In silico directed chemical probing of the adenosine receptor family

Article abstract of DOI:10.1016/j.bmc.2010.03.048

One of the grand challenges in chemical biology is identifying a small-molecule modulator for each individual function of all human proteins. Instead of targeting one protein at a time, an efficient approach to address this challenge is to target entire protein families by taking advantage of the relatively high levels of chemical promiscuity observed within certain boundaries of sequence phylogeny. We recently developed a computational approach to identifying the potential protein targets of compounds based on their similarity to known bioactive molecules for almost 700 targets. Here, we describe the direct identification of novel antagonists for all four adenosine receptor subtypes by applying our virtual profiling approach to a unique synthesis-driven chemical collection composed of 482 biologically-orphan molecules. These results illustrate the potential role of in silico target profiling to guide efficiently screening campaigns directed to discover new chemical probes for all members of a protein family.

Full text of DOI:10.1016/j.bmc.2010.03.048

Bioorganic & Medicinal Chemistry 18 (2010) 3043–3052
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry
journal homepage: www.elsevier.com/locate/bmc
In silico directed chemical probing of the adenosine receptor family
Filipe M. Areias a,b, Jose Brea b, Elisabet Gregori-Puigjané c, Magdi E. A. Zaki a, M. Alice Carvalho a,
Eduardo Domínguez b, Hugo Gutiérrez-de-Terán b, M. Fernanda Proença a, María I. Loza b, Jordi Mestres c,
*
a Center of Chemistry, Campus de Gualtar, Universidade do Minho, 4710-057 Braga, Portugal
b Department of Pharmacology, Universidade de Santiago de Compostela, 15782 Santiago de Compostela, Spain
c Chemotargets SL and Chemogenomics Laboratory, Research Unit on Biomedical Informatics, Institut Municipal d’Investigació Mèdica and Universitat Pompeu Fabra, 08003 Barcelona,
Catalonia, Spain
a r t i c l e i n f o
a b s t r a c t
Article history:
One of the grand challenges in chemical biology is identifying a small-molecule modulator for each indi-
vidual function of all human proteins. Instead of targeting one protein at a time, an efficient approach to
address this challenge is to target entire protein families by taking advantage of the relatively high levels
of chemical promiscuity observed within certain boundaries of sequence phylogeny. We recently devel-
oped a computational approach to identifying the potential protein targets of compounds based on their
similarity to known bioactive molecules for almost 700 targets. Here, we describe the direct identification
of novel antagonists for all four adenosine receptor subtypes by applying our virtual profiling approach to
a unique synthesis-driven chemical collection composed of 482 biologically-orphan molecules. These
results illustrate the potential role of in silico target profiling to guide efficiently screening campaigns
directed to discover new chemical probes for all members of a protein family.
Received 12 January 2010
Revised 12 March 2010
Accepted 20 March 2010
Available online 27 March 2010
Keywords:
Target profiling
Adenosine antagonists
Chemogenomics
Computational chemical biology
Ó 2010 Elsevier Ltd. All rights reserved.
1. Introduction
graphical sources and making them accessible also in the public
domain. Additionally, a growing number of entries containing a li-
The identification of the complete list of proteins to which small
molecules could potentially have affinity is a cornerstone in the use
of chemistry to probe biology, with important implications for
modern drug discovery.1 The compilation of large compound col-
lections and the implementation of high-throughput screening, in
both industry and academia, have increased dramatically our
capacity to probe the chemical space for a single target. However,
our capacity to probe the biological space of a single molecule is
still limited, mainly due to the huge logistics involved in the sys-
tematic screening of molecules on a large panel of in vitro assays
which, until recently, made these activities feasible only within
the privacy of biotechnology and pharmaceutical industry.2–4 In
this respect, the launch of global coordinated initiatives, such as
the NIH Molecular Libraries Screening Center Network (MLSCN)5
and the Psychoactive Drug Screening Program (PDSP),6 opens an
avenue towards the consistent generation of screening data for
molecules and the ultimate deposition of all chemical and biolog-
ical data in public databases. In parallel, several informatics initia-
tives, such as DrugBank,7 IUPHAR-DB,8 and BindingDB,9 are
complementing these experimental screening programs by collect-
ing all pharmacological data being published in multiple biblio-
gand forming a complex inside a protein cavity are being deposited
in the Protein Data Bank,10 the major public repository for protein
structure information.
The availability of an increasing amount of protein–ligand inter-
action data has promoted the development of a variety of computa-
tional methods aiming at predicting the pharmacological profile of
compounds,11–13 and thus offering a perfect means to expand in sil-
ico our capacity to probe the biological space of molecules. Depend-
ing on the source of information being exploited, target profiling
methods can be divided into ligand-based and structure-based
methods: ligand-based methods rely essentially on comparing a tar-
get compound to a database of hundreds of thousands of chemical
structures with known targets, whereas structure-based methods
require three-dimensional information of the protein binding cavity
to dock compounds in or assess their fitness relative to the exposed
pharmacophoric features. Ligand-based target profiling methods
were successfully applied recently to several drugs to detect addi-
tional proteins other than their recognized primary targets14 and
to predict the mechanism of action of antimalarials discovered in a
high-throughput cell-based screen.15 Likewise, structure-based tar-
get profiling methods allowed also for identifying a true target for
some compounds in a scaffold-focused library16 and to discover
three targets associated with constituents of a medicinal plant.17
Here, we report the application of our own implementation of a
ligand-based approach to in silico pharmacological profiling18 to
* Corresponding author.
E-mail address: jmestres@imim.es (J. Mestres).
0968-0896/$ - see front matter Ó 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmc.2010.03.048

3044
F. M. Areias et al. / Bioorg. Med. Chem. 18 (2010) 3043–3052
identify the potential biological targets of a synthesis-driven com-
pound collection and its subsequent use to complete the chemical
probing of an entire protein family.
A2B, and A3 subtypes transfected cell lines (Table 2). Screening data
confirmed the existence of two hits with specific binding displace-
ments greater than 50% at the A2B and A3 adenosine receptors.
These results allow for getting a fair assessment of the overall per-
formance of the predictions made by in silico profiling. In this
respect, it can be observed that not all compound–receptor interac-
tions predicted to be active could be confirmed (2/15) and likewise
not all inactive interactions were correctly anticipated (25/38).
However, it is worth stressing that all 25 interactions predicted
to be inactive were ultimately confirmed and, most importantly,
the receptor affinities for the two hits identified were correctly
predicted.
The structures of the two hits obtained from the primary screen
for the A2B and A3 receptors are presented in Figure 1a. An analysis
of these chemotypes against all bioactive compounds for the aden-
osine receptors present in the annotated chemical libraries used in
this work confirmed the structural novelty of the hits. For the sake
of clarity, the compounds will be referred to henceforth as MSB-
A2B and MSB-A3. Further confirmation of the affinity and func-
tional activity of the two compounds for their respective receptors
is also included in Figure 1. Concentration-response binding com-
petition curves revealed that both compounds had affinity values
around the micromolar range, MSB-A2b and MSB-A3 showing pKi
values of 5.8 0.3 and 6.5 0.1 at the A2B and A3 receptors, respec-
tively (Fig. 1b). Furthermore, the agonist/antagonist behaviour of
these compounds was examined by measuring their modulation
of NECA-dependent intracellular cAMP formation, which con-
firmed that the two compounds are antagonists of the A2B and A3
receptors with pKb values of 5.7 0.1 and 6.3 0.4 for MSB-A2b
and MSB-A3, respectively (Fig. 1c).
2. Results and discussion
2.1. Composition of the chemical library
Chemistry plays a fundamental role in our quest to probe and
understand the biology of proteins and thus access to novel struc-
tural cores covering unexplored areas of chemical space is of utmost
importance. Accordingly, our source of chemical matter was a un-
ique collection composed of 482 biologically-orphan molecules syn-
thesized and stored over the last few years at the Centre of
Chemistry of the Universidade do Minho. The structural features of
those molecules contained mainly aromatic heterocycles decorated
with phenolic substituents obtained from commercially available
starting materials, usually in multi-step processes, using a synthetic
approach based primarily on methods developed internally for the
synthesis of highly functionalised nitrogen heterocycles, which are
difficult and costly to obtain otherwise. The exclusivity of the collec-
tion is reflected by the fact that only 11 of those 482 molecules were
found available in the catalogues of chemical providers. The main
core structures of this collection are compiled in Table 1. Substituted
purines, including 8-oxopurines and imidazo-pyridines (S1, S2 and
S3), correspond to approximately 43% of the compounds in the
collection. Substituted imidazoles, 2-oxoimidazoles, imidazolyl-
imidazolonesand imidazo-imidazoles (S4, S5, S6 and S7) correspond
to 24%. Pyrimido-pyrimidines and pyrido-pyrimidines (S8, S9 and
S10) were generated by an ANRORC type rearrangement from some
of the purine and imidazo-pyridine structures and correspond to
10%. A number of diaminomaleonitrile derivatives (S11) were also
prepared, some of them were used as linear synthetic precursors
of the cyclic compounds used in this study, and represent an
additional 8% of the collection. The remaining 15% contain a
diverse range of nitrogen heterocycles, including fused tricyclic
structures.
2.4. Complete probing of the adenosine receptor family
At this stage, our aim of chemically probing the entire adeno-
sine receptor family had partially failed, since the primary screen
of the compounds selected from our chemical collection did not
deliver hits on the A1 and A2A receptors. Therefore, given the struc-
tural novelty of the hits identified for the A2B and A3 receptors, we
decided to proceed by taking those structures as templates for a
similarity-based virtual screening of the chemical catalogue com-
posed of 7.5 million compounds available from commercial ven-
dors using a set of low-dimensional descriptors that reflect the
overall distribution of pharmacophoric features in a molecule
(Supplementary Fig. 1).22 A total of 9622 compounds were re-
trieved within a similarity cut-off of any of the two hits. In order
to reduce the numbers further, those compounds were then pro-
filed in silico against all four adenosine receptors and, to maximize
the chances of identifying hits, those that could not be annotated to
both A1 and A2A receptors were filtered out. The remaining 283
compounds were visually inspected and a set of eight compounds
was finally agreed to be selected for purchase and testing based on
their arrangement of the main pharmacophore features relative to
those present in the most similar bioactive compounds to A1 and
A2A. Complete in vitro screening of the eight molecules against
the A1 and A2A receptors was carried out at the same conditions
as described earlier. Screening data ultimately confirmed the exis-
tence of five hits with specific binding displacements greater than
50% at the A1 and/or A2A receptors (Table 3).
2.2. In silico identification of putative targets
Given its structural composition, we decided to start profiling in
silico the compound collection against a panel of 86 diverse G pro-
tein-coupled receptors (GPCRs) for which ligand-based models
were available. Of the 482 molecules processed, only 23 received
an annotation to at least one GPCR target. In total, 37 annotations
were assigned to nine targets (Supplementary Table 1). Remark-
ably, all four adenosine receptors were included in the target hit
list, collectively gathering 15 of those annotations from 10 differ-
ent compounds. We saw in this outcome an opportunity to attempt
for the first time the identification of novel hits for all members of
a protein family by means of in silico profiling. Focus on adenosine
receptors was also justified by the fact that they constitute a family
of utmost biological importance and broad therapeutic relevance
in cardiovascular, inflammatory, immune, and neurodegenerative
diseases.19,20 In addition, many of the adenosine receptor leads re-
ported to date are xanthine or xanthine-like derivatives,21 a core
structure known to have issues with solubility and poor overall
drug-like profile. Therefore, the discovery of novel chemical scaf-
folds in potent and selective bioactive compounds for adenosine
receptors remains critical for a wide range of therapeutic areas.
In agreement with the predicted in silico target profiling, three
of the hits showed binding displacements greater than 50% for
both A1 and A2A receptors, but most interestingly some degree of
selectivity for A1 and A2A, respectively, was unexpectedly observed
for the other two hits (Table 3). The structures of the two most
selective hits obtained from the primary screen for the A1 and
A2A receptors are presented in Figure 2a, which will be referred
henceforth as MSB-A1 and MSB-A2a. Once more, an analysis of
2.3. In vitro screening against adenosine receptors
Complete in vitro screening of the 10 molecules against all four
adenosine receptors was carried out at 10
lM concentration for
competitive binding of specific radioligands to human A1, A2A
,

F. M. Areias et al. / Bioorg. Med. Chem. 18 (2010) 3043–3052
3045
Table 1
Main core structures and substitution patterns of the chemical library processed by in silico target profiling
Label
Core structure
Substitution pattern
R = alkyl, aryl (including hydroxyphenyl)
R
N
N
R3
R
R
R
1 = H, aryl
2 = amidino, amidrazono, oxazolidinyl, dialkylamino (including cyclic amines)
3 = H, aryl (including hydroxyphenyl)
R1
S1
N
N
R2
R
N
R = alkyl
R1
R
1 = H, aryl
N
R
2 = H, cyano, amidino, triazolyl
O
R
S2
S3
N
N
H
R2
R = aryl
N
R1
R2
R
R
R
1 = phenyl, indolyl, amino, hydroxyl
2 = cyano, sulfonyl
N
3 = amino, cyano
N
R3
R = H, alkyl, aryl,
R
N
R3
R2
R
R
R
1 = H, aryl
2 = cyano, carbamoyl, cyanoformimidoyl, alkoxyformamidino
3 = cyano, alkoxyalkylideneamine, amino, arylmethyleneamino, alkoxyformamidino
R1
O
S4
S5
N
R = alkyl
1 = arylaminomethyleneamino, arylmethyleneamino, alkoxymethyleneamino
R
N
NH
R1
R
N
H
NC
NR1
R
N
R = alkyl
R
R
R
1 = H, arylamino
2 = H, methyl
O
3 = aryl, amino, arylamino, alkyl (including hydroxyalkyl)
N
N
S6
S7
R2
N
H2N
R3
R = alkyl, aryl (including hydroxyphenyl)
1 = hydroxyphenyl
R
N
NH
R
N
CO2NH
R1
N
N
R = aryl
R
HN
R
R
1 = H, acylamino, benzyloxy, hydroxyl
2 = H, benzyloxy
N
S8
S9
N
NR2
N
R1
R = aryl
1 = acylhydrazino
R
HN
R
N
N
N
N
NHR1
N
R = aryl
1 = alkyl (including hydroxyalkyl)
R
HN
R
Ph
S10
S11
N
CN
NH
N
R1
R, R1 = amino, arylaminomethyleneamino, arylmethyleneamino, alkoxyalkylideneamine, ureylene
2 = cyano, carbamoyl
R
CN
R2
R
R1
their chemotypes against all bioactive compounds for adenosine
receptors present in the annotated chemical libraries used in this
work confirmed the structural novelty of the hits. Figure 2 includes
also further confirmation of the affinity and functional activity of
the two compounds for their respective receptors. Concentration-
response binding competition curves (Fig. 2b) returned pKi values
of 7.8 0.1 and 5.6 0.1 at the A1 and A2A receptors for MSB-A1
and MSB-A2a, respectively, and measurements of the modulation
of NECA-dependent intracellular cAMP formation by MSB-A1 and
MSB-A2a (Fig. 2c) confirmed their antagonist behaviour with pKb

3046
F. M. Areias et al. / Bioorg. Med. Chem. 18 (2010) 3043–3052
Table 2
In vitro screening of in silico-selected compounds on each adenosine receptor subtype

F. M. Areias et al. / Bioorg. Med. Chem. 18 (2010) 3043–3052
3047
Table 2 (continued)
The 15 predictions made in silico are marked in light-grey. The two predictions with experimental values over 50% displacement of specific radioligand binding at 10
highlighted in dark-grey.
l
M are
MSB-A2b
MSB-A3
HO
N
OH
N
N
N
N
O
N
NH
N
H
H2N
O
2500
2000
1500
1000
500
3000
2000
1000
0
pKi= 5.8 0.3
pKi= 6.5 0.1
0
-9
-8
-7
-6
-5
-4
-3
-9
-8
-7
-6
-5
-4
-3
Log Concentration (M)
Log Concentration (M)
4000
3000
2000
1000
0
2000
1500
1000
500
pKb= 5.7 0.1
pKb= 6.3 0.4
-10
-9
-8
-7
-6
-5
-4
-3
-10
-9
-8
-7
-6
-5
-4
-3
log Concentration (M)
log Concentration (M)
Figure 1. (a) Structures of the two hits for the (left) A2B and (right) A3 receptors; (b) binding competition curves of (left) MSB-A2b (j) against 25 nM [3H]DPCPX at human A2B
receptors, using 100 M R-PIA for
M NECA for unmasking non-specific binding (N) and (right) MSB-A3 (j) against 30 nM [3H]NECA at human A3 receptors, using 100
M NECA-induced cAMP accumulation at human A2B receptors and
M forskolin-stimulated human A3 receptors. Points represent the mean S.E.M. (vertical bars) of two
l
l
unmasking non-specific binding (N); and (c) modulation of cAMP formation of (left) MSB-A2b (j) over 10
(right) MSB-A3 (j) over 10 lM NECA-induced cAMP decrease of 10 l
l
separate experiments.

3048
F. M. Areias et al. / Bioorg. Med. Chem. 18 (2010) 3043–3052
Table 3
properties, all hits are found to be within the boundaries of the
rule-of-five.23 With respect to the binding profile, all antagonists
were found to be reasonably selective for their corresponding
adenosine receptors. Only MSB-A2a showed similar affinity values
for the A2A and A3 receptors. Considering the fact that no optimisa-
tion activities on the binding and selectivity of the four antagonists
took place, this is a remarkable outcome. In addition, little is
known in the literature about the polypharmacology of adenosine
receptor antagonists.21 To further assess the degree of selectivity of
those hits beyond its own protein family, they were tested in pri-
mary screens against three diverse amine GPCRs, namely, the
In vitro screening of the compounds purchased from commercial vendors on the A1
and A2A adenosine receptors
adrenergic
tors. The results show that the percentage of displacement of radi-
oligand binding at 10 M for those amine GPCRs is in all cases
a2A, the histamine H1, and the serotonin 5-HT7 recep-
l
below 50%. In fact, only for the interaction between antagonists
MSB-A1 and MSB-A2a and the H1 receptor values close to 40%
are obtained, whereas all other interactions show values below
30%.
3. Conclusions
In silico target profiling methods are emerging as efficient alter-
natives to the currently unaffordable high-throughput in vitro tar-
get profiling of millions of compounds over hundreds of biological
targets. All these methods capitalize on the vast amount of prior
knowledge on bioactive ligands and protein structures available
for many targets of therapeutic relevance,12 having led recently
to the successful identification of novel targets for known drugs.17
As both pharmacological and structural data on ligand–target
interactions are becoming increasingly available, it is expected that
the performance of these knowledge-based methods will improve
significantly reaching enough maturity to be able to not only iden-
tify novel targets but to characterize the entire pharmacological
profile of compounds. In addition, exploration of the biological
space for molecules has been traditionally focussed on commer-
cially available compounds and target-directed libraries from
internal corporate projects, which represent just a minute portion
of the potentially synthetically accessible chemical space. How-
ever, large numbers of synthesis-driven compounds, often covering
novel areas of chemical space, are sitting on the shelves of aca-
demic chemistry laboratories without ever being considered for
biological screening. In this respect, the development of novel in
silico methods for target profiling might help directing the biolog-
ical space relevant for each chemical collection and thus promoting
and strengthening collaborative efforts between chemists and biol-
ogists.24 A good example has been presented here, leading to the
identification of novel antagonists covering all members of the
adenosine receptor family. This is, to the best of our knowledge,
the first time that in silico directed chemical probing of an entire
protein family has been attempted and successfully achieved.
4. Materials and methods
4.1. In silico target profiling
Our in silico target profiling approach relies on the assumption
that the set of bioactive ligands collected for a given target pro-
vides a complementary description of the target from a ligand per-
spective. In order to be able to process this information efficiently,
molecular structures need to be encoded using some sort of math-
ematical descriptors. In this respect, a novel set of low-dimension
molecular descriptors called SHED were used.22 SHED are derived
from distributions of atom-centered feature pairs extracted
directly from the topology of molecules. The collection of SHED
values reflecting the overall distribution of pharmacophoric fea-
tures in a molecule constitutes its SHED profile. The ensemble of
The eight predictions with experimental values over 50% displacement of specific
radioligand binding at 10 lM are highlighted in grey.
values of 6.8 0.2 and 5.1 0.3 at the A1 and A2A receptors,
respectively.
Finally, the property and binding profiles of the four antagonists
identified are summarized in Table 4. In terms of molecular

F. M. Areias et al. / Bioorg. Med. Chem. 18 (2010) 3043–3052
3049
MSB-A1
MSB-A2a
O
N
H
N
N
O
N
N
N
N
O
HO
3000
2000
1000
0
3000
2000
1000
0
pKi= 7.8 0.1
pKi= 5.6 0.1
-10
-9
-8
-7
-6
-5
-4
-3
-10
-9
-8
-7
-6
-5
-4
-3
Log Concentration (M)
Log Concentration (M)
750
500
250
0
3000
2000
1000
0
pKi= 6.8 0.2
pKi= 5.1 0.3
-10 -9
-8
-7
-6
-5
-4
-3
-2
-10
-9
-8
-7
-6
-5
-4
-3
log Concentration (M)
log Concentration (M)
Figure 2. (a) Structures of the two hits for the (left) A1 and (right) A2A receptors; (b) binding competition curves of (left) MSB-A1 (j) against 1.4 nM [3H]DPCPX at human A1
receptors, using 10 M NECA for
M R-PIA for unmasking non-specific binding (N) and (right) MSB-A2a (j) against 3 nM [3H]ZM241385 at human A2A receptors, using 50
M NECA-induced cAMP decrease of 10 M forskolin-stimulated
M NECA-induced cAMP accumulation at human A2A receptors. Points represent the mean S.E.M. (vertical bars) of two
l
l
unmasking non-specific binding (N); and (c) modulation of cAMP formation of (left) MSB-A1 (j) over 0.1
l
l
human A1 receptors and (right) MSB-A2a (j) over 1
l
separate experiments.
Table 4
Property and binding profiles of the novel antagonists identified for each one of the four adenosine receptors
O
N
H
N
N
HO
N
O
N
N
N
N
N
O
OH
N
N
N
O
N
N
H
NH
HO
H2N
O
MSB-A1
MSB-A2a
MSB-A2b
MSB-A3
Property profile
MW
clogP
HBA
HBD
361.4
4.58
3
346.4
4.45
3
350.4
1.58
4
316.4
4.44
3
1
1
4
1
Binding profile
A1
7.8 0.1
11%
5%
39%
14%
39%
5.6 0.1
19%
5.5 0.2
12%
39%
27%
15%
5.8 0.3
26%
28%
17%
38%
20%
3%
6.5 0.1
18%
A2A
A2B
A3
a2A
H1
42%
17%
5-HT7
20%
18%
11%
1%

3050
F. M. Areias et al. / Bioorg. Med. Chem. 18 (2010) 3043–3052
SHED profiles representing all bioactive molecules annotated to a
particular protein target forms a mathematical description of the
target from a chemical perspective. On this basis, ligand-based
descriptor models were derived for 86 GPCRs using the interaction
data present in BindingDB9 and PDSP,25 supplemented with an
additional GPCR annotated chemical library assembled internally.
The probability of any molecule to interact with a particular target
is assumed to be related to the degree of similarity relative to the
set of known bioactive ligands for that target. Accordingly, Euclid-
ean distances between the SHED profile of a molecule and all SHED
profiles associated to a target are first calculated and a final target
scoring is given by the minimum value of all Euclidean distances.
Following a previous validation analysis,26 if that minimum Euclid-
ean distance is below 0.6, the molecule receives an annotation to
the target.
characterization of MSB-A3 yielded the following: 1H NMR
(300 MHz, DMSO-d6): d 11.48 (broad s, 1H), 8.31 (s, 1H), 8.20 (d,
2H, J = 8.2 Hz), 7.40 (d, 2H, J = 7.2 Hz), 7.32 (t, 2H, J = 8.0 Hz), 7.26
(brd, 2H, J = 8.2 Hz), 5.05 (s, 2H), 2.33 (s, 3H). 13C NMR (75 MHz,
DMSO-d6): d 21.01, 42.58, 120.48, 127.13, 127.71, 127.85, 128.71,
129.24, 133.28, 134.99, 136.70, 139.54, 150.34, 153.36, 156.23.
Anal. Calcd C19 H16N4ꢀO. 0.2H2O requires: C, 71.32; H, 5.17; N,
17.51. Found: C, 71.20; H, 5.38; N, 17.57.
9-Benzyl-2-(4-methylphenyl)-7,9-dihydro-8H-purin-8-one,
used as starting material, was obtained from the reaction of diami-
nomaleonitrile with benzylisocyanate, followed by addition of 4-
tolualdehyde and DBU, in three consecutive steps.28
4.3. Radioligand binding assays
Compounds were assayed at the concentration of 10 lM at all
4.2. Chemical synthesis and characterization
receptors following the conditions stated above. Concentration-re-
sponse binding competition curves at all adenosine receptors were
carried out by assaying six different concentrations (range be-
4.2.1. Compound MSB-A1
5-Ethyl-1-hydroxy-3-(3-(vinyloxy)phenyl)imidazo[5,1-a]phthal-
azin-6(5H)-one was purchased from Enamine Ltd with guaranteed
purity of 96%.
tween 10 nm to 100
MSB-A2b and MSB-A3. The ꢁlog of inhibition constant (pKi) of each
compound was calculated by the Cheng–Prusoff equation, Ki = IC50
lM) of the compounds MSB-A1, MSB-A2a,
/
(1 + [L]/KD), where IC50 is the concentration of compound that dis-
places the binding of radioligand by 50%, [L] is the free concentra-
tion of radioligand and KD is the dissociation constant of each
radioligand. IC50 values were obtained by fitting the data with
non-linear regression, with Prism 2.1 software (GraphPad, San Die-
go, CA).
4.2.2. Compound MSB-A2a
6,7,8,9-Tetrahydro-N-(pyridin-2-yl)-3-p-tolyl-5H-imidazo[1,5-
a]azepine-1-carboxamide was purchased from ChemDiv Inc. with
guaranteed purity of 95%.
4.2.3. Compound MSB-A2b
(1-Carbamoyl-7-imino-6,7-dihydro-3-(20-hydroxyphenyl)-6-
(400-hydroxyphenyl)-imidazo[1,5-e]imidazole) was synthesized
using the following experimental procedure: Triethylamine
(0.37 mL, 2.74 mmol) was added to a solution of N1-(2-amino-
4.3.1. Human A1 receptors
Adenosine A1 receptor competition binding experiments were
carried out in membranes from CHO-A1 cells (Euroscreen, Gosse-
lies, Belgium). On the day of assay, membranes were defrosted
and re-suspended in incubation buffer 20 mM Hepes, 100 mM
NaCl, 10 mM MgCl2, 2 units/ml adenosine deaminase (pH 7.4).
Each reaction well of a GF/C multiscreen plate (Millipore, Madrid,
Spain), prepared in duplicate, contained 15
[3H]DPCPX and test compound. Non-specific binding was deter-
mined in the presence of 10 M (R)-PIA. The reaction mixture
1,2-dicyanovinyl)-N2-(4’-hydroxyphenyl)formamidine
(0.10 g,
0.44 mmol) and 2-hydroxybenzaldehyde (0.05 mL, 0.47 mmol) in
ethyl acetate (3 mL) and ethanol (0.3 mL). The mixture was stirred
at 0 to 4 °C, under a nitrogen atmosphere, until the starting mate-
rial was totally consumed (17 days). The cream solid suspension
was filtered and washed with ethyl acetate and diethyl ether.
The solid was recrystallized from acetone, leading to an analyti-
cally pure sample of the product (0.05 g, 0.15 mmol, 34%). Spectro-
scopic characterization of MSB-A2b yielded the following: 1H NMR
(300 MHz, DMSO-d6): d 11.38 (s, 1H), 9.33 (s, 1H), 9.08 (s, 1H), 8.27
(s, 1H), 7.82 (d, 2H, J = 8.4 Hz), 7.69 (d, 1H, J = 6.9 Hz), 7.63 (s, 1H),
7.39 (t, 1H, J = 7.2 Hz), 7.04 (d, 1H, J = 8.1 Hz), 6.93–7.00 (m, 1H),
6.80 (d, 2H, J = 8.4 Hz), 6.18 (s, 2H). 13C NMR (75 MHz, DMSO-
d6): d 64.74, 112.50, 115.22, 117.14, 119.38, 120.62, 126.94,
128.82, 131.17, 131.40, 132.86, 139.87, 150.72, 153.34, 156.35,
163.46. HRMS (m/z): calcd C18H15N5O3 requires MH+, 350.1208;
found MH+, 350.1249.
lg of protein, 2 nM
l
was incubated at 25 °C for 60 min, after which samples were fil-
tered and measured in a microplate beta scintillation counter
(Microbeta Trilux, Perkin Elmer, Madrid, Spain).
4.3.2. Human A2A receptors
Adenosine A2A receptor competition binding experiments were
carried out in membranes from HeLa-A2A cells. On the day of assay,
membranes were defrosted and re-suspended in incubation buffer
50 mM Tris–HCl, 1 mM EDTA, 10 mM MgCl2 and 2 UI/mL adeno-
sine deaminase (pH 7.4). Each reaction well of a GF/C multiscreen
plate (Millipore, Madrid, Spain), prepared in duplicate, contained
N1-(2-Amino-1,2-dicyanovinyl)-N2-(40-hydroxyphenyl)form-
amidine, used as starting material, was obtained from the reaction
of ethyl-N-(2-amino-1,2-dicyanovinyl)formimidate and 2-hydrox-
yaniline, using a synthetic method previously described for the
preparation of analogous compounds.27
10
C0036E08. Non-specific binding was determined in the presence
of 50 M NECA. The reaction mixture was incubated at 25 °C for
lg
of protein, 3 nM [3H]ZM241385 and test compound
l
30 min, after which samples were filtered and measured in a
microplate beta scintillation counter (Microbeta Trilux, Perkin El-
mer, Madrid, Spain).
4.2.4. Compound MSB-A3
(9-Benzyl-2-(4-methylphenyl)-7,9-dihydro-8H-purin-8-one)
was synthesized through the following method: piperidine (0.03 g,
0.35 mmol) was added to a solution of 2-(1-benzyl-5-imino-2-oxo-
imidazolidin-4-ylidene)-3-(phenylmethyleneamino)propanenitrile
(0.12 g, 0.35 mmol) in ethanol (20 mL) and the mixture was stirred
at room temperature for 5 days. The solvent was partially removed
in the rotary evaporator leading to a white solid. The product was
filtered, washed with diethyl ether and ethanol and identified as
the title compound (0.10 g, 0.32 mmol, 91%). Spectroscopic
4.3.3. Human A2B receptors
Adenosine A2B receptor competition binding experiments were
carried out in membranes from HEK-293-A2B cells (Euroscreen,
Gosselies, Belgium) prepared following the provider’s protocol.
On the day of assay, membranes were defrosted and re-suspended
in incubation buffer 50 mM Tris–HCl, 1 mM EDTA, 10 mM MgCl2,
0.1 mM benzamidine, 10
deaminase (pH 6.5). Each reaction well prepared in duplicate, con-
tained 18
g of protein, 35 nM [3H]DPCPX and test compound.
lg/ml bacitracine and 2 UI/mL adenosine
l

F. M. Areias et al. / Bioorg. Med. Chem. 18 (2010) 3043–3052
3051
Non-specific binding was determined in the presence of 400
l
M
A3 receptors are Gi/o-coupled receptors and therefore competition
curves were performed over the NECA-induced inhibition of for-
skolin-induced cAMP accumulation.
NECA. The reaction mixture was incubated at 25 °C for 30 min,
after which samples were filtered through a multiscreen GF/C
microplate and measured in a microplate beta scintillation counter
(Microbeta Trilux, Perkin Elmer, Madrid, Spain).
The antagonist potency was expressed as pKB (ꢁlog of the dis-
sociation constant, KB), calculated for one concentration of antago-
nist following the equation proposed by Leff and Dougall, KB = IC50
/
4.3.4. Human A3 receptors
((2 + ([A]/[A50])n)1/n ꢁ 1), where IC50 is the concentration of antag-
onist that inhibits receptor activation in a 50%, [A] is the agonist
concentration used to stimulate the receptor under study, [A50] is
the concentration of agonist that elicits a half-maximal stimulation
of the receptor and n is the slope of the antagonist concentration-
response curve.
Adenosine A3 receptor competition binding experiments were
carried out in membranes from HeLa-A3 cells. On the day of assay,
membranes were defrosted and re-suspended in incubation buffer
50 mM Tris–HCl, 1 mM EDTA, 5 mM MgCl2 and 2 UI/mL adenosine
deaminase (pH 7.4). Each reaction well of a GF/B multiscreen plate
(Millipore, Madrid, Spain), prepared in triplicate, contained 90
of protein, 30 nM [3H]NECA and test compound. Non-specific bind-
ing was determined in the presence of 100 M (R)-PIA. The reac-
tion mixture was incubated at 25 °C for 180 min, after which
samples were filtered and measured in a microplate beta scintilla-
tion counter (Microbeta Trilux, Perkin Elmer, Madrid, Spain).
lg
4.4.1. Human A1 receptors
l
CHO-A1 cells were seeded (20,000 cells/well) in 96-well culture
plates and incubated at 37 °C in an atmosphere with 5% CO2 in Ea-
gle’s Medium Nutrient Mixture F-12 (EMEM F-12), containing 10%
Foetal Calf Serum (FCS) and 1%
with 200 l assay medium (EMEM-F12 and 25 mM HEPES pH 7.4)
and pre-incubated with assay medium containing 20 M rolipram
and test compounds at 37 °C for 15 min. 0.1 M NECA was incu-
bated for 15 min at 37 °C and 3 M forskolin was incubated for
L
-Glutamine. Cells were washed 3ꢂ
l
4.3.5. Human a2A receptors
l
Adrenergic
a
2A receptor competition binding experiments were
l
carried out in membranes from CHO-
a2A cells. On the day of assay,
l
membranes were defrosted and re-suspended in incubation buffer
25 mM Na2PO4 (pH 7.4). Each reaction well of a GF/C multiscreen
plate (Millipore, Madrid, Spain), prepared in duplicate, contained
3 min (total incubation time 30 min). Reaction was stopped with
lysis buffer supplied in the kit and the enzyme-immunoassay
was carried out for detection of intracellular cAMP at 450 nm in
an Ultra Evolution detector (Tecan). Data were fitted by non-linear
regression using GraphPad Prism v2.01 (GraphPad Software).
25
specific binding was determined in the presence of 100
l
g of protein, 0.37 nM [3H]MK-912 and test compound. Non-
l
M norepi-
nephrine. The reaction mixture was incubated at 25 °C for 30 min,
after which samples were filtered and measured in a microplate
beta scintillation counter (Microbeta Trilux, Perkin Elmer, Madrid,
Spain).
4.4.2. Human A2A receptors
CHO-A2A cells were seeded (10,000 cells/well) in 96-well cul-
ture plates and incubated at 37 °C in an atmosphere with 5% CO2
in Dulbecco’s Modified Eagle’s Medium Nutrient Mixture F-12
4.3.6. Human H1 receptors
(DMEM F-12), containing 10% Foetal Calf Serum (FCS) and 1%
Glutamine. Cells were washed 3ꢂ with 200 l assay medium
(DMEM-F12 and 25 mM HEPES pH 7.4) and pre-incubated with as-
say medium containing 30 M rolipram and test compounds at
37 °C for 15 min. 1 M NECA was incubated for 15 min at 37 °C (to-
tal incubation time 30 min). Reaction was stopped with lysis buffer
supplied in the kit and the enzyme-immunoassay was carried out
for detection of intracellular cAMP at 450 nm in an Ultra Evolution
detector (Tecan). Data were fitted by non-linear regression using
GraphPad Prism v2.01 (GraphPad Software).
L-
Histamine H1 receptor competition binding experiments were
carried out in membranes from CHO-H1 cells. On the day of assay,
membranes were defrosted and re-suspended in incubation buffer
50 mM Tris–HCl (pH 7.4). Each tube prepared in duplicate, con-
l
l
l
tained 1
l
g of protein, 2 nM [3H]pyrilamine and test compound.
Non-specific binding was determined in the presence of 10
l
M tri-
prolidine. The reaction mixture was incubated at 27 °C for 60 min,
after which samples were filtered through GF/C filters using a
Brandel Harvester and measured in a beta scintillation counter
(Beckman LS600 LL, Beckman Coulter, Madrid, Spain).
4.4.3. Human A2B receptors
4.3.7. Human 5-HT7 receptors
HEK-293 cells were seeded (10,000 cells/well) in 96-well cul-
ture plates and incubated at 37 °C in an atmosphere with 5% CO2
in Eagle’s Medium Nutrient Mixture F-12 (EMEM F-12), containing
Serotonin 5-HT7 receptor competition binding experiments
were carried out in membranes from HEK-293-5-HT7 cells. On
the day of assay, membranes were defrosted and re-suspended in
incubation buffer 50 mM Tris–HCl, 4 mM CaCl2, 1 mM ascorbic
acid, 0.1 mM pargyline (pH 7.4). Each reaction well of a GF/C mul-
tiscreen plate (Millipore, Madrid, Spain), prepared in duplicate,
10% Foetal Calf Serum (FCS) and 1%
washed 3ꢂ with 200 l assay medium (EMEM-F12 and 25 mM
HEPES pH 7.4) and pre-incubated with assay medium containing
30 M rolipram and test compounds at 37 °C for 15 min. 10
L-Glutamine. Cells were
l
l
lM
contained 5
l
g of protein, 2 nM [3H]SB269970 and test compound.
M clo-
NECA was incubated for 15 min at 37 °C (total incubation time
30 min). Reaction was stopped with lysis buffer supplied in the
kit and the enzyme-immunoassay was carried out for detection
of intracellular cAMP at 450 nm in an Ultra Evolution detector (Te-
can). Data were fitted by non-linear regression using GraphPad
Prism v2.01 (GraphPad Software).
Non-specific binding was determined in the presence of 25
l
zapine. The reaction mixture was incubated at 37 °C for 60 min,
after which samples were filtered and measured in a microplate
beta scintillation counter (Microbeta Trilux, Perkin Elmer, Madrid,
Spain).
4.4. cAMP prodution measurement
4.4.4. Human A3 receptors
CHO-A3 cells were seeded (20,000 cells/well) in 96-well culture
plates and incubated at 37 °C in an atmosphere with 5% CO2 Dul-
becco’s Modified Eagle’s Medium Nutrient Mixture F-12 (DMEM
These assays were performed at adenosine receptors transfec-
ted using a cAMP enzyme-immunoassay kit (Amersham Biosci-
ences) and competition curves were obtained for compounds
MSB-A1, MSB-A2a, MSB-A2b and MSB-A3. Since A2A and A2B recep-
tors are Gs-coupled receptors competition curves where performed
over the NECA-induced cAMP accumulation. By other side, A1 and
F-12), containing 10% Foetal Calf Serum (FCS) and 1%
Cells were washed 3x with 200 l assay medium (DMEM-F12 and
25 mM HEPES pH 7.4) and pre-incubated with assay medium con-
taining 30 M rolipram and test compounds at 37 °C for 15 min.
L-Glutamine.
l
l

3052
F. M. Areias et al. / Bioorg. Med. Chem. 18 (2010) 3043–3052
Patel, H. K.; Pritchard, S.; Wodicka, L. M.; Zarrinkar, P. P. Nat. Biotechnol. 2008,
26, 127.
5. Austin, C. P.; Brady, L. S.; Collins, F. S. Science 2004, 306, 1138.
6. Jensen, N. H.; Roth, B. L. Comb. Chem. High Throughput Screening 2008, 11, 420.
7. Wishart, D. S.; Knox, C.; Guo, A. C.; Cheng, D.; Shrivastava, S.; Tzur, D.; Gautam,
B.; Hassanali, M. Nucleic Acids Res. 2008, 36, D901.
8. Harmar, A. J.; Hills, R. A.; Rosser, E. M.; Jones, M.; Buneman, O. P.; Dunbar, D. R.;
Greenhill, S. D.; Hale, V. A.; Sharman, J. L.; Bonner, T. I.; Catterall, W. A.;
Davenport, A. P.; Delagrange, P.; Dollery, C. T.; Foord, S. M.; Gutman, G. A.;
Laudet, V.; Neubig, R. R.; Ohlstein, E. H.; Olsen, R. W.; Peters, J.; Pin, J. P.;
Ruffolo, R. R.; Searls, D. B.; Wright, M. W.; Spedding, M. Nucleic Acids Res. 2009,
37, D680.
1 lM NECA was incubated for 15 min at 37 °C and 10 lM forskolin
was incubated for 3 min (total incubation time 30 min). Reaction
was stopped with lysis buffer supplied in the kit and the en-
zyme-immunoassay was carried out for detection of intracellular
cAMP at 450 nm in an Ultra Evolution detector (Tecan). Data were
fitted by non-linear regression using GraphPad Prism v2.01
(GraphPad Software).
Acknowledgements
9. Liu, T.;Lin,Y.;Wen,X.;Jorissen,R. N.;Gilson,M. K. NucleicAcidsRes. 2007,35, D198.
10. Berman, H. M.; Westbrook, J.; Feng, Z.; Gilliland, G.; Bhat, T. N.; Weissig, H.;
Shindyalov, I. N.; Bourne, P. E. Nucleic Acids Res. 2000, 28, 235.
11. Jenkins, J. L.; Bender, A.; Davies, J. W. Drug Discovery Today: Technol. 2006, 3,
413.
12. Ekins, S.; Mestres, J.; Testa, B. Br. J. Pharmacol. 2007, 152, 9.
13. Jenwitheesuk, E.; Horst, J. A.; Rivas, K. L.; van Voorhis, W. C.; Samudrala, R.
Trends Pharmacol. Sci. 2008, 29, 62.
14. Keiser, M. J.; Roth, B. L.; Armbruster, B. N.; Ernsberger, P.; Irwin, J. J.; Shoichet,
B. K. Nat. Biotechnol. 2007, 25, 197.
15. Plouffe, D.; Brinker, A.; McNamara, C.; Henson, K.; Kato, N.; Kuhen, K.; Nagle,
A.; Adrián, F.; Matzen, J. T.; Anderson, P.; Nam, T. G.; Gray, N. S.; Chatterjee, A.;
Janes, J.; Yan, S. F.; Trager, R.; Caldwell, J. S.; Schultz, P. G.; Zhou, Y.; Winzeler, E.
A. Proc. Natl. Acad. Sci. U.S.A. 2008, 105, 9059.
This research was funded by the Spanish Ministerio de Ciencia e
Innovación (Grants HF2007-0055 and BIO2008-02329), the Portu-
guese Fundação para a Ciência e Tecnologia (PPCDT/QUI/59356/
2004), the Xunta de Galicia (07CSA003203PR and 08CSA02020
3PR), and the Instituto de Salud Carlos III. J.B. is the recipient of
an Isabel Barreto Contract from the Xunta de Galicia. FMA and
MEAZ gratefully acknowledge Post-doc grants from the Portuguese
FCT (SFRH/BPD/26106/2005 and SFRH/BPD/27029/2006).
Supplementary data
16. Muller, P.; Lena, G.; Boilard, E.; Bezzine, S.; Lambeau, G.; Guichard, G.; Rognan,
D. J. Med. Chem. 2006, 49, 6768.
Supplementary data (full list of the 86 G protein-coupled recep-
tors against which in silico target profiling was performed, to-
gether with the predicted number of annotations assigned to
each target) associated with this article can be found, in the online
version, at doi:10.1016/j.bmc.2010.03.048.
17. Rollinger, J. M.; Schuster, D.; Danzl, B.; Schwaiger, S.; Markt, P.; Schmidtke, M.;
Gertsch, J.; Raduner, S.; Wolber, G.; Langer, T.; Stuppner, H. Planta Med. 2009,
75, 195.
18. Gregori-Puigjané, E.; Mestres, J. Comb. Chem. High Throughput Screening 2008,
11, 669.
19. Jacobson, K. A.; Gao, Z. G. Nat. Rev. Drug Disc. 2006, 5, 247.
20. Haskó, G.; Linden, J.; Cronstein, B.; Pacher, P. Nat. Rev. Drug Disc. 2008, 7, 759.
21. Baraldi, P. G.; Tabrizi, M. A.; Gessi, S.; Borea, P. A. Chem. Rev. 2008, 108, 238.
22. Gregori-Puigjané, E.; Mestres, J. J. Chem. Inf. Model. 2006, 46, 1615.
23. Lipinski, C. A. Drug Discovery Today: Technol. 2004, 1, 337.
24. Muir, T. W. ACS Chem. Biol. 2009, 4, 241.
References and notes
1. Schreiber, S. L. Nat. Chem. Biol. 2005, 1, 64.
25. Roth, B. L.; Lopez, E.; Beischel, S.; Westkaemper, R. B.; Evans, J. M. Pharmacol.
Ther. 2004, 102, 99.
2. Krejsa, C. M.; Horvath, D.; Rogalski, S. L.; Penzotti, J. E.; Mao, B.; Barbosa, F.;
Migeon, J. C. Curr. Opin. Drug Discov. Devel. 2003, 6, 470.
26. Mestres, J.; Martin-Couce, L.; Gregori-Puigjané, E.; Cases, M.; Boyer, S. J. Chem.
Inf. Model. 2006, 46, 2725.
3. Whitebread, S.; Hamon, J.; Bojanic, D.; Urban, L. Drug Discovery Today 2005, 10,
1421.
27. Booth, B. L.; Dias, A. M.; Proença, M. F. J. Chem. Soc., Perkin Trans. I 1992, 2119.
28. Booth, B. L.; Dias, A. M.; Proença, M. F.; Zaki, M. E. A. J. Org. Chem. 2001, 66,
8436.
4. Karaman, M. W.; Herrgard, S.; Treiber, D. K.; Gallant, P.; Atteridge, C. E.;
Campbell, B. T.; Chan, K. W.; Ciceri, P.; Davis, M. I.; Edeen, P. T.; Faraoni, R.;
Floyd, M.; Hunt, J. P.; Lockhart, D. J.; Milanov, Z. V.; Morrison, M. J.; Pallares, G.;