19 and using n-propanol as solvent and reacting alcohol. The
reaction yielded 20 (66%), a white solid (mp: 273-275 °C). Anal.
(C20H19NO3) C, H, N.
Diethyl 4-ethylanilinoethylidenemalonate (21a). 4-Ethylaniline
(80.4 mg, 0.663 mmol), diethyl acetylmalonate (0.134 g, 0.334
mmol), and a catalytic amount of p-toluenesulfonic acid were
refluxed in 10 mL of pentane. Water produced during the reaction
was removed using a Dean-Stark trap filled with 4 Å molecular
sieves. After 24 h, 5 mL of sat. NaHCO3(aq) was added and the
mixture was extracted with diethyl ether (210 mL). The combined
organic extract was dried with MgSO4. The crude product was
purified by chromatography (SiO2, petroleum ether:ethyl acetate
6:1-1:1) yielding 21a (102 mg, 0.334 mmol, 50%), a yellow solid
(mp: 54-56 °C).
3-Ethoxycarbonyl-6-ethyl-2-methyl-4-quinolone (21) was pre-
pared and purified according to the procedure described for 2,
starting from 21a. The reaction yielded 21 (27%), a white solid
(mp: 248-250 °C). Anal. (C15H17NO3) C, H, N.
Diethyl 4-bromoanilinoethylidenemalonate (22a) was prepared
and purified according to the same procedure described for 21a,
using 4-bromoaniline as reacting amine. The reaction yielded 22a
(35%), a white solid (mp: 61-63 °C).
6-Ethyl-3-(3-pentoxycarbonyl)-4-quinolone (12) was prepared
and purified according to the procedure described for 7, using
3-pentanol as solvent and reacting alcohol. The reaction yielded
12 (43%), a white solid (mp: 231-232 °C). Anal. (C17H21NO3)
C, H, N.
3-Carboxy-6-ethyl-4-quinolone (13) was prepared and purified
according to the procedure described for 3, using the mixture
i-propanol:H2O (9:1) as solvent and water as the reacting species.
The reaction yielded 13 (68%), a white solid (mp: 285-287 °C).
Anal. (C12H11NO3) C, H, N.
Diethyl anilinomethylenemalonate (14a) was prepared accord-
ing to the procedure described for 2a, using aniline as reacting
6-Bromo-3-ethoxycarbonyl-2-methyl-4-quinolone (22) was
prepared and purified according to the procedure described for 2,

2532 Journal of Medicinal Chemistry, 2006, Vol. 49, No. 8
Lager et al.
starting from 22a. The reaction yielded 22 (45%), a white solid
(mp: 271-273 °C). Anal. (C13H12NO3Br) C, H, N.
6-Ethyl-3-propylaminocarbonyl-4-quinolone (23) was prepared
and purified according to the procedure described for 7, using
n-propylamine as solvent and reacting amine. The reaction yielded
23 (64%), a white solid (mp: 205-207 °C). Anal. (C15H18N2O2)
C, H, N.
3-Butylaminocarbonyl-6-ethyl-4-quinolone (24) was prepared
and purified according to the procedure described for 7, using
n-butylamine as solvent and reacting amine. The reaction yielded
24 (92%), a white solid (mp: 180-182 °C). Anal. (C16H20N2O2)
C, H, N.
6-Benzyl-3-propylaminocarbonyl-4-quinolone (25) was pre-
pared and purified according to the procedure described for 7, using
n-propylamine as solvent and reacting amine. The reaction yielded
25 (76%), a white solid (mp: 209-211 °C). Anal. (C20H20N2O2)
C, H, N.
Propyl 2-methyl-7-oxo-4,7-dihydropyrazolo[1,5-a]pyrimidine-
6-carboxylate (27) was prepared according to the procedure
described for 7, starting from 26 and using n-propanol as solvent
and reacting alcohol. The crude product was purified by chroma-
tography (CH2Cl2:MeOH 8:1) yielding 27 (quantitative yield), a
white solid (mp: 267-269 °C). Anal. (C11H13N3O3) C, H, N.
Ethyl 7-Oxo-4,7-dihydropyrazolo[1,5-a]pyrimidine-6-carbox-
ylate (28). 3-Aminopyrazole (432 mg, 5.2 mmol) was dissolved in
diethyl ethoxymethylenemalonate (1.12 g, 5.2 mmol) and heated
at 200 °C for 10 min. The crude product was purified by
recrystallization from DMF and filteration on silica (CH2Cl2:EtOH
10:1), yielding 28 (630 mg, 3.04 mmol, 58.5%), a white solid
(mp: 295-296 °C). Anal. (C9H9N3O3) C, H, N.
method21 with the default setting as implemented in MacroModel
7.0.22 Force field calculations were carried out using the MMFF94s
force field23 with solvation effects calculated by the GB/SA
hydration model.24 The energy minimizations were carried out using
the Polak-Ribiere conjugate gradient algorithm (PRCG) as imple-
mented in MacroModel 7.0. LogP calculations were performed by
using the ClogP algorithm as implemented in ChemDrawUltra
version 8.0.3 (CambridgeSoft). The free energies of hydration for
compounds 14 and 27 were calculated by using AM1/SM220 as
implemented in Spartan 02 for Macintosh (Wavefunction, Inc.)
Acknowledgment. Financial support from the Carlsberg
Foundation and the Swedish Natural Science Research Council
is gratefully acknowledged.
Supporting Information Available: Spectral data (1H NMR,
13C NMR and HRMS) of all synthesized compounds and elemental
analysis for all new target compounds. This material is available
free of charge via the Internet at http://pubs.acs.org.
References
(1) Sieghart, W. Structure and pharmacology of gamma-aminobutyric
acid A receptor subtypes. Pharmacol. ReV. 1995, 47 (2), 181-234.
(2) Kaupmann, K.; Huggel, K.; Heid, J.; Flor, P. J.; Bishoff, 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) Johnston, G. A. R. GABAC receptors: relatively simple transmitter-
gated ion channels. Trends Pharmacol. Sci. 1996, 17, 319-323.
(5) Barnard, E.; Skolnick, P.; Olsen, R.; Mohler, H.; Sieghart, W.; Biggio,
G.; Braestrup, C.; Bateson, A.; Langer, S., International union of
pharmacology. XV. Subtypes of γ-aminobutyric acidA receptors:
Classification of the basis of subunit structure and receptor function.
Pharmacol. ReV. 1998, 50, 291-313.
(6) Rudolph, U.; Crestani, F.; Benke, D.; Brunig, I.; Benson, J. A.;
Fritschy, J. M.; Martin, J. R.; Bluethmann, H.; Mohler, H. Benzo-
diazepine actions mediated by specific gamma-aminobutyric acid-
(A) receptor subtypes. Nature 1999, 401 (6755), 796-800.
(7) Rudolph, U.; Crestani, F.; Mohler, H. GABAA receptor subtypes:
dissecting their pharmacological functions. Trends Pharmacol. Sci.
2001, 22, 188-194.
(8) Doble, A.; Martin, I. L. The GABAA/Benzodiazepine receptor as
Target for PsychoactiVe Drugs; Springer-Verlag: Heidelberg, Ger-
many, 1996.
(9) He, X.; Huang, Q.; Ma, C.; Yu, S.; McKernan, R.; Cook, J. M.
Pharmacophore/receptor models for GABAA/BzR R2â3γ2, R3â3γ2
and R4â3γ2 recombinant subtypes. Included volume analysis and
comparison to R1â3γ2, R5â3γ2 and R6â3γ2 subtypes. Drug Des.
DiscoVery 2000, 17 (2), 131-171.
(10) Gardner, C. R.; Tully, W. R.; Hedgecock, J. R. The rapidly expanding
range of neuronal benzodiazepine ligands. Prog. Neurobiol. 1993,
40, 1-61.
(11) Dekermendijan, K.; Kahnberg, P.; Witt, M.; Sterner, O.; Nielsen,
M.; Liljefors, T. Structure-activity relationship and molecular
modeling analysis of flavonoids binding to the benzodiazepine site
of the rat brain GABAA receptor complex. J. Med. Chem. 1999, 42,
4343-4350.
(12) Kahnberg, P.; Lager, E.; Rosenberg, C.; Schougaard, J.; Camet, L.;
Sterner, O.; Østergaard Nielsen, E.; Nielsen, M.; Liljefors, T.
Refinement and evaluation of a pharmacophore model for flavone
derivatives binding to the benzodiazepine site of the GABAA receptor.
J. Med. Chem. 2002, 45, 4188-4201.
(13) Zhang, W.; Koehler, K. F.; Zhang, P.; Cook, J. M., Development of
a comprehensive pharmacophore model for the benzodiazepine
receptor. Drug Des. DiscoVery 1995, 12, 193-248.
(14) Kahnberg, P.; Howard, M.; Liljefors, T.; Nielsen, M.; Østergaard
Nielsen, E.; Sterner, O.; Pettersson, I. The use of a pharmacophore
model for identification of novel ligands for the benzodiazepine
binding site of the GABAA receptor. J. Mol. Graphics Modelling
2004, 23, 253-261.
(15) Gould, G.; Jacobs, W. The synthesis of certain substituted quinolines
and 5,6-benzoquinolines. J. Am. Chem. Soc. 1939, 61, 2890-2895.
(16) Seebach, D.; Hungerbu¨hler, E.; Naef, R.; Schnurrenberger, P.;
Weidmann, B.; Zu¨ger, M. Titanate-mediated trans-esterifications with
functionalized substrates. Synthesis 1982, 13, 138-141.
Propyl 7-oxo-4,7-dihydropyrazolo[1,5-a]pyrimidine-6-car-
boxylate (29) was prepared according to the procedure described
for 7, starting from 28 and using n-propanol as solvent and reacting
alcohol. The crude product was purified by chromatography (CH2-
Cl2:EtOH 10:1) yielding 29 (89%), a white solid (dec: 300 °C).
Anal. (C10H11N3O3) C, H, N.
3
Benzodiazepine Receptor Binding in Vitro. Binding of H-
flumazenil (87 Ci/mmol) to rat cortical membranes and to a
membrane suspension of HEK 293 cells expressing human R1â2γ2S
or R3â2γ2S GABAA receptors was done following methods previ-
ously described by Kahnberg et al.12 In brief, tissue is homogenized
in 20 mL of Tris, HCl (30 mM, pH 7.4) using an Ultra-Turrax
homogenizer. The suspensions are centrifuged at 27000g for 15
min followed by three centrifugation resuspension cycles. The
washed pellet is resuspended in 20 mL of buffer, incubated at 37
°C for 30 min, and then centrifuged for 10 min (27000g). The pellet
is washed once, and the final pellet is resuspended in 30 mL of
Tris-HCl buffer (50 mM, pH 7.4) and stored at -20 °C until use.
For binding studies frozen membrane suspensions were thawed and
centrifuged (27000g, 10 min). The pellet was resuspended into Tris-
citrate buffer (50 mM, pH 7.1) at a tissue concentration: cortex
preparation ca. 50 µg protein/0.55 mL assay (1 mg original tissue/
0.55 mL assay) and HEK cells ca. 25 µg protein per 0.55 mL assay.
Aliquots of 0.5 mL of membrane preparation are added to 25 µL
3
of H-flumazenil solution (1 nM final concentration) and 25 µL
containing test substance and incubated at an ice-bath (0-4 °C)
for 40 min. The incubated samples were added to 5 mL of ice-
cold buffer (Tris-citrate, 50 mM, pH 7.1), poured direcly onto
Whatman GF/C glass fiber filters under suction, and immediately
washed with 5 mL of ice-cold buffer. Nonspecific binding was
determined by adding clonazepam (1 µM final concentration) to
separate samples. Protein was estimated by conventional protein
assay method using bovine serum albumin as standard.
IC50 values were determined using four to six different concen-
trations of test substance. Ki values were calculated according to
Ki ) IC50/(1 + [L]/KD), ([L] is the concentration of 3H-flumazenil;
KD is binding affinity constant of 3H-flumazenil (1.6 nM). All assays
were done in triplicate (cortical assays) or duplicate (HEK cells).
Computational Methods. Conformational analyses were per-
formed by using the Monte Carlo Multiple Minimum (MCMM)

4-Quinolone Brain GABAA Receptor Ligands
Journal of Medicinal Chemistry, 2006, Vol. 49, No. 8 2533
(17) Nagahara, K.; Kawano, H.; Sasaoka, S.; Ukawa, C.; Hirama, T.;
Takada, A.; Cottam, H. B.; Robins, R. K. Reaction of 5-aminopyr-
azole derivatives with ethoxymethylenemalononitrile and its ana-
logues. J. Heterocycl. Chem. 1994, 31, 239-243.
(21) Chang, G.; Guida, W. C.; Still, W. C. An Internal Coordinate Monte
Carlo Method for Searching Conformational Space. J. Am. Chem.
Soc. 1989, 111, 4379-4386.
(22) Mohamadi, F.; Richards, N. G. J.; Guida, W. C.; Liskamp, R.; Lipton,
M.; Caulfield, C.; Chang, G.; Hendrickson, T.; Still, W. C. Macro-
Model - An Integrated Software System for Modeling Organic and
Bioorganic Molecules Using Molecular Mechanics. J. Comput. Chem.
1990, 11, 440-467.
(23) Halgren, T. A. MMFF94s option for energy minimization studies. J.
Comput. Chem. 1999, 29, 720-729.
(24) Still, W. C.; Tempczyk, A.; Hawley, R. C.; Hendrickson, T.
Semianalytical Treatment of Solvation for Molecular Mechanics and
Dynamics. J. Am. Chem. Soc. 1990, 112, 6127-6129.
(18) Albaugh, P. A.; Currie, K. S.; Rosewater, D.; Cai, G. Substituted
4-oxo-Quinoline-3-carboxamides: GABA Brain Receptor Ligands.
Patent Application WO 2000/68202.
(19) Albaugh, P. A.; Marshall, L.; Gregory, J.; White, G.; Hutchison, A.;
Ross, P. C.; Gallagher, D. W.; Tallman, J. F.; Crago, M.; Casella, J.
V. Synthesis and biological evaluation of 7,8,9,10-tetrahydroimidazo-
[1,2-c]pyrido[3,4-e]pyrimidin-5(6H)-ones as functionally selective
ligands of the benzodiazepine receptor site on the GABAA receptor.
J. Med. Chem. 2002, 45, 5043-5051.
(20) Cramer, C. J.; Truhlar, D. G. AM1-SM2 and PM3-SM3 parametrized
SCF solvation models for free energies in aqueous solution. J.
Comput.-Aided Mol. Des. 1992, 6, 629-666.
JM058057P