Synthesis and biological activity of tricyclic cycloalkylimidazo-, pyrimido- and diazepinopurinediones

Article abstract of DOI:10.1016/j.ejmech.2011.05.023

Syntheses and physicochemical properties of N-cycloalkyl-substituted imidazo-, pyrimido- and 1,3-diazepino[2,1-f]purinediones are described. These derivatives were synthesized by cyclization of 7-halogenoalkyl-8-bromo-1,3- dimethylxanthine derivatives with aminocycloalkanes. The obtained compounds (1-33) were evaluated for their affinity to rat adenosine A₁ and A_2A receptors. Selected compounds were additionally investigated for affinity to the human A₁, A_2A, A_2B and A ₃ receptor subtypes. The results of the radioligand binding assays at adenosine A₁ and A_2A receptors showed that most of the compounds exhibited adenosine A_2A receptor affinity at micromolar or submicromolar concentrations; an annelated pyrimidine ring was beneficial for A_2A affinity. The most potent A_2A ligands of the present series were compounds 6 (K_i 0.33 μM rat A_2A, 0.31 μM human A_2A), 8 (K_i 0.98 μM rat A_2A, 0.42 μM human A_2A) and 15 (K_i 0.24 μM rat A_2A, 0.61 μM human A_2A) with the latter one showing high A _2A selectivity. In NaCl shift assay, 15 was shown to be an antagonist at A_2A receptors. This result was confirmed for the best compounds 6, 8, 15 in cAMP accumulation studies. A 3D-QSAR equation with a good predicting power (q² = 0.88) for A_2A AR affinity was obtained. The compounds were evaluated in vivo as anticonvulsants in MES and ScMet tests and examined for neurotoxicity in mice (i.p.). Most of them showed anticonvulsant activity in chemically induced seizures; among them the diazepinopurinediones were the best (e.g. 31) showing protection in both tests on short time symptoms, without signs of neurotoxicity. Five compounds, 8, 17, 20, 29, and 31, exhibited anticonvulsant activity after peroral application in rats. Structure-activity relationships are discussed including the analysis of lipophilic and spatial properties. The new compounds, which contain a basic nitrogen atom and can therefore be protonated, may be good starting points for obtaining A_2A antagonists with good water-solubility.

Full text of DOI:10.1016/j.ejmech.2011.05.023

European Journal of Medicinal Chemistry 46 (2011) 3590e3607
Contents lists available at ScienceDirect
European Journal of Medicinal Chemistry
journal homepage: http://www.elsevier.com/locate/ejmech
Original article
Synthesis and biological activity of tricyclic cycloalkylimidazo-, pyrimido- and
diazepinopurinediones
a
a
b
b
b
ꢀ
Anna Drabczynska , Olga Yuzlenko , Meryem Köse , Minka Paskaleva , Anke C. Schiedel ,
Janina Karolak-Wojciechowska c, Jadwiga Handzlik a, Tadeusz Karcz a, Kamil Kuder a, Christa E. Müller b,
a,
*
ꢀ
Katarzyna Kiec-Kononowicz
a Jagiellonian University Medical College, Faculty of Pharmacy, Department of Technology and Biotechnology of Drugs, Medyczna 9, PL 30-688 Kraków, Poland
b PharmaCenter Bonn, Pharmaceutical Institute, Pharmaceutical Chemistry I, University of Bonn, An der Immenburg 4, D-53121 Bonn, Germany
c
_
ꢀ
ꢀ
Institute of General and Ecological Chemistry, Technical University of Łódz, Zwirki 36, PL 90-924 Łódz, Poland
a r t i c l e i n f o
a b s t r a c t
Article history:
Syntheses and physicochemical properties of N-cycloalkyl-substituted imidazo-, pyrimido- and 1,3-
diazepino[2,1-f]purinediones are described. These derivatives were synthesized by cyclization of
7-halogenoalkyl-8-bromo-1,3-dimethylxanthine derivatives with aminocycloalkanes. The obtained
compounds (1e33) were evaluated for their affinity to rat adenosine A1 and A2A receptors. Selected
compounds were additionally investigated for affinity to the human A1, A2A, A2B and A3 receptor
subtypes. The results of the radioligand binding assays at adenosine A1 and A2A receptors showed that
most of the compounds exhibited adenosine A2A receptor affinity at micromolar or submicromolar
concentrations; an annelated pyrimidine ring was beneficial for A2A affinity. The most potent A2A ligands
Received 26 October 2010
Received in revised form
10 May 2011
Accepted 10 May 2011
Available online 23 May 2011
Keywords:
Adenosine A1 and A2A receptor antagonists
Tricyclic xanthine derivatives
Cycloalkyl-1,3-diazepino[2,1-f]purinediones
of the present series were compounds 6 (Ki 0.33
m
M rat A2A, 0.31
mM human A2A), 8 (Ki 0.98 mM rat A2A,
0.42 M human A2A) and 15 (Ki 0.24 M rat A2A, 0.61 m
m
m
M human A2A) with the latter one showing high
A2A selectivity. In NaCl shift assay, 15 was shown to be an antagonist at A2A receptors. This result was
confirmed for the best compounds 6, 8, 15 in cAMP accumulation studies. A 3D-QSAR equation with
a good predicting power (q2 ¼ 0.88) for A2A AR affinity was obtained. The compounds were evaluated
in vivo as anticonvulsants in MES and ScMet tests and examined for neurotoxicity in mice (i.p.). Most of
them showed anticonvulsant activity in chemically induced seizures; among them the diazepinopur-
inediones were the best (e.g. 31) showing protection in both tests on short time symptoms, without signs
of neurotoxicity. Five compounds, 8, 17, 20, 29, and 31, exhibited anticonvulsant activity after peroral
application in rats. Structureeactivity relationships are discussed including the analysis of lipophilic and
spatial properties. The new compounds, which contain a basic nitrogen atom and can therefore be
protonated, may be good starting points for obtaining A2A antagonists with good water-solubility.
Ó 2011 Elsevier Masson SAS. All rights reserved.
1. Introduction
Adenosine A1 receptors are particularly ubiquitous within the
central nervous system (CNS), with high levels being expressed in
The adenosine receptor (AR) family consists of four subtypes: A1,
A2A, A2B and A3 [1]. The responses of these four ARs are mediated by
receptor-coupled G proteins, which may activate several different
effector systems including adenylate cyclase, potassium and
calcium channels, phospholipase A2 or C, and guanylate cyclase.
Recent studies indicate a widening role for adenosine receptors in
many therapeutic areas, including immunology, the cardiovascular
system, and various CNS-mediated events such as sleep, neuro-
protection, and pain [1e7].
many regions of the brain. The distribution of adenosine A2A
receptors is more restricted, comprising lymphocytes, platelets,
specific brain areas (striatum, nucleus accumbens, olfactory
tubercle), vascular smooth muscle and endothelium [8].
Selective A1 AR antagonists have demonstrated promising
therapeutic potential for the treatment of cognitive diseases, renal
failure, Alzheimer’s disease and cardiac failure [9]. Adenosine A2A
receptor antagonists may be useful for the treatment of acute and
chronic neurodegenerative disorders such as cerebral ischemia,
Parkinson’s, Huntington’s, and Alzheimer’s disease, as drugs
controlling motor functions and exhibiting neuroprotective prop-
erties [10e18].
* Corresponding author. Tel: þ48 12 6205581; fax: þ48 12 6205596.
ꢀ
E-mail address: mfkonono@cyf-kr.edu.pl (K. Kiec-Kononowicz).
0223-5234/$ e see front matter Ó 2011 Elsevier Masson SAS. All rights reserved.
doi:10.1016/j.ejmech.2011.05.023

ꢀ
A. Drabczynska et al. / European Journal of Medicinal Chemistry 46 (2011) 3590e3607
3591
Since ARs offer an attractive target for drug development
a number of studies have been carried out in order to find subtype-
selective AR antagonists 2[19] especially in the group of xanthine
derivatives [20]. Among A1 antagonists with a xanthine structure
the most potent ligands were in the group of 1,3-dipropyl-
substituted derivatives bearing bulky residues in position 8 of the
xanthine core (e.g. compound 1 [21] and KW-3902 [2]), whereas
active A2A AR antagonists were found among 8-styrylxanthines
(e.g. MSX-2 [22,23], KW-6002 [22e26]) (Fig. 1). Furthermore,
non-xanthine AR antagonists have been developed [19]. A major
problem with adenosine receptor antagonists has been their
generally low water-solubility [9,18,20].
Fig. 2. General structure of the tricyclic xanthine ligands with N-cycloalkyl
substituents.
Our groups have made considerable efforts to develop new
selective xanthine adenosine receptor ligands. Our main focus was
the investigation of tricyclic xanthine derivatives [27e32]. The
most active A1 AR ligands were found among 1,3-dipropyl-
substituted benzylpyrimidopurinediones [31], while A2A adenosine
receptor ligands were mostly 1,3-dimethyl-substituted aryl- [30]
and phenalkylpyrimidopurinediones [32].
As a continuation of our studies on tricyclic annelated purine-
diones the systematic examination of imidazo-, pyrimido- and
diazepinopurinediones with cycloalkyl substituents was per-
formed. Four elements in the target compounds were varied
(Fig. 2): the size of the annelated ring (imidazo, pyrimido, dia-
zepino), the size of the cycloalkyl moiety (from 3- to 8-membered),
the spacer (from 0 to 2 C atoms), and the substituents (CH3, OH,
OCOCH3) in the cycloalkyl ring.
the N-8, N-9 or N-10 position, respectively, of the annelated ring
was accomplished as shown in Fig. 3.
The following starting materials were used, which were
obtained according to described procedures [30,33,34]: 7-(2-
bromoethyl)-8-bromotheophylline,
7-(3-chloropropyl)-8-
bromotheophylline and 7-(4-bromobutyl)-8-bromotheophylline.
These were cyclized with amines under various reaction conditions
with regard to the amount of amine, solvent and reaction time.
The data are summarized in Table 1. Compounds 3 and 11 were
described previously [35], but no biological activity has been
reported. The structures of the synthesized compounds were
confirmed by UV, IR and 1H NMR spectra: UV spectra showed
a bathochromic shift typical for 8-aminoxanthine derivatives with
lmax of about 300 nm [36]. The IR absorption bands were typical of
xanthine derivatives [37] and in the 1H NMR spectra, the expected
chemical shifts were observed. All compounds were purified by
recrystallization.
Syntheses, structure elucidation by means of X-ray analysis,
physicochemical properties, and in vitro evaluation of the potency
at ARs, and in vivo examination of anticonvulsant properties were
performed.
3. X-ray structure analysis
2. Chemistry
It is well known that hydrogen bonds are very important among
all intermolecular interactions [38e42]. They are responsible for
molecular recognition and/or self-organization of the molecules
[43]. The H-bonds take part in the formation of complexes between
The synthesis of 1,3-dimethylperhydroimidazo-, -pyrimido- and
-1,3-diazepino[2,1-f]purinediones with cycloalkyl substituents in
Fig. 1. Potent and selective xanthine adenosine A1 and A2A receptor antagonists. Ki values are given in nM. aRadioligand [3H]CPX, rat brain cortex membranes; bradioligand [125I]ZM
241385, rat brain striatum membranes; cradioligand [3H]MRS 1754, HEK 293 cells expressing human A2B AR; dradioligand [125I]AB-MECA, HEK 293 cells expressing human A3 AR;
eradioligand [3H]CGS21680, rat brain striatal membranes; fradioligand [3H]CHA, rat brain cortical membranes; gradioligand [3H]CCPA, rat brain cortical membranes; hradioligand
[3H]MSX-2, rat brain striatal membranes.

ꢀ
A. Drabczynska et al. / European Journal of Medicinal Chemistry 46 (2011) 3590e3607
3592
biological receptors and their ligands and they are responsible for
O
O
H
N
H
N
packing motif formation in nearly all crystals of organic substances.
The most important information about H-bond geometry and
topology can be taken from crystallographic data. Therefore, X-ray
structure analysis of small, biologically active molecules is highly
useful. Previous X-ray studies on imidazo-, pyrimido- and 1,3-
diazepino[2,1-f]purinediones with various aromatic substituents in
the N-8, N-9 or N-10 position respectively, have suggested that
CeH/O]C interactions are crucial for crystal structure architecture
[27e32]. Such H-bonds were mainly formed between the O2 purine
oxygen atom and protons from the annelated rings. It should be
noted that aromatic groups at the nitrogen atom participated only
incidentally in structural motif formation, usually via their substit-
uents (for example OCH3) [31]. Recently, among derivatives with
various cycloalkyl substituents at N-8, N-9 or N-10 (Fig. 2), mono-
crystals of three compounds e 7,13 and 33 e were selected for X-ray
structure analysis (Fig. 4). These comprise the smallest 3-membered
(7) and the largest 8-membered cykloalkyl derivative (33).
H3C
H3C
N
N
i
ii
Br
N
N
O
N
O
N
I
II
CH3
CH3
O
O
H ) X
n
2
(C
H )
2
(C
N
n
H3C
H3C
N
N
N
N
N
N
iii
Br
R
O
N
O
N
IV
III
CH3
CH3
n = 1 compds 1-6
n = 2 compds 7-24
n = 3 compds 25-33
R = cycloalkyl;
alkylcycloalkyl
n = 2, 3, 4
X = Cl, Br
,
i: 40% HBr, NaClO AcOH; ii: X(CH
3
)
CO
X, TEBA, K , acetone; iii: R-NH
2
3
2
2 n
In the studied structures the bicyclic xanthine moiety is planar.
Annelated pyrimidone rings in 7 and 13 are half-chairs, while the
diazepine ring in 33 adopts a boat conformation. Superimposition
Fig. 3. Synthesis of imidazo-, pyrimido- and 1,3-diazepino[2,1-f]purinediones.
Table 1
Physical data and reaction conditions of N-cycloalkylimidazo-, pyrimido-, and diazepino[2,1-f]purinediones.
O
H3C
N
N
(CH2)n
N
O
N
N
R
CH3
Compound
R
n
Formula, MW
M.p. (ꢀC)
Yield of
cyclization
(%)
Reaction
medium
(excess of
amine)
Reaction
time
[h] reflux
Crystal.
solvent
TLC Rf
eluent
1
1
C14H19N5O2, 289.33
167e169
46
63
Xylene (5)
10
10
Ethanol
Ethanol
0.30a
2
3
1
1
C16H23N5O2, 317.37
201e203
Butanol (5)
0.55a
[35]
4
5
1
1
C17H23N5O2, 329.39
162e164
215e218
53
40
Xylene (4)
6
Methanol
0.49a
C15H21N5O3$H2O,
e (20)
10
50% Ethanol
0.35b
337.37
OH
cis trans

ꢀ
A. Drabczynska et al. / European Journal of Medicinal Chemistry 46 (2011) 3590e3607
3593
Table 1 (continued )
Compound
R
n
Formula, MW
M.p. (ꢀC)
Yield of
cyclization
(%)
Reaction
medium
(excess of
amine)
Reaction
time
[h] reflux
Crystal.
solvent
TLC Rf
eluent
6
7
1
2
C17H25N5O2, 331.41
153e155
220e221
68
69
Xylene (5)
Butanol (5)
10
10
Ethanol
0.42a
0.50a
C13H17N5O2, 275.31
50% Ethanol
8
2
2
2
2
C14H19N5O2, 289.33
C14H19N5O2, 289.33
C15H21N5O2, 303.36
[35]
183e185
251e253
234e236
59
48
82
Me-Digol (2)
Me-Digol (2)
Me-Digol (2)
5
10
5
70% Ethanol
Ethanol
0.57a
0.51a
9
10
11
Ethanol
12
2
C16H23N5O3, 333.38
231e233
83
e (20)
5
20% Ethanol
0.11a
OH
cis trans
13
2
C16H23N5O3, 333.38
262e264
62
Me-Digol (2)
5
Methoxyethanol
0.11a
OH
trans
3 trans
Acetic
anhydride (10)
14
2
C18H25N5O4, 357.42
305e306
79
5
Methoxyethanol
0.53a
O
O
CH
15
2
C17H25N5O2, 331.41
182e184
63
Butanol (3)
10
50% Ethanol
(continued on next page)
0.55a
CH
3 cis trans

ꢀ
A. Drabczynska et al. / European Journal of Medicinal Chemistry 46 (2011) 3590e3607
3594
Table 1 (continued )
Compound
R
n
Formula, MW
M.p. (ꢀC)
Yield of
cyclization
(%)
Reaction
medium
(excess of
amine)
Reaction
time
[h] reflux
Crystal.
solvent
TLC Rf
eluent
16
2
C17H25N5O2, 331.41
172e174
95
94
Butanol (5)
10
Ethanol/H2O
70% Ethanol
0.55a
cis trans
H3C
17
18
2
2
C17H25N5O2, 331.41
178e180
148e149
Me-Digol (2)
5
0.57a
C18H27N5O2$½H2O,
c
ethanol
0.60a
354.44
R, S
C
18H27N5O2$½H2O,
19
2
149e151
69
DMF (2)
5
50% Ethanol
0.60a
354.44
R
C
18H27N5O2$½H2O,
20
21
2
2
149e150
162e164
65
62
DMF (2)
DMF (2)
5
8
50% Ethanol
Ethanol/H2O
0.60a
0.56a
354.41
S
C18H25N5O2, 343.42
22
2
C17H25N5O2, 331.41
200e202
51
Me-Digol (2)
10
Ethanol
0.57a
23
24
2
2
C18H27N5O2, 345.44
186e188
293e295
84
62
Me-Digol (2)
Me-Digol (2)
5
8
70% Ethanol
0.52a
0.60a
C20H25N5O2, 367.44
Methoxyethanol
25
3
C16H23N5O2, 317.38
182e184
79
DMF (5)
10
Ethanol
0.68a

ꢀ
A. Drabczynska et al. / European Journal of Medicinal Chemistry 46 (2011) 3590e3607
3595
Table 1 (continued )
Compound
R
n
Formula, MW
M.p. (ꢀC)
Yield of
cyclization
(%)
Reaction
medium
(excess of
amine)
Reaction
time
[h] reflux
Crystal.
solvent
TLC Rf
eluent
26
27
3
C17H25N5O2, 331.47
228e230
55
62
DMF (5)
10
Ethanol
0.59a
3
C17H25N5O, 347.41
233e235
e(20)
5
30% Ethanol
0.77d
OH
cis trans
Acetyl
anhydride (10)
0.69a
O
28
3
C19H27N5O4, 389.45
137e140
82
5
50% Ethanol
O
CH
cis trans
29
3
C18H27N5O2, 345.44
212e214
74
Butanol (5)
10
Ethanol/H2O
0.66a
CH
3 cis trans
30
3
3
3
C13H27N5O2, 345.44
C18H27N5O2, 345.44
C19H27N5O2, 357.45
214e217
133e134
113e115
89
80
91
Butanol (5)
10
10
8
Ethanol
0.67a
0.65a
0.62a
H3C
cis trans
31
32
DMF (4)
Methanol/H2O
DMF (4)
Methanol
33
3
C19H29N5O2, 359.46
164e165
56
DMF (5)
10
Ethanol/H2O
0.46e
Me-Digol e diethylene glycol monomethyl ether.
a
Benzene/acetone e 7:3.
Butanol/butyl acetate/CHCl3 e 5:1:1.
Stoichiometric mixture of 19 and 20.
Benzene/acetone/methanol e 1:1:1.
b
c
d
e
Hexane/dioxane e 5:2.

ꢀ
3596
A. Drabczynska et al. / European Journal of Medicinal Chemistry 46 (2011) 3590e3607
Fig. 4. Superimposition of 7, 13 and 33 (molecule A) with respect to the bicyclic xanthine moieties.
of all three structures (Fig. 4) shows that the cycloalkyl residues are
oriented in a similar mode e trans with respect to the N9e(or
from cycloalkyl) join molecules into a ribbon (Fig. 6 and Tables in
Supplementary data). The C7eH/O4 bond is forming a connection
resulting in two ribbons which form the main structural motif in the
structure of 7 and 13. The cyclohexyl-OH substituent in 13 has been
the origin of strong H-bonds which join the molecules along the c-
N10e)C12 bond. In the third structure (13)
e cyclohexyl-
substituted e the hydroxyl group is positioned trans with respect
to the N9eC12 bond. In that prominent localization, OH seems to be
an evident H-bond partner.
ꢁ
axis [O17eH/O17(x,1/2 ꢁ y,1/2 þ z) ¼ 2.856 A]. As a consequence
As it often happens for conformationally flexible groups, pyr-
imidone in 13, and diazepine and cyclooctyl in 33 show a confor-
mational disorder in the structure. This is evident in the structure of
33 with three molecules in a crystallographically independent unit,
designated A (atom numbering without indices), B (with indices 0)
and C (with indices 00 and # for disordered atoms). Two of them (A
and B) differ insignificantly in the cyclooctyl conformation (Fig. 5).
The third one, with disordered diazepine and cyclooctyl (both with
s.o.f. ¼ 0.5), consists in fact of two identical molecules which are
mirror images (Fig. 5).
between the double ribbons there is enough space for a methanol
molecule which is located in the canal down the c-axis. Solvent
molecules form a H-bonding chain, similar to the one observed for
the drug molecules. Even if CeH/O]C hydrogen bonds are also
identified in 33, the main structural motif is based on stacked
bicycles of three independent molecules (Fig. 6). These molecules
form a sandwich of A and B with C in between. In the sequence
ꢁ
ꢁ
A/C/B distances between bicycles equal 3.56 A and 5.58 A,
respectively. The main motifs form a column down the c-axis with
ꢁ
a distance between sandwiches of 3.60 A. Nevertheless, H-bonds
The structure of the crystals for three studied compounds is
based on CeH/O]C hydrogen bonds. In 7 and 13, two CeH/O2
interactions (where one carbon is from pyrimidone and the other
of CeH/O (Table in Supplementary data) are responsible for
the molecules position. All six oxygen atoms are proton acceptors,
while protons mainly come from the azepine ring carbons. Columns
Fig. 5. Superimposition of diazepine and cyclooctyl residues from the structure of 33.

ꢀ
A. Drabczynska et al. / European Journal of Medicinal Chemistry 46 (2011) 3590e3607
3597
Fig. 6. Main structural motifs in the structures of 7, 13 and 33.
are getting together through the weak H-bonds of the cyclooctyl
performed in mice. The MES assays have predictive value for agents
as potential therapeutics in the management of grand mal epilepsy,
whereas the ScMet test is for those likely to be effective against
petit mal [52,53]. Minimal motor impairment was measured by the
rotorod toxicity test. The results are given in Table 4.
Some compounds were administered orally to rats and exam-
ined in the MES, ScMet screen and rotorod tests. For two
compounds (17, 31) a quantitative test was performed and ED50
values were determined. Results of this test were compared with
literature data available for valproate [54] as reported in Table 5.
00
00
ꢁ
carbon hydrogen atoms C13 eH/O2 (ꢁ0.5 þ x,1.5 ꢁ y,z) ¼ 3.625 A.
X-ray structure analysis of pyrimido- and 1,3-diazepino[2,1-f]
purinediones with various cycloalkyl substituents in the N-9 or N-
10 position respectively, confirmed our earlier observation that
both oxygen atoms, O2 and O4, can be competent H-bond accep-
tors. Moreover in the studied compounds carbon atoms from
cycloalkyl substituents may serve as additional proton-donating
centers.
4. Pharmacology
4.1. In vitro tests
All compounds were tested in vitro in radioligand binding assays
for affinity to A1 and A2A ARs at rat cortical membrane and rat striatal
membrane preparations, respectively. Selected compounds were
further tested for their affinity to human A1, A2A, A2B and A3 receptors
recombinantly expressed in CHO cells. As A1 AR radioligand [3H]2-
chloro-N6-cyclopentyladenosine ([3H]CCPA) [44] was used and as
A2A radioligand [3H]1-propargyl-3-(3-hydroxypropyl)-7-methyl-8-
(m-methoxystyryl)xanthine ([3H]MSX-2) was applied [45]; [3H]4-(2-
[7-amino-2-(2-furyl-[1,2,4]-triazolo[2,3-a]-[1,3,5]-triazin-5ylamino]
ethyl)phenol ([3H]ZM241385) [46] and [3H]8-(4-(4-(4-chlorophenyl)
piperazine-1-sulfonyl)phenyl)-1-propylxanthine ([3H]PSB-603) [47]
were used as radioligands in A2B binding studies and [3H]2-phenyl-
8-ethyl-4-methyl-(8R)-4,5,7,8-tetrahydro-1H-imidazo[2,1-i]purine-
5-one ([3H]PSB-11) was used as A3 adenosine selective receptor
ligand [48]. The results are presented in Tables 2 and 3. The potent
A2A antagonist KW-6002 and the non-selective ligand caffeine were
included for comparison. Sodium chloride shift experiments [49,50]
were performed for one of the most potent and A2A-selective
compounds of the present series in order to confirm that the
N-cycloalkyl-substituted tricyclic purinediones are antagonists at
the A2A ARs (Fig. 7). Furthermore, some compounds were investi-
gated in functional cAMP accumulation studies. Thus, the most
potent compounds were investigated for their potency to inhibit
cAMP accumulation induced by the agonist NECA in CHO cells
expressing the human A2A receptor.
The results of the radioligand binding assays at adenosine A1
and A2A receptors (Table 2) showed that N-cycloalkylpurinediones
(with the exception of: 4, 26, 31, 32, which were non-selective)
exhibited affinity to A2A receptors but poor A1 affinity. It was
noticed that for both imidazo- and diazepinopurinediones
substituted with a cyclohex-1-enylethyl moiety (4, 32) selectivity
was lost, while for pyrimidopurinedione with the same substituent
(21) high selectivity (>40) was maintained. Adamantyl derivative
24 was inactive at both ARs subtypes. A decisive influence on A2A
affinity had the size of the annelated ring. As observed previously
[30] also in this series of compounds, the pyrimidine ring was
beneficial for A2A affinity. Decrease to 5-membered (compounds
1e6) or enlargement to 7-membered rings (compounds 25e33) led
to a reduction in adenosine receptor affinity (only two compounds,
6 and 29, showed submicromolar affinity).
The size of the cycloalkyl moiety had a smaller influence on A2A
affinity, however bigger rings were favourable (larger than 5 C, but
no cycloheptyl). Introducing a 1e2 carbon atom linker between the
cycloalkyl moiety and the annelated ring increased affinity
(compare the affinity of compounds 2 and 3, 7 and 8, 11 and 17 or
21). The influence of substituents, OH or CH3, in the cyclohexyl ring
was interesting: a methyl group was beneficial only in position 4
producing the best ligand of the present series, compound 15
(Ki ¼ 0.24
mM). The configuration of the OH substituent was crucial
The compounds were additionally evaluated in vivo as anti-
convulsants by the Antiepileptic Drug Development Program (ADD)
of the National Institute of Neurological Disorders and Stroke
(NINDS) in Bethesda according to the Antiepileptic Screening
Project (ASP) [52,53]. Phase I of the evaluation included three tests,
maximal electroshock (MES), subcutaneous pentylentetrazol
(ScMet) and the rotorod test for neurological toxicity (TOX)
for the affinity to the adenosine A2A receptor. The mixture of cis and
trans isomers of 4-hydroxycyclohexyl derivative 12 was more
potent than the trans-isomer 13 alone, which indicates the pref-
erence of the A2A receptor for the cis-configuration. The S-config-
uration of the optically active compounds 19 (R) and 20 (S) was
favourable increasing affinity to the A2A receptor to the sub-
micromolar range (compound 20).

ꢀ
A. Drabczynska et al. / European Journal of Medicinal Chemistry 46 (2011) 3590e3607
3598
Table 2
Affinities of cycloalkylimidazo-, pyrimido-, and diazepino[2,1-f]purinediones at adenosine A1 and A2A receptors.
Compound
Adenosine A1 receptor
Adenosine A2A receptor
A2A AR selectivity
A1/A2A
(rat brain cortical
(rat brain striatal membranes) vs.
membranes) vs. [3H]CCPA Ki ꢂ SEM (
m
M,
M)
[3H]MSX-2 Ki ꢂ SEM (
mM)
(% inhibition ꢂ SEM at 25
respectively) (n ¼ 3)
m
M or 10
m
(% inhibition ꢂ SEM at 25 M) (n ¼ 3)
m
KW-6002 [29] (Fig. 1)
Caffeine [29]
0.230 ꢂ 0.030
0.00515 ꢂ 0.00025
32.8 ꢂ 8.0
45
18.8 ꢂ 5.6
0.6
O
H3C
O
N
N
N
N
R
N
CH3
Imidazopurinediones
1
2
3
4
5
6
Cyclopentyl
Cyclohexylmethyl
Cyclohexyl
Cyclohex-1-enylethyl
4-Hydroxycyclohexyl
Cyclooctyl
ca. 25 (50 ꢂ 3%)
ꢃ25 (41 ꢂ 2%)
3.10 ꢂ 1.18
2.00 ꢂ 0.70
2.58 ꢂ 0.68
2.02 ꢂ 0.05
4.48 ꢂ 0.41
0.33 ꢂ 0.09
w8
ꢃ13
>4
1
>6
ꢃ76
>10 (27 ꢂ 4% at 10
2.45 ꢂ 0.18
mM)
>25 (4 ꢂ 3%)
ꢃ25 (40 ꢂ 2%)
O
H3C
O
N
N
N
N
R
N
CH3
Pyrimidopurinediones
7
8
9
Cyclopropyl
Cyclopropylmethyl
Cyclobutyl
Cyclopentyl
Cyclohexyl
4-Hydroxycyclohexyl
trans-4-Hydroxycyclohexyl
4-Acetocyclohexyl
4-Methylcyclohexyl
2-Methylcyclohexyl
Cyclohexylmethyl
Cyclohexylethyl
(R)-Cyclohexylethyl
(S)-Cyclohexylethyl
Cyclohex-1-enylethyl
Cycloheptyl
ꢃ25 (43 ꢂ 6%)
2.39 ꢂ 0.58
1.24 ꢂ 0.33
0.98 ꢂ 0.24
1.93 ꢂ 0.37
1.00 ꢂ 0.45
0.81 ꢂ 0.01
2.04 ꢂ 0.18
3.63 ꢂ 0.25
6.20 ꢂ 0.62
0.24 ꢂ 0.15
1.08 ꢂ 0.26
0.61 ꢂ 0.24
1.24 ꢂ 0.04
1.30 ꢂ 0.10
0.887 ꢂ 0.053
0.50 ꢂ 0.18
2.67 ꢂ 0.71
0.57 ꢂ 0.15
>10 (13 ꢂ 11)
ca. 20
2
ꢃ10 (39 ꢂ 3%)
5.31 ꢂ 1.29
ꢃ5
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
5
ꢃ25 (39 ꢂ 3%)
>25 (8 ꢂ 3%)
>10 (11 ꢂ 5%)
>10 (5 ꢂ 5%)
ꢃ25 (38 ꢂ 1%)
>25 (31 ꢂ 3%)
2.49 ꢂ 0.35
ꢃ31
>12
>3
>2
ꢃ100
>23
4
6
6
13
44
7.58 ꢂ 0.54
7.89 ꢂ 0.14
11.9 ꢂ 1.38
14.7 ꢂ 1.76
>10 (4 ꢂ 2%)
ꢃ25 (33 ꢂ 1%)
>10 (7 ꢂ 9%)
4
Cyclooctyl
1-Adamantyl
ꢃ44
1
O
N
H3C
O
N
N
N
N
R
CH3
Diazepinopurinediones
25
26
27
28
29
30
31
32
33
Cyclopentyl
Cyclohexyl
5.06 ꢂ 0.64
1.38 ꢂ 0.26
2.96 ꢂ 1.20
3.81 ꢂ 0.09
1.74 ꢂ 0.85
0.83 ꢂ 0.20
3.46 ꢂ 1.81
3.5 ꢂ 0.75
4
1
w7
3
4
>7
1
1
2.13 ꢂ 0.44
4-Hydroxycyclohexyl
4-Acetocyclohexyl
4-Methylcyclohexyl
2-Methylcyclohexyl
Cyclohexylmethyl
Cyclohex-1-enylethyl
Cyclooctyl
ꢃ25 (44 ꢂ 3%)
4.87 ꢂ 0.84
3.44 ꢂ 0.98
>25 (27 ꢂ 13%)
5.18 ꢂ 1.15
1.66 ꢂ 0.17
1.17 ꢂ 0.13
6.86 ꢂ 0.94
>10 (16 ꢂ 9%)
>1

ꢀ
A. Drabczynska et al. / European Journal of Medicinal Chemistry 46 (2011) 3590e3607
3599
Table 3
Affinities of selected compounds and standard antagonists at human adenosine A1, A2A, A2B and A3 receptors recombinantly expressed in CHO cells.
Compound
Adenosine A1 receptor
Adenosine A2A
Adenosine A2B receptor
(human recombinant)
vs. [3H]ZM241385* vs.
[3H]PSB-603** Ki ꢂ SEM
Adenosine A3
(human recombinant) vs.
receptor (human
recombinant) vs.
[3H]MSX-2 Ki ꢂ SEM
receptor (human
recombinant) vs.
[3H]PBS-11 Ki ꢂ SEM
[3H]CCPA Ki ꢂ SEM [
(% inhibition ꢂ SEM
mM]
at 10
m
M) (n ¼ 3)
[
m
M] (% inhibition ꢂ SEM
[
m
M] (% inhibition ꢂ SEM
[
m
M] (% inhibition ꢂ SEM
at 10
23.4 ꢂ 7.1#
m
M) (n ¼ 3)
at 10
m
M) (n ¼ 3)
at 10
m
M) (n ¼ 3)
Caffeine
44.9 ꢂ 6.2#
20.5 ꢂ 2.2#
>100 [51]
KW-6002
2.07 ꢂ 0.43#
0.0908 ꢂ 0.0228#
ꢃ10 (47 ꢂ 2%)a [49]
4.47 ꢂ 4.06#
Imidiazopurinediones
3
5
6
Cyclohexyl
4-Hydroxy-cyclohexyl
Cyclooctyl
>10 (41 ꢂ 4%)
>10 (7 ꢂ 3%)
>10 (21 ꢂ 9%)
1.61 ꢂ 0.34
3.27 ꢂ 1.43
0.306 ꢂ 0.024
>10 (11 ꢂ 7%)**
>10 (24 ꢂ 24%)*
0.80 ꢂ 0.16**
>10 (12 ꢂ 6%)
>10 (0 ꢂ 0%)
>10 (16 ꢂ 2%)
Pyrimidopurinediones
8
9
Cyclopropyl-methyl
Cyclobutyl
Cyclohexyl
trans-4-Hydroxy-cyclohexyl
4-Aceto-cyclohexyl
4-Methyl-cyclohexyl
Cyclohex-1-enylethyl
Cycloheptyl
7.13 ꢂ 1.04
0.417 ꢂ 0.199
0.847 ꢂ 0.650
1.61 ꢂ 0.41
19.0 ꢂ 1.5
12.5 ꢂ 1.6*
>10 (1 ꢂ 1%)
>10 (5 ꢂ 4%)
>10 (12 ꢂ 5%)
>10 (27 ꢂ 5%)
>10 (27 ꢂ 9%)
>10 (16 ꢂ 6%)
>10 (23 ꢂ 3%)
>10 (6 ꢂ 1%)
>10 (52 ꢂ 5%)
1.29 ꢂ 0.30
10.3 ꢂ 2.76
>10 (21 ꢂ 4%)**
>10 (0 ꢂ 5%)**
>10 (0 ꢂ 4%)**
>10 (0 ꢂ 10%)**
>10 (16 ꢂ 16%)*
>10 (6 ꢂ 4%)**
>10 (0 ꢂ 18%)**
>10 (32 ꢂ 0%)*
>10 (0 ꢂ 3%)**
11
13
14
15
21
22
23
24
>10 (37 ꢂ 8%)
>10 (21 ꢂ 2%)
>10 (5 ꢂ 3%)
>10 (28 ꢂ 4%)
6.90 ꢂ 1.34
21.9 ꢂ 7.0
0.61 ꢂ 0.21
0.933 ꢂ 0.141
1.16 ꢂ 0.31
1.53 ꢂ 0.59
>10 (0 ꢂ 23%)
>10 (41 ꢂ 5%)
>10 (17 ꢂ 15%)
>10 (15 ꢂ 2%)
Cyclooctyl
1-Adamantyl
Diazepinopurinediones
28
33
4-Aceto-cyclohexyl
Cyclooctyl
>10 (16 ꢂ 7%)
>10 (29 ꢂ 3%)
>10 (8 ꢂ 1%)
2.40 ꢂ 0.49
>10 (29 ꢂ 1%)*
>10 (0 ꢂ 6%)**
>10 (28 ꢂ 6%)
1.67 ꢂ 0.22
a
[3H]PSB-298 was used as a radioligand [49].
*
**
#
[3H]ZM241385 [46].
[3H]PSB-603 [47].
Ref. [18].
Several compounds were more potent at the rat as compared to
a sodium chloride shift assay. While the curve for the agonist NECA
was significantly shifted to the right in the presence of 100 mM of
sodium chloride, the curve for 15 was unaltered in the same
experiment. This clearly indicates that 15 acts as an antagonist at
A2A receptors (Fig. 7). To confirm these results, the three most
potent compounds 6, 8, 15 were evaluated in functional experi-
ments. They were investigated for their potency to inhibit NECA-
induced cAMP accumulation in CHO cells expressing the human
A2A receptor (Fig. 8). The compounds clearly behaved as competi-
tive antagonists as the concentrationeresponse curve of NECA was
shifted to the right in a parallel fashion in their presence. Kb values
determined in living CHO cells expressing the human adenosine
A2A receptor were well in accordance with Ki values determined in
radioligand binding studies at membrane preparations of the same
cell line. Owing to the structural similarity of all compounds in this
series we suppose that they are all antagonists.
the human A2A receptor subtype: 13 and 28 (almost five-fold), 14,
15 and 23 (about three-fold), 11 and 21 (two-fold) (Tables 2 and 3).
While compound 6 was similarly potent in both species, some of
the compounds showed higher affinity for the human as compared
to the rat A2A AR, including 3, 5, 8, 9, 22, and 33 (up to three-fold).
Affinity for A1 AR, when it was observed, was as well worse for
human AR (for 8, 28) as better for 21 (two times). Affinity to the
human A2B and A3 receptor of selected compounds (3, 5, 6, 8, 9, 11,
13, 14, 15, 21, 22, 23, 24, 28, and 33) was very weak only cyclooctyl
derivative 6 showed submicromolar affinity and cyclopropylmethyl
derivative 8 exhibited micromolar affinity to the human A2B AR.
Compounds 24 and 33 were the only derivatives which displayed
good, micromolar affinity for the human A3 AR (Ki 1.29 and 1.67 mM,
respectively) (Table 3). One of the most potent A2A-selective ligands
(compound 15) was investigated for its functional properties using
Fig. 7. Radioligand binding curves of the agonist N-ethylcarboxamidoadenosine (NECA) and of compound 15 at rat adenosine A2A receptors in the absence and in the presence of
NaCl (100 mM); IC50 NECA (ꢁNaCl): 7.05 ꢂ 0.70 nM, IC50 NECA (þNaCl): 366 ꢂ 197 nM (49-fold shift); IC50 15 (ꢁNaCl): 263 ꢂ 152 nM, IC50 15 (þNaCl): 251 ꢂ19 nM.

ꢀ
A. Drabczynska et al. / European Journal of Medicinal Chemistry 46 (2011) 3590e3607
3600
Table 4
Anticonvulsant activity and neurotoxicity of cycloalkylimidazo-, pyrimido-, and 1,3-diazepino[2,1-f]purinediones (i.p., mice).
Compounda
MESb,c
0.25 h
ScMetb,c
0.25 h
Toxicityb,c
ASPd class
0.5 h
1 h
4 h
0.5 h
1 h
4 h
0.25 h
0.5 h
1 h
4 h
1
2
3
4
5
6
7
8
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
300 (4/5)
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
300 (2/4)
e
e
e
2
3
2
3
3
2
3
1
3
2
2
2
2
3
2
3
1
3
3
1
1
3
3
3
2
3
2
2
1
2
1
3
2
e
e
e
e
e
e
e
e
e
300 (1/1)
e
e
100 (1/8)
300 (1/4)
e
e
e
e
e
e
e
300 (1/2)
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
300 (2/5)e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
300 (4/4)
300 (4/4)
e
100 (2/3)
300 (1/1)
e
e
100 (3/5)
e
300f
300 (4/5)
300 (4/5)g
300 (1/5)e
e
e
e
9
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
e
e
300 (3/5)
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
g
e
e
e
e
300 (3/5)
e
300 (1/4)
e
e
e
e
e
e
e
100 (3/5)
e
300 (5/5)
e
100 (1/5)
e
300 (1/4)
300 (1/4)
300 (1/4)
300 (1/4)
300 (1/4)
e
e
e
e
e
e
e
e
e
100 (1/3)
e
e
100 (1/5)
100 (1/5)
e
e
e
e
e
e
e
e
e
e
e
e
300 (5/5)
e
e
100 (1/8)
300 (1/4)
e
e
e
e
e
e
e
e
300 (1/5)
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
300 (1/1)
300 (5/5)
300 (5/5)
300 (2/5)
100 (4/5)
e
e
100 (1/8)
100 (2/8)
e
e
e
e
e
e
e
100 (3/5)
100 (3/5)
e
e
e
e
e
e
e
e
e
e
e
e
300 (1/1)
e
300 (1/5)e
e
e
100 (1/8)
100 (1/8)
100 (1/4)
e
e
The dash (e) indicates an absence of activity/toxicity at maximum dose administration (300 mg/kg).
a
Administered as suspension in 0.5% methylcellulose.
b
Doses of 30, 100, 300 mg/kg. The figures in the table indicate the minimum dose whereby activity was demonstrated. The animals were examined 0.5 and 4 h after
injections were made. For compounds 8, 17, 29 the biological response was observed after 0.25 h.
c
Meaning of figures in the anticonvulsant test: e.g. 1/5 means the number of animals protected/number of animals tested; in toxicity tests: number of animals that
exhibited toxicity/number of animals tested.
d
Classification is as follows: 1 e anticonvulsant activity at 100 mg/kg or less; 2 e anticonvulsant activity at 300 mg/kg; 3 e lack of anticonvulsant activity at 300 mg/kg.
Myoclonic jerks.
Death following clonic seizure.
Death following tonic extension.
e
f
g
4.2. In vivo tests
size of the annelated ring and the character and position of the
cycloalkyl ring and its substituents seemed to have a significant
influence on the strength of the anticonvulsant activity. The rank
order of potency according to the size of the annelated ring was as
follows: diazepine > pyrimidine > imidazole. Among investigated
diazepinopurinediones most compounds (78%) showed protective
Unsubstituted imidazo-, pyrimido-, and diazepinopurinediones
did not show protective activity in both electric and chemical
seizures [30]. Introduction of cycloalkyl substituents at the nitrogen
atom of the annelated ring resulted in anticonvulsant activity. The
Table 5
Anticonvulsant activity and neurotoxicity of selected compounds after oral administration (30 mg/kg or 50 mg/kg) to rats.
Compounda
MESb
0.5 h
ScMetb
0.25 h
Toxicityb
0.25 h
ED50 [mg/kg]
1 h
2 h
4 h
0.5 h
1 h
2 h
4 h
0.5 h
1 h
4 h
8
e
e
e
e
e
30 (1/4)
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
e
17
20
29
31
e
e
50 (3/4)
50 (2/4)
e
50 (1/4)c
50 (2/4)
50 (3/4)
50 (1/4)
e
50 (2/4)
>200 ScMet 0.25 h
30 (2/4)
e
30 (1/4)
e
30 (1/4)
e
e
e
e
50 (1/4)c
50 (1/4)c
e
50 (1/4)c
e
e
e
e
e
50 (1/4)
e
>250 (MES)
>100 (ScMet)c
287 (MES)
Valproate
209 (ScMet)
The dash (e) indicates an absence of activity/toxicity at the given dose administration.
a
Given as a suspension in 0.5% methylcellulose.
Meaning of figures in anticonvulsant test: e.g. 1/4 means the number of animals protected/number of animals tested.
Death following continuous seizures.
b
c

ꢀ
A. Drabczynska et al. / European Journal of Medicinal Chemistry 46 (2011) 3590e3607
3601
A
100
NECA
80
NECA + 6 (3 µM)
60
40
20
0
10-12
10-10
10-8
10-6
10-4
10-2
[NECA], M
B
120
NECA
NECA + 8 (3 µM)
100
80
60
40
20
0
10-11 10-10 10-9 10-8 10-7 10-6 10-5 10-4 10-3 10-2
[NECA], M
C
NECA
NECA + 15 (5 µM)
120
100
80
60
40
20
0
10-11 10-10 10-9 10-8 10-7 10-6 10-5 10-4 10-3 10-2
[NECA], M
Fig. 8. cAMP accumulation studies in CHO cells expressing the human adenosine A2A receptor by different concentrations of the agonist NECA in the absence and presence of test
compound 6 (A), 8 (B) or 15 (C). All three investigated compounds shifted the concentrationeresponse curves for NECA to the right. Apparent Kb values could be calculated as
follows: 210 ꢂ 51 nM (6), 220 ꢂ 39 nM (8), and 644 ꢂ156 nM (15).
activity in the chemical test (compound 31 in both tests at short
time, 0.5 h). Introducing a substituent at the cyclohexyl ring
resulted in anticonvulsant activity while the unsubstituted cyclo-
hexyl compound 26 was inactive. Among active substances three
compounds 27, 28, 33 showed neurotoxicity at a dose of 100 mg/kg,
which was lower than the effective dose (300 mg/kg). All the other
compounds were nontoxic. The two most active derivatives 29 and
31 were also administered orally to rats (Table 5). Compound 31
showed protection in ScMet test at 0.5 h and 2 h time at a dose of
50 mg/kg with no symptoms of neurotoxicity. The ED50 determined
for compound 31 was >100 mg/kg in the chemical test and
>250 mg/kg in the electric test (Table 5). Compound 29 showed
protection in the ScMet test at 0.5e4 h time at a dose of 50 mg/kg
but caused death following seizures.
introducing substituents to the cyclohexyl ring did not increase
anticonvulsant potency in comparison to unsubstituted compound
11. The majority of substances in this group (except compounds 10,
11, 12, 13) showed neurotoxicity at 300 mg/kg.
Three compounds: 8, 17, 20 were administered orally to rats,
showing protection from 1 to 4 h at 30 mg/kg (compounds 8, 20) in
the electric test or at 50 mg/kg (compound 17) in the chemical test
without symptoms of neurotoxicity. For compound 17 the ED50
(>200 mg/kg) was determined. Imidazopurinediones were the least
active compounds as anticonvulsants, only 33% of the investigated
compounds showed weak anticonvulsant activity in chemical
seizures at a high dose of 300 mg/kg with neurotoxicity at the same
dose (compounds 1, 4) or even at lower dose (compound 3).
The anticonvulsant activity of the examined compounds was
analysed for correlation with AR affinity and some coincidence of
adenosine A2A affinity and anticonvulsant activity was observed.
The best A2A ligands among pyrimido- and diazepinopurinediones
Among pyrimidopurinedione derivatives 59% of investigated
substances showed anticonvulsant activity in chemical tests or in
both tests (compounds 8, 20). In this group of compounds

ꢀ
A. Drabczynska et al. / European Journal of Medicinal Chemistry 46 (2011) 3590e3607
3602
concluded that the proper range of lipophilicity is clearly not the
only parameter relevant for adenosine receptor affinity.
6. QSAR prediction
The process of QSAR model development can be generally
divided into three stages: data preparation, data analysis, and
model validation. The first stage includes the selection of a molec-
ular dataset, calculation of molecular descriptors, and the choice of
the QSAR approach in terms of the statistical methods of data
analysis and correlation. The second part involves the building of
models that correlate descriptor values with those of biological
activity. The final part of QSAR model development is the model
validation [58]. In the present study the investigations have been
performed for the dataset of 32 active A2A adenosine receptor
ligands.
Fig. 9. Correlation between RM0 and log P values calculated by CAChe 6.1.
showed also the best anticonvulsant activity (compounds 8, 15, 17,
20, 21, 29). This correlation was significantly distinct in compound
20. S-Configuration was beneficial for both activities. Suggestions in
recent work [55] of proconvulsant but not anticonvulsant effects of
adenosine receptor antagonists were not confirmed in this study
and the correlation is not clear. The selectivity profile appears to be
important, A2A-selective AR antagonists being neuroprotective
while A1 antagonists may show proconvulsant effects.
6.1. Model building
The dataset was randomly split into two sets: training set (TS)
for model building and prediction set (PS) for the model validation
prediction. The TS contained 82% of the tested ligands and the PS
18% covering the whole range of binding affinities [59]. For devel-
oping QSAR equations for A2A AR binding affinity prediction the set
of 39 ligands TS and PS contained 32 and
respectively.
The structures of the synthesized compounds were built in
CAChe 7.6 workspace [60]. The topological, geometric and elec-
tronic descriptors [59] were calculated in CAChe 7.6 Project Leader
application.
Determination of the equation that best represented the
dependence of the A2A AR binding affinity on several input
descriptors was carried out using the MLR (multiple linear regres-
sion) technique incorporated into the Project Leader CAChe 7.6
application [60] using a randomized training set and a described
procedure [32].
7 compounds,
5. Physicochemical properties
Parameters of lipophilicity expressed by RMO values were
determined using planar RP-TLC. The theoretical partition coeffi-
cients (log P) were also calculated in Project Leader application
incorporated into the CAChe 6.1 software [56] using the atom
typing scheme of Ghose and Crippen [57].
The calculated and experimental lipophilicity values of the
examined compounds are collected in the Table in Supplementary
data. The RM0 values are in the range from ꢁ0.085 (compound 5) to
3.442 (compound 33). For a number of compounds the RM0 values
are below 2, which may suggest weak BBB permeation. It can be
observed that for the analogous compounds, e.g.1,10, 25 or 2, 17, 31,
the RM0 values increase with the size of the annelated ring. For
structural isomers 15e17 as well as 29e31 the values of lip-
ophilicity are the same. Interestingly for the mixtures of cis/trans
isomers (compounds 5, 15, 16, 28, 29, 30) values of RM0 did not
differ and only one spot on the plate could be observed, instead of
the expected two spots, for the cis and the trans form. This differ-
entiation was seen only for two substances 12 and 27, where two
spots for cis and transisomers were obtained.
All the models having cross-validated q2 > 0.50 and a regression
coefficient r2 > 0.60 were collected. Two outliers within the TS, that
did not fit the model and which could probably act involving
another receptor binding mode, were identified and excluded. The
total amount of these outliers was not exceeding 7% of the TS. More
that 30 models were collected.
6.2. Model validation
The predictive ability of a QSAR model can only be estimated
using an external test set of compounds (PS) that was not used for
building the model. The following criteria for a QSAR model to have
high predictive power should be satisfied [58,61]:
The correlation coefficient R2 was estimated for experimental
and calculated partition coefficients by means of linear regression
analysis. The determined R2 was 0.87 suggesting high correlation of
the calculated log P with the experimental RMO values (Fig. 9).
However, there was no correlation between adenosine receptor
affinity of the compounds and their partition coefficients. It can be
Table 6
A2A AR QSAR model. Summary of the statistics.
Number of compounds
31
r2a
q2b
kc
0.77
0.68
0.70
1.00
c
k0
a
Regression coefficient.
Cross-validated regression coefficient.
Slopes of regression lines.
b
Fig. 10. Experimental vs. predicted Ki values [mM] obtained by using the QSAR
equation.
c

ꢀ
A. Drabczynska et al. / European Journal of Medicinal Chemistry 46 (2011) 3590e3607
3603
Table 7
Summary of the statistics for the A2AAR QSAR model. Validation protocol.
Number of compounds
6
q2
R
0.88
0.96
0.92
0.92
0.99
R2
k
k0
1. high value of cross-validated coefficient after leave-one-out
procedure q2 > 0.5, non-cross-validated regression coefficient
r2 > 0.6 (Table 6, Fig. 10);
2. correlation coefficient R between the predicted and observed
activities of compounds from an external test set (PS) close to 1
(Table 7, Figs. 11 and 12);
3. at least one of the correlation coefficients for regressions
through the origin (predicted versus observed activities, or
Fig. 12. Comparison of experimental Ki values (-)[
QSAR model (A).
m
M] with those predicted by the
2
2
observed versus predicted activities), i.e. R0 or R00 should be
close to R2 (Table 7, Figs. 11 and 12) or
4. slopes k and k0 of regression lines (predicted versus observed
activities, or observed versus predicted activities) through the
origin 0.85 ꢄ k ꢄ 1.15, 0.85 ꢄ k0 ꢄ 1.15 (Table 7, Figs. 11 and 12);
preference is given to the 6-membered third cycle and large
cycloalkyl substituents. The surface area of the negatively charged
electrostatic potential (SAiꢁ) represents the distance from the
structure at which a proton experiences a set attraction. The value
of the electrostatic potential on the surface is ꢂ18 kcal/mol. A
positive contribution coefficient implies that this descriptor should
have a small SAiꢁ value to possess higher affinity. Indeed, the
ligands in this series with polar electronegative substituents are not
favourable for A2A AR affinity. The last descriptor in Eq. (1) is the
surface area of the electrostatic potential on electron density that
represents charge distribution in a molecule and also is a more
accurate representation of the true shape of the ligand. Its positive
contribution coefficient suggests that this surface area should be
small. Taking into account the preferences for the ligand e which is
a 6- or 7-membered annelated ring with a bulky cycloalkyl
substituent, the last descriptor might suggest that the molecule
should also have a twisted conformation in order possess a smaller
surface area.
To validate the obtained 3D-QSAR models we selected 6 mole-
cules with a different spectrum of structure and affinity (prediction
test PS). The statistics for the best model is shown in Table 6 and
Fig. 7. The best 3D-QSAR model obtained is:
Ki(A2A AR) ¼ ꢁ2.546*log P ꢁ 18.674*EHOMO þ 0.007*SAiꢁ þ
0.049*SED ꢁ 92.3163,
(1)
where log P e octanol/water partition coefficient, EHOMO
e
HOMO energy, SAiꢁ e surface area of the negative-charged elec-
trostatic isopotential, SED e surface area of the electrostatic
potential on electron density.
The best equation (1) predicted the affinity in the PS with
a correlation coefficient between predicted and observed affinity of
R ¼ 0.96, Table 6, Fig. 11. The predicted Ki values were close to the
experimental values, as shown in Fig. 12. For the best model
obtained the statistical criteria are as follows (Table 7).
The first descriptor in the A2A AR QSAR model is log P with
a negative contribution coefficient, i.e. the higher the log P value of
the molecule, the better the A2A AR affinity of the ligand. Preferable
would be a six- or seven-membered annelated ring with cyclohexyl
and larger substituents. The second descriptor in Eq. (1) is EHOMO e
the energy required to remove an electron from the highest occu-
pied molecular orbital. In general, this term characterizes the
reactivity of the molecule, the ability to form bonds. As EHOMO has
a negative value and a negative contribution coefficient, the higher
energy HOMO would result in better A2A AR affinity. Here the
In conclusion, the obtained 3D-QSAR model (1) accurately pre-
dicted A2A AR affinity of cycloalkyl-substituted tricyclic xanthine
derivatives. Introduction of 3D-properties such as surface areas of
the potentials significantly improved the predictive ability of the
equation. It appears that the ligand binding affinity strongly
depends on the electron surface properties of the molecule. The
analysis of the obtained equation outlined the preferences for A2A
AR ligands and therefore will be helpful for further design of A2A
selective AR antagonists.
-
7. Conclusions
A series of 33 new cycloalkyl derivatives of imidazo-, pyrimido-
and 1,3-diazepino[2,1-f]purinediones were obtained. The new
compounds were tested for their adenosine receptor affinity and
exhibited selectivity for adenosine A2A receptors. The most potent
A2A antagonists were found among the pyrimido[2,1-f]purine-
diones, while 1,3-diazepino[2,1-f]purinediones showed the best
anticonvulsant properties. Lipophilicity was not correlated with the
observed pharmacological activities. It was assumed that small
differences in the shape and significant changes in the electrostatic
potentials are responsible for the differences in activity. A QSAR
model, able to predict the A2AR affinity within the group of tricyclic
xanthine derivatives possessing cycloalkyl substituents, was
proposed and validated. This approach can be considered useful for
further search for tricyclic xanthine antagonists possessing the
desired high A2A AR binding affinity. In addition, the cycloalkyl-
Fig. 11. Experimental vs. predicted Ki values [mM] for the external test set.

ꢀ
A. Drabczynska et al. / European Journal of Medicinal Chemistry 46 (2011) 3590e3607
3604
substituted tricyclic xanthine derivatives contain a basic nitrogen
atom, which can easily be protonated, and thus leads to increased
water-solubility in comparison to most other xanthine derivatives
as well as non-xanthine AR antagonists.
T ¼ 293 K,
m
¼ 0.12 mmꢁ1
,
l
¼ 1.54178, data/parameters ¼ 3368/252;
final R1 ¼0.0527.
Crystal data for 33: C19H29N5O2, M ¼ 359.47, orthorhombic, space
ꢁ
ꢁ
ꢁ
group Pna21, a ¼ 23.9625(14) A, b ¼ 11.4933(6) A, c ¼ 20.9751(16) A,
3
V ¼ 5776.7(6) A ,
Z ¼ 12,
Dx ¼ 1.240 g cmꢁ3
,
T ¼ 293 K,
ꢁ
¼ 0.083 mmꢁ1
,
l
¼ 0.71073 A, data/parameters ¼ 9093/839; final
ꢁ
8. Experimental protocols
m
R1 ¼0.0540.
8.1. Chemistry
The crystals of 7, 13 and 33 were obtained by slow evaporation
from methanol and propanol (1:1) solutions. The measurements of
Melting points were determined on a Mel-Temp II apparatus. IR
spectra were taken as KBr discs on an FT Jasco IR 410 apparatus. 1H
NMR spectra were performed with a Varian Mercury 300 MHz
spectrometer in DMSO-d6 (5, 12, 13) or CDCl3 (the remaining
compounds) with TMS as an internal standard. UV spectra were
recorded on a UVeVis V530 spectrophotometer at a concentration
of 1 ꢅ10ꢁ5 mol/L in methanol. Elemental analyses (C, H and N)
were performed on an Elemental Vario-EL III apparatus and were in
accordance with theoretical values within ꢂ0.4%. TLC data were
obtained with Merck Silica Gel 60F254 aluminum sheets with
developing systems A, B, C. Spots were detected under UV light.
Measurements of optical rotation were carried out on a Jasco Dipol
1000 polarimeter conc. 2 in ethanol (c ¼ 2 g/100 ml, ethanol).
Spectroscopic data are presented in Table 2.
7 and 33 crystals were performed on a Kuma4CCD
tometer with graphite-monochromated Mo
k
-axis diffrac-
Ka radiation
ꢁ
(l
¼ 0.71073 A) at room temperature. While for 13 measurements
were performed on
monochromated Cu K
a SMART diffractometer with graphite-
ꢁ
a
radiation (
l
¼ 1.154178 A) at room temper-
ature. The structures were solved by direct method using the
SHELXTS program [62]and refined with SHELXTL [63]. E-map
provided positions for all non-H-atoms. The full-matrix least-
squares refinement was carried out on F2 using anisotropic
temperature factors for all non-H-atoms in 7 and 33. While in 13
mobile solvent atoms, located in the channel down c-axis, were left
with isotropic temperature factors. In structures 13 and 33 posi-
tional disorder for carbon atoms was identified and refined. In
structure 13, the H-atoms attached to O17 and O1s were located in
difference Fourier map and their positions were refined freely with
Uiso ¼ 1.5Ueq(O). For all three structures H-atoms bonded to carbons
were placed in idealized position and considered to ride on their
parent atoms.
8.1.1. General procedure for the synthesis of cycloalkyl-substituted
1,3-dimethyl-6,7,8,9-tetrahydro-(8H)-imidazopyrimido-6,7,8,9-
tetrahydro(10H)-1,3-diazepino[2,1-f]purine-2,4-(1H,3H)-diones
(1e33) (except: 14, 18, 28)
Special comments to structure 33: The structure was solved and
refined in Pna21 non-centrosymmetric space group with three
molecules in independent crystallographic unit. Two of them differ
insignificantly in cyclooctyl conformation. The third one, with
disordered diazepine and cyclooctyl (both with s.o.f. ¼ 0.5), consists
in fact two identical molecules repeat by mirror. The final R1-factor
equals 0.0540. After transformation to Pnma centrosymmetric
space group with one molecule in general and the other in special
position, final R1 stopped at 0.12. Therefore refinement of the
structure 33 was left in a non-centrosymmetric space group.
Crystallographic data (excluding structural factors) for the
structure reported in this paper have been deposited with the
Cambridge Crystallographic Data Centre and allocated the deposi-
tion numbers: (CCDC 741160e741162) for compounds 7, 13 and 33
respectively. Copies of the data can be obtained free of charge on
application to CCDC, 12 Union Road, Cambridge CB2 1EW, UK (fax:
þ44 1223 336 033; e-mail: deposit@ccdc.cam.ac.uk).
A mixture of 0.73 g (2 mmol) of 7-(2-bromoethyl)-8-bromoth-
eophylline, 0.66 g (2 mmol) of 7-(3-chloropropyl)-8-bromo-
theophylline, 0.79 g (2 mmol) of 7-(4-bromobutyl)-8-bromoth
eophylline and the appropriate amine (2e20 fold excess) was
refluxed in DMF, butanol, xylene, or Me-Digol, or without solvent
(compounds 5, 11, 26) for 5e10 h (see Table 1). The progress of the
cyclization was monitored by thin layer chromatography (TLC). The
reaction was carried out until the spot of the starting material had
disappeared. After cooling the precipitate was separated
(compounds 1, 2, 4, 7, 10, 13, 15, 17, 22, 23, 24, 30) and washed with
ethanol and water. Other compounds precipitated by cooling and
adding water (8, 9, 12, 16, 19, 20, 21, 25, 26, 27, 31, 32, 33) to the
reaction mixture. Some compounds were separated after distilla-
tion of solvent and adding water (5, 29) or water and NaOH (6).
Acetyl derivatives 14, 28 were prepared by refluxing compounds 13
and 27 with acetic anhydride for 5 h, removing the excess of
anhydride by distillation under reduced pressure and crystalliza-
tion of the residue.
8.3. Pharmacology
Optically active compounds 19, 20 were obtained by cyclization
of 7-(3-chloropropyl)-8-bromotheophylline with commercially
8.3.1. Adenosine receptor binding assays
available enantiomerically pure S(þ)-1-cyclohexylethylamine
Adenosine binding assays were performed as previously
described [49,64] using rat brain cortical membrane preparations for
A1 AR assays and rat brain striatial membrane preparations for A2A
assays. Frozen rat brains (unstripped) were obtained from Pel-FreezÒ,
Rogers, Arkansas, USA. For assays at human A1, A2A, A2B and A3, ARs,
CHO cell membranes expressing the human receptors were used as
described [65]. [3H]2-chloro-N6-cyclopentyladenosine ([3H]CCPA)
was used as the A1 radioligand, [3H]3-(3-hydroxypropyl)-7-methyl-
8-(m-methoxystyryl)-1-propargylxanthine ([3H]MSX-2) as the A2A
radioligand, [3H]4-(2-[7-amino-2-(2-furyl)-[1,2,4]-triazolo-[2,3-a]-
[1,3,5]-triazin-5ylamino]ethyl)phenol ([3H]ZM241385) and [3H]8-(4-
(4-(4-chlorophenyl)piperazine-1-sulfonyl)phenyl)-1-propylxanthine
([3H]PSB-603)asA2B receptorradioligandsand[3H]phenyl-8-ethyl-4-
methyl-(8R)-4,5,7,8-tetrahydro-1H-imidazo[2,1-i]purine-5-one ([3H]
PSB-11) as the A3 AR radioligand. Initially, a single high concentration
20
([a]
]
¼ þ3.8 ꢂ 0.3 (Fluka)) and R(ꢁ)-1-cyclohexylethylamine
¼ ꢁ3.8 ꢂ 0.3 (Fluka)). Racemic compound 18 was a stoi-
D
20
([
a
D
chiometric mixture of compounds 19 and 20. Spectroscopic data
are presented in Table 8. The elemental analyses can be found in
Supplementary data.
8.2. X-ray structure analysis
Crystal data for 7: C13H17N5O2, M ¼ 275.32, monoclinic, space
ꢁ
ꢁ
ꢁ
group P21/c, a ¼ 8.3797(9) A, b ¼ 16.6762(16) A, c ¼ 9.6895(8) A,
3
ꢁ
b
¼ 105.482(10)ꢀ, V ¼ 1304.9(2) A , Z ¼ 4, Dx ¼ 1.401 g cmꢁ3
,
T ¼ 293 K,
m
¼ 0.099 mmꢁ1
,
l
¼ 0.71073 A, data/parameters ¼ 2286/
ꢁ
185; final R1 ¼0.0460.
Crystal data for 13: C16H23N5O3$CH3OH, M ¼ 365.44, monoclinic,
ꢁ
ꢁ
ꢁ
space group P21/c, a ¼ 9.7248(2) A, b ¼ 36.4797(7) A, c ¼ 5.1337(1) A,
of compound (25
mM at A1 and A2A, 10 mM at A2B and A3 receptors)
3
ꢁ
b
¼ 92.671(1)ꢀ,
V ¼ 1819.24(6) A ,
Z ¼ 4,
Dx ¼ 1.334 g cmꢁ3
,
was tested in three (A1, A2A) or two (A2B, A3) independent

ꢀ
A. Drabczynska et al. / European Journal of Medicinal Chemistry 46 (2011) 3590e3607
3605
Table 8
Spectral data of cycloalkylpyrimido-, imidazo- and diazepino[2,1-f]purinediones.
Compound UV lmax IR (ppm)
(cmꢁ1 1H NMR
n
)
d
1
300
302
1693 e CO (pos. 2) 1.64e1.99 (2 m, 8H) cyclopentyl; 3.36 (s, 3H, N3CH3); 3.51 (s, 3H, N1CH3); 3.84 (t, 2H, J ¼ 8.10 Hz, N5CH2); 4.05e4.20
1648 e CO (pos. 4) (m, 3H, CH2N2 þ N9CH); 0.98e1.77 (3 m, 11H, cyclohexyl); 3.16 (d, 2H, J ¼ 7.15 Hz, N8CH2); 3.36 (s, 3H, N3CH3); 3.50
(s, 3H, N1CH3); 3.86 (q, 2H, J ¼ 7.43 Hz, CH2N8); 4.19 (q, 2H, J ¼ 7.29 Hz, N5CH2)
2
1700 e CO (pos. 2) 0.98e1.77 (3 m, 11H, cyclohexyl); 3.16 (d, 2H, J ¼ 7.15 Hz, N8CH2); 3.50 (s, 3H, N1CH3); 3.86
1655 e CO (pos. 4) (q, 2H, J ¼ 7.43 Hz, CH2N8); 4.19 (q, 2H, J ¼ 7.29 Hz, N5CH2)
3
4
[35]
303
1700 e CO (pos. 2) 1.49 (2 m, 8H, cyclohexenyl); 2.26 (t, 2H, J ¼ 7.18 Hz, CH2CH2); 3.36 (s, 3H, N3CH3); 3.43 (t, 2H, J ¼ 7.31 Hz, N8CH2); 3.50
1651 e CO (pos. 4) (s, 3H, N1CH3); 3.87 (q, 2H, J ¼ 7.31 Hz, CH2N8); 4.15e4.20 (m, 2H, N5CH2); 5.47 (s, 1H, cyclohexenyl)
5
302
3391 e OH
1.17e1.98 (3, m, 8H, cyclohexyl; 3.13 (s, 3H, N3CH3); 3.22e3.26 (m, 2H, CH2N9); 3.32
1700 e CO (pos. 2) (s, 4H, N1CH3 þ 40-CH cyclohexyl); 3.98e4.1 (m, 3H, N5CH2 þ N8CH); 4.57 (d, 1H, J ¼ 5.00 Hz, OH)
1654 e CO (pos. 4)
6
7
8
302
1693 e CO (pos. 2) 1.59e1.85 (m, 14H, cyclooctyl; 3.37 (s, 3H, N3CH3); 3.52 (s, 3H, N1CH3); 3.82e3.90 (33H, CH2N8 þ N8CH); 4.14
1648 e CO (pos. 4) (q, 2H, J ¼ 7.2 Hz, N5CH2)
299
1702 e CO (pos. 2) 0.72e0.88 (2, m, 4H, cyclopropyl); 2.08e2.16 (m, 2H, CH2eCH2eCH2); 2.65e2.72 (m, 1H, N9CH); 3.35e3.39
1653 e CO (pos. 4) (m, 5H, N3CH3 þ CH2N9); 3.54 (s, 3H, N1CH3); 4.19 (t, 2H, J ¼ 6.16 Hz, N5CH2)
1697 e CO (pos. 2) 0.28 (d, 2H, J ¼ 4.87 Hz, cyclopropyl; 0.55 (d, 2H, J ¼ 7.4 Hz, 2H, cyclopropyl); 1.05e1.038 (m, 1H, cyclopropyl);
2.13e2.20 (m, 2H, CH2eCH2eCH2); 3.36 (s, 3H, N3CH3); 3.40e3.48 (m, 4H, CH2N9 þ N9CH2); 3.49 (s, 3H, N1CH3); 4.22
(t, 2H, J ¼ 5.90 Hz, N5CH2)
301.5
9
302
1699 e CO (pos. 2) 1.64e1.75 (m, 2H, CH2eCH2eCH2); 2.09e2.25 (m, 6H, cyclobutyl); 3.35e3.40 (m, 5H, N3CH3 þ CH2N9); 3.50
1639 e CO (pos. 4) (s, 3H, N1CH3); 4.19 (2H, J ¼ 6.03 Hz, N5CH2); 4.77e4.89 (q, 1H, N9CH)
1697 e CO (pos. 2) 1.56e1.93 (m, 8H, cyclopentyl); 2.05e2.13 (m, 2H, CH2eCH2eCH2); 3.28 (t, 2H, J ¼ 5.63 Hz, CH2N9); 3.34
1664 e CO (pos. 4) (s, 3H, N3CH3); 3.48 (s, 3H, N1CH3); 4.18 (t, 2H, J ¼ 6.04 Hz, N5CH2); 4.73e4.83 (m, 1H, N9CH)
10
301.5
11
12
[35]
301
3484 e OH
1.21e2.01 (m, 10H, cyclohexyl þ CH2eCH2eCH2); 3.13 (s, 3H, N3CH3); 3.23e3.37 (m, 5H, N1CH3 þ CH2N9); 3.80
1701 e CO (pos. 2) (s, 1H, 4-cyclohexyl); 3.98e4.1 (m, 3H, N5CH2 þ N9CH); 4.38 (s, 1H, OH ax); 4.56 (s, 1H, OH eq)
1652 e CO (pos. 4)
13
14
302
301
3384 e OH
1.16e1.30 (m, 2H, cyclohexyl); 1.58e1.68 (m, 4H, cyclohexyl); 1.85e1.98 (2m, 2H, CH2eCH2eCH2 þ cyclohexyl); 3.11
1702 e CO (pos. 2) (s. 3H, N3CH3); 3.24 (t, 2H, J ¼ 6.33 Hz, CH2N9); 3.30 (s, 3H, N1CH3); 3.40e3.49 (m, 1H, 40-cyclohexyl); 3.97e4.11
1660 e CO (pos. 4) (m, 3H, N5CH2 þ N9CH); 4.56 (t, 1H, J ¼ 4.38 Hz. OH)
1731 e (OCOCH3)
1.51e1.73 (m, 4H, cyclohexyl); 1.81e1.85 (m, 2H, CH2CH2CH2); 2.04e2.14 (m, 7H, OCOCH3 þ cyclohexyl); 3.27
1698 e CO (pos. 2) (t, 2H, J ¼ 5.52 Hz, CH2N9); 3.36 (s, 3H, N3CH3); 3.51 (s, 3H, N1CH3); 4.17e4.29 (m, 3H, N5CH2 þ 40-CH cyclohexyl);
1650 e CO (pos. 4) 4.63e4.70 (m, 1H, N9CH)
15
16
304
303
1720 e CO (pos. 2) 0.921e1.03 (d, 3H, J ¼ 6.25 Hz. J ¼ 7.00 Hz, cis/trans CH3); 1.51e1.98 (m, 9H, cyclohexyl); 2.06e2.15
1655 e CO (pos. 4) (m, 2H, CH2eCH2eCH2); 3.08e3.36 (m, 5H, N3CH3 þ CH2N9); 3.50 (s, 3H, N1CH3); 4.17e4.22 (m, 3H, N5CH2, N9CH)
1698 e CO (pos. 2) 0.86 (d, 2H, J ¼ 6.41 Hz, CH3); 0.95 (d, 1H, J ¼ 7.18 Hz, CH3); 1.14e1.87 (m, 9H, cyclohexyl); 2.05e2.13
1655 e CO (pos. 4) (m, 2H, CH2eCH2eCH2); 3.24e3.34 (m, 2H, CH2N9); 3.36 (s. 3H, N3CH3); 3.50 (s, 3H, N1CH3); 4.07e4.31
(m, 3H, N5CH2, N9CH)
17
303.5
303
303
303
303
1705 e CO (pos. 2) 0.97e1.27 (m, 10H, cyclohexyl); 1.66e1.76 (m, 2H, CH2eCH2eCH2); 3.35 (d. 7H, J ¼ 5.77 Hz, N3CH3 þ CH2N9 þ N9CH2);
1651 e CO (pos. 4) 3.50 (s, 3H, N1CH3); 4.21 (t, 2H, J ¼ 6.04 Hz, N5CH2)
18
1699 e CO (pos. 2) 0.94e1.80 (m, 14H, cyclohexyl þ CH3); 2.06e2.15 (m, 2H, CH2eCH2eCH2); 3.18e3.31 (m, 2H, CH2N9); 3.36
1656 e CO (pos. 4) (s, 3H, N3CH3); 3.50 (s, 3H, N1CH3); 4.11e4.26 (m, 3H, N5CH2 þ N9CH)
19a
20b
21
1699 e CO (pos. 2) 0.95e1.79 (m, 14H, cyclohexyl þ CH3); 2.07e2.14 (m, 2H, CH2eCH2eCH2); 3.16e3.30 (m, 2H, CH2N9); 3.36
1656 e CO (pos. 4) (s, 3H, N3CH3); 3.49 (s, 3H, N1CH3); 4.11e4.27 (m, 3H, N5CH2 þ N9CH)
1699 e CO (pos. 2) 0.95e1.80 (m, 14H, cyclohexyl þ CH3); 2.07e2.14 (m, 2H, CH2eCH2eCH2); 3.16e3.31 (m, 2H, CH2N9); 3.36
1656 e CO (pos. 4) (s, 3H, N3CH3); 3.49 (s, 3H, N1CH3); 4.11e4.27 (m, 3H, N5CH2); 4.49 (t, 1H, J ¼ 7.06 Hz, N9CH)
1698 e CO (pos. 2) 1.46e1.62 (m, 4H, cyclohexenyl); 1.92e2.01 (m, 4H, cyclohexenyl); 2.14e2.17 (2H, CH2eCH2eCH2); 2.25
1658 e CO (pos. 4) (t, 2H, J ¼ 6.92 Hz, CH3CH2); 3.28e3.47 (m, 5H, N3CH3 þ CH2N9); 3.50 (s, 3H, N1CH3); 3.60 (t, 2H, J ¼ 7.18 Hz, N9CH2);
4.19 (t, 2H, J ¼ 6.08 Hz, N5CH2); 5.13 (s, 1H, cyclohexenyl)
22
23
24
25
26
27
303
1698 e CO (pos. 2) 1.51e1.87 (m, 12H, cycloheptyl); 2.04e2.11 (m, 2H, CH2eCH2eCH2); 3.30 (t, 2H, J ¼ 5.62 Hz, CH2N9); 3.35
1652 e CO (pos. 4) (s, 3H, N3CH3); 3.51 (s, 3H, N1CH3); 4.17 (t, 2H, J ¼ 6.04 Hz, N5CH2); 4.34e4.40 (m, 1H, N9CH)
1702 e CO (pos. 2) 1.55e1.83 (m, 14H, cyclooctyl); 2.04e2.12 (m, 2H, CH2CH2CH2); 3.31 (t, 2H, J ¼ 5.63 Hz, CH2N9); 3.36 (s, 3H, N3CH3);
1653 e CO (pos. 4) 3.51 (s, 3H, N1CH3); 4.18 (t, 2H, J ¼ 6.04 Hz, N5CH2); 4.48e4.52 (m, 1H, N9CH)
303.5
304
1700 e CO (pos. 2) 1.71 (s, 6H, adamantyl); 2.04e2.09 (m, 2H, CH2CH2CH2); 2.16e2.27 (m, 9H, adamantyl); 3.36 (s, 3H, N3CH3); 3.43
1655 e CO (pos. 4) (t, 2H, J ¼ 5.64 Hz, CH2N9); 3.51 (s, 3H, N1CH3); 4.16 (t, 2H, J ¼ 6.16 Hz, N5CH2)
301
1699 e CO (pos. 2) 1.46e2.05 (3m, 12H, cyclopentyl þ CH2CH2CH2CH2); 3.12 (t, 2H, J ¼ 5.13 Hz, CH2N10); 3.37 (s, 3H, N3CH3); 3.52
1644 e CO (pos. 4) (s, 3H, N1CH3); 4.30 (t, 2H, J ¼ 5.26 Hz, N5CH2); 4.38e4.64 (m, 1H, N10CH)
302.5
301
1698 e CO (pos. 2) 1.32e1.51 (m, 10H, cyclohexyl); 1.80e2.00 (m, 4H, CH2CH2CH2CH2); 3.20 (t, 2H, J ¼ 4.72 Hz, CH2N10); 3.37
1656 e CO (pos. 4) (s, 3H, N3CH3); 3.51 (s, 3H, N1CH3); 3.89e3.91 (m, 1H, N10CH); 4.26e4.30 (m, 2H, N5CH2)
3442 e OH
1.43e1.94 (m, 12H, cyclohexyl þ CH2CH2CH2CH2); 2.06 (d, 1H, J ¼ 11.8 Hz, 40-cyclohexyl); 3.14e3.23
1696 e CO (pos. 2) (dt, 2H, J ¼ 5.52 Hz, J ¼ 5.39 Hz, CH2N10); 3.36 (s, 3H, N3CH3); 3.50 (s, 3H, N1CH3); 3.51e3.62 (m, 1H, OH);
1657 e CO (pos. 4) 3.87e3.98 (m, 1H, N10CH); 4.08 (s, 1H, OH trans); 4.28 (t, 2H, J ¼ 4.8 Hz, N5CH2)
28
29
30
31
300
301
303
300
1731 e OCOCH3
1.51e1.69 (m, 10H, cyclohexyl); 1.70e2.00 (m, 7H, cyclohexyl þ OCOCH3); 2.04e2.11 (m, 4H, CH2CH2CH2CH2);
1698 e CO (pos. 2) 3.13e3.22 (m, 2H, CH2N10); 3.37 (s, 3H, N3CH3); 3.51 (s, 3H, N1CH3); 3.91e3.96 (m, 1H, N10CH); 4.27e4.32
1669 e CO (pos. 4) (m, 2H, N5CH2)
1699 e CO (pos. 2) 0.91e1.02 (dd, 3H, J ¼ 6.41 Hz, J ¼ 7.18 Hz, cis/trans CH3); 1.01e1.18 (m, 1H ax, 40-cyclohexyl); 1.33e1.48
1656 e CO (pos. 4) (1H eq, 40-cyclohexyl); 1.67e1.91 (m, 12H, cyclohexyl þ CH2eCH2eCH2eCH2); 3.16e3.23 (m, 2H, CH2N10); 3.37
(s. 3H, N3CH3); 3.51 (s, 3H, N1CH3); 3.79e3.92 (m, 1H, N10CH); 4.26e4.30 (m, 2H, N5CH2)
1698 e CO (pos. 2) 0.92e1.01 (dd, 3H, J ¼ 7.8 Hz, J ¼ 6.41 Hz, cis/trans CH3); 1,1e2.01 (m, 12H, cyclohexyl þ CH2eCH2eCH2); 3.02e3.35
1656 e CO (pos. 4) (m, 2H, CH2N10); 3.36 (s, 3H, N3CH3); 3.51 (s, 3H, N1CH3); 3.60e3.64 (m, 1H, N10CH); 3.99e4.01 (m, 1H, 2cyclohexyl ax);
4.1e4.13 (m, 1H, 20-cyclohexyl eq); 4.29e4.65 (2m, 2H, N5CH2)
1698 e CO (pos. 2) 0.88e1.72 (3m, 11H, cyclohexyl); 1.75e1.95 (m, 4H, CH2eCH2CH2eCH2); 3.25 (t, 2H, J ¼ 5.13 Hz, CH2N10); 3.37
1663 e CO (pos. 4) (s, 5H, N3CH3 þ N10CH2); 3.51 (s, 3H, N1CH3); 4.32 (t, 2H, J ¼ 5.13 Hz, N5CH2)
(continued on next page)

ꢀ
A. Drabczynska et al. / European Journal of Medicinal Chemistry 46 (2011) 3590e3607
3606
Table 8 (continued )
Compound UV lmax IR
n
(cmꢁ1
)
1H NMR
d (ppm)
32
302
302
1695 e CO (pos. 2) 1.48e1.63 (m, 4H, cyclohexenyl); 1.86e1.97 (m, 8H, cyclohexenyl þ CH2CH2CH2CH2); 2,26 (t, 2H, J ¼ 7.30 Hz, CHCH2);
1651 e CO (pos. 4) 3.23e3.36 (m, 2H, CH2N10); 3.37 (s, 3H, N3CH3); 3.52 (s, 3H, N1CH3); 3.52e3.60 (m, 2H, N10CH2); 4.28 (t, 2H, J ¼ 5.00 Hz,
N5CH2), 5.44 (s, 1H, 20-cyclohexenyl)
33
1696 e CO (pos. 2) 1.51e1.88 (m, 18H, cyclooctyl þ CH2CHCH2CH2); 3.13 (t, 2H, J ¼ 5.13 Hz, CH2N10); 3.37 (s, 3H, N3CH3); 3.51
1655 e CO (pos. 4) (s, 3H, N1CH3); 4.16e4.27 (m, 3H, N5CH2 þ N10CH)
a
20
20
[
[
a
a
]
]
¼ þ33.26 (c ¼ 2 g/100 ml, ethanol) (conc. 2 in ethanol).
¼ ꢁ33.29 (c ¼ 2 g/100 ml, ethanol) (conc. 2 in ethanol).
D
D
b
experiments. For potent compounds, curves were determined using
6e7 different concentrations of test compounds spanning 3 orders
of magnitude. Data were analysed using the PRISM program version
3.0 or 4.0 (Graph Pad, San Diego, CA, USA).
groups of 4 rats were used. The test compounds were suspended in
a 0.5% methylcellulose/water mixture. In the preliminary screening
each compound was administered as an ip injection at three dose
levels (30, 100 and 300 mg/kg) with anticonvulsant activity and
neurotoxicity assessed at 0.5 and 4 h intervals after administration.
For some compounds also intervals 0.25, 1 and 2 h were applied.
Anticonvulsant efficacy was measured by maximal electroshock
(MES) and subcutaneous pentylenetetrazole (ScMet), neurological
deficit was investigated in the rotorod test; the data are presented in
Table 4. Compounds 8,17, 20, 29, 31 were examined fororal activity in
the rat ScMet and neurotoxicity screen at 30 mg/kg doses (Table 5).
The pharmacological parameters estimated in the preliminary
screening were quantified for compounds 17 and 31. Anticonvulsant
8.3.1.1. Functional assays. Stably transfected CHO cells expressing
the human A2A receptor were grown in DMEM-F12 medium (Invi-
trogen) with 10% fetal calf serum, 100 U/ml penicillin G, 100 mg/ml
streptomycin and 1% ultraglutamine at 37 ꢀC and 5% CO2. For the
experiment they were transferred to 24-well plates at a density of
200,000 cells per well. After 24 h the medium was removed and the
cells were washed with 500 m
l of 37 ꢀC warm Hank’s Balanced Salt
Solution (HBSS; 20 mM HEPES, 13 mM NaCl, 5.5 mM glucose,
5.4 mM KCl, 4.2 mM NaHCO3, 1.25 mM CaCl2, 1 mM MgCl2, 0.8 mM
MgSO4, 0.44 mM KH2PO4 and 0.34 mM Na2HPO4, pH adjusted to 7.3)
containing 1 U/ml of adenosine deaminase (ADA, Sigma). The cells
activity was expressed in terms of the median effective dose (ED50
)
in rats after oral administration. For determination of the ED50 value
groups of 8 rats were given a range of p.o. doses of the test drug until
at least three points were established in the range of 10e90% seizure
protection. From the plot of this data, the respective ED50 values, 95%
confidence intervals, slope of the regression line, and the standard
error of the slope were calculated by means of a computer program
written at NINDS, NIH.
were then incubated in 300
for 2 h. Then, 100 l of the phosphodiesterase inhibitor Ro20-1724
(Hoffmann La Roche; final concentration 40 M) was added to
each well and the cells were incubated for 15 min at 37 ꢀC and 5%
CO2. Then 100
l of various dilutions of the agonist 50-N-ethyl-
m
l of HBSS with ADA at 37 ꢀC and 5% CO2
m
m
m
carboxamidoadenosine (NECA; Sigma) in the presence or absence of
a single concentration of test compound in HBSS containing 5%
DMSO were added in triplicates. After 15 min of incubation at 37 ꢀC
8.4. Determination of lipophilicity
and 5% CO2 the supernatant was removed and 500
m
l of 90 ꢀC hot
8.4.1. Chromatographic determination of lipophilicity
lysis buffer consisting of 4 mM EDTA and 0.01% Triton X-100 with
the pH adjusted to 7.3 were added. After 1 h of mixing on ice, cAMP
amounts of the lysates were determined by competitive radioligand
binding experiments [66]. cAMP competition experiments were
Methanolic solutions of all compounds were prepared at 1 mg/
ml concentration. The solutions were spotted onto TLC plates
(20 ꢅ 10 cm) precoated with RP-18 silica gel F254 (MERCK) using
a Hamilton syringe in an amount of 10 ml. A mixture of acetone with
performed in a final volume of 120
30
of [3H]cAMP radioligand solution in lysis buffer (final
concentration 3 nM) and 40 l of cAMP binding protein [66] diluted
in the same buffer (50 g per sample). For determining cAMP
concentrations 50 of various cAMP concentrations were
measured instead of cell lysates, to obtain a standard curve. Total
binding was determined by adding radioligand and binding protein
to lysis buffer, and the background was determined without addi-
tion of binding protein. The mixture was incubated for 60 min on ice
and filtered through a GF/B glass fiber filter using a cell harvester
(Brandel). The filters were washed three times with 2e3 ml of ice-
cold 50 mM TriseHCl buffer, pH 7.4 and subsequently transferred
into scintillation vials. The liquid scintillation counting of the filters
started after 9 h of incubation in 2.5 ml of scintillation cocktail
(Lumag AG, Basel). Three separate experiments were performed. The
amount of cAMP was determined by comparison to a standard curve
generated for each experiment.
m
l containing 50
m
l of cell lysates,
water in the range of 50e85% (v/v) in 5% increments was used as
a mobile phase. The plates were evaluated in horizontal chro-
matographic chambers (CHROMODES), saturated with eluent for
45 min (þ15 min with plates). The RM values were calculated from
Rf values using the equation: RM ¼ log(1/Rf ꢁ 1). RM values were
then extrapolated to zero acetone concentration (pure water) by
use of equation: RM ¼ RMO þ bc.
ml
m
m
ml
8.4.2. Calculated lipophilicity
The structures of the ligands were built in CAChe 6.1 [56], the
geometry was optimized in MOPAC with PM5 parameters. The
lipophilicity was calculated using the CAChe 6.1 Project Leader
application with the atom typing scheme of Ghose and Crippen
[57].
8.5. QSAR
The structures of the synthesized compounds were built in
CAChe 7.6 workspace [60] and the geometry was optimized in
MOPAC (Molecular Orbital Package) using semi-empirical Hamil-
tonian (PM5) to minimize the energy. To find the low-energy
conformers the CONFLEX application was used. The energy of
each conformation was plotted in a three-dimensional graph. The
low-energy conformations, separated by high energy barriers, were
collected and optimized using DFT methods B88-LYP functional
8.3.2. Anticonvulsant screening
The anticonvulsant evaluation was carried out using reported
procedures [52,53]. Male albino mice (F-1 strain,18e25 g) were used
as experimental animals. For testing compounds 8,17, 20, 29, 31 male
albino rats (Sprague-Dawley 100e150 g) were used. Groups of 1e5
mice were used in MES, ScMet tests, groups of 2e8 animals in the
rotorod test. For the evaluation of activity after oral administration,

ꢀ
A. Drabczynska et al. / European Journal of Medicinal Chemistry 46 (2011) 3590e3607
3607
with 6-31G** basis set. The topological, geometric and electronic
descriptors [59] were calculated in CAChe 7.6 Project Leader
application by the atom typing scheme; by Mechanics using
Augmented MM3; by MOPAC with PM5 parameters; by DFT
methods: D-VWN LDA functional with DZVP basis set or B88-LYP
GGA functional with 6-31G** basis set. The statistical parameters
were calculated using Project Leader application (for R2 and q2) of
CAChe 7.6 and Microsoft Excel (for k and k0).
[22] R. Sauer, J. Maurinsh, U. Reith, F. Fülle, N. K-Klotz, C.E. Müller, J. Med. Chem. 43
(2000) 440e448.
[23] M. Yang, D. Soohoo, S. Soelaiman, R. Kalla, J. Zablocki, N. Chu, K. Leung, L. Yao,
I. Diamond, L. Belardinelli, J.C. Shryock, Naunyn-Schmiedeberg’s Arch. Phar-
macol. 375 (2007) 133e144.
[24] E. Ongini, A. Monopoli, B. Cacciari, P.G. Baraldi, Farmaco 56 (2001) 87e90.
[25] T.N. Chase, F. Bibbiani, W. Bara-Jimenez, T. Dimitrowa, J.D. Oh-Lee, Neurology
61 (2003) 107e111.
[26]
S.S. Wu, S.J. Frucht, CNS Drugs 19 (2005) 723e743.
ꢀ
ꢀ
[27] K. Kiec-Kononowicz, A. Drabczynska, E. Pe˛kala, B. Michalak, C.E. Müller,
B. Schumacher, J. Karolak-Wojciechowska, H. Duddeck, S. Rockitt,
R. Wartchow, Pure Appl. Chem. 73 (2001) 1411e1420.
ꢀ
[28] A. Drabczynska, B. Schumacher, C.E. Müller, J. Karolak-Wojciechowska,
Acknowledgements
ꢀ
B. Michalak, E. Pe˛kala, K. Kiec-Kononowicz, Eur. J. Med. Chem. 38 (2003)
397e402.
ꢀ
[29] A. Drabczynska, C.E. Müller, B. Schumacher, S. Hinz, J. Karolak-Wojciechow-
The authors are grateful to Professor James Stables for providing
the biological data through the Antiepileptic Drug Development
(ADD) Program of the National Institute of Neurological and
Communicative Disorders and Stroke at the National Institute of
Health, Bethesda, MD, USA. We thank Dr. Britta Schumacher for
performing some of the radioligand binding experiments. This
work was partly supported by the Polish Ministry of Science and
Higher Education e grants No 2 P05F 022 26 and N N405 297 836
and by the Deutsche Forschungsgemeinschaft (GRK 677).
ꢀ
ska, B. Michalak, E. Pe˛kala, K. Kiec-Kononowicz, Bioorg. Med. Chem. 12 (2004)
4895e4908.
ꢀ
[30] A. Drabczynska, C.E. Müller, S.K. Lacher, B. Schumacher, J. Karolak-Wojcie-
ꢀ
chowska, A. Nasal, P. Kawczak, O. Yuzlenko, K. Kiec-Kononowicz, Bioorg. Med.
Chem. 14 (2006) 7258e7281.
ꢀ
[31] A. Drabczynska, C.E. Müller, J. Karolak-Wojciechowska, B. Schumacher,
ꢀ
A. Schiedel, O. Yuzlenko, K. Kiec-Kononowicz, Bioorg. Med. Chem. 15 (2007)
5003e5017.
ꢀ
[32] A. Drabczynska, C.E. Müller, A. Schiedel, B. Schumacher, J. Karolak-Wojcie-
ꢀ
ꢀ
chowska, A. Fruzinski, W. Zobnina, O. Yuzlenko, K. Kiec-Kononowicz, Bioorg.
Med. Chem. 15 (2007) 6956e6974.
[33] S. Rockitt, R. Wartchow, H. Duddeck, A. Drabczynska, K. Kiec-Kononowicz,
ꢀ
ꢀ
Z. Naturforsch. 56B (2001) 319e324.
[34] M. Paw1owski, A. Drabczynska, M. Gorczyca, D. Malec, J. Modzelewski, Acta
Appendix. Supplementary data
ꢀ
Polon. Pharm.-Drug Res. 51 (1994) 385e391.
[35] M. Eckstein, Dissert. Pharm. 14 (1962) 425e434.
[36] F. Bergmann, S. Dickstein, J. Am. Chem. Soc. 77 (1955) 691.
[37] A. Rybár, K. Antos, Collect. Czech. Chem. Commun. 35 (1970) 1415e1425.
Supplementary data associated with this article can be found in
the online version, at doi:10.1016/j.ejmech.2011.05.023.
ꢂ
[38] G.R. Desiraju, T. Steiner, The Weak Hydrogen Bond in Structural Chemistry
and Biology. Oxford University Press, Oxford, 1999.
[39] G.R. Desiraju, J. Hulliger, Curr. Opin. Solid State Mater. Sci. 5 (2001) 105e115.
[40] G.R. Desiraju, Acc. Chem. Res. 35 (2002) 565e573.
References
[41] R.P. Sijbesma, E.W. Meijer, Chem. Commun. (2003) 5e16.
[42] F.F. Awwadi, R.D. Willett, B. Twamley, J. Mol. Struct. 918 (2009) 116e122.
[43] J.-M. Lehn, Angew. Chem. 27 (1988) 89e112.
[1] B.B. Fredholm, A.P. IJzerman, K.A. Jacobson, K.N. Klotz, J. Linden, Pharmacol.
Rev. 53 (2001) 527e552.
[2] S. Hess, Exp. Opin. Ther. Pat. 11 (2001) 1533e1561.
[44] K.-N. Klotz, I. Hessling, J. Hegler, C. Owman, B. Kull, B.B. Fredholm, M.J. Lohse,
Naunyn-Schmiedeberg’s Arch. Pharmacol. 357 (1998) 1e9.
[45] C.E. Müller, J. Maurinsh, R. Sauer, Eur. J. Pharm. Sci. 10 (2000) 259e265.
[46] X.-D. Ji, K.A. Jacobson, Drug Des. Discov. 16 (1999) 217e226.
[47] T. Borrmann, S. Hinz, D.C.G. Bertarelli, W. Li, N.C. Florin, A.B. Scheiff,
C.E. Müller, J. Med. Chem. 52 (2009) 3994e4006.
[3] L. Yu, H.-Y. Shen, J.E. Coelho, I.M. Araújo, Q.-Y. Huang, Y.-J. Day, N. Rebola,
P.M. Canas, E.K. Rapp, J. Ferrara, D. Taylor, C.E. Müller, J. Linden, R.A. Cunha,
J.-F. Chen, Ann. Neurol. 63 (2008) 338e346.
[4] D. Elmenhorst, P.T. Meyer, O.H. Winz, A. Matusch, J. Ermert, H.H. Coenen,
R. Basheer, H.L. Haas, K. Zilles, A. Bauer, J. Neurosci. 27 (2007) 2410e2415.
[5] A. Bilkei-Gorzo, O.M. Abo-Salem, A.M. Hayallah, K. Michel, C.E. Müller,
A. Zimmer, Naunyn-Schmiedeberg’s Arch. Pharmacol. 377 (2008) 65e76.
[6] O.M. Abo-Salem, A.M. Hayallah, A. Bilkei-Gorzo, B. Filipek, A. Zimmer,
C.E. Müller, J. Pharmacol. Exp. Ther. 308 (2004) 358e366.
[7] B. Schumacher, S. Scholle, J. Hölzl, N. Khudeir, S. Hess, C.E. Müller, J. Nat. Prod.
65 (2002) 1479e1485.
[8] V. Ralevic, G. Burnstock, Pharmacol. Rev. 50 (1998) 413e492.
[9] S.-A. Poulsen, R.J. Quinn, Bioorg. Med. Chem. 6 (1998) 619e641.
[10] J.-L. Moreau, G. Huber, Brain Res. Rev. 31 (1999) 65e82.
[11] C.E. Müller, Drugs Future 25 (2000) 1043e1052.
[48] C.E. Müller, M. Diekmann, M. Thorand, V. Ozola, Bioorg. Med. Chem. Lett. 12
(2002) 501e503.
[49] D.C.G. Bertarelli, M. Diekmann, A.M. Hayallah, D. Rüsing, J. Iqbal, B. Preiss,
E.J. Verspohl, C.E. Müller, Purinergic Signal. 2 (2006) 559e571.
[50] Z.G. Gao, A.P. IJzerman, Biochem. Pharmacol. 60 (2000) 669e676.
[51] P.J. van Galen, A.H. van Bergen, C. Gallo-Rodriguez, N. Melman, M.E. Olah,
A.P. IJzerman, G.L. Stiles, K.A. Jacobson, Mol. Pharmacol. 45 (1994) 1101e1111.
[52] H.J. Kupferberg, Epilepsia 30 (1989) S51eS56.
[53] A. Le Tiran, J.P. Stables, H. Kohn, J. Med. Chem. 45 (2002) 4762e4773.
[54] D. Mulzac, K. Scott, Epilepsia 34 (1993) 1141e1146.
[55] S. McGaraughty, M. Cowart, M.F. Jarvis, R.F. Berman, Curr. Top. Med. Chem. 5
(2005) 43e58.
[12] E. Ongini, P. Schubert, Drug Dev. Res. 45 (1998) 387e393.
[13] F. Impagnatiello, E. Bastia, E. Ongini, A. Monopoli, Emerg. Ther. Targets 4
(2000) 635e663.
[56] CAChe Worksystem Pro 6.1. CAChe Group, Fujitsu Limited, Japan, 2003.
[57] A.K. Ghose, A. Pritchett, G.M. Crippen, J. Comput. Chem. 9 (1988) 80e90.
[58] A. Golbraikh, M. Shen, Z. Xiao, Y.D. Xiao, K.H. Lee, A. Tropsha, J. Comput.-Aided
Mol. Des. 17 (2003) 241e253.
[14] J.A. Ribeiro, A.M. Sebastião, A. de Mendonça, Prog. Neurobiol. 68 (2003) 377e392.
[15] J.D. Salamone, A.J. Betz, K. Ishiwari, J. Felsted, L. Madson, B. Mirante, K. Clark,
L. Font, S. Korbey, T.N. Sager, J. Hockemeyer, C.E. Müller, Front. Biosci. 13
(2008) 3594e3605.
[59] L. He, P.C. Jurs, J. Mol. Graph. Model. 23 (2005) 503e523.
[60] CAChe Worksystem Pro 7.6. CAChe Group, Fujitsu Limited, Japan, 2007.
[61] A. Golbraikh, A. Tropsha, J. Mol. Graph. Model. 20 (2002) 269e276.
[62] G.M. Sheldrick, SHELXTL PCMT. Siemens Analytical X-Ray Instruments Inc.,
Madison, Wisconsin, USA, 1990.
[16] K. Ishiwari, L.J. Madson, A.M. Farrar, S.M. Mingote, J.P. Valenta,
M.D. DiGianvittorio, L.E. Frank, M. Correa, J. Hockemeyer, C.E. Müller,
J.D. Salamone, Behav. Brain Res. 178 (2007) 190e199.
[17] A.M. Farrar, M. Pereira, F. Velasco, J. Hockemeyer, C.E. Müller, J. Salamone,
Psychopharmacology 191 (2007) 579e586.
[63] G.M. Sheldrick, SHELXTL-97. A FORTRAN-77 Program for the Refinement of
Crystal Structures from Diffraction Data. University of Göttingen, Germany,
1997.
[64] L. Yan, D.C.G. Bertarelli, A.M. Hayallah, H. Meyer, K.-N. Klotz, C.E. Müller,
J. Med. Chem. 49 (2006) 4384e4391.
[65] J. Bulicz, D.C.G. Bertarelli, D. Baumert, F. Fülle, C.E. Müller, D. Heber, Bioorg.
Med. Chem. 14 (2006) 2837e2849.
[66] C. Nordstedt, B.B. Fredholm, Anal. Biochem. 189 (1990) 231e234.
[18] C.E. Müller, S. Ferré, Recent Pat. CNS Drug Discov. 2 (2007) 1e21.
[19] (a) O. Yuzlenko, K. Kiec-Kononowicz, Curr. Med. Chem. 13 (2006) 3609e3625;
ꢀ
(b) C.E. Müller, K.A. Jacobson, Biochim. Biophys. Acta. 1808 (2011)
1290e1308.
[20] B.B. Fredholm (Ed.), Methylxanthines, Handbook of Experimental Pharma-
cology, vol. 200, Springer-Verlag, Berlin, Heidelberg, 2011 ISBN: 3642134424.
[21] W.F. Kiesman, J. Zhao, P.R. Conlon, R.C. Petter, X. Jin, G. Smits, F. Lutterodt,
G. Sullivan, J. Linden, Bioorg. Med. Chem. 14 (2006) 3654e3661.