Structure-based drug design using GPCR homology modeling: Toward the discovery of novel selective CysLT2 antagonists

Article abstract of DOI:10.1016/j.ejmech.2013.01.041

3D structure of CysLT2 receptor was constructed by using homology modeling and molecular simulations. The binding pocket of CysLT2 receptor and the proposition of the interaction mode between CysLT2 and HAMI3379 were identified. A series of dicarboxylated chalcones was then virtually evaluated through molecular docking studies. A total of six compounds 13a-f with preferable scores was further synthesized and tested for CysLT2 antagonistic activities by determination of the cytosolic free Ca²⁺ levels in HEK293 cells. Compounds 13e and 13f exhibited potent and selective CysLT2 antagonistic activities with IC₅₀ values being 7.5 and 0.25 μM, respectively.

Full text of DOI:10.1016/j.ejmech.2013.01.041

European Journal of Medicinal Chemistry 62 (2013) 754e763
Contents lists available at SciVerse ScienceDirect
European Journal of Medicinal Chemistry
journal homepage: http://www.elsevier.com/locate/ejmech
Original article
Structure-based drug design using GPCR homology modeling:
Toward the discovery of novel selective CysLT2 antagonists
Xiaowu Dong a, Yanmei Zhao a, Xueqin Huang b, Kana Lin b, Jianzhong Chen a,
,
a,
*
Erqing Wei b, Tao Liu a , Yongzhou Hu
*
a ZJU-ENS Joint Laboratory of Medicinal Chemistry, College of Pharmaceutical Sciences, Zhejiang University, Hangzhou 310058, China
b Department of Pharmacology, School of Medicine, Zhejiang University, Hangzhou 310058, China
a r t i c l e i n f o
a b s t r a c t
Article history:
3D structure of CysLT2 receptor was constructed by using homology modeling and molecular simula-
tions. The binding pocket of CysLT2 receptor and the proposition of the interaction mode between CysLT2
and HAMI3379 were identified. A series of dicarboxylated chalcones was then virtually evaluated
through molecular docking studies. A total of six compounds 13aef with preferable scores was further
synthesized and tested for CysLT2 antagonistic activities by determination of the cytosolic free Ca2þ
levels in HEK293 cells. Compounds 13e and 13f exhibited potent and selective CysLT2 antagonistic ac-
Received 15 November 2012
Received in revised form
16 January 2013
Accepted 24 January 2013
Available online 9 February 2013
tivities with IC50 values being 7.5 and 0.25
m
M, respectively.
Ó 2013 Elsevier Masson SAS. All rights reserved.
Keywords:
G-protein-coupled receptors
CysLT2 antagonists
Homology model
Molecular docking
Structure-based design
1. Introduction
may relate to cerebral spinal fluid (CSF) circulation in leukotrienes-
dependent vascular reactions, thus, being a potential target for
treatment of cardiovascular disease. However, there have been no
reports delineating the structureeactivities relationships of
HAMI3379 derivatives and possible interaction modes between
CysLT2 and HAMI3379.
Cysteinyl leukotrienes (CysLTs) LTC4, LTD4, and LTE4 are in-
flammatory mediators derived from arachidonic acid. They are
generated in response to different immune and inflammatory
stimuli [1e4], and exert biological effects via two major classes of
receptors: CysLT1 and CysLT2 [5]. The physiological roles of CysLT1
receptor, which are coherent with anti-bronchoconstrictive and
anti-inflammatory activities of CysLT1 antagonists, such as pran-
lukast [6], montelukast [7] and zafirlukast [8], are well documented
[9] (Fig. 1A). In contrast, the pharmacological role of CysLT2 re-
ceptor is yet less defined due to the lacking of specific antagonists.
Several previously identified nonselective CysLT1/CysLT2 receptor
antagonists, such as BAY u9773 [10] and DUO-LT [11], were com-
monly used as tools to characterize the physiological roles of
CysTL2 receptor, while the poor selectivity and weak potency of
these antagonists hampered further progresses. HAMI3379 (Fig.1B)
was recently reported as the potent and selective CysLT2 receptor
antagonist that can effectively reverse the LTC4-induced perfusion
pressure increase and contractility decrease in isolated
Langendorff-perfused guinea pig hearts [12]. Thus, CysLT2 receptor
Although the X-ray crystal structure of CysLT2 has not been
revealed yet, the bovine rhodopsin (bRh)-based homology model-
ing has been a well-established approach to study 3D structures of
the GPCR receptors [13,14]. Previously, site-directed mutagenesis
has been applied to the elucidation the structures of P2Y receptors
[15,16], which showed close structural and phylogenetic relation-
ships between P2Ys and CysLTs [17]. Recently, Parravicini et al.
reported the molecular modeling of GPR17 receptor, a G-protein-
coupled receptor located at intermediate phylogenetic position
between two distinct receptor families: the P2Ys and CysLTs re-
ceptors [18,19]. These methods and results provided useful guid-
ance in the construction of 3D structure of CysLT2 receptor.
A series of monocarboxylated chalcones (e.g. compounds 7 and
8, Fig. 2A) was previously identified as good CysLT1 antagonists
[20], and none of them exhibited CysLT2 antagonistic activities
(Supplementary Table S1). This study was consistent with reported
result that CysLT2 receptor was not sensitive to classical CysLT1
antagonists [21,22]. A review of the structures of reported CysLT2
* Corresponding authors. Tel./fax: þ86 571 88208460.
E-mail addresses: lt601@zju.edu.cn (T. Liu), huyz@zju.edu.cn (Y. Hu).
0223-5234/$ e see front matter Ó 2013 Elsevier Masson SAS. All rights reserved.
http://dx.doi.org/10.1016/j.ejmech.2013.01.041

X. Dong et al. / European Journal of Medicinal Chemistry 62 (2013) 754e763
755
Fig. 1. Structures of CysLT1 antagonists (A) and CysLT2 antagonists (B).
antagonists revealed several common features among CysLT2 an-
tagonists that are concluded and highlighted as shown in Fig. 2B.
Thus, two carboxyl groups were introduced onto the A-ring of
chalcones and a lipophilic chain was introduced on the B-ring of
chalcones to give dicarboxylated chalcones as potential CysLT2
antagonists (Fig. 2C). Herein, with the aim to enhance accuracy and
efficiency in the development of novel CysLT2 antagonists, 3D
structure of CysLT2 was constructed through homology modeling
and dynamic simulations. The binding pocket of CysLT2 was
defined and the interaction mode between CysLT2 and HAMI3379
was proposed. The newly designed dicarboxylated chalcones were
then virtually evaluated by molecular docking and preferable
screened hits 13aef were synthesized and biologically evaluated
for CysLT2 antagonistic activities.
alignment tool in Discovery studio 2.5 (Accelrys, Inc. San Diego, CA).
Manual multiple sequence alignments (Fig. 3) were then conducted
to locate the homology aligned regions taking into account both
primary and secondary structures according to reported literature
[18,23], in which the authors constructed a reliable homology
model for GPR17, a G-protein-coupled receptor located at the in-
termediate phylogenetic position between two distinct receptor
families: P2Ys and CysLTs receptors. Moreover, two conserved
disulphide bridges (Cys111eCys187 and Cys31eCys279) were
proposed, since that these disulfide bonds were considered as an
important factor in constraining the conformation of the extrac-
ellular and transmembrane domains of the receptor. Finally, ho-
mology modules of the Discovery Studio 2.5 program were applied
to build an initial approximate 3D structure of CysLT2 receptor,
while the first high resolution structure of bovine rhodopsin (PDB
ID: 1U19) [24] was used as a template.
2. Computational methods
2.1. Homology modeling
2.2. Hydrophobic profile analysis
The amino acidic sequences of CysLT2, bRh and pharmacological
related CysLT1 and P2Y1,4,6 receptors were obtained from Swiss-
Prot/TrEMBL database (http://expasy.org/). At first, we aligned the
CysLT2 sequence with the other GPCRs’ sequences using automatic
Since the hydrophobic residues for GPCRs tend to be buried in
the interior of the protein and the hydrophilic residues are more
exposed to cellular membrane, a profile of these values can indicate
the overall folding pattern. Herein, we applied KyteeDoolittle (KD)
Fig. 2. Rational design and development of dicarboxylated chalcone derivatives as CysLT2 antagonists, A) the structure of identified monocarboxylated chalcones 7 and 8 as CysLT1
antagonists in our previous study [20]; B) the common structural features of CysLT2 antagonists; C) the evolution of dicarboxylated chalcone derivatives as novel specific CysLT2
antagonists.

756
X. Dong et al. / European Journal of Medicinal Chemistry 62 (2013) 754e763
Fig. 3. Multiple alignment of the sequences of bovine rhodopsin (bRh), CysLT2, CysLT1 and P2Y1,4,6 receptors. The cysteines involved in the formation of the two assumed
disulphide bridges are highlighted in blue; the conserved residues and motif are highlighted in green.(For interpretation of the references to colour in this figure legend, the reader
is referred to the web version of this article.)
method [25], a widely used technique for detecting hydrophobic
regions in proteins, to perform the hydrophobic profile analysis of
the identified transmembrane regions by visualizing the hydro-
phobicity of CysLT2 along its sequence. As defined in the Kytee
Doolittle scale, the hydrophobicity is calculated by sliding a fixed
size window (size ¼ 9) over the protein sequence. At the central
position of the window, the average hydrophobicity of the entire
windows is plotted.
molecules, for a total of 52,279 atoms in a rectangular box of
62 ꢀ 78 ꢀ 92 A.
ꢀ
2.4. Molecular dynamics simulations
The final ensemble was submitted to energy minimization cy-
cles, followed by simulated annealing in order to lead the system to
a more favorable energetic condition before starting the pure MD
simulation, the backbone of the seven TM helices and the struc-
tured EL-2 motifs were constrained to maintain the overall
arrangement of the helical bundle and the structural conserved
organization of EL-2. The protocol by which the assembly was
prepared for the MD run was composed of separated cycles as
described below. For the earlier minimization steps, the steepest
descent algorithm was applied. The minimization sequence of the
various component of the system was the following: first the lipids,
then the water, then both lipids and water, and finally the whole
system. Then, further the conjugate gradient algorithm was used to
minimize the whole system, and then a first run of 200 ps of MD
simulation at 5 K was performed. At the end of this first mini-
mization and relaxation protocol, a simulated annealing procedure
was performed as follows. The system was heated from 5 to 310 K in
400 ps, and kept at this temperature for further 240 ps. Then, an
MD simulation with CHARMM force field was performed at
2.3. Protein/membrane complex building process
The crude model of CysLT2 receptor was locally minimized us-
ing CHARMM force field [26] in vacuo by constraining the backbone
of the helices in order to give a first optimization of the rough ge-
ometry derived from homology modeling. The model of DPPC in
the liquid-crystalline phase proposed by Tieleman and Berendsen
[27,28] was used to produce the membrane environment. The
structure of CysLT2 was inserted in the centre of the 128 lipid
bilayer using membrane Builder in the CHARMM-GUI website [29],
in such a way that the principal axes of the helical bundle was
parallel to the membrane axis (z) and perpendicular to the mem-
brane plane (xy), and the extracellular and intracellular loops were
at the lipid interface. The resulting complex system consisted of
346 amino acid residues, 122 DPPC molecules, 10,305 water

X. Dong et al. / European Journal of Medicinal Chemistry 62 (2013) 754e763
757
a temperature of 310 K for a total of 3 ns of dynamics simulation
without any constraints.
run with CHARMM force field for 500 iterations of steepest de-
scents, followed by a 500 iterations conjugate gradients opti-
mization. The other parameters of MD/MM simulation are
maintained at their Discovery studio default configuration.
2.5. Molecular docking and MM/MD simulation
3. Results and discussion
2.5.1. Molecular docking with LigandFit [30] program
The docking site of CysLT2 receptor was derived using the
LigandFit site search utility in Discovery studio 2.5 [30]. For the
generation of the ligand’s conformations, we used variable num-
bers of Monte Carlo simulations. All the calculations during the
docking steps were performed under the PLP Force Field formalism.
A short rigid body minimization was then performed and 50 poses
for each ligand were saved. Scoring was performed with a set of
scoring functions (including Dock_score, -PLP1 and -PMF) imple-
mented in LigandFit module. The combination of consensus scoring
method and the interaction mode was applied to select the pref-
erable output conformation.
3.1. The initial crude structure of CysLT2 receptor
3.1.1. Seven TM helices bundle
As shown in Fig. 4, the results from KyteeDoolittle (KD) analysis
indicated that the hydrophobic properties of the amino acid resi-
dues of CysLT2 were consistent with the structural features of seven
transmembrane domains. The amino acid residues from seven TM
helices (from TM1 to TM7) are much more hydrophobic, since that
these residues tend exposed to cellular membrane and result in
formation of a complex weave of polar interactions. It has been
demonstrated that there are close structural and phylogenetic re-
lationships between the P2Ys and CysLTs families which have been
characterized by consisting of the seven TM helices bundle (TM1-
7), an extracellular N-terminus region (NT), a cytoplasmic C-ter-
minus tail (CT) and three extracellular (ECs) and three intracellular
(ICs) loops [33,34].
2.5.2. Molecular docking with Flexidock [31] program
ꢀ
The binding pocket was defined all residues within 4 A of the
ligand within the CysLT2 complex proposed by LigandFit. All single
bonds of residue side chains inside the binding pocket were
regarded as rotatable or flexible. The docked ligand was allowed to
rotate on all of single bonds and move flexibly within the tentative
binding pocket. The obtained enzymeeligand complex was further
minimized with 500 iterations of steepest descent and followed by
a conjugate gradients optimization by using CHARMM force field
within Discovery Studio 2.5 program. Then, the binding energy
[32] between CysLT2 receptor and its bound ligand was calculated
using “Calculate binding energies module” in Discovery studio 2.5
program
3.1.2. Conserved residues and motifs
The amino acid sequences of both CysLT1 and CysLT2 receptors
have only 24e32% identity to members of the purinergic (P2Y)
receptor family, however, several conserved structural motifs of
these nucleotide receptors (e.g. CysLTs and P2Ys receptors) were
readily found, especially in the nucleotides bound domains (EC1,
TM2, TM3 and TM6) [23,35]. The conserved domains of these re-
ceptors are highlighted in Fig. 3, including a WXFG motif in the first
extracellular loop (EC1) of each receptor, a hydrophobic motif
(VXMXNLAXXDLL) in the second transmembrane domain (TM2),
a YVNXY motif in the third transmembrane domain (TM3) and
a conserved HXXRT motif in the sixth transmembrane domain
(TM6). Therefore, all of them were taken into consideration in
manual multi-sequence alignment, homology modeling and mo-
lecular simulation.
Eint ¼ EðA; BÞ ꢁ ðEðAÞ þ EðBÞÞ;
where E(A) and E(B) are the energies of CysLT2 and ligand respec-
tively, and E(A,B) the energy of docked complex.
2.5.3. MD/MM simulation on HAMI3379-bound CysLT2 complex
The HAMI3379-bound CysLT2 complex proposed by Flexidock
study was further refined by molecular dynamic and mechanical
(MD/MM) calculations, with aim to examine the reliability of the
interaction mode of HAMI3379: (a) An NVT ensemble molecular
dynamic simulations with CHARMM force field were then per-
formed at a constant temperature of 300 K with a time step of 1 fs
for a total of 1.5 ns, while atomic constraints of backbone of CysLT2
were applied to retain; (b) the output conformation was initially
3.1.3. Disulfide bond bridge
The disulfide bonds play important role in constraining the
conformation of extracellular domains of the protein and stabilize
the transmembrane structures [36,37]. As commonly found in
others’ GPCR receptors, the outmost part of TM3 of CysLT2 seemed
to be permanently engaged in a conserved disulphide bridge with
Fig. 4. Plot of hydrophobicity along the amino acid sequence for CysLT2 receptor by using KyteeDoolittle analysis, window size ¼ 9, max ¼ 3.50, min ¼ ꢁ2.75.

758
X. Dong et al. / European Journal of Medicinal Chemistry 62 (2013) 754e763
3.2. Homology model refined by molecular simulation
3.2.1. Molecular dynamics within protein/membrane complex
The locally minimized structure was embedded in a fully hy-
drated phospholipidic bilayer (dipalmitoylphosphatidylcholine,
DPPC, hydrated with water). The structure of the proteinelipidse
solvent system derived from the simulated annealing simulation
was used as input for 3 ns of molecular dynamics. After an initial
decrease, the total energy of the system kept a stable trend, indi-
cating that an equilibrium state had been reached (Supplementary
Fig. S1). The final picture of the protein after 3 ns of MD is shown in
Fig. 5. It reveals that the 346 amino acids of CysLT2 fold into seven
transmembrane helices, three extracellular and three intracellular
loops, as well as a cytoplasmic helix. The inter-helix hydrogen bond
and non-bonded interactions were observed that are essential for
the stabilization of the seven transmembrane domains of CysLT2
receptor structure. For example, TM3 positions itself in close con-
tact with the TM2, TM4, TM5 and TM7 through inter-helix hydro-
gen bonds (e.g., Ser117eLeu91, Asn121eAsp84, Tyr95eGly181,
Tyr99eVal180, Asn93eAsn269). Other hydrogen bond in-
teractions were also detected including TM1 with adjacent TM7
(Try17eLeu263, Asn28eAla270), TM2 with adjacent TM3, TM4 and
TM7 (Ser117eLeu91, Asn121eAsp84, Leu52eTrp135, Asp56e
Asn273). Besides, the hydrogen bonds between EL-2 and TM1,
TM3 and TM6 (e.g. Ser155eArg10, Cys159eMet86, Glu161eTyr91,
Leu162eArg267, Asn163e His242) play important role in stabili-
zation of the seven transmembrane domains. Also, non-bonded
inter-helix interactions, such as the hydrophobic interaction be-
tween non-polar side chains of each pair of adjacent helices,
Fig. 5. Typical structure of CysLT2 embedded in the fully hydrated lipid bilayer. A
frame of the system extracted from the 3 ns MD simulations is shown. The helix and
loops of CysLT2 are represented in green, the DPPC are in orange and white, water is in
red/blue.(For interpretation of the references to colour in this figure legend, the reader
is referred to the web version of this article.)
including
pep interaction (e.g. Phe41eTyr45, Phe87eTyr116e
Tyr163, Phe93eTyr97ePhe106, Phe214ePhe257 and Tyr301e
Phe312eTyr302) and lots of hydrophobic alkylealkyl interaction
(e.g. Leu129eLeu117eLeu253, Val131eLeu158eIle162, Leu132e
Ile249, Val53eLeu85eIle60, Leu206eLeu271eThr268, Leu224e
Ile250eLeu244 and Ile251eIle255eLeu304).
EC2 of CysLT2 involving the Cys111 and Cys187 residues. In addi-
tion, CysLT2 has another pair of cysteines which are conserved in all
the P2Ys and CysLTs receptors; these two residues are positioned at
the end of the Nt (Cys31) and at the middle of the EC3 domain
(Cys279), respectively. It has been demonstrated by site-directed
mutagenesis and ligand affinity data that corresponding cysteines
in P2Y1 form a disulphide bridge which is important for receptor
activation [38]. Thus, we also included this additional disulphide
bridge into the 3D structure of CysLT2 receptor.
3.2.2. Definition of the binding site
On the basis of the homology-constructed CysLT2 structure,
LigandFit/active site finding module was applied to find the most
promising binding domains, which is located in a cavity among the
TM1, TM3, and TM5eTM7 on the extracellular side of the seven TMs
bundles. This cavity consists of a hydrophilic center pointed toward
Fig. 6. Model of the complex formed by CysLT2 receptor and HAMI3379 after 1 ns of MD simulation. (A) HAMI3379 is displayed as spheres in the schematic representation of the
entire ligandereceptor complex; (B) HAMI3379 is displayed in stick within the binding pocket of CysLT2 in the detailed picture, and the polar interactions between HAMI3379 and
CysLT2 are highlighted using yellow dash line.(For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

X. Dong et al. / European Journal of Medicinal Chemistry 62 (2013) 754e763
759
the TM6 and EL-2 domains (His264, Arg267 and Glu189) and a large
hydrophobic cleft surrounded by hydrophobic residues (Leu118,
Ala295, Ala296, Ala298, Cys299 and Ile48). The distance between
the hydrophilic and hydrophobic centers of the identified cavity is
ꢀ
estimated to be 15e17 A, which is approximately typical size of
a CysLT2 antagonist, such as the HAMI3379. Furthermore, the
defined binding pocket is also congruent with the biochemical
studies, by comparing with observed characteristics of other
transmembrane receptors (e.g. bovine Rhodopsin [37] and P2Y1
[38]). As found in P2Y1, both His277 and Lys280 are essential for
ligand recognition and/or receptor activation [15,39]. In our con-
structed CysLT2 structure, these two crucial residues are His264
and Arg267, which would be critical to the binding affinity of
CysLT2 ligands. The interaction analysis between TM6 and adjacent
helices also validated the binding cavity of CysLT2, since that TM6
engaged only few interactions with other helices. It is suggested
that this helix may maintain a dynamic behavior needed to evoke
receptor activation without constraints from the other helices. In
fact, in bRh, TM6 is believed to move away from TM3 thereby
starting the activation process [40].
Fig. 7. Structures of promising hits 13aef identified by docking studies.
structural diversity and synthetic feasibility. The docking results of
them are shown in Table 1. All of them show good consensus re-
sults, with good performance in more than three scoring functions.
As exemplified by 13f, the best one predicted in docking studies, the
dock scores, -PLP1 and -PMF (Dock_score: 136.6; -PLP1: 145.9 and
-PMF: 191.9) obtained from LigandFit are promising and compa-
rable to that of HAMI3379 (Dock_score: 153.0; -PLP1: 164.9 and
-PMF: 209.0). Also, the result of Flexidock indicated that the com-
pound 13f showed the lowest interaction energy (ꢁ411.7 kcal/mol),
which is even lower than that of HAMI3379 (ꢁ377.0 kcal/mol).
Furthermore, the synthesis and biological evaluation of screened
hits 13aef with good docking results were performed to validate
robustness and predictive ability of the homology model as well as
molecular docking.
3.2.3. The interaction mode between CysLT2 and HAMI3379
The overall picture of the CysLT2 and HAMI3379 complex
(Fig. 6A) was obtained by means of molecular dynamics (MD) runs
of 1 ns starting from the best-docked conformation of CysLT2 and
HAMI3379 proposed by LigandFit and followed by Flexidock. A total
energy of the system kept a stable trend, indicating that an equi-
librium state had been reached (Supplementary Fig. S2). The pu-
tative arrangement of HAMI3379 in CysLT2 binding site is displayed
in Fig. 6B. The guanidine group of Arg267 seems to form an ionic
bond with carboxyl on the phenyl group of HAMI3379 (CaOe
3.4. Chemistry
The synthetic route for compound 13 is outlined in Scheme 1.
ClaiseneSchmidt condensation of acetophenone 9 with appro-
priate benzaldehyde 10 in the presence of 10% potassium hydroxide
in a solvent of EtOH-THF-H2O (5:5:2) for 5e6 days, following by the
esterization with ethanol in the presence of sulfuric acid afforded
chalcone 11. Alkylation of 11 with ethyl bromoacetate in the pres-
ence of potassium carbonate and acetone gave in 12, which was
further alkaloid hydrolyzed using 5% sodium hydroxide in EtOHe
H2O mixture solvent at room temperature to get dicarboxylated
chalcone 13.
ꢀ
Arg267: 2.89A), while the imidazol group of His264 seems to
form another ionic bond interaction with carboxyl group on
ꢀ
cyclohexane moiety of HAMI3379 (CeOeHis264: 2.97A). A hydro-
gen bond between the carboxyl group on the phenyl group of
ꢀ
HAMI3379 with the hydroxyl group of Tyr119 (CaOeTyr119: 2.02A)
was observed to form another hydrophilic interaction. In addition
to the polar interactions, hydrophobic contacts were also found
within the pocket: these involved hydrophobic resides Leu291,
Leu93, Leu118, Ala295, Ala296, Ala298 and Cys299, and aromatic
residues Try119, Tyr263 and Tyr45. Although holding some peculiar
features, the overall configuration of the bound ligand agreed with
the general configuration reported in previous computational study
for P2Y receptors [41]. Furthermore, these interactions (such as
polar and non-polar interactions) were further analyzed in the time
evolution of HAMI3379-bound CysLT2 complex using molecular
dynamic simulation (Supplementary Fig. S3AeC), showing all of
the interactions are stable and would be essential for ligand
binding.
3.5. Pharmacological activities
The CysLT2 antagonistic potency of the screened hits 13aef was
evaluated by determination of cytosolic free Ca2þ levels in HEK293
cells which were stably transfected with pcDNA3.1 (þ)-hCysLT2. In
this biological evaluation model, PT-PCR and Western blot analysis
showed a higher expression of CysLT2 receptor in HEK293 cells
(Supplementary Fig. S4). ATP at 50
and was used as a positive control for measurement of [Ca2þ]i. LTD4
at 0.1
M also significantly increased [Ca2þ]i, which could be par-
m
M significantly elevated [Ca2þ]i,
m
3.3. Evaluation of dicarboxylated chalcones through molecular
docking
tially blocked by Bay u9773 (a CysLT1 and CysLT2 dual antagonist)
but could not blocked by selective CysLT1 receptor antagonists
Once the HAMI3379-bound CysLT2 complex was obtained, the
newly designed dicarboxylated chalcones were virtually evaluated
using molecular docking by combination of LigandFit and Flex-
idock. At first, the LigandFit was used to fast examine the fitness
between the shape of the dicarboxylated chalcones and topological
feature of the binding pocket of CysLT2. Then, the Flexidock, an
advanced molecular docking program, was applied to investigate
more accurate binding mode between CysLT2 and screened hits. It
is indicated that 4(or 5)-carboxyl group on A-ring and more than
eight atomic alkyl chain on B-ring of chalcones are preferable for
their interacting with CysLT2. Then, a total of six promising hits
13aef (Fig. 7) was picked out considering the good docking scores,
Table 1
The docking results of screened hits 13aef and reference HAMI3379.
Compd.
LigandFit
Binding energy
(kcal/mol)
Dock_score
-PLP1
-PMF
HAMI3379
13a
13b
13c
13d
153.0
119.9
122.8
120.8
128.9
127.2
136.6
164.9
125.4
131.2
133.4
134.9
130.1
145.9
209.0
188.0
175.4
166.2
184.1
205.9
191.9
ꢁ377.0
ꢁ312.7
ꢁ378.8
ꢁ390.4
ꢁ262.6
ꢁ298.8
ꢁ411.7
13e
13f

760
X. Dong et al. / European Journal of Medicinal Chemistry 62 (2013) 754e763
Scheme 1. Synthesis of dicarboxylated chalcones 13aef. Reagents and conditions:(a) appropriate alkyl bromide, K2CO3, acetone; (b) 10% KOH, ethanol/THF/H2O, r.t.; (c) H2SO4,
EtOH; (d) BrCH2COOEt, K2CO3, acetone, reflux; (e) 5% NaOH, EtOH/H2O.
montelukast and pranlukast (CysLT1 antagonists). As shown in
Fig. 8A, the antagonistic effects of tested compounds 13aef at
and can remarkably improve efficiency in lead optimization.
Although monomonocarboxylated chalcones haven been reported
as CysLT1 antagonists [20,42], it is also interestingly that the
dicarboxylated chalcones are the firstly reported as CysLT2
antagonists.
a concentration of 1 mM were observed, especially for compounds
13e and 13f, which show 44.0% and 70.5% decreased [Ca2þ]i by
blocking the agonistic effect of LTD4, respectively. As shown in
Fig. 8B, the antagonistic effects of 13e and 13f were observed in
a dose-dependent manner with the maximal effect (13e and 13f:
72.8% and 80.5%, respectively), and their half antagonistic concen-
3.6. Docking mode analysis of compound 13f
tration are also good (IC50of 13e and 13f: 7.5 and 0.25 mM, respec-
The interaction mode of compound 13f proposed by molecular
docking studies is given in Fig. 9A and B. The binding pocket was
rendered from the molecular surface color-coded by hydropho-
bicity (brown to blue: a scale of hydrophobic to hydrophilic prop-
erties). Overall, the interaction mode of 13f is consistent with that
of HAMI3379.A hydrophilic region (His264, Arg267 and Glu189)
and a large hydrophobic cleft (Leu118, Ala295, Ala296, Ala298,
Cys299 and Ile48) of the binding pocket of CysLT2 tightly embrace
compound 13f and form polar and non-polar interactions. The
guanidine group of Arg267 and the imidazol group of His264 seem
to form ionic bonds with two carboxyl group of 13f (CaOeArg267:
tively). In order to examine the selective profiles of the tested
compounds 13aef against CysLT1 and CysLT2 receptors, the CysLT1
antagonistic activities of them were evaluated according to our
previous method [20]. The results demonstrated that the most
promising compounds 13e and 13f also show good selectivity for
CysLT2 receptor, while they exhibit very low antagonistic potency
on CysLT1 receptor (Supplement Table S2). Thus, as expected, the
introduction of two carboxyl groups onto the A-ring of chalcones is
preferable for CysLT2 antagonistic activities, by comparing with our
previous monocarboxylated chalcones (e.g. compounds 7 and 8). By
optimization of the length and substituents of the lipophilic chain
on B-ring of chalcones can significantly modulate the CysLT2
antagonistic activities. Obviously, the homology model based
structural modification of CysLT2 antagonists is a rational protocol,
ꢀ
ꢀ
2.39A, CaOeHis264: 3.09A). A further hydrogen bond is possibly
established between the carboxyl oxygen atom of 13f and the hy-
ꢀ
droxyl oxygen atom of Tyr119 (CaOeTyr119: 2.23A). Moreover, the
presence of a significant hydrophobic cleft (Leu291, Leu93, Leu118,
Fig. 8. The CysLT2 antagonistic activities of tested compounds on the LTD4 (0.1
The CysLT2 antagonistic activities of dicarboxylated chalcones 13aef, montelukast, pranlukast and bay u9773 at a concentration of 1
antagonistic activities of 13e and 13f.
m
M)-elevated [Ca2þ]i in HEK293 cells which were stably transfected with pcDNA3.1 (þ)-hCysLT2. A)
m; B) the concentrationedependent CysLT2

X. Dong et al. / European Journal of Medicinal Chemistry 62 (2013) 754e763
761
Fig. 9. Interaction mode of 13f (CysLT2 antagonistic IC50 ¼ 0.25
mM) with CysLT2 proposed by molecular docking. A) Compound 13f was embraced in the CysLT2 binding pocket,
which was rendered from the molecular surface and color-coded by hydrophobicity (brown to blue: a scale of hydrophobic to hydrophilic properties); B) 13f is displayed in stick
within the binding pocket of CysLT2 in the detailed picture, and the polar interactions between l3f and CysLT2 are highlighted using yellow dash line.(For interpretation of the
references to colour in this figure legend, the reader is referred to the web version of this article.)
Ala295, Ala296, Ala298 and Cys299, Try45) in the CysLT2 binding
site that is capable of accommodating a long hydrophobic chain on
B-ring of chalcone. In addition, the aromatic stack interaction be-
tween the phenol ring of Tyr119 and A-ring of 13f also serves to
increase the thermal stability of the complex.
the homology model of CysLT2 receptor and its application in
structure-based lead optimization are reported for the first time.
Also, the discovery of novel dicarboxylated chalcones, such as
compounds 13e and 13f, as CysLT2 antagonists would be interest-
ing leads to do further structureeactivity exploration and struc-
turally optimization.
4. Conclusion
5. Experimental
The 3D structure of CysLT2 was established using homology
modelling and applied in virtually evaluation of newly designed
dicarboxylated chalcones using molecular docking. This approach
led to the identification of six promising screened hits 13aef, which
were synthesized and biologically evaluated for CysLT2 antago-
nistic activities. The identification of compounds 13e and 13f with
good and selective CysLT2 antagonistic activities indicated that
combination of homology model and structure-based development
of novel specific CysLT2 antagonists is a rational strategy. The
interaction mode between CysLT2 and its ligands proposed in
present study would be useful and valuable for further structural
optimization of CysLT2 antagonists. To the best of our knowledge,
5.1. Chemistry
For experimental procedures and spectral characterization data
see the Supplementary Material.
5.2. CysLT2 antagonistic activities
The CysLT2 antagonistic potency of the test compounds was
evaluated by determination of cytosolic free Ca2þ levels in HEK293
cells which were stably transfected with pcDNA3.1 (þ)-hCysLT2
[43]. In brief, cells were seeded in 96-well culture plates (BMG

762
X. Dong et al. / European Journal of Medicinal Chemistry 62 (2013) 754e763
LABTECHNOLOGIES 96) and grown at 37 ꢂC in 5% CO2-containing
humidified air. Twenty-four hours after plating, the growth me-
[15] Q. Jiang, D. Guo, B.X. Lee, A.M. Van Rhee, Y.C. Kim, R.A. Nicholas, J.B. Schachter,
T.K. Harden, K.A. Jacobson, A mutational analysis of residues essential for
ligand recognition at the human P2Y1 receptor, Mol. Pharmacol. 52 (1997)
499e507.
[16] L. Erb, R. Garrad, Y. Wang, T. Quinn, J.T. Turner, G.A. Weisman, Site-directed
mutagenesis of P2U purinoceptors. Positively charged amino acids in trans-
membrane helices 6 and 7 affect agonist potency and specificity, J. Biol. Chem.
270 (1995) 4185e4188.
[17] R. Fredriksson, M.C. Lagerstrom, L.G. Lundin, H.B. Schioth, The G-protein-
coupled receptors in the human genome form five main families. Phylogenetic
analysis, paralogon groups, and fingerprints, Mol. Pharmacol. 63 (2003)
1256e1272.
dium was removed and replaced with 50
140, KCl 4, MgCl2 1, CaCl2 1.25, HEPES 5, glucose 11, NaH2PO4 1,
ascorbic acid 5.7; pH 7.4) containing 1 M Fluo4 AM and 0.03%
pluronic F-127 for 45 min at 37 ꢂC. After loading, Fluo4 AM was
removed and cells were washed three times with BSS. Then 40
ml BSS (BSS in mM: NaCl
m
ml
BSS containing tested compounds or not were added to the wells
and cells were further incubated for 30 min. After that, cells were
transferred to the FLUOstar OPTIMA (BMG Germany) and 50
ATP which as a positive stimulus or 0.1
mM
[18] C. Parravicini, G. Ranghino, M.P. Abbracchio, P. Fantucci, GPR17: molecular
modeling and dynamics studies of the 3-D structure and purinergic ligand
m
M LTD4 was added to the
binding features in comparison with P2Y receptors, BMC Bioinform.
(2008) 263.
9
wells. Calcium concentration ([Ca2þ]i) was measured as the fluo-
rescence monitored at 37 ꢂC (485 nm excitation, 525 nm emission)
and the LTD4 response was obtained by calculating peak minus
basal fluorescence values, and reported as percentages of control
values (BSS only).
[19] I. Eberini, S. Daniele, C. Parravicini, C. Sensi, M.L. Trincavelli, C. Martini,
M.P. Abbracchio, In silico identification of new ligands for GPR17: a promising
therapeutic target for neurodegenerative diseases, J. Comput. Aided Mol. Des.
25 (2011) 743e752.
[20] X. Dong, L. Wang, X. Huang, T. Liu, E. Wei, L. Du, B. Yang, Y. Hu, Pharmaco-
phore identification, synthesis, and biological evaluation of carboxylated
chalcone derivatives as CysLT1 antagonists, Bioorg. Med. Chem. 18 (2010)
5519e5527.
Acknowledgements
[21] C.E. Heise, B.F. O’Dowd, D.J. Figueroa, N. Sawyer, T. Nguyen, D.S. Im, R. Stocco,
J.N. Bellefeuille, M. Abramovitz, R. Cheng, D.L. Williams Jr., Z. Zeng, Q. Liu,
L. Ma, M.K. Clements, N. Coulombe, Y. Liu, C.P. Austin, S.R. George, G.P. O’Neill,
K.M. Metters, K.R. Lynch, J.F. Evans, Characterization of the human cysteinyl
leukotriene 2 receptor, J. Biol. Chem. 275 (2000) 30531e30536.
[22] J. Takasaki, M. Kamohara, M. Matsumoto, T. Saito, T. Sugimoto, T. Ohishi,
H. Ishii, T. Ota, T. Nishikawa, Y. Kawai, Y. Masuho, T. Isogai, Y. Suzuki,
S. Sugano, K. Furuichi, The molecular characterization and tissue distribution
of the human cysteinyl leukotriene CysLT(2) receptor, Biochem. Biophys. Res.
Commun. 274 (2000) 316e322.
[23] E.A. Mellor, A. Maekawa, K.F. Austen, J.A. Boyce, Cysteinyl leukotriene receptor
1 is also a pyrimidinergic receptor and is expressed by human mast cells, Proc.
Natl. Acad. Sci. U.S.A. 98 (2001) 7964e7969.
[24] T. Okada, M. Sugihara, A.N. Bondar, M. Elstner, P. Entel, V. Buss, The retinal
conformation and its environment in rhodopsin in light of a new 2.2 Å crystal
structure, J. Mol. Biol. 342 (2004) 571e583.
The authors are grateful to the support provided by the National
Natural Science Foundation of China (No. 81001359) and the
China Postdoctoral Science Foundation (No. 201003734) and the
Fundamental Research Funds for the Central Universities
(No. 2011QNA7010).
Appendix A. Supplementary data
Supplementary data related to this article can be found at
http://dx.doi.org/10.1016/j.ejmech.2013.01.041.
References
[25] J. Kyte, R.F. Doolittle, A simple method for displaying the hydropathic char-
acter of a protein, J. Mol. Biol. 157 (1982) 105e132.
[26] B.R. Brooks, R.E. Bruccoleri, B.D. Olafson, D.J. States, S. Swaminathan,
M. Karplus, CHARMM e a program for macromolecular energy, minimization,
and dynamics calculations, J. Comput. Chem. 4 (1983) 187e217.
[27] D.P. Tieleman, H.J.C. Berendsen, molecular dynamics simulations of a fully
hydrated dipalmitoylphosphatidylcholine bilayer with different macro-
scopic boundary conditions and parameters, J. Chem. Phys. 105 (1996)
4871e4880.
[28] D.P. Tieleman, S.J. Marrink, H.J. Berendsen, A computer perspective of mem-
branes: molecular dynamics studies of lipid bilayer systems, Biochim. Bio-
phys. Acta 1331 (1997) 235e270.
[1] S. Poulin, C. Thompson, M. Thivierge, S. Veronneau, S. McMahon, C.M. Dubois,
J. Stankova, M. Rola-Pleszczynski, Cysteinyl-leukotrienes induce vascular
endothelial growth factor production in human monocytes and bronchial
smooth muscle cells, Clin. Exp. Allergy 41 (2011) 204e217.
[2] E. Mayatepek, G.F. Hoffmann, Leukotrienes: biosynthesis, metabolism, and
pathophysiologic significance, Pediatr. Res. 37 (1995) 1e9.
[3] K.F. Chung, Leukotriene receptor antagonists and biosynthesis inhibitors:
potential breakthrough in asthma therapy, Eur. Respir. J. 8 (1995) 1203e1213.
[4] H.E. Claesson, S.E. Dahlen, Asthma and leukotrienes: antileukotrienes as novel
anti-asthmatic drugs, J. Intern. Med. 245 (1999) 205e227.
[29] S. Jo, T. Kim, W. Im, Automated builder and database of protein/membrane
complexes for molecular dynamics simulations, PLoS One 2 (2007) e880.
[30] C.M. Venkatachalam, X. Jiang, T. Oldfield, M. Waldman, LigandFit: a novel
method for the shape-directed rapid docking of ligands to protein active sites,
J. Mol. Graph. Model. 21 (2003) 289e307.
[5] R.A. Coleman, R.M. Eglen, R.L. Jones, S. Narumiya, T. Shimizu, W.L. Smith,
S.E. Dahlen, J.M. Drazen, P.J. Gardiner, W.T. Jackson, et al., Prostanoid and
leukotriene receptors: a progress report from the IUPHAR working parties on
classification and nomenclature, Adv. Prostaglandin Thromboxane Leukot.
Res. 23 (1995) 283e285.
[31] T. Ahnfeldt, D. Gunzelmann, T. Loiseau, D. Hirsemann, J. Senker, G. Ferey,
N. Stock, Synthesis and modification of a functionalized 3D open-framework
structure with MIL-53 topology, Inorg. Chem. 48 (2009) 3057e3064.
[32] R. Vijayalakshmi, V. Subramanian, B.U. Nair, A study of the interaction of
Cr(III) complexes and their selective binding with B-DNA: a molecular mod-
eling approach, J. Biomol. Struct. Dyn. 19 (2002) 1063e1071.
[33] F. Horn, J. Weare, M.W. Beukers, S. Horsch, A. Bairoch, W. Chen, O. Edvardsen,
F. Campagne, G. Vriend, GPCRDB: an information system for G protein-
coupled receptors, Nucleic Acids Res. 26 (1998) 275e279.
[6] S.H. Yoo, S.H. Park, J.S. Song, K.H. Kang, C.S. Park, J.H. Yoo, B.W. Choi,
M.H. Hahn, Clinical effects of pranlukast, an oral leukotriene receptor antag-
onist, in mild-to-moderate asthma: a 4 week randomized multicentre con-
trolled trial, Respirology 6 (2001) 15e21.
[7] G.L. Yu, E.Q. Wei, M.L. Wang, W.P. Zhang, S.H. Zhang, J.Q. Weng, L.S. Chu,
S.H. Fang, Y. Zhou, Z. Chen, Q. Zhang, L.H. Zhang, Pranlukast, a cysteinyl leu-
kotriene receptor-1 antagonist, protects against chronic ischemic brain injury
and inhibits the glial scar formation in mice, Brain Res. 1053 (2005) 116e125.
[8] S.L. Spector, Management of asthma with zafirlukast. Clinical experience and
tolerability profile, Drugs 52 (Suppl. 6) (1996) 36e46.
[34] F. Horn, E. Bettler, L. Oliveira, F. Campagne, F.E. Cohen, G. Vriend, GPCRDB
information system for G protein-coupled receptors, Nucleic Acids Res. 31
(2003) 294e297.
[9] V. Capra, Molecular and functional aspects of human cysteinyl leukotriene
receptors, Pharmacol. Res. 50 (2004) 1e11.
[35] D.T. Major, B. Fischer, Molecular recognition in purinergic receptors. 1. A
comprehensive computational study of the h-P2Y(1)-receptor, J. Med. Chem.
47 (2004) 4391e4404.
[36] S.S. Karnik, H.G. Khorana, Assembly of functional rhodopsin requires a disul-
fide bond between cysteine residues 110 and 187, J. Biol. Chem. 265 (1990)
17520e17524.
[10] M. Back, Functional characteristics of cysteinyl-leukotriene receptor subtypes,
Life Sci. 71 (2002) 611e622.
[11] H. Galczenski, J. Hutchinson, D.J. Figueroa, J. Scheigetz, R. Young, J.F. Evans,
American Thoracic Society, in: 98th International Conference (May 2002,
Atlanta) [D38] [Poster F4] Characterization of DUO-LT: A Novel Dual Cysteinyl
Receptor Antagonist.
[37] K. Palczewski, T. Kumasaka, T. Hori, C.A. Behnke, H. Motoshima, B.A. Fox, I. Le
Trong, D.C. Teller, T. Okada, R.E. Stenkamp, M. Yamamoto, M. Miyano, Crystal
structure of rhodopsin: a G protein-coupled receptor, Science 289 (2000)
739e745.
[12] F. Wunder, H. Tinel, R. Kast, A. Geerts, E.M. Becker, P. Kolkhof, J. Hutter,
J. Erguden, M. Harter, Pharmacological characterization of the first potent and
selective antagonist at the cysteinyl leukotriene 2 (CysLT(2)) receptor, Br. J.
Pharmacol. 160 (2010) 399e409.
[38] C. Gerard, C. Mollereau, G. Vassart, M. Parmentier, Nucleotide sequence of
a human cannabinoid receptor cDNA, Nucleic Acids Res. 18 (1990) 7142.
[39] A.A. Ivanov, I. Fricks, T. Kendall Harden, K.A. Jacobson, Molecular dynamics
simulation of the P2Y14 receptor. Ligand docking and identification of a pu-
tative binding site of the distal hexose moiety, Bioorg. Med. Chem. Lett. 17
(2007) 761e766.
[13] S. Moro, F. Deflorian, G. Spalluto, G. Pastorin, B. Cacciari, S.K. Kim,
K.A. Jacobson, Demystifying the three dimensional structure of G protein-
coupled receptors (GPCRs) with the aid of molecular modeling, Chem. Com-
mun. (Camb) (2003) 2949e2956.
[14] A.M. Van Rhee, B. Fischer, P.J. Van Galen, K.A. Jacobson, Modelling the P2Y
purinoceptor usingrhodopsinastemplate, DrugDes. Discov. 13(1995)133e154.

X. Dong et al. / European Journal of Medicinal Chemistry 62 (2013) 754e763
763
[40] J. Saam, E. Tajkhorshid, S. Hayashi, K. Schulten, Molecular dynamics inves-
tigation of primary photoinduced events in the activation of rhodopsin, Bio-
phys. J. 83 (2002) 3097e3112.
[41] S. Costanzi, B.V. Joshi, S. Maddileti, L. Mamedova, M.J. Gonzalez-Moa,
V.E. Marquez, T.K. Harden, K.A. Jacobson, Human P2Y(6) receptor: molecular
modeling leads to the rational design of a novel agonist based on a unique
conformational preference, J. Med. Chem. 48 (2005) 8108e8111.
[42] M.E. Zwaagstra, H. Timmerman, M. Tamura, T. Tohma, Y. Wada, K. Onogi,
M.Q. Zhang, Synthesis and structure-activity relationships of carboxylated
chalcones: a novel series of CysLT1 (LTD4) receptor antagonists, J. Med. Chem.
40 (1997) 1075e1089.
[43] V. Capra, S. Ravasi, M.R. Accomazzo, M. Parenti, G.E. Rovati, CysLT1 signal
transduction in differentiated U937 cells involves the activation of the small
GTP-binding protein Ras, Biochem. Pharmacol. 67 (2004) 1569e1577.