High affinity central benzodiazepine receptor ligands. Part 2: Quantitative structure-activity relationships and comparative molecular field analysis of pyrazolo[4,3-c]quinolin-3-ones

Article abstract of DOI:10.1016/S0968-0896(00)00262-5

A large series of 2-aryl(heteroaryl)-2,5-dihydropyrazolo[4,3-c]quinolin-3(3H)-ones (PQ, 106 compounds), carrying appropriate substituents at the quinoline and N₂-phenyl rings, were designed, prepared and tested as central benzodiazepine recepto

Full text of DOI:10.1016/S0968-0896(00)00262-5

Bioorganic & Medicinal Chemistry 9 (2001) 431±444
High Anity Central Benzodiazepine Receptor Ligands. Part 2:
Quantitative Structure±Activity Relationships and Comparative
Molecular Field Analysis of Pyrazolo[4,3-c]quinolin-3-ones
L. Savini,b L. Chiasserini,b C. Pellerano,b G. Biggio,c E. Maciocco,c
M. Serra,c N. Cinone,a,d A. Carrieri,a C. Altomarea and A. Carottia,*
a
Á
Dipartimento Farmaco Chimico, Universita degli Studi, via E. Orabona 4, I-70125 Bari, Italy
Á
Dipartimento di Biologia Sperimentale, via Palabanda 12, Universita degli Studi, I-09123 Cagliari, Italy
b
Dipartimento Farmaco Chimico Tecnologico, Universita degli Studi, A. Moro, I-53100 Siena, Italy
Á
c
d
Á
Dipartimento di Scienze del Farmaco, Universita degli Studi ``G. D'Annunzio'', via dei Vestini 31, I-66013 Chieti Scalo (CH), Italy
Received 17July 2000; accepted 26 September 2000
AbstractÐA large series of 2-aryl(heteroaryl)-2,5-dihydropyrazolo[4,3-c]quinolin-3(3H)-ones (PQ, 106 compounds), carrying appro-
priate substituents at the quinoline and N2-phenyl rings, were designed, prepared and tested as central benzodiazepine receptor ligands.
Compounds with an anity signi®cantly higher than the parent compound CGS-8216 were obtained, the most active ligand showing a
pIC50=10.35. Hansch and comparative molecular ®eld analyses gave coherent results suggesting the main structural requirements
of high receptor binding anity. The possible formation of a three-centred hydrogen bond (HB) at the HB donor site H2, as a key
interaction for high receptor binding anity, was assessed by the calculation and comparison of the molecular electrostatic potentials
of a series of selected ligands. # 2001 Elsevier Science Ltd. All rights reserved.
Introduction
types has been identi®ed yet.6 Recent evidence suggests
that subtype receptors with a1b2g2 subunit combina-
tions show many of the pharmacological e€ects asso-
ciated with the type I BzR, whereas a2b2g2, a3b2g2 and
a5b2g2 subunit compositions characterise the biological
e€ects of the so called type II BzR.7
Benzodiazepine receptor (BzR) ligands are structurally
diverse compounds that bind to the GABAA/BzR complex
evoking numerous neurological e€ects including con-
vulsion, anxiety and sleep, and in¯uencing memory and
learning processes.1À3
The possibility to carry out nowadays binding and
pharmacological studies on cloned BzR subtypes with a
well de®ned subunit composition should lead to a better
understanding of which subtypes mediate which speci®c
physiological response(s). Unfortunately, highly potent
and subtype-selective ligands, designed on the basis of
SAR studies on cloned receptor subtypes with well
de®ned subunit combinations, did not show the expected
pharmacological pro®le.8
The GABAA/BzR complex contains a chloride channel
and it is a membrane-bound heteropentameric protein
composed principally of a, b, and g subunits. Presently,
a total of 21 subunits (6a, 4b, 4g, 1e, 1d, 3r, 1W, and 1p)
of the GABAA receptor have been identi®ed by mole-
cular cloning4 and 16 of them have been found in the
mammalian CNS.5 It has been found that a functional
GABAA/BzR must contain a, b, and g subunits. The
heterogeneity of GABAA/BzR subtypes has been indi-
cated as the main factor responsible for the multiplicity
of the pharmacological properties displayed by benzo-
diazepines but no clear and de®nitive link of the observed
pharmacological activities to the diverse receptor sub-
While this new approach seems very rational and pro-
mising, these initial discouraging results demonstrate
that there is still a long way ahead to reach a better
comprehension of the structure±function relationships
of the diverse BzR subtypes.
In this context, research based on traditional binding
anity and intrinsic ecacy studies can still play a role
*Corresponding author. Tel.: +39-080-544-2782; fax: +39-080-544-
2230; e-mail: carotti@farmchim.uniba.it
0968-0896/01/$ - see front matter # 2001 Elsevier Science Ltd. All rights reserved.
PII: S0968-0896(00)00262-5

432
L. Savini et al. / Bioorg. Med. Chem. 9 (2001) 431±444
to improve our knowledge on the main structural
requirements necessary for a high binding anity and
possibly a well de®ned intrinsic activity.9À14 Indeed, sev-
eral pharmacophore models proved useful for the design
of new and potent BzR ligands.15 In this ®eld, a signi®cant
contribution also came from 2-D and 3-D QSAR stu-
dies16,17 on a new class of BzR ligands, the 2-aryl-2,5-
dihydropyridazino[4,3-b]indol-3(3H)-ones and from a
preliminary 2-D QSAR study18 on a structurally related
class of ligands, the well known 2-aryl-2,5-dihydropyr-
azolo[4,3-c]quinoline-3(3H)-ones (PQs).19
and well suited to perform QSAR studies. In addition
the examined data set covers a wide range of activity
spanning quite regularly more than four orders of mag-
nitude intervals.
Chemistry
The general synthetic procedure to prepare the pyr-
azolo[4,3-c]quinoline derivatives (2) is shown in Scheme 1.
Compounds 2 were obtained according to previously
described methods18 by reacting ethyl-4-chloroquinoline-
3-carboxylates (1) with substituted phenylhydrazines or
a-(N)-heterocyclichydrazines.
PQs constitute a biologically interesting class of BzR
ligands because of their peculiar pharmacological activity,
lacking many unwanted side e€ects observed in classical
benzodiazepines. Moreover, the net shift of intrinsic
activity, from agonist, through antagonist, to inverse
agonist, caused by small structural modi®cations of PQs
render this class of compounds particularly appealing
for deep structure±ecacy relationship studies.20À24
Finally, PQs represent ideal model compounds to carry
out SAR investigations in that they can be easily pre-
pared through simple and well consolidated synthetic
methods,25,26 thus allowing a proper exploration of their
physicochemical and structural properties.
All the synthesised PQs were isolated in satisfactory yields
1
(40±60%). Their spectral data (IR and H NMR) are in
accordance with literature19 data and allow unequivocal
1
structural assignments to all new compounds. In the H
NMR spectra, the characteristic signals for PQs (2) were
a singlet between 8.4 and 8.9 d, attributable to H-4
proton, and a broad singlet, exchangeable with D2O,
between 12.3 and 13.2 d due to the N5-H proton. Due to
a possible tautomeric equilibrium, sometimes the H-4
proton appeared as a doublet and collapsed to a singlet
upon treatment with D2O (see compound 281). In
A preliminary account of a QSAR study of PQs has been
recently reported by us in this journal.18 In the present
paper, we describe the design, synthesis, 2-D and 3-D
QSAR studies of a further set of PQ BzR ligands. The
molecular skeleton of PQs was appropriately sub-
stituted aiming at the identi®cation of the primary
ligand±receptor interactions, all around the quinoline
and the N2-phenyl rings, responsible for recognition and
di€erent activation of BzR. To reach this important
objective, a high number of PQ analogues (listed in
Table 1) were synthesised and their BzR anity mea-
sured by radioligand binding assay. In particular, several
mono- and polysubstituted derivatives at the quinoline
ring and monosubstituted ortho, meta, and para con-
geners at the N2-phenyl ring were prepared. The aryl
substituents at position 2 were also replaced by nitrogen
heterocycles. The overall structural modi®cations were
such that the substituent physicochemical space was
more thoroughly explored for meta and para congeners
and for positions 6, 7, and 8 of the quinoline moiety.
1
Table 2, the IR and H NMR spectral data of some
representative PQs are reported.
The known compounds CGS-8216 and CGS-9896 were
also synthesised as biological standards. Their physical
and spectral properties are consistent with literature
values.20
The hydroxy derivatives 223, 236, 237 and 277 were con-
veniently obtained from the corresponding alkoxy deri-
vatives by hydrolysis with 48% HBr in glacial acetic
acid. Catalytic reduction with 10% Pd/C of nitro com-
pounds 25, 28, 221 and 280 a€orded the target amino
derivatives 24, 27, 222 and 279, as already described.18
Compounds 2101 and 2102 were prepared from the corre-
sponding chloro derivatives (1a and 1b) and phenylhy-
drazine in analogy with all other described PQs (2). The
dimethylamino derivative 271 was obtained by reductive
methylation of the corresponding amino derivative 268
with formaldehyde, on Pd/C.
Our main goals were to reinforce and complement our
previous SAR and QSAR results and to derive meaningful
and highly predictive 3-D QSAR models, by means of
Comparative Molecular Field Analysis (CoMFA). The
development of signi®cant structure±ecacy relation-
ship (SER) models is also being pursued by analysing
intrinsic activity data of a properly selected set of PQ
Compound 8-tri¯uoromethoxy-2-(4-hydroxy-3-nitro-
phenyl) pyrazolo[4,3-c]quinolin-3-one was synthesised
by nitration of 8-tri¯uoromethoxy-2-(4-hydroxyphenyl)
pyrazolo[4,3-c]quinolin-3-one18 using KNO3, aqueous
sulfuric acid and a catalytic amount of sodium nitrite;
its subsequent catalytic reduction with 10% Pd/C yiel-
ded the corresponding amino derivative 273.
ligands, measured through
a binding assay with
[35S]TBPS,27,28 and an electrophysiological assay on the
cloned receptor subtype a2b2g2. The results from this
SER study will be reported soon in this journal.
Biochemistry
It is worth underlining that, with the only exception of
compounds 22 and 26, all the binding data reported in
Table 1 have been measured by the same method and in
the same laboratory and therefore are very homogeneous
The compounds listed in Table 1 were tested for their
ability to displace [3H]¯unitrazepam binding from rat
brain membranes according to standard procedures.34

L. Savini et al. / Bioorg. Med. Chem. 9 (2001) 431±444
Table 1. Chemical structures and binding anities of pyrazolo[4,3-c]quinolin-3-ones (21±2107
433
)
Mp (ꢀC)b
Cryst. solventc
a
0
R6 (2 )
0
0
0
R4
Compound
R6
R7
R8
R9
R5
pIC50
d
21
22
23
24
25
26
27
28
H
H
H
H
H
H
H
H
H
H
H
H
F
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
OCH3
Cl
F
CF3
NO2
NH2
OH
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
F
F
F
F
F
Cl
Cl
OCH3
OCH3
OCH3
OC2H5
OC2H5
OH
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
Unsubstituted at N2
6.30
9.35
8.16
9.77
8.53
9.40
8.67>300
7.96
9.00
9.17Ð
6.84
6.75
8.16
7.18
5.73
5.66
8.37>300
8.67>300
9.01
7.64
7.69
8.86
8.88
9.54
8.70
9.26
9.48
9.34
9.37Ð
9.37>300
9.17Ð
8.80
7.75
8.85
8.56
9.70
9.28
6.30
9.00
9.18
5.93
6.86
9.50
8.77
9.22
8.35
5.55
8.60
8.36
8.16
7.75
6.94
9.15
9.40
8.60
>300
Ð
Ð
>300
>300
>300
A
±
±
A
D
A
e,f
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
Cl
NH2
NO2
H
H
H
H
f
n
H
H
H
H
OH
NH2
NO2
Cl
n
A
n
>300
Ð
A
±
e,f
29
210
211
f,g
212
e,f
H
H
H
OCH3
H
Cl
±
f,g
Ð
Ð
>300
>300
294±296
288±290
±
±
A
A
B
A
h
213
214
H
H
l
CH3
CF3
OCH3
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
h
215
216
H
H
l
H
H
f
217
H
H
A
A
m
218
219
H
H
H
H
>300
>300
>300
285±288
>300
>300
>300
>300
A
A
C
B
A
A
D
A
A
A
m
220
H
H
221
222
223
H
H
H
H
H
H
n
224
225
H
NO2
H
H
h
h
226
227
NH2
H
H
H
n
OCH3
OH
H
290 (dec)
>300
h
228
e
229
230
H
±
±
2-Pyrid-20-yl
E
e
231
H
F
H
H
H
H
232
233
234
>300
>300
>300
>300
>300
250 (dec)
Ð
Ð
240±243
>300
286 (dec)
263±267B
>300
270 (dec)
Ð
>300
>300
>300
>300
Ð
A
B
B
A
A
E
±
2-Pyrimid-20-yl
h
H
H
H
H
H
H
H
H
H
Cl
H
235
236
237
OH
Cl
i
238
n-C4H9
n-C4H9
n-C4H9
n-C4H9
n-C4H9
n-C4H9
n-C4H9
n-C4H9
c-C6H11
c-C6H11
c-C6H11
c-C6H11
c-C6H11
OBn
OCF3
OCF3
OCF3
OCF3
OCF3
OCF3
OCF3
OCF3
Unsubstituted at N2
j
239
240
H
F
H
H
H
H
H
H
H
H
±
A
E
A
h
241
242
COOH
COOCH3
243
244
2-Pyrid-20-yl
2-Pyrimid-20-yl
2-Pyrazin-20-yl
H H
h
h
A
B
±
A
A
C
B
±
B
A
F
B
h
245
246
j
H
H
h
247
248
H
COOH
2-Pyrid-20-yl
h
249
250
2-Pyrimid-20-yl
2-Pyrazin-20-yl
H H
h
h
f
251
252
H
h
Unsubstituted at N2
>300
>300
268±271
291±294
h
253
254
H
F
H
H
H
H
H
H
H
H
H
H
h
h
255
256
Cl
CH3
H
h
H
8.47290±292
B
h
257
258
Br
CH3
Cl
7.46
8.20
7.62
>300
295 (dec)
>300
A
B
A
h
H
H
h
259
(continued on next page)

434
L. Savini et al. / Bioorg. Med. Chem. 9 (2001) 431±444
Table 1 (continued)
a
0
R6 (2 )
0
0
0
R4
Compound
R6
R7
R8
R9
R5
F
pIC50
MP
(ꢀC)b
Cryst. solventc
h
260
261
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
F
F
F
F
F
F
F
H
F
F
F
F
F
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
Cl
F
F
F
OCF3
OCF3
OCF3
OCF3
OCF3
OCF3
OCF3
OCF3
OCF3
OCF3
OCF3
OCF3
OCF3
OCF3
OCF3
OCF3
H
H
H
H
H
H
H
F
F
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
OH
OH
OCH3
OCH3
OCH3
OCH3
OCH3
H
H
H
H
H
H
H
Cl
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
Br
CH3
Cl
9.40
7.20
9.62
7.82
8.79
7.90
9.00
7.40
9.10
9.22
9.63
8.26
5.50
7.98
10.35
8.50
9.62
8.82
8.84
8.85
8.45
8.67>300
8.94
7.87
8.02
6.79
8.12
7.82
6.47>300
6.94
8.43
7.70
7.15
7.13
7.68
8.14
8.90
5.52
8.50
7.24
8.94
6.00
8.14
8.52
8.14
8.54
9.08
7.45
>300
>300
>300
>300
>300
>300
>300
>300
>300
>300
284±287A
285 (dec)
>300
>300
>300
>300
Ð
A
D
B
B
A
A
A
C
B
B
h
NO2
NH2
H
h
262
263
h
h
264
265
H
H
H
H
H
H
H
H
h
h
266
267
F
h
NO2
NH2
OCH3
OH
N(CH3)2
COOH
OH
h
268
269
h
h
270
271
272
273
274
275
B
A
G
A
A
±
A
±
G
D
H
NH2
2-Pyrid-20-yl
2-Pyrimid-20-yl
f
276
277
H
H
H
H
H
H
H
H
H
H
H
H
NH2
H
NH2
NO2
H
H
H
H
H
H
OH
OCH3
H
H
Br
>300
Ð
265±269
300
f
278
279
280
281
282
A
H
H
F
H
>300
Ð
Ð
Ð
Ð
B
±
±
±
±
k
283
284
k
k
285
286
F
F
F
F
F
H
F
F
k
H
OCH3
h
287
288
2-Pyrid-20-yl
>300
A
2-Pyrimid-20-yl
A
h
h
289
290
2-Pyrazin-20-yl
>300
Ð
A
±
k
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
CH3
Cl
h
291
292
>300
>300
>300
>300
>300
298±301
>300
>300
297 (dec)
297(dec)
>300
>300
Ð
A
A
A
A
A
A
D
A
A
h
h
293
294
F
F
F
h
F
F
F
h
295
296
OCH3
H
COOH
h
OCH3
OCH3
OCH3
OCH3
OCH3
OCH3
OCH3
OCH3
OCH3
OCH3
H
OCH3
OCH3
OCH3
OCH3
OCH3
H
h
297
298
2-Pyrid-20-yl
h
h
299
2100
2-Pyrimid-20-yl
2-Pyrazin-20-yl
A
h
2101
2102
2103
2104
2105
2106
2107
6,7-BENZOFUSED
H
H
H
H
H
F
H
H
H
H
H
H
H
H
H
H
OCH3
H
E
B
±
H
H
H
H
H
H
H
8,9-BENZOFUSED
i
i
7,8-OCH2O
7,8-OCH2O
OC4H9
H
H
H
H
H
Ð
±
H
H
H
237±40
240±243
>300
B
A
A
OC4H9
OC4H9
H
Cl
H
aBinding data from the displacement of [3H]¯unitrazepam, ref 32.
bReported only for the newly synthesised compounds and for compounds already described but lacking those data.
cCrystallization solvent(s): A: EtOH; B: EtOH+H2O; C: DMF; D: DMF+H2O; E: MeOH; F: AcOEt; G: puri®ed by column chromatography
(dichloromethane:methanol, 18:2).
dRef 29.
eRef 23, pIC50s of compounds 22 and 29 have been redetermined. Compounds 22, 29 and 210 are better known as CGS-8216, CGS-9896 and CGS-
9895, respectively.
fRef. 22, pIC50 of compound 217 has been redetermined.
gN5-Methyl derivatives.
hMolecules from our previous paper (ref 18).
iRef 28.
jRef 26.
kRef 21.
lRef 30.
mRef 31, pIC50 of compounds 218 and 220 have been redetermined.
nRef 33, pIC50 has been redetermined.
Quantitative structure±activity relationship (QSAR) studies
The pIC50 values reported in Table 1 have been nor-
malized using CGS-8216 (22) and CGS-9896 (29) as
internal reference compounds. Literature data in Table 1
were also normalized by the same method.
An extensive review on the quantitative structure±
activity relationships of several classes of central ben-

L. Savini et al. / Bioorg. Med. Chem. 9 (2001) 431±444
Table 2. 1H NMR and IR spectral data of some representative PQs (2)
435
Compound
1H NMR (d, ppm; J=Hz)
IR (cmÀ1
)
1618 (CO)
1620 (CO)
1616 (CO)
25
(DMSO-d6) 7.50±7.74 (m, 4H), 7.97 (dd, 1H), 8.22 (d, 1H), 8.63 (dd, 1H), 8.75 (s, 1H, H-4), 9.08
(n t, 1H, H-20), 12.88 (br s, 1H, NH, D2O exchangeable).
(DMSO-d6) 7.17 (t, 1H), 7.42 (t, 2H, H-(30,50)), 8.13 (d, 2H, Jorto=8.2 Hz), 8.20 (dd, 1H, H-8, J8-9=8.8, J8-6=2.0 Hz), 8.32
(d, 1H, H-9, J9-8=8.8 Hz), 8.44 (n d, 1H, H-6, J6-8=2.0 Hz), 8.81 (s, 1H, H-4), 12.94 (br s, 1H, NH, D2O exchangeable).
(DMSO-d6) 7.10 (dd, 1H, H-8, J8-9=1.8), 7.20 (d,1H, H-6, J6-8=8.7, J8-6=1.8 Hz), 7.23 (d, 1H, PhH), 7.50 (t, 2H, PhH),
8.12 (d, 1H, H-9, J9-8=8.7Hz), 8.3 (d, 2H, PhH), 8.65 (s, 1H, H-4), 10.44 (br s, 1H, OH, D 2O exchangeable).
221
223
233
235
(DMSO-d6) 3.88 (s, 3H, OCH3), 7.24±7.36 (m, 2H, H-7 and H-5 pyrimidinyl), 7.51 (d, 1H, H-9, J9-7=2.6 Hz), 7.63 (d, 1H, H-6, 1647(CO)
J6-7=9.2 Hz), 8.61 (s, 1H, H-4), 8.82 (d, 2H, H-(4, 6) pyrimidinyl), 12.65 (br s, 1H, NH, D2O exchangeable).
(DMSO-d6) 1.39 (t, 3H, CH3-CH2O), 4.18 (q, 2H, CH3-CH2O), 7.28 (dd, 1H, H-7, J7-6=9.0, J7-9=2.4 Hz), 7.48 (d, 2H, PhH,
Jorto=8.8 Hz), 7.55 (d, 1H, H-9, J9-7=2.4 Hz), 7.66 (d, 1H, H-6, J6-7=9.0 Hz), 8.29 (d, 2H, PhH, Jorto=8.8 Hz)
8.65 (s, 1H, H-4), 12.82 (br s, 1H, NH, D2O exchangeable).
1618 (CO)
242
(DMSO-d6) 0.89 (t, 3H, CH3-CH2-CH2-CH2-), 1.31 (st, 2H, CH3-CH2-CH2-CH2-), 1.61 (qt, 2H, CH3-CH2-CH2-CH2-),
2.73 (t, 2H, CH3-CH2-CH2-CH2-), 3.83 (s, 3H, OCH3), 7.51 (dd, 1H, H-7, J7-6=8.4, J7-9=1.3 Hz), 7.61 (d, 1H, H-6,
J6-7=8.4 Hz), 8.01 (d, 3H, H-9 and 2PhH), 8.39 (d, 1H, H-4, Jorto=8.7Hz), 8.70 (s, 1H, H-4), 12.82 (br s, 1H, NH,
D2O exchangeable).
1667(CO)
274
(DMSO-d6) 7.29 (td, 1H, H-5 pyridyl), 7.72 (dd, 1H, H-7, J7-6=9.1, J7-9=2.0 Hz), 7.85±7.99 (m, 2H, H-6 and 1H pyridyl),
8.07(u d, 1H, H-9), 8.25 (d, 1H, H-3 pyridyl, J3-4=8.1 Hz), 8.55 (dd, 1H, H-6 pyridyl), 8.83 (s, 1H, H-4),
12.99 (br s, 1H, NH, D2O exchangeable).
1665 (CO)
277
281
(DMSO-d6) 5.21 (br s, 2H, NH2, D2O exchangeable), 6.41 (d, 1H), 6.98 (d, 1H), 7.07 (t, 1H, H-7), 7.20 (d, 1H), 7.33±7.40
(m, 2H), 7.52 (t, 1H, H-20), 8.72 (s, 1H, H-4), 9.58 (s, 1H, OH, D2O exchangeable), 12.86 (br s, 1H, NH, D2O exchangeable).
(DMSO-d6) 4.05 (s, 3H, OCH3), 6.90 (d, 2H, PhH, Jorto=8.8 Hz), 7.14 and 7.31 (u dd, 2H, H-(6,8), 7.62 (t, 1H, H-7),
8.02 (d, 2H, PhH, Jorto=8.8 Hz), 8.63 (d, 1H, H-4, JCH-NH=5.7Hz), 9.40 (s, 1H, OH, D 2O exchangeable), 12.61 (d, 1H, NH,
JNH-CH=5.7Hz, D 2O exchangeable).
1624 (CO)
1639 (CO)
2101
2102
2106
(DMSO-d6) 7.22 (t, 1H, H-40), 7.50 (t, 2H, H-(30, 50)), 7.75±7.89 (m, 2H), 8.06±8.31 (m, 5H), 8.70 (s, 1H, H-4),
8.78 (u dd, 1H, H-6), 13.01 (br s, 1H, NH, D2O exchangeable).
1636 (CO)
(DMSO-d6) 7.25 (t, 1H, H-40), 7.54 (t, 2H, H-(30, 50)), 7.72±7.98 (m, 3H), 8.14±8.26 (m, 3H), 8.39 (d, 1H, H-7, J7-6=8.4 Hz), 8.89 1636 (CO)
(s, 1H, H-4), 9.93 (d, 1H, H-6, J6-7=8.4 Hz), 13.17(br s, 1H, NH, D 2O exchangeable).
(DMSO-d6) 0.93 (t, 3H, CH3-CH2-CH2-CH2O), 1.44 (st, 2H, CH3-CH2-CH2-CH2O), 1.74 (qt, 2H, CH3-CH2-CH2-CH2O),
4.08 (t, 2H, CH3-CH2-CH2-CH2O), 7.24 (d, 1H, Jmeta=2.7 Hz), 7.28±7.42 (m, 3H), 7.47 (d, 1H, H-9, J9-7=2.7Hz),
7.57 (dd, 1H), 7.65 (d, 1H, H-6, J6-7=8.9 Hz), 8.66 (s, 1H, H-4), 12.80 (br s, 1H, NH, D2O exchangeable ).
1654 (CO)
Scheme 1. R and R0 as reported in Table 1 (i) R0-phenylhydrazine (ethanol re¯ux); a-(N)-heterocyclichydrazines (rt).
zodiazepine receptor (CBR) ligands, not inclusive of PQ
compounds, has been reported by Hansch.35 Subse-
quently, a limited number of PQs has been subjected to a
QSAR analysis leading to some interesting results, pre-
liminarily reported by us in this journal.18
The electronic and hydrophobic e€ects of substituents
were assessed by the Hammett (s) and Hansch (p) sub-
stituent constants, whereas the molar refractivity (MR),
the van der Waals volume (vW) and the STERIMOL
Verloop parameters (L, B1 and B5) were employed to
model bulkiness and polarizability e€ects.37
In order to gain further indications at a quantitative level
on the topography of the BzR binding sites, a QSAR
study of a larger series of PQs, including all the com-
pounds reported in our previous paper,18 was carried out
by the classical Hansch approach.36 Unfortunately, the
lack of parameters for some substituents and the pre-
sence of several polisubstituted congeners, quite dicult
to be correctly parameterised, prevented a safe application
of the classical QSAR approach to the overall data set.
The Hansch approach, on the other hand, is well suited
for congeneric series and for this reason the QSAR ana-
lysis was conducted separately on the 6-, 7-, and 8-sub-
stituted 2-aryl congeners and on meta and para substituted
2-phenyl congeners of the 8-OCF3 derivative 253.
The following equations were derived for the indicated
subsets:
. 8 substituted, 2 phenyl congeners
pIC50 ˆ À0:312ꢀÆ0:04†vW ‡ 9:662ꢀÆ0:11†
n ˆ 11; r2 ˆ 0:879; s ˆ 0:211; Q2 ˆ 0:781; F ˆ 65:14
ꢀ1†
. 7substituted, 2-phenyl congeners
pIC50 ˆ À1:066ꢀÆ0:32†vW ‡ 9:372ꢀÆ0:30†
n ˆ 8; r2 ˆ 0:618; s ˆ 0:412; Q2 ˆ 0:292;
F ˆ 9:69
The multiple linear regression analysis (MLR) with
cross-validation was performed on the binding data in
Table 1 using the chemical descriptors listed in Table 4.
ꢀ2†

436
L. Savini et al. / Bioorg. Med. Chem. 9 (2001) 431±444
. 6-substituted, 2-phenyl congeners
pIC50 ˆ À2:613ꢀÆ0:47†vW ‡ 9:347ꢀÆ0:45†
n ˆ 5; r2 ˆ 0:912; s ˆ 0:543; Q2 ˆ 0:801;
F ˆ 31:23
ꢀ3†
. 8-OCF3, 2-phenyl meta and para substituted con-
geners
pIC50 ˆ À1:343ꢀÆ0:32†ꢀm; p À 0:960ꢀÆ0:30†vWm; p
‡ 9:479ꢀÆ0:28†
n ˆ 15; r2 ˆ 0:774; s ˆ 0:448;
Q2 ˆ 0:723; F ˆ 20:50
ꢀ4†
Figure 1. Proposed pharmacophore model13 for BzR. The main bind-
ing sites for pyrazoloquinoline CGS-8216 (22) are indicated as H1 and
H2 (HB donor sites), A2 (HB acceptor site), L1 and L2 (hydrophobic
sites). L3 is another lipophilic region reached by the 5-phenyl ring of
classical benzodiazepine. S1 represents a sterically inaccessible region.
Binding to H2 and A2 is not necessary for inverse agonist activity.39
In eqs. (1)±(4), n is the number of compounds, r2 is the
squared correlation coecient, s is the standard devia-
tion, Q2 is the squared crossvalidation coecient and F
is the Fisher ratio. The standard errors of the regression
coecients are reported in parentheses. Eqs. (1), (3),
and (4) present acceptably good statistics in terms of
®tting power, whereas eq. (2) gave poorer statistics,
especially in terms of predictive capacity. Eqs. (1), (2),
and (3) point out a signi®cant steric hindrance at 8-, 7-,
and 6-positions, respectively, as accounted for by the
substituent van der Waals volume vW. The negative
coecient with vW is much higher in eq. (3) than in eqs.
(1) and (2) and this suggests a stronger steric e€ect at
position 6, which may hamper the formation of a
hydrogen bond of the N5-H group, as hypothesised in
several pharmacophore models.38
The negative sign of s and vW in eq. (4) indicates that
the binding of the 2-phenyl-substituted ring at the L1
(L2) receptor regions (see pharmacophore model in
Fig. 1) is favoured by small electron donor substituents
at both meta and para positions. An equation of com-
parable statistical value can be obtained by substituting
vW with MR. Compounds 272, bearing at para position
a carboxylic group, resulted in a strong outlier and was
excluded from the regression. The drop of 272 may be
justi®ed considering that the carboxylic group is ionised
at physiological pH, and that the carboxylate anion has
less accurately de®ned electronic and steric parameters.
Moreover, it would be the only charged compound of
this set of ligands likely constituting a possible leverage
point, all the other data being referred to uncharged
ligands. The lipophilic character of the substituents
seems to play a marginal role in the receptor±ligand
interaction and this casts some doubt about the lipo-
philic nature assigned in several pharmacophore models to
some receptor regions, especially the one named L2, which
is close to the para position of the 2-phenyl ring (Fig. 1).
Figure 2. Pictorial QSAR of PQs showing regression and squared
correlation coecients of the descriptors in each positional QSAR.
From the previous Hansch equations based on the sub-
stituent e€ects on single positions, a comprehensive
regression equation was derived to describe the pIC50
changes by considering at once all the relevant e€ects
already seen, singularly, at positions 6, 7, 8, 30 (meta)
and 40 (para), and parameterised now by vW6, vW7,
vW8, sm,p and vWm,p, respectively.
The following comprehensive equation was formulated:
. 6,7,8-monosubstituted-2-phenyl congeners and 8-
OCF3 2-phenyl, substituted congeners:
pIC50 ˆ À2:999ꢀÆ0:26†vW6 À 1:251ꢀÆ0:24†vW7
À 0:268ꢀÆ0:07†vW8 À 1:405ꢀÆ0:30†ꢀm; p
In addition, eq. (4) illustrates a limited steric accessi-
bility at the L2 region, as also noticed in a previous
study,16 and suggests that a p±p stacking interaction,40
in which the substituted phenyl ring acts as p electron
donor, might take place at the L1 receptor region.
À 0:683ꢀÆ0:17†vWm; p ‡ 9:955ꢀÆ0:20†
ꢀ5†
n ˆ 36; r2 ˆ 0:849; s ˆ 0:421; Q2 ˆ 0:750;
F ˆ 33:89
Also in the derivation of eq. (5), compounds 272 was
omitted.
Eqs. (1)±(4) can be more eciently interpreted by the
pictorial QSAR reported in Figure 2.

L. Savini et al. / Bioorg. Med. Chem. 9 (2001) 431±444
437
Eq. (5) gives a straightforward overall picture of the
main physicochemical interactions, underlying the BzR
binding of PQs and, as expected, it ®ts well with the pic-
torial QSAR in Figure 2. On the basis of these results, two
new compounds, namely compounds 271 and 273, were
designed, synthesised and tested by taking into account, in
particular, the favourable e€ects on the binding anity
exerted by electron donor groups (negative s values) in
positions meta and para.
the steric and electrostatic ®elds and also includes an
entropy component.49 For these reasons, the CoMFA
methodology, with the inclusion of the MLP, was selec-
ted as the more appropriate tool to study the SARs of
our CBR ligands.
Molecular models were constructed from the fragment
library of SYBYL 6.6 and their geometry optimised by
the AM1 Hamiltonian within the suite of program
MOPAC.50 Each ligand molecule was subjected to a
conformational analysis, when necessary, through a sys-
tematic search and the minimum energy conformers
were selected for the superposition.
Indeed, we obtained two very active compounds whose
binding anity was acceptably well predicted by eq. (4):
271 (pIC50(obs)=8.53, pIC50(pred.)=8.42) and 273
(pIC50(obs)=8.70, pIC50(pred.)=9.07).
The molecular alignment is the most critical step in a
CoMFA study.51 In the present work, dealing with
congeneric compounds, the molecular overlay was per-
formed by choosing as the anchor moiety the common
pyrazolo-quinoline ring. With few exceptions, the sub-
stituents in the quinoline ring were used in their mini-
mum energy conformations. The relative orientation of
the phenyl and azaheterocyclic rings in position 2, as
well as the diverse topology of the ortho and meta sub-
stituents, were analysed in full detail.
3-D QSAR study: comparative molecular ®eld analysis
Since the valuable indications coming from the preced-
ing 2-D QSAR study refer to a relatively small number
of congeneric ligands (36) we decided to extend our
investigations to the whole set of ligands (106 com-
pounds) by using a 3-D QSAR approach, that is the
Comparative Molecular Field Analysis (CoMFA41).
CoMFA is a widely used tool for studying quantitative
structure±activity (property) relationships (QSAR,
QSPR) at the three-dimensional level.42 Unlike the tra-
ditional Hansch analysis, which makes use of substituent
parameters, CoMFA relates the biological activity (tar-
get property) of a series of molecules with their steric
and electrostatic ®elds sampled at grid points de®ning a
large 3-D box around the molecule. CoMFA descriptors
are commonly constituted by steric (Lennard-Jones)
and electrostatic (Coulomb) potentials computed for
each molecule, at each grid point, by means of a suitable
probe, usually an sp3 carbon atom with a charge of +1.
Partial Least Squares (PLS43) is used as the regression
method to ®nd the relations between independent vari-
ables (steric and electrostatic potentials) and biological
activity (dependent variable). PLS analysis produces
model equations which explain the variance in the target
property in terms of the independent variables.44 The
optimum number of components (latent variables) is
determined by cross-validation and the model predictive
ability is assessed by cross-validated correlation coe-
cient (r2cv, Q2).45 The graphical representation of
CoMFA results as isocontour maps indicated the
regions where the variation in steric and electrostatic
properties of di€erent molecules in a data set is corre-
lated with the variation of biological activity.
For the sake of clarity and a more prompt comparison,
the binding anities of selected 8-substituted ligands
bearing phenyl, ortho-¯uorophenyl and nitrogen het-
erocyclic substituents at 2 position were taken from
Table 1 and collected in Table 3. The analysis of data in
Table 3 reveals that the introduction of a ¯uorine atom
in ortho position, as well as the replacement of the
phenyl ring with a 2-pyrid-20-yl ring, always yielded
higher anities, the strongest e€ect being seen for 274. A
less de®nite e€ect was observed for the 2-pyrimidin-20-yl
and 2-pyrazin-20-yl substituted ligands.
It was interesting to note that the pyridyl nitrogen and
the ¯uoro atom in ortho position can occupy close
regions and that, when facing the pyrazole N1, a likely
three-centred HB could be formed. Strong steric and
repulsive electrostatic interactions could prevent the
formation of a similar HB with larger sized ortho-sub-
stituents. In principle, a similar three-centred HB could
also be formed with the oxygen of the 3-carbonyl group,
in place of the N1 pyrazole nitrogen, with the phenyl
Table 3. Chemical structures and binding anity of selected 2,8-di-
substituted PQs 2
Classical CoMFA studies rely on standard steric and
electrostatic molecular ®elds to model receptor±ligand
interactions. Unfortunately, as demonstrated in several
investigations,46 these two ®elds are not always able to
appropriately describe all binding forces. Furthermore,
CoMFA describes only the enthalpic component of the
ligand±receptor interactions. Introducing the molecular
lipophilicity potential (MLP47) as an additional ®eld has
been shown to signi®cantly improve the descriptive,
interpretative and predictive power of CoMFA in many
cases.48 The MLP encodes indeed hydrogen bonds and
hydrophobic interactions not suciently described by
R2/R8
n-C4H9 9.00 (239
c-C6H11 8.35 (246
OCF3
OC4H9 8.54 (2105
Cl 9.37( 229)
)
)
)
9.18 (240
Ð
9.40 (254
9.08 (2106
Ð
)
)
9.50 (243
8.60 (248
10.35 (274
Ð
)
)
)
8.77 (244
8.36 (249
8.50 (275
Ð
)
)
)
9.22 (245
8.16 (250
)
)
9.15 (253
)
)
Ð
Ð
9.39 (230
)
Ð

438
L. Savini et al. / Bioorg. Med. Chem. 9 (2001) 431±444
Table 4. Substituent parameters used for the derivation of regression equations (1)±(5)a
b
Subs.
sp
p
MRc
vW
L
B1
B5
H
0.00
0.00
1.04
2.13
2.51
0.10
0.79
1.96
2.674.26
0.60
0.79
3.174.95
0.49
0.09
1.25
0.57
0.50
0.89
0.74
0.54
2.174.276.86
0.08
1.60
3.12
6.17 1.91
1.07
1.49
8.20
2.7 1.35
0.36
2.42
1.01
1.11
1.32
1.20
0.672.7 1.35
1.35
2.06
4.57
6.171.52
1.00
1.35
1.00
3.61
OCF3
n-C4H9
c-C6H11
Cl
OCH3
OBn
OH
0.35
À0.16
À0.22
4.54
3.49
1.93
0.23 (0.37)
0.71
À0.02
3.52
3.98
1.80
1.35
1.61d
1.80
3.07
4.44d
À0.27
À0.23
À0.37
d
1.66
À0.670.28
0.14
0.38
0.56
0.88
4
F
0.06 (0.34)
2.65
4.80
2.87
3.30
3.82
3.44
1.35
1.35
1.52
1.99
1.95
1.70
1.35
3.36
2.04
2.61
1.95
2.44
OC2H5
CH3
CF3
À0.24
À0.17( À0.07)
0.54
Br
0.23 (0.39)
0.78 (0.71)
À0.66 (À0.16)
À0.32 (0.10)
0.86
NO2
NH2
OC4H9
À0.28
À1.23
1.55
8
1.97
4.7
9
aData taken from ref 37. See text for the signi®cance of parameter symbols.
bs values for meta substituents (sm) are reported in parentheses.
cScaled by 0.1.
dReferring to the benzyloxy group in a planar conformation.
ring in a di€erent conformation. Taking into account
the most widely accepted pharmacophore model of
CBR ligands (see Fig. 1) this latter orientation should
be less favoured because of the steric bulk at S1. Inter-
estingly, a three-centred hydrogen bond has been used
to explain the high binding anities observed for other
BZ-receptor ligands such as certain benzodiazepines
and b-carbolines, which are thought to interact with the
same HB donor site (H1).52
around it. In our case, the proton atom was chosen as
the probe and therefore our calculated MIPs are indeed
classical Molecular Electrostatic Potentials (MEPs).
To compute and locate the MEP minima, the ab initio
program GAMESS60 was used within the MIN module
of MIPSIM. The MIN calculation was performed on a
40Â33Â19 A cubic grid lattice (default spacing=0.5 A)
giving a total number of points equal to 25,080. The
conjugate gradient algorithm was selected as the opti-
misation method and all the energy minima under a
value of 5 kcal/mol were optimised with a tolerance of
the gradient=0.01 kcal/mol A. The charges of the
ligands were calculated using the RHF/6-31G basis set
implemented in the GAMESS package.
Also the 2-pyrimidin-20-yl and 2-pyrazin-20-yl sub-
stituted ligands might form, in principle, a three-centred
HB analogously to the 2-pyrid-20-yl congeners and
therefore a higher anity could be expected. Data in
Table 3 show that this is not the case; most likely the
presence of a second heterocyclic nitrogen might gen-
erate additional important interactions leading to more
complex structure±anity relationships.
The results of our study are reported in Figure 3, where
the main minima, expressed in kcal/mol and indicated
by violet balls, are represented only for the moiety of
interest, that is the 2-substituted 3-pyrazolone moiety.
In order to furnish a signi®cant support to the hypoth-
esis of a three-centred HB formation, a quantum
mechanical study was undertaken on selected PQs
ligands, namely the 8-OCF3 derivatives 253, 254, 274 and
275 (Table 3). It should be anticipated that three-centred
HBs have been studied, both theoretically and experi-
mentally, only when nitrogen and oxygen atoms are
involved.53 In contrast, no investigation has been made
yet on a three-centred HB involving a ¯uorine atom.
Indeed, for the F-C group, the so called `organic ¯uor-
ine', even the formation of a simple HB of OH-FC and
NH-FC type has been the object of a strong dispute in
the scienti®c chemical community and only very
recently has it been proved and accepted.54À56
Interestingly, Figure 3 shows only three main minima for
each moiety, two close to the two carbonyl lone pairs and
the third located between the N1 and the N01 heterocyclic
nitrogens or the N1 and the ortho-¯uoro atom.
The minima near the carbonyl groups, whose intensities
are comparable for all the analysed compounds, are
positioned in a symmetric way with respect to the car-
bonyl oxygen atom, with the notable exception of the
pyrimidinyl congeners, where the ``upper'' minimum is
close to the N03 nitrogen.
The existence of only one minimum between N1 and N01
(or ortho-¯uoro) atoms may support the hypothesis of a
formation of a three-centred HB where the hydrogen
atom is very near to those minima.
Our theoretical study was carried out by means of
MIPSIM, a useful tool for the analysis and comparison
of Molecular Interaction Potentials (MIPs). MIPSIM57
is an evolution of the MEPSIM58,59 program that allows
the calculation of the potentials to be analysed and
compared, as the interaction energies between a molecule
and proper probes, computed in each node of a grid all
Interestingly, the depth of the minima seems related to the
observed binding anities, with the exception of the pyr-
imidin-20-yl derivative (275), that, as already anticipated,

L. Savini et al. / Bioorg. Med. Chem. 9 (2001) 431±444
439
could give rise to further interactions driven by the pre-
sence of the second heterocyclic nitrogen.
The statistical results of CoMFA models are repre-
sented in Table 5 and the corresponding isocontour
maps are shown in Figures 4, 5, and 6. The colour code
used to characterise favourable and unfavourable zones
of each ®eld is described in Table 6 and it is the same
used by us and others in past studies.48
The previous analysis gave us precise indications as to
the alignment of ortho-substituted 2-phenyl and 2-het-
erocyclic congeners.
As far as the alignment of the meta derivatives is con-
cerned, the two possible diverse topologies were taken
into account and the one giving the best statistical
model was retained.
The introduction of the MLP as third ®eld did not
improve the statistical value of the models. This was not
very surprising since in the QSAR eqs. (1)±(5) derived
by the classical Hansch approach no lipophilic para-
meters was found signi®cant.
In summary, the ortho-¯uoro substituents were placed
in position 20, all the other ortho-substituents being in
position 60; meta-substituents were in position 50 and all
the heterocyclic substituents were aligned with one
nitrogen in position 20 (see Fig. 1 for numbering).
In a ®rst step we derived the two-®eld PLS model ES1
by analysing the same data set used in the 2-D QSAR
study. Model ES1 and regression eq. (5) show compar-
able statistics in terms of ®tting power, being however
the predictive ability of eq. (5) signi®cantly better than
model ES1. Interestingly, the results coming from 2-D
and 3-D QSAR approaches are in quite good agreement
All the selected ligands conformations were super-
imposed by the RIGIDFIT option of SYBYL (poorer
results always came from the FIELDFIT option) and
subjected to PLS analysis in conjunction with cross-
validation (CV; leave-one-out method61) to derive the
optimal number of components (ONC) to be used in the
subsequent analyses. The PLS run was repeated with the
ONC (see below) and a number of CV groups set to
zero. However, in following strictly this procedure there
is a risk of obtaining over®tted models due to a rela-
tively high number of components to be used.62 We
therefore chose, from a plot of the squared cross-vali-
dated Q2 versus ONC, the ®rst maximum in the curve.
The selection of a lower number of components than
the ONC given by default in SYBYL yields poorer sta-
tistics in terms of r2 and standard deviation (worse ®t-
ting), but more realistic and trustworthy models.63
Table 5. Statistical results from the CoMFA study
c
Model
ES1
n
NOCa
4
Q2b
sdcv
r2d
sde
Fields
36
0.536 02.770.9170.308
Steric and
Electrostatic
Steric
Electrostatic
S2
E2
ES2
106
106
106
6
3
0.404 0.838
0.434 0.805
70.632 0.662 0.886 0.369
Steric and
Electrostatic
aNumber of optimal components.
bSquared cross-validation correlation coecient.
cStandard deviation of errors of predicted values.
dSquared correlation coecient.
eStandard deviation of errors of ®tted values.
Figure 4. Electrostatic and steric isocontour maps from model ES1.
Electrostatic contour levels: white=0.04, magenta=À0.034. Steric
contour levels: red=À0.070, green=0.044. Molecules shown to help
interpretation are: 215 (pIC50=5.73), 216 (pIC50=5.66), 261
(pIC50=7.20), 220 (pIC50=7.64), 236 (pIC50=9.70), 260 (pIC50=9.40),
270 (pIC50=9.63), and 224 (pIC50=9.54).
Figure 3. MEP minima (kcal/mol, violet balls) on the 2-substituted 3-
pyrazolone moiety of (clockwise) 253, 254, 274 and 275 ligands are
marked as violet balls.

440
L. Savini et al. / Bioorg. Med. Chem. 9 (2001) 431±444
Table 6. Colour code of CoMFA isocountour maps in Figures 3, 4,
and 5
Field or ®eld component
Increased
anity
Decreased
anity
Steric ®eld
Electrostatic ®eld (positive charge)
Electrostatic ®eld (negative charge)
Green
White
Magenta
Red
Magenta
White
interesting to note that to the electronic e€ects accounted
for by s in eqs (4) and (5), correspond to two favourable
electrostatic signals, coloured in magenta in Figure 4,
close to positions 50 and 40 of the 2-phenyl ring. They
may be ascribed to electron-rich groups in those posi-
tions of the 2-phenyl ring.
In a second step the CoMFA study was extended to the
whole set of PQ ligands. The four 40-COOH derivatives
(241, 247, 272 and 297) were considered in their ionised
form in the derivation of the two-®eld PLS model ES2,
which presents satisfactory statistics in terms of both ®t-
ting and predictive power and agrees well with the 2-D
QSAR models derived by MLR. At physiological (bind-
ing) pH these acidic ligands should indeed be completely
ionised. However, their elimination from the model, as
well as their inclusion as unionised species, led to poorer
PLS models (results not shown).
Figure 5. Steric isocontour maps from model ES2; contour levels:
green=0.035, red=À0.053. Molecules shown to help interpretation
are: 211 (pIC50=6.84), 215 (pIC50=5.73), 220 (pIC50=7.64), 242
(pIC50=6.86), and 258 (pIC50=8.20).
The isocontour maps were therefore developed from
PLS model ES2. In the steric map of Figure 5, sterically
allowed regions (green polyedra) can be seen close to 50,
8 and, to a lesser extent, 20 positions. Strong unfavour-
able steric interactions, indicated by red polyedra, can
be localised near positions 5 and 6 of the quinoline ring
and position 40 of the 2-phenyl ring. Additional detri-
mental steric interactions can be located at positions 7
and 8. Most of these signals (see also Fig. 4) are easily
interpretable considering the strong drop of anity
observed for N-5 substituted (compound 211), 6-sub-
stituted (compound 215), 7-substituted (compound 220)
and 8-substituted (compound 242) ligands. The strong
negative steric signal at position 40 may be due to very
low activity ligands such as the acidic compounds 241,
247, 272 and 297, and the ester 242, the latter being shown
on the map. The electrostatic signals indicated mainly
unfavourable interactions (white polyedra) for electron-
rich moieties located at 60, 50, and 40 positions of the 2-
phenyl ring (compounds 255, 261 and 272 in Fig. 6) and
at 6 and 7positions of the quinoline moiety (com-
pounds 2101 and 215 in Fig. 6). The magenta polyedra,
whose occupation by electron-rich groups increases the
binding anity, are near positions 9 (1), 40 and 10 (20).
Interestingly, an informative magenta polyhedron is
placed just between positions 1 and 9, where the for-
mation of a three-centred HB involving the N1 nitrogen
and small electron donor group like the OH at position 9
in compounds 276 and 277 may be responsible for the
observed high anity. The synthesis and testing of deri-
vatives bearing small sized HB accepting substituents at
position 9 might prove the validity of such a hypothesis.
Figure 6. Electrostatic isocontour maps from model ES2; contour levels:
white=0.05, magenta=À0.025. Molecules shown to help interpretation
are: 215 (pIC50=5.73), 255 (pIC50=8.60), 261 (pIC50=7.20), 272
(pIC50=5.50), 276 (pIC50=9.62), and 2101 (pIC50=6.00).
suggesting similar receptor±ligand interactions, as can
be seen by a simple visual comparison of the negative
steric signals (red polyhedra) of the contour map in
Figure 4 with the pictorial QSAR in Figure 2. In the case
of the electrostatic ®elds, no straightforward comparison
can be made with the 2-D QSAR map, since, by de®ni-
tion, electrostatic ®eld indicates local electronic e€ects,
whereas classical electronic substituent parameters, like the
Hammett s used in the Hansch analysis, measure an e€ect
on a formally distant ``reaction centre''. Nevertheless, it is
The favourable e€ects on the binding anity caused by
the replacement of the phenyl ring in position 2 with a

L. Savini et al. / Bioorg. Med. Chem. 9 (2001) 431±444
441
pyridine ring (compare 239 versus 243 and 246 versus 248),
although not completely highlighted by the isocontour
maps in Figure 6, prompt us to design, synthesise and
test a new potent ligand, namely compound 274, whose
binding anity was much higher than that predicted by
the ES2 model: pIC50(obs)=10.35, pIC50(pred)=8.80.
monitoring. Column chromatography on silica-gel
(Merck, 70±230 mesh and 230±400 mesh ASTH for
¯ash chromatography) was applied, when necessary, to
isolate and purify the di€erent reaction products. The
unknown ethyl 4-chloroquinoline-3-carboxylates (1a
and 1b) are also described in this section. Physicochem-
ical properties of the newly synthesised PQs are sum-
marised in Table 1.
Conclusion and perspectives
Taken together, the results from our QSAR studies gave
us coherent insights about the nature and spatial loca-
tion of the main interactions underlying the binding of
potent ligands at the CBR. As observed in the past in
several SAR studies, the coordinated application of 2-D
and 3-D QSAR methods furnishes signi®cant insights
which may complement each other in a synergic manner
yielding at the 3-D molecular level a clear picture of the
main forces modulating important chemical64,65 and
biological processes.66À70 Our results con®rm that for
strictly congeneric series the traditional QSAR approach
(Hansch analysis) still yields more valuable and physi-
cochemically better interpretable models than 3-D
QSAR methods.71 However, 3-D QSAR approaches42
play a decisive and indispensable role when structurally
diverse compounds have to be examined.
3-Carbethoxy-4-chlorobenzo(h)quinoline (1a). A mixture
of 3-carbethoxy-4-hydroxybenzo(h)quinoline (10 g, 71
mmol) and phosphorus oxychloride (25 mL) was heated
in a sand bath at 135 ꢀC for 45 min. After cooling the
mixture was poured on ice water and made basic with
sodium carbonate. The resulting precipitate, ®ltered o€
and puri®ed by recrystallisation from ethanol, gave a
white solid, mp 115±117ꢀC. 1H NMR (CDCl)3 d 1.52 (t,
3H, CH3-CH2O), 4.52 (q, 2H, CH3-CH2O), 7.75±7.79
and 7.92±7.97 (2 m, 4H), 8.26 (d, 1H, Jortho=9.5 Hz),
9.28±9.33 (m, 2H). Anal. calcd for C16H12ClNO2: C,
67.26; H,4.23; N, 4.90. Found: C, 67.56; H, 4.33; N, 4.86.
3-Carbethoxy-4-chlorobenzo(f)quinoline (1b). The title
compound (1b) was obtained following the same proce-
dure as described above for compound 1a; in this case
the resulting crude oil was immediately used without
further puri®cation for the subsequent reaction with
phenylhydrazine.
Our ®ndings are in good accordance with previous SAR
studies on CBR ligands and may be used to complement
existing pharmacophore models and to guide future
structural modi®cations. Being based on a rather large
array of PQ ligands, properly modi®ed to ®nd out sig-
ni®cant substituent e€ects in the diverse positions of the
quinoline and 2-phenyl rings of the lead compound
CGS-8216, our results appear quite robust and reliable.
8n-Butyl-2-(4-carbomethoxyphenyl)pyrazolo[4,3-c]quinolin-
3(5H)-one (242). The starting 8n-butyl-2-(4-carboxy-
phenyl)pyrazoloquinolinone18 (0.3 g, 0.83 mmol), added
with methanol (5 mL) and concd H2SO4 (0.1 mL), was
re¯uxed in an oil bath for 5±6 h. The reaction mixture
was evaporated under reduced pressure, the residue was
diluted with water and then washed with aqueous
NaHCO3 saturated solution. The resulting crude pro-
duct was puri®ed by recrystallisation from ethanol, mp
Finally, several ligands with a binding anity sig-
ni®cantly higher than the lead compound CGS-8216,
such as compounds 24, 236 and 276, were identi®ed. They
might constitute new chemical tools to probe CBR
ligand binding site(s) and are presently being used to
study in more detail structure±activity and structure±
ecacy relationships.
ꢀ
1
286 C dec. H NMR (DMSO-d6) d 0.89 (t, 3H, CH3-
CH2-CH2-CH2-), 1.31 (st, 2H, CH3-CH2-CH2-CH2-),
1.61 (qt, 2H, CH3-CH2-CH2-CH2-), 2.73 (t, 2H, CH3-
CH2-CH2-CH2-), 3.83 (s, 3H, OCH3), 7.51 (dd, 1H, H-
7, J7À6=8.4, J7À9=1.3 Hz); 7.61 (d, 1H, H-6, J6À7
=
8.4 Hz), 8.01 (d, 3H, H-9 and 2PhH), 8.39 (d, 2H, PhH,
Jorto=8.7Hz), 8.70 (s, 1H, H-4), 12.82 (br s, 1H, NH,
D2O exchangeable). IR 1667cm À1 (CO). Anal. for
C22H21N3O3 (C, H, N).
Chemical Experimental Section
Melting points were determined on a Gallenkamp
capillary melting point apparatus and are uncorrected.
Elemental analyses were performed on a Perkin±Elmer
elemental analyser Mod. 240 and the data for C, H, and
N are within Æ0.40% of calculated values. Spectral
analyses (IR and 1H NMR) are consistent with the
indicated chemical structures. IR spectra were deter-
mined in Nujol mulls on a Perkin±Elmer 398 or a Per-
kin±Elmer FT-IR 1600 spectrophotometer. 1H NMR
spectra were recorded on an AC 200 Bruker instrument.
The chemical shifts (d) are relative to Me4Si used as
internal standard. The following abbreviations were
used: s=singlet, d=doublet, t=triplet, dd=double
doublet, q=quartet, qt=quintet, st=sextet, m=multi-
plet, u=unresolved, br=broad, n=narrow. The cou-
pling constants J are in hertz. TLC on silica-gel plates
(Merck, 60-F254) was used for purity check and reaction
8-Tri¯uoromethoxy-2-(4-N,N0-dimethylaminophenyl)pyr-
azolo[4,3-c]quinolin-3(5H)-one (271). To a stirred solution
of 8-tri¯uoromethoxy-2-(4-aminophenyl)pyrazolo[4,3-c]
quinolin-3(5H)-one18 (576mg, 1.6 mmol) in methanol
(5 mL), containing 40% aqueous formaldehyde (0.4 mL,
5 mmol), was added a solution of sodium cyanoboro-
hydride (105 mg, 1.6 mmol) and zinc chloride (110 mg,
0.8 mmol) in methanol (5 mL). After stirring at room
temperature for 4 h, the reaction mixture was taken up
in 0.1 N NaOH (10 mL), and the excess of methanol was
removed under reduced pressure; the residue, puri®ed
through ¯ash chromatography (eluent CH2Cl2:CH3OH,
19:1), was recrystallised from EtOH:H2O, mp 285 ꢀC dec.
1H NMR (DMSO-d6) d 2.88 (s, 6H, 2CH3), 6.78 (d, 2H,
PhH, Jortho=9 Hz), 7.62 (dd, 1H, H-7, J7À8=8.9, J7À9
=

442
L. Savini et al. / Bioorg. Med. Chem. 9 (2001) 431±444
1.7Hz), 7.8 (d, 1H, H-6, J6À7=8.9 Hz), 7.9 (d, 2H, PhH,
Jortho=9 Hz), 7.99 (u d, 1H, H-9), 8.68 (s, 1H, H-4), 12.8
(br s, 1H, NH, D2O exchangeable). IR 1620 cmÀ1 (CO).
Anal. for C19H15F3N4O5 (C, H, N).
The animals were killed by decapitation and the brains
were rapidly removed, the cerebral cortex was dissected
out and was used for the measurement of [3H]¯uni-
trazepam.
8-Tri¯uoromethoxy-2-(4-hydroxy-3-nitrophenyl)pyrazolo
[4,3-c]quinolin-3(5H)-one. To a stirred solution of 8-tri-
¯uoromethoxy-2-(4-hydroxyphenyl)pyrazoloquinolin-3-
one18 (1.2 g, 3.3 mmol) in absolute ethanol, KNO3
(0.34 g, 3.3 mmol), NaNO2 (0.12 g, 1.7mmol) as the
catalyst and aqueous sulfuric acid were added. After
stirring at 70 ꢀC for 2±3 h, following the reaction course
by TLC (CHCl3:CH3OH, 18:2, as eluent), the mixture
was taken up in ice water. The resulting precipitate was
®ltered o€ and puri®ed by chromatography on silica gel
column (CHCl3:CH3OH, 18:2, as eluent) to a€ord the
[3H]Flunitrazepam binding assay32
Cerebral cortices were homogenised in 50 volumes of
ice-cold 50 mM Tris±HCl bu€er with a polytron PT 10
(setting 5, for 20 s), centrifuged at 48,000 g for 10min and
washed one time. The pellet was resuspended in 50
volumes of 50mM Tris±HCl bu€er (pH 7.40) and aliquots
of 400 mL tissue homogenate (400±500mg of protein) were
incubated in the presence of [3H]¯unitrazepam at a ®nal
concentration of 0.5 nM, in a total incubation volume of
1000mL. The compounds were dissolved in dimethylsulf-
oxide and serial dilutions were made up in dimethylsulf-
oxide and added in 5 mL aliquots. After 60 min
incubation at 4 ꢀC, the assay was terminated by rapid
®ltration through glass-®bre ®lter strips (Whatman GF/
B). The ®lters were rinsed with 2 Â 4 mL ice-cold 50 mM
Tris±HCl bu€er with a cell Harvester ®ltration manifold
(Model M-24, brandel) and transferred in plastic mini-
vials with 3 mL scintillation ¯uid (AtomLight, New
England Nuclear). Six to eight concentrations of the
samples in triplicate were used to determine the IC50
values. Non-speci®c binding was determined as the
binding in the presence of 5 mM diazepam and was 85±
90% of the total binding.
pure title compound, mp>300 ꢀC. H NMR (DMSO-
1
d6) d 7.20 (d, 1H, H-50, J5 À6 =9 Hz), 7.65 (dd, 1H, H-7,
J7À6=9.1 Hz), 7.80 (d, 1H, H-6, J6À7=9.1 Hz), 8.01 (d,
0
0
1H, H-9), 8.32 (dd, 1H, H-60, J6 À5 =9, J6 À2 =1.8 Hz),
0
0
0
0
8.72 (d, 1H, H-20, J2 À6 =1.8 Hz), 8.77 (s, 1H, H-4),
10.45 (br s, 1H, OH, D2O exchangeable), 13.02 (br s,
1H, NH, D2O exchangeable). IR 1618 cmÀ1 (CO). Anal.
for C17H9F3N4O5 (C, H, N).
0
0
8-Tri¯uoromethoxy-2-(3-amino-4-hydroxyphenyl)pyra-
zolo[4,3-c]quinolin-3(5H)-one (273). To a suspension of 8-
tri¯uoromethoxy-2-(4-hydroxy-3-nitrophenyl) pyrazolo-
[4,3-c]quinolin-3(5H)-one (160 mg, 3.9 mmol), in abso-
lute ethanol (120 mL), 10% Pd/C (0.02 g) was added.
The reaction mixture was hydrogenated at room tem-
perature in a Parr apparatus at 40 psi for 4 h, then the
catalyst was ®ltered o€ through Celite and the solution
taken to dryness. The resulting residue was puri®ed
through chromatography on silica gel column (CHCl3:
CH3OH, 18:2, as eluent) to give 273 as yellow powder,
Acknowledgements
mp >300 ꢀC. H NMR (DMSO-d6) d 4.67(br s, 2H,
The ®nancial support from MURST and CNR (Rome)
is gratefully acknowledged. The authors wish to thank
Prof. Ferran Sanz of the University Pompeu Favra,
Barcelona (Spain) for the free access to the MIPSIM
software package.
1
NH2, D2O exchangeable),0 6.68 (d, 1H, H-50, J5 À6
0
0
=
0
0
0
0
8.7Hz), 7.20 (dd, 1H, H-6 , J6 À5 =8.7, J6 À2 =2.3 Hz),
7.41 (d, 1H, H-20, J2 À6 =2.3 Hz), 7.65 (dd, 1H, H-7,
0
0
J7À6=9.2, J7À9=2.1 Hz), 7.83 (d, 1H, H-6, J6À7
=
9.2 Hz), 7.98 (d, 1H, H-9, J9À7=1.6 Hz), 8.70 (s, 1H, H-
4), 8.95 (s, 1H, OH, D2O exchangeable), 12.89 (br s, 1H,
NH, D2O exchangeable). IR 1618 cmÀ1 (CO). Anal. for
C17H11F3N4O5 (C, H, N).
References and Notes
1. Kerr, I. B.; Ong, J. J. Med. Res. Rev. 1992, 12, 593.
2. GABA and Benzodiazepine Receptors; Squires, R., Ed.;
CRC: Boca Raton, 1988; Vols. 1 and 2.
Biochemical Experimental Section
Chemicals
3. Bernard, E. A. In GABAA Receptor and Anxiety. From
Neurobiology to Treatment; Biggio, G., Sanna, E., Costa, E.,
Eds.; Raven: New York, 1995; pp 1±16.
4. Metha, A. K.; Ticku, M. K. Brain Res. Rev. 1999, 29, 196.
5. Sieghart, W. Pharm. Rev. 1995, 47, 181.
[3H]Flunitrazepam (New England Nuclear, Boston,
USA) had a speci®c activity of 84.3 Ci/mmol and a
radiochemical purity >74.6%.
6. Verdoorn, T. A.; Draguhn, A.; Ymer, S.; Seeburg, P. H.;
Sakmann, B. Neuron 1990, 4, 919.
Animals
7. Chang, Y.; Wang, R.; Barot, S.; Weiss, D. S. J. Neurosci.
1996, 16, 5415.
8. Huang, Q.; He, X.; Ma, C.; Liu, R.; Yu, S.; Dayer, C. A.;
Wenger, G. R.; McKernan, R.; Cook, J. M. J. Med. Chem.
2000, 43, 7 1.
Male Sprague±Dawley rats (Charles River, Como,
Italy) with body weights 150±200 g were kept under a
12 h light/dark cycle at a temperature of 23 Æ 2 ꢀC and
65% humidity. Upon arrival at the animal facilities
there was a minimum of 7days acclimatisation, during
which the animals had a free access to food and water.
9. Villar, H. O.; Davies, M. F.; Loew, G. H.; Maguire, P. A.
Life Sci. 1991, 48, 593.

L. Savini et al. / Bioorg. Med. Chem. 9 (2001) 431±444
443
10. Diaz-Arauzo, H.; Koehler, K. F.; Hagen, T. J.; Cook, J.
M. Life Sci. 1991, 49, 207.
11. Allen, M. S.; Skolnick, P.; Cook, J. M. J. Med. Chem.
1992, 35, 368.
42. 3-D QSAR in Drug Design: Theory, Methods and Applica-
tions; Kubinyi, H., Ed.; Escom: Leiden, 1993.
43. Dunn, W. J. III; Wold, S.; Edlund, U.; Helberg, S. Quant.
Struct.-Act. Relat. 1984, 3, 131.
12. Schove, L. T.; Perez, J. J.; Loew, G. H. Bioorg. Med.
Chem. 1994, 2, 1029.
13. Zhang, F.; Koehler, K. F.; Zhang, P.; Cook, J. M. Drug
Des. Dis. 1995, 12, 193.
14. Harris, D. L.; Loew, G. Bioorg. Med. Chem., in press.
15. Gardner, C. R. Progr. Neuro-Psychopharmacol. Biol. Psy-
chiatry 1992, 16, 755.
16. Campagna, F.; Carotti, A.; Casini, G.; Palluotto, F.;
Genchi, G.; De Sarro, G. B. Bioorg. Med. Chem. 1993, 1,
437.
44. Wold, S.; Johansson, E.; Cocchi, M. In 3D QSAR in Drug
Design: Theory, Methods and Applications; Kubinyi, H., Ed.;
Escom: Leiden, 1993; p 523.
45. Cramer, R. D. III; Bunce, J. D.; Patterson, D. E. Quant.
Struct.-Act. Relat. 1988, 7, 18.
46. Thibaut, U. In 3D QSAR in Drug Design: Theory, Meth-
ods and Applications; Kubinyi, H., Ed.; Escom: Leiden, 1993;
p 661.
47. Gaillard, P.; Carrupt, P.-A.; Testa, B.; Boudon, A. J.
Comput.-Aided Mol. Des. 1994, 8, 83.
17. Palluotto, F.; Carotti, A.; Casini, G.; Campagna, F.;
Genchi, G.; Rizzo, M.; De Sarro, G. B. Bioorg. Med. Chem.
1996, 4, 2091.
18. Savini, L.; Massarelli, P.; Nencini, C.; Pellerano, C.; Big-
gio, G.; Maciocco, A.; Tuligi, G.; Carrieri, A.; Cinone, N.;
Carotti, A. Bioorg. Med. Chem. 1998, 6, 389.
19. Yokoyama, N.; Ritter, B.; Neubert, A. D. J. Med. Chem.
1982, 25, 337.
20. Takada, S.; Shindo, H.; Sasatani, T.; Chomei, N.; Matsa-
shita, A.; Masami, E.; Kawasaki, K.; Murata, S.; Takahara,
Y.; Shintaku, H. J. Med. Chem. 1988, 31, 1738.
21. Shindo, H.; Takada, S.; Murata, S.; Masami, E.; Mat-
sushita, A. J. Med. Chem. 1989, 32, 1213.
48. Gaillard, P.; Carrupt, P.-A.; Testa, B.; Schambel, P. J.
Med. Chem. 1996, 39, 126.
49. Testa, B.; Carrupt, P.-A.; Gaillard, P.; Billois, F.; Weber,
P. Pharm. Res. 1996, 13, 126.
50. Stewart, J. J. P. J. Comput.-Aided Mol. Des. 1990, 4, 1.
51. Cramer R. D.; De Priest S. A.; Patterson D. E.; Hecht, P.
In 3-D QSAR in Drug Design: Theory, Methods and Applica-
tions; Kubinyi, H., Ed.; Escom: Leiden, 1993; p 443.
52. 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. J.
Med. Chem. 1990, 33, 2343.
53. Barnes, A. J.; Ratajczak, H. J. Mol. Struct. 1997, 416,
PAGES.
54. Steiner, T.; Desiraju, G. R. The Weak Hydrogen Bond: In
Structural Chemistry and Biology; Oxford University: Oxford
1999.
22. Wong, G.; Zi-Qiang, G.; Fryer, R. I.; Skolnick, P. Med.
Chem. Res. 1992, 2, 217.
23. Fryer, R. I.; Zhang, P.; Rios, R.; Gu, Z. Q.; Basile, A. S.;
Skolnick, P. J. Med. Chem. 1993, 36, 1669.
24. Schove, L. T.; Perez, J. I.; Maguire, P. A.; Loew, G. I.
Med. Chem. Res. 1994, 4, 307.
25. Savini, L.; Massarelli, P.; Pellerano, C.; Fiorini, I.; Bruni,
G.; Romeo, M. R. Il Farmaco 1993, 48, 65.
26. Savini, L.; Massarelli, P.; Corti, P.; Pellerano, C.; Bruni,
G.; Romeo, M. R. Il Farmaco 1993, 48, 1675.
27. Trapani, G.; Franco, M.; Latrofa, A.; Carotti, A.; Genchi,
G.; Serra, M.; Biggio, G.; Liso, G. Eur. J. Med. Chem. 1996,
31, 575 and references therein.
55. Doyon, J. B.; Jain, A. Org. Lett. 1999, 1, 183.
56. Deerhovanessian, A.; Doyon, J. B.; Jain, A.; Rablen, P.
R.; Sapse, A. M. Org. Lett. 1999, 1, 1360.
57. De Caceres, M., Villa, J., Lozano, J. J., Sanz, F. Bioinfor-
matics (in press).
58. Sanz, F.; Manaut, F.; Rodrõguez, J.; Lozoya, E.; Lopez de
Brinas, E. J. Comput.-Aided Mol. Des. 1993, 7, 337.
59. Rodrõguez, J.; Manaut, F.; Sanz, F. J. Comput. Chem.
1993, 14, 922.
28. Ananthan, S.; Clayton, D. S.; Ealick, E. S.; Wong, G.;
Evoniuk, E. G.; Skolknick, P. J. Med. Chem. 1993, 36, 479.
29. Pietrogrande, M. C.; Borea, P. A.; Biagi, G. L. J. Chro-
mat. 1988, 447, 404.
30. Leeson L. J. (Ciba-Geigy Corp.) US 4,758,427; 19 July
1988. Chem. Abstr. 1989, 110, 199215s.
31. Yokoyama N. (Ciba-Geigy A.-G.), Eur. Pat. Appl.
22,078; 07 Jan 1981. Chem. Abstr. 1981, 95, 7278s.
32. Mohler, H.; Okada, T. Life Sci. 1977, 20, 2101.
33. Yokoyama, N. US 4,312,870; 26 Jan 1982. Chem. Abstr.
1982, 96, 18123v.
34. Trapani, G.; Franco, M.; Latrofa, A.; Carotti, A.; Genchi,
G.; Serra, M.; Biggio, G.; Liso, G. Eur. J. Med. Chem. 1996,
31, 575.
60. Schmidt, M. W.; Baldridge, K. K.; Boatz, J. A.; Elbert, S.
T.; Gordon, M. S.; Jensen, J. H.; Koseki, S.; Matsunaga, N.;
Nguyen, K. A.; Su, S. J.; Windus, T. L.; Dupuis, M.; Mon-
tgomery, J. A. J. Comput. Chem. 1993, 14, 1347.
61. Wold, S.; Eriksson, L. In Chemometric Methods in Mole-
cular Design; van de Waterbeemd, H., Ed.; VCH: Weinheim,
1996; p 309.
62. Folkers, G.; Merz, A.; Rognan, D. In 3D QSAR in Drug
Design: Theory, Methods and Applications; Kubinyi, H., Ed.;
Escom: Leiden, 1993; p 583.
63. Wold, S. «PLS for Multivariate Linear Modeling» in ref.
61, p 195.
64. Carotti, A.; Altomare, C.; Cellamare, S.; Monforte, A.
M.; Bettoni, G.; Loiodice, F.; Tangari, N.; Tortorella, V. J.
Comput.-Aided Mol. Des. 1995, 9, 131.
35. Hadjipavlon-Litina, D.; Hansch, C. Chem. Rev. 1994, 94,
1483.
36. Hansch, C.; Leo, A. Exploring QSAR; American Chemi-
cal Society: Washington, DC, 1995; Vol. 1.
65. Altomare, C.; Cellamare, S.; Carotti, A.; Barreca, M. L.;
Chimirri, A.; Monforte, A-M.; Gasparrini, F.; Villani, C.;
Cirilli, M.; Mazza, F. Chirality 1996, 8, 556.
37. Hansch, C., Leo, A. and Hoeckman, D. Exploring
QSAR; American Chemical Society: Washington, DC, 1995;
Vol. 2.
66. Altomare, C.; Carrupt, P.-A.; Gaillard, P.; El Tayar, N.;
Testa, B.; Carotti, A. Chem. Res. Toxicol. 1992, 5, 366.
67. Carrieri, A.; Altomare, C.; Barreca, M. L.; Contento, A.;
Carotti, A.; Hansch, C. Il Farmaco 1994, 49, 573.
68. Kneubuhler, S.; Thull, U.; Altomare, C.; Carta, V.; Gail-
lard, P.; Carupt, P.; Carotti, A.; Testa, B. J. Med. Chem. 1995,
38, 3874.
69. Carrieri, A.; Brasili, L.; Leonetti, F.; Pigini, M.; Gian-
nella, M.; Bousquet, P.; Carotti, A. Bioorg. Med. Chem. 1997,
5, 843.
38. Mickelson, J. W.; Jacobsen, E. J.; Carter, D. B.; Im, H.
K.; Im, W. B.; Schreur, P. J. K. D.; Sethy, V. H.; Tang, A. H.;
McGee, J. E.; Petke, J. D. J. Med. Chem. 1996, 39, 4654.
39. Fryer, R. I.; Rios, R.; Zhang, P.; Gu, Z.; Wong, G.;
Basile, S. A.; Skolnick, P. Med. Chem. Res. 1993, 3, 122.
40. Jones, G. B.; Chapman, B. J. Synthesis 1995, 475.
41. Cramer, R. D.; Patterson, D. E.; Bunce, J. D. J. Am.
Chem. Soc. 1988, 110, 5959.
70. Pigini, M.; Bousquet, P.; Brasili, L.; Carrieri, A.;

444
L. Savini et al. / Bioorg. Med. Chem. 9 (2001) 431±444
Cavagna, R.; Dontenwill, M.; Gentili, F.; Giannella, M.;
Leonetti, F.; Piergentili, A.; Quaglia, W.; Carotti, A. Bioorg.
Med. Chem. 1998, 6, 2245.
71. Kim, K. H. In 3D QSAR in Drug Design: Theory, Meth-
ods and Applications; Kubinyi, H., Ed.; Escom: Leiden, 1993;
p 619.