Synthesis, biological activity and molecular modelling of new trisubstituted 8-azaadenines with high affinity for A1 adenosine receptors

Article abstract of DOI:10.1016/j.ejmech.2006.08.014

We describe here the synthesis and biological activity of new 8-azaadenines bearing both a phenyl group on C(2) and a 9-benzyl group substituted in the ortho position with a Cl or a F atom or a CF₃ group, to verify the synergistic effect of a c

Full text of DOI:10.1016/j.ejmech.2006.08.014

European Journal of Medicinal Chemistry 42 (2007) 1e9
http://www.elsevier.com/locate/ejmech
Original article
Synthesis, biological activity and molecular modelling of new
trisubstituted 8-azaadenines with high affinity for A1
adenosine receptors
a
a
a
a
Irene Giorgi a, , Anna Maria Bianucci , Giuliana Biagi , Oreste Livi , Valerio Scartoni ,
Michele Leonardi a, Daniele Pietra a, Alessio Coi a, Ilaria Massarelli b, Fatena Ahmad Nofal a,
Francesca Lidia Fiamingo a, Paola Anastasi a, Giuliano Giannini a
*
a Dipartimento di Scienze Farmaceutiche, Universita` di Pisa, via Bonanno 6, 56126 Pisa, Italy
b Dipartimento di Chimica e Chimica Industriale, Universita` di Pisa, Via Risorgimento 35, 56126 Pisa, Italy
Received 23 June 2006; received in revised form 21 July 2006; accepted 11 August 2006
Available online 5 October 2006
Abstract
We describe here the synthesis and biological activity of new 8-azaadenines bearing both a phenyl group on C(2) and a 9-benzyl group
substituted in the ortho position with a Cl or a F atom or a CF3 group, to verify the synergistic effect of a combination of these substitution
patterns on binding with the A1 adenosine receptors. In position N6 aliphatic and cycloaliphatic substituents were chosen which had been shown
to bind well with the A1 receptors. Because of the high lipophilicity of these kinds of molecules, we also introduced a hydroxyalkyl substituent in
the same position. The compounds obtained generally showed a very good affinity and selectivity for A1 receptors. Some of the compounds
showed Ki in the nanomolar range, one even in the subnanomolar range (0.6 nM). Molecular docking calculations were performed in order
to evaluate the interaction energies between the bovine A1 receptor model and the selected ligands, and then to correlate these energies with
biological activities of the ligands as obtained from the experiments. Molecular docking analysis suggests different binding modes towards
A1 receptors that are plausible for these ligands.
Ó 2006 Elsevier Masson SAS. All rights reserved.
Keywords: 8-Azaadenines; 1,2,3-Triazolo[4,5-d]pyrimidines; A1 adenosine receptor ligands
1. Introduction
central nervous system and in many peripheral tissues and
mediate diverse physiological effects.
Adenosine receptors are member of the P1 family and are
coupled with more subtypes of G proteins which regulate
the activity of adenylyl cyclase and other effector systems
including calcium or potassium ion channels, phospholipase
A or C, and guanylate cyclase [1]. They are classified as A1,
The ubiquitous distribution of adenosine receptors in mam-
malian cell types and the existence of four distinct subtypes
together with the variability of physiological responses mean
that agonists and antagonists have to be highly selective in
their action (with respect to receptor subtype and tissue
type) to be of value as therapeutics [2].
The importance of A1 adenosine receptors is demonstrated
by a great number of papers published in these last years about
this subject [3e5]. Hundreds of A1 agonists and antagonists
have been assayed to obtain information about the receptor ac-
tive site and to select more potent and selective compounds
A2A, A2B, A3 subtypes, and have been characterised using
selective ligands. A1 receptors are widely distributed in the
* Corresponding author. Tel.: þ39 0502219549.
E-mail address: igiorgi@farm.unipi.it (I. Giorgi).
0223-5234/$ - see front matter Ó 2006 Elsevier Masson SAS. All rights reserved.
doi:10.1016/j.ejmech.2006.08.014

2
I. Giorgi et al. / European Journal of Medicinal Chemistry 42 (2007) 1e9
with potential therapeutic effects. A1 antagonists are studied as
antihypertensives and potassium sparing diuretics with kidney-
protection properties [4], and could be useful in Alzheimer’s
disease and other CNS disorders [6], and in cardiac therapy
and kidney diseases [7].
In the last years we have synthesised a great number of A1
adenosine receptor ligands, especially 8-azaadenines, varying
the substituents in positions 2, 6 and 9. We have shown that
an unsubstituted phenyl group, linked in the 2 position,
improves very much the activity of 9-benzyl-8-azaadenine,
which is inactive with no substituent in the same position; and
that the phenyl group is the best one among a series of 2-aryl
substituents (2-phenyl-9-benzyl-8-azaadenine Ki ¼ 40 nM) [8].
Starting from this very effective nucleus, we have intro-
duced various substituents in position N6: some N6-alkyl or
N6-cycloalkyl substituted 2-phenyl-9-benzyladenines demon-
strated a very good affinity for A1 receptors [9,10].
Comparing the affinity of these compounds with that of the
corresponding 2-n-butyl derivatives we concluded that in posi-
tion 2 an aromatic group was better than an aliphatic one; in-
stead, in position N6 the situation is reversed [9]. When the
substituent in the N6 position is a cyclopentyl group, the affin-
ity for the A1 receptor is very high for both the series, 2-phe-
nyl-9-benzyl and 2-butyl-9-benzyl-8-azaadenines (Ki ¼ 11 nM
and Ki ¼ 170 nM, respectively) showing that the receptor con-
tains at least three lipophilic pockets capable of interacting
with lipophilic groups substituted in the 2, N6 and 9 positions
of the 8-azaadenine nucleus [9]. Many other trisubstituted 8-
azaadenines were synthesised [11e14] and assayed and
some demonstrated a high affinity towards A1 receptors con-
firming our hypothesis about the three lipophilic pockets of
the A1 receptors (Fig. 1, A) [11e14].
(Ki ꢀ 43 nM). Comparing these results with those of the anal-
ogous lacking the halogen in the ortho position of the 9-benzyl
substituent, we observed an increase in affinity due to the pres-
ence of the halogen. So we hypothesised that the halogen in
this position must influence the interaction of the ligand
with the receptor by a positive electronic effect on the pep
interaction of the phenyl moiety of the 9-benzyl substituent,
or, as in the case of the 9-ortho-fluorobenzyl substituent, by
a hydrogen bond with the receptor [16].
Considering the results obtained in the past, we describe
here the synthesis and biological activity of new 8-azaadenines
bearing both a phenyl group on C(2) and a 9-benzyl group
substituted in the ortho position with a Cl or a F atom or
a CF3 group to verify the synergistic effect of a combination
of these substitution patterns on binding with A1 receptors
(Fig. 1, C). For the position N6 aliphatic and cycloaliphatic
substituents were chosen which had been shown to bind well
with the A1 receptors. Because of the high lipophilicity of
these kinds of molecules, we also introduced hydroxyalkyl
substituents in the same position, as in the past we had shown
that an hydroxyl group on the N6 substituent does not lower
the affinity of trisubstituted 8-azaadenines [13].
2. Chemistry
The synthetic route to prepare compounds 9a, 5e16b and
5e16c (Scheme 1) employed a known two-step reaction in
the presence of sodium ethoxide [16]: the first step was the
1,3 dipolar addition reaction of the suitable benzylazide and
cyanoacetamide to give 1-benzyl or 1-(2-chloro)benzyl or
1-(2 fluoro)benzyl-4-carbamoyl-5-amino-1H-1,2,3-triazole (2a,
2b, 2c) which were not isolated; then, in the same flask, ethyl
benzoate was added to obtain the 8-azahypoxanthines 3a [17],
3b or 3c by annulation reaction.
This protocol gave very low yields when 2-trifluorobenzy-
lazide was used, recovering a large amount of 1,2,3-triazole
2d. Probably the steric hindrance of the ortho substituent hin-
ders the electrophile attack of the ethyl benzoate on the amino
group of the triazole. So, we put ethyl benzoate in the starting
mixture, together with cyanoacetamide and then we added
azide dropwise: by this method we obtained the desired prod-
uct 3d (25% yield).
In other papers we have described the synthesis and biolog-
ical results of 8-azaadenines lacking the substituent in the 2
position but having an o-Cl- or o-F-benzyl moiety in position
9 (Fig. 1, B). Among them, the best compounds as ligands for
A1 receptors were the ones having a cyclohexyl or cyclopentyl
ring on the N6 position [15,16], showing a very good affinity
The reaction of 3aed with phosphorus oxychloride pro-
duced the 6-chloro-8-azapurines 4aed which, by nucleophilic
displacement of the chlorine atom by the suitable amine, gave
the desired N6-substituted-8-azaadenines.
3. Biochemistry
All the new compounds (9a, 5e16b, c, d) were tested by
radioligand binding assays for affinity towards A1, A2A, and
A3 adenosine receptors. In Table 1 the results of A1 adenosine
receptor binding are reported; in the A2A and A3 adenosine re-
ceptor binding assays, all the compounds showed inhibition
values < 50% at 1 mM, so were considered to have very
weak affinity.
Fig. 1. Design of the new compounds C.

I. Giorgi et al. / European Journal of Medicinal Chemistry 42 (2007) 1e9
3
Scheme 1. Reagents: (a) CNCH2CONH2, EtONa; (b) C6H5COOC2H5; (c) POCl3; (d) RNH2.
4. Molecular modelling
screening, based on force field grids, contains the highest scor-
ing compounds ranked in order of their scores. Generalized-
Born Surface Area (GBSA), a new scoring function within
the most recent version of the program, accounts for the effects
of solvation by water. In several systems the DOCK program
has been shown to give quite good correlations with respect to
the docking scores [19] (see later in Section 7.2).
The X-ray crystal structure of bovine Rhodopsin with retinal
(PDB code: 1U19) was used as a template for complex con-
struction [18]. A homology model for the bovine A1 receptor
was built from this crystal structure, after previous sequence
alignment. The construction of a model for the complex of
the bovine A1 receptor subtype with the antagonist DPCPX
was accomplished after the binding site had been properly iden-
tified. A single-point ab initio calculation with no further geom-
etry optimisation was performed for DPCPX, using the
HartreeeFock method, in order to calculate electrostatic poten-
tials at a grid of points around the ligand. The GAFF force field
parameters and partial charges were assigned. Preliminary en-
ergy minimizations and molecular dynamics (MD) simulations
were then performed in order to optimise the geometry of the
model. The lowest energy conformer, collected after MD simu-
lations, was further energy minimised and taken as a reliable 3D
theoretical model for the bovine A1 receptor. It was exploited
for the subsequent molecular docking procedure. The com-
plexes, created for each ligand of Table 1, were then constructed
starting from DPCPX and screened by molecular docking
within the DOCK approach. This approach, in the most recent
version of its implementation, allows docking of flexible li-
gands in the active site of a receptor in order to search for their
favourable orientations and conformations. A pre-calculated
force field-scoring grid, based on molecular mechanic interac-
tion energies consisting of van der Waals and electrostatic com-
ponents, is generated to rank each ligand. The energy score is
determined both by types and positions of the atoms of the
ligand on the energy grids. Subsequently, score optimisation al-
lows the conformation and orientation of the molecule to be ad-
justed to improve the score. The resulting output file for each
5. Results and discussion
5.1. Structureeactivity relationships
The compounds obtained generally showed very good af-
finity and selectivity for A1 receptors. Some of the compounds
showed Ki values in the nanomolar range, one even in the sub-
nanomolar range (0.6 nM). We compared the affinities of the
new compounds (series b, c, d) with the affinities of similar
compounds obtained in the past (series a). As A1 affinity
data of series a (Table 1) have been performed on bovine
membranes, we decided to again use the same membranes to
compare the results of the new compounds with those previ-
ously described (series a, Table 1). To evaluate selectivity, af-
finities for A2A and A3 were measured using cloned human
receptors. In fact for A1 subtype there is a good amino acid se-
quence homology, since standard antagonists as theophilline
and DPCPX showed an affinity at bovine A1 receptors compa-
rable to those reported at the cloned human ones [20,21]. All
the compounds were considered to be very weak ligands (%
inhibition < 50% at 1 mM) towards A2A and A3 receptors
and consequently, selective for A1 subtype.
The trend of the A1 Ki values is interestingly the same for the
first three series (series a, b, c) leading to the assumption that these
compounds interact with the receptor in a very similar manner. In

4
I. Giorgi et al. / European Journal of Medicinal Chemistry 42 (2007) 1e9
Table 1
Biological results of A1 adenosine receptor binding
NHR
N
NHR
NHR
N
NHR
N
N
N
N
N
N
N
N
N
N
N
N
N
F
N
N
N
N
N
CF3
Cl
R
Compound
Ki (nM)
Compound
Ki (nM)
Compound
Ki (nM)
Compound
Ki (nM)
eCH3
eC2H5
enC3H7
enC4H9
enC6H13
5a [8]
6a [8]
7a [8]
8a [9]
9a
353 ꢁ 30
91 ꢁ 8
5b
6b
7b
8b
9b
191 ꢁ 15
463 ꢁ 18
101 ꢁ 9.5
79 ꢁ 8
5c
6c
7c
8c
9c
25 ꢁ 2
11 ꢁ 1
5d
6d
7d
8d
9d
14.9 ꢁ 12
n.d.
n.d.
37 ꢁ 4
3.2 ꢁ 0.2
4.3 ꢁ 0.3
25 ꢁ 2
11 ꢁ 1.5
13.2 ꢁ 1.7
6.5 ꢁ 4
n.d.
31 ꢁ 3
10a [13]
11.3 ꢁ 1
10b
11 ꢁ 1
10c
15 ꢁ 2
10d
7.8 ꢁ 8
11a [13]
5 ꢁ 0.5
11b
24 ꢁ 2
12a [8]
13a [9]
11
12b
13b
9 ꢁ 1
12c
13c
6.7 ꢁ 0.8
12d
13d
0.6 ꢁ 0.2
6.8 ꢁ 5
1.6
34 ꢁ 3
1.15 ꢁ 0.3
eCH2CH2OH
eCH2CHOHCH3
eCH2CH2CH2OH
14a [12]
15a [12]
16a [12]
38
14b
15b
16b
116 ꢁ 10
32 ꢁ 3
14c
15c
16c
54 ꢁ 5
14d
15d
16d
n.d.
n.d.
7.5
21.4
4.8 ꢁ 0.5
3.7 ꢁ 0.2
33 ꢁ 3
147 ꢁ 20
n.d. ¼ not determined.
order to investigate the structure activity relationships we com-
pared the affinity of compounds 12aec and 13aec (see Table 1)
with the analogous not substituted in position 2 obtained in the
past: 9-benzyl-N6-cyclopentyl-8-azaadenine (Ki ¼ 127 ꢁ 8.7 nM)
[15], 9-o-chlorobenzyl-N6-cyclopentyl-8-azaadenine (Ki ¼ 21 ꢁ
1.5 nM) [15], 9-o-fluorobenzyl-N6-cyclopentyl-8-azaadenine
(Ki ¼ 10.5 ꢁ 0.7 nM) [16], 9-benzyl-N6-cyclohexyl-8-azaadenine
(Ki ¼ 121 ꢁ 5.1 nM) [15], 9-o-chlorobenzyl-N6-cyclohexyl-
8-azaadenine (Ki ¼ 43 ꢁ 2.5 nM) [15], 9-o-fluorobenzyl-N6-
cyclohexyl-8-azaadenine (Ki ¼ 19.5 ꢁ 1.2 nM) [16]. 2-Phenyl
substituted compounds (see Ki values of the compounds 12aec
and 13aec in Table 1) show better affinity than unsubstituted
analogues in the same position. These observations confirm our
hypothesis of the presence in the A1 receptor of three
lipophilic pockets in front of positions 2, N6 and 9 of adenine
which can well bind a phenyl, an alkyl and a benzyl group, respec-
tively [11].
Regarding the compounds of the fourth series (d), the intro-
duction of a CF3 group further increased the lipophilicity of
the molecules; this fact made the biological assays difficult
to perform, because the compounds precipitated in the aque-
ous environment of the assays, even if in the compound a hy-
droxyl group is present on the N6 substituent (compounds
14de16d). So we performed the assays after reducing the con-
centration of the tested compounds and repeated every test till
we obtained at least three quantitatively homogeneous results
to mediate among them. This technique helped to overcome
the problem of low water solubility in some cases. So, proba-
bly due to their high affinity, we could measure the Ki values
for compounds 5d (Ki ¼ 14.9 nM), 8d (Ki ¼ 6.5 nM), 10d
(7.8 nM), 12d (Ki ¼ 0.6 nM) and 13d (Ki ¼ 6.8 nM). Promis-
ing results were obtained.
Regarding the substituent on the 9-benzyl group, we can
see that the series 9-(o-chlorobenzyl) substituted compounds
showed in general less good results, followed by the 9-(unsub-
stituted) benzyl-8-azaadenines. The 9-(o-fluorobenzyl) and
the 9-(o-trifluoromethylbenzyl) substituted compounds are in
general the best ones. Obviously, the order of activity with re-
spect to the ortho substituent is not directly related to its steric
hindrance or electronegativity. So the high affinity of the com-
pounds of the series c and d could be partially attributable to
a possible hydrogen bond of a fluorine atom, as is confirmed
by the molecular modelling study. However, the activity trend
cannot be explained in terms of simple molecular descriptors,
since the sum of the different contributions to the interaction en-
ergy must take into account the possibility of different binding
modes, as was suggested by the molecular modelling studies.
5.2. Molecular modelling
The molecular docking procedure was performed in order
to evaluate the interaction energies between the bovine A1

I. Giorgi et al. / European Journal of Medicinal Chemistry 42 (2007) 1e9
5
receptor model and the selected ligands, and then to correlate
these energies with biological activities of the ligands. The to-
tal interaction energy estimated by means of the DOCK pro-
gram consisted of electrostatic and van der Waals (vdW)
contributions.
The values of the correlation coefficient R2 referred to the
total energy, and the two terms with respect to the entire set
of ligands and to classes a, b, c, and d, are reported in Table 2.
The vdW term seems to give good correlation with the ex-
perimental data and the total energy shows the same trend,
since the electrostatic term does not give a substantial contri-
bution towards the total energy. The binding of the ligands to
the bovine A1 receptor seems to be driven by favourable vdW
interactions and this energetic term could reproduce the exper-
imental data quite well. On the contrary, the electrostatic term
does not give a correlation with Ki except for class d where the
vdW contribution is the highest in terms of mean value along
the class (ꢂ50.30 kcal/mol). The molecular docking procedure
also suggested the binding mode of the ligands. The most
straightforward orientation implies that the bicyclic nuclei of
the ligands take the same location as the nucleus of the known
antagonist DPCPX. Fig. 2 shows the ligand 7c in the site of the
bovine A1 receptor.
Some major interactions justify its good activity. In partic-
ular, hydrogen bonds are observed between the N(3) and the
hydroxyl group of Thr91 and between the fluorine of the or-
tho-substituted benzyl group in N(9) and the hydroxyl group
of Ser246, thus stabilising the molecule in the binding site.
Furthermore, lipophilic interactions involving the propyl
group in N6 and residues Leu88, Leu250, Val271 are reported;
moreover, the phenyl group in C(2) is able to interact with
Val189, Trp247, and His251.
The molecular docking procedure gives interesting sugges-
tions about different binding modes of the ligands to the bind-
ing site. In a previous work we refer to this above-described
orientation as ‘‘adenosine-like’’ or AL [11]. A second orienta-
tion was considered, in which the N(9)-substituent arranges
itself in the same binding site as the N6-substituent. This im-
plies that the phenyl substituent in position 2 of the bicyclic
nuclei retains analogous locations in both orientations. This
second orientation may be identified as the one obtained
from the AL orientation after a 180ꢃ rotation of the plane con-
taining the bicyclic nucleus around a pseudo-symmetry axis
aligned along the straight line containing the C(2) and C(5)
atoms. (Fig. 3).
Fig. 2. Stereoview of ligand 7c and the characteristic residues of its binding
site. Thr91 and Ser246 coloured by atom type; residues Leu88, Leu250,
Val271 in yellow; and residues Val189, Trp247, and His251 in magenta.
(For interpretation of the references to colour in this figure legend, the reader
is referred to the web version of this article.)
is supposed to give a similar hydrophobic contribution as the
substituted phenyl at N(9). Furthermore, the CF3 substituent
alone seems to lead to the RE orientation, as observed from
the results of DOCK within the class d (Fig. 4). This could
be explained hypothesizing that the CF3-ortho-substituted ben-
zyl group gives better interactions with residues Leu88,
Leu250, Val271 than any N6 substituents characterising the
series 5e16.
6. Conclusions
We described herein the synthesis and biological activity of
new 8-azaadenines bearing both a phenyl group on C(2) and
a 9-benzyl group substituted in the ortho position with a Cl
or a F atom or a CF3 group.
This orientation was referred to as ‘‘reverse’’ or RE orien-
tation. From the molecular docking, class d and series 13ae
d seem to be arranged in RE orientation. This fact was
expected for 13aed, since the cyclohexyl substituent at N6
The compounds obtained generally showed very good af-
finity and selectivity for A1 receptors. The 9-(o-fluorobenzyl)
(series c) and the 9-(o-trifluoromethylbenzyl)substituted com-
pounds (series d) are more active than the 9-(o-chloro or
Table 2
Values of correlation coefficient R2 between biological activity (Ki) and total interaction energy calculated by DOCK or its single components (electrostatic and van
der Waals contributions) for the entire dataset and for each class (aed)
Dataset
Total
Class a
Total
Class b
Total
Class c
Total
0.64
Class d
Total
vdW
0.66
es
vdW
0.78
es
vdW
0.78
es
vdW
0.63
es
vdW
0.67
es
0.68
0.002
0.82
0.05
0.79
0.12
0.02
0.64
0.60

6
I. Giorgi et al. / European Journal of Medicinal Chemistry 42 (2007) 1e9
docking analysis suggests different binding modes towards the
A1 receptors that are plausible for these antagonists.
7. Experimental
7.1. Chemistry
Melting points were determined on a Kofler hot-stage appa-
ratus and are uncorrected. IR spectra in Nujol mulls were re-
1
corded on a Mattson Genesis series FTIR spectrometer. H
NMR spectra were recorded on a Bruker AC 200 spectrometer
in d units using TMS as an internal standard; the compounds
were dissolved in the solvent indicated in Table 3. TLC was
performed on precoated silica gel F254 plates (Merck). Micro-
analyses (C, H, N) were carried out on a Carlo Erba elemental
analyser (Model 1106) and were within ꢁ 0.4% of the theoret-
ical values.
7.1.1. Benzylazide (1a), 2-chlorobenzylazide (1b) and
2-fluorobenzylazide (1c)
These compounds were prepared as described in the litera-
ture [22].
7.1.2. 2-Trifluoromethyl-benzylazide (1d)
Fig. 3. Starting AL orientation of 13c (red), and its relative RE orientation after
molecular docking (blue). (For interpretation of the references to colour in this
figure legend, the reader is referred to the web version of this article.)
A
mixture of 2-(trifluoromethyl)benzyl bromide
(0.026 mol), ethanol (18 ml), water (2 ml) and sodium azide
(2.80 g, 0.043 mol) was refluxed for 24 h. After cooling, the
mixture was filtered and the filtrate was concentrated to
10 ml; then water (5 ml) was added and the mixture was ex-
tracted with chloroform (3 ꢄ 10 ml). The organic phases
were dried and evaporated to obtain a yellow oil; 3.18 g, yield
60%; b.p. 200e201 ꢃC. IR 2102 cmꢂ1 (N3). 1H NMR (DMSO-
d6) 4.625 (s, 2H, CH2); 7.59e7.80 (m, 4H, arom). Elemental
analysis (C, H, N). Rf 0.53 (n-hexaneeCHCl3 8:2).
unsubstituted) benzyl ones (series a and b). The high affinity
of the compounds of the series c and d could be partially at-
tributable to a possible hydrogen bond of a fluorine atom, as
is confirmed by the molecular modelling study. The molecular
7.1.3. 5-Amino-1-(2-trifluoromethyl-benzyl)-1H-[1,2,3]
triazole-4-carboxylic acid amide (2d)
To a solution of EtONa (0.05 mol) in 10 ml of absolute
EtOH, cyanoacetamide (1.05 g, 0.0125 mol) was added and
the mixture was heated at 60 ꢃC for 0.5 h. Then 2-(trifluorome-
thyl)benzylazide (2.5 g, 0.0125 mol) was added and the mix-
ture was refluxed for 6 h, cooled and concentrated under
reduced pressure. To the residue, water (20 ml) was added
and the solution acidified with 4 N acetic acid at pH ¼ 5 to ob-
tain a white solid which was crystallised from EtOHeCHCl3
1:10; yield: 81%; m.p. 198e199 ꢃC. IR 3396, 3307, 3177
1
(NH); 1640 (C]O) cmꢂ1. H NMR 5.60 (s, 2H, CH2); 6.51
(s, 2H, exch); 7.19 (broad s, 2H, exch); 7.48e7.81 (m, 4H,
arom). Elemental analysis (C, H, N).
7.1.4. 2-Phenyl-9-(2-substituted)benzyl-8-azahypoxanthines
(3b, c)
To a stirred solution of EtONa (2.76 g, 0.12 g atom of Na)
in 30 ml of absolute EtOH, cyanacetamide (2.51 g, 0.03 mol)
was added. The mixture was refluxed for 0.5 h then a solution
of the suitable azide (0.03 mol) and ethyl benzoate (0.3 mol)
in absolute EtOH (10 ml) was added drop by drop and the
Fig. 4. Starting AL orientation of 5a (red), and the RE orientations of class d re-
sulting after molecular docking (coloured by atom). (For interpretation of the
references to colour in this figure legend, the reader is referred to the web ver-
sion of this article.)

I. Giorgi et al. / European Journal of Medicinal Chemistry 42 (2007) 1e9
7
Elemental analysis (C, H, N). 3c: yield: 50%; m.p. 285 ꢃC.
IR 1695 (C]O) cmꢂ1 1H NMR (DMSO) 12.85 (s, 1H,
exch); 8.16 (m, 2H, arom); 7.59e7.21 (m, 7H, arom); 5.84
Table 3
Chemical and physical properties of the newly synthesised compounds
.
Compound Amine
Yield Rf
(%)
m.p. (ꢃC) Analysis
(C, H, N)
(s, 2H, CH2). Elemental analysis (C, H, N).
9a
5b
n-Hexylamine
Methylaminef
Ethylamineg
58
43
27
41
53
80
0.34b 175e176h C25H26N6
0.55a 212e213h C18H15ClN6
0.27b 170e172h C19H17ClN6
0.30b 203e204h C20H19ClN6
0.30b 174e176h C21H21ClN6
0.36b 152e154h C23H25ClN6
0.18b 202e203i C20H17ClN6
7.1.5. 2-Phenyl-9-(2-trifluoromethyl)benzyl-8-
azahypoxanthine (3d)
6b
7b
n-Propylamine
n-Butylamine
n-Hexylamine
Cyclopropylamine 33
Cyclobutylamine 25
8b
9b
To a stirred solution of EtONa (1.10 g, 0.048 g atom of Na)
in 15 ml of absolute EtOH, cyanacetamide (0.950 g, 0.01 mol)
and ethyl benzoate (15 g, 0.1 mol) were added. The mixture
was refluxed for 0.5 h then a solution of the suitable azide
(0.03 mol) in absolute EtOH (5 ml) was added drop by drop
and the mixture refluxed for 6 h, then cooled and concentrated
under reduced pressure. To the residue, water (20 ml) was
added and the solution acidified with 4 N acetic acid at
pH ¼ 5. The solid precipitated was filtered and crystallised
from isopropanol to afford 0.705 g. Yield: 22%; m.p.
10b
11b
12b
13b
14b
15b
0.25b 208i
C21H19ClN6
Cyclopentylamine 31
Cyclohexylamine 66
0.36b 156e157i C22H21ClN6
0.39b 151e152i C23H23ClN6
0.25c 223e224i C19H17ClN6O
0.14a 206e208i C20H19ClN6O
Ethanolamine
(ꢁ)-1-Amino-2-
propanolj
83
38
16b
3-Amino-1-
propanolj
63
0.45e 202e203i C20H19ClN6O
5c
6c
Methylaminef
Ethylamineg
n-Propylamine
n-Butylamine
n-Hexylamine
78
82
59
66
61
0.43a 223e224h C18H15FN6
0.21b 195e197h C19H17FN6
0.31b 200e201h C20H19FN6
0.34b 148e150h C21H21FN6
0.38b 113e114h C23H25FN6
0.26b 180e181i C20H17FN6
0.44a 127e128i C22H21FN6
0.43b 153e155i C23H23FN6
1
272 ꢃC. IR 1695 (C]O) cmꢂ1. H NMR (DMSO) 12.89 (s,
7c
8c
1H, exch); 8.13 (m, 2H, arom); 7.85e7.27 (m, 7H, arom);
5.96 (s, 2H, CH2). Elemental analysis (C, H, N).
9c
10c
12c
13c
14c
15c
Cyclopropylamine 15
Cyclopentylamine 58
Cyclohexylamine 57
7.1.6. 6-Chloro-2-phenyl-9-(2-substituted)benzyl-8-
azapurines (4aed)
Ethanolamine
(ꢁ)-1-Amino-2-
propanolj
65
60
0.25c 230i
0.18a 202i
C19H17FN6O
C20H19FN6O
A mixture of 2-phenyl-9-(2-substituted)benzyl-8-azahy-
poxanthine (3aed) (3.3 mmol), N,N-diethylaniline (0.8 g,
5.36 mmol) and POCl3 (3.0 ml, 32.55 mmol) was heated at
90 ꢃC for 4 h, then evaporated under reduced pressure. The
residue was crystallised from chloroformeether to give the ti-
tle compounds (2.18 mmol, 60e70% yield). 4a: m.p. 184 ꢃC.
1H NMR (CDCl3) d: 5.94 (s, 2H, NeCH2); 7.38 (m, 3H,
arom); 7.57 (m, 5H, arom); 8.59 (m, 2H, arom). 4b: m.p.
16c
3-Amino-1-
propanolj
39
0.13a 195e196i C20H19FN6O
0.20d 226e227i C19H15F3N6
5d
6d
Methylaminef
Ethylamineg
n-Propylamine
n-Butylamine
n-Hexylamine
31
48
87
54
36
0.25d 234i
0.29d 169e170i C21H19F3N6
0.34d 194i
C22H21F3N6
0.46d 169e170i C24H25F3N6
0.28d 223i
C21H17F3N6
0.47d 163e164i C23H21F3N6
0.49d 123i
C24H23F3N6
0.52e 245e246i C20H17F3N6O
C20H17F3N6
7d
8d
9d
1
165 ꢃC. H NMR (DMSO) d: 6.10 (s, 2H, NeCH2); 7.38e
10d
12d
13d
14d
15d
Cyclopropylamine 84
Cyclopentylamine 80
Cyclohexylamine 78
7.62 (m, 7H, arom); 8.45 (m, 2H, arom). 4c: m.p. 163 ꢃC.
1H NMR (CDCl3) d: 6.04 (s, 2H, NeCH2); 7.23e7.64 (m,
Ethanolamine
(ꢁ)-1-Amino-2-
propanolj
30
64
1
7H, arom); 8.46 (m, 2H, arom). 4d: m.p. 149 ꢃC. H NMR
(CDCl3) d: 6.16 (s, 2H, NeCH2); 7.15e7.79 (m, 7H); 8.52
(m, 2H).
0.60e 241i
C21H19F3N6O
16d
3-Amino-1-
propanolj
65
0.41e 213i
C21H19F3N6O
a
b
c
d
e
f
7.1.7. 2-Phenyl-N6-n-hexyl-9-benzyl-8-azaadenines (9a)
In a well-stoppered flask a mixture of 4a (0.623 mmol), tol-
uene (2 ml), n-hexylamine (1 mmol) and N,N-diethylaniline
(0.63 mmol) was heated at 130 ꢃC for 16 h. The mixture was
then evaporated, diluted with chloroform, washed with 10%
HCl and water. The evaporation of the organic phase gave
an oil which was crystallised (see Tables 3 and 4).
Eluent: chloroform.
Eluent: n-hexaneeethyl acetate 9:1.
Eluent: chloroformemethanol 9.8:0.2.
Eluent: hexaneeethyl acetate 8:2.
Eluent: chloroformemethanol 9.9:0.1.
70% in water.
g
h
i
40% in water.
Crystallisation solvent: 2-propanoleethyl ether.
Crystallisation solvent: 2-propanol.
The free base was obtained from the commercial hydrochloride by treating
with 10% NaOH and extracting with ethyl acetate.
j
7.1.8. General procedure for preparing 2-phenyl-N6-
alkyl or cycloalkyl-9-(2-substituted-benzyl)-8-azaadenines
(5e16b, 5e16c, 5e16d)
mixture was refluxed for 6 h, then cooled and concentrated un-
der reduced pressure. To the residue, water (20 ml) was added
and the solution acidified with 4 N acetic acid at pH ¼ 5. The
solid precipitated was filtered and crystallised from eꢂth1anol.
In a well-stoppered flask a mixture of 4bed (0.623 mmol),
toluene (2 ml), the suitable amine (1 mmol) and N,N-diethyla-
niline (0.09 g, 0.63 mmol) was heated at 130 ꢃC for 16 h. Then
the solution was evaporated under reduced pressure, diluted
with chloroform and washed with 10% HCl and water. After
evaporation of the organic layer the residue was crystallised
(see Tables 3 and 4).
3b: yield: 50%; m.p. 285 ꢃC. IR 1695 (C]O) cm
.
1H
NMR (DMSO) 12.96 (s, 1H, exch); 8.15 (m, 2H, arom);
7.55 (m, 4H, arom); 7.36 (m, 3H arom); 5.88 (s, 2H, CH2).

8
I. Giorgi et al. / European Journal of Medicinal Chemistry 42 (2007) 1e9
Table 4
1H NMR of the newly synthesised compounds
Compound (solvent)
Aliphatic H
Aromatic H
Benzylic H
5.90 (s, 2H)
Exch. H
9a (DMSO-d6)
3.66 (m, 2H), 1.73 (m, 2H), 1.60e1.31
(m 6H), 0.94 (t, J ¼ 7.4 Hz, 3H)
8.40 (m, 2H), 7.59e7.33 (m, 7H)
9.02 (br t, 1H)
5b (DMSO-d6)
6b (DMSO-d6)
7b (DMSO-d6)
8b (DMSO-d6)
3.15 (d, J ¼ 4.6 Hz, 2H)
8.45 (m, 2H), 7.60e7.31 (m, 7H)
8.43 (m, 2H), 7.60e7.30 (m, 7H)
8.45 (m, 2H), 7.64e7.35 (m, 7H)
8.43 (m, 2H), 7.60e7.31(m, 7H)
5.91 (s, 2H)
5.90 (s, 2H)
5.91 (s, 2H)
5.90 (s, 2H)
8.93 (br q, 1H)
9.01 (br t, 1H)
9.04 (br t, 1H)
9.02 (br t, 1H)
3.71 (m, 2H), 1.29 (t, J ¼ 7.0 Hz, 3H)
3.64 (m, 2H), 1.73 (m, 2H), 0.97 (t, J ¼ 7.4 Hz, 3H)
3.68 (m, 2H), 1.70 (m, 2H), 1.41 (m, 2H), 0.94
(t, J ¼ 2.8 Hz, 3H)
9b (DMSO-d6)
3.67 (m, 2H), 1.70 (m, 2H), 1.34 (m, 6H), 0.85
(t, J ¼ 7.4 Hz, 3H)
8.43 (m, 2H), 7.56e7.32 (m, 7H)
5.90 (s, 2H)
9.03 (br t, 1H)
10b (DMSO-d6)
11b (DMSO-d6)
12b (DMSO-d6)
13b (DMSO-d6)
14b (DMSO-d6)
3.26 (m, 1H), 0.87 (m, 2H), 0.75 (m, 2H)
4.56 (m, 1H), 3.73 (m, 2H), 3.54 (m, 2H), 1.87 (m, 2H)
4.73 (m, 1H), 2.06 (m, 2H), 1.73 (m, 4H)
4.33 (m, 1H), 1.99e1.20 (m, 10H)
3.71 (m, 4H)
8.47 (m, 2H), 7.61e7.35 (m, 7H)
8.43 (m, 2H), 7.60e7.35 (m, 7H)
8.41 (m, 2H), 7.64e7.28 (m, 7H)
8.40 (m, 2H), 7.52e7.32 (m, 7H)
8.42 (m, 2H), 7.56e7.30 (m, 7H)
5.91 (s, 1H)
5.91 (s, 2H)
5.90 (s, 2H)
5.90 (s, 2H)
5.91 (s, 2H)
9.13 (br d, 1H)
8.98 (br d, 1H)
9.03 (br d, 1H)
8.93 (br d, 1H)
8.92 (br t, 1H),
4.84 (br t, 1H)
8.90 (br t, 1H),
4.87 (d, 1H)
15b (DMSO-d6)
16b (DMSO-d6)
4.02 (m, 1H), 3.61 (m, 2H), 1.4 (d, J ¼ 6.2 Hz, 3H)
8.43 (m, 2H), 7.52e7.32 (m, 7H)
8.43 (m, 2H), 7.61e7.35 (m, 7H)
5.91 (s, 2H)
5.91 (s, 2H)
3.72 (m, 2H), 3.56 (m, 2H), 1.87 (m, 2H)
8.97 (br t, 1H),
456 (br t, 1H)
8.92 (br q, 1H)
6.32 (br t, 1H)
6.40 (br t, 1H)
6.32 (br t, 1H)
6.37 (br t, 1H)
6.48 (br d, 1H)
6.29 (br d, 1H)
6.22 (br d, 1H)
8.91 (br t, 1H),
4.85 (m, 1H)
5c (DMSO-d6)
6c (CDCl3)
7c (CDCl3)
3.13 (d, J ¼ 3.8 Hz, 3H)
8.47 (m, 2H), 7.52e7.17 (m, 7H)
8.54 (m, 2H), 7.50e7.07 (m, 7H)
8.54 (m, 2 H), 7.51e7.07 (m, 7 H)
8.55 (m, 2H), 7.49e7.10 (m, 7H)
8.53 (m, 2H), 7.49e7.10 (m, 7H)
8.56 (m, 2H), 7.50e7.05 (m, 7H)
8.54 (m, 2H), 7.51e7.10 (m, 7H)
8.53 (m, 2H), 7.51e7.08 (m, 7H)
8.43 (m, 2 H), 7.52e7.32 (m, 7H)
5.87 (s, 2H)
5.89 (s, 2H)
5.89 (s, 2H)
5.89 (s, 2H)
5.88 (s, 2H)
5.89 (s, 2H)
5.89 (s, 2H)
5.89 (s, 2H)
5.91 (s, 2H)
3.87 (m, 2H), 1.42 (t, J ¼ 7.2 Hz, 3H)
3.80 (m, 2H), 1.84 (m, 2H), 1.08 (t, J ¼ 7.6 Hz, 3 H)
3.85 (m, 2H), 1.78e1.53 (m, 4H), 1.01 (t, J ¼ 7.6 Hz, 3H)
3.80 (m, 2H), 1.74e1.33 (m, 8H), 0.90 (t, J ¼ 7.0 Hz, 3H)
3.25 (m, 1H), 1.06 (m, 2H), 0.78 (m, 2H)
4.82 (m, 2H), 2.26 (m, 2 H), 1.78 (m, 6H)
4.43 (m, 1H), 2.24 (m, 2H), 1.84e1.27 (m, 8H)
3.72 (m, 4H)
8c (CDCl3)
9c (CDCl3)
10c (CDCl3)
12c (CDCl3)
13c (CDCl3)
14c (DMSO-d6)
15c (DMSO-d6)
16c (DMSO-d6)
4.01 (m, 1H), 3.62 (m, 2H), 1.14 (d, J ¼ 4.2 Hz, 3H)
8.47 (m, 2H), 7.52e7.20 (m, 7H)
8.45 (m, 2H), 7.50e7.20 (m, 7H)
5.87 (s, 2H)
5.85 (s, 2H)
8.89 m, 1H),
4.0 (m, 1H)
3.72 (m, 2H), 3.52 (m, 2H), 1.85 (m, 2H)
8.95 (br t, 1H),
4.54 (m, 1 H)
6.48 (br t, 1H)
6.30 (br t, 1H)
6.32 (m, 1H)
5d (CDCl3)
6d (CDCl3)
7d (CDCl3)
8d (DMSO-d6)
3.38 (d, J ¼ 4.6 Hz, 3H)
8.54 (m, 2H), 7.78e7.06 (m, 7H)
8.51 (m, 2H), 7.80e7.06 (m, 7H),
8.50 (m, 2H), 7.76e7.07 (m, 7H)
8.45 (m, 2H), 7.24e7.60 (m, 7H)
6.07 (s, 2H)
6.06 (s, 2H)
6.04 (s, 2H)
5.99 (s, 1H)
3.89 (m, 2H), 1.44 (t, J ¼ 7.3 Hz, 3H)
3.79 (m, 2H), 1.82 (m, 2H), 1.09 (t, J ¼ 7.4 Hz, 3H)
3.69 (m, 2H), 1.70 (m, 2H), 1.44 (m, 2H), 0.94
(t, J ¼ 7.4 Hz, 3H)
9.02 (br t, 1H)
9d (CDCl3)
3.81 (m, 2H), 1.79 (m, 2H), 1.40e1.25 (m 6H), 0.90
(t, J ¼ 7.2 Hz, 3H)
8.48 (m, 2H), 7.76e7.05 (m, 7H)
6.05 (s, 2H)
6.30 (m, 1H)
10d (CDCl3)
3.24 (m, 1H), 1.05 (m, 2H), 0.83 (m, 2H)
4.75 (m, 1H), 2.06 (m, 2 H), 1.89e1.57 (m, 6H)
4.33 (m, 1H), 2.02e1.21 (m, 10 H)
3.73 (m, 4H)
8.53 (m, 2H), 7.44e7.10 (m, 7H)
8.38 (m, 2H), 7.85e7.21 (m, 7H)
8.37 (m, 2H), 7.85e7.21 (m, 7H)
8.40 (m, 2H), 7.85e7.18 (m, 7H)
6.07 (s, 2H)
5.99 (s, 2H)
5.98 (s, 2H)
5.99 (s, 2H)
6.41 (br d, 1H)
9.04 (br d, 1H)
8.94 (br d, 1H)
8.89 (br t, 1H),
4.84 (br t, 1H)
8.89 (br t, 1H),
4.88 (br d, 1H)
6.80 (m, 1H)
12d (DMSO-d6)
13d (DMSO-d6)
14d (DMSO-d6)
15d (DMSO-d6)
16d (CDCl3)
4.03 (m, 1H), 3.61 (m, 2H), 1.14 (d, J ¼ 5.8 Hz, 3H)
8.41 (m, 2H), 7.84e7.22 (m, 7H)
8.42 (m, 2H), 7.77e7.10 (m, 7H)
5.98 (s, 2H)
6.05 (s, 1H)
4.06 (m, 2H), 3.73 (m, 2H), 1.98 (m, 2H)
7.2. Technical details of molecular modelling
Energy minimization, performed by means of the AM-
BER8.0 program, was initially run using positional restraints
only for the enzyme heavy atoms with a harmonic potential
All models were built by means of the Sybyl7.0 program
(Tripos) on an SGI Octane R12000 workstation [23].
Standard PARM99 [24] force field parameters were as-
signed to the protein, using the LEAP program in AMBER8.0.
The single-point ab initio calculation on DPCPX was per-
formed by using the Gaussian03 program with the 6-31G* ba-
sis set [25]. The GAFF force field parameters and partial
charges were assigned using the ANTECHAMBER program
as implemented in AMBER8.0.
2
˚
of 1000 kcal/mol A . Subsequently, a second restrained mini-
mization was performed only restraining the backbone atoms
2
with a force constant of 5.0 kcal/mol A . Finally, a further
˚
minimization was made to relax all the atoms without con-
straints. The minimization employed a distance dependent di-
electric constant (4r) and the convergence criterion for the
˚
energy gradient was 0.01 kcal/mol A. After the minimization,
MD simulations were initiated using the pair wise

I. Giorgi et al. / European Journal of Medicinal Chemistry 42 (2007) 1e9
9
Generalised-Born (GB) continuum solvent model introduced
7.3.2. Statistical analysis
Binding parameters were estimated by GraphPAD Prism
software (GraphPAD, San Diego, CA, USA).
by Hawkins and co-workers and implemented in the SANDER
module of AMBER8.0 [26,27]. Simulations employed a 1 fs
time step for 20 000 steps corresponding to a total of
20.00 ps of GB-MD. The final desired temperature of 298 K
was obtained by requesting a heating cycle from 0 to 298 K
over the course of the first 5000 MD steps and Langevin dy-
namics were used to simulate solvent frictional effects with
a collision frequency g ¼ 1.0 psꢂ1. The SHAKE algorithm
was applied to constrain bonds involving hydrogen atoms [28].
Dielectric constants of 1 (interior) and 80 (exterior) were
employed in all GB-MD simulations. The MD simulation of
the bovine A1 receptor in complex with DPCPX, consisted
of 5 ps of heating, 15 ps for equilibration, and 20 ps of MD
Acknowledgement
This research has been supported by the Italian MIUR
`
(Ministero Istruzione Universita Ricerca).
References
[1] B.B. Fredholm, A.P. Ijzerman, K.-N. Klotz, Linden, J. Pharmacol. Rev.
53 (2001) 1e26.
[2] S.-A. Poulsen, J.R. Quinn, Bioorg. Med. Chem. 6 (1998) 619e641.
[3] C.E. Muller, Farmaco 56 (2001) 77e80.
¨
[4] C.E. Muller, Expert Opin. Ther. Patents 7 (1997) 419e440.
¨
[5] K.A. Jacobson, F. Suzuki, Drug Dev. Res. 39 (1996) 289e300.
~
[6] J.A. Ribeiro, A.M. Sebastiao, A. de Mendoc¸a, Prog. Neurobiol. 68 (2003)
˚
simulations for data collection. No cutoff (cutoff ¼ 999 A)
was employed during the GB-MD simulations.
All the complexes, each containing one member of the en-
tire class of ligands, were subjected to molecular docking by
means of the DOCK approach [29] using the DOCK5.2 ver-
sion of the program. The GBSA exploited in the calculations
is a new scoring function, within the DOCK5.2 release, which
has been developed with the aim of accounting for the effects
of solvation by water. It works by implicitly using the GB ap-
proximation. In spite of the many approximations used, it has
been shown that the DOCK program in many cases offers
quite good correlations not only with respect to the orienta-
tions of molecules inside the receptor cavity but also with re-
spect to the docking scores.
377e392.
[7] K. Dalla, J.C. Shiryock, R. Shreeniwas, L. Belardinelli, Curr. Top. Med.
Chem. 32 (2003) 369e385.
[8] G. Biagi, I. Giorgi, O. Livi, V. Scartoni, A. Lucacchini, C. Martini,
P. Tacchi, Farmaco 49 (1994) 93e96.
[9] G. Biagi, I. Giorgi, O. Livi, V. Scartoni, A. Lucacchini, C. Martini,
P. Tacchi, Farmaco 49 (1994) 187e191.
[10] G. Biagi, I. Giorgi, O. Livi, V. Scartoni, A. Lucacchini, Farmaco 51
(1996) 395e399.
[11] A.M. Bianucci, G. Biagi, A. Coi, I. Giorgi, O. Livi, F. Pacchini,
V. Scartoni, A. Lucacchini, B. Costa, Drug Dev. Res. 54 (2001) 52e65.
[12] G. Biagi, I. Giorgi, O. Livi, F. Pacchini, P. Rum, V. Scartoni, B. Costa,
M.R. Mazzoni, L. Giusti, Farmaco 57 (2002) 221e233.
[13] G. Biagi, I. Giorgi, M. Leonardi, O. Livi, F. Pacchini, V. Scartoni,
B. Costa, A. Lucacchini, Eur. J. Med. Chem. 38 (2003) 801e810.
[14] G. Biagi, I. Giorgi, A. Nardi, F. Pacchini, V. Scartoni, A. Lucacchini, Eur.
J. Med. Chem. 38 (2003) 983e990.
7.3. Biochemical assays
[15] G. Biagi, I. Giorgi, O. Livi, C. Manera, V. Scartoni, L. Betti,
G. Giannaccini, A. Lucacchini, J. Med. Chem. 41 (1998) 668e673.
[16] G. Biagi, I. Giorgi, O. Livi, C. Manera, V. Scartoni, L. Betti,
G. Giannaccini, A. Lucacchini, Eur. J. Med. Chem. 34 (1999) 867e875.
[17] P.L. Barili, G. Biagi, O. Livi, V. Scartoni, J. Heterocycl. Chem. 22 (1985)
1607e1609.
7.3.1. Radioligand binding assays
Membranes of bovine cerebral cortex, that contain adeno-
sine A1 receptors, were prepared as previously described [30].
Membranes (40 mg of protein) were incubated at 25 ꢃC for
60 min with 0.4 nM of [3H]DPCPX (Kd 0.4 nM), and with in-
creasing concentrations of the compounds in duplicate, in a final
volume of 0.4 ml of TriseHCl buffer. Nonspecific binding was
measured in the presence of CPA 10 mM. Binding reactions
were terminated by dilution with ice-cold 50 mM TriseHCl
buffer. Samples were then filtered through Whatman GF/C
glass-fibre filters using a Brandell cell harvester or a Millipore
manifold. Filters were washed three times with 2e3 ml of the
same buffer. Bound radioactivity was measured in a liquid scin-
tillation counter (1600 TR Packard) after the addition of 4 ml of
scintillation liquid (Emulsifier-Safe, Packard).
The experimental conditions for adenosine A2A and A3 re-
ceptor binding essays were slightly different. Membranes of
CHO cells expressing recombinant human A2A or A3 receptors
were prepared as previously described [31]. Membranes (40 mg
of protein) were incubated with [3H]ZM241385 4 nM or
[3H]NECA 6 nM in the A2A or A3 experiment, respectively,
and compoundsat aconcentration of 1 mM in duplicate, in a final
volume of 0.4 ml of TriseHCl buffer for 120 min at 25 ꢃC. Non-
specific binding was measured in the presence of 100 mM NECA
in the case of A2A binding assay and 100 mM R-PIA in the case
of A3 binding assay. Samples were handled as mentioned before.
[18] T. Okada, M. Sugihara, A.N. Bondar, M. Elstner, P. Entel, V. Buss, J.
Mol. Biol. 342 (2004) 571e583.
[19] A.C. Illapakurthy, Y.A. Sabnis, B.A. Avery, M.A. Avery, C.M. Wyandt, J.
Pharm. Sci. 92 (2003) 649e655.
[20] H. Nakata, J. Biochem. 206 (1992) 171e177.
[21] K.N. Klotz, J. Hessling, J. Hegler, C. Owman, B. Kull, B.B. Fredholm,
M.J. Lohes, Naunyn Schiedebergs Arch. Pharmacol. 357 (1998) 1e9.
[22] R. Kreher, U. Bergman, Tetrahedron Lett. 17 (1976) 4259e4262.
[23] SYBYL, Version 7.0 TRIPOS Assoc. Inc., St. Louis, MO., http://
www.tripos.com/
[24] J.M. Wang, P. Cieplak, P.A. Kollman, J. Comput. Chem. 21 (2000)
1049e1074.
[25] M.J. Frisch, G.W. Trucks, H.B. Schlegel, G.E. Scuseria, M.A. Robb,
J.R. Cheeseman, et al., Gaussian 03, Revision C.02, Gaussian, Inc., Wall-
ingford CT, 2004.
[26] G.D. Hawkins, C.J. Cramer, D.G. Truhlar, Chem. Phys. Lett. 246 (1995)
122e129.
[27] G.D. Hawkins, C.J. Cramer, D.G. Truhlar, J. Phys. Chem. 100 (1996)
19824e19839.
[28] J.P. Ryckaert, G. Ciccotti, H.J.C. Berendsen, J. Comput. Phys. 23 (1977)
327e341.
[29] T.J.A. Ewing, I.D. Kuntz, J. Comput. Chem. 18 (1997) 1175e1189.
[30] C. Martini, M. Poli, A. Lucacchini, Bull. Mol. Biol. Med. 11 (1986) 1e10.
[31] G. Biagi, A.M. Bianucci, A. Coi, B. Costa, L. Fabbrini, I. Giorgi, O. Livi,
I. Micco, F. Pacchini, E. Santini, M. Leonardi, F. Nofal Ahamad, O. Salerni
LeRoy, V. Scartoni, Bioorg. Med. Chem. 13 (2005) 4679e4693.