Synthesis, molecular modeling studies, and pharmacological activity of selective A1 receptor antagonists

Article abstract of DOI:10.1021/jm0209580

We present a combined computational study aimed at identifying the three-dimensional structural properties required for different classes of compounds to show antagonistic activity toward the A₁ adenosine receptor (AR). Particularly, an approac

Full text of DOI:10.1021/jm0209580

J . Med. Chem. 2002, 45, 4875-4887
4875
Syn th esis, Molecu la r Mod elin g Stu d ies, a n d P h a r m a cologica l Activity of
Selective A1 Recep tor An ta gon ists
Francesco Bondavalli,† Maurizio Botta,*,‡ Olga Bruno,† Andrea Ciacci,‡ Federico Corelli,‡ Paola Fossa,*,†
Antonio Lucacchini,§ Fabrizio Manetti,‡ Claudia Martini,§ Giulia Menozzi,† Luisa Mosti,† Angelo Ranise,†
Silvia Schenone,† Andrea Tafi,‡ and Maria Letizia Trincavelli§
Dipartimento di Scienze Farmaceutiche, Universita` degli Studi di Genova, Viale Benedetto XV n.3, I-16132 Genova, Italy,
Dipartimento Farmaco Chimico Tecnologico, Universita` degli Studi di Siena, Via Aldo Moro, I-53100 Siena, Italy, and
Dipartimento di Psichiatria, Neurobiologia, Farmacologia e Biotecnologie, Universita` degli Studi di Pisa,
Via Bonanno, I-56126, Italy
Received J une 10, 2002
We present a combined computational study aimed at identifying the three-dimensional
structural properties required for different classes of compounds to show antagonistic activity
toward the A1 adenosine receptor (AR). Particularly, an approach combining pharmacophore
mapping, molecular alignment, and pseudoreceptor generation was applied to derive a
hypothesis of the interaction pathway between a set of A1 AR antagonists taken from the
literature and a model of the putative A1 receptor. The pharmacophore model consists of seven
features and represents an improvement of the N6-C8 model, generally reported as the most
probable pharmacophore model for A1 AR agonists and antagonists. It was used to build up a
pseudoreceptor model able to rationalize the relationships between structural properties and
biological data of, and external to, the training set. In fact, to further assess its statistical
significance and predictive power, the pseudoreceptor was employed to predict the free energy
of binding associated with compounds constituting a test set. While part of these molecules
was also taken from the literature, the remaining compounds were designed and synthesized
by our research group. All of the new compounds were tested for their affinity toward A1, A2a,
and A3 AR, showing interesting antagonistic activity and A1 selectivity.
In tr od u ction
lease) may therefore be clinically useful as neuropro-
tective agents. On the contrary, adenosine antagonists
(such as the alkylxanthines) stimulate the activity of
the CNS and have proven to be effective as cognition
enhancers. This is the joint action of antagonism of the
sedative effects caused by adenosine and of increasing
cerebral blood flow, thus increasing glucose and oxygen
availability to the brain.
In the last two decades, many efforts have been
invested in the synthesis of selective AR ligands for their
potential therapeutic use. This research has resulted
in the synthesis of a number of AR agonists and
antagonists.5,6 Particularly, selective AR subtype an-
tagonists are sought as antiinflammatory, antiasth-
matic, and antiischemic agents.7,8
In addition, A1 selective antagonists may have thera-
peutic potential in the treatment of various forms of
dementia, for example, in Alzheimer’s6 and Parkinson’s9
disease. Some compounds have been developed as
kidney protective diuretics and for the treatment of
asthma and depression.10 Moreover, on the basis of the
fact that adenosine plays a role in mediating the
haemodynamic changes associated with acute renal
failure, compounds that antagonize the renal effects of
adenosine are potential renal protective agents.11,12 As
an example, the antagonist 1,3-dipropyl-8-(3-norada-
mantyl)xanthine 13 is currently undergoing clinical
trials as a renal protective agent.13
Adenosine is a ubiquitous neuromodulator in both the
periphery and the central nervous system (CNS). The
effects elicited by adenosine are mediated by its interac-
tions with four receptor subtypes termed A1, A2a, A2b,
and A3, which can be distinguished pharmacologically,1
based on the rank order of potency of agonists and
antagonists.2 These receptors belong to the superfamily
of G-protein-coupled receptors and contain seven trans-
membrane domains (R-helices), interconnecting loops,
an extracellular terminal amino residue, and a cyto-
plasmic terminal carboxylate residue.3 Adenosine recep-
tors (ARs) from different species show a very high amino
acid sequence homology (82-93%), with the only excep-
tion of the A3 subtype, which exhibits a 74% primary
sequence homology between rat and human or sheep.4
The physiological significance and function of endog-
enous adenosine have been extensively researched.
Adenosine has been described as a neuromodulator in
the CNS, possessing global importance in the modula-
tion of the molecular mechanisms underlying many
aspects of brain function by mediating central inhibitory
effects. The development of agonists for the adenosine
A1 receptor able to mimic the central inhibitory effects
of adenosine (and so inhibiting neurotransmitter re-
* To whom correspondence should be addressed. M.B.: Tel: +39
0577 234306. Fax: +39 0577 234333. E-mail: botta@unisi.it. P.F.:
Tel: +39 010 3538361. Fax: +30 010 3538358. E-mail: fossap@unige.it.
† Universita` degli Studi di Genova.
The first AR antagonists reported were the natural
xanthines, caffeine and theophylline, but potent and
selective antagonists have stemmed from multiple
‡ Universita` degli Studi di Siena.
§ Universita` degli Studi di Pisa.
10.1021/jm0209580 CCC: $22.00 © 2002 American Chemical Society
Published on Web 09/24/2002

4876 J ournal of Medicinal Chemistry, 2002, Vol. 45, No. 22
Bondavalli et al.
pyrazolo-pyridines 10-12 have been reported in Scheme
1 and Table 1.
Finally, a two step computational protocol has been
applied to build a pharmacophore model for A1 AR
antagonists and a pseudoreceptor model of the A1 AR.
The latter model, able to rationalize the relationships
between the chemical features of A1 AR antagonists and
their binding affinity data, shows a good statistical
significance (correlation coefficient, r ) 0.9; rms devia-
tion, rmsd ) 0.6 kcal/mol) and successfully estimates
the affinities of the molecules of, and external to, the
training set.
F igu r e 1.
substitution of the parent heterocycle. On the other
hand, many structurally different nonxanthine deriva-
tives have been synthesized and studied as A1 AR and
A2 AR antagonists. Comparison of compounds belonging
to different classes of antagonists highlighted that
despite being structurally diverse, most of the known
ligands show some common features. In general, the
structures are planar, aromatic, or π-electron rich and
nitrogen-containing heterocycles. The heterocycles are
most often 6:5 fused bicycles or 6:6:5 fused tricycles,
substituted with hydrophobic moieties. Additionally,
antagonists lack the ribose moiety, which is essential
for agonist activity.
Ch em istr y
Scheme 1 reports the synthesis of the new compounds
10-12. 2-Hydrazino-1-phenylethanol 3a , prepared ac-
cording to a literature procedure,22 reacted with ethyl-
ethoxymethylene-cyanoacetate 4 in anhydrous toluene
to give 5-amino-1-(2-hydroxy-2-phenylethyl)-1H-pyra-
zole-4-carboxylic acid ethyl ester 5a in a very good yield
(80%). Basic hydrolysis (EtOH/NaOH) of 5a led to the
carboxy intermediate 6a , which by thermal decarboxy-
lation at 185 °C19 quantitatively afforded 2-(5-aminopy-
razol-1-yl)-1-phenylethanol 7a .
Condensation of 7a with diethyl ethoxymethylenemalo-
nate gave the intermediate 8a , which upon treatment
with POCl3 at reflux (36 h) underwent the cyclization
to the pyrazolo-pyridine nucleus with a concurrent
chlorination of the hydroxy side chain. Chromatographic
purification with Florisil and CHCl3 as the eluant gave
9a in a 60% overall yield.
Important classes of A1 selective antagonists are the
prototypic xanthine derivatives (preferably with bulky
cycloalkyl substituent at the C8-position), adenine
derivatives including aza and deaza analogues of ad-
enine, and other various heterocyclic compounds such
as pyrrolo-pyrimidines and pyrimido-indoles.14,15 More-
over, some literature reports revealed that the pyrazolo-
[3,4-b]pyridine scaffold provides compounds that effec-
tively bind A1 AR.
The same reaction sequence was applied to 3b to
afford 9b with a 40% yield in the last step, probably
because of a partial hydrolysis involving the phenoxy
substituent. Regioselective substitution of the C4 chlo-
rine of compounds 9 with an excess of various amines
(method A, Experimental Section) afforded the desired
products 10a -n and 11a -h in good yield (Table 1).
Compounds 12a -i have been obtained in a 70-90%
yield (Table 1) by treating 10a -c,e-g,j-l with an
excess of 1,8-diazabicyclo[5.4.0]undec-7-ene (DBU) at 80
°C (method B, Experimental Section).
As an example, tracazolate 1a , etazolate 1b, and
cartazolate 1c (Figure 1) are among the first nonxan-
thine antagonists reported in the literature,16 compound
1c being the most potent and quite selective antagonist
even more potent than theophylline at both A1 and A2
AR. They were also found to inhibit binding at A1
adenosine brain receptors.17,18 An additional example
is represented by a series of (substituted)-4-aminopy-
razolo[3,4-b]pyridines 219 (Figure 1), showing interesting
affinity for A1 and A2 receptors with the most active
compound possessing affinity of 0.3 and 0.5 µM for A1
and A2a AR, respectively, without selective antagonist
activity. Finally, many other products with similar
activity have been recently patented.20
It is interesting to point out that the chlorine atom
at the side chain of all of the new compounds has never
been substituted by the amino group. This hypothesis
1
was proved by the H NMR chemical shifts of the CH2-
On the basis of this experimental evidence, within our
research project on A1 AR antagonists, we have recently
designed new pyrazolo[3,4-b]pyridine derivatives21 with
the aim of obtaining compounds possibly characterized
by high affinity and selectivity toward the A1 AR. In
detail, a long lipophilic chain with an aromatic moiety
(chloroalkylphenyl, chloroalkylphenoxy, and styryl) has
been placed at the 1-position of the bicyclic nucleus,
instead of both the small methyl and the ethyl groups
previously reported (see compounds 1 and 2). Moreover,
various alkylamino, arylamino, and cycloalkylamino
moieties with different length and bulkiness, as well as
heterocyclic substituents with sizes variable from five
to seven members in the ring, have been added to the
4-position of the scaffold with the purpose of exploring
steric and electronic properties that a group in this
position should have to improve affinity toward A1 AR.
Synthetic pathways and biological data of the new
CH side chain protons, which gave an ABX complex
pattern owing to their nonequivalence and by the
subsequent dehydrochlorination with DBU to give the
corresponding styryl derivatives.
Biology
Compounds were tested for their ability to displace
[3H]-N6-cyclohexyladenosine (CHA) on A1 AR in bovine
cortical membranes, [3H]-2-{[4-(2-carboxyethyl)phen-
ethyl]amino}-5′-(N-ethylcarbamoyl)adenosine (CGS21680)
on A2a AR in bovine striatal membranes, and [125I]-N6-
(3-iodo-4-aminobenzyl)-5′-N-methylcarboxamidoadeno-
sine (AB-MECA) to A3 AR in bovine cortical membranes,
following a reported procedure.23 The A1, A2a, and A3
receptor binding affinities, expressed as Ki or percent
of binding for compounds 10-12, are reported in Table
2.

Selective Receptor Antagonists
J ournal of Medicinal Chemistry, 2002, Vol. 45, No. 22 4877
Sch em e 1a
a
Compounds: 3a , R1 ) Ph; 3b, R1 ) CH2OPh; 5a , R1 ) Ph; 5b, R1 ) CH2OPh; 6a , R1 ) Ph; 6b, R1 ) CH2OPh; 7a , R1 ) Ph; 7b, R1
) CH2OPh; 8a , R1 ) Ph; 8b, R1 ) CH2OPh; 9a , R1 ) Ph; 9b, R1 ) CH2OPh; 10a , R1 ) Ph, R2 ) NHn-Pr; 10b, R1 ) Ph, R2 ) NHcyclopropyl;
10c, R1 ) Ph, R2 ) NHn-Bu; 10d , R1 ) Ph, R2 ) NHt-Bu; 10e, R1 ) Ph, R2 ) NHCH2CH2OEt; 10f, R1 ) Ph, R2 ) NHcyclohexyl; 10g,
R1 ) Ph, R2 ) 1-pyrrolidinyl; 10h , R1 ) Ph, R2 ) 4-morpholinyl; 10i, R1 ) Ph, R2 ) NHPh; 10j, R1 ) Ph, R2 ) NHCH2Ph; 10k , R1 ) Ph,
R2 ) NHCH2CH2Ph; 10l, R1 ) Ph, R2 ) 1-piperidinyl; 10m , R1 ) Ph, R2 ) 1-hexahydroazepinyl; 10n , R1 ) Ph, R2 ) 1-(4-methyl)piperazinyl;
11a , R1 ) CH2OPh, R2 ) NHcyclopropyl; 11b, R1 ) CH2OPh, R2 ) NHn-Bu; 11c, R1 ) CH2OPh, R2 ) NHCH2CH2OEt; 11d , R1 ) CH2OPh,
R2 ) NHcyclohexyl; 11e, R1 ) CH2OPh, R2 ) 1-pyrrolidinyl; 11f, R1 ) CH2OPh, R2 ) 4-morpholinyl; 11g, R1 ) CH2OPh, R2 ) NHCH2Ph;
11h , R1 ) CH2OPh, R2 ) NHCH2CH2Ph; 12a , R2 ) NHn-Pr; 12b, R2 ) NHcyclopropyl; 12c, R2 ) NHn-Bu; 12d , R2 ) NHCH2CH2OEt;
12e, R2 ) NHcyclohexyl; 12f, R2 ) 1-pyrrolidinyl; 12g, R2 ) 4-morpholinyl; 12h , R2 ) NHCH2Ph, 12i, R2 ) NHCH2CH2Ph. Reagents: (a)
anhydrous toluene; (b) EtOH, NaOH; (c) 185 °C, HCl 6N; (d) 120 °C, Et2O; (e) POCl3, reflux; (f) anhydrous toluene, NHR2; (g) DBU,
absolute EtOH.
Moreover, to determine the intrinsic activity of 10g,k ,
found to be the most active compounds toward A1 ARs,
competition studies were performed in the presence and
in the absence of 1 mM guanosine 5′-triphosphate (GTP)
using the radiolabeled antagonist [3H]DPCPX. The GTP
shift is an in vitro parameter often indicative of intrinsic
activity. GTP shift represented the ratio between the
compound affinity constant in the presence and in the
absence of GTP. GTP modulates the affinity of agonist
compound whereas it does not affect the affinity for an
antagonist compound. A GTP shift value >1 is indicative
of an agonist profile; a GTP shift near to 1 is indicative
of an antagonist profile. In Table 3, the GTP shift values
of the selected compounds and R-PIA, included as
standard, were reported. At the A1 ARs, the selected
compounds displayed no significant GTP shift, suggest-
ing that they elicited an antagonist profile. In contrast,
the standard agonist R-PIA exhibited a larger GTP shift
value of 4.7. Intrinsic activity of compounds 10g,k was
also assessed by adenylyl cyclase functional assay
evaluating their ability to reverse the inhibition of
forskolin-stimulated adenylyl cyclase activity induced
by the agonist CHA (100 nM). In rat cerebral cortex
membranes, the A1 adenosine agonist CHA induced a
maximal inhibition of adenylyl cyclase activity of 15-
20% of total activity, under conditions of stimulation
(typically 3-4-fold) in the presence of 0.1 mM forskolin,
with an IC50 value of 1.4 ( 7 nM.24 The inhibiton effect
of CHA (100 nM) on adenylyl cyclase activity was
antagonized completely and in a concentration-depend-
ent manner by derivatives 10g,k with an IC50 value of
153.7 ( 9.8 nM and 52.2 ( 3.3 nM, respectively (Figure
2). The affinity constant values of compounds 10g,k
were also determined on rat cerebral cortex with results
similar to those obtained from bovine tissues. In fact,
while 10g possessed an affinity of 140 nM toward rat
A1 AR vs a value of 98 nM toward bovine A1 AR, 10k
showed an affinity of 73 nM toward rat A1 AR vs a value
of 50 nM toward bovine A1 AR.
Str u ctu r e-Activity Relation sh ip Con sider ation s
on th e New Com p ou n d s. Table 1 reports the A1, A2a,
and A3 AR binding affinities, expressed as Ki or,
alternatively, percent values, of the new pyrazolo-
pyridine compounds 10-12. From the binding data, it
can be seen that some of these compounds demonstrated
moderate to high affinity for A1 AR. Moreover, all of
these derivatives exhibited no affinity toward both the
A2a and the A3 AR, with consequent high selectivity
against A1 AR.
As a general rule, compounds 10, bearing a chlo-
rophenylethyl side chain at the 1-position, represented
the most active compounds within the newly synthe-
sized molecules. In fact, 10k showed the lowest affinity
value against A1 AR (50 nM), while compounds 10a ,b,e,-
g,j were characterized by affinity data ranging from 98
to 152 nM.
The length of the side chain at the C4 was of great
importance for A1 affinity. Reduction of the phenylethyl
moiety of 10k to a benzyl or phenyl group of 10j,i,
respectively, caused a relevant decrease in affinity for
A1 AR, probably due to the reduction of hydrophobic
contacts with the receptor (see below). On the contrary,
a shorter alkyl chain (n-propyl or cyclopropyl of com-
pounds 10a ,b, respectively) was associated with higher

4878 J ournal of Medicinal Chemistry, 2002, Vol. 45, No. 22
Bondavalli et al.
Ta ble 1. Physicochemical Data and Affinity at ARs of Compounds 10-12
Ki (nM)a or % inhibition
b
c
no.
R2
formula
mp (°C)
yield (%)
A1
A2a
A3d (%)
23
10a
10b
10c
10d
10e
10f
10g
10h
10i
NHC3H7
C20H23N4O2Cl
20H21N4O2Cl
82-83
102-103
81-82
149-150
108-109
104-105
164-165
87-88
90
93
70
60
80
75
75
70
88
75
85
75
70
80
86
73
50
83
65
63
70
75
82
75
83
98
80
90
90
94
95
100 ( 8.4
112 ( 9.6
4100 ( 23
4800 ( 32
151 ( 10
1490 ( 107
98.2 ( 7.3
470 ( 29
348 ( 21
139 ( 10
50 ( 3.7
41%
11%
10%
3%
19%
2%
0%
17%
0%
11%
0%
4%
23%
22%
NHcyclopropyl
NHC4H9
NHC(CH3)3
C
C21H25N4O2Cl
C21H25N4O2Cl
C21H25N4O3Cl
23H27N4O2Cl
21H23N4O2Cl
21H23N4O3Cl
C23H21N4O2Cl
C24H23N4O2Cl
C25H25N4O2Cl
22H26N4O2Cl2
23H27N4O2Cl
22H26N5O2Cl
21H24N4O3Cl2
C22H27N4O3Cl
C22H28N4O4Cl2
24H30N4O3Cl2
22H26N4O3Cl2
22H26N4O4Cl2
C25H25N4O3Cl
C26H27N4O3Cl
C20H22N4O2
NH(CH2)2OC2H5
NHcyclohexyl
1-pyrrolidinyl
4-morpholinyl
NHC6H5
NHCH2C6H5
NHCH2CH2C6H5
1-piperidinyl
1-hexahydroazepinyl
1-(4-methylpiperazinyl)
NHcyclopropyl
NHC4H9
NH(CH2)2OC2H5
NHcyclohexyl
1-pyrrolidinyl
4-morpholinyl
NHCH2C6H5
NHCH2CH2C6H5
NHC3H7
C
C
C
0
137-138
143-144
121-122
135-136e
170-171
142-143
184-185e
75-76
10j
10k
10l
34
C
C
C
C
10m
10n
11a
11b
11c
11d
11e
11f
11g
11h
12a
12b
12c
12d
12e
12f
12g
12h
12i
35%
22%
2140 ( 112
34%
23%
15%
1%
1047 ( 97
1219 ( 104
2690 ( 192
3710 ( 240
33%
145-146e
75-76e
C
C
C
179-180e
180-181e
129-130
105-106
147-148
137-138
130-131
139-140
143-144
151-152
123-124
172-173
156-157
15%
0%
12%
5%
0%
0%
0%
0%
0%
0%
40%
36%
456 ( 37
40%
67%
60%
51%
21%
39%
40%
39%
NHcyclopropyl
NHC4H9
C
20H20N4O2
C21H24N4O2
C21H24N4O3
23H26N4O2
21H22N4O2
21H22N4O3
C24H22N4O2
C25H24N4O2
NH(CH2)2OC2H5
NHcyclohexyl
1-pyrrolidinyl
4-morpholinyl
NHCH2C6H5
NHCH2CH2C6H5
C
C
C
0%
2%
2%
58%
a
b
The Ki values are means ( SEM of three separate assays, each performed in triplicate. Displacement of specific [3H]CHA binding
in bovine cortical membranes or percentage of inhibition of specific binding at 10 µM concentration. c Displacement of specific [3H]CGS21680
d
binding in bovine striatal membranes or percentage of inhibition of specific binding at 10 µM concentration. Displacement of specific
[
125I]AB-MECA binding in bovine cortical membranes or percentage of inhibition of specific binding at 10 µM concentration. Only compounds
10a ,g,k were tested. e As hydrochloride.
Ta ble 2. Experimental and Calculated Binding Affinity of
Compounds 13-29 (Taken from the Literature), and 10a ,e,i,k
and 11h Belonging to the New Class of A1AR Antagonists
Introduction of a long alkoxyalkyl chain at C4 par-
tially restored the affinity for A1 AR, being 151 nM the
value measured for affinity of compound 10e. Finally,
variation on the nature of the amine (from secondary
to tertiary) at the C4 position significantly influenced
the affinity. The general trend showed a marked drop
in affinity with tertiary amines, with the only exception
of compound 10g that showed appreciable affinity for
A1 AR.
When the 2-chloro-2-phenylethyl side chain at the N1
position was changed to the 2-chloro-3-phenoxypropyl
moiety or to the styryl group, a dramatic decrease in
affinity was observed. In fact, both compounds 11 and
12 were all characterized by very low affinity data.
These findings led to the suggestion that the exten-
sion of the alkyl chain at the C4, combined with the
substituent at the N1 position, had some relevant effects
on the binding of compounds 10-12 to the A1 AR. When
a 2-chloro-2-phenylethyl side chain was linked to the
N1 position of the pyrazole ring, a gradual increase in
affinity was observed by lengthening the side chain at
C4 from a phenyl to a phenylethyl moiety. On the
contrary, the highest affinity for the 4-alkylamino
derivatives was found when the chain was characterized
by three carbon atoms (n-propyl or cyclopropyl).
Analogous considerations cannot be made for com-
pounds bearing a 2-chloro-3-phenoxypropyl moiety at
N1. In fact, as a general rule, compounds 11 were less
active than the corresponding molecules belonging to
10. As an example, while affinity of 11h was about 3-fold
∆Gexp
∆Gpred
∆∆G
compd
ref
Ki (nM)
(kcal/mol)a
(kcal/mol) (kcal/mol)
13
13
33
15
34
35
36
37
38
39
40
41
66
66
19
15
68
41
21
21
21
21
0.19
0.46
2.6
7.3
7.9
-13.030
-12.510
-11.500
-10.900
-10.860
-10.720
-10.570
-9.050
-8.400
-10.090
-12.480
-11.910
-10.480
-7.190
-9.374
-8.337
-10.961
-9.380
-9.140
-8.650
-9.780
-8.500
-12.884
-11.033
-10.694
-10.557
-10.527
-10.444
-9.668
-8.118
-9.034
-10.045
-12.913
-11.628
-10.845
-8.408
-11.356
-9.383
-9.379
-8.138
-7.595
-8.720
-10.045
-8.504
0.146
1.477
0.806
0.343
0.333
-0.276
0.902
0.932
-0.634
0.045
-0.433
0.282
-0.365
-1.217
-1.982
-1.046
1.582
14b
15b
16
17b
18b
19
20
21
22
23
24
25
10
13
176
540
29.5
0.49
1.3
15
26
4300
101
600
6.6
100
151
348
50
27b
28b
29b
10a b
10eb
10ib
10k b
11h b
1.242
1.545
-0.070
-0.262
-0.007
456
a
Free energies of binding derived from the Ki values according
b
to the Gibbs-Helmholtz equation. Test set ligands used for the
prediction of ∆Gpred values.
affinity with respect to longer or bulky substituents such
as the butyl or tert-butyl moieties of compounds 10c,d ,
respectively.

Selective Receptor Antagonists
J ournal of Medicinal Chemistry, 2002, Vol. 45, No. 22 4879
Ta ble 3. Intrinsic Activity of 10g,k toward A1 ARs Expressed
as GTP Shift
Ki (nM)a
compd
-GTP
+GTP
GTP shift
R-PIA
10g
10k
4.2 ( 0.3
103 ( 10.4
69.7 ( 4.9
19.9 ( 1.4
90 ( 6.3
56 ( 3.6
4.7
0.96
0.8
a
Displacement of [3H]DPCPX from bovine cortical membranes
in the absence (-GTP) and in the presence (+GTP) of 1 mM GTP.
Values are taken from three separate experiments and expressed
as means ( SEM.
lower than the corresponding 10j, affinity of 11a ,c was
about 1 order of magnitude lower than 10b,e, respec-
tively.
F igu r e 2. Concentration-dependent reversal of CHA adenylyl
cyclase activity inhibition by 10g (9) and 10k (2) derivatives.
The enzyme activity was assayed at the concentrations of the
antagonists indicated in the presence of 100 nM CHA and 0.1
mM forskolin as described in biologic methods. Each data point
is expressed as a percentage of adenylyl cyclase activity and
represents the mean ( SEM of at least three independent
experiments.
This last evidence supported the hypothesis that the
side chain at the N1 position was also a crucial key in
determining the affinity values of these compounds
toward A1 AR, in agreement with results obtained from
the pseudoreceptor modeling (see below). In fact, the
surrogate of the A1 AR generated by means of PrGen25
software showed that a phenylethyl side chain at N1
possessed the optimal structural requirements (in terms
of extension and bulkiness) to have profitable hydro-
phobic interactions with Ile6(89), Ile14(252), and His23
of the putative receptor.26
(AS), and two donor sites (DS) (tolerance 2.0 Å) was
derived. Figure 4 shows the seven feature pharmaco-
phore model with compounds 13-23 superimposed on
it. As an example, compound 13, chosen by the program
as the reference molecule, is characterized by a complete
mapping onto the pharmacophore model. In fact, while
HY2 is matched by both the five-membered heterocyclic
ring and the alkyl substituent at the 3-position, the HY3
feature is mapped by both the six-membered heterocy-
clic ring and the alkyl chain at the 1-position. On the
other hand, HY1 is fulfilled by the adamantyl moiety
at the 8-position. Moreover, the nitrogen atom at the
9-position corresponds to the AA feature of the model,
with DS1 representing its counterpart on the putative
receptor. Finally, the NH group at the 7-position and
the carbonyl group at the 6-position are the molecular
counterparts of the corresponding AS and DS2 features
of the putative receptor.
The validity of the new pharmacophore was tested by
means of molecular field analysis. Accordingly, for each
conformer of compounds 13-23 selected by DISCO in
deriving the seven point model, the corresponding
electrostatic map was generated from the atomic partial
charges calculated with MOPAC45 (AM1) and compared
each other with the Isopotential Contour option in
Sybyl.46 As a result, a good superimposition in isopo-
tential contours for the selected compounds was found.
Moreover, the HINT program47 was applied to localize
and display common hydrophobic areas for compounds
13-23. The superimposition of HINT maps of the most
selective compounds highlighted three hydrophobic por-
tions of the molecules able to fit three pockets (labeled
as P1, P2, and P3 in Figure 5) on the putative receptor.
In detail, P1 could represent the hydrophobic region of
the receptor contacted by HY1 (i.e., the alkyl substituent
at the 8-position of 13). Similarly, P2 defines a hydro-
phobic cavity where alkyl or aryl substituents lie and
where the N3 substituent of xanthinic antagonists, such
as 13, is located. Finally, P3 is the region of the receptor
able to accept various substituents, in particular the N1
substituent of compound 13.
Resu lts a n d Discu ssion
A computational study that combines a ligand-
based drug design (pharmacophore development) method
and a pseudoreceptor generation approach aimed at
rationalizing the relationships between structures and
affinity data of A1 AR antagonists is presented. In
particular, the combined computational approach can
be summarized as follows: (i) generation of a pharma-
cophore model to be intended as identification and
superposition of the structural features shared by the
molecules and potentially important for biological activ-
ity; (ii) building of a pseudoreceptor model (based on the
above alignment and on site-directed mutagenesis ex-
periments) and its validation by prediction of the
activity of the test set molecules, aligned to the phar-
macophore model.
Molecu lar Align m en t an d P h ar m acoph or e Model
Gen er a tion . Several pharmacophore models for the A1
AR ligands have already been presented in the litera-
ture,27-35 all based on the assumption that A1 agonists
and antagonists share a common binding site on the
biomolecule.32 Among them, the most accredited is the
N6-C8 model,30 which is derived from the greatest
overall steric and hydrophobic overlap of xanthine
antagonists with respect to adenosine, the natural
substrate of the A1 receptor.
Eleven structurally diverse A1 selective adenosine
antagonists 13-2313,15,35-43 were taken from the litera-
ture (Figure 3), and the DISCO (DIStance COmpari-
son)44 strategy was employed to derive a meaningful
pharmacophoric model for these compounds.
In the first DISCO runs, among all of the solutions
proposed by the program, both the standard, the flipped,
and the N6-C8 models were found. By increasing the
minimum input number of common pharmacophoric
points, among all of the different hypotheses found, a
seven point model with three hydrophobic centers (HY),
one hydrogen bond acceptor atom (AA), one acceptor site
All of these findings showed a good agreement be-
tween the properties of the pharmacophore model
generated by DISCO and the results obtained by means

4880 J ournal of Medicinal Chemistry, 2002, Vol. 45, No. 22
Bondavalli et al.
F igu r e 3. Compounds 13-29 collected from the literature and used for pharmacophore generation and pseudoreceptor modeling.
F igu r e 4. DISCO superposition of compounds 13-23. Phar-
macophore features are labeled with AA (hydrogen bond
acceptor atom), HY (hydrophobic center), DS (donor site), and
F igu r e 5. HINT maps for compounds 13-23. Green volumes
represent the hydrophobic regions occupied by the ligand
corresponding to the P1, P2, and P3 pockets on the putative
receptor.
AS (acceptor site).
of HINT calculations. In fact, while P2 and P3 pockets
partially fit the HY1 and HY3 features of the pharma-
cophore model, HY1 and P1 are perfectly superposed.
In summary, DISCO and HINT calculations improved
the previous N6-C8 model with a novel seven point
pharmacophoric map whose properties can be sum-
marized as follows: (i) a hydrogen bond acceptor atom
(AA) corresponding to a nitrogen or oxygen atom, able
to interact with a donor site (DS1), which represents
the counterpart on the receptor; (ii) two hydrophobic
centers (HY2 and HY3) in the bicyclic planar nucleus
filling the corresponding receptor pockets P2 and P3 and
a third hydrophobic center (HY1) on the side chain
matching the P1 receptor pocket; (iii) one acceptor site
(AS) and two donor sites (DS1 and DS2), which define
the putative active site on the A1 receptor.

Selective Receptor Antagonists
J ournal of Medicinal Chemistry, 2002, Vol. 45, No. 22 4881
P seu d or ecep tor Gen er a tion . The pharmacophore
model (corresponding to the alignment of the ligands
deduced by DISCO calculations) was used to perform
the second computational step of our work. Particularly,
by application of the pseudoreceptor modeling software
PrGen, an atomistic binding site model for A1 AR was
built taking into account the structure and biological
activity of known ligand molecules.
For the generation of the pseudoreceptor model, with
the aim of covering a range of about 4-5 orders of
magnitude in affinity, six additional antagonists (24-
29)15,19,43,48,49 taken from the literature were considered.
They all express antagonist activity toward A1 AR and
were collected under the assumption that all of these
substances were acting through the same binding site.
Biological data of the whole set of compounds 13-
29, expressed as Ki, were in the range between 0.19
(compound 13) and 4300 nM (compound 26). According
to the Gibbs-Helmholtz equation, Ki values were
converted into free energies of binding, hereafter re-
ported as ∆Gexp. Ten molecules of the whole set (namely,
compounds 13, 16, and 19-26 reported in Table 2) have
been automatically chosen by PrGen to build the train-
ing set.
F igu r e 6. Complex between the A1 adenosine pseudoreceptor
(green) and the compound 13 (black), the most active inhibitor
considered in this study. For the sake of clarity, only few amino
acids of the model have been displayed.
Sch em e 2. Schematic representation of residues
constituting the pseudoreceptor model and their major
interactions with compound 13a
Next, to choose appropriate residues for pseudorecep-
tor construction, information derived from site-directed
mutagenesis experiments3c,6,50 and from the primary
amino acid sequence of the rat A1 AR,51 were used.
Moreover, the receptor-mediated ligand alignment tech-
nique52,53 was applied to generate the primordial model
(see the Experimental Section for further details). In
the next step, the remaining seven ligands of the
training set were docked into the receptor cavity thus
generating the final pseudoreceptor-inhibitors assem-
bly.
a
To achieve the optimum positions of the manually
placed residues, a receptor equilibration was subse-
quently performed allowing for translation, rotation,
and torsional variations of receptor residues, whereas
the ligands were kept fixed in their original arrange-
ments (correlation-coupling protocol). Finally, the phar-
macophore was allowed to relax within the binding
pocket. Repeating these two steps several times (the
computational protocol called ligand equilibration) yielded
a pseudoreceptor model with an r value of 0.93 and rmsd
between experimental and predicted free energies of
ligand binding of 0.641 kcal/mol, corresponding to an
uncertainty factor of 3.0 in the inhibitory constants.
Com p a r ison betw een th e P seu d or ecep tor a n d
th e P h a r m a cop h or e Mod el. Figure 6 and Scheme 2
show the complex between the final pseudoreceptor
model and the compound 13, taken as representative
of all of the A1 AR inhibitors considered in this study.
The pseudoreceptor is mainly characterized by a large
hydrophobic pocket defined by Ile14(252), Leu15(253),
Ile18(272), Ala19(273), and Ile20 (details on residue
numbering are given in ref 26), and two distinct
hydrogen-bonding sites involving Thr1(91) and His12-
(251), respectively.
Amino acid labels are from the primary sequence of rat A1AR
(see ref 26 for further detail). In parentheses, pharmacophore
features identified by DISCO and HINT calculations.
the terminal amino group of the long hydrophilic side
chain of 24 is engaged in a hydrogen-bonding contact
with the carbonyl moiety of Ala21. Ile14(252) and
Leu15(253) of the hydrophobic pocket are located in a
region of the space in front of the five-membered ring
of the xanthine nucleus, thus corresponding to the
receptor counterpart of the HY1 hydrophobic feature
identified by DISCO. Ile18(272) and Ala19(273), con-
stituting a portion of the hydrophobic pocket containing
the C8 substituent, are located at the opposite site of
the five-membered ring, with respect to Ile14(252) and
Leu15(253).
The nitrogen atom at the 9-position interacts by a
hydrogen bond with the NH group of the imidazole ring
of His12(251), the NH-N9 distance being 3.01 Å. As a
consequence, the xanthine N9 and the His12(251) NH
group could be identified as the AA and DS1 features
of the DISCO pharmacophore, respectively. In a similar
way, the xanthine NH group at the 7-position is involved
in a hydrogen-bonding interaction with the hydroxy
oxygen of Thr1(91) (NH-O distance of 1.76 Å) corre-
sponding to the AS pharmacophore feature.
The hydrophobic cavity is able to accommodate bulky
cycloalkyl substituents, such as the noradamantyl moi-
ety of 13, mainly interacting with the alkyl side chain
of Ile20. Moreover, a polar substitution at the 8-position
of the xanthine nucleus is also tolerated. As an example,
Finally, while the aromatic portion of Tyr17(271) is
located at the proper distance of about 4.6 Å to have
profitable interactions with the six-membered ring of
xanthines (based on a T-tilted orientation of the two

4882 J ournal of Medicinal Chemistry, 2002, Vol. 45, No. 22
Bondavalli et al.
cyclic moieties), Gln2(92) and Leu7(88) approach the
region accommodating the N1 and C6 positions of 13,
thus corresponding to the HY3 feature.
affinities of such compounds were predicted to be -8.138
and -7.595 kcal/mol vs actual values of -9.380 and
-9.140 kcal/mol, respectively. Also in this case, we may
interpret this decrease in affinity taking into account
that both the alkyl and the alkoxyalkyl chains lack some
of the profitable hydrophobic contacts with Leu15(253),
Ala19(273), and Ile20 found for compound 10k .
In summary, to depict the major interaction keys
between pseudoreceptors and inhibitors, we can ad-
ditionally report that the hydroxy oxygen of Thr1(91)
acts as a hydrogen-bonding acceptor for the hydrogen
atom bound to N7 in the xanthine derivatives (such as
13, 24, and 25), as well as for the hydrogen atom bound
to N6 of nonxanthine derivatives 16 and 21 and at the
1-position of 20. This result is in accordance with site-
directed mutagenesis data suggesting an interaction
between Thr1(91) and the N6 substituents of adenine
moiety in antagonist compounds.3c On the other hand,
the imidazole NH hydrogen of His12(251) (reported as
an important amino acid for interactions with A1
antagonists)50b is engaged in a hydrogen bond interac-
tion with the N9 nitrogen atom of 13 and 19. Moreover,
the terminal amino group on the long C8 side chain of
compounds 24 is involved in an additional hydrogen
bond with the carbonyl oxygen of Ala 21. Finally, the
hydrophobic cavity accommodates the bulky substituent
at the 8-position of xanthine antagonists and at the N6-
position of adenosine analogues, according to the N6-
C8 pharmacophore model.
Va lid a tion of th e P seu d or ecep tor Mod el. With
the purpose of testing the predictive power of the model,
12 A1 antagonists defining the test set (Table 2) were
docked into the receptor model following the DISCO-
derived alignment and subjected to a free ligand relax-
ation with the same settings used for the training set.
This procedure yielded a rmsd between experimental
and predicted free energies of ligand binding of 1.1 kcal/
mol, corresponding to an uncertainty factor of 6.6 in the
inhibitory constants.
Although the theoretically derived pseudoreceptor
model is unlikely to fully describe the real binding site
of A1 AR, it was able to explain the properties of the
new synthesized antagonists. In fact, the most active
ligand of the series, 10k , bearing a 2-chloro-2-phenyl-
ethyl side chain at the N1 nitrogen atom together with
a phenylethylamino substituent at the C4 carbon atom,
is predicted by the model to have a ∆G of -10.045 vs
an experimental value of -9.780 kcal/mol. Compound
10k interacts with the pseudoreceptor model in such a
way as its C4 side chain contacts a hydrophobic region
mainly defined by Leu15(253), Ala19(273), and Ile20.
On the other hand, while the side chain at the N1
position is located inside a second pocket surrounded
by Gln2(92), Ser13(281), and Glu22, the ethyl ester
chain at the 5-position is embedded between His10(278)
and His12(251). It is also important to point out that
the pyridine ring of the bicyclic nucleus is involved in a
π-π interaction with the aromatic side chain of Tyr17-
(271).
Moreover, the molecular portion corresponding to the
C5-C6 sequence of the heterocycle is accommodated
within a pseudoreceptor cavity mainly defined by His12-
(251), His13(278), Tyr17(271), and Ala19(273) (contain-
ing the HY2 pharmacophore feature described below in
the text), quite unexplored by our derivatives. In fact,
while the carbethoxy substituent of compounds 10-12
corresponded to the propyl chain at the 4-position of 13,
the condensed phenyl ring of active compounds such as
16 and 17 is located in front of the unsubstituted C6 of
the pyrazolo-pyridine nucleus. These findings led to the
suggestion that variations on the stereoelectronic prop-
erties of the carbethoxy substituent are required, as well
as insertion of lipophilic moieties into the C6 position
of the pyrazole-pyridine ring.
Finally, variation on the length of the N1 substituent
also affects the affinities of these derivatives. Particu-
larly, the lengthening (from a 2-chloro-2-phenylethyl to
a 2-chloro-3-phenoxypropyl moiety) of the side chain at
the N1 position of 11h lead to a decreased affinity
(-8.505 estimated vs -8.497 kcal/mol experimental
value) mainly due to unfavorable contacts between the
phenoxy moiety and the His23 residue.
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 seu d or ecep tor Mod els a n d P r evi-
ou sly P u blish ed Mod els for A1 AR. A recent paper23
described both a pharmacophore model for A1 AR and
the three-dimensional theoretical models of the com-
plexes between compounds taken from the literature
(i.e., compound 18) and a putative A1 AR, as determined
by homology modeling and molecular dynamics calcula-
tions. Having no structural information (i.e., distances
between the key structural elements of these complexes)
in our hands, we could make only a qualitative com-
parison between these structures and the pharmaco-
phore model proposed in this paper.
In particular, a perfect agreement has been found
between the pharmacophore model proposed by Da
Settimo and results derived from both our DISCO and
our HINT studies. In fact, the hydrogen bond acceptor-
donor features identified by DISCO and the three
hydrophobic pockets found with HINT were all de-
scribed in the cited paper. On the other hand, some
differences have been highlighted by comparing our
pseudoreceptor and the complexes previously reported.
In detail, the Asn254 side chain has been reported by
Da Settimo as responsible for a hydrogen bond accep-
tor-donor motif involving both N6 and N7 of compound
18. On the contrary, the corresponding Asn16(254) of
our pseudoreceptor is at the periphery of the binding
site, while the N6 atom of 18 interacts by a hydrogen
bond with the hydroxy group of Thr1(91), in agreement
with Rivkees and co-workers3c reporting the same
residue as interacting with the N6 substituents of A1
AR ligands. Moreover, to the best of our knowledge, no
experimental data have been published on the literature
supporting the hypothesis that Asn254 can represent
The shortening of the C4 phenylethylamino to a
benzylamino or phenylamino substituent (compound
10j,i, respectively) lead to reduced hydrophobic contacts
between the ligand and the pseudoreceptor (particu-
larly, Ile20). Accordingly, the estimated binding free
energy of 10i was -8.720 kcal/mol vs an actual value
of -8.650 kcal/mol.
On the other hand, when the C4 substituent is an
alkylamino (10a ) or alkoxyalkylamino (10e) group, the

Selective Receptor Antagonists
J ournal of Medicinal Chemistry, 2002, Vol. 45, No. 22 4883
measured in KBr with a Perkin-Elmer 398 spectrophotometer.
1H NMR spectra were recorded in (CD3)2SO solution on a
Varian Gemini 200 (200 MHz) instrument. Chemical shifts are
reported as δ (ppm) relative to tetramethylsilane (TMS) as
internal standard, J in Hz. 1H patterns are described using
the following abbreviations: s ) singlet, d ) doublet, t )
triplet, q ) quartet, sx ) sextect, m ) multiplet, br ) broad.
All compounds were tested for purity by thin-layer chroma-
tography (TLC) (Merk, Silica gel 60 F254, CHCl3 as eluant).
Analyses for C, H, N were within (0.3% of the theoretical
value.
5-Am in o-1-(2-h yd r oxy-2-p h en ylet h yl)-1H -p yr a zole-4-
ca r boxylic Acid Eth yl Ester (5a ). The starting hydrazine
3a (3.04 g, 20.0 mmol) was added to a solution of ethyl-
ethoxymethylene cyanoacetate 4 (3.38 g, 20.0 mmol) in anhy-
drous toluene (20 mL), and the mixture was heated at 80 °C
for 8 h. The solution was then concentrated by rotatory
evaporation to half of the volume and allowed to cool to room
temperature.
The separated yellow pale solid was filtered and recrystal-
lized from toluene to afford 5a (4.40 g, 80%) as a white solid;
mp 136-137 °C. 1H NMR (CDCl3): δ 1.33 (t, J ) 7.0, 3H, CH3),
3.53 (m, 1H, OH, disappears with D2O), 3.92-4.20 (m, 2H,
CH2N), 4.25 (q, J ) 7.0, 2H, CH2O), 5.02-5.13 (m, 1H, CHOH),
5.30 (br s, 2H, NH2, disappears with D2O), 7.23-7.42 (m, 5H
Ar), 7.58 (s, 1H, H-3). IR (CHCl3) cm-1: 3470, 3330 (NH2),
3300-3000 (OH), 1685 (CO). Anal. (C14H17N3O3) C, H, N.
the receptor residue interacting with the hydrogen-
bonding acceptor-donor motif of A1 antagonists. As a
consequence, the pharmacophore model proposed by our
research group, reproducing the well-known N6-Thr91
interaction, could be considered as an improved model
with respect to the Da Settimo hypothesis.
Moreover, the hydrophobic region corresponding to
the P3 pocket described in this paper has been sug-
gested by Da Settimo as a relatively small cavity, not
in full agreement with our results. In fact, the pseu-
doreceptor model is able to well-accommodate within the
P3 pocket the phenylethyl side chain of compound 10k ,
the most active molecule in our hands. Particularly,
while the phenyl ring is mainly interacting with His23,
the ethyl portion of the side chain maps a region defined
by His10(270), Ser13(281), and Ile14(252). These find-
ings led to the suggestion that the size of P3 is able to
accommodate extended, not too bulky, side chains.
Finally, while the Da Settimo model, based on phar-
macophore generation and molecular docking protocols,
allowed for the rationalization of A1 AR antagonist SAR
only at a qualitative level, our pseudoreceptor provided
quantitative relationships between the structure of A1
AR antagonists and their biological data.
5-Am in o-1-(2-h yd r oxy-3-p h en oxy-p r op yl)-1H-p yr a zole-
4-ca r boxylic Acid Eth yl Ester (5b). The compound was
prepared according to the synthetic sequence described for
compound 5a starting from 3b to give 5b as a white solid (4.88
g, 80%); mp 94-95 °C. 1H NMR (CDCl3): δ 1.33 (t, J ) 7.1,
3H, CH3), 2.5-3.5 (very br s, 1H, OH, disappears with D2O),
3.82-4.06 and 4.15-4.33 (2m, 4H, CH2N + CH2OAr), 4.26 (q,
J ) 7.1, 2H, CH2OCO), 4.36-4.49 (m, 1H, CHO), 5.1-5.7 (very
br s, 2H, NH2, disappears with D2O), 6.85-7.05 and 7.22-
7.38 (2m, 5H Ar), 7.63 (s, 1H, H-3). IR (CHCl3) cm-1: 3600-
3450 and 3350 (OH + NH2), 1680 (CO). Anal. (C15H19N3O4) C,
H, N.
Con clu sion s
A pharmacophore model generation protocol has been
successfully applied to build a three-dimensional model
of the chemical features responsible for A1 AR antago-
nist activity. The seven pharmacophore features cor-
responded to four structural portions on the ligands and
three points of the receptor, mapping the most impor-
tant interaction between the A1 receptor and its an-
tagonists.
The pharmacophore model, combined with the find-
ings derived from experimental data (site-directed mu-
tagenesis and primary amino acid sequence of rat A1
AR), was used to build a pseudoreceptor, to be intended
as the putative binding site model for the structurally
uncharacterized A1 AR. Such a three-dimensional re-
ceptor surrogate has been subsequently validated using
an external set of compounds (test set), leading to high
correlation and predictive power as well as a good
agreement with the pharmacophore model.
The newly synthesized compounds showed an inter-
esting antagonistic profile and selectivity toward A1
ARs, with respect to molecules belonging to the same
class of pyrazolo-pyridine derivatives reported in the
literature. The pseudoreceptor appeared to be an im-
proved model with respect to both the N6-C8 hypothesis
and a three-dimensional model of A1 AR recently
described in the literature. It also furnished some
suggestions on variations that should be made on the
structure of the pyrazolo-pyridine compounds to better
fit the pseudoreceptor model. In particular, C5 and C6
were identified as the positions to be further investi-
gated. Accordingly, enlargement of the pyrazolo-pyridine
class is ongoing by synthesis of some new derivatives
that will be published in due time with their biological
data.
5-Am in o-1-(2-h yd r oxy-2-p h en ylet h yl)-1H -p yr a zole-4-
ca r boxylic Acid (6a ). To a solution of 5a (2.7 g, 10 mmol) in
ethanol 96% (15 mL), a solution 3.5 M of NaOH (10 mL) was
added. The reaction mixture was refluxed for 4 h, and then,
the ethanol was evaporated under reduced pressure. The
mixture was acidified with HCl 6 N; the precipitated white
solid was collected by filtration and washed with water. The
crude product was then recrystallized from absolute ethanol
to give 6a as a white solid (2.34 g, 95%); mp 180-182 °C (dec).
1H NMR ((CD3)2SO): δ 3.92-4.25 (m, 2H, CH2N), 4.91-5.03
(m, 1H, CHOH), 5.68-5.76 (m, 1H, OH, disappears with D2O),
6.03-6.13 (br s, 2H, NH2, disappears with D2O), 7.23-7.44
(m, 5H Ar), 7.45 (s, 1H, H-3), 11.50-12.00 (br s, 1H, COOH,
disappears with D2O). IR (KBr) cm-1: 3385, 3280 (NH2), 3250-
2800 (COOH + OH), 1650 (CO). Anal. (C13H13N3O3) C, H, N.
5-Am in o-1-(2-h yd r oxy-3-p h en oxy-p r op yl)-1H-p yr a zole-
4-ca r boxylic Acid (6b). The compound was prepared accord-
ing to the synthetic sequence described for compound 6a
starting from 5b, to give 6b (2.49 g, 90%) as a white solid; mp
1
152-153 °C (dec). H NMR ((CD3)2SO): δ 3.82-4.35 (m, 5H,
2CH2 + CH), 5.41-5.62 (m, 1H, OH, disappears with D2O),
5.95-6.25 (br s, 2H, NH2, disappears with D2O), 6.80-7.10
and 7.15-7.32 (2m, 5H Ar), 7.50 (s, 1H, H-3), 11.50-12.05 (br
s, 1H, COOH, disappears with D2O). IR (KBr) cm-1: 3435,
3310 (NH2), 3200-2500 (OH), 1670 (CO). Anal. (C13H15N3O4)
C, H, N.
2-(5-Am in o-p yr a zol-1-yl)-1-p h en yl Eth a n ol (7a ). Com-
pound 6a (2.47 g, 10 mmol) was heated to 185 °C. When the
development of CO2 had finished, the residue was cooled to
room temperature, dissolved in HCl 6 N, and neutralized with
solid NaHCO3. A light brown solid precipitated, which was
collected by filtration. The crude product was then recrystal-
lized from CHCl3 to give 7a (1.99 g, 98%) as a light yellow
solid; mp 130-132 °C. 1H NMR (CDCl3): δ 1.70-3.20 (br s,
3H, NH2 + OH, disappears with D2O), 4.03-4.31 (m, 2H, CH2),
Exp er im en ta l Section
Ch em istr y. Starting materials were purchased from Ald-
rich-Italia (Milan). Melting points were determined with a
Bu¨chi 530 apparatus and are uncorrected. IR spectra were

4884 J ournal of Medicinal Chemistry, 2002, Vol. 45, No. 22
Bondavalli et al.
1
5.09-5.19 (dd, 1H, CHOH), 5.54 (d, 1H, H-4), 7.20-7.50 (m,
6H, 5H Ar + H-3). IR (CHCl3) cm-1: 3500-3000 (OH + NH2).
Anal. (C11H13N3O) C, H, N.
10a in 90% yield; mp 82-83 °C. H NMR (CDCl3): δ 1.10 (t,
J ) 7.4, 3H, CH3 prop), 1.39, (t, J ) 7.1, 3H, CH3), 1.82 (sx, J
) 7.4, 2H, CH2 prop), 3.52-3.67 (m, 2H, CH2NH), 4.33 (q, J
) 7.1, 2H, CH2O), 4.71-4.85 and 4.97-5.12 (2 dd, 2H, CH2N),
5.53-5.68 (m, 1H, CHCl), 7.25-7.40 and 7.42-7.52 (2m, 5H
Ar), 8.03 (s, 1H, H-3), 8.87 (s, 1H, H-6), 9.21 (br s, 1H, NH,
exchanges with D2O). IR (CHCl3) cm-1: 3280 (NH), 1663 (CO).
1-(5-Am in o-p yr a zol-1-yl)-3-p h en oxy-2-p r op a n -2-ol (7b).
The compound was prepared according to the synthetic
sequence described for compound 7a , starting from 6b to give
7b (1.98 g, 85%) as a light yellow solid; mp 112-113 °C. 1H
NMR (CDCl3): δ 1.70 (br s, 1H, OH, disappears with D2O),
3.72-3.99 (m, 4H, CH2N + NH2, 2H disappear with D2O),
4.13-4.29 (m, 2H, CH2O), 4.30-4.48 (m, 1H, CHOH), 5.50-
5.53 (m, 1H, H-4), 6.38-7.06 and 7.21-7.39 (2m, 6H, 5H Ar
Meth od B. Exa m p le. 4-P r op yla m in o-1-styr yl-1H-p yr a -
zolo[3,4-b]p yr id in e-5-ca r boxylic Acid Eth yl Ester (12a ).
DBU (5 g, 33.44 mmol) was added to 10a (10 mmol), and the
mixture was heated at 90 °C for 8 h. Absolute ethanol (5 mL)
was added to give the crude 12a , which was then recrystallized
+ H-3). IR (CHCl3) cm-1
(C12H15N3O2) C, H, N.
: 3500-3100 (OH + NH2). Anal.
1
from absolute ethanol with a 40% yield; mp 147-148 °C. H
2-{[2-(2-Hydr oxy-2-ph en yl-eth yl)-2H-pyr azol-3-ylam in o]-
m et h ylen e}m a lon ic Acid Diet h yl E st er (8a ). Diethyl
ethoxymethylenemalonate (2.27 g, 10 mmol) was added to 7a ,
and the mixture was heated to 120 °C for 2 h and then cooled
to room temperature. After diethyl ether (20 mL) was added,
a white solid precipitated. The crude product was filtered off
and then recrystallized from absolute ethanol to give 8a (3.47
g, 93%) as a white solid; mp 128-129 °C. 1H NMR (CDCl3): δ
1.31 and 1.37 (2t, 6H, 2CH3), 1.6-1.8 (br s, 1H, OH disappears
with D2O), 4.12-4.38 (m, 6H, 3CH2), 5.07-5.18 (dd, 1H, CHO),
6.06 (d, 1H, H-3), 7.25-7.47 (m, 6H, 5H Ar + H-4), 8.02 (d,
1H,CHd), 11.16 (d, 1H, NH, exchanges with D2O). IR (CHCl3)
cm-1: 3400-3100 (OH + NH), 1690, 1655 (CO and CdC).
Anal. (C19H23N3O5) C, H, N.
2-{[2-(2-H yd r oxy-3-p h e n oxy-p r op yl)-2H -p yr a zol-3-
yla m in o]m eth ylen e}m a lon ic Acid Dieth yl Ester (8b). The
compound was prepared according to the synthetic sequence
described for compound 10a starting from 7b, to give 8b (2.82
g, 70%) as a light yellow solid; mp 72-73 °C. 1H NMR
(CDCl3): δ 1.30 and 1.36 (2t, 6H, 2CH3), 2.60-3.00 (very br s,
1H, OH, disappears with D2O), 3.71-4.02 and 4.17-4.40 (2m,
8H, 4CH2), 4.42-4.52 (m, 1H, CH), 6.08 (d, 1H, H-4), 6.85-
7.03 and 7.21-7.32 (2m, 5H Ar), 7.45 (d, 1H, H-3), 8.08 (d,
1H, CHd), 11.16 (d, 1H, NH, exchanges with D2O). IR (CHCl3)
cm-1: 3400-3100 (OH + NH), 1685, 1655 (CO and CdC).
Anal. (C20H25N3O6) C, H, N.
NMR (CDCl3): δ 1.23 (t, J ) 7.4, 3H, CH3 prop), 1.41 (t, J )
7.2, 3H, CH3), 1.85 (sx, J ) 7.4, 2H, CH2 prop), 3.56-3.70 (m,
2H, CH2N), 4.35 (q, J ) 7.2, 2H, CH2O), 7.20-7.43 and 7.51-
7.60 (2m, 6H, 5H Ar + CHd), 8.12 (d, J ) 14.8, 1H, CHd),
8.14 (s, 1H, H-3), 8.88 (s, 1H, H-6), 9.29 (br s, 1H, NH,
exchanges with D2O). IR (CHCl3) cm-1: 1655 (CO).
Biologica l Meth od s. [3H]CHA, [125I]AB-MECA, [3H]CGS
21680, and [R32P]ATP were obtained from DuPont-NEN
(Boston, MA). DPCPX was purchased from RBI (Natik, MA).
Adenosine deaminase was from Sigma Chemical Co. (St. Louis,
MO).
A1 a n d A2A Recep tor Bin d in g. Displacement of [3H]CHA
(31 Ci/mmol) from A1 AR in bovine cortical membranes and of
[3H]CGS 21680 (42.1 Ci/mmol) from A2A AR in bovine striatal
membranes was performed as described.54 Adenosine A1 recep-
tor affinities with [3H]DPCPX as radioligand were determined
according to Pirovano et al.55 Measurements with [3H]DPCPX
were performed in the presence and in the absence of 1 mM
GTP.
A3 AR Recep tor Bin d in g. [125I]AB-MECA binding to A3
AR in bovine cortical membranes was performed in 50 mM
Tris, 10 mM MgCl2, and 1 mM EDTA buffer (pH 7.4) contain-
ing 0.2 mg of proteins, 2 U/mL adenosine deaminase, and 20
nM DPCPX.23a Incubations were carried out in duplicate for
90 min at 25 °C. Nonspecific binding was determined in the
presence of 50 µM R-PIA and represented approximately 30%
of the total binding. The binding reaction was terminated by
filtration through a Whatman GF/C filter, washing three times
with 5 mL of ice-cold buffer.
4-Ch lor o-1-(2-ch lor o-2-p h en yl-eth yl)-1H-p yr a zolo[3,4-
b]p yr id in e-5-ca r boxylic Acid Eth yl Ester (9a ). POCl3 (14
g, 91 mmol) was added to 8a (3.73 g, 10 mmol), and the
mixture was refluxed for 12 h and then cooled to room
temperature. The excess of POCl3 was removed by distillation
under reduced pressure. H2O (20 mL) was then carefully added
to the residue, and the suspension was extracted with CHCl3
(3 × 20 mL). The organic solution was washed with H2O (10
mL), dried (MgSO4), filtered, and concentrated under reduced
pressure. The crude brown oil was purified by column chro-
matography (Florisil 100-200 Mesh) using CHCl3 as eluant
to afford the pure product 9a (2.18 g, 60%) as a white solid;
All compounds were routinely dissolved in dimethyl sulfox-
ide (DMSO) and diluted with assay buffer to the final
concentration, where the amount of DMSO never exceeded 2%.
At least six different concentrations spanning 3 orders of
magnitude, adjusted appropriately for the IC50 of each com-
pound, were used. IC50 values, computer-generated using a
nonlinear regression formula on a computer program (Graph-
Pad, San Diego, CA), were converted to Ki values, knowing
the Kd values of radioligands in the different tissues and using
the Cheng and Prusoff equation.56 The dissociation constants
(Kd) of [3H]CHA, [3H]CGS 21680, and [125I]AB-MECA were 1.2,
14, and 1.02 nM, respectively.
1
mp 72-73 °C. H NMR (CDCl3): δ 1.44 (t, J ) 7.1, 3H, CH3),
4.45 (q, J ) 7.1, 2H, CH2O), 4.85-4.95 and 5.05-5.20 (2 dd,
2H, CH2N), 5.52-5.62 (m, 1H, CHCl), 7.25-7.51, (m, 5H Ar),
8.21 (s, 1H, H-3), 9.02 (s, 1H, H-6). IR (CHCl3) cm-1: 1710
(CO). Anal. (C17H15N3O2Cl2) C, H, N.
Ad en ylyl Cycla se Assa y. The adenylyl cyclase assay was
performed as previously described.57 The adenylyl cyclase
activity was measured by monitoring the conversion of [R32P]-
ATP to [R32P]cAMP.58 The method involved addition of [R32P]-
ATP to membranes in the presence of forskolin to stimulate
adenylyl cyclase and papaverine as a phosphodiesterase
inhibitor. Briefly, enzyme activity was routinely assayed in a
100 µL reaction mixture containing 50 mM HEPES/NaOH
buffer, pH 7.4, 2 mM MgCl2, 1 mM DTT, 0.1 mg/mL creatine
phosphokinase, 0.1 mg/mL bacitracin, 0.5 mg/mL creatine
phosphate, 0.1 mM ATP, 0.05 mM cAMP, 15 units/mL myo-
kinase, 2 units/mL adenosine deaminase, 10 M GTP, 1 µCi
[R32P]ATP, 0.2 mM papaverine, and 0.1 mM forskolin. The
incubation was started by the addition of membranes (10-20
g of proteins) and carried out for 15 min at 23 °C. The reaction
was terminated by placing assay tubes in an ice bath and
adding 0.5 mL of a stop solution containing 120 mM Zn-
(C2H3O2)2/[3H]cAMP (10 000-20 000 cpm/sample) and then 0.6
mL of 144 mM Na2CO3. The total radiolabeled cAMP was
4-Ch lor o-1-(2-ch lor o-3-p h en oxy-p r op yl)-1H-p yr a zolo-
[3,4-b]p yr id in e-5-ca r boxylic Acid Eth yl Ester (9b). The
compound was prepared according to the synthetic sequence
described for compound 9a starting from 8b, to give 9b (1.97
g, 50%) as a white solid; mp 70-71 °C. 1H NMR (CDCl3): δ
1.45 (t, J ) 7.1, 3H, CH3), 4.28 (d, 2H, CH2N), 4.46 (q, J )
7.1, 2H, CH2OCO), 4.76-4.92 (m, 1H, CHCl), 4.97-5.04 (d,
2H, CH2OAr), 6.84-7.03 and 7.22-7.57 (2m, 5H Ar), 8.25 (s,
1H, H-3), 9.03 (s, 1H, H-6). IR (CHCl3) cm-1: 1720 (CO). Anal.
(C18H17N3O3Cl2) C, H, N.
Meth od A. Exa m p le. 4-P r op yla m in o-1-(2-ch lor o-2-p h e-
n yleth yl)-1H-p yr a zolo[3,4-b]p yr id in e-5-ca r boxylic Acid
Eth yl Ester (10a ). To a solution of 9a (10 mmol) in anhydrous
toluene (20 mL), propylamine (40 mmol) was added, and the
reaction mixture was stirred at room temperature for 24 h.
After it was extracted with H2O, the organic phase was dried
(MgSO4) and evaporated under reduced pressure; the oil
residue crystallized by adding absolute ethanol (10 mL) to give

Selective Receptor Antagonists
J ournal of Medicinal Chemistry, 2002, Vol. 45, No. 22 4885
isolated on columns of Dowex 50 ion-exchange resin and
albumina as described.58
the technique referred to as receptor-mediated ligand align-
ment was used.52,53 In particular, at the beginning, only three
compounds of the whole training set (13, ∆Gexp ) -13.030 kcal/
mol; 21, ∆Gexp ) -8.400 kcal/mol; and 24, ∆Gexp ) -11.910
kcal/mol), superimposed according to the DISCO-derived
model, were used to develop the pseudoreceptor around the
ligands. In detail, at the tips of the vectors generated by PrGen
for each functional group of the ligands, residue templates
were docked and oriented. When possible, amino acids were
chosen on the basis of published studies on the binding site
for A1 ligands, performed by site-directed mutagenesis experi-
ments. Otherwise, the knowledge of the primary amino acid
sequence of the rat A1 AR reported by Mahan and co-workers51
was applied to pick up residues. Accordingly, (i) the side chain
of Thr1(91) acting as a hydrogen bond acceptor was placed as
the complementary counterpart to the xanthine NH. In fact,
it has been reported that Thr1(91) interacts with the N6
substituent of A1 ligands.3c To grow the amino acid sequence
of the pseudoreceptor, some additional residues were added
to the first one. In particular, Gln2(92), Ser3(93), and Ser4-
(94) were attached at the C terminus of Thr1(91), while Leu5-
(90), Ile6(89), Leu7(88), Val8(87), and Pro9(86) were attached
at the corresponding N terminus. (ii) The lone pair vector
present on the xanthine N9 was neutralized by a hydrogen
bond vector of the imidazole NH of His12(251), reported as
an important residue in A1 AR-antagonist interactions,50b
while Ile14(252), Leu15(253), and Asn16(254) were added to
the C terminus of His12(251). (iii) The carbonyl oxygen at the
2-position of 13 was involved in a hydrogen bond with the
imidazole NH of His10(278), found to critically influence the
binding of both agonists and antagonists to A1AR.50b An
additional residue, Thr11(277), reported as an essential amino
acid for agonist binding (interacting with the sugar portion of
the ligands),50c,63 has been added to His10(278) to fill the region
of space around the N9 atom of both xanthine and adenosine
A1 antagonists. In a similar way, Ser13(281) has been placed
in front of the nitrogen at the 9-position, according to the model
reported by Poulsen and co-workers.6 (iv) Hydrophobic vectors
were found by PrGen, perpendicular to the adenosine or
xanthine planar nucleus of the ligands, suggesting π-π
interactions with the receptor. On the other hand, mutagenesis
experiments on the human A2A AR found that mutation of
Tyr271 (conserved in the rat A1 AR) with nonaromatic residues
led to a great decrease in ligand affinity.50a On the basis of
these findings, the adenosine or xanthine portion of the ligands
was engaged in a π-π stacking interaction with Tyr17(271),
while Ile18(272) and Ala19(273) (taken from the primary
sequence of rat A1 AR) were connected to its C terminus to
build, together with Ile20, a wall of the cavity accommodating
the C8 substituent. (v) Finally, the above-mentioned Ile20,
together with Ala21, Glu22, and His23, were arbitrarily chosen
and added to the growing pseudoreceptor with the aim of
counterbalancing all of the remaining vectors generated by the
program on the ligands, not yet saturated by the other
residues. The final pseudoreceptor consisted of 23 amino acids.
Next, all of the remaining ligands of the training set were
inserted into the pseudoreceptor cavity to obtain the final
model with embedded inhibitors.
The antagonist behavior of some compounds was examined
for their ability to completely reverse the inhibition of forsko-
lin-stimulated adenylyl cyclase activity induced by the A1
selective agonist CHA. Experiments were performed evaluat-
ing the effects of multiple antagonist concentrations (10 nM
to 10 µM) on the inhibition of adenylyl cyclase activity induced
by 100 nM CHA. The compounds tested were dissolved in
DMSO and then diluted with 50 mM HEPES/NaOH buffer,
pH 7.4, so that the final DMSO concentration never exceeded
1%. The data were analyzed as competition curves and by
nonlinear regression analysis for models of one or two nonin-
teracting sites (GraphPad).
Molecu la r Mod elin g. All calculations were performed in
vacuo, on Indigo 2 and Octane R12000 Silicon Graphics
workstations. A set of seventeen structurally diverse A1 AR
selective antagonists 13-29 (Figure 3 and Table 2) was taken
from the literature. The selection was performed on the basis
of two rules: (i) the greatest structural diversity, essential
requirement for obtaining a meaningful model, and (ii) Ki
values spanning among ∼4-5 orders of magnitude, to evaluate
which pharmacophoric elements were necessary to the an-
tagonist for displaying the highest degree of selectivity. When
available,59 X-ray data were used as starting geometries;
otherwise, the interactive building procedure of MacroModel60
was applied to draw the initial geometries to be submitted to
the energy minimization protocol. A conformational analysis
was carried out with the AMBER* force field, using two
different strategies, which depended on the number of signifi-
cant rotatable bonds present in the molecule. Systematic
nested rotation of each selected dihedral angle was performed
for compounds with two or less rotatable bonds, employing 10
degree increments and energy-minimizing the resulting con-
formations. On the contrary, the conformational space of more
flexible compounds was explored by random search, using the
Monte Carlo option implemented in MacroModel. Starting
from different randomly generated initial conformations,
several parallel Monte Carlo cycles were run. For each cycle,
the following parameters were used automatic setup and 1000
as the maximum number of search interactions. Both in the
systematic and in the random approach, conformations with
energy higher than 5 kcal/mol above the minimum energy
conformer were discarded. The selection of this energy cutoff
value was based on the hypothesis that for a majority of
ligand-protein complexes, the bioactive conformations are
within such a threshold.61 The search was stopped when
results from different runs were nearly identical.
The high number of conformations produced by each cycle
was reduced by means of a cluster analysis (XCluster option).
Resulting geometries of the selected low energy conformers
were reoptimized with semiempirical quantum mechanics
calculations, using the Hamiltonian AM1 as implemented in
MOPAC package.
The DISCO approach, as implemented in Sybyl, was sub-
sequently applied to derive an optimal superimposition of the
selected structures. Among the diverse solutions provided by
the program, the selection of a meaningful pharmacophore
model was done choosing the one with the highest number of
pharmacophoric points and the lowest tolerance value, an
index of the validity of the alignment, usually ranging from
0.5 to 2.5 Å. For all of the conformers of compounds 13-23
selected by DISCO, molecular electrostatic potentials and
hydropathic fields were calculated, using MOPAC (AM1) and
HINT computational packages, respectively. The HINT hy-
drophobic fields were calculated with the Essential Hydrogen
Treatment and via Bond Polar Proximity. Next, while the
pseudoreceptor generator software PrGen was employed to
build an atomistic binding site model for the A1 AR, a method
originally developed by Marengo and Todeschini62 and adapted
for pseudoreceptor modeling by Vedani and co-workers52 was
applied to select a training set from compounds 13-29.
To achieve a high correlation between experimentally
derived and calculated binding energies (∆Gexp vs ∆Gcalcd), the
correlation coupling protocol was applied, leading to the
optimization of the pseudoreceptor, without changing position,
orientation, and conformation of the ligands. In the next step,
the pharmacophore was allowed to relax by minimizing the
ligands without constraints while the receptor remained fixed
(ligand relaxation). This allows one to remove the strain
possibly imposed to the ligands by the receptor during cor-
relation-coupled refinement but usually leads to a less highly
correlated model. Therefore, correlation-coupled receptor mini-
mization followed by unconstrained ligand relaxation was
repeated until a highly correlated pseudoreceptor model was
obtained in the relaxed state (designated as the equilibrated
receptor). To validate the equilibrated receptor, its potency to
predict free energies of binding (∆Gpred) for an external set of
To circumvent problems associated with the mutual obscur-
ing of functional groups within a pharmacophore hypothesis,

4886 J ournal of Medicinal Chemistry, 2002, Vol. 45, No. 22
Bondavalli et al.
(12) Suzuki, F.; Shimada, J .; Mizumoto, H.; Karasawa, A.; Kubo, K.;
Nonaka, H.; Ishii, A.; Kawakita, T. Adenosine A1 Antagonists.
2. Structure-Activity Relationships on Diuretic Activities and
Protective Effects against Acute Renal Failure. J . Med. Chem.
1992, 35, 3066-3075.
(13) Nonaka, H.; Ichimura, M.; Takeda, M.; Kanda, T.; Shimada, J .;
Suzuki, F.; Kase, H. KW-3902, a Selective High Affinity An-
tagonist for Adenosine A1 Receptors. Br. J . Pharmacol. 1996,
117, 1645-1652.
(14) Mu¨ller, C. E. A1 Adenosine Receptors Antagonists. Exp. Opin.
Ther. Pat. 1997, 7, 419-440.
(15) Mu¨ller, C. E.; Geis, U.; Grahner, B.; Lanzner, B.; Eger, K. Chiral
Pyrrolo[2,3-d]pyrimidine and Pyrimido[4,5-b]indole Deriva-
tives: Structure-Activity Relationships of Potent, Highly Ste-
reoselective A1 Adenosine Receptors Antagonists. J . Med. Chem.
1996, 39, 2482-2491.
(16) Daly, J . W.; Hong, O.; Padgett, W. L.; Shamin, M. T.; J acobson,
K. A.; Ukena, D. Non-Xanthine Heterocycles: Activity as
Antagonists of A1- and A2-Adenosine Receptors. Biochem. Phar-
macol. 1988, 37, 655-664.
(17) Murphy, K. M.; Snyder, S. H. Adenosine Receptors in Rat
Testes: Labeling with 3H-Cyclohexyladenosine. Life Sci. 1981,
28, 917-920.
(18) Williams, M.; Risley, E. A.; Huff, J . R. Interaction of Putative
Anxiolytic Agents with Central Adenosine Receptors. Can. J .
Physiol. Pharmacol. 1981, 59, 897-900.
ligands (the test set reported in Table 2) was examined. For
this purpose, the test set ligands were relaxed within the fixed
equilibrated pseudoreceptor, applying the same refinement
protocol as described for the training set ligands (see ligand
relaxation). The linear regression obtained for the training set
was used to estimate free energies of binding for the test set
derivatives.52
In the present study, a coupling constant of 1.0 and a
maximum allowed rmsd of 0.100 kcal/mol for the predicted vs
experimental inhibition constants of all correlation-coupled
minimization procedures were used. The target rmsd was
limited to a maximum of 0.200 kcal/mol. Solvation energies of
the ligands were calculated according to Still,64 and entropy
corrections were considered following Searle.65 Compounds 13,
16, and 19-26 taken from the literature, characterized by A1
AR antagonist activity ranging from 0.19 to 4300 nM, have
been used in this study to build a 10 compound training set.
Affinities of the investigated compounds (constituting both the
training and the test set) were in part collected from the
literature (compounds 13-29) under the assumption that all
of these substances are acting through the same mechanism
and binding site and in part experimentally determined
(compounds 10a ,e,i,k and 11h ). Taking into account the
Gibbs-Helmholtz equation, conversion of experimental inhibi-
tion constants (Ki) to free energies of binding were calculated
as follows: ∆Gexp ) RTln(Ki) ) 1.34 (kcal/mol) log(Ki) at 20
°C. The complex between the pseudoreceptor and compound
13 was saved with PrGen as a pdb file and then transferred
to the Viewer module of Insight II (2000) software,66 in turn
used to generate Figure 6.
(19) Daly, J . W.; Hutchinson, K. D.; Secunda, S. I.; Shi, D.; Padgett,
W. L.; Shamin, M. T. 1-Methyl-4-substituted-1H-pyrazolo[3,4-
b]pyridine-5-carboxylic Acid Derivatives: Effect of Structural
Alterations on Activity at A1 and A2 Adenosine Receptors. Med.
Chem. Res. 1994, 4, 293-306.
(20) (a) Akane, A.; Kuroda, S.; Itani, H.; Shimizu, Y. Patent NO
WO9803507, 29-1-1998, CA 128:154090. (b) Akane, A.; Nish-
imura, S.; Kuroda, S.; Itani, H. Patent NO J P10182643, 7-7-
1998, CA 129:144876.
(21) Schenone, S.; Bruno, O.; Fossa, P.; Ranise, A.; Menozzi, G.; Mosti,
L.; Bondavalli, F.; Martini, C.; Trincavelli, L. Synthesis and
Biological Data of 4-Amino-1-(2-chloro-2-phenylehtyl)-1H-pyra-
zolo[3,4-b]pyridine-5-carboxylic Acid Ethyl Ester Derivatives, a
New Series of A1-Adenosine Receptor (A1AR) Ligands. Bioorg.
Med. Chem. Lett. 2001, 11, 2529-2531.
Ack n ow led gm en t. This work was financially sup-
ported by MIUR (cofin.2000, ex 40%, prot. MM032-
37197_002).
Su p p or tin g In for m a tion Ava ila ble: 1H NMR and IR
data of some representative compounds. This material is
available free of charge via the Internet at http://pubs.acs.org.
(22) Benoit, G. Hydroxyalcoylhydrazine. Bull. Soc. Chim. Fr. 1939,
6, 708-715.
(23) Da Settimo, F.; Primofiore, G.; Taliani, S.; Marini, A. M.; La
Motta, C.; Novellino, E.; Greco, G.; Lavecchia, A.; Trincavelli,
L.; Martini, C. 3-Aryl[1,2,4]triazino[4,3-a]benzimidazol-4(10H)-
ones: A New Class of Selective A1 Adenosine Receptor Antago-
nists. J . Med. Chem. 2001, 44, 316-327.
(24) Franchetti, P.; Cappellacci, L.; Marchetti, S.; Trincavelli, L.;
Martini, C.; Mazzoni, M. R.; Lucacchini, A.; Grifantini, M. 2′C-
methyl Analogues of Selective Adenosine Receptor Agonists:
Synthesis and Binding Studies. J . Med. Chem. 1998, 41, 1708-
1715.
Refer en ces
(1) Collis, M. G.; Hourani, S. M. O. Adenosine Receptor Subtypes.
Trends Pharmacol. Sci. 1993, 14, 360-366.
(2) Ralevic, V.; Burnstock, G. Receptors for Purines and Pyrim-
idines. Pharmacol. Rev. 1998, 50, 413-492.
(3) (a) Palczewski, K.; Kumasaka, T.; Hori, T.; Behnke, C. A.;
Motoshima, H.; Fox, B. A.; Le Trong, I.; Teller, D. C.; Okada,
T.; Stenkamp, R. E.; Yamamoto, M.; Miyano, M. Crystal
Structure of Rhodopsin: A G Protein-Coupled Receptor. Science
2000, 289, 739-745. (b) Biancucci, A.-M.; Bigi, M.; Biagi, G.;
Giorgi, I.; Livi, O.; Scartoni, V. A 3D Model of the Human A1
Adenosine Receptor. An Evaluation of the Binding Free-Energy
with Ligands. Drug Des. Discovery 1998, 15, 149-156. (c)
Rivkees, S. A.; Barbhaiya, H.; IJ zerman, A. P. Identification of
the Adenine Binding Site of the Human A1 Adenosine Receptor.
J . Biol. Chem. 1999, 274, 3617-3621.
(25) Zbinden, P. PrGen 2.1.1; Biographics Laboratory: Basel, CH,
1997.
(26) Numbering of pseudoreceptor residues is given as follows. In
parentheses, the amino acid sequence number taken from ref
69 is indicated. On the contrary, the number directly associated
with the amino acid three letter notation is assigned automati-
cally by PrGen. As an example, Thr1(91) is the amino acid Thr91
of the primary sequence and the residue number 1 of the PrGen
labeling.
(27) van Galen, P. J . M.; Leusen, F. J . J .; Ijzerman, A. P.; Soudijn,
W. Mapping the N6 Region of the Adenosine A1 Receptor with
Computer Graphics. Eur. J . Pharmacol. Mol. Pharmacol. Sect.
1989, 172, 19-27.
(28) van der Wenden, E. M.; van Galen, P. J . M.; Ijzerman, A. P.;
Soudijn, W. Mapping the Xanthine C8-Region of the Adenosine
A1 Receptor with Computer Graphics. Eur. J . Pharmacol. Mol.
Pharmacol. Sect. 1991, 206, 315-323.
(29) van Galen, P. J . M.; van Vlijmen, H. W. T.; Ijzerman, A. P.;
Soudijn, W. A Model for the Antagonist Binding Site on the
Adenosine A1 Receptor, Based on Steric, Electrostatic, and
Hydrophobic Properties. J . Med. Chem. 1990, 33, 1708-1713.
(30) Peet, N. P.; Lentz, N. L.; Meng, E. C.; Dudley, M. W.; Ogden, A.
M. L.; Demeter, D. A.; Weintraub, H. J . R.; Bey, P. A Novel
Synthesis of Xanthines: Support for a New Binding Mode for
Xanthines with Respect to Adenosine at Adenosine Receptors.
J . Med. Chem. 1990, 33, 3127-3130.
(4) J i, X.-D.; von Lubitz, D.; Olah, M. E.; Stiles, G. L.; J acobson, K.
A. Species Differences in Ligand Affinity at Central A3-Adenos-
ine Receptors. Drug Dev. Res. 1994, 33, 51-59.
(5) Mu¨ller, C. E.; Stein, B. Adenosine Receptors Antagonists:
Structures and Potential Therapeutic Applications. Curr. Pharm.
Des. 1996, 2, 501-530.
(6) Poulsen, S.-A.; Quinn, R. J . Adenosine Receptors: New Op-
portunities for Future Drugs. Bioorg. Med. Chem. 1998, 6, 619-
641.
(7) von Lubitz, D. K. J . E.; Lin, R. C. S.; Popik, P.; Carter, M. F.;
J acobson, K. A. Adenosine A3 Receptor Stimulation and Cerebral
Ischemia. Eur. J . Pharmacol. 1994, 263, 59-67.
(8) Liang, B. T.; Swierkosz, T. A.; Herrmann, H. C.; Kimmel, S.;
J acobson, K. A. Adenosine and Ischemic Preconditioning. Curr.
Pharm. Des. 1999, 5, 1029-1041.
(9) Mu¨ller, C. E. A2a Adenosine Receptor Antagonists. Future Drugs
for Parkinson’s Disease? Drugs Future 2000, 25, 1043-1052.
(10) Mu¨ller, C. E. A1 Adenosine Receptors and their Ligands:
Overview and Recent Developments. Farmaco 2001, 56, 77-
80.
(11) Kuroda, S.; Akahane, A.; Itani, H.; Nishimura, S.; Durkin, K.;
Tenda, Y.; Sakane, K. Novel Adenosine A1 Receptor Antagonists.
Synthesis and Structure-Activity Relationships of a Novel
Series of 3-(2-Cyclohexenyl-3-oxo-2,3-dyhydropyridazin-6-yl)-2-
phenylpyrazolo[1,5-a]pyridines. Bioorg. Med. Chem. 2000, 8, 55-
64.
(31) Dooley, M. J .; Quinn, R. J . The Three Binding Domain Model of
Adenosine Receptors: Molecular Modeling Aspects. J . Med.
Chem. 1992, 35, 211-216.
(32) van der Wenden, E. M.; Ijzerman, A. P.; Soudijn, W. A Steric
and Electrostatic Comparison of Three Models for the Agonist/
Antagonist Binding Site on the Adenosine A1 Receptor. J . Med.
Chem. 1992, 35, 629-635.

Selective Receptor Antagonists
J ournal of Medicinal Chemistry, 2002, Vol. 45, No. 22 4887
(33) Dooley, M. J .; Kono, M.; Suzuki, F. Conformational Search for
the N6-Substituted Adenosine Analogues and Related Adenosine
A1 Receptor Antagonists. Bioorg. Med. Chem. 1996, 4, 917-921.
(34) Doytchinova, I.; Petrova, S. “N6-N7”sA Modification of the "N6-
C8" Model for the Binding Site on Adenosine A1 Receptors with
Improved Steric and Electrostatic Fit. Med. Chem. Res. 1998,
8, 143-152.
(35) Bruns, R. F.; Fergus, J . H.; Badger, E. W.; Bristol, J . A.; Santay,
L. A.; Hartman, J . D.; Hays, S. J .; Huang, C. C. Binding of the
A1-Selective Adenosine Antagonist 8-Cyclopentyl-1,3-dipropyl-
xanthine to Rat Brain Membranes. Naunyn-Schmiedeberg’s
Arch. Pharmacol. 1987, 335, 59-63.
(36) Trivedi, B. K.; Bruns, R. F. [1,2,4]Triazolo[4,3-a]quinoxalin-4-
amines: A New Class of A1 Receptor Selective Adenosine
Antagonists. J . Med. Chem. 1988, 31, 1011-1014.
(37) Ceccarelli, S.; D’Alessandro, A.; Prinzivalli, M.; Zanarella, S.
Imidazo[1,2-a]quinoxalin-4-amines: A Novel Class of Nonxan-
thine A1-Adenosine Receptor Antagonists. Eur. J . Med. Chem.
1998, 33, 943-955.
(38) van Galen, P. J . M.; Nissen, P.; van Wijngaarden, I.; Ijzerman,
A. P. 1H-Imidazo[4,5-c]quinolin-4-amines: Novel Non-Xanthine
Adenosine Antagonists. J . Med. Chem. 1991, 34, 1202-1206.
(39) Colotta, V.; Cecchi, L.; Catarzi, D.; Filacchioni, G.; Martini, C.;
Tacchi, P.; Lucacchini, A. Synthesis of Some Tricyclic Heteroaro-
matic Systems and their A1 and A2a Adenosine Binding Activity.
Eur. J . Med. Chem. 1995, 30, 133-139.
(40) Hamilton, H. W.; Ortwine, D. F.; Worth, D. F.; Bristol, J . A.
Synthesis and Structure-Activity Relationships of Pyrazolo[4,3-
d]pyrimidin-7-ones as Adenosine Receptor Antagonists. J . Med.
Chem. 1987, 30, 91-96.
(41) Thompson, R. D.; Secunda, S.; Daly, J . W.; Olsson, R. A. N6,9-
Disubstituted Adenines: Potent, Selective Antagonists at A1
Adenosine Receptor. J . Med. Chem. 1991, 34, 2877-2882.
(42) Poulsen, S.-A.; Quinn, R. J . Pyrazolo[3,4-d]pyrimidines: C4, C6
Substitution Leads to Adenosine A1 Receptor Selectivity. Bioorg.
Med. Chem. Lett. 1996, 6, 357-360.
Nicholson, A.; Schofield, P. R. A Threonine Residue in the
Seventh Transmembrane Domain of the Human A1 Adenosine
Receptor Mediates Specific Agonist Binding. J . Biol. Chem. 1994,
269, 2373-2376.
(51) Mahan, L. C.; McVittie, L. D.; Smyk-Randall, E. M.; Nakata,
H.; Monsma, F. J ., J r.; Gerfen, C. R.; Sibley, D. R. Cloning and
Expressing of an A1 Adenosine Receptor from Rat Brain. Mol.
Pharmacol. 1992, 40, 1-7.
(52) Zbinden, P.; Dobler, M.; Folkers, G.; Vedani, A. PrGen: Pseu-
doreceptor Modeling Using Receptor-Mediated Ligand Aligne-
ment and Pharmacophore Equilibration. Quant. Struct.-Act.
Relat. 1998, 17, 122-130.
(53) Bassoli, A.; Merlini, L.; Morini, G.; Vedani, A.
A Three-
Dimensional Receptor Model for Isovanillic Sweet Derivatives.
J . Chem. Soc., Perkin Trans. 1998, 2, 1449-1454.
(54) Colotta, V.; Catarzi, D.; Varano, F.; Cecchi, L.; Filacchioni, G.;
Martini, C.; Trincavelli, L.; Lucacchini, A. 1,2,4-Triazolo[4,3-a]-
quinoxalin-1-one: A Versatile Tool for the Synthesis of Potent
and Selective Adenosine Receptor Antagonists. J . Med. Chem.
2000, 43, 1158-1164.
(55) Pirovano, I. M.; IJ zerman, A. P.; van Galen, P. J . M.; Soudijn,
W. The Influence of Molecular Structure of N6-(aminoalkyl)-
adenosines on Adenosine Receptor Affinity and Intrinsic Activity.
Eur. J . Pharmacol. 1989, 172, 185-193.
(56) Cheng, Y. C.; Prusoff, W. H. Relation between the Inhibition
Constant Ki and the Concentration of Inhibitor which Causes
Fifty Percent Inhibition (IC50) of an Enzyme Reaction. Biochem.
Pharmacol. 1973, 22, 3099-3108.
(57) Ferrarini, P. L.; Mori, C.; Manera, C.; Martinelli, A.; Mori, F.;
Saccomanni, G.; Barili, P. G.; Betti, L.; Giannaccini, G.; Trin-
cavelli, L.; Lucacchini, A. A Novel Class of Highly Potent and
Selective A1 Adenosine Antagonists: Structure-Affinity Profile
of a Series of 1,8-Naphthyridine Derivatives. J . Med. Chem.
2000, 43, 2814-2823.
(58) J ohnson, R. G.; Alvarez, R.; Salomon, Y. Determination of
Adenylyl Cyclase Catalytic Activity Using Single and Double
Column Procedures. Methods Enzymol. 1994, 238, 31-56.
(59) Allen, F. H.; Bellard, S.; Brice, M. D.; Cartwright, B. A.;
Doubleday, A.; Higgs, H.; Hummelink, T.; Hummelink-Peters,
B. G.; Kennard, O.; Motherwell, W. D. S.; Rodgers, J . R.; Watson,
D. G. The Cambridge Crystallographic Data Centre: Computer-
based Search, Retrieval, Analysis and Display of Information.
Acta Crystallogr. 1979, B35, 2331-2332.
(43) Akahane, A.; Katayama, H.; Mitsunaga, T.; Kato, T.; Kinoshita,
T.; Kita, Y.; Kusunoki, T.; Terai, T.; Yoshida, K.; Shiokawa, Y.
Discovery of 6-Oxo-3-(2-phenylpyrazolo[1,5-a]pyridin-3-yl)-1(6H)-
pyridazinebutanoic Acid (FK 838): A Novel Non-Xanthine
Adenosine A1 Receptor Antagonist with Potent Diuretic Activity.
J . Med. Chem. 1999, 42, 2, 779-783.
(44) Martin, Y. C.; Bures, M. G.; Dahaner, E. A.; De Lazzer, J .; Lico,
I.; Pavlik, P. A. A Fast Approach to Pharmacophore Mapping
and its Application to Dopaminergic and Benzodiazepine Ago-
nists. J . Comput.-Aided Mol. Des. 1993, 7, 83-102.
(45) AM1-MOPAC: QCPE-Program No. 455, Version 6.0; Quantum
Chemistry Program Exchange, Indiana University: Blooming-
ton, IN.
(60) MacroModel Version 5.5; Columbia University: New York, 1996.
(61) Ho¨ltje, H. D.; Folkers, G. Small Molecules. In Molecular Model-
ing. Basic Principles and Applications; Mannhold, R., Kubinyi,
H., Timmerman, H., Eds.; VCH Publishers: New York, 1996;
pp 23-35.
(62) Marengo, E.; Todeschini, R. A New Algorithm for Optimal,
Distance-Based Experimental Design. Chemom. Intell. Lab. Syst.
1992, 16, 37-44.
(63) Tucker, A. L.; Robeva, A. S.; Taylor, H. E.; Holeton, D.; Bockner,
M.; Lynch, K. R.; Linden, J . A1 Adenosine Receptor. Two Amino
Acids are Responsible for Species Differences in Ligand Recogni-
tion. J . Biol. Chem. 1994, 269, 27900-27906.
(46) SYBYL Version 6.5; Tripos Inc.: St. Louis, MO, 1999.
(47) Hint! Version 2.30S; EduSoft, L. C.: Ashland, VA, 19.
(48) J acobson, K. A.; Suzuki, F. Recent Developments in Selective
Agonists and Antagonists Acting at Purine and Pyrimidine
Receptors. Drug Dev. Res. 1996, 39, 289-300.
(49) Hess, S.; Mu¨ller, C. E.; Frobenius, W.; Reith, U.; Klotz, K.-N.;
Eger, K. 7-Deazaadenines Bearing Polar Substituents: Structure-
Activity Relationships of New A1 and A3 Adenosine Receptor
Antagonists. J . Med. Chem. 2000, 43, 4636-4646.
(50) (a) Kim, J .; Wess, J .; van Rhee, A. M.; Scho¨neberg, T.; J acobson,
K. A. Site-directed Mutagenesis Identifies Residues Involved in
Ligand Recognition in the Human A2a Adenosine Receptor. J .
Biol. Chem. 1995, 270, 13987-13997. (b) Olah, M. E.; Ren, H.;
Ostrowski, J .; J acobson, K. A.; Stiles, G. L. Cloning, Expression,
and Characterization of the Unique Bovine A1 Adenosine Recep-
tor. J . Biol. Chem. 1992, 267, 10764-10770. (c) Townsend-
(64) Still, W. C.; Tempczyk, A.; Hawley, R. C.; Hendrickson, T.
Semianalytical Treatment of Solvation for Molecular Mechanics
and Dynamics. J . Am. Chem. Soc. 1990, 112, 6127-6129.
(65) Searle, M. S.; Williams, D. H. The Cost of Conformational
Order: Entropy Changes in Molecular Associations. J . Am.
Chem. Soc. 1992, 114, 10690-10697.
(66) Insight II 2000; Accelrys, Inc.: Scranton Road, San Diego, CA.
J M0209580