Virtual screening leads to the discovery of novel non-nucleotide P2Y 1 receptor antagonists

Article abstract of DOI:10.1016/j.bmc.2012.06.044

The P2Y₁ receptor (P2Y₁R) is a G protein-coupled receptor naturally activated by extracellular ADP. Its stimulation is an essential requirement of ADP-induced platelet aggregation, thus making antagonists highly sought compounds for the development of antithrombotic agents. Here, through a virtual screening campaign based on a pharmacophoric representation of the common characteristics of known P2Y₁R ligands and the putative shape and size of the receptor binding pocket, we have identified novel antagonist hits of μM affinity derived from a N,N′-bis-arylurea chemotype. Unlike the vast majority of known P2Y ₁R antagonists, these drug-like compounds do not have a nucleotidic scaffold or highly negatively charged phosphate groups. Hence, our compounds may provide a direction for the development of receptor probes with altered physicochemical properties.

Full text of DOI:10.1016/j.bmc.2012.06.044

Bioorganic & Medicinal Chemistry 20 (2012) 5254–5261
Contents lists available at SciVerse ScienceDirect
Bioorganic & Medicinal Chemistry
journal homepage: www.elsevier.com/locate/bmc
Virtual screening leads to the discovery of novel non-nucleotide P2Y1
receptor antagonists
Stefano Costanzi a, , T. Santhosh Kumar b, Ramachandran Balasubramanian b, T. Kendall Harden c,
⇑
,
Kenneth A. Jacobson b,
⇑
a Laboratory of Biological Modeling, National Institutes of Health, Bethesda, MD 20892, USA
b Molecular Recognition Section (Laboratory of Biooorganic Chemistry), National Institutes of Diabetes and Digestive and Kidney Diseases,
National Institutes of Health, Bethesda, MD 20892, USA
c Department of Pharmacology, University of North Carolina, School of Medicine, Chapel Hill, NC 27599, USA
a r t i c l e i n f o
a b s t r a c t
Article history:
The P2Y1 receptor (P2Y1R) is a G protein-coupled receptor naturally activated by extracellular ADP. Its
stimulation is an essential requirement of ADP-induced platelet aggregation, thus making antagonists
highly sought compounds for the development of antithrombotic agents. Here, through a virtual screen-
ing campaign based on a pharmacophoric representation of the common characteristics of known P2Y1R
ligands and the putative shape and size of the receptor binding pocket, we have identified novel antag-
Received 15 May 2012
Revised 18 June 2012
Accepted 25 June 2012
Available online 3 July 2012
onist hits of l
M affinity derived from a N,N0-bis-arylurea chemotype. Unlike the vast majority of known
Keywords:
P2Y1 receptor
G protein-coupled receptor
Antagonist
Virtual screening
Molecular modeling
P2Y1R antagonists, these drug-like compounds do not have a nucleotidic scaffold or highly negatively
charged phosphate groups. Hence, our compounds may provide a direction for the development of recep-
tor probes with altered physicochemical properties.
Published by Elsevier Ltd.
1. Introduction
clinical tools, while antagonists of the P2Y1R are still under study
as an alternative platform for the development of antiaggregatory
The P2Y receptor (P2YR) family comprises eight subtypes of G
protein-coupled receptors (GPCRs) that are activated by extracellu-
lar nucleotides.1 Two subtypes, namely the Gq-coupled P2Y1R and
the Gi-coupled P2Y12R, play a prominent role in platelet aggrega-
tion, since their stimulation by adenosine 50-diphosphate (ADP)
synergistically mediates a proaggregatory response.2 Since the
simultaneous activation of both receptors is required to elicit
aggregation, antagonists of either subtype act as effective anti-
thrombotic agents. In particular, antagonists of the P2Y12R, such
as liver-activated clopidogrel (PlavixÒ), are already widespread
drugs.3,4 With the exception of a few compounds recently identi-
fied by researchers at GlaxoSmithKline5 and Pfizer,6 most of the
known P2Y1R antagonists are based on the structure of adeno-
sine-30,50-bisphosphate (A3P5P).7,8 For example, the selective
antagonist (10R,20S,40S,50S)-4-(2-iodo-6-methylamino-purin-9-yl)-
1-[(phosphato)-methyl]-2-(phosphato)-bicyclo[3.1.0]hexane
(MRS2500), compound 1 (Fig. 1), is used widely as a pharmacolog-
ical probe of the P2Y1R. It contains a substituted ribose ring moiety,
namely, a methanocarba ring system (bicyclo[3.1.0]hexane), which
is constrained in the P2Y1R-preferring Northern (N) conformation.9
Notably, compound 1 was shown to be effective in reducing
thrombus formation in vivo and to be more resistant to degrada-
tion than related 9-riboside derivatives.10
Although a crystal structure of the P2Y1R in complex with its li-
gands has not been solved, we have extensively studied the molec-
ular recognition at the P2Y1R using a combination of site-directed
mutagenesis, structure-activity relationships (SARs), and molecu-
lar modeling based on the known structure of homologue
GPCRs.1,8,11–13 In the course of these studies, we have docked sev-
eral nucleotide antagonists to the putative ligand binding site of
the P2Y1R. Here, in the quest for P2Y1R antagonists with more
drug-like characteristics, we generated a three-dimensional phar-
macophore that reflected the common chemical features shown
Abbreviations: ADP, adenosine 50-diphosphate; A3P5P, adenosine-30,50-bisphos-
phate; DMEM, Dulbecco0s modified Eagle medium; ESI-HRMS, electrospray
ionization-high resolution mass spectroscopy; GPCR, G protein-coupled receptor;
2-MeSADP, 2-methylthio-adenosine 50-diphosphate; MOE, molecular operating
environment; MRS2500, (10R,20S,40S,50S)-4-(2-iodo-6-methylamino-purin-9-yl)-1-
[(phosphato)-methyl]-2-(phosphato)-bicyclo[3.1.0]hexane; PLC, phospholipase C;
QSAR, quantitative structure activity relationship; RMSD, root mean square
deviation; THF, tetrahydrofuran; TGT, typed graph triangle.
⇑
Corresponding authors. Tel.: +1 301 496 9024; fax: +1 301 480 8422 (K.A.J.);
tel.: +1 202 885 1722 (S.C.).
E-mail addresses: costanzi@american.edu (S. Costanzi), kajacobs@helix.nih.gov
(K.A. Jacobson).
Present address: Department of Chemistry, American University, Washington, DC
20016, USA.
0968-0896/$ - see front matter Published by Elsevier Ltd.
http://dx.doi.org/10.1016/j.bmc.2012.06.044

S. Costanzi et al. / Bioorg. Med. Chem. 20 (2012) 5254–5261
5255
Figure 1. Representation of the pharmacophore at the base of the virtual screening, embedded within the homology model of the P2Y1R: the hydrogen bond donor feature is
in light purple; its projection is in dark purple; the aromatic or hydrogen bond acceptor feature is in indigo; the negatively charged or hydrogen bond acceptor features are in
pink. The ribbon representation of the backbone of the receptor is colored with a continuum spectrum going from red at the N-terminus to purple at the C-terminus, with
TM1 in orange, TM2 in yellow/green, TM3 in green, TM4 in blue/green, TM5 in cyan, TM6 in blue and TM7 in purple. The high affinity antagonist MRS2500 (1) appears in the
figure, with its carbon atoms in yellow, while a schematic 2D representation is shown in the lower left inset. Three cationic residues that, according to experimental evidence,
are fundamental for ligand recognition are also shown, with their carbon atoms in grey.
by these docked known antagonists. We then performed a pharma-
cophore-based virtual screen that led us to the discovery of novel
antagonists derived from non-nucleotide chemical scaffolds.
2.2. Virtual screening
A database of approximately 250,000 commercially available
compounds was then virtually screened according to the scheme
represented in Figure 2. Prior to the pharmacophore search, the
database was subjected to a descriptor-based filtering to rapidly
eliminate the compounds devoid of the characteristics necessary
to match the pharmacophore. In particular, we eliminated all the
compounds endowed with less than one hydrogen bond donor
and less than four hydrogen bond acceptors (or one aromatic ring
and three hydrogen bond acceptors). The resulting filtered data-
base, amounting to about half of the original size, was then sub-
jected to the pharmacophore search and yielded 362 virtual hits.
Aiming at testing about one hundred compounds, we subjected
these virtual hits to a molecular fingerprint-based clustering,
which yielded 51 clusters of molecules with similar structures
and 59 unique singletons. We then selected a total of 110 mole-
2. Results and discussion
2.1. Construction of the pharmacophore
We have recently published a docking-based QSAR analysis of
53 in-house developed nucleotide-based antagonists of the
P2Y1R.8 On the basis of the docked conformations of said com-
pounds, here we generated a pharmacophore that reflected their
common chemical characteristics as well as the shape of the puta-
tive ligand binding pocket in our P2Y1R homology model (Fig. 1).
The number, size and position of the pharmacophoric features
were then adjusted in order to maximize the number of known ac-
tive compounds able to pass the pharmacophoric search, leading to
a final pharmacophoric query consisting of: (a) a hydrogen-bond
donor feature with a radius of 1 Å corresponding to the exocyclic
amino group of the known bisphosphate antagonists, supple-
mented by a projected feature with a radius of 1.4 Å—this was nec-
essary to ensure that the H-bond donor feature maintained the
orientation observed in the docked P2Y1 antagonists; (b) an aro-
matic or hydrogen bond acceptor feature with a radius of 2.6 Å,
corresponding to the adenine moiety of the known antagonists;
(c) six negatively charged or hydrogen-bond acceptor features with
a radius of 1.3 Å, corresponding to the phosphate moieties of the
known antagonists. The criteria for the successful passing of the
pharmacophore search by a given compound were defined as the
essential matching of the hydrogen-bond donor feature and its
projection, and the additional matching of four of the remaining
features. The shape of the P2Y1R binding pocket was taken into ac-
count by adding to the pharmacophore query an excluded volume
corresponding to the residues that line it in our model. These re-
gions were treated as impenetrable by the ligands. Under these
conditions, all the above mentioned 53 known inhibitors were suc-
cessfully recognized as ligands through the pharmacophore search.
Figure 2. Schematic representation of the virtual screening process.

5256
S. Costanzi et al. / Bioorg. Med. Chem. 20 (2012) 5254–5261
cules, namely all 59 singletons and a representative compound for
each cluster, choosing the member that in the pharmacophore
search yielded the lowest root mean square deviation (RMSD) with
the query.
dust in acetic acid to obtain the amine 14 in 69% yield. The amine
14 was then used as a common intermediate to generate the vari-
ous urea derivatives 2a and 15a–i, by reacting with the corre-
sponding isocyanates.
The compounds were evaluated for their ability to inhibit P2Y1
antagonist radioligand binding and to increase intracellular Ca2+
concentration. Selected compounds were found to inhibit 2-Me-
SADP-stimulated phospholipase C (PLC) signaling. The binding
was measured in membranes of hP2Y1R-expressing Sf9 insect cells,
and the functional assays were performed in hP2Y1R-expressing
1321N1 human astrocytoma cells. The results, shown in Table 2,
indicated that several of the in-house synthesized analogs of 2a
2.3. Biological evaluation of the virtual hits and identification of
novel ligands
The selected compounds were evaluated for their ability to in-
hibit the binding of a P2Y1 antagonist radioligand in membranes
of human (h) P2Y1R-expressing Sf9 insect cells. The experimental
screening, which was initially conducted studying the inhibition
of radioligand binding by a 20
lM concentration of the tested com-
bind to the P2Y1R with an affinity in the low lM range, although
pounds, yielded the identification of several novel ligand hits with
none of the compounds appeared to be a better binder than the
parent compound. Moreover, the data show that antagonism of
P2Y1R-induced effects was present. In particular, when adminis-
P30% inhibition of [125I]MRS2500 binding (2–4).16 Among them,
N,N0-bis-arylurea 2a was found to have an affinity in the low
lM
range, thus constituting a very good starting basis for hit optimiza-
tion. In a subsequent search after the chemical modification of 2a
was initiated, we tested commercially available analogs of com-
pound 2a and found two structurally related compounds, 2b and
2c, of comparable affinity. The structure of other tentative library
hits (5–10) that inhibited only 20% of radioligand binding are
shown in Supporting Information (Table S1).
tered at a concentration of 10 lM, the compounds endowed with
higher affinity for the receptor inhibited about 30% of the P2Y1R-
mediated calcium mobilization induced by the agonist 2-MeSADP.
The only exception appears to be compound 15b that, despite a rel-
atively high binding affinity, seemed to be a weaker antagonist.
2.5. Analysis of the SAR data
2.4. Synthesis and biological evaluation of analogs of compound
2a
The SAR data indicate that the presence of chlorine substituents
at the meta and para positions of the phenyl ring is crucial for the
affinity and the antagonistic activity of compound 2a. The removal
of both chlorine substituents (compound 15a) from compound 2a
leads to a complete loss of affinity. Even the sole removal of the
chlorine at the para position (compound 15c) leads to a substantial
loss of affinity, which however is partially regained when a second
chlorine atom is reintroduced at the other meta position (com-
pound 15b). The substitution of the chlorine atoms with an elec-
tron withdrawing trifluoromethyl group (compound 15g) is
tolerated by the receptor, while their substitution with electron
donating methyl or methoxy groups is not (compounds 15h and
15i). Moreover, the substitution of the chlorine in para with a fluo-
rine (compound 15d), although somewhat detrimental, is toler-
ated. However, the substitution of both chlorines with fluorines
leads to a complete loss of affinity (compound 15e and 15f). Taken
together, these data suggest that the region of the receptor that
surrounds the phenyl ring is not inclined to the recognition of elec-
tron rich moieties, thus supporting the structural results of our
pharmacophore search that placed the disubstituted phenyl ring
As mentioned, before the identification of the activity of com-
pounds 2b and 2c we conducted a preliminary exploration of the
SAR of the novel antagonists by synthesizing analogs of compound
2a. Compound 2a was also resynthesized in order to confirm the
structure provided. In particular, taking advantage of the commer-
cial availability of several aryl isocyanates, we replaced the di-
chloro-substituted phenyl ring of 2a with six different
disubstituted analogs. The central phenylsulfonamido or isoxazole
rings remained unmodified. The synthetic route for the preparation
of various urea derivatives 2a and 15a–i is shown in Scheme 1. A
coupling reaction between 5-amino-3,4-dimethylisoxazole 11
and p-nitrobenzenesulfonyl chloride 12 generated the nitro com-
pound 13 in 41% yield. Our efforts to reduce the nitro group under
Pd/C catalyzed hydrogenation reaction conditions resulted in the
nitro reduction along with reduction of the less substituted double
bond of the 3,4-dimethylisoxazol-5-yl moiety (results not shown).
However, the nitro group was chemo-selectively reduced using Zn
Zn Powder,
AcOH:CH3CN
(1:1, v/v)
N
O
O
O
O
Et3N,
DMAP, an DCM
O
NH2
O
S
S
Cl
N
H
O2N
O2N
N
11
12
13
N
N
O
R
O
O
O
DIPEA and
THF, 65 °C
O
O
S
S
O
N
H
N
H
N
H2N
N
H
NCO
H
R
14
2a, 15a-i
2a
R = 3,4-Cl2
15d R = 3-Cl-4-F 15g R = 4-CF3
15e R = 3,4-F2 5h R = 3,4-(OCH3)2
15f R = 3,5-F2 5i R = 3,4-(CH3)2
15a R = H
15b R = 3,5-Cl2
15c R = 3-Cl
Scheme 1.

S. Costanzi et al. / Bioorg. Med. Chem. 20 (2012) 5254–5261
5257
Table 1
In vitro pharmacological data for compounds selected through an in silico screen (life chemicals code shown) for inhibition of radioligand binding at the hP2Y1R
Compound No.
Structure
Binding Ki, l
M, or % inhibitiona
H
N
H
N
Cl
Cl
O
2a, F3099-0019
13
6
3
O
N
S
O
O
N
H
O
H
N
H
N
O
Cl
N
O
2b, F0348-3162
2c, F1766-0041
O
S
N
N
O
H
H
H
N
N
N
N
O
10
O
CF3
S
S
N
O
H
N
O
N
3, F2019-1559
4, F1795-0008
40%b
N
O
O
S
O
O
O
S
-O
HN
N
O
S
O
30%b
HN
NH
O
S
O
O-
Assays were with membranes from Sf9 insect cells expressing the hP2Y1R and performed with a radioligand ([125I]MRS2500) concentration of approximately 0.2 nM.
a
b
Percent inhibition of specific binding of [125I]MRS2500 by a 20
lM concentration of test compound.
of compound 2a distantly from the negatively charged phosphate
groups of 1. In particular, our pharmacophore search suggests that
the ureic nitrogen of 2a closer to the dichlorophenyl ring mimics
the exocyclic amino group of 1, while the 5-sulfonamido-isoxazole
moiety of 2a substitutes the two phosphates groups of 1 (Fig. 3).
This hypothesis is consistent with our published models of the
P2Y1R in complex with 1, according to which the phosphates of
this nucleotide antagonist bind to a portion of the binding cavity
lined by cationic residues, which are notoriously electrophilic
and fluorophilic residues.14 Notably, compound 2a is more drug-
like in its physicochemical properties (cLogp = 3.79) than antago-
nists having negatively-charged phosphate groups.
tions that the nucleotide antagonists establish with the P2Y1R are
not essential for ligand recognition. Our pharmacophore model
suggests that the phosphates of nucleotide antagonists are re-
placed by 5-sulfonamido-isoxazole moiety of the novel antago-
nists, which most likely establish hydrogen-bonds and cation-
aromatic interactions with the receptor. The confirmation that
these substances bind as suggested by the pharacophore model
will depend on the identification of analogues with enhanced affin-
ity. These molecules may now be subjected to further structural
optimization and more extensive pharmacological characterization
in platelet aggregation and other models.
4. Materials and methods
4.1. Molecular modeling
3. Conclusions
In conclusion, our pharmacophore-based virtual screening led
to the identification of novel P2Y1R antagonists endowed with
drug-like structures. One of the more promising hits, a N,N0-bis-
phenylurea 2a, from the in silico screen was validated by de novo
synthesis and the preparation of closely related analogues to gen-
erate family of non-nucleotide P2Y1 antagonists, some of which
The molecular modeling study was performed with the molec-
ular operating environment (MOE), Chemical Computing Group,
Inc. (www.chemcomp.com). The molecular database subjected to
the virtual screening process was the catalogue of compounds
commercially available from Life Chemicals, Inc. (Burlington, ON,
Canada, www.lifechemicals.com).
displayed affinities and antagonistic potencies in the low
lM
range. Thus, these molecules represent good candidates for a fol-
low-up hit optimization study. Our initial attempts to explore
the SAR of the phenyl ring, did not improve the potency as antag-
onists, but shed light onto the requirements of ligand recognition.
Moreover, there are additional sites on the molecules, such as the
isoxazole ring, that can be explored, with additional information
on other heterocycles from library hit compounds 2b and 2c.
The novel P2Y1 antagonists identified through this work lack
negatively charged phosphate groups, thus providing more suit-
able scaffolds for the development of development of receptor
probes with different physicochemical properties from the canon-
ical A3P5P-based antagonists. However, being devoid of ionized
groups, these novel antagonists demonstrate that the ionic interac-
4.1.1. Construction of the pharmacophore
The pharmacophore query was generated with the ‘pharmaco-
phore query editor’ of MOE. A set of 53 in house-developed
A3P5P-based P2Y1 antagonists8 were loaded into MOE, and an ini-
tial query was generated using the ‘consensus’ function, according
to the PCHD scheme. Only the phosphate groups, the purine ring,
and the exocyclic amino group were taken into account. The result-
ing query was then simplified by unifying the aromatic/hydrogen
bond acceptor feature relative to the purine ring and deleting all
the projected features. However, we preserved the projected donor
feature relative to the exocyclic amine, in order to ensure its direc-
tionality. Moreover, an excluded volume was added on the basis of

5258
S. Costanzi et al. / Bioorg. Med. Chem. 20 (2012) 5254–5261
Table 2
In vitro pharmacological data for the synthesized analogs of compound 2a at the hP2Y1R in binding, determined as in Table 1, and in the antagonism increase of intracellular
calcium ion concentration induced by 30 nM 2-MeSADP in 1321N1 astrocytoma cells expressing the hP2Y1R
N
O
O
O
S
O
N
H
R
N
H
N
H
Compound No.
R
Binding Ki,
lM
% inhibition in Ca2+ assay at 10
36.8 2.2a
lM
Cl
Cl
2a (F3099-0019)
13
3
Cl
15a
15b
NE
<5
27 10
17.6
1
Cl
Cl
15c
15d
15e
P100
42 15
NE
21.7 3.5
31.3 2.5
11.3 2.3
Cl
F
F
F
F
15f
P100
10.2 2.6
F
15g
15h
45
8
32.7 3.7
9.1 0.4
CF3
OCH3
NE
OCH3
CH3
15i
NE
9.4 0.2b
CH3
a
IC50 in the IP assay: 0.14 0.04
IC50 in the IP assay: 0.13 0.01
l
l
M.
M; NE–less than 10% inhibition at 20
b
lM.
all the atoms of the residues lining the binding pocket in our rho-
dopsin-based model of the P2Y1R,8 thus accounting for its shape
and size. Specifically, the generation of the excluded volume was
based on the following residues, indicated through their sequence
number as well as their GPCR residue index: L54(1.35), V57(1.38),
Y58(1.39), V61(1.42), Y100(2.53), L105(2.57), L108(2.61),
R128(3.29), F131(3.32), H132(3.33), L135(3.36), K196(EL2.44),
Chemicals database with the ‘calculate descriptors’ function of
MOE. To expedite the screening, we then deleted all the com-
pounds that did not the have the necessary features to match the
pharmacophore query.
4.1.3. Conformational explosion
The resulting filtered Life Chemicals database was then sub-
jected to a conformational search with the ‘conformation import’
function of MOE, in order to generate for each compound multiple
conformers to be evaluated in the pharmacophore search. The
maximum number of conformations per compound was set to
250, and the maximum strain energy for a conformation to be ac-
cepted was set to 100 kcal/mol. A screenshot with the complete
settings is provided in Figure S1 of the Supplementary data.
N197(EL2.45),
I200(EL2.48),
T201(EL2.49),
Y203(EL2.51),
D204(EL2.52), F276(6.51), H277(6.52), K280(6.55), N283(6.58),
Q307(7.36), R310(7.39), G311(7.40), S314(7.43)—for more infor-
mation on the GPCR residue index, see Ballesteros and Weinstein15
and Costanzi and coworkers.1 Finally, we tested the generated
pharmacophore query for its ability to correctly recognize the 53
known antagonists and adjusted the size and the position of the
features in order to ensure the correct recognition of the entire
set of the known ligands. For this test, the 53 known antagonists
where sketched from scratch in MOE and subjected to the same
conformational search as the Life Chemicals database, thus recreat-
ing the condition used in the actual pharmacophore search.
4.1.4. Fingerprint-based clustering
The compounds that passed the pharmacophore search were
clustered on the basis of their molecular similarity according to
two different fingerprints: the MACCS structural keys and the
typed graph triangle (TGT) fingerprint. Both fingerprints are con-
formation-independent and can be calculated from a two-dimen-
sional representation of the molecules. The molecular similarity
was calculated according to the Tanimoto coefficient, and the
4.1.2. Descriptor-based filtering of the database
The number of hydrogen bond acceptors, hydrogen bond donors
and aromatic atoms was calculated for all compounds in the Life

S. Costanzi et al. / Bioorg. Med. Chem. 20 (2012) 5254–5261
5259
4.2.3. General experimental procedure for compounds 2a and
15a–i
Method A: Amine 14 (25 mg, 0.094 mmol) was coevaporated
with anhydrous toluene (3 ꢀ 3 mL) and dissolved in anhydrous
THF (2 mL). Diisopropylethylamine (66 lL, 0.38 mmol) and the
corresponding isocyanate (0.140 mmol) were added. After stirring
for 2 h at 70 °C, the reaction mixture was diluted with EtOAc
(25 mL) and washed with saturated aq. NaCl (1 ꢀ 15 mL). The
aqueous phase was back extracted with EtOAc (3 ꢀ 20 mL), and
the combined organic phases were evaporated to dryness. The
resulting residue was purified by silica gel column chromatogra-
phy (0–10% MeOH in EtOAc, v/v) to afford target compounds 2a
and 15b, 15d–f, and 15h–i, as solid materials.
Method B: Amine 14 (15 mg, 0.056 mmol) was coevaporated
with anhydrous toluene (3 ꢀ 3 mL) and dissolved in anhydrous
THF (2 mL). Diisoproylethylamine (40 lL, 0.224 mmol) and corre-
sponding isocyanate (0.224 mmol) were added. After stirring for
16 h at room temperature, the reaction mixture was evaporated
to dryness. The resulting residue was purified by silica gel column
chromatography (0–6% MeOH in EtOAc, v/v) to afford target com-
pounds 15a, 15c and 15g as solid materials.
Figure 3. Superimposition of MRS2500 (1) and compound 2, as resulting from the
pharmacophore search. MRS2500 is shown in ball and stick representation, with the
carbon atoms colored in green. Compound 2 is shown in stick representation, with
the carbon atoms colored in grey.
4.2.3.1. 4-(3-(3,4-Dichlorophenyl)ureido)-N-(3,4-dimethyli sox-
clustering was done setting the threshold for similarity and overlap
to 80% for the MACCS and 88% for the TGT fingerprints. From the
clusters independently calculated with the two fingerprints, a con-
sensus cluster was then generated with a union operation via an ad
hoc written script.
azol-5-yl)benzenesulfonamide (2a).
Rf = 0.5 (10% MeOH in
EtOAc, v/v); ESI-HRMS m/z 455.0331 ([M+H]+, C18H16N4O4SCl2ꢁH+:
Calcd 455.0348); 1H NMR (CD3OD) d 7.81 (d, J = 2.8 Hz, 1H), 7.72
(d, J = 8.8 Hz, 2H), 7.63 (d, J = 8.8 Hz, 2H), 7.42 (d, J = 8.8 Hz, 1H),
7.31 (dd, J = 2.8, 8.8 Hz, 1H), 2.14 (s, 3H), 1.78 (s, 3H).
4.2. Chemical synthesis
4.2.3.2. N-(3,4-Dimethylisoxazol-5-yl)-4-nitrobenzenesulfonami
de (13).
Rf = 0.2 (10% MeOH in EtOAc, v/v); ESI-HRMS m/z
The synthetic procedures for intermediates are described here,
while analytical and spectroscopic data for each compound tested
biologically are given below. Compounds subjected to biological
testing were of >95% purity. For structural confirmation, 1H NMR
spectra were recorded at 300 Mz. Chemical shifts are reported in
parts per million (ppm) relative to deuterated solvent as the inter-
nal standard. ESI-High resolution mass spectroscopic measure-
298.0492 ([M+H]+, C11H11N3O5SꢁH+: Calcd 298.0498); 1H NMR
(CDCl3) d 8.37 (d, J = 8.8 Hz, 2H), 8.01 (d, J = 8.8 Hz, 2H), 2.22 (s,
3H), 1.98 (s, 3H), 1.57 (s, 1H).
4.2.3.3. 4-Amino-N-(3,4-dimethylisoxazol-5-yl)benzenesulfon-
amide (14).
Rf = 0.4 (7% MeOH in EtOAc, v/v); ESI-HRMS m/z
268.0767 ([M+H]+, C11H13N3O3SꢁH+: Calcd 268.0756); 1H NMR
(CDCl3) d 7.56 (d, J = 8.8 Hz, 2H), 6.64 (d, J = 8.8 Hz, 2H), 4.19 (s,
2H), 2.18 (s, 3H), 1.91 (s, 3H), 1.59 (s, 1H).
ments were performed on
a proteomics optimized Q-TOF-2
(micromass-waters) using external calibration with polyalanine.
cLogp was calculated using ChemBioDraw Ultra v. 12.03 (PerkinEl-
mer, Boston, MA).
4.2.3.4. N-(3,4-Dimethylisoxazol-5-yl)-4-(3-phenylureido)benze
nesulfonamide (15a).
Rf = 0.2 (5% MeOH in CH2Cl2, v/v); ESI-
4.2.1. N-(3,4-Dimethylisoxazol-5-yl)-4-nitrobenzenesulfona
mide (13)
HRMS m/z 387.1118 ([M+H]+, C18H18N4O4SꢁH+: Calcd 387.1127);
1H NMR (CD3OD) d, 7.72 (d, J = 9.1 Hz, 2H), 7.64 (d, J = 9.1 Hz,
2H), 7.40–7.48 (m, 2H), 7.27–7.34 (m, 2H), 7.02–7.07 (m, 1H),
2.16 (s, 3H), 1.79 (s, 3H).
5-Amino-3,4-dimethylisoxazole 11 (0.20 g, 1.78 mmol) and p-
nitrobenzenesulfonyl chloride 12 (0.39 g, 1.78 mmol) were dis-
solved in anhydrous CH2Cl2 (20 mL). Et3N (0.75 mL, 5.36 mmol)
and N,N-dimethylaminopyridine (44 mg, 0.36 mmol) were then
added. After stirring the reaction mixture for 42 h, it was diluted
with EtOAc (50 mL) and washed with sat. aq. NaHCO3
(1 ꢀ 20 mL) and sat. aq. NaCl (1 ꢀ 20 mL). The aqueous phase
was back extracted with EtOAc (3 ꢀ 20 mL), and the combined or-
ganic phases were evaporated to dryness. The resulting residue
was purified by silica gel column chromatography (0–10% MeOH
in EtOAc, v/v) to afford compound 13 (0.22 g, 41%) as a solid.
4.2.3.5. 4-(3-(3,5-Dichlorophenyl)ureido)-N-(3,4-dimethylisoxa
zol-5-yl)benzenesulfonamide (15b).
Rf = 0.2 (5% MeOH in
EtOAc, v/v); ESI-HRMS m/z 455.0348 ([M+H]+, C18H16N4O4SCl2ꢁH+:
Calcd 455.0348); 1H NMR (CD3OD) d 7.93 (d, J = 8.8 Hz, 2H), 7.66
(d, J = 8.8 Hz, 2H), 7.50 (d, J = 1.7 Hz, 2H), 7.06–7.08 (m, 1H), 2.07
(s, 3H), 1.19 (s, 3H).
4.2.3.6. 4-(3-(3-Chlorophenyl)ureido)-N-(3,4-dimethylisoxazol-
5-yl)benzenesulfonamide (15c).
Rf = 0.3 (5% MeOH in
4.2.2. 4-Amino-N-(3,4-dimethylisoxazol-5-yl)benzenesulfon am
ide (14)
CH2Cl2, v/v); ESI-HRMS m/z 421.0725 ([M+H]+, C18H17N4O4SClꢁH+:
Calcd 421.0737); 1H NMR (CD3OD) d, 7.74 (d, J = 8.1 Hz, 2H),
7.60–7.69 (m, 3H), 7.23–7.32 (m, 2H), 7.02–7.07 (m, 1H), 2.16 (s,
3H), 1.79 (s, 3H).
Nitro compound 13 (0.71 g, 2.40 mmol) was dissolved in a
mixture of AcOH:CH3CN (20 mL, 1:1, v/v) and cooled to 0 °C. Zn
powder (1.5 g) was added in small portions over a period of
10 min. After stirring the reaction for 3 h at rt, reaction mixture
was filtered off and filtrate was evaporated to dryness. The result-
ing residue was purified by silica gel column chromatography (0–
10% MeOH in EtOAc, v/v) to afford amine 14 (0.65 g, 69%) as a solid.
4.2.3.7. 4-(3-(3-Chloro-4-fluorophenyl)ureido)-N-(3,4-dimethyli
soxazol-5-yl)benzenesulfonamide (15d).
Rf = 0.2 (5% MeOH
in EtOAc, v/v); ESI-HRMS m/z 439.0641 ([M+H]+, C18H16N4O4
FClSꢁH+: Calcd 439.0643); 1H NMR (CD3OD) d 7.70–7.80 (m, 3H),

5260
S. Costanzi et al. / Bioorg. Med. Chem. 20 (2012) 5254–5261
7.65 (d, J = 8.8 Hz, 2H), 7.20–7.30 (m, 1H), 7.05–7.15 (m, 1H), 2.13
(s, 3H), 1.79 (s, 3H).
were measured as reported.20 Specifically, 1321N1 human astrocy-
toma cells stably expressing the hP2Y1R were cultured in DMEM
(JRH Biosciences, Inc., Lenexa, KS) supplemented with 5% fetal bo-
4.2.3.8. 4-(3-(3,4-Difluorophenyl)ureido)-N-(3,4-dimethyli sox-
vine serum, 100 units penicillin mLꢂ1, 100
l
g streptomycin mLꢂ1
g geneticin mLꢂ1. For the assay,
lL of media in 96-well flat-bot-
,
azol-5-yl)benzenesulfonamide (15e).
Rf = 0.2 (7% MeOH in
2
l
mol glutamine mLꢂ1, and 800
cells were grown overnight in 100
l
EtOAc, v/v); ESI-HRMS m/z 423.0924 ([M+H]+, C18H16N4O4SF2ꢁH+:
Calcd 423.0939); 1H NMR (CD3OD) d 7.76 (d, J = 8.8 Hz, 2H), 7.62
(d, J = 8.8 Hz, 2H), 7.18–7.21 (m, 2H), 7.09–7.12 (m, 1H), 2.16 (s,
3H), 1.77 (s, 3H).
tom plates at 37 °C at 5% CO2 or until approximately 60–80% con-
fluency. The calcium 4 assay kit (Molecular Devices, Sunnyvale, CA)
was used as directed without washing of the cells. Cells were
loaded with 30 lL of dye to each well and incubated for 1 h at
4.2.3.9. 4-(3-(3,5-Difluorophenyl)ureido)-N-(3,4-dimethyli sox-
room temperature. All the tested compounds were diluted to the
appropriate concentration in Hank0s buffer. Dose-response curves
for the agonist 2-MeSADP were obtained treating the cells with
six different concentration of the compound in the range between
(10ꢂ9 M and 10ꢂ5 M). For the antagonist studies, the same dose-re-
sponse curves were registered in the presence of the newly identi-
fied P2Y1 ligands, pretreating the cells for 20 min at room
azol-5-yl)benzenesulfonamide (15f).
Rf = 0.2 (5% MeOH in
EtOAc, v/v); ESI-HRMS m/z 423.0934 ([M+H]+, C18H16N4O4F2SꢁH+:
Calcd 423.0939); 1H NMR (CD3OD) d 7.75 (d, J = 8.8 Hz, 2H), 7.62
(d, J = 8.8 Hz, 2H), 7.14 (d, J = 8.2 Hz, 2H), 6.54–6.61 (m, 1H), 2.13
(s, 3H), 1.78 (s, 3H).
4.2.3.10. N-(3,4-Dimethylisoxazol-5-yl)-4-(3-(4-(trifluorometh
temperature with a 10 lM concentration of each candidate antag-
yl)phenyl)ureido)-benzenesulfonamide (15g).
Rf = 0.2 (5%
onist before application of the agonist 2-MeSADP. Samples were
run in triplicates using a Molecular Devices FLIPRTETRA at room
temperature. Cell fluorescence (excitation = 485 nm, emis-
sion = 525 nm) was monitored following exposure to the com-
pound. Increases in intracellular calcium are reported as the
maximum fluorescence value after exposure minus the basal fluo-
rescence value before exposure.
MeOH in CH2Cl2, v/v); ESI-HRMS m/z 387.1118 ([M+H]+, C19H17N4
O4SꢁH+: Calcd 387.1127); 1H NMR (CD3OD) d, 7.73 (d, J = 8.6 Hz,
2H), 7.60–7.68 (m, 4H), 7.58 (d, J = 8.7 Hz, 2H), 2.15 (s, 3H), 1.79
(s, 3H).
4.2.3.11. 4-(3-(3,4-Dimethoxyphenyl)ureido)-N-(3,4-dimethylis
oxazol-5-yl)benzenesulfonamide (15h).
Rf = 0.2 (5% MeOH
Binding and functional parameters were estimated using
GraphPAD Prism software (GraphPAD, San Diego, CA).
in EtOAc, v/v); ESI-HRMS m/z 447.1343 ([M+H]+, C20H23N4O6SꢁH+:
Calcd 447.1338); 1H NMR (CD3OD) d 7.72 (d, J = 8.8 Hz, 2H), 7.60
(d, J = 8.8 Hz, 2H), 7.22 (d, J = 2.2 Hz, 1H), 6.83–6.92 (m, 2H), 3.84
(s, 3H), 3.81 (s, 3H), 2.13 (s, 3H), 1.77 (s, 3H).
Acknowledgments
This research was supported in part by the Intramural Research
Program of NIDDK and by the National Institutes of General Med-
ical Sciences (GM38213).
4.2.3.12. N-(3,4-Dimethylisoxazol-5-yl)-4-(3-(3,4-dimethylphe
nyl)ureido)benzenesulfonamide (15i).
Rf = 0.2 (5% MeOH in
EtOAc, v/v); ESI-HRMS m/z 415.1447([M+H]+, C20H22N4O4SꢁH+:
Calcd 415.1440); 1H NMR (CD3OD) d 7.70 (d, J = 8.8 Hz, 2H), 7.60
(d, J = 8.8 Hz, 2H), 7.11–7.21 (m, 2H), 7.04 (d, J = 8.2 Hz, 1H), 2.24
(s, 3H), 2.21 (s, 3H), 2.14 (s, 3H), 1.77 (s, 3H).
Supplementary data
Supplementary data associated with this article can be found, in
the online version, at http://dx.doi.org/10.1016/j.bmc.2012.06.044.
4.3. Pharmacological evaluation
References and notes
4.3.1. Radioligand binding assay
Binding of [125I]MRS2500, a potent, selective, antagonist of the
P2Y1R was performed as reported elsewhere.16,17 2-MeSADP was
purchased from Sigma (St. Louis, MO). Myo-[3H]inositol (20 Ci/
mmol) was obtained from American Radiolabeled Chemicals (St.
Louis, MO).
1. Costanzi, S.; Mamedova, L.; Gao, Z.; Jacobson, K. J. Med. Chem. 2004, 47,
5393.
2. Gachet, C.; Hechler, B. Semin. Thromb. Hemost. 2005, 31, 162.
3. Shankar, H.; Kunapuli, S. P. Drug Discov. Today 2005, 2, 285.
4. Jacobson, K.; Boeynaems, J. Drug Discov. Today 2010, 15, 570.
5. Morales-Ramos, A.; Mecom, J.; Kiesow, T.; Graybill, T.; Brown, G.; Aiyar, N.;
Davenport, E.; Kallal, L.; Knapp-Reed, B.; Li, P.; Londregan, A.; Morrow, D.;
Senadhi, S.; Thalji, R.; Zhao, S.; Burns-Kurtis, C.; Marino, J. J. Bioorg. Med. Chem.
Lett. 2008, 18, 6222.
4.3.2. Assay of inositol phosphates
Selected compounds were examined for inhibition of P2Y1R-pro-
moted stimulation of inositol phosphate formation measured at the
hP2Y1R stably expressed in 1321N1 human astrocytoma cells as
previously described.8,18,19 The IC50 values were averaged from 3
independently determined concentration-effect curves for each
compound. Briefly, cells plated in 24-well dishes were labeled in
inositol-free medium (DMEM; Gibco, Gaithersburg, MD) containing
6. Pfefferkorn, J.; Choi, C.; Winters, T.; Kennedy, R.; Chi, L.; Perrin, L.; Lu, G.; Ping,
Y.; McClanahan, T.; Schroeder, R.; Leininger, M.; Geyer, A.; Schefzick, S.;
Atherton, J. Bioorg. Med. Chem. Lett. 2008, 18, 3338.
7. Boyer, J.; Romero-Avila, T.; Schachter, J.; Harden, T. Mol. Pharmacol. 1996, 50,
1323.
8. Costanzi, S.; Tikhonova, I.; Ohno, M.; Roh, E.; Joshi, B.; Colson, A.; Houston, D.;
Maddileti, S.; Harden, T.; Jacobson, K. J. Med. Chem. 2007, 50, 3229.
9. Kim, H.; Ohno, M.; Xu, B.; Kim, H.; Choi, Y.; Ji, X.; Maddileti, S.; Marquez, V.;
Harden, T.; Jacobson, K. J. Med. Chem. 2003, 46, 4974.
10. Hechler, B.; Nonne, C.; Roh, E.; Cattaneo, M.; Cazenave, J.; Lanza, F.; Jacobson,
K.; Gachet, C. J. Pharmacol. Exp. Ther. 2006, 316, 556.
11. Ivanov, A.; Costanzi, S.; Jacobson, K. J. Comput. Aided Mol. Des. 2006, 20, 417.
12. Ohno, M.; Costanzi, S.; Kim, H.; Kempeneers, V.; Vastmans, K.; Herdewijn, P.;
Maddileti, S.; Gao, Z.; Harden, T.; Jacobson, K. Bioorg. Med. Chem. 2004, 12,
5619.
13. de Castro, S.; Maruoka, H.; Hong, K.; Kilbey, S. M., 2nd; Costanzi, S.; Hechler, B.;
Brown, G. G., Jr.; Gachet, C.; Harden, T. K.; Jacobson, K. A. Bioconjug. Chem. 2010,
21, 1190.
1.0 l
Ci of 2-[3H]myo-inositol (20 Ci/mmol; American Radiolabeled
Chemicals, Inc., St. Louis, MO) for 18–24 h in a humidified atmo-
sphere of 95% air/5% CO2 at 37 °C. PLC activity was measured the fol-
lowing day by quantitating [3H]inositol phosphate accumulation
after a 10 min incubation at 37 °C in the presence of 10 mM LiCl. To-
tal [3H]inositol phosphates were quantified by anion exchange
chromatography as previously described.
14. Müller, K.; Faeh, C.; Diederich, F. Science 2007, 317, 1881.
15. Ballesteros, J. A.; Weinstein, H. In Methods in Neurosciences; Stuart, C. S., Ed.;
Academic Press, 1995; Vol. 25, p 366.
16. Ohlmann, P.; de Castro, S.; Brown, G. G., Jr.; Gachet, C.; Jacobson, K. A.; Harden,
T. K. Pharmacol. Res. 2010, 62, 344.
4.3.3. Calcium mobilization assay
Changes in intracellular calcium ion concentration induced by
the hP2Y1R stably expressed in 1321N1 human astrocytoma cells

S. Costanzi et al. / Bioorg. Med. Chem. 20 (2012) 5254–5261
5261
17. Waldo, G. L.; Corbitt, J.; Boyer, J. L.; Ravi, G.; Kim, H. S.; Ji, X. D.; Lacy, J.;
Jacobson, K. A.; Harden, T. K. Mol. Pharmacol. 2002, 62, 1249.
18. Harden, T. K.; Hawkins, P. T.; Stephens, L.; Boyer, J. L.; Downes, C. P. Biochem. J.
1988, 252, 583.
19. Boyer, J. L.; Downes, C. P.; Harden, T. K. J. Biol. Chem. 1989, 264, 884.
20. Balasubramanian, R.; Ruiz de Azua, I.; Wess, J.; Jacobson, K. A. Biochem.
Pharmacol. 2010, 79, 1317.