Synthesis, biological evaluation, and pharmacophore generation of new pyridazinone derivatives with affinity toward α1- and α2-adrenoceptors

Article abstract of DOI:10.1021/jm010821u

A series of new pyridazin-3(2H)-one derivatives (3 and 4) were evaluated for their in vitro affinity toward both α₁- and α₂-adrenoceptors by radioligand receptor binding assays. All target compounds showed good affinities for the α₁-adrenoceptor, with K_i values in the low nanomolar range. The polymethylene chain constituting the spacer between the furoylpiperazinyl pyridazinone and the arylpiperazine moiety was shown to influence the affinity and selectivity of these compounds. Particularly, a gradual increase in affinity was observed by lengthening the polymethylene chain up to a maximum of seven carbon atoms. In addition, compound 3k, characterized by a very interesting α₁-AR affinity (1.9 nM), was also shown to be a highly selective α₁-AR antagonist, the affinity ratio for α₂- and α₁-adrenoceptors being 274. To gain insight into the structural features required for α₁ antagonist activity, the pyridazinone derivatives were submitted to a pharmacophore generation procedure using the program Catalyst. The resulting pharmacophore model showed high correlation and predictive power. It also rationalized the relationships between structural properties and biological data of, and external to, the pyridazinone class.

Full text of DOI:10.1021/jm010821u

2118
J . Med. Chem. 2001, 44, 2118-2132
Syn th esis, Biologica l Eva lu a tion , a n d P h a r m a cop h or e Gen er a tion of New
P yr id a zin on e Der iva tives w ith Affin ity tow a r d r1- a n d r2-Ad r en ocep tor s1
Roberta Barbaro,† Laura Betti,‡ Maurizio Botta,*,§ Federico Corelli,§ Gino Giannaccini,‡ Laura Maccari,§
Fabrizio Manetti,§ Giovannella Strappaghetti,*,† and Stefano Corsano†
Istituto di Chimica e Tecnologia del Farmaco, Universita` di Perugia, Via del Liceo 1, 06123 Perugia, Italy, Dipartimento
di Psichiatria, Neurobiologia, Farmacologia e Biotecnologie, Universita` di Pisa, Via Bonanno 6, 56126 Pisa, Italy, and
Dipartimento Farmaco Chimico Tecnologico, Universita` degli Studi di Siena, Via Aldo Moro, 53100 Siena, Italy
Received J anuary 15, 2001
A series of new pyridazin-3(2H)-one derivatives (3 and 4) were evaluated for their in vitro
affinity toward both R1- and R2-adrenoceptors by radioligand receptor binding assays. All target
compounds showed good affinities for the R1-adrenoceptor, with Ki values in the low nanomolar
range. The polymethylene chain constituting the spacer between the furoylpiperazinyl
pyridazinone and the arylpiperazine moiety was shown to influence the affinity and selectivity
of these compounds. Particularly, a gradual increase in affinity was observed by lengthening
the polymethylene chain up to a maximum of seven carbon atoms. In addition, compound 3k ,
characterized by a very interesting R1-AR affinity (1.9 nM), was also shown to be a highly
selective R1-AR antagonist, the affinity ratio for R2- and R1-adrenoceptors being 274. To gain
insight into the structural features required for R1 antagonist activity, the pyridazinone
derivatives were submitted to a pharmacophore generation procedure using the program
Catalyst. The resulting pharmacophore model showed high correlation and predictive power.
It also rationalized the relationships between structural properties and biological data of, and
external to, the pyridazinone class.
In tr od u ction
a small, but significant, incidence of side effects, such
as orthostatic hypotension, which is considered a critical
disadvantage in BPH patients. On the other hand, some
of these effects, such as a significant reduction of both
systolic and diastolic blood pressure in hypertensive
patients,6 and the favorable reduction in the level of
LDL cholesterol as well as serum triglyceride profiles,7
could be construed as beneficial.
Molecular cloning studies8,9 have shown that the R1-
and R2-adrenoceptors have many common features
which could reflect their similar mechanisms of action.
As a consequence of such similarities, synthetic com-
pounds with affinity toward R-AR are expected to
potentially bind to both R1 and R2 receptors. On the
other hand, many literature reports revealed that the
addition of arylpiperazinylalkyl side chains into differ-
ent heterocycles, like uracils or the pyrimido[5,4-b]-
indole moiety,10 provides compounds that effectively
lower blood pressure by antagonizing the R1-AR. More-
over, great attention has been paid to the compounds
containing a pyridazin-3(2H)-one moiety, due to their
potential biological activities as antihypertensive agents
(compound GYKI-12743 is an example).11 This literature
survey led to the suggestion that both the arylpiperazi-
nyl and the pyridazinone moiety are key elements for
R1-AR affinity.
The R1- and R2-adrenoceptors (termed R1-AR and R2-
AR, respectively, in the text) are members of the seven-
transmembrane-spanning domain group sharing the
common structural motif of seven putative R-helical
segments traversing the cell membrane.
It is now clear that R1-ARs are comprised of multiple
subtypes that have been identified by both pharmaco-
logical and binding studies. To date, they are classified
2
into R1A, R1B, and R1D and the corresponding cloned
counterparts termed R1a-, R1b-, and R1d-AR, respectively.3
In addition, the existence of an additional subtype (R1L),
characterized by a low affinity for prazosin, has been
postulated.
In a similar way, R2-ARs have been classified into four
subtypes, called R2A-R2D, respectively.2c
In recent years, the search for new selective R1-AR
antagonists has intensified, due to their importance in
the treatment of hypertension and of benign prostatic
hyperplasia (BPH). In fact, R1-AR blockers have been
employed in the treatment of BPH for more than two
decades, due to the significant improvements in symp-
toms and flow rates in patients with bladder outflow
obstruction.4,5 Nevertheless, the efficacy of R1-AR an-
tagonists in the treatment of BPH is balanced against
On the basis of this experimental evidence, in the
course of our studies in the field of new and potentially
selective R1-AR antagonists containing a pyridazin-
3(2H)-one ring, we have recently synthesized com-
pounds 1 and 212 (Figure 1 and Table 1) bearing an
arylpiperazinylalkyl chain at the 2-position of the py-
ridazinone moiety. While they all exhibited a good
affinity toward R1-AR, with values ranging from 0.6
* To whom correspondence should be addressed. M.B.: Dipartimento
Farmaco Chimico Tecnologico, Universita` degli Studi di Siena, Via Aldo
Moro, I-53100 Siena, Italy; telephone, +39 0577 234306; fax, +39 0577
234333; e-mail, botta@unisi.it. G.S.: Istituto di Chimica e Tecnologia
del Farmaco, Universita` di Perugia, Via del Liceo 1, I-06123 Perugia,
Italy; telephone, +39 075 5855136; fax, +39 075 5855129; e-mail,
noemi@unipg.it.
† Universita` di Perugia.
‡ Universita` di Pisa.
§ Universita` degli Studi di Siena.
10.1021/jm010821u CCC: $20.00 © 2001 American Chemical Society
Published on Web 05/26/2001

New Pyridazinone Derivatives
J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 13 2119
AR, to investigate the structure-affinity relationships
of these ligands, we have used the program Catalyst13
to develop a pharmacophore model for R1-AR antago-
nists. The calculated model, able to rationalize the
relationships between the chemical features of new
structures 3 and 4 and their binding affinity data at
R1-AR, consists of a positive ionizable portion, three
hydrophobic features, and a hydrogen bond acceptor
group. It shows a good statistical significance (r ) 0.92,
rmsd ) 0.89) and successfully predicts the affinities of
the molecules of, and external to, the training set.
F igu r e 1. General structures of compounds 1 and 2.
Ch em istr y
Ta ble 1. R1- and R2-Adrenergic Receptor Binding Affinities for
Pyridazin-3(2H)-one Derivatives 1 and 2
The target compounds 3 and 4 listed in Table 2 were
synthesized as outlined in Scheme 1. The first series of
compounds 3 were prepared starting from 4-chloro-5-
[4-(2-furoyl)piperazin-1-yl]pyridazin-3(2H)-one (5), ob-
tained by condensation of 4,5-dichloropyridazin-3(2H)-
one and 1-(2-furoyl)piperazine in DMF and potassium
carbonate or EtOH and Et3N according to the procedure
reported by Go`mez-Gil.14
Kia (nM)
compd
n
X
R1-ARb
R2-AR
R2/R1
1a
1b
2
3
4
5
6
7
2
3
4
5
6
7
2
3
4
5
6
7
8
2
3
4
5
6
7
8
OCH3 38.0 ( 5.0
1261.0 ( 210.0
254.0 ( 40.3
62.0 ( 8.0
33.2
34.8
103.3
5.4
10.2
20.0
20.5
46.5
86.7
20.0
16.5
9.3
25.5
17.0
53.0
3.8
2.3
3.2
6.5
5.0
7.8
4.0
6.4
6.0
2.7
4.0
OCH3
OCH3
7.3 ( 0.6
1cc
0.6 ( 0.1 (1.8)
OCH3 12.9 ( 1.5
1d
69.8 ( 9.2
1e
OCH3
OCH3
Cl
Cl
Cl
Cl
Cl
Cl
7.0 ( 0.8
71.8 ( 6.3
Alkylation of 5 with 1-(2-methoxyphenyl)-4-(3-chlo-
ropropyl)piperazine (6a )15 or 1-(2-chlorophenyl)-4-(3-
chloropropyl)piperazine (6b)15 in dry ethanol in the
presence of sodium hydroxide (method A) afforded
compounds 3a and 3b, respectively, in moderate yield.
Alternatively, 5 was alkylated with 1,2-dibromoethane
under phase transfer catalysis, according to the proce-
dure reported by Yamada,16 to give intermediate 7a ,
which in turn was converted to final compounds 3c and
3d by reaction with 1-(2-methoxyphenyl)piperazine (8)
or 1-(2-chlorophenyl)piperazine (9) (Na2CO3/isoamyl
alcohol, method B) in 20-30% overall yield. R,ω-Dibro-
moalkanes having four to seven methylene groups were
employed to prepare intermediates 7b -e (K2CO3/
acetone, method C), from which compounds 3e-l were
synthesized following method B. Analogously, compound
4d was prepared by direct alkylation of 10,17 according
to the procedure previously reported for the synthesis
of 4a -c,17 while pyridazinones 4e-l were obtained via
intermediates 11a -d , in turn prepared by treatment
of 10 with dibromoalkanes (method C).
1fd
2.3 ( 0.2 (19)
18.0 ( 2.0
45.8 ( 6.3
1g
370.0 ( 50.2
1h c
6.8 ( 0.6 (7.5) 316.0 ( 40.3
0.8 ( 0.1 (3.3)
7.0 ( 0.9
8.4 ( 1.2
15.0 ( 2.5
1ic
69.4 ( 6.0
138.0 ( 25.0
138.8 ( 20.3
139.0 ( 15.3
409.0 ( 35.7
1j
1k
1l
2a
OCH3 16.0 ( 1.7
OCH3 14.5 ( 1.3 (4.9) 245.0 ( 25.3
2bc
2c
OCH3
OCH3
OCH3
OCH3
OCH3
Cl
Cl
Cl
Cl
Cl
4.3 ( 0.3
3.9 ( 0.2
1.5 ( 0.1 (14)
1.4 ( 0.1 (9.7)
3.5 ( 0.4
58.8 ( 3.7 (39)
27.8 ( 3.0 (5.6) 219.0 ( 15.8
230.0 ( 20.0
15.0 ( 1.9
3.5 ( 0.4
4.6 ( 0.5
22.7 ( 3.0
292.3 ( 30.2
2d
2ed
2fc
2g
2h c
2id
2jd
10.0 ( 1.5 (2.8)
4.5 ( 0.2 (8.6)
4.2 ( 0.6
39.3 ( 5.3
29.0 ( 4.5
25.2 ( 3.8
7.4 ( 0.5
2k c
2l
2m
2n
Cl
Cl
2.7 ( 0.3
5.6 ( 0.7
22.8 ( 2.5
prazosin
rauwolscine
0.24 ( 0.05
4.0 ( 0.3
a
The Ki binding data were calculated as described in the
Experimental Section. The Ki values are means ( standard
deviation (SD) of separate series assays, each performed in
triplicate. Inhibition constants (Ki) were calculated according to
the equation of Cheng and Prusoff:36 Ki ) IC50/[1 + ([L]/Kd)], where
[L] is the ligand concentration and Kd its dissociation constant.
The Kd of [3H]prazosin binding to rat cortex membranes was 0.24
nM (R1), and the Kd of [3H]rauwolscine binding to rat cortex
The chemical and physical data of the new compounds
are reported in Table 3.
Molecu la r Mod elin g Stu d ies
P h a r m a cop h or e Gen er a tion . Our goal is to gain
insight into the structural factors responsible for R1
affinity, and to design new ligands with possibly higher
selectivity for the R1 receptor, with respect to the R2-
AR. In particular, here we report a study that applies
a ligand-based drug design (pharmacophore develop-
ment) method to rationalize the relationships between
new pyridazinone-arylpiperazine structures and their
affinity data at the R1 receptor. Experimentally deter-
mined affinities are used to derive a pharmacophore
model that describes the three-dimensional structural
properties required to have profitable interactions with
R1 receptor sites.
b
membranes was 4 nM (R2). In parentheses are estimated and
predicted affinity values calculated by Catalyst for the training
set and test set, respectively. c Compounds used to build the
d
training set. Compounds used to build the test set.
(compound 1c) to 58.8 nM (compound 2h ), a good
selectivity was found for compound 1i [R2/R1 ratio of 86.7
(Table 1)] and 1c [R2/R1 ratio of 103.3 (Table 1)].
The goal of this work was the synthesis of new
pyridazinone derivatives possibly characterized by high
affinity and selectivity toward the R1-adrenoceptors. In
this paper, we report the synthesis and biological
evaluation of pyridazinone derivatives 3 and 4 (Scheme
1 and Table 2) bearing at the 2-position an ortho-
substituted arylpiperazinylalkyl side chain and, at the
5- or 6-position, a furoylpiperazinyl moiety. In addition,
taking into account the fact that the new pyridazinones
showed a higher affinity toward R1-AR than toward R2-
Since we were primarily interested in the design of
new selective ligands for R1-AR, based on pyridazinone-
arylpiperazine lead compounds 1-4, and various other
molecules collected from the literature (Figure 2 and
Table 4), we decided to employ a computational ap-

2120 J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 13
Barbaro et al.
Sch em e 1a
a
Compounds: 3a , n ) 3, X ) OCH3; 3b, n ) 3, X ) Cl; 3c, n ) 2, X ) OCH3; 3d , n ) 2, X ) Cl; 3e, n ) 4, X ) OCH3; 3f, n ) 4, X )
Cl; 3g, n ) 5, X ) OCH3; 3h , n ) 5, X ) Cl; 3i, n ) 6, X ) OCH3; 3j, n ) 6, X ) Cl; 3k , n ) 7, X ) OCH3; 3l, n ) 7, X ) Cl; 4a , n ) 2,
X ) OCH3; 4b, n ) 2, X ) Cl; 4c, n ) 3, X ) OCH3; 4d , n ) 3, X ) Cl; 4e, n ) 4, X ) OCH3; 4f, n ) 4, X ) Cl; 4g, n ) 5, X ) OCH3; 4h ,
n ) 5, X ) Cl; 4i, n ) 6, X ) OCH3; 4j, n ) 6, X ) Cl; 4k , n ) 7, X ) OCH3; 4l, n ) 7, X ) Cl; 7a , n ) 2; 7b, n ) 4; 7c, n ) 5; 7d , n )
6; 7e, n ) 7; 11a , n ) 4; 11b, n ) 5; 11c, n ) 6; 11d , n ) 7. Reagents: (a) 6a or 6b, dry EtOH, NaOH; (b) BrCH2CH2Br, benzene, TBAB,
KOH; (c) 8 or 9, Na2CO3, isoamyl alcohol; (d) Br(CH2)nBr (n ) 4-7), K2CO3, acetone.
proach to build an “inclusive” pharmacophore for R1-
AR antagonists, without considering the additional
variable of the selectivity between the R1-AR subtypes.
In fact, we provided the program with the structure of
the compounds to be analyzed and their affinity data,
without any information about the selectivity of the R1-
AR subtypes.
cluded in the training set, taking into account the
findings showing that para substituents (particularly,
nitro and amino groups) are unfavorable to ligand-
receptor binding. As an example, 12 [Ki ) 0.21 nM
(Figure 2)] has been reported to be much more active
than its p-methoxy counterpart (Ki > 5000 nM).10
Moreover, a CoMFA study19 on hydantoin-phenylpip-
erazine derivatives has shown that the affinity of the
para-substituted compounds is modulated mainly by
steric factors. On the basis of this hypothesis, one may
account for the dramatic drop in affinity observed when
a nitro or amino group is introduced in the para position
with respect to a fluoro substituent that leads to a
moderate increase in affinity. In contrast, a different
CoMFA study recently published20 led to the conclusion
that the para position is forbidden to electronegative
substituents.
As a consequence of these choices, 10 additional
compounds [namely, 12-15, 18, and 20-24 (Figure 2
and Table 4)] have been added to the training set to
obtain a total number of 24 structures to be used for
pharmacophore generation.
Biological data associated with the training set,
expressed as Ki, were in the range between 0.21
(compound 12) and 2396 nM (compound 13). While the
biological activities of compounds 1 and 2 (Table 1) and
3 and 4 (Table 2) were experimentally determined (see
Experimental Section), data associated with the remain-
ing compounds were assembled from the literature
(Table 4) under the assumption that all these sub-
stances are acting through the same binding site.
Because no experimental data on the biologically
The training set for the pharmacophore development
has been chosen according to the Catalyst guidelines.18
Fourteen molecules of the whole set of pyridazinones
(namely, compounds 1c, 1h , 1i, 2b, 2f, 2h , 2k , 3e, 3g,
3l, 4a , 4e, 4h , and 4l) have been selected as a part of
the training set. They are characterized by affinity
values spanning ∼2.5 orders of magnitude, the mini-
mum value of 0.6 nM being associated with compound
1c and the maximum value of 184 nM being associated
with compound 4e. In addition, with the aim of covering
the optimal value of 4 orders of magnitude in affinity
required by the program, we investigated literature
reports to find some additional R1-AR antagonists with
the appropriate biological properties. Within the large
set of compounds that were found, to ensure the highest
degree of homogeneity of biological data with respect
to those of the pyridazinone derivatives, we have focused
our attention on such compounds whose antagonist
activity on R1-AR was tested on rat cortex homogenates
and expressed as Ki. Moreover, because we are mainly
interested in compounds sharing common chemical
features with pyridazinone derivatives 1-4, we have
decided to select only some of the arylpiperazine-bearing
molecules found in the literature. Between them, para-
substituted derivatives have not been arbitrarily in-

New Pyridazinone Derivatives
J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 13 2121
Ta ble 3. Chemical and Physical Data of the New Compounds
Ta ble 2. R1- and R2-Adrenergic Receptor Binding Affinities for
Pyridazin-3(2H)-one Derivatives 3 and 4
compd
X
n
formula
mp (°C)
yield (%)
Kia (nM)
3a
3b
3c
3d
3e
3f
3g
3h
3i
3j
3k
3l
4d
4e
4f
4g
4h
4i
OCH3
Cl
OCH3
Cl
OCH3
Cl
OCH3
Cl
OCH3
Cl
OCH3
Cl
Cl
OCH3
Cl
3
C27H33ClN6O4
183-185a
95-100b
214-216a
162-165a
106-110b
228-230b
62-65a
35
55
70
35
30
25
60
60
50
35
70
65
40
20
70
45
40
65
60
50
35
40
40
65
50
40
30
50
30
30
35
compd
R1-ARb
R2-AR
R2/R1
3
2
2
4
4
5
5
6
6
7
7
3
4
4
5
5
6
6
7
7
-
2
4
5
6
7
4
5
6
7
C
26H30Cl2N6O3
C26H31ClN6O4
25H28Cl2N6O3
C28H35ClN6O4
27H32Cl2N6O3
C29H37ClN6O4
28H34Cl2N6O3
C30H39ClN6O4
29H36Cl2N6O3
C31H41ClN6O4
3a
20.4 ( 2.1
3.5 ( 0.4 (11)
33.0 ( 5.3 (36)
3.9 ( 0.2
680.0 ( 120
85.0 ( 18.3
440.0 ( 36
150.0 ( 15.5
870 ( 165
280.0 ( 38.5
23.3 ( 3.2
18.0 ( 2.5
24.5 ( 3.6
66.0 ( 9.2
520.0 ( 4.2
36.2 ( 5.3
NDf
52.6 ( 3.7
631.0 ( 40.5
260.0 ( 24.5
790.0 ( 54.3
24.0 ( 3.8
255.0 ( 25.3
105.0 ( 20.3
100.0 ( 15.2
23.7 ( 2.5
96.0 ( 8.4
23.5 ( 3.0
33.3
24.3
13.3
38.4
26.0
14.7
12.3
5.0
C
3bc
3cc
3d
C
3ed
3f
33.5 ( 4.7 (15)
19.0 ( 1.7
C
40-45a
96-98b
3gd
3h
1.9 ( 0.1 (10)
3.6 ( 0.3
C
70-75b
128-130b
35-40b
3i
3j
4.1 ( 0.6
6.0
C
C
30H38Cl2N6O3
26H31ClN6O3
5.7 ( 0.5
11.6
273.7
3.1
-
1.5
9.8
20.3
4.4
3.0
5.8
102-105b
68-70c
3k
1.9 ( 0.1
C28H36N6O4
27H33ClN6O3
C29H38N6O4
28H35ClN6O3
C30H40N6O4
29H37ClN6O3
C31H42N6O4
3ld
11.5 ( 1.7 (19)
94.5 ( 8.5 (31)
34.0 ( 5.3
C
35-40b
4a d,e
4be
4ce
4d
OCH3
Cl
OCH3
Cl
OCH3
Cl
148-150d
48-52d
C
64.5 ( 5.7
12.8 ( 1.8
180.0 ( 15.3 (88)
7.9 ( 0.5 (10)
43.7 ( 5.2
13.3 ( 1.2 (34)
17.8 ( 2.5 (20)
5.9 ( 0.6
54.7 ( 6.7
5.6 ( 0.5 (6.3)
0.24 ( 0.05
55-58b
4j
4k
4l
C
70-74c
4ed
4fc
140-147d
C
C
C
C
C
C
C
C
C
C
30H39ClN6O3
13H13ClN4O3
75-78b
4g
5
-
214-216
4h d
7.9
5.6
4.0
1.7
7a
7b
7c
7d
7e
9a
9b
9c
9d
-
16H18BrClN4O3 130-135
18H22BrClN4O3 oil
15H24BrClN4O3 oil
16H26BrClN4O3 oil
17H28BrClN4O3 oil
17H21BrN4O3
18H23BrN4O3
18H25BrN4O3
4ic
-
4j
-
4k
-
4ld
4.2
-
prazosin
rauwolscine
-
oil
oil
oil
oil
4.0 ( 0.3
-
a
-
The Ki binding data were calculated as described in the
Experimental Section. The Ki values are means ( SD of separate
series assays, each performed in triplicate. Inhibition constants
(Ki) were calculated according to the equation of Cheng and
Prusoff:36 Ki ) IC50/[1 + ([L]/Kd)], where [L] is the ligand
concentration and Kd its dissociation constant. The Kd of [3H]pra-
zosin binding to rat cortex membranes was 0.24 nM (R1), and the
Kd of [3H]rauwolscine binding to rat cortex membranes was 4 nM
-
C19H27BrN4O3
a
b
As dihydrochloride monohydrate. As dihydrochloride. c As
d
trihydrochloride. As trihydrochloride monohydrate.
have chosen to analyze the three highest-scoring hy-
potheses among the 10 pharmacophore models gener-
ated by the program.
b
(R2). In parentheses are estimated and predicted affinity values
calculated by Catalyst for the training set and test set, respec-
All these three pharmacophores exhibit five chemical
features and are characterized by either similar com-
position or spatial location of the features (Table 5). The
major difference in composition arises from the replace-
ment of one of the hydrophobic features found in both
hypotheses 1 and 2 with a more specific aromatic ring
feature (see Table 5 for the composition of hypothesis
3). Nevertheless, the RA-HY1 assembly (these two
pharmacophore features were found at a 3.0 Å distance)
of hypothesis 3 constitutes a constraint for the matching
of the ligands with the pharmacophore. In fact, when
the program tries to fit the substituted phenyl ring of
the arylpiperazine moiety into the RA-HY1 system, the
only possible orientation is that where RA is mapped
by the phenyl group and HY1 by the methoxy substitu-
ent on the same ring. Thus, hypothesis 3 seems to be
too restrictive with respect to the orientation of the
substituted phenyl ring. Moreover, it was found to be
characterized by a higher cost and rmsd, and by a lower
correlation coefficient with respect to the first two
hypotheses (Table 5). These findings, combined with the
aim of broadening the possible orientations of the
substituted phenyl moiety into the above-mentioned
portion of the pharmacophore, led us to prefer hypoth-
eses 1 and 2 over hypothesis 3. Hence, only hypotheses
1 and 2, bearing the HY1 and HY2 features (instead of
RA and HY1 found in hypothesis 3) potentially mapped
by either the phenyl ring or its substituent(s) in
alternate conformations, were further investigated.
The vector describing HBA makes the difference
between hypotheses 1 and 2. The three-dimensional
tively. c Compounds used to build the test set. Compounds used
d
to build the training set. e Compounds described elsewhere by our
research group.17 f ND, not determined.
relevant conformations of the selected compounds (for
example, atomic coordinates derived from X-ray crystal-
lographic studies of their complexes with the putative
receptor) are available, we resorted to a molecular
mechanics approach to build the conformational models
to be used for pharmacophore generation. All the
conformers of each compound, within a range of 20 kcal/
mol with respect to the global minimum, have been
employed to derive a set of pharmacophore hypotheses.
P h a r m a cop h or e Descr ip tion . As a result of phar-
macophore generation, 100.9 bits as the total fixed cost
(ideal hypothesis) and 154.9 bits as the cost of the null
hypothesis have been found. The cost range over the
10 best generated hypotheses was 20.4 bits, suggesting
that there was a homogeneous set of hypotheses and
that the signal generated by this training set was
strong. Moreover, the fact that the hypotheses were
much closer to the fixed cost (i.e., the cost of hypothesis
1 is 112.9) meant that the signal could be interpreted.
The composition (in terms of chemical features that
are necessary for activity), ranking score, and statistical
parameters associated with the pharmacophore hypoth-
eses are reported in Table 5. Even if the highest scoring
hypothesis is usually the most likely to yield relevant
information about the pharmacophore elements of a set
of compounds, with the aim of extracting the maximum
amount of information from the computational run, we

2122 J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 13
Barbaro et al.
F igu r e 2. Structures of R1-adrenoceptor antagonists collected from the literature.
Ta ble 4. Actual and Calculated (Estimated or Predicted)
indicate a reliable ability to estimate the affinities of
the training set. Moreover, the cost difference between
hypothesis 1 and the null hypothesis is 42 bits, corre-
sponding to an ∼75% chance of true correlation in the
data.
R1-Adrenergic Receptor Binding Affinities for Arylpiperazines
Collected from the Literature, As Calculated by Catalyst
Ki (nM)
actual calcd
0.21 0.11
2395 1200
24.6
Ki (nM)
calcd
0.36
140
compd
compd
actual
12a,b
13a,c
14a,c
15a,c
16c,d
17c,d
18a,c
19c,d
20a,c
21a,b
22a,b
23a,b
24a,c
26b,d
27b,d
28b,d
33d
0.34
374
258
516
0.8
11.9
17.5
0.74
Figure 3 shows the five-feature hypothesis (character-
ized by three hydrophobic regions, a hydrogen bond
acceptor, and a positive ionizable group) with compound
12, the most active and small compound of the training
set, superposed on it. The affinity of 12 is correctly
estimated by the model, due to the fact that a complete
mapping of the molecule onto the pharmacophore is
possible. In detail, the tricyclic system fulfils both HY3
(with the phenyl group) and HBA (with the 1-carbonyl
group of the six-membered heterocyclic ring). Both HY1
and HY2 are mapped by the o-methoxyphenyl moiety,
and finally, the N-1 atom of the piperazine ring corre-
sponds to the PI feature of the hypothesis. The strong
antagonist effect of this molecule suggests that it
possesses many or all of the molecular features required
for activity. On the basis of the experimentally deter-
mined activity [0.21 nM (Table 3)] and taking into
account the fact that the pharmacophore model correctly
estimated the antagonist property of this compound
[0.11 nM (Table 3)], one may conclude that all the
pharmacophore features possibly account for the major
interactions between antagonists and the receptor.
15
11
11
380
620
0.46
190
6.7
9.8 (1.1)e
46.7
42.7
15.6
9.2
68
0.86
1000
147
7.9
2308
a
b
Compounds belonging to the training set. Compounds taken
from ref 22. c Compounds taken from ref 24. Compounds belong-
d
ing to the test set. e In parentheses is the affinity calculated using
the manipulated pharmacophore model (see ref 31).
coordinates of the HBA projecting point (corresponding
to the position of the hydrogen atom involved in
hydrogen bonding with the hydrogen bond acceptor of
the ligand) are in fact slightly different between the two
hypotheses. On the basis of the very similar composition
of the two hypotheses, hypothesis 1, characterized by
the best statistical parameters (Table 5), has been
chosen to represent “the pharmacophore model”.
The regression line (Figure 3) of “true” versus “esti-
mated” R1-AR antagonist affinity for the training set,
based on the lowest cost Catalyst-generated hypothesis,
exhibited a correlation coefficient r of 0.92, and a root-
mean-square deviation (rmsd) of 0.89. Comparison
between the estimated affinity of the compounds in the
training set relative to their experimentally measured
values shows, in the worst case, a <7-fold difference
and, in most cases, a <3-fold difference. These findings
Compound 1c, taken as a representative example of
all pyridazinone derivatives, overlaps the pharmaco-
phore model in a manner similar to that of 12. In fact,
the chemical functionalities of the hypothesis are all
matched by the chemical groups of the ligand (Figure
5). Particularly, the arylpiperazinyl moiety maps the
region where a cluster of features, known as crucial

New Pyridazinone Derivatives
J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 13 2123
Ta ble 5. Composition (Features), Cost (Bits), and Statistical Parameters (rmsd and Correlation) Associated with the 10 Best
Hypotheses (Pharmacophore Models)
hypothesis
feature 1a
feature 2a
feature 3a
feature 4a
feature 5a
costb
rmsd
correlation
1
2
3
4
5
6
7
8
9
HBA
HBA
HBA
HBA
HBA
HBA
HBA
HBA
HBA
HBA
HY1
HY1
HY1
HY1
HY1
HY1
HY1
HY1
HY1
HY1
HY2
HY2
HY2
HY2
HY2
HY2
HY2
HY2
HY2
HY2
HY3
HY3
RA
RA
RA
PI
PI
PI
PI
PI
PI
PI
PI
PI
PI
112.9
118.7
125.4
126.0
126.0
127.1
127.5
132.2
132.8
133.3
0.89
1.11
1.35
1.36
1.38
1.39
1.43
1.54
1.56
1.57
0.92
0.87
0.80
0.80
0.79
0.79
0.78
0.74
0.74
0.72
RA
HY3
HY3
RA
10
RA
a
b
HBA, hydrogen bond acceptor; HY, hydrophobic; PI, positive ionizable; RA, ring aromatic. Expressed in bits.
F igu r e 5. Compound 1c mapped to the pharmacophore
model. The sphere colors are described in the legend of Figure
4.
Sch em e 2. Structures of Compounds 1-4
F igu r e 3. Regression line for hypothesis 1 (the pharmaco-
phore model). The experimental vs estimated R1-adrenoceptor
binding affinity is reported for each member of the training
set.
all the orientations of 1c in the pharmacophore, it was
found that the remaining two features (HBA and HY3)
of the model are fulfilled by the ligand in two different
ways. In most cases, while the carbonyl oxygen (or,
sometimes, the N-1 atom) of the pyridazinone moiety
(C in Scheme 2) overlaps the HBA feature (correspond-
ing to the polar region in the De Marinis pharmacoph-
ore),21 the piperazine directly bound to the pyridazinone
ring reaches HY3 of the pharmacophore. Alternatively,
the HY3 feature is sometimes mapped by the ortho
substituent of the methoxyphenoxyethyl portion of the
ligand.
F igu r e 4. Compound 12, the most active of the whole training
set, mapped to the pharmacophore hypothesis. Pharmacophore
features are color-coded: cyan for hydrophobic regions (HY1-
HY3), green for a hydrogen bond acceptor (HBA), and red for
a positive ionizable feature (PI).
elements for interacting with the putative receptor, lies.
While the ortho-substituted phenyl ring (identified as
A in the sketch reported in Scheme 2, and corresponding
to the aromatic feature of the De Marinis pharmaco-
phore model)21 occupies both HY1 and HY2, the piper-
azinyl N-4 atom is located inside the positive ionizable
feature PI, also identified by all the previously proposed
pharmacophore models for R1-AR antagonists. Among
From the analysis of the fitting mode of 1c and other
pyridazinone derivatives onto the best hypothesis, it
turned out that this class of compounds is characterized
by an “extra-size” portion of the molecule with respect

2124 J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 13
Barbaro et al.
to 12 and other small compounds, exceeding the model
itself. In the first orientation described above, the extra-
size portion of 1c is represented by the methoxyphe-
noxyethyl portion of the molecule (Figure 5), while in
the second model, the exceeding moiety is the piperazi-
nyl ring bound to the pyridazinone. Similar consider-
ations can be made for compounds 3 and 4. In fact, the
furoyl moiety has been found to alternatively represent
the chemical feature interacting with HY3 and the
extra-size portion of the ligands. On the other hand,
small and very active ligands have been found to
perfectly overlap the pharmacophore model (Figure 4).
As an additional example, the affinity of compound 25,
found to be ∼1 order of magnitude more active than 4b,
4c, and 4e,17 has been predicted to be 5 nM, in good
agreement with experimental data. In conclusion, it is
possible that the exceeding moiety of the pyridazinone
ligands 1-4 could represent a molecular portion able
to contact particular regions of the receptor assigned
to modulate the biological properties of these com-
pounds.
program in a similar way into the pharmacophore, and
any fluctuation of their affinity data is well accounted
for on the basis of how well each compound overlays
onto each of the pharmacophore features.
Finally, compounds characterized by an octyl spacer
(2g and 2n ) exhibit a decreased affinity. We may
interpret this observation by assuming that the confor-
mation of the spacer, forced to bear the pyridazinone
and piperazine rings at the PI-HBA distance, causes
an intrachain steric hindrance between C-2 and C-6,
thus reducing the overall affinity.
The substitution pattern on the phenyl ring of the
arylpiperazine moiety is a crucial element for which to
account with the purpose of rationalizing the relation-
ships between the structural properties and affinities
of the studied compounds. In particular, pharmacologi-
cal tests highlighted the fact that the presence of a
methoxy group or a chlorine atom at the ortho position
of the arylpiperazine fragment is almost equivalent. In
fact, while some subsets are characterized by decreased
affinity corresponding to the presence of a chloride
instead of a methoxy group (2a -f vs 2h -m , 1e,f vs 1k ,l,
and 3g,i,k vs 3h ,j,l), in the others the general trend goes
in the opposite sense (1a -d vs 1g-j, 3a ,c,e vs 3b,d ,f,
and 4a ,c,e,g,i,k vs 4b,d ,f,h ,j,l). The effect of a substitu-
tion of the phenyl ring is worthy of consideration also
from the computational point of view. In particular, the
o-methoxy group seems to be the optimum substituent
that is able to interact with HY1 or HY2 of the
pharmacophore model. Alternatively, an o-chloro or
m-chloro (see compounds 16, 17, and 19) substituent is
also profitable for interaction with the model. In addi-
tion, HY1 or HY2 (depending on the orientation of the
phenyl ring) is able to locate a hydrophobic group larger
than a methoxy substituent (data not shown), according
to the literature where o-isopropyl-substituted com-
pounds with high R1-AR affinity have been reported.22
From the above considerations, the three-dimensional
structural properties required for an ideal R1-antagonist
can be summarized as follows. (i) A basic, positive
ionizable nitrogen accessible to the receptor and easily
protonable at physiological pH is required. This is the
structural feature interacting with an aspartate side
chain in the third transmembrane (TM) helix of the
G-protein-coupled receptors (GPCRs). (ii) The ortho- or
meta-substituted phenyl ring occupies both the HY1 and
HY2 features of the pharmacophore, located 6.69 and
6.17 Å, respectively, from PI. In addition, the phenyl
ring and each of its substituent can alternatively match
HY1 or HY2 with a profitable fit value. These findings
suggest that both HY1 and HY2 might constitute a
unique and large hydrophobic pocket accommodating
the ortho(meta)-substituted phenyl ring linked to the
piperazine. (iii) A polar group (the pyridazinone ring in
compounds 1-4) at the other end of the molecule
relative to the arylpiperazine moiety corresponds to a
hydrogen bond acceptor (HBA, located 5.62 Å from PI)
substituent. These results are in good agreement with
the generally accepted statement that an R1-AR antago-
nist interacts with the corresponding receptor by a
three-pocket binding pathway;21,23,24 while the most
important structural element of the R1-AR inhibitors is
a protonated nitrogen atom interacting with the car-
boxylate group of an aspartic residue, the remaining
The distance between the A-B and C-D portions of
the ligand structures (Scheme 2) is influenced by the
length of the polymethylene chain acting as a spacer
linking B and C of pyridazinone derivatives 1-4. Both
the intrinsic high conformational flexibility of the Csp3-
Csp3 polymethylene sequence and the length of the
whole chain are crucial parameters influencing the
goodness of fit of each compound to the chemical
features of the pharmacophore model. In fact, while the
PI, HY1, and HY2 features are mapped by most ligands,
HBA and HY3 can be mapped only depending on the
length of the polymethylene chain. In particular, n ) 2
compounds are estimated to be weakly active due to the
difficulty in mapping the HBA feature. This is in
agreement with the experimental data of 1a , 1g, 2a ,
2h , 3c, and 4b, the affinity values of each compound
being the highest within the corresponding subset.
Alternative orientations of compounds bearing the eth-
ylene chain are characterized by an enhanced mapping
to HBA, with the piperazinyl ring in B undergoing a
deep conformational rearrangement leading to a very
highly strained structure of the ligand.
The lengthening of the ethyl chain to a propyl spacer
led to several compounds characterized by a higher
conformational flexibility. This structural variation is
responsible for the enhanced fit to the pharmacophore
with a consequent improvement of the calculated affin-
ity values. In fact, among compounds 1-4, the propyl
spacer-bearing derivatives are all more active than the
corresponding ethylene analogues (Tables 1 and 2).
When 4 < n < 7, by variation of dihedral angles of
the polymethylene chain, the program is able to gener-
ate conformations of the most active ligands character-
ized by the optimum distance between the N-4 atom of
the piperazine ring in B and the pyridazinone, allowing
for a good overlapping to the features of the pharma-
cophore model. Thus, the distance between the A-B and
C-D moieties of the ligands has been identified by the
program as a key element in determining the goodness
of fit to the pharmacophore features and, consequently,
the estimated/predicted affinity values with respect to
those experimentally determined. Moreover, molecules
with slight structural differences are oriented by the

New Pyridazinone Derivatives
J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 13 2125
molecular portions of the ligand fit two binding pockets
almost symmetrically located with respect to the aspar-
tate. (iv) Finally, an additional hydrophobic pharma-
cophore element (HY3) accommodates (portions of) the
terminal moieties linked to the pyridazinone. HY3 was
found 9.78 Å from PI.
of the piperazine ring and/or a different orientation of
the substituted phenyl ring (i.e., allowing for a match
on both HY1 and HY2, but leading to the lack of a fit
into HY3) have been considered unprofitable and, then,
investigated no further. The affinity of 17 for the R1-
AR has been predicted by the model to be 9.2 nM, in
good agreement with the experimentally determined
value of 15.6 nM.
P h a r m a cop h or e Va lid a tion . (a ) F isch er Test.
Due to the relative small range between the cost for an
ideal versus null hypothesis (54 bits), and due moreover
to the small difference between the cost of hypothesis 1
and null hypothesis (42 bits), special care was taken to
test the model for chance correlation. In fact, the 42 bits
difference in the cost of the hypothesis representing the
pharmacophore model and the null hypothesis can only
ensure a 75% chance of true correlation. As a conse-
quence, the pharmacophore model corresponding to
hypothesis 1 was evaluated for the statistical signifi-
cance using a randomization trial procedure derived
from the Fischer method.25 Experimental activities of
the training set were scrambled 19 times to obtain
spreadsheets with randomized activity data. Nineteen
hypothesis generation experiments were performed
using the scrambled training sets, and among the 190
resulting hypotheses (10 for each computational run),
none was found with a lower cost than hypothesis 1.
Thus, there was at least a 95% chance that hypothesis
1 represented true correlation in the data.
Compound 18, lacking the chlorine atom at the meta
position on the phenyl ring, is characterized by an
orientation into the pharmacophore model very similar
to that of 17. Also in this case, the three-dimensional
features of the fitted conformers bearing the phenyl ring
orthogonal to the piperazine plane, combined with the
matching of the methyl group of the isoxazole ring into
HY3, led to the impossibility of mapping the HY1
feature of the model. In fact, the program was unable
to find any conformation characterized by the superpo-
sition of the o-methoxyphenyl moiety to both HY1 and
HY2. As a consequence, the expected decrease in the
antagonistic affinity of 18 with respect to 17 has been
estimated with a value of 68 nM versus an actual
affinity of 147.1 nM.
The lengthening of the polymethylene chain to a
propyl moiety, combined with a twisted conformation
of the piperazine ring, led to the complete mapping of
19 into the pharmacophore model. A reorientation of the
phenyl group of 19 has been highlighted, the o-meth-
oxyphenyl substituent assuming the usual orientation
to fulfill both HY1 and HY2. As a consequence of this
enhanced fit to the pharmacophore, the predicted af-
finity for 19 was 0.86 nM versus an actual value of 7.9
nM.
(b) Test Set. The validity and predictive power of
hypothesis 1 were further assessed using molecules with
known affinities for R1-AR, outside the training set. This
set of compounds (the test set) consists of pyridazinone
derivatives and other compounds taken from the litera-
ture. According to the same guidelines that were fol-
lowed to obtain the training set, the biological activities
of compounds belonging to the test set have been
determined as the ability to displace [3H]prazosin from
R1 binding sites on rat cerebral cortex. The predicted
affinities, calculated by Catalyst on the basis of the
pharmacophore model, are reported in Tables 1, 2, and
4.
Finally, the shortening of the polymethylene chain to
a methylene spacer led to a partial superposition of 20
onto the pharmacophore, the methyl group of the
isoxazole ring being unable to reach HY3. Thus, the
estimated affinity of 20 was 1000 nM versus an actual
value of 2308 nM.
Compound 13, the open analogue of 20, fits into the
model in an orientation similar to that of 20 and is also
unable to map HY3. As a consequence, a poor affinity
for 13 has been estimated by the program (1200 vs 2395
nM).
In particular, 17 is characterized by molecular frag-
ments fulfilling all the features of the pharmacophore
model. In fact, the 3-methyl and 7-carbonyl groups of
the isoxazolo[3,4-d]pyridazin-7(6H)-one moiety map HY3
and HBA, respectively. Moreover, while the positive
ionizable group PI is represented by the N-1 atom of
the piperazine ring, the m-chlorophenyl substituent of
the molecule corresponds to the HY1 and HY2 features
of the model. It is interesting to note that the o-methoxy
group, usually mapping to HY1 or HY2 in the pyridazi-
none derivatives, lies in an “empty” region of the space,
due to the particular conformation assumed by the
phenylpiperazine moiety. In fact, the piperazine ring is
characterized by a N-1-N-4 diequatorial chair confor-
mation with the plane of the C-2-C-3-C-5-C-6 system
perfectly orthogonal to the phenyl ring. In this three-
dimensional context, a conformation where the methoxy
group is allowed to map one of the HY1 and HY2
hydrophobic features is forbidden because of steric
clashes with the axial protons at both the C-2 and C-6
atoms of the piperazine. When some literature reporting
an orthogonal conformation of the piperazine and the
aryl ring as being highly profitable for the R1-adreno-
ceptor affinity is considered,22 alternative conformations
Estimated (14 and 15) or predicted (16) affinities (15,
11, and 11 nM, respectively) of the open analogues arise
from how well each compound is able to fit into HY3 as
a consequence of the transformation of the condensed
isoxazole ring into both the 4-amino and 5-acetyl side
chains on the pyridazinone ring. In fact, the conforma-
tional rigidity of the isoxazole ring serves as a constraint
to orient the methyl group of 17-19 toward the HY3
feature of the pharmacophore. When the isoxazole was
broken, the conformationally free acetyl side chain
undergoes a conformational rearrangement to avoid
steric clashes between its methyl group and the amino
moiety at the 4-position. As a consequence, the mea-
sured dihedral angle between the plane passing through
the acetyl moiety and the plane of the pyridazinone was
∼40°, only leading to a partial fit into HY3.
The phenyl ring directly linked to the pyridazinone
moiety of compounds 13-20 failed to contact the phar-
macophore model, due to the fact that this group is

2126 J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 13
Barbaro et al.
located in the region of space where the extra-size
portion of compounds 1-4 has been found.
is the actual value), mainly due to the adamantyl moiety
(instead of the ethyl group of 23) that is able to reach
HY3.
A comparison of the estimated/predicted Ki values of
18 versus 14, 17 versus 16, and 19 versus 15 shows a
good correlation with actual affinities. The affinity of
14, where a good fit into HY1 was restored with respect
to 18, and a very partial fit into HY3 was found, has
been estimated to be 15 nM instead an actual value of
24.6 nM. Moreover, decreased values calculated for 16
and 15 with respect to 17 and 19 derive from a partial
mapping of HY3 by the methyl group of the acetyl
moiety at the 5-position of the pyridazinone ring.
As a summary of the validation step based on the
prediction of affinity data for a set of compounds
external to the training set, the pharmacophore model
has accommodated much of the SAR for R1-AR antago-
nists belonging to the arylpiperazine class. In fact, the
importance of the ortho substituent has emerged, as
well as the influence of the terminal heterocyclic moiety
in defining the R1-AR antagonist activity. Moreover, the
distance between the arylpiperazine and the terminal
fragment was demonstrated to be crucial for affinity.
In summary, the pharmacophore model accounts for
the trend in the biological data associated with com-
pounds 17-20 and their open analogues 13-16. Com-
pounds with a methylene spacer (20 and 13) exhibit a
markedly decreased antagonistic activity with respect
to compounds with ethyl or propyl chains. This decrease
is reflected in the estimated/predicted Ki values gener-
ated by means of our pharmacophore model where both
20 and 13 are completely unable to match the HY3
feature. Moreover, while compounds 14-16 are char-
acterized by a partial fit to HY3, the conformational
rigidity of 17-19 restores the ability of the methyl group
of the isoxazole ring to locate itself within the HY3
feature of the pharmacophore model.
(c) Da ta ba se Sea r ch . With the purpose of further
assessing the validity of the model, the five-feature
pharmacophore model was used as a three-dimensional
query to perform a database search to find other
structural motifs that fulfill the functional and spatial
constraints imposed by the model itself. The goal of this
search was the identification of compounds previously
reported in the literature and characterized by R1-AR
antagonist affinity. In particular, NCI (National Cancer
Institute), Maybridge, MiniBioByte, and Sample data-
bases (provided by MSI along with Catalyst 4.5),
containing approximately 178 600 compounds, were
searched using the pharmacophore model as a three-
dimensional query and applying a molecular weight
cutoff set to 700. As a result, 486 compounds (∼0.27%
of the total) have been retrieved from the databases and
investigated to find known R1-AR antagonists. Within
them, the pharmacophore model has identified com-
pounds exhibiting R1-AR affinity, and belonging to
different structural classes with respect to that of the
training set. As an example, the widely used antide-
pressant drug trazodone (2-{3-[4-(3-chlorophenyl)-1-
piperazinyl]propyl}-1,2,4-triazolo[4,3-a]pyridin-3-one, 29,
Figure 6) has been recently reported to have an affinity
of 281 nM (IC50 value) expressed as the ability to
displace [3H]prazosin from R1-rat cortical membranes.26
The affinity of 29 for R1-AR has been predicted by the
model to be 220 nM, with a partial mapping of the
m-chlorophenyl moiety to HY1 and HY2, and a poor fit
of the condensed pyridine ring to the HY3 feature.
Moreover, an open analogue of 29, etoperidone (30,
Figure 6),27 characterized by an anti-R1-adrenergic
activity, has been also identified during the database
search.
Several other compounds such as carpipramine (31,
Figure 6)28 and ergotamine (32, Figure 6),29 reported
to be blockers of R1-adrenoceptors in the brain and at
the periphery, respectively, have been also found in the
databases.
Interestingly, prazosin (33, Figure 6), a compound
with a very high affinity for R1-AR, has been discarded
because of its very poor fit to the pharmacophore
hypothesis. This is due to the fact that the N-1 atom of
the quinazoline ring has been drawn in the database
as an unprotonated, uncharged atom. Thus, since
Catalyst does not identify this atom as a positive
ionizable feature, it tries to fit one of the nitrogen atoms
of the piperazine ring to the PI feature of the pharma-
cophore model without finding any profitable mode of
interaction. A previous study30 highlighted the fact that
the N-2 of prazosin is unlikely to be protonated, the
The model also accounts for the contribution of the
terminal heterocyclic moiety (opposite the piperazine
ring) to the affinity of each compound. As an example,
compound 21, whose terminal fragment linked through
the polymethylene spacer to the piperazine ring is
structurally different from that of 12 (estimated value
of 0.11 nM vs an actual affinity of 0.21 nM), shows
identical chemical features interacting with the phar-
macophore. In particular, the condensed phenyl ring
and the carbonyl group lie within HY3 and HBA,
respectively, in a manner similar to that of the con-
densed phenyl ring and the 1-carbonyl group of 12. PI,
HY1, and HY2 are matched by the N-1 atom of the
piperazine ring, and by the o-methoxyphenyl moiety,
respectively. Compound 21 exhibits an estimated affin-
ity of 0.36 nM versus an actual value of 0.34 nM.
Compound 22, bearing a pyrimidine ring instead of the
o-methoxyphenyl substituent at the N-4 atom of the
piperazine ring, is completely unable to map HY2. The
model accounts for this modification with a predicted
affinity of 140 nM versus an actual value of 374 nM.
In a similar way, the terminal fragment of NAN-190
(26, Figure 2), represented by a cyclic imide, possesses
chemical substituents that are able to map both the HY3
and HBA features, with a predicted affinity of 0.46 nM
versus an actual value of 0.8 nM. When the condensed
phenyl ring was omitted, compound 27 (Figure 2), in
the same orientation as 26, cannot reach HY3, but one
of the carbonyl groups is able to perfectly map HBA.
As a consequence, the calculated affinity for this com-
pound is 190 nM versus an actual value of 11.9 nM.
Similarly, 23 is characterized by an ethyl spacer that
is able to bring the carbonyl group of the aliphatic amide
to interact with HBA, but the terminal ethyl group is
far away from HY3 with a consequent predicted affinity
of 380 nM versus an actual value of 258 nM. Compound
28 is predicted to have an affinity of 6.7 nM (17.5 nM

New Pyridazinone Derivatives
J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 13 2127
trigonal planar geometry, in accordance with a plausible
conjugation effect of the piperazine nitrogens with both
the quinazoline and furoyl moieties, and with X-ray
crystallographic data reported in the Cambridge Struc-
ture Database32 for N-arylpiperazine derivatives. In
addition, the same structural features have been con-
sidered in recent three-dimensional QSAR studies of
piperazines acting as R1-AR and 5HT1A antagonists.20,33
Resu lts a n d Discu ssion
Tables 1 and 2 report the R1- and R2-adrenoceptor
binding affinity, expressed as Ki values, of pyridazinone
derivatives 1-4 containing an (ortho-substituted phe-
nyl)piperazinylalkyl moiety.
Being aware that both the R1- and R2-AR are hetero-
genic species, but taking into account the fact that our
major interest is the synthesis of selective R1-AR
antagonists (with respect to R2-AR antagonists), we
make the following comments about the structural
features of compounds 1-4 that only refer to the native
R1- and R2-adrenoceptors, and not to their relative
subtypes.
From the binding data, it can be seen that most of
these compounds demonstrated moderate to high affin-
ity for R1- and R2-AR binding sites. Moreover, all the
pyridazinone derivatives (1-4) exhibited higher affinity
toward the R1-AR than toward the R2-AR, but poor R2/
R1 selectivities. The only exception was found for
compounds 1c and 3k with R2/R1 ratios corresponding
to 103 and 274, respectively.
With respect to the phenylpiperazine substitution,
compounds with a methoxy group at the ortho position
display the best R2/R1 selectivity profile of the overall
pyridazinone class, 1c and 3k being the most selective
compounds.
F igu r e 6. Structure of R1-adrenoceptor antagonists identified
using the pharmacophore model as a three-dimensional query
in a database search.
The effect of the replacement of the o-methoxy group
with a chlorine atom was difficult to rationalize. The
introduction of a chlorine atom instead of the methoxy
group led to a decreased affinity for the R1-AR in several
subsets (2h -n , 3h ,j,l, and 1k ,l). On the other hand, the
opposite trend was found for compounds 3 and 4 bearing
a furoyl moiety instead of the phenoxyethyl group. In
fact, the affinities for the R1-AR underwent an improve-
ment by replacement of the methoxy with a chlorine
group, with 3g,i,k constituting an exception. Increased
affinity was also observed for compounds 1a -d with
respect to chlorinate analogues 1g-j. On the basis of
these findings and taking into account reports10,34
showing that the introduction of these two substituents
(MeO and Cl) to the ortho position of the phenyl ring of
the arylpiperazine chain may produce potent R1-AR
antagonists, we suggest that a methoxy or chlorine
group is characterized by the same interaction pattern
with the R1-AR. These results are consistent with the
molecular modeling studies showing either the methoxy
or the chloro substituent interacting equally weighted
with the same hydrophobic portion of the pharmaco-
phore model built for R1-AR antagonists. As a conse-
quence, the diverse affinity for the R1-AR highlighted
for compounds 1-4 is mainly dependent on the terminal
fragment linked through the spacer to the piperazine
ring, and on the length of the polymethylene chain (the
spacer) itself.
F igu r e 7. Prazosin (33) mapped to the pharmacophore model.
The sphere colors are described in the legend of Figure 4.
energy difference for protonation at N-1 and N-2 being
∼20 kcal/mol. In addition, the best superposition of N-1-
protonated prazosin to the pharmacophore hypothesis
led to a predicted affinity of 9.8 nM instead of the actual
value of 0.74 nM.
The orientation of prazosin allows for a full matching
of HY1 and HY2 with the methoxy substituents, the
mapping of PI by the protonated nitrogen, the identi-
fication of the carbonyl oxygen as a HBA, and, finally,
the partial mapping of HY3 by the C-4-C-5 portion of
the furan ring (Figure 7).31 Interestingly, the conformer
used by the program to obtain the best superposition is
characterized by the piperazine nitrogen atoms in a
The length of the spacer linking the phenylpiperazine
and pyridazinone moieties is of great importance for R1

2128 J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 13
Barbaro et al.
affinity and selectivity. As a general trend, the maxi-
mum affinity for both the R1- and R2-AR is reached with
n ranging from 4 to 7. Reduction of the alkyl chain to n
) 2 or 3 causes a relevant decrease in affinity in both
receptors. This decrease is especially marked in deriva-
tives bearing an ethylene or propylene spacer, for R2
receptor binding.
HBA with the pyridazinone. These three features (HBA,
PI, and the HY1-HY2 assembly) constitute the mini-
mum requirements for having R1-AR antagonist activity,
according to previously published literature. In addition,
as reported above, different affinities also derived from
variable matching to HY3 were considered as an ad-
ditional pharmacophore feature with respect to the De
Marinis model.21
The most active R1-AR antagonists within the furoyl
series are characterized by a five-carbon (3g and 3h ) or
seven-carbon (3k and 4l) atom spacer. An exception to
these general observations was found in 3d and 3b
where an ethylene or propylene spacer led to an affinity
of 3.9 or 3.5 nM, respectively. On the other hand, in the
phenoxyethyl series, the most active compounds are 1c
and 1i (n ) 4, 6-substituted pyridazinone), 2e and 2l
(n ) 6, 5-substituted pyridazinone), and, finally, 2f and
2m (n ) 7, 5-substituted pyridazinone).
These findings led to the suggestion that the substi-
tution pattern on the pyridazinone ring, combined with
the extension of the alkyl spacer, had some relevant
effects on the binding to the R-adrenoceptors. When a
phenoxyethyl piperazine moiety is linked to the 5-posi-
tion of the pyridazinone ring, a gradual increase in
affinity was observed by lengthening the polymethylene
chain up to a maximum of six (2e and 2l) or seven (2f
and 2m ) carbon atoms. On the contrary, the highest
affinity for the 6-substituted pyridazinone derivatives
has been found when the polymethylene chain is
characterized by four carbon atoms (1c and 1i).
These findings led to the conclusion that the substitu-
tion pattern on the pyridazinone ring influences the
affinity, which is also strictly dependent on the poly-
methylene chain length.
Com p a r ison betw een th e P r op osed P h a r m a -
cop h or e a n d P r eviou sly P u blish ed Mod els for r1-
AR. Recently published literature described the three-
dimensional theoretical models of the complexes between
compounds 13-20 and the three R1-AR subtypes, as
determined by molecular dynamics calculations. Having
no structural information (i.e., distances between the
key structural elements of these complexes) in our
hands, we could make only qualitative comparisons
between these structures and the pharmacophore model
proposed in this paper. According to the DeMarinis R1-
AR model, each of the three complexes is characterized
by three major binding subsites accommodating the
inhibitors. In detail, an expected hydrogen bond between
the basic nitrogen atom of the ligand and an aspartate
lying in the third transmembrane domain (TM3) has
been found. This interaction is also accounted for by the
PI feature of our pharmacophore model. Moreover, an
additional common structural feature is a hydrophobic
pocket accommodating the substituted phenyl ring of
the arylpiperazine moiety of inhibitors. This cavity is
mainly formed by residues of TM4-TM6, and corre-
sponds to both the HY1 and HY2 features of our
pharmacophore model. Finally, a second binding pocket
where, in turn, either the pyridazinone or the isoxazolo-
pyridazinone moieties of the ligand lies, allows for
different pathways of interaction with the inhibitor, for
each of the three R1-AR subtypes. In particular, a
hydrogen bond involving both the carbonyl group of the
pyridazinone moiety of the inhibitors and an arginine
of TM7 has been evidenced in the DeBenedetti model
for the R1D subtype. The same interaction was also found
in the pharmacophore model proposed by us, the car-
bonyl group of the pyridazinone ring of compounds 1-4
being the hydrogen bond acceptor.
Analogous considerations can be made for the series
bearing a 1-(2-furoyl)piperazinyl fragment linked to the
5-position of the pyridazinone ring (compounds 3a -l).
A gradual increase in affinity was observed by length-
ening the polymethylene spacer between the pyridazi-
none and arylpiperazine moieties of the molecule. While
the highest affinity was found for compound 3g (1.88
nM) characterized by a five-carbon alkyl chain and
bearing a methoxy group at the ortho position of the
phenylpiperazine fragment, the best selectivity is as-
sociated with compound 3k (affinity of 1.95 nM, R2/R1
ratio of 274, seven-carbon alkyl chain). When the
furoylpiperazine moiety is linked to the 6-position of the
pyridazinone ring (compounds 4a -l), a decreased af-
finity was observed toward R1-AR with respect to
5-substituted pyridazinone derivatives 3a -l. The high-
est affinity was observed with a six- or seven-carbon
spacer. The chlorine atom at the ortho position of the
arylpiperazine group led to an increased affinity with
respect to a methoxy substituent.
Moreover, compounds 13-20, some of which have
been included in the hypothesis generation step, are
generally characterized by an affinity for the R1D-AR
that is higher than that for both the R1A- and R1B-AR.
In a similar way, evaluation of the R1-AR subtype
inhibitor properties for some of the compounds 1-4
(namely, 2a and 4a -c) showed affinity values for the
We may interpret these observations by analyzing the
fitting properties of each series to the pharmacophore
model. In fact, the position of both HBA and PI
constrained the spatial location of both the pyridazinone
ring (whose carbonyl group was identified as the hy-
drogen bond acceptor feature of the pharmacophore for
R1-AR antagonists) and the N-1 atom of the piperazine
moiety (the positive ionizable feature), at a defined
distance. This distance constraint is well satisfied in
most compounds as a consequence of the conformational
freedom allowed by the polymethylene chain, and any
fluctuation in affinity within the different series is the
consequence of how well the compounds fit the most
important pharmacophore elements around PI: HY1
and HY2 with the ortho-substituted phenyl ring and
R
1D- and R1A-AR that were higher than that for R1B-AR.
All these considerations led to the suggestion that the
pharmacophore model proposed in this paper might
incorporate the most relevant structural features of the
R1D-adrenoceptor, as suggested by a qualitative com-
parison with a theoretical model of the R1D-AR reported
in the recent literature. On the other hand, the phar-
macophore model is distinguishable from that reported
as the R1D-AR model by, at least, two features. The first
one is the inability to account for the interaction of the

New Pyridazinone Derivatives
J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 13 2129
phenyl ring directly linked to the pyridazinone with the
receptor, reported as an important interaction key in
determining the affinity for both R1A- and R1D-ARs. The
second difference is related to the molecular portion
corresponding to the isoxazole ring in compounds 17-
20 or to the 4-acetyl and 5-amino side chains in
compounds 13-16. In fact, on the basis of how well
these moieties can fit the HY3 feature, the pharma-
cophore model is able to account for different R1-AR
affinity values of compounds 13-20. On the contrary,
no discussion has been carried out on the isoxazole ring
by DeBenedetti, probably because isoxazolopyridazino-
nes 17-20 and trisubstituted pyridazinones 13-16
“give quite similar interaction patterns with each of the
three R1-AR subtypes”.
shape of the ligand, and on the basis of the use of a
spatial region forbidden to the ligand (excluded volume
features), we are currently carrying out some compu-
tational experiments to define the overall three-dimen-
sional properties of the binding site for R1-antagonists.
Exp er im en ta l Section
Ch em istr y. Melting points were determined using a Kofler
hot-stage apparatus and are uncorrected. 1H NMR spectra
were recorded on a Bruker AC 200 MHz instrument in the
solvent indicated below. Chemical shift values (parts per
million) are relative to tetramethylsilane used as an internal
reference standard. Elemental analyses are within (0.4% of
theoretical values. Precoated Kiesegel 60 F254 plates (Merck)
were used for TLC. The corresponding hydrochlorides were
prepared by bubbling dry HCl into the dry solution of the
compound.
Con clu sion s
Syn th eses. Specific examples presented below illustrate
general synthetic methods A-C.
We have prepared a new series of (ortho-substituted
phenyl)piperazinylalkyl pyridazinone derivatives struc-
turally related to previously described R1- and R2-AR
antagonists. The pharmacophore model generation pro-
tocol provided by Catalyst has been successfully applied
to some of these compounds to generate a three-
dimensional model of the pharmacophore features re-
sponsible for R1-AR antagonist activity. The resulting
model provided significant correlation between the
chemical structures of the studied compounds and their
biological data. Some structural features of the py-
ridazinone derivatives have been demonstrated to be
very important for the affinity and selectivity for the
R1-AR with respect to the R2-AR. (i) The methoxy group
at the ortho position of the phenylpiperazine moiety led
to the best R1 affinity-selectivity profile. In fact, two
o-methoxyphenylpiperazinyl derivatives (namely, 1c
and 3k ) exhibited high affinity (0.6 and 1.9 nM, respec-
tively) and interesting selectivity (103 and 274 as the
R2/R1 ratios, respectively) values. These findings led to
the conclusion that the ortho position could play a
crucial role in the improvement of the R1-AR antagonist
properties in terms of affinity and selectivity. As a
consequence, new substituents at the ortho position of
the phenyl ring are currently under investigation. (ii)
The polymethylene chain linking the arylpiperazine
moiety to the pyridazinone ring is a critical structural
feature in determining affinity and selectivity properties
of compounds 1-4. In fact, as demonstrated by modeling
studies, the alkyl moiety serves as a spacer to bring both
the pyridazinone and the arylpiperazine moiety to the
optimal distance to interact with a hydrogen bond donor
and a hydrophobic pocket of the putative receptor,
respectively. This is in accordance with the De Marinis
pharmacophore model, where an aromatic moiety is
present instead of the more general hydrophobic phar-
macophore feature of our model. (iii) Fitting of com-
pounds bearing a long alkyl spacer to the computation-
ally built pharmacophore model highlighted the fact
that a portion of the polymethylene spacer or the
terminal aromatic fragment (the phenoxyethyl or the
furoyl moiety) could represent a portion of the molecule
that is unable to interact with any pharmacophore
feature. Moreover, the substitution pattern of the phenyl
ring in the arylpiperazine moiety is worthy of further
investigation. On the basis of these considerations,
taking into account the fact that Catalyst does allow
for pharmacophore generation based on the overall
4-Ch lor o-5-[4-(2-fu r oyl)p ip er a zin -1-yl]p yr id a zin -3(2H)-
on e (5). A mixture of 4,5-dichloropyridazin-3(2H)-one (4.5 g,
27 mmol) and 1-(2-furoyl)piperazine (6.3 g, 27 mmol) in DMF
in the presence of dry potassium carbonate was heated at
reflux for 6 h. The same compound can be also obtained with
refluxing ethanol and Et3N for 20 h. The mixture was
evaporated under reduced pressure and purified by chroma-
tography on a silica gel column eluting with an EtOH/CH2Cl2
mixture (6:94) to give a 40% yield of an ivory solid: mp 214-
1
216 °C; H NMR (CDCl3) δ 3.50-3.60 (m, 4H, H-pip), 3.90-
4.10 (m, 4H, H-pip), 6.50 (dd, J ) 1.7 and 3.4 Hz, 1H, H-fur),
7.10 (dd, J ) 0.8 and 3.4 Hz, 1H, H-fur), 7.50 (dd, J ) 0.8 and
1.7 Hz, 1H, H-fur), 7.70 (s, 1H, H6-pyrid), 12.00 (s, 1H, H2-
pyrid).
2-(2-Br om oet h yl)-4-ch lor o-5-[4-(2-fu r oyl)p ip er a zin -1-
yl]p yr id a zin -3(2H)-on e (7a ). This compound was prepared
starting from 5 (3.08 g, 10 mmol) with 1,2-dibromoethane (5.63
g, 30 mmol) in benzene (50 mL). Potassium hydroxide pellets
(5.61 g, 10 mmol) and tetrabutylammonium bromide (3.2 g,
10 mmol) were then added to the mixture, using the method
reported by Yamada.16 The residue was purified by chroma-
tography on a silica gel column eluting with an EtOH/CH2Cl2
mixture (6:94) to give a 40% yield of a yellow solid: mp 130-
1
135 °C; H NMR (CDCl3) δ 3.45-3.55 (m, 4H, H-pip), 3.75 (t,
J ) 7 Hz, 2H, CH2), 3.90-4.00 (m, 4H, H-pip), 4.55 (t, J ) 7
Hz, 2H, CH2), 6.45 (dd, J ) 1.7 and 3.4 Hz, 1H, H-fur), 7.00
(dd, J ) 0.8 and 3.4 Hz, 1H, H-fur), 7.50 (dd, J ) 0.8 and 1.7
Hz, 1H, H-fur), 7.60 (s, 1H, H-pyrid).
Meth od A Exa m p le. 2-{3-[4-(2-Meth oxyp h en yl)p ip er -
a zin -1-yl]p r op yl}-4-ch lor o-5-[4-(2-fu r oyl)p ip er a zin -1-yl]-
p yr id a zin -3(2H)-on e (3a ). To 20 mL of dry ethanol were
added sodium hydroxide (0.4 g, 10 mmol) and 5 (3.08 g, 10
mmol). This mixture was refluxed for 30 min, and then
compound 6a (2.68 g, 10 mmol) was summed up before
dissolution in dry ethanol. The mixture was again refluxed
under stirring for 15 h. After evaporation under reduced
pressure, the residue was purified by chromatography on a
silica gel column eluting with an EtOH/CH2Cl2 mixture (5:
95) to give a 35% yield of a dense oil: 1H NMR (CDCl3) δ 2.10
(quint, J ) 7 Hz, 2H, CH2), 2.50 (t, J ) 7 Hz, 2H, CH2), 2.65-
2.80 (m, 4H, H-pip), 3.10-3.20 (m, 4H, H-pip), 3.40-3.50 (m,
4H, H-pip), 3.85 (s, 3H, OCH3), 3.95-4.10 (m, 4H, H-pip), 4.25
(t, J ) 7 Hz, 2H, CH2), 6.50 (dd, J ) 1.7 and 3.4 Hz, 1H, H-fur),
6.80-7.00 (m, 4H, H-arom), 7.10 (dd, J ) 0.8 and 3.4 Hz, 1H,
H-fur), 7.50 (dd, J ) 0.8 and 1.7 Hz, 1H, H-fur), 7.60 (s, 1H,
H-pyrid). For the corresponding hydrochloride: mp 183-185
°C. Anal. (C27H33ClN6O4‚2HCl‚H2O) C, H, N.
Meth od B Exa m p le. 2-{4-[4-(2-Meth oxyp h en yl)p ip er -
a zin -1-yl]bu tyl}-6-[4-(2-fu r oyl)p ip er a zin -1-yl]p yr id a zin -
3(2H)-on e (4e). This compound was obtained from 11a with
1-(2-methoxyphenyl)piperazine in isoamyl alcohol and dry
sodium carbonate. 4e was purified by cromatography on a
silica gel column eluted with an EtOH/CH2Cl2 mixture (11:
89) to give a 20% yield of a dense oil: 1H NMR (CDCl3) δ 1.70-

2130 J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 13
Barbaro et al.
1.90 (m, 4H, 2CH2), 2.60-2.80 (m, 2H, CH2), 2.90-3.00 (m,
4H, H-pip), 3.10-3.40 (m, 8H, H-pip), 3.85 (s, 3H, OCH3),
3.95-4.00 (m, 4H, H-pip), 4.10 (t, J ) 7 Hz, 2H, CH2), 6.50
(dd, J ) 1.7 and 3.4 Hz, 1H, H-fur), 6.80-7.05 (m, 6H, 4H-
arom, 1H-pyrid, 1H-fur), 7.15 (d, J ) 9 Hz, 1H, H-pyrid), 7.50
(dd, J ) 0.8 and 1.7 Hz, 1H, H-fur). For the corresponding
hydrochloride: mp 68-70 °C. Anal. (C28H36N6O4‚3HCl) C, H,
N.
Meth od C Exa m p le. 2-(4-Br om obu tyl)-4-ch lor o-5-[4-(2-
fu r oyl)p ip er a zin -1-yl]p yr id a zin -3(2H)-on e (7b). A mixture
of 5 (6.2 g, 20 mmol) with 1,4-dibromobutane (5.1 g, 30 mmol)
and dry potassium carbonate (4.1 g, 30 mmol) in 50 mL of
acetone was refluxed under stirring for 20 h. The filtered
residue was evaporated under reduced pressure and purified
by chromatography on a silica gel column eluting with an
EtOH/CH2Cl2 mixture (5:95) to give a 65% yield of a dense
oil: 1H NMR (CDCl3) δ 1.60-2.00 (m, 4H, 2CH2), 3.30-3.55
(m, 6H, H-pip, CH2), 3.80-3.90 (m, 4H, H-pip), 4.15 (t, J ) 7
Hz, 2H, CH2), 6.45 (dd, J ) 1.7 and 3.4 Hz, 1H, H-fur), 7.00
(dd, J ) 0.8 and 3.4 Hz, 1H, H-fur), 7.50 (dd, J ) 0.8 and 1.7
Hz, 1H, H-fur), 7.60 (s, 1H, H-pyrid).
Ra d ioliga n d Bin d in g Assa ys. The pharmacological profile
of compounds 1-4 was evaluated for their affinities for R1-
and R2-adrenoceptors by determining for each compound its
ability to displace [3H]prazosin or [3H]rauwolscine from specific
binding sites on rat cerebral cortex. Ki values were determined
on the basis of three competition binding experiments in which
seven drug concentrations, run in triplicate, were used.
r1-Recep tor Bin d in g. Rat cerebral cortex was homog-
enized in 20 volumes of ice-cold 50 mM Tris-HCl buffer at pH
7.7 containing 5 mM EDTA (buffer T1) in an ultra-turrax
homogenizer. The homogenate was centrifuged at 48000g for
15 min at 4 °C. The pellet (P1) was suspended in 20 volumes
of ice-cold buffer T1. It was then homogenized and centrifuged
at 48000g for 15 min at 4 °C. The resulting pellet (P2) was
frozen at -80 °C until the time of assay.
Pellet P2 was suspended in 20 volumes of ice-cold 50 mM
Tris-HCl buffer at pH 7.7 (T2 buffer), and the R1 binding assay
was performed in triplicate by incubating at 25 °C for 60 min
in 1 mL of T2 buffer containing aliquots of the membrane
fraction (0.2-0.3 mg of protein) and 0.1 nM [3H]prazosin in
the absence or presence of unlabeled 1 µM prazosin. The
binding reaction was terminated by filtering through What-
man GF/C glass fiber filters under suction and washing twice
with 5 mL of ice-cold Tris buffer. The filters were placed in
scintillation vials, and 4 mL of Ultima Gold MN Cocktail
solvent scintillation fluid (Packard) was added. The radioactiv-
ity was assessed with a Packard 1600 TR scintillation counter.
The level of specific binding was obtained by subtracting the
level of nonspecific binding from the total level of binding and
was approximated to be 85-90% of the total level of binding.
r2-Recep tor Bin d in g. Cerebral cortex was dissected from
rat brain, and the tissue was homogenized in 20 volumes of
ice-cold 50 mM Tris-HCl buffer at pH 7.7 containing 5 mM
EDTA, as reported above (buffer T1). The homogenate was
centrifuged at 48000g for 15 min at 4 °C. The resulting pellet
was diluted in 20 volumes of 50 mM Tris-HCl buffer at pH
7.7 and used in the binding assay.
experiment was carried out to determine the effect of the
solvent on binding. Protein estimation was based on a reported
method,35 after solubilization with 0.75 N sodium hydroxide,
using bovine serum albumin as a standard. The concentration
of tested compound that produces 50% inhibition of specific
[3H]prazosin or [3H]rauwolscine binding (IC50) was determined
by log-probit analysis with seven concentrations of the dis-
placer, each performed in triplicate. Inhibition constants (Ki)
were calculated according to the equation36 Ki ) IC50/[1 + ([L]/
Kd)], where [L] is the ligand concentration and Kd its dissocia-
tion constant. The Kd for [3H]prazosin binding to cortex
membranes was 0.24 nM (R1), and the Kd for [3H]rauwolscine
binding to cortex membranes was 4 nM (R2).
Com p u ta tion a l Meth od s. All calculations and graphic
manipulations were performed on a Silicon Graphics O2
workstation by means of the Catalyst 4.5 software package.
Fourteen compounds belonging to a new class of pyridazinone
derivatives (Tables 1 and 2) and 10 additional compounds
taken from the literature (namely, compounds 12-15, 18, and
20-24, reported in Figure 2), all characterized by R1-AR
antagonistic activity, have been used in this study to build a
24-compound training set.
All the compounds of the training and test set were built
using the two- and three-dimensional sketcher of the program.
A representative family of conformations were generated for
each molecule using the poling algorithm and the “best quality
conformational analysis” method37 (CHARMm force field).38,39
The “best quality” is considered as the method of choice also
for systems with flexible rings. In this context, among Catalyst
users, this conformational search procedure seems to have
caused some problems about the axial/equatorial distribution
of substituents in satured six-membered rings.40 In addition,
distorted molecular geometries have been occasionally found.40
In our case, almost a balanced distribution of axial and
equatorial (representing the most favorable conformations)
substituents on the piperazine ring was highlighted. Moreover,
the structures of the generated conformers showed a quasi-
coplanar orientation of the piperazine ring with respect to the
pyridazinone moiety directly linked to it, in agreement with
some literature reports for piperazine derivatives.20 In addi-
tion, twisted or even orthogonal conformations of the pipera-
zine with respect to the phenyl ring of the arylpiperazinyl
moiety of the ligands, demonstrated to be highly profitable for
R1-AR antagonism, were also found.22
Conformational diversity was emphasized by selection of the
conformers that fell within the 20 kcal/mol range above the
lowest-energy conformation that was found.
The training set of molecules, with their associated confor-
mational models, were submitted to Catalyst hypothesis
generation with the aim of building potential pharmacophore
models. The chemical functions used in this generation step
are only surface-accessible features (i.e., the environment
around the feature is sufficiently uncrowded). In particular,
hydrogen bond acceptor lipid (HBA), hydrogen bond donor
(HBD), positive ionizable (PI), ring aromatic (RA), and hydro-
phobic (HY) features41 have been included in the calculations.
Hydrogen bond acceptor lipid includes basic nitrogen not
considered in the hydrogen bond acceptor feature.
The binding assay was perfomed in triplicate, by incubating
aliquots of the membrane fraction (0.2-0.3 mg of protein) in
Tris-HCl buffer at pH 7.7 with approximately 2 nM [3H]-
rauwolscine in a final volume of 1 mL. Incubation was carried
out at 25 °C for 60 min. Nonspecific binding was defined in
the presence of 10 µM rauwolscine. The binding reaction was
concluded by filtration through Whatman GF/C glass fiber
filters under reduced pressure. Filters were washed four times
with 5 mL aliquots of ice-cold buffer and placed in scintillation
vials. The level of specific binding was obtained by subtracting
the level of nonspecific binding from the total level of binding
and approximated to be 85-90% of the total level of binding.
The receptor-bound radioactivity was assessed as described
above.
A couple of constraints have been imposed on the hypothesis
generator. (i) Only hypotheses with five features have been
kept because of the molecular flexibility and functional
complexity of the training set. (ii) On the basis of the literature
reporting a positive ionizable group as a critical key for
antagonistic activity,23,30,42,43 the program was forced to include
such a feature in the composition of hypotheses.
A database search has been performed using the “best
flexible search” method provided by Catalyst. It is allowed to
flex the conformational models stored in the databases to map
the three-dimensional query represented by the pharmaco-
phore hypothesis (hypothesis 1). Only molecules mapping all
the features of the hypothesis were considered as hits. A total
of 486 compounds were identified and ranked by descending
predicted affinities for the R1-AR.
Compounds were dissolved in buffer or DMSO (2% buffer
concentration) and added to the assay mixture. A blank

New Pyridazinone Derivatives
J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 13 2131
(21) De Marinis, R. M.; Wise, M.; Hieble, J . P.; Ruffolo, R. R., J r.
Structure-Activity Relationships for R-1 Adrenergic Receptor
Agonists and Antagonists. In The Alpha-1 Adrenergic Receptor;
Ruffolo, R. R., J r., Ed.; Humana Press: Clifton, NJ , 1987; pp
211-265.
(22) Mokrosz, M. J .; Paluchowska, M. H.; Charakchieva-Minol, S.;
Bien, S. Effect of Structural Modifications in 1-Arylpiperazine
Derivatives on R1-Adrenoceptor Affinity. Arch. Pharm. Pharm.
Med. Chem. 1997, 330, 177-180.
Ack n ow led gm en t. Financial support provided by
the Italian “Ministero dell’Universita` e della Ricerca
Scientifica e Tecnologica” (MURST), and Italian Re-
search National Council (CNR) “Progetto Finalizzato
Biotecnologie” (CNR Target Project on “Biotechnology)
is acknowledged.
(23) (a) Bremner, J . B.; Coban, B.; Griffith, R. Pharmacophore
Development for Antagonists at R1 Adrenergic Receptor Sub-
types. J . Comput.-Aided Mol. Des. 1996, 10, 545-557. (b)
Griffith, R.; Bremner, J . B. Modelling of Adrenoceptor Ligand
Targets Based on Novel Medium- or Macro-sized Fused Nitrogen
Heterocyclic Systems. J . Comput.-Aided Mol. Des. 1999, 13, 69-
78. (c) Bremner, J . B.; Coban, B.; Griffith, R.; Groenewoud, K.
M.; Yates, B. F. Ligand Design for R1 Adrenoceptor Subtype
Selective Antagonists. Bioorg. Med. Chem. 2000, 8, 201-214.
(24) Montesano, F.; Barlocco, D.; Dal Piaz, V.; Leonardi, A.; Poggesi,
E.; Fanelli, F.; De Benedetti, P. G. Isoxazolo-[3,4-d]-pyridazin-
7-(6H)-ones and their Corresponding 4,5-Disubstituted-3-(2H)-
pyridazinone Analogues as New Substrates for R1-Adrenoceptor
Selective Antagonists: Synthesis, Modeling, and Binding Stud-
ies. Bioorg. Med. Chem. 1998, 6, 925-935.
(25) Fischer, R. The Principle of Experimentation, Illustrated by a
Psycho-Physical Experiment. In The Design of Experiments;
Hafner Publishing Co.: New York, 1966; Chapter II.
(26) Giannangeli, M.; Cazzolla, N.; Luparini, M. R.; Magnani, M.;
Mabilia, M.; Picconi, G.; Tomaselli, M.; Baiocchi, L. Effect of
Modification of the Alkylpiperazine Moiety of Trazodone on
5HT2A and R1 Receptor Binding Activity. J . Med. Chem. 1999,
42, 336-345.
(27) Costa, A.; Martignoni, E.; Blandini, F.; Petraglia, F.; Genazzani,
A. R.; Nappi, G. Effects of Etoperidone on Sympathetic and
Pituitary-adrenal Responses to Diverse Stressors in Humans.
Clin. Neuropharmacol. 1993, 16, 127-138.
(28) Setoguchi, M.; Sakamori, M.; Takehara, S.; Fukuda, T. Effects
of Iminodibenzyl Antipsychotic Drugs on Cerebral Dopamine and
Alpha-adrenergic Receptors. Eur. J . Pharmacol. 1985, 112, 313-
322.
(29) (a) Roquebert, J .; Grenie, B. Alpha 2-Adrenergic Agonist and
Alpha 1-Adrenergic Antagonist Activity of Ergotamine and
Dihydroergotamine in Rats. Arch. Int. Pharmacodyn. Ther. 1986,
284, 30-37. (b) Bonuso, S.; Di Stasio, E.; Marano, E.; Covelli,
V.; Testa, N.; Tetto, A.; Buscaino, G. A. The Antimigraine Effect
of Ergotamine: a Role for Alpha-adrenergic Blockade? Acta
Neurol. 1994, 16, 1-10.
(30) De Benedetti, P. G.; Menziani, M. C.; Rastelli, G.; Cocchi, M.
Molecular Orbital Study of the Nitrogen Basicity of Prazosin
Analogues in Relation to their R1-Adrenoceptor Binding Affinity.
THEOCHEM 1991, 233, 343-351.
(31) Considering the 13-fold difference in the predicted/actual affinity
of prazosin 29, mainly due to a partial mapping into HY3, we
have increased by 0.5 Å the radius of the sphere representing
this hydrophobic feature. As a result, an improved fit (from 9.0
to 10.4) and a predicted affinity value of 1.1 nM, in good
agreement with the experimental data (0.74 nM), were found.
Although this finding is to be considered as a preliminary result,
it suggested that a hydrophobic region larger than HY3 might
improve the goodness of ligand affinity versus the receptor.
Evaluation of the influence of the HY3 size on the affinity is
currently under investigation.
(32) Allen, F. H.; Kennard, O. 3D Search and Research Using the
Cambridge Structural Database. Chem. Des. Autom. News 1993,
8, 31-37.
(33) Gaillard, P.; Carrupt, P. A.; Testa, B.; Schambel, P. Binding of
Arylpiperazines, (Aryloxy)propanolamines, and Tetrahydropy-
ridylindoles to the 5-HT1A Receptor: Contribution of the Molec-
ular Lipophilicity Potential to Three-Dimensional Quantitative
Structure-Affinity Relationship Models. J . Med. Chem. 1996, 39,
126-134.
Refer en ces
(1) A portion of this research was presented at the XVIth Interna-
tional Symposium on Medicinal Chemistry, Bologna, Italy,
September 18-22, 2000 (Abstract PC-14).
(2) (a) Ford, A. P. D. W.; Williams, T. J .; Blue, D. R.; Clarke, D. E.
R1-Adrenoceptor Classification: Sharpening Occam’s Razor.
Trends Pharmacol. Sci. 1994, 15, 167-170. (b) Hieble, J . P.;
Bylund, D. B.; Clarke, A. E.; Eikenberg, D. C.; Langer, S. Z.;
Lefkowitz, R. J .; Minneman, K. P.; Ruffolo, R. R., J r. Interna-
tional Union of Pharmacology X. Recommendation for Nomen-
clature of R1-Adrenoceptors: Consensus Update. Pharmacol.
Rev. 1995, 47, 266-270. (c) Hieble, J . P.; Ruffolo, R. R., J r.
Subclassification and Nomenclature of R1- and R2-Adrenoceptors.
Prog. Drug Res. 1996, 47, 81-130.
(3) Price, D. T.; Schwinn, D. A.; Lomasney, J . W.; Allen, L. F.; Caron,
M. G.; Lefkowitz, R. J . Identification, Quantitation, and Local-
ization of mRNA for Three Distinct R1 Adrenergic Subtypes in
Human Prostate. J . Urol. 1993, 250, 546-551.
(4) Lepor, H. The Emerging Role of Alpha Antagonists in the
Therapy of Benign Prostatic Hyperplasia. J . Androl. 1991, 12,
389-394.
(5) Caine, M. Alpha-Adrenergic Blockers for the Treatment of
Benign Prostatic Hyperplasia. Urol. Clin. North Am. 1990, 17,
641-649.
(6) Gifford, R. W., J r. The fifth report of the J oint National
Committee on Detection, Evaluation, and Treatment of High
Blood Pressure: Insights and Highlights from the Chairman.
Cleveland Clinic J ournal of Medicine 1993, 60, 273-277.
(7) Ames, R. P.; Kiyasu, J . Y. R1-Adrenoceptor Blockade with
Doxazosin in Hypertension: Effects on Blood Pressure and
Lipoproteins. J . Clin. Pharmacol. 1989, 29, 123-127.
(8) Strader, C. D.; Sigal, I. S.; Dixon, R. A. Structural Basis of
â-Adrenergic Receptor Function. FASEB J . 1989, 3, 1825-1832.
(9) Strader, C. D.; Sigal, I. S.; Dixon, R. A. Genetic Approaches to
the Determination of Structure-Function Relationships of G
Protein-Coupled Receptors. Trends Pharmacol. Sci. 1989 (Dec.
Suppl.), 26-30.
(10) Russo, F.; Romeo, G.; Guccione, S.; De Blasi, A. Pyrimido[5,4-
b]indole derivatives. 1. A new Class of Potent and Selective R1
Adrenoceptor Ligands. J . Med. Chem. 1991, 34, 1850-1854.
(11) J aszlits, L.; Rabloczky, G.; Csokas, G.; Bodi, I.; Kurthy, M.;
Horvath, E.; Kovacs, A.; J ednakovits, A.; Aranyi, P.; Matyus, P.
GYKI-12 743, a Novel Antihypertensive Compound with Special
R-Adrenergic Blocking Profile. Acta Physiol. Hung. 1990, 75
(Suppl.), 155-156.
(12) Corsano, S.; Strappaghetti, G.; Barbaro, R.; Giannaccini, G.;
Betti, L.; Lucacchini, A. Synthesis of New Pyridazinone Deriva-
tives and their Affinity Towards R1-R2-Adrenoceptors. Bioorg.
Med. Chem. 1999, 7, 933-941.
(13) Catalyst 4.5; Molecular Simulations, Inc.: San Diego.
(14) Gomez-Gil, M. C.; Gomez-Parra, V.; Sanchez, F.; Torres, T. 4,5-
Disubstituted Pyridazin-3(2H)-ones as Hypotensive Drugs: In-
corporation of a â-Blocking Moiety. Arch. Pharm. 1986, 319, 60-
64.
(15) Bourdais, J . Alkylation of Piperazine in N,N-Dimethylforma-
mide. Bull. Soc. Chim. Fr. 1968, 8, 3246-3249.
(16) Yamada, T.; Tsukamoto, Y.; Shimamura, H.; Banno, S.; Sapo,
M. Eur. J . Med. Chem. 1983, 18, 209-214.
(17) Corsano, S.; Strappaghetti, G.; Leonardi, A.; Rhazri, K.; Barbaro,
R. New 3(2H)-Pyridazinone Derivatives: Synthesis and Affinity
Towards R1AR Subtypes and 5HT1A Receptors. Eur. J . Med.
Chem. 1997, 32, 339-342.
(18) The rules for picking training sets are reported at http://
www.msi.com/support/catalyst/hypogen.html.
(34) Chern, J .-W.; Yen, M.-H.; Lu, G.-Y.; Shiau, C.-Y.; Lai, Y.-J .;
Chan, C.-H. Studies on Quinazoline. 5. 2,3-Dihydroimidazo[1,2-
c]quinazoline Derivatives: A Novel Class of Potent and Selective
R1-Adrenoceptor Antagonists and Antihypertensive Agents. J .
Med. Chem. 1993, 36, 2196-2207.
(19) Lopez-Rodriguez, M. L.; Rosado, M. L.; Benhamu, B.; Morcillo,
M. J .; Fernandez, E.; Schaper, K.-J . Synthesis and Structure-
Activity Relationships of a New Model of Arylpiperazines. 2.
Three-Dimensional Quantitative Structure-Activity Relation-
ships of Hydantoin-Phenylpiperazine Derivatives with Affinity
for 5-HT1A and R1 Receptors. A Comparison of CoMFA Models.
J . Med. Chem. 1997, 40, 1648-1656.
(20) Cinone, N.; Carrieri, A.; Strappaghetti, G.; Corsano, S.; Barbaro,
R.; Carotti, A. Comparative Molecular Field Analysis of Some
Pyridazinone-containing R1-Antagonists. Bioorg. Med. Chem.
1999, 7, 2615-2620.
(35) Lowry, O. H.; Rosenbrough, N. J .; Farr, A. L.; Randall, R. J .
Protein Measurement with the Folin Phenol Reagent. J . Biol.
Chem. 1951, 193, 265-275.
(36) Cheng, Y. C.; Prusoff, W. H. Relationship between the Inhibition
Constant (Ki) and the Concentration of Inhibitor which Causes
50% Inhibition (IC50) of an Enzymatic Reaction. Biochem.
Pharmacol. 1973, 22, 3099-3108.
(37) (a) Smellie, A.; Teig, S. L.; Towbin, P. Poling: Promoting
Conformational Coverage. J . Comput. Chem. 1995, 16, 171-187.
(b) Smellie, A.; Kahn, S. D.; Teig, S. L. An Analysis of Confor-
mational Coverage 1. Validation and Estimation of Coverage.

2132 J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 13
Barbaro et al.
J . Chem. Inf. Comput. Sci. 1995, 35, 285-294. (c) Smellie, A.;
Kahn, S. D.; Teig, S. L. An Analysis of Conformational Coverage
2. Applications of Conformational Models. J . Chem. Inf. Comput.
Sci. 1995, 35, 295-304.
(41) Greene, J .; Kahn, S. D.; Savoj, H.; Sprague, P.; Teig, S. L.
Chemical Function Queries for 3D Database Search. J . Chem.
Inf. Comput. Sci. 1994, 34, 1297-1308.
(42) De Benedetti, P. G.; Cocchi, M.; Menziani, M. C.; Fanelli, F.
Theoretical Quantitative Size and Shape Activity Selectivity
Analyses of 5-HT1A Serotonin and R-Adrenergic Receptor Ligands.
J . Mol. Struct. 1994, 305, 101-110.
(43) Cocchi, M.; Fanelli, F.; Menziani, M. C.; De Benedetti, P. G.
Theoretical Quantitative Structure-Activity Relationship Analy-
sis of Congeneric and Non-Congeneric 1-Adrenoceptor Antago-
nists: a Chemometric Study. J . Mol. Struct. 1995, 331, 79-93.
(38) Brooks, B. R.; Bruccoleri, R. E.; Olafson, B. D.; States, D. J .;
Swaminathan, S.; Karplus, M. CHARMm: A Program for
Macromolecular Energy, Minimization, and Dynamics Calcula-
tions. J . Comput. Chem. 1983, 4, 187, 217.
(39) Catalyst force field and potential energy functions are described
in the file REFenergy.doc.html provided by MSI along with the
program Catalyst 4.5.
(40) Further information is available at the Catalyst mail-
ing list (http://www.msi.com/user/groups/catalyst/cat-archive-dir/
index.html) under the subject “Catalyst 4.5 conformations”.
J M010821U