Mapping histamine H4 receptor-ligand binding modes

Article abstract of DOI:10.1039/c2md20212c

The increasing number of G protein-coupled receptor (GPCR) crystal structures offers new opportunities for histamine receptor homology modeling. However, computational prediction of ligand binding modes in GPCRs such as the histamine H₄ receptor (H₄R), a receptor that plays an important role in inflammation, remains a challenging task. In the current work we have combined complementary in silico receptor modeling approaches with in vitro ligand structure-activity relationship (SAR) and protein site-directed mutagenesis studies to elucidate the binding modes of different ligand classes in H₄R. By systematically considering different H₄R modelling templates, ligand binding poses, and ligand protonation states in combination with docking and MD simulations we are able to explain ligand-specific mutation effects and subtle differences in ligand SAR. Our studies confirm that a combined theoretical and experimental approach represents a powerful strategy to map ligand-protein interactions. The Royal Society of Chemistry 2013.

Full text of DOI:10.1039/c2md20212c

MedChemComm
View Article Online
View Journal | View Issue
CONCISE ARTICLE
Mapping histamine H4 receptor–ligand binding modes†
Sabine Schultes,‡ab Saskia Nijmeijer,‡a Harald Engelhardt,ab Albert J. Kooistra,a
Cite this: Med. Chem. Commun., 2013,
4, 193
Henry F. Vischer,a Iwan J. P. de Esch,a Eric E. J. Haaksma,ab Rob Leursa and
a
*
Chris de Graaf
The increasing number of G protein-coupled receptor (GPCR) crystal structures offers new opportunities for
histamine receptor homology modeling. However, computational prediction of ligand binding modes in
GPCRs such as the histamine H4 receptor (H4R), a receptor that plays an important role in inflammation,
remains a challenging task. In the current work we have combined complementary in silico receptor
modeling approaches with in vitro ligand structure–activity relationship (SAR) and protein site-directed
mutagenesis studies to elucidate the binding modes of different ligand classes in H4R. By systematically
considering different H4R modelling templates, ligand binding poses, and ligand protonation states in
combination with docking and MD simulations we are able to explain ligand-specific mutation effects
and subtle differences in ligand SAR. Our studies confirm that a combined theoretical and experimental
approach represents a powerful strategy to map ligand–protein interactions.
Received 26th July 2012
Accepted 4th September 2012
DOI: 10.1039/c2md20212c
www.rsc.org/medchemcomm
(Fig. 2)29–31 allows different plausible H4R–ligand binding
modes. Site-directed mutagenesis studies have identied
Introduction
D943.32 (a conserved binding residue in bioaminergic GPCRs)32
as an essential residue in H4R to bind both the non-imidazole
antagonist JNJ 7777120 1b3 and the small imidazole-containing
agonist histamine 3.3,7 In addition, site-directed mutagenesis
(SDM) studies showed that the anionic carboxylate group of
E1825.46 plays an important role in binding of H4R ligands that
contain two basic groups, including 3, VUF 8430 4, and to a
smaller degree clobenpropit 5 and clobenpropit analogue VUF
5228 6, while 1b and clozapine 7 only require a H-bond acceptor
at position 1825.46 (e.g. present in the E1825.46Q mutant).2,3 In
the H1R co-crystal structure23 the inverse agonist doxepin forms
an ionic H-bond interaction with the conserved D1073.32, but
does not form a H-bond with N1985.46. The aromatic ring
systems of doxepin bind to two connected subpockets in the
hydrophobic cavity between transmembrane (TM) helices 3–6
(pocket II28), in the current study dened as subpocket IIa and
subpocket IIb (Fig. 1A). Receptor mutagenesis, ligand SAR, and
receptor–ligand interaction modeling studies have indicated
that different ligands bind to these different subpockets.2,4,30,33
Ligands 6 and 7 occupy both subpockets IIa and IIb simul-
taneously (Fig. 1B and C).2,4 Triazole ligand 8 is proposed to
bind only in subpocket IIa,33 while combined modeling and
mutagenesis studies support two alternative binding modes of
ligand 5 in which it accommodates its aromatic moiety in
subpocket IIb, or, alternatively, in subpocket I.2
Rational drug design requires detailed knowledge and under-
standing of the molecular interactions between ligands and
proteins. Both ligand- and protein-based in silico modeling
approaches have been successfully applied to rationalize
structure–activity relationships (SAR) and receptor site-directed
mutagenesis studies, for the histamine H4 receptor (H4R),1–7 a G
protein-coupled receptor (GPCR) that plays an important role in
inammation.8–20 The increasing number of GPCR X-ray struc-
tures,21,22 including the recently solved histamine H1 receptor
(H1R) crystal structure23 (Fig. 1A), offers new opportunities for
histamine receptor homology modeling and the structure-based
design of new histamine receptor ligands.24–27 However,
computational H4R–ligand binding mode prediction still
remains a challenging task. The symmetric distribution of the
two acetic residues (aspartate D943.32 and glutamate
E1825.46
)
in combination with different hydrophobic sub-
2,3,7
pockets (I and II,28 Fig. 1B and C) that are complementary to
(two) basic and several hydrophobic groups in H4R ligands
aLeiden/Amsterdam Center for Drug Research (LACDR), Division of Medicinal
Chemistry, Department of Pharmacochemistry, Faculty of Exact Sciences, VU
University Amsterdam, De Boelelaan 1083, 1081 HV Amsterdam, The Netherlands.
E-mail: c.de.graaf@vu.nl
bBoehringer Ingelheim RCV GmbH & Co KG, Department of Medicinal Chemistry, Dr
Boehringer-Gasse 5-11, Vienna, Austria
† Electronic supplementary information (ESI) available: Additional analyses of
molecular dynamics simulations, purity of compounds by LC-MS and NMR, and
comparison of ADRB2 vs. H1R crystal structure based H4R homology models.
See DOI: 10.1039/c2md20212c
In this study we have focused on the elucidation of the
binding interactions of two different selective H4R ligand
classes: indolecarboxamides4,6,27,36–39 (1a,b) and 2-amino-
pyrimidines40–46 (2a,b) (Fig. 2). Recent ligand-binding mode
‡ These authors contributed equally.
This journal is ª The Royal Society of Chemistry 2013
Med. Chem. Commun., 2013, 4, 193–204 | 193

View Article Online
MedChemComm
Concise Article
Fig. 1 Binding modes of doxepin in the H1R co-crystal structure23 (A), and clozapine4 (B) and VUF 5228 (ref. 2) (C) in (ADRB2 crystal structure34 based) H4R homology
models. Compounds and pocket residues are depicted as ball-and-sticks, whereas for clarity Y953.33 is shown as lines. H-bonds between the ligand and pocket residues
are represented as black dotted lines. The backbone TM helices 5, 6, and 7 (right to left) are presented as yellow helices. Helix 3 is presented by yellow ribbons.
Subpockets I, IIa and IIb are labeled in red.
predictions of dual H3R/H4R clobenpropit ligands 5 and 6 molecular modeling and MD-simulation revealed similarities as
demonstrated that the integration of complementary in vitro well as differences in the molecular H4R binding determinants
and in silico modeling approaches is an efficient way to map for both ligand classes. These insights are useful for rational
protein–ligand interactions.2 In the current study we used in optimization of indolecarboxamides and 2-aminopyrimidines,
silico guided mutation studies to compare the binding modes of and the design of new H4R ligands.
substituted and unsubstituted indolecarboxamides and 2-ami-
nopyrimidines. To obtain more detailed insights into ligand-
specic molecular recognition features, molecular dynamics
Results
(MD) simulations were employed, a technique that has been
successfully applied to describe subtle differences in ligand
binding (e.g. ligand regio-, stereo-, and protein-selectivity47,48).
Furthermore, different ligand protonation states (i.e. 2-amino-
pyrimidines 2a,b with one versus two positively ionized basic
groups) and different H4R modeling templates (i.e. the bio-
aminergic GPCR b2-adrenergic receptor (ADRB2)34 versus the
more closely related H1R)23 were considered to investigate the
inuence of modeling parameters. The combination of protein
mutagenesis data with different binding mode hypotheses,
SAR (Table 1, Fig. 3) and mutagenesis (Table 2, Fig. 4) studies in
combination with docking and MD-simulations (Fig. 5–7,
Tables 3 and 4) were used to elucidate the binding modes of
indolecarboxamides 1a,b and 2-aminopyrimidines 2a,b (Fig. 2)
in the H4R binding site and to give insights into the molecular
determinants of H4R–ligand interactions. For this procedure
different starting poses of the ligands were generated (Fig. 5) in
different H4R models (Fig. S1†). For the 2-aminopyrimidine
ligands different protonation states were considered (Tables 3
and 4). The H4R–ligand binding mode models that could best
Fig. 2 Structures of H4R ligands 1–9. Affinities (pKi) for H4R of 1 and 2 are from the current study, 3, 4 and 7 from ref. 4, 5 and 6 from ref. 2, 8 from ref. 33 and 9 from
ref. 35.
194 | Med. Chem. Commun., 2013, 4, 193–204
This journal is ª The Royal Society of Chemistry 2013

View Article Online
Concise Article
MedChemComm
Table 1 pKi values of indolecarboxamides (1) and 2-aminopyrimidines (2) with
different R substitutionsa
Ligands (R) a (H)
b (Cl)
c (NH2)
d (NO2)
e (OCH3)
1
2
7.1(0.07) 7.7(0.07) 7.6(0.10) 7.0(0.20) 5.3(0.16)
8.3(0.09) 8.7(0.15) 8.3(0.04) 8.1(0.00) 7.8(0.05)
a
Data shown are mean of at least two independent experiments
performed in triplicate (SEM given between brackets).
Fig. 4 Radioligand displacement curves of ligands 1a (A), 1b (B), 2a (C), 2b (D)
on H4R WT ( ), H4R-L5.39V ( ), H4R-E5.46Q ( ). E5.46Q binding studies were
performed with the radioligand [3H]-JNJ 7777120, whereas [3H]-histamine was
used for H4R and L5.39V. Data shown are representative specific binding curves of
at least two experiments performed in triplicate. Error bars indicate SEM values.
To enable a better visualization of the pKi shifts, each curve is corrected for the
respective radioligand concentration and Kd values.
Fig. 3 Reagents and conditions: (i) dichloromethane, N-ethyl-diisopropylamine,
20 ꢀC, 16 h; (ii) dichloro[1,10-bis(diphenylphosphino)ferrocene]palladium(II)
dichloromethane adduct, Cs2CO3, tetrahydrofuran, 1-methyl-2-pyrrolidinone,
water, 100 ꢀC, 16 h.
nucleophilic aromatic substitution reaction. In a second step
the accordant phenyl moieties were installed using a Suzuki
reaction.
Table 2 pKi values of ligands 1a,b and 2a,b for the H4R, L1755.39V and
E1825.46Qa
Radioligand displacement studies of [3H]-histamine binding
to the human H4R of the indolecarboxamide (1a–e) and 2-
aminopyrimidine (2a–e) analogues indicated that the SAR of the
two ligand series is comparable for small substituents but
deviates for the larger –OCH3 substituent (Table 1). Substitution
of the indolecarboxamide with the methoxy group (1e) resulted
in a 63-fold decrease compared to the unsubstituted ligand (1a),
while comparison of the methoxy- (2e) and unsubstituted (2a)
2-aminopyrimidines gave only a 3-fold affinity change. The
Cl-substituted ligands showed the highest H4R binding affinity
for both the indolecarboxamide and 2-aminopyrimidine ligand
classes.
Ligands
WT
L1755.39
V
E1825.46
Q
1a
1b
2a
2b
7.1 (0.07)
7.7 (0.07)
8.3 (0.09)
8.7 (0.15)
6.5 (0.13)
6.4 (0.07)
7.7 (0.29)
7.2 (0.12)
6.7 (0.09)
7.5 (0.08)
7.2 (0.02)
6.7 (0.15)
a
Data shown are mean of at least two independent experiments
performed in triplicate (SEM given between brackets).
explain the (ligand specic) mutation data were used to identify
similarities and differences in the binding characteristics for
the different ligand classes.
Ligand binding affinities for H4R, L1755.39V and E1825.46
Q
Structure–activity relationships of substituted
indolecarboxamides and 2-aminopyrimidines
The binding affinities (pKi) of the unsubstituted and Cl-
substituted indolecarboxamide and 2-aminopyrimidine ligands
To determine the binding mode of indolecarboxamides and 2- were analysed for H4R, L1755.39V and E1825.46Q (Table 2, Fig. 4)
aminopyrimidines we synthesized H, Cl, NH2, NO2, and OCH3 using heterologous [3H]histamine displacement binding
substituted indolecarboxamide (1a–e) and 2-aminopyrimidine experiments. Previous studies have shown that E1825.46Q affects
(2a–e) analogues and investigated the effect of different the binding of H4R ligands that contain two basic groups
substituents for both ligand classes (Table 1). The synthesis of (3–6),2,3,7 but does not affect binding of ligands that contain only
the indolecarboxamide ligands is described in the litera- one basic moiety (1b, 7).3 The L1755.39V mutant affects the
ture36,37,39 and Fig. 3 illustrates the synthesis route for the 2- affinity for larger ligands 1b and 7, but does not affect the
pyrimidine-2-amine derivatives. Starting from 4,6-dichloro- binding of 3.4 The fact that these two mutants show ligand-
pyrimidine-2-amine, methylpiperidine was introduced via a dependent effects and that they have previously been used to
This journal is ª The Royal Society of Chemistry 2013
Med. Chem. Commun., 2013, 4, 193–204 | 195

View Article Online
MedChemComm
Concise Article
Fig. 5 Initial binding poses of indolecarboxamide ligand 1a (A–C) and 2-aminopyrimidine ligand 2a (D–F) in H1R crystal structure based H4R models used as starting
structures for MD simulations. Rendering and labels are the same as in Fig. 1.
Fig. 6 Shortest distance between any heavy atom of the L1755.39 and (A) 1b, and (B) mono-protonated (1+) and (C) double-protonated (2+) 2b in MD-simulations
starting from pose 1 (black), pose 2 (blue), and pose 3 (red) (see Fig. 5).
investigate the binding mode of 1b makes them attractive tools
for investigating similarities and differences in binding modes
of 1a,b and 2a,b. Both ligand classes lost affinity for L1755.39V.
The Cl-substituted ligands, 1b and 2b, were more affected
(DpKi(1b): ꢁ1.3, DpKi(2b): ꢁ1.5) than the unsubstituted ligands,
1a and 2a, (DpKi(1a): ꢁ0.6, DpKi(2a): ꢁ0.6). In contrast,
E1825.46Q had a different effect on the two ligand classes. While
the affinities of the 2-aminopyrimidines for E1825.46Q signi-
cantly decreased (DpKi(2a): ꢁ1.1, DpKi(2b): ꢁ2) the mutations
did not have a signicant effect on the affinities of the indole-
carboxamides (DpKi(1a): ꢁ0.4, DpKi(1b): ꢁ0.2). Again the
Cl-substituted 2-aminopyrimidine ligand 2b was more affected
than the unsubstituted ligand 2a.
Molecular docking studies
We used molecular docking simulations49,50 to generate
different plausible poses for the two indolecarboxamides (1a
and 1b) and 2-aminopyrimidines (2a and 2b) in a H4R homology
model that is based on the recently resolved H1R crystal struc-
ture (Fig. 5).23 For comparison we also investigated docking
poses in a H4R model that is based on the ADRB2 crystal
structure34 (Fig. S1†) as this model was as good as the H1R-based
H4R model in retrospective virtual screening studies.51 While
the TM helices of ADRB2 and H1R crystal structures share the
same overall fold,23,34 the conformations of in particular the
second extracellular loops (EL2) are different (Fig. S1†).
196 | Med. Chem. Commun., 2013, 4, 193–204
This journal is ª The Royal Society of Chemistry 2013

View Article Online
Concise Article
MedChemComm
accommodated in the upper hydrophobic subpocket IIb. The
binding pose in Fig. 5B (pose 2) is similar to the pose proposed
by Kiss et al.38 in which the piperazine and indole nitrogen
atoms of the ligand donate H-bonds to E1825.46 and D943.32
,
respectively, while the aromatic moiety of the ligand is directed
towards TM7 (subpocket I28). The binding pose in Fig. 5C (pose
3) is similar to the binding mode reported by Schneider et al.,6
in which the piperazine and indole nitrogen atoms of the ligand
donate H-bonds to D943.32 and E1825.46, respectively (similar to
pose 1), while the aromatic ring of the ligand binds to subpocket
IIa. To obtain this latter pose the c1 torsional angle of C983.36
was rotated from its t-conformation to the g+-conformation.
Three docking poses were selected for the 2-amino-
pyrimidines that correspond to the respective poses of the
indolecarboxamides (Fig. 5D–F). The amino group of the
pyrimidine interacts in a similar manner as the NH of the
indole group in these poses. A binding pose for 2-amino-
pyrimidines that is comparable to pose 1 was proposed by
Werner et al.,40–46 with the exception that the 2-amino-
pyrimidine moiety in the Werner H4R model is differently
oriented in the H4R binding pocket and H-bonds to E1825.46 are
formed via water molecules.40–46 The 2-aminopyrimidines can
form an additional hydrogen bond with either E1825.46, in pose
1 or 3, or D943.32 in pose 2 when the pyrimidine ring is
protonated. pKa measurements for ligand 2a (see Experimental
section for procedure) revealed a pKa value of 7.5 for the
piperazine nitrogen atom and a pKa value of 6.0 for the pyrim-
idine, indicating that in solution approximately 2.3% of ligand
2a will be double-protonated at physiological pH 7.4.
Molecular dynamics simulations
Binding mode elucidation. All three binding poses were
subjected to 1 ns MD simulations. During the course of the
MD-simulation 500 snapshots were collected. H-bond interac-
tions between the ligands 1a,b and 2a,b and the carboxylate
groups of D943.32 and E1825.46, as well as hydrophobic contacts
between the ligands and L1755.39 were monitored (Table 3,
Fig. 6) to validate the different binding mode models with
mutagenesis data. Our data showed that the E1825.46Q mutant
affects binding of 2a and 2b, but not of 1a and 1b, while the
L1755.39V mutation affects the affinity of all ligands (Table 2 and
Fig. 4). In addition, previous mutation studies3,7,32,52,53 indicated
Fig. 7 Representative snapshots of MD simulations starting from pose 1 in the
˚
H1R-based H4R model showing the pocket volume within 3.5 A of the chlorine
atom of ligand 1b (red, A) and ligand 2b (green, B). The snapshots presented in
panels A (red triangle) and B (green circle) are indicated in panel C, showing
˚
pocket volumes within 3.5 A of the chlorine atom of ligand 1b (red) and double-
protonated ligand 2b (green) along the full 1 ns MD trajectories.
As a result, the H4R models based on ADRB2 and H1R crystal that the conserved D943.32 residue plays an important role in
structures have very similar TM domains and show overall the histamine receptor binding, while an H-bond donor in the side-
same receptor–ligand interactions, except in the EL2 loop chain of residue 1825.46 in H4R (e.g. present in the E1825.46
Q
region (Fig. S1†). Only docking poses where the ligands interact mutant) is required to bind 1b.3 Table 3 shows the frequency of
with both, aspartate D943.32 and glutamate E1825.46 (essential hydrophobic contacts of the ligand with L1755.39 in combination
residues for binding of H4R ligands)3,7 were considered. Both with H-bond formation with both D943.32 and E1825.46 (Tables
residues were modelled in the deprotonated (negatively S1 and S2† report the individual and combined frequencies of
charged) state. Three binding poses were generated for both the different interactions). Frequent interactions of indole-
ligand classes in the two different H4R models as starting carboxamides (1a and 1b) and 2-aminopyrimidines (2a and 2b)
structures for MD simulations. The binding pose of indole- with L1755.39 in combination with stable H-bond formation
carboxamides in Fig. 5A (pose 1) is similar to the binding mode with D943.32 and E1825.46 were only observed in the MD-simu-
proposed by Lim et al.,4 in which the piperazine and indole lations starting from pose 1 (Table 3, Fig. 6). Starting from pose
nitrogen atoms of the ligand donate H-bonds to D943.32 and 2, only the MD-simulations of ligands 2a and 2b showed
E1825.46, respectively, while the aromatic moiety of the ligand is interactions with L1755.39 in combination with H-bond
This journal is ª The Royal Society of Chemistry 2013
Med. Chem. Commun., 2013, 4, 193–204 | 197

View Article Online
MedChemComm
Concise Article
Table 3 Frequency (%) of hydrophobic interactionsa between the ligand and L1755.39 in combination with H-bond formationb with D943.32 and E1825.46 in H1R-based
and ADRB2-based H4R models that occur during 1 ns MD simulations
Structure
Protonation
1+
1+
2+
R
H
Cl
H
Cl
H
Cl
Name
1a
1b
2a
2b
2a
2b
H1R based H4R model
Pose 1c
56.2
0
0
89.4
0
0
91.0
15.4
0
94.6
14.4
0
87.6
22.6
0
75.2
24.0
0
Pose 2c
Pose 3c
ADRB2 based H4R model
Pose 1c
21.0
0
0
70.6
0
0
90.2
38.0
0
80.0
0
0
51.2
0.2
0
88.8
0
0
Pose 2c
Pose 3c
a
b
A hydrophobic interaction is counted if the shortest distance between any heavy atom of the ligand and L1755.39 is #4 A. An H-bond is counted if
˚
˚
the distance between the H-bond donor and acceptor heavy atom is below 3.5 A and the angle between the H-bond donor heavy atom, hydrogen and
H-bond acceptor heavy atom is between 135ꢀ and 225ꢀ. c See Fig. 5.
formation with D943.32 and E1825.46, but the frequency of these the affinities of the 2-aminopyrimidines were signicantly
interactions is much lower than in pose 1 (0–38% vs. 51–95%, affected by this mutation (DpKi(2a): ꢁ1.1, DpKi(2b): ꢁ2.0).
Table 3), as demonstrated for ligand 2b in Fig. 6B and C. In the These data indicate that ionic interactions with the negatively
ADRB2-H4R model only for the mono-protonated ligand 2a ionisable E1825.46 residue play a role in binding of 2a and 2b
interactions with L1755.39 (in combination with stable H-bonds (but not of 1a and 1b) and suggest that the basic 2-amino-
with D943.32 and E1825.46) were observed, but with signicantly pyrimidine moieties of 2a-b are protonated. To investigate this
lower frequency (38%) than in pose 1 (90%, Table 3).
hypothesis, the mono and double-protonated 2a and 2b were
Protonation state elucidation of 2-aminopyrimidines bound modelled in E1825.46Q mutated H4R (pose 1) and subsequently
to H4R. Our mutagenesis studies (Table 2 and Fig. 3) showed subjected to MD simulations. As anticipated and shown in
that the affinities of indolecarboxamides were hardly affected by Table 4, only the H4R binding models in which ligands 2a
the E1825.46Q mutation (DpKi(2a): ꢁ0.4, DpKi(2b): ꢁ0.2), while and 2b are double-protonated (i.e. both piperazine and
Table 4 Average H-bond interactions between the ligand and residue 1825.46 in wild-type and E1825.46Q H4R models during 1 ns MD-simulation starting from pose 1
(see Fig. 5)
Structure
Protonation
1+
1+
2+
R
H
Cl
H
Cl
H
Cl
Name
1a
1b
2a
2b
2a
2b
H1R based H4R model
WT
E5.46
D3.32
Q5.46
D3.32
1.2
0.9
1.0
1.0
1.1
0.9
1.0
0.9
1.1
1.0
1.0
0.1
1.3
1.0
0.7
0.6
2.5
1.0
0.6
1.0
2.2
1.0
0.3
0.9
E5.46
Q
ADRB2 based H4R model
WT
E5.46
D3.32
Q5.46
D3.32
1.0
1.0
0.9
1.0
1.2
1.0
1.2
1.0
1.1
1.1
1.3
1.0
1.3
1.2
1.4
1.0
2.3
1.0
1.1
1.0
2.4
1.0
0.7
1.0
E5.46
Q
198 | Med. Chem. Commun., 2013, 4, 193–204
This journal is ª The Royal Society of Chemistry 2013

View Article Online
Concise Article
MedChemComm
aminopyrimidine groups in their positively ionized forms) can obtained into the similarities and differences of molecular
explain the observed affinity decrease for 2a and 2b upon interactions between H4R and these two ligand classes.
E1825.46Q mutation (Table 2). In these models the average
number of hydrogen bonding interactions with residue 1825.46
Overlap in indolecarboxamides and 2-aminopyrimidines
binding modes
observed in the 500 snapshots was decreased in the E1825.46
Q
mutant compared to the H4R (difference in H-bond formation
with 1825.46 in H1R–H4R: 1a(2+): ꢁ1.9, 1b(2+): ꢁ1.9), ADRB2-
H4R: 2a(2+): ꢁ1.2, 2b(2+): ꢁ1.7), while H-bond interactions with
D943.32 were unaffected. To compare the two ligand classes, also
indolecarboxamides 1a and 1b in the E1825.46Q models were
subjected to 1 ns MD-simulation. In both E1825.46Q models
H-bond interactions with 1a and 1b were hardly affected
compared to the H4R models (Table 4). This is in line with the
fact that the E1825.46Q mutation had no signicant effect on
binding affinity for these ligands (Table 2, Fig. 4).
Our combined SDM and in silico studies indicate that the
binding modes of indolecarboxamides and 2-amino-
pyrimidines overlap, but also show that there are subtle differ-
ences in the importance of different H4R interaction features
within these two ligand series. Ligands of both ligand classes
accommodate their aromatic ring moieties in subpocket IIb
that is located in the upper half of the extracellular regions
between TM helices 3–6 (pose 1), (Fig. 5A and D and Fig. S1†).
Only in this pose all ligands (1a,b and 2a,b) form stable H-bond
interactions with D943.32 and E1825.46 in combination with
hydrophobic interactions with L1755.39. Current (Table 2 and
Fig. 4) and previous SDM studies show that these interactions
are important binding determinants for ligands 1a,b and 2a,b
as well as other H4R ligands.2,3,7 The overall binding orientation
was stable and not inuenced by interactions with water
molecules.
It should be noted that effects of H4R mutations are oen
ligand specic. While the E1825.46Q mutant has a decreased
affinity for ligands 2a, 2b, 3, and 4, this mutation does not affect
binding of ligands 1a, 1b and 7 (Table 2 and Fig. 4).2,3 The
L1755.39V mutant that mimics the monkey H4R has a decreased
affinity for 1 and 2 (Table 2 and Fig. 4) but an increased affinity
for ligand 7 (Fig. 1B), while binding of ligand 3 is not affected by
this mutation.4 This reects the diversity in H4R ligand binding
modes (Fig. 1 and 7) that can match different parts of the H4R
ligand pharmacophore.30 This pharmacophore contains: (i) 1 to
2 H-bond donors (e.g. piperazine and indole/aminopyrimidine
nitrogen atoms of ligands 1 and 2) that are complementary to
D943.32 (ref. 3 and 7) and E1825.46, (ii) a central aromatic moiety
(e.g., indole (1) and 2-aminopyrimidine (2) moieties) that stacks
between Y3.33 and Y6.51, and (iii) 1 to 2 hydrophobic moieties
that can accommodate subpockets IIa (e.g., cyclohexyl groups of
ligands 6 (Fig. 1C) and 8, the chlorinated benzene ring of ligand
7 (Fig. 1B)),2,4,33 IIb (benzene rings of ligands 1 and 2 (Fig. 7) and
ligands 5 and -7 (Fig. 1B and C)2,4 between TM helices 3–6, or
alternatively, subpocket I between TM helices 3 and 5–7
(benzene ring of ligand 5).2
Rationalization of differences in ligand structure–activity
relationships
The SAR for different substitutions in both ligand series
(Table 1) indicated that the larger substituent –OCH3 is less
compatible with the H4R binding site than smaller apolar (–Cl)
or polar (–NH2, –NO2) groups. This affinity decrease was much
more pronounced for the indolecarboxamides than for 2-ami-
nopyrimidines (Table 1). To explain this effect we monitored the
˚
pocket volume around 3.5 A distance of the chlorine atom for 1b
and 2b (Fig. 7). We observed that in the H1R-based H4R model
the chlorine atom of 2b (independent of the modeled proton-
ation state) occupies more frequently larger pockets than the
chlorine atom of 1b. Average pocket volumes around the chlo-
rine atom in the H1R-based H4R model were signicantly
3
3
˚
˚
smaller for ligand 1b (0.6 A ) than for ligand 2b (8.4 A (mono-
3
˚
protonated) and 3.2 A (double-protonated)) while in the
ADRB2-based H4R model the volume for 1b (11.0 A ) was
signicantly larger than for 2b (6.4 A (mono-protonated) and
3.9 A (double-protonated)). The H1R-based H4R models there-
3
˚
3
˚
3
˚
fore explain the fact that 2-aminopyrimidines can accommo-
date larger substituents than indolecarboxamides.
A
comparison of representative snapshots at the end of the MD-
simulation showed that although the overall orientation of the
ligands in the binding pocket is comparable, the chlorine atoms
of both ligands are oriented in different pockets (Fig. 7A and B).
The chlorine atom of 1b is directed towards the extracellular
loop (EL2) and TM6 (surrounded by F16845.54, F16945.55
,
Our MD-simulation showed that although the overall
orientation of the ligands (1a,b and 2a,b) in the H4R binding
pocket is comparable, the chlorine atoms of indolecarboxamide
and 2-aminopyrimidine ligands are located in subpockets of
different size (Fig. 7). This size difference explains why larger
substituents have a more negative effect on the affinity for
indolecarboxamides than for 2-aminopyrimidines (Table 1).
Interestingly, SDM studies have shown ligand-dependent
effects towards mutations in the same EL2 region that can
explain H4R species selectivity differences.5
E16345.49, L1755.39, T3236.55) in a relatively tight binding pocket
(Fig. 7A). The chlorine atom of 2b occupies a pocket, which is
located more towards TM5 (surrounded by I1745.38, L1755.39 and
T1785.42 in TM5, as well as F16845.54 and backbone atoms of
EL2) and has a larger volume (Fig. 7B) that allows larger
substituents (Table 1).
Discussion
In the current study, ligand SAR and H4R SDM studies were
combined with molecular docking and MD simulations to
The role of ligand protonation states in H4R binding
elucidate the binding modes of indolecarboxamides and In silico evaluation of different protonation states of 2-amino-
2-aminopyrimidines in the H4R binding pocket. Insights were pyrimidines showed that the H4R binding models in which both
This journal is ª The Royal Society of Chemistry 2013
Med. Chem. Commun., 2013, 4, 193–204 | 199

View Article Online
MedChemComm
Concise Article
the piperazine and 2-aminopyrimidine moieties are positively GPCR–ligand co-crystal structures,61,62 and emphasized by the
ionized best explain the affinity decrease of ligands 2a and 2b different extracellular loop structures displayed in the currently
(but not of 1a and 1b) for the E1825.46Q mutant (Tables 2–4). The available GPCR crystal structures.21,63 Despite some recent
pKa value of the pyrimidine ring of 6.0, however, suggests that reports of successful automated retrospective prediction of
under physiological pH 7.4 only a small portion (ꢂ2.3%) of individual loops of GPCR crystal structures,64 GPCR loop
ligands 2a and 2b contains a positively ionized pyrimidine ring. modeling should be approached with much caution and should
On the other hand, it has been reported that pKa values of be reserved for cases where loop building can be guided and
functional groups in ligands can change upon binding, as validated by experimental data.65 Moreover, structure-based
demonstrated for the pteridine ring of methotrexate that virtual screening studies have shown that loopless TM models of
becomes protonated upon binding to dihydrofolate reduc- GPCRs can be suitable targets for virtual screening as well.63
tase.54,55 Thus, it cannot be excluded that the pKa value of the Nevertheless, our H4R modeling studies indicate that the crystal
2-aminopyrimidine ring is shied towards higher values upon structure of the more closely related H1R provides a better
binding to H4R.
template for the EL2 region (despite the fact that some parts of
It should furthermore be noted that also other ligands have this loop were not resolved in the H1R crystal structure),23 and to
been proposed to bind in a double-protonated state to H4R. The construct H4R models that can describe subtle differences
two basic groups of ligand 9 have pKa values of 8.3 and 9.4 between ligand binding modes.
(ref. 56) indicating that most parts of this ligand will be double-
protonated at pH 7.4. Agonist 4 (ref. 57), derived from ligand 9,
Docking in combination with MD simulations explains H4R
has two chemically similar basic moieties, while the pKa values
mutation studies
of the amine and imidazole groups of 3 (9.4–9.9 and 5.9–6.5,
respectively)58–60 suggest that also 3 might bind in a double-
protonated form. This is supported by the fact that the
E1825.46Q H4R mutant has a signicantly decreased affinity for
small agonists 3 and 4 (ref. 3) that depend on strong H-bond
interactions. The effects of the E1825.46Q mutation on H4R
binding affinity are relatively smaller (5 and 6) or not signicant
(1a,b and 7) for ligands with larger hydrophobic moieties (Table
2, Fig. 4),2,3 as the binding of these molecules is more deter-
mined by hydrophobic interactions. Interestingly, in this
current study we show that also the 2-aminopyrimidines (con-
taining a hydrophobic benzene ring) are to a large extend
In particular the symmetric distributions of two acetic residues
(aspartate D943.32 and glutamate E1825.46) and two hydrophobic
subpockets IIa and IIb that are complementary to (two) basic
and several hydrophobic groups in H4R ligands allow different
plausible binding H4R–ligand modes (Fig. 5). Previous66 and
current studies suggest that molecular docking alone is not
suitable for accurate determination of relative probabilities of
different ligand binding modes. To explain the in vitro SDM
data (Table 2, Fig. 4), which are an average of different ligand
and receptor orientations with time, a dynamic treatment of
both ligand and receptor is required (Tables 3 and 4). MD
simulations can account for distributions of protein–ligand
interactions and thus can give a more comprehensive explana-
tion of ligand-dependent mutation effects (Tables 2–4) and
subtle differences in ligand structure–activity relationships
(Table 1, Fig. 7).67,68 By systematically considering different
ligand binding mode hypotheses and protein models, and
combining ligand SAR and protein mutagenesis experiments
with extensive MD-simulation studies (48 independent MD runs
in total) we could identify similarities and differences in the
molecular determinants of H4R binding for both indole-
carboxamide and 2-aminopyrimidine ligand classes.
dependent on ionic interactions with E1825.46
.
Ligand-binding mode predictions in H4R homology models
were modelling template independent
To elucidate H4R-ligand binding modes we considered two H4R
models, one based on the ADRB2 crystal structure34 and another
based on the more recent H1R crystal structure.23 The ADRB2
and H1R crystal structures share a similar fold23 and conse-
quently both models are constructed using the same modeling
approach.2,4 It is therefore not surprising that the overall
structure of the homology models is relatively similar, and gives
comparable results in ligand-binding mode prediction2 and
virtual screening studies.51 Also in the current study, the
Conclusion
conformational ensembles derived from MD simulations of We have successfully combined complementary in silico H4R
ADRB2- and H1R-based H4R models show the same overall structural modeling approaches and ligand SAR with in vitro
binding mode (pose 1) for indolecarboxamides and 2-amino- mutagenesis data to investigate the binding mode of indole-
pyrimidines that matches our SDM studies (Table 2 and Fig. 4). carboxamides and 2-aminopyrimidines in H4R. By systemati-
In contrast, only the H1R-based H4R model is able to explain cally considering different H4R modelling templates, ligand
that larger substituents have a larger negative effect on the binding poses, and ligand protonation states in combination
affinity for indolecarboxamides than for 2-aminopyrimidines with molecular docking and MD simulations we are able to
(Table 1) because the chlorine atoms of indolecarboxamide and explain ligand-specic mutation effects and subtle differences
2-aminopyrimidine ligands are located in different subpockets in ligand SAR. Our results improve the knowledge of H4R–
close to EL2 (as discussed above). Accurate GPCR loop ligand binding and provide valuable information for optimizing
modelling is still very challenging, as demonstrated by the and developing other H4R-specic ligands. The combined in
recent GPCR DOCK competitions to predict the coordinates of silico modelling and in vitro SAR and mutagenesis approach
200 | Med. Chem. Commun., 2013, 4, 193–204
This journal is ª The Royal Society of Chemistry 2013

View Article Online
Concise Article
MedChemComm
presents a promising approach to explore protein–ligand Purity by method A1: >95%; RT ¼ 1.35 min; MS (ESI) m/z 270
interactions and can be considered as a general method to [M + H]+; HRMS (ESI+) m/z found 270.1715 [M + H]+, C15H20N5
elucidate protein ligand-binding modes.
requires M+ 270.1718; 1H NMR (500 MHz, DMSO) d (ppm) 8.05–
8.01 (m, 2H), 7.45–7.41 (m, 3H), 6.57 (s, 1H), 6.06 (s, 2H), 3.64–
3.59 (m, 4H), 2.37–2.34 (m, 4H), 2.21 (s, 3H).
Using the same method 2b–e were synthesized. Analytic data
for these ligands are given in the ESI.†
Experimental section
Synthetic methods
General remarks. Chemicals and reagents were obtained
from commercial suppliers and were used without further
purication. Proton and carbon NMR spectra were obtained on
a Bruker Advance 400 FT-NMR or Bruker Advance 500 FT-NMR
Pharmacological assays
Cell culture, transfection and membrane preparation.
instrument with chemical shis (d) reported relative to tetra- HEK293T cells were cultured in Dulbecco's modied Eagles
methylsilane as an internal standard.
medium (DMEM) supplemented with 10% fetal bovine serum
High resolution mass spectroscopy data were obtained on a (FBS), 50 IU mlꢁ1 penicillin and 50 mg mlꢁ1 streptomycin at
LTQ Orbitrap XL (Thermo Scientic) equipped with a NSI 37 ꢀC/5% CO2. One day prior to transfection, the HEK293T cells
Source (Advion Nanomate) in ESI positive mode.
were seeded at 2 ꢃ 106 cells per 10 cm dish. The poly-
Analytical HPLC-MS analyses were conducted using an Agi- ethyleneimine (PEI) transfection method was used to tran-
lent 1100 series LC/MSD system. The analytic method A1 is siently transfect HEK293T cells with H4R, L1755.39V or E1825.46
Q
dened in Table S3.† Ligand purities were calculated as the cDNA. Briey, for each 10 cm dish 2.5 mg H4R (mutant) cDNA
percentage peak area of the analyzed ligand by UV detection at and 2.5 mg pcDEF3 (empty vector) were incubated with 20 mg
254 nm. If purity data are not explicitly mentioned the ligand 25 kDa linear PEI in a total volume of 500 ml 150 mM NaCl for 30
displays a purity >95%. Flash column chromatography was minutes at room temperature. The transfection mix (cDNA–PEI
carried out using hand packed silica gel 60 (230–400 mesh) or mix) was subsequently added drop wise to the 10 cm dish with 6
pre-packed silica gel columns from Biotage and the product was ml fresh culture medium. Two days post transfection, cells were
eluted under medium pressure liquid chromatography. washed once with phosphate-buffered saline (PBS) and subse-
Preparative high performance chromatography was carried out quently scraped from their culture dish in 1 ml of PBS. Crude
on a Gilson system (pump system: 333 and 334 prep-scale HPLC membrane extracts were collected by centrifugation at ꢂ2000g
pump; fraction collector: 215 liquid handler; detector: Gilson for 10 min at ꢀ4 ꢀC. The crude membrane extract pellets were
UV/VIS 155) using pre-packed reversed phase silica gel columns stored at ꢁ20 C until further use.
from waters. The method for preparative high performance
chromatography P1 is dened in Table S3.†
6-Chloro-4-(4-methylpiperazin-1-yl)pyrimidin-2-amine (2h). 4,6- described.3,4
Site-directed mutagenesis. The construction of the H4R-
L1755.39V and H4R-E1825.46Q mutants has been previously
Dichloro-pyrimidine-2-amine 2f (1.00 g, 5.79 mmol), N-ethyl-
[3H]-Radioligand binding assay. The displacement binding
diisopropylamine (1.33 g, 11.58 mmol) and 1-methylpiperazine assays were performed using crude membrane extracts from
2g (0.64 g, 6.37 mmol) were suspended in 30 ml dichloro- transiently transfected HEK293T cells in 50 mM Tris–HCl
methane and stirred for 16 h at 20 ꢀC. The solvent was removed binding buffer (pH 7.4 at room temperature). Crude membrane
under reduced pressure and the crude material was puried extracts were co-incubated with the 2-aminopyrimidine or
using silica gel ash column chromatography with a solvent indolecarboxamide ligands (Table 1) and ꢂ10 nM radioligand
mixture of dichloromethane, methanol and 25% aqueous ([3H]-histamine for H4R-WT and H4R-L1755.39V or [3H]-JNJ
ammonia of 90 : 9 : 1 for elution. The solvent of the corre- 7777120 for H4R-E1825.46Q) in a total volume of 100 ml per well.
sponding fractions was evaporated under reduced pressure, The ligand–membrane mixtures were incubated for 1.5 h at
yielding 0.9 g (70%) of the title ligand. Purity by method A1: room temperature on a shaking table (750 rpm). Bound radio-
>95%; RT ¼ 1.01 min; MS (ESI+) m/z 228/230 [M + H]+, Cl ligands were separated from free radioligands via rapid ltra-
distribution; HRMS (ESI+) m/z found 228.1015 [M + H]+, tion over a 0.3% PEI-pre-soaked glass ber C plate (GF/C, Perkin
C9H15ClN5 requires M+ 228.1016; 1H NMR (DMSO) d (ppm) 6.45 Elmer). GF/C plates were subsequently washed three times with
(s, 2H), 6.08 (s, 1H), 3.56–3.47 (m, 4H), 2.30 (t, J ¼ 5.1 Hz, 4H), ice-cold 50 mM Tris–HCl wash buffer (pH 7.4 at 4 ꢀC). The
2.19 (s, 3H).
retained radioactivity on the GF/C plates was counted by liquid
4-phenyl-6-(4-methylpiperazin-1-yl)pyrimidin-2-amine (2a). 4- scintillation counting in a Wallac Microbeta (Perkin Elmer).
Chloro-6-(4-methylpiperazin-1-yl)pyrimidin-2-amine 2h (100
Materials. [3H]-Histamine (10.6–13.4 Ci mmolꢁ1
) was
mg, 0.44 mmol), 2-phenylboronic acid (66 mg, 0.53 mmol), purchased from Perkin Elmer and [3H]-JNJ 7777120 (56.1 Ci
caesium carbonate (286 mg, 0.88 mmol) and dichloro[1,10- mmolꢁ1) was a kind gi of Robin Thurmond (JNJ, La Jolla, CA).
bis(diphenylphosphino)ferrocene]palladium(II)
dichloro- PEI was bought from Polysciences and cell cultures media were
methane adduct (36 mg, 0.04 mmol) were suspended in 1 ml of obtained from PAA (Pasching, Austria).
a 3 : 1 : 1 mixture of tetrahydrofuran, 1-methyl-2-pyrrolidinone
pKa Measurements. For pKa measurements a GLpKa auto-
and water. The reaction mixture was ushed with argon and mated pKa analyser (Sirius Analytical Instruments Ltd) was used
stirred for 16 h at 100 ꢀC. The crude product was puried using with standard methods described by Allen et al.69 and Volgyi
method P1, yielding 76 mg (64%, 0.28 mmol) of the title ligand. et al.70
This journal is ª The Royal Society of Chemistry 2013
Med. Chem. Commun., 2013, 4, 193–204 | 201

View Article Online
MedChemComm
Concise Article
adjusted and extended for 2-aminopyrimidine ligands to retain
the amino group (and in the case of double protonation also the
additional proton on the pyrimidine ring) in the plane of the
pyrimidine ring. Partial charges for the ligand were derived
using the AM1-BCC procedure in ANTECHAMBER. The tLEaP
module of AMBER10 was used to generate the topology and
coordinate les of the protein–ligand complex.
Computational methods
H4R model. The construction of the ADRB2-H4R model has
been described previously.2,4 The H1R-H4R model was built in
the same manner. The mutated H4R model (E1825.46Q) was
constructed by mutating the corresponding residues of the H4R
model with the mutate function in MOE 2009.10.71 The H4R
model was used either with the c1 torsional angle of C983.36 in
the t-conformation or in the g+-conformation, where an addi-
tional pocket is assessable. Ligands were docked into both H4R
models using PLANTS49,50 without restraints. As aspartate
D943.32 and glutamate E1825.46 are reported to be important for
binding,3,4 only docking poses where the ligands interact with
both of these residues were selected.
During the MD-simulation trajectories of every 1000th MD-
step were saved. This results in 500 snapshots for a 1 ns MD-
simulation. Using the PTRAJ module of AMBER10 distances and
angles could be extracted which were used to analyse H-bond
interactions between the protein and the ligand and interactions
of the ligand with L1755.39. An H-bond was counted if the
distance between the H-bond donor and acceptor was below
Molecular dynamics simulations. The protein–ligand
complexes were minimized using Amber10,72 including
restraints for the experimentally supported hydrogen bond
interactions with aspartate D943.32 and glutamate E1825.46 (ref.
3 and 4) (distance between the donor and acceptor atom and
angle between the donor, hydrogen and acceptor atom). The
number of H-bond interactions between ligands and the
E1825.46 side chain were generally low for all ligands in MD
simulations in which the carboxylate group of E1825.46 was
considered in the (neutral) protonated state. This is not in line
with the ligand-dependent effects of the E5.46Q mutation and
therefore both residues were modelled in the deprotonated
(negatively charged) state. The minimized protein–ligand
complex was rened by a second minimization without
restraints. The minimized complex was embedded in a pre-
equilibrated lipid bilayer consisting of 1-palmitoyl-2-oleoyl-
phosphatidylcholine (POPC) molecules and solvated with TIP3P
water molecules as described by Urizar et al.73 The whole
complex was again minimized using restraints for the expected
hydrogen bonds. Additionally positional harmonic restraints of
10 kcal molꢁ1 on Ca atoms of the helical domains of the H4R
receptor were applied during this minimization step. Each
minimization was performed using a certain number of steps
steepest descent followed by conjugate gradient until the root
mean square gradient of the potential energy was lower than
0.0001 kcal molꢁ1. Non-bonded interactions were calculated
˚
3.5 A and the angle between the H-bond donor, hydrogen and
H-bond acceptor was between 180 ꢄ 45ꢀ.78 Hydrophobic inter-
actions of the ligand with L1755.39 were counted if the distance
between any heavy atom of the ligand and L1755.39 was #4 A.
˚
Calculation of the pocket volumes. The pocket volume
around the chlorine atoms of ligands 1b and 2b was determined
using POVME.79 A radius of 3.5 A was chosen for the calcula-
tions. The chlorine atom in the pdb snapshots taken during the
MD-simulation was deleted for these calculations. Padding was
on a standard value of 1.09. This is the radius of a hydrogen
atom. Basically, only the volume of the binding site that can be
occupied by hydrogen atoms was measured. As the van der
Waals radius of the carbon atom (where the chorine atom is
attached to) and the diameter of a hydrogen atom overcome the
˚
3
˚
diameter of the chlorine atom also pocket volumes of 0 A can
be observed.
Acknowledgements
This research was nancially supported by The Netherlands
Organization for Scientic Research [NWO VENI Grant
700.59.408 to C. de G.], and COST Action BM0806. Obbe Zui-
derveld is acknowledged for technical assistance.
References
˚
within a cut-off of 10 A.
1 P. Igel, R. Geyer, A. Strasser, S. Dove, R. Seifert and
A. Buschauer, J. Med. Chem., 2009, 52, 6297–6313.
The entire system was subjected to a constant pressure (1
bar) MD-simulation. All bonds involving hydrogen atoms were
frozen with the SHAKE algorithm. In the rst 100 ps the
temperature was increased in a stepwise manner to 300
K. During this procedure positional harmonic restraints on
helical Ca atoms of 10 kcal molꢁ1 were applied which were
reduced to 0.5 kcal molꢁ1 for the subsequent 1 ns production
run (at 300 K). The temperature was controlled using the weak
coupling approach74 with a coupling constant of 0.2 ps. For
calculating the long range electrostatic interactions the Particle
Mesh Ewald (PMW) method was applied. van der Waals inter-
2 E. P. Istyastono, S. Nijmeijer, H. D. Lim, A. van de Stolpe,
L. Roumen, A. J. Kooistra, H. F. Vischer, I. J. de Esch,
R. Leurs and C. de Graaf, J. Med. Chem., 2011, 54, 8136–8147.
3 A. Jongejan, H. D. Lim, R. A. Smits, I. J. de Esch, E. Haaksma
and R. Leurs, J. Chem. Inf. Model., 2008, 48, 1455–1463.
4 H. D. Lim, C. de Graaf, W. Jiang, P. Sadek, P. M. McGovern,
E. P. Istyastono, R. A. Bakker, I. J. de Esch, R. L. Thurmond
and R. Leurs, Mol. Pharmacol., 2010, 77, 734–743.
5 H. D. Lim, A. Jongejan, R. A. Bakker, E. Haaksma, I. J. de Esch
and R. Leurs, J. Pharmacol. Exp. Ther., 2008, 327, 88–96.
6 E. H. Schneider, A. Strasser, R. L. Thurmond and R. Seifert,
J. Pharmacol. Exp. Ther., 2010, 334, 513–521.
˚
actions were calculated within a cut-off of 8 A.
For the protein, the POPC and the water molecules the
AMBER03 force eld75 was used. For the ligand atoms the
GAFF76 force eld was used. Ligand force eld parameters were
derived using ANTECHAMBER.77 These parameters were
7 N. Shin, E. Coates, N. J. Murgolo, K. L. Morse, M. Bayne,
C. D. Strader and F. J. Monsma, Jr, Mol. Pharmacol., 2002,
62, 38–47.
202 | Med. Chem. Commun., 2013, 4, 193–204
This journal is ª The Royal Society of Chemistry 2013

View Article Online
Concise Article
MedChemComm
8 G. Coruzzi, M. Adami, E. Guaita, I. J. de Esch and R. Leurs,
Eur. J. Pharmacol., 2007, 563, 240–244.
Symposium: Trends in Drug Research, http://f1000.com/
posters/browse/summary/1090195.
9 J. M. Cowden, J. P. Riley, J. Y. Ma, R. L. Thurmond and 27 Y. Tanrikulu, E. Proschak, T. Werner, T. Geppert,
P. J. Dunford, Respir. Res., 2010, 11, 86.
N. Todoroff, A. Klenner, T. Kottke, K. Sander, E. Schneider,
R. Seifert, H. Stark, T. Clark and G. Schneider,
ChemMedChem, 2009, 4, 820–827.
10 P. J. Dunford, N. O'Donnell, J. P. Riley, K. N. Williams,
L. Karlsson and R. L. Thurmond, J. Immunol., 2006, 176,
7062–7070.
28 J. S. Surgand, J. Rodrigo, E. Kellenberger and D. Rognan,
Proteins, 2006, 62, 509–538.
11 P. J. Dunford, K. N. Williams, P. J. Desai, L. Karlsson,
D. McQueen and R. L. Thurmond, J. Allergy Clin. Immunol., 29 H. Engelhardt, R. A. Smits, R. Leurs, E. Haaksma and I. J. de
2007, 119, 176–183. Esch, Curr. Opin. Drug Discovery Dev., 2009, 12, 628–643.
12 F. Gantner, K. Sakai, M. W. Tusche, W. W. Cruikshank, 30 E. P. Istyastono, C. de Graaf, I. J. de Esch and R. Leurs, Curr.
D. M. Center and K. B. Bacon, J. Pharmacol. Exp. Ther.,
2002, 303, 300–307.
13 C. L. Hofstra, P. J. Desai, R. L. Thurmond and W. P. Fung-
Leung, J. Pharmacol. Exp. Ther., 2003, 305, 1212–1221.
14 G. C. Hsieh, P. Chandran, A. K. Salyers, M. Pai, C. Z. Zhu,
Top. Med. Chem., 2011, 11, 661–679.
31 R. A. Smits, R. Leurs and I. J. de Esch, Drug Discovery Today,
2009, 14, 745–753.
32 L. Shi and J. A. Javitch, Annu. Rev. Pharmacol. Toxicol., 2002,
42, 437–467.
E. J. Wensink, D. G. Witte, T. R. Miller, J. P. Mikusa, 33 M. Wijtmans, C. de Graaf, G. de Kloe, E. P. Istyastono,
S. J. Baker, J. M. Wetter, K. C. Marsh, A. A. Hancock,
M. D. Cowart, T. A. Esbenshade, J. D. Brioni and
P. Honore, Pharmacol., Biochem. Behav., 2010, 95, 41–50.
J. Smit, H. Lim, R. Boonnak, S. Nijmeijer, R. A. Smits,
A. Jongejan, O. Zuiderveld, I. J. de Esch and R. Leurs, J.
Med. Chem., 2011, 54, 1693–1703.
15 H. D. Lim, R. A. Smits, R. Leurs and I. J. De Esch, Curr. Top. 34 V. Cherezov, D. M. Rosenbaum, M. A. Hanson,
Med. Chem., 2006, 6, 1365–1373.
S. G. Rasmussen, F. S. Thian, T. S. Kobilka, H. J. Choi,
P. Kuhn, W. I. Weis, B. K. Kobilka and R. C. Stevens,
Science, 2007, 318, 1258–1265.
16 P. Ling, K. Ngo, S. Nguyen, R. L. Thurmond, J. P. Edwards,
L. Karlsson and W. P. Fung-Leung, Br. J. Pharmacol., 2004,
142, 161–171.
17 C. Liu, X. Ma, X. Jiang, S. J. Wilson, C. L. Hofstra, J. Blevitt,
J. Pyati, X. Li, W. Chai, N. Carruthers and T. W. Lovenberg,
Mol. Pharmacol., 2001, 59, 420–426.
18 M. O'Reilly, R. Alpert, S. Jenkinson, R. P. Gladue, S. Foo,
S. Trim, B. Peter, M. Trevethick and M. Fidock, J. Recept.
Signal Transduction, 2002, 22, 431–448.
35 H. D. Lim, R. M. van Rijn, P. Ling, R. A. Bakker,
R. L. Thurmond and R. Leurs, J. Pharmacol. Exp. Ther.,
2005, 314, 1310–1321.
36 H. Engelhardt, I. J. P. de Esch, D. Kuhn, R. A. Smits,
O. P. Zuiderveld, J. Dobler, M. Mayer, S. Lips, H. Arnhof,
D. Scharn, E. E. J. Haaksma and R. Leurs, Eur. J. Med.
Chem., 2012, 54, 660–668.
19 K. Rossbach, S. Wendorff, K. Sander, H. Stark, R. Gutzmer, 37 J. A. Jablonowski, C. A. Grice, W. Chai, C. A. Dvorak,
T. Werfel, M. Kietzmann and W. Baumer, Exp. Dermatol.,
2009, 18, 57–63.
20 R. L. Thurmond, P. J. Desai, P. J. Dunford, W. P. Fung-Leung,
C. L. Hofstra, W. Jiang, S. Nguyen, J. P. Riley, S. Sun,
J. D. Venable, A. K. Kwok, K. S. Ly, J. Wei, S. M. Baker,
P. J. Desai, W. Jiang, S. J. Wilson, R. L. Thurmond,
L. Karlsson, J. P. Edwards, T. W. Lovenberg and
N. I. Carruthers, J. Med. Chem., 2003, 46, 3957–3960.
K. N. Williams, J. P. Edwards and L. Karlsson, J. Pharmacol. 38 R. Kiss, B. Noszal, A. Racz, A. Falus, D. Eros and
Exp. Ther., 2004, 309, 404–413. G. M. Keseru, Eur. J. Med. Chem., 2008, 43, 1059–1070.
21 V. Katritch, V. Cherezov and R. C. Stevens, Trends Pharmacol. 39 J. D. Venable, H. Cai, W. Chai, C. A. Dvorak, C. A. Grice,
Sci., 2012, 33, 17–27.
22 A. J. Kooistra, L. Roumen, R. Leurs, I. J. P. de Esch and C. de
Graaf, Methods Enzymol., 2012, in press.
23 T. Shimamura, M. Shiroishi, S. Weyand, H. Tsujimoto,
G. Winter, V. Katritch, R. Abagyan, V. Cherezov, W. Liu,
J. A. Jablonowski, C. R. Shah, A. K. Kwok, K. S. Ly, B. Pio,
J. Wei, P. J. Desai, W. Jiang, S. Nguyen, P. Ling,
S. J. Wilson, P. J. Dunford, R. L. Thurmond,
T. W. Lovenberg, L. Karlsson, N. I. Carruthers and
J. P. Edwards, J. Med. Chem., 2005, 48, 8289–8298.
G. W. Han, T. Kobayashi, R. C. Stevens and S. Iwata, 40 R. J. Altenbach, R. M. Adair, B. M. Bettencourt, L. A. Black,
Nature, 2011, 475, 65–70.
S. R. Fix-Stenzel, S. M. Gopalakrishnan, G. C. Hsieh,
H. Liu, K. C. Marsh, M. J. McPherson, I. Milicic,
T. R. Miller, T. A. Vortherms, U. Warrior, J. M. Wetter,
N. Wishart, D. G. Witte, P. Honore, T. A. Esbenshade,
A. A. Hancock, J. D. Brioni and M. D. Cowart, J. Med.
Chem., 2008, 51, 6571–6580.
24 C. de Graaf, A. J. Kooistra, H. F. Vischer, V. Katritch,
M. Kuijer, M. Shiroishi, S. Iwata, T. Shimamura,
R. C. Stevens, I. J. de Esch and R. Leurs, J. Med. Chem.,
2011, 54, 8195–8206.
25 R. Kiss, B. Kiss, A. Konczol, F. Szalai, I. Jelinek, V. Laszlo,
B. Noszal, A. Falus and G. M. Keseru, J. Med. Chem., 2008, 41 M. D. Cowart, R. J. Altenbach, H. Liu, G. C. Hsieh, I. Drizin,
51, 3145–3153.
I. Milicic, T. R. Miller, D. G. Witte, N. Wishart, S. R. Fix-
Stenzel, M. J. McPherson, R. M. Adair, J. M. Wetter,
B. M. Bettencourt, K. C. Marsh, J. P. Sullivan, P. Honore,
T. A. Esbenshade and J. D. Brioni, J. Med. Chem., 2008, 51,
6547–6557.
26 F. Sirci, E. P. Istyastono, H. F. Vischer, A. J. Kooistra,
S. Nijmeijer, M. Kuijer, M. Wijtmans, R. Mannhold,
R. Leurs, I. J. P. de Esch and C. de Graaf, 2012, Poster
presentation:
30th
Noorwijkerhout–Camerino–Cyprus
This journal is ª The Royal Society of Chemistry 2013
Med. Chem. Commun., 2013, 4, 193–204 | 203

View Article Online
MedChemComm
Concise Article
42 J. Liu, Y. Li, H. X. Zhang, S. W. Zhang and L. Yang, J. Mol. 61 I. Kufareva, M. Rueda, V. Katritch, Participants, G. D.,
Model., 2012, 18, 991–1001. R. C. Stevens and R. Abagyan, Structure, 2011, 19, 1108–1126.
43 S. P. Meduna, B. M. Savall, H. Cai, J. P. Edwards, 62 M. Michino, E. Abola, Participants, G. D., C. L. Brooks, III,
R. L. Thurmond and P. M. McGovern, Bioorg. Med. Chem.
Lett., 2011, 21, 3113–3116.
J. S. Dixon, J. Moult and R. C. Stevens, Nat. Rev. Drug
Discovery, 2009, 8, 455–463.
44 C. E. Mowbray, A. S. Bell, N. P. Clarke, M. Collins, 63 C. de Graaf and D. Rognan, Curr. Pharm. Des., 2009, 15,
R. M. Jones, C. A. Lane, W. L. Liu, S. D. Newman, 4026–4048.
M. Paradowski, E. J. Schenck, M. D. Selby, N. A. Swain and 64 D. A. Goldfeld, K. Zhu, T. Beuming and R. A. Friesner, Proc
D. H. Williams, Bioorg. Med. Chem. Lett., 2011, 21, 6596–
6602.
45 K. Sander, T. Kottke, Y. Tanrikulu, E. Proschak, L. Weizel,
Natl Acad Sci USA, 2011, 108, 8275–8280.
65 C. de Graaf, N. Foata, O. Engkvist and D. Rognan, Proteins,
2008, 71, 599–620.
E. H. Schneider, R. Seifert, G. Schneider and H. Stark, 66 E. Stjernschantz and C. Oostenbrink, Biophys. J., 2010, 98,
Bioorg. Med. Chem., 2009, 17, 7186–7196. 2682–2691.
46 T. Werner, K. Sander, Y. Tanrikulu, T. Kottke, E. Proschak, 67 M. Bruysters, A. Jongejan, A. Akdemir, R. A. Bakker and
H. Stark and G. Schneider, Chembiochem, 2010, 11, 1850–
1855.
47 P. H. Keizers, C. de Graaf, F. J. de Kanter, C. Oostenbrink,
K. A. Feenstra, J. N. Commandeur and N. P. Vermeulen,
J. Med. Chem., 2005, 48, 6117–6127.
R. Leurs, J. Biol. Chem., 2005, 280, 34741–34746.
68 A. Heifetz, G. B. Morris, P. C. Biggin, O. Barker, T. Fryatt,
J. Bentley, D. Hallett, D. Manikowski, S. Pal, R. Reifegerste,
M. Slack and R. Law, Biochemistry, 2012, 51, 3178–3197.
69 R. I. Allen, K. J. Box, J. E. Comer, C. Peake and K. Y. Tam,
J. Pharm. Biomed. Anal., 1998, 17, 699–712.
48 A. C. Kruse, J. Hu, A. C. Pan, D. H. Arlow, D. M. Rosenbaum,
E. Rosemond, H. F. Green, T. Liu, P. S. Chae, R. O. Dror, 70 G. Volgyi, R. Ruiz, K. Box, J. Comer, E. Bosch and K. Takacs-
D. E. Shaw, W. I. Weis, J. Wess and B. K. Kobilka, Nature,
2012, 482, 552–556.
49 O. Korb, T. Stutzle and T. E. Exner, J. Chem. Inf. Model., 2009,
49, 84–96.
Novak, Anal. Chim. Acta, 2007, 583, 418–428.
71 Chemical Computing Group (CCG) MOE (Molecular
Operating Environment) v.2009.10, Montreal, 2009, http://
www.chemcomp.com/soware.htm.
¨
50 O. Korb, T. Stutzle and T. E. Exner, Lect Notes Comput Sci, 72 D. A. Case, T. A. Darden, T. E. Chatman III,
2006, 4150, 247–258.
51 E. P. Istyastono. PhD Thesis, 2012.
C. W. J. Simmerling, R. E. Duke, R. Luo, M. Crowley,
R. C. Walker, W. Zhang, K. M. Merz, B. Wang, S. Hayik,
A. Roitberg, G. Seabra, I. Kolossvary, K. F. Wong,
F. Paesani, J. Vanicek, X. Wu, S. R. Brozell, T. Steinbrecher,
H. Gohlke, L. Yang, C. Tan, J. Mongan, V. Hornak, G. Cui,
D. H. Mathews, M. G. Seetin, C. Sagui, V. Babin and
P. A. Kollman, AMBER10, 2008.
52 I. Gantz, J. DelValle, L. D. Wang, T. Tashiro, G. Munzert,
Y. J. Guo, Y. Konda and T. Yamada, J. Biol. Chem., 1992,
267, 20840–20843.
53 K. Ohta, H. Hayashi, H. Mizuguchi, H. Kagamiyama,
K. Fujimoto and H. Fukui, Biochem. Biophys. Res. Commun.,
1994, 203, 1096–1101.
54 L. Cocco, J. P. Groff, C. Temple, Jr, J. A. Montgomery,
R. E. London, N. A. Matwiyoff and R. L. Blakley,
Biochemistry, 1981, 20, 3972–3978.
55 P. Czodrowski, I. Dramburg, C. A. Sotriffer and G. Klebe,
Proteins, 2006, 65, 424–437.
73 E. Urizar, S. Claeysen, X. Deupi, C. Govaerts, S. Costagliola,
G. Vassart and L. Pardo, J. Biol. Chem., 2005, 280, 17135–
17141.
74 H. J. C. Berendsen, J. P. M. Postma, W. F. van Gunsteren,
A. DiNola and J. R. Haak, J. Chem. Phys., 1984, 81, 3684–3690.
75 Y. Duan, C. Wu, S. Chowdhury, M. C. Lee, G. Xiong,
W. Zhang, R. Yang, P. Cieplak, R. Luo, T. Lee, J. Caldwell,
J. Wang and P. Kollman, J. Comput. Chem., 2003, 24, 1999–
2012.
56 G. J. Sterk, H. van der Goot and H. Timmerman, Eur. J. Med.
Chem., 1984, 19, 545.
57 H. D. Lim, R. A. Smits, R. A. Bakker, C. M. van Dam, I. J. de
Esch and R. Leurs, J. Med. Chem., 2006, 49, 6650– 76 J. Wang, R. M. Wolf, J. W. Caldwell, P. A. Kollman and
6651. D. A. Case, J. Comput. Chem., 2004, 25, 1157–1174.
58 B. Lenarcik and M. Wisniewski, Polish. J. Chem, 1983, 57, 77 J. Wang, W. Wang, P. A. Kollman and D. A. Case, J. Mol.
735–744. Graphics Modell., 2006, 25, 247–260.
59 P. R. Reddy, N. Raju, K. S. Rao and A. Shilpa, Indian J. Chem., 78 G. Marcou and D. Rognan, J. Chem. Inf. Model., 2007, 47,
2009, 48, 761–768. 195–207.
60 T. Vitali, M. Impicciatore, C. Ferrari and G. Morini, Farmaco, 79 J. D. Durrant, C. A. de Oliveira and J. A. McCammon, J. Mol.
Ed. Sci., 1980, 35, 366–379.
Graphics Modell., 2011, 29, 773–776.
204 | Med. Chem. Commun., 2013, 4, 193–204
This journal is ª The Royal Society of Chemistry 2013