Development of a pharmacophore model for histamine H3 receptor antagonists, using the newly developed molecular modeling program SLATE

Article abstract of DOI:10.1021/jm001109k

New molecular modeling tools were developed to construct a qualitative pharmacophore model for histamine H₃ receptor antagonists. The program SLATE superposes ligands assuming optimum hydrogen bond geometry. One or two ligands are allowed to flex in the procedure, thereby enabling the determination of the bioactive conformation of flexible H₃ antagonists. In the derived model, four hydrogen-bonding site points and two hydrophobic pockets available for binding antagonists are revealed. The model results in a better understanding of the structure-activity relationships of H₃ antagonists. To validate the model, a series of new antagonists was synthesized. The compounds were designed to interact with all four hydrogen-bonding site points and the two hydrophobic pockets simultaneously. These ligands have high H₃ receptor affinity, thereby illustrating how the model can be used in the design of new classes of H₃ antagonists.

Full text of DOI:10.1021/jm001109k

1666
J . Med. Chem. 2001, 44, 1666-1674
Articles
Develop m en t of a P h a r m a cop h or e Mod el for Hista m in e H3 Recep tor
An ta gon ists, Usin g th e New ly Develop ed Molecu la r Mod elin g P r ogr a m SLATE
Iwan J . P. De Esch,*,§ J ames E. J . Mills,† Tim D. J . Perkins,† Giuseppe Romeo,‡,# Marcel Hoffmann,‡
Kerstin Wieland,‡ Rob Leurs,‡ Wiro M. P. B. Menge,‡ Paul H. J . Nederkoorn,‡ Philip M. Dean,§ and
Henk Timmerman‡
De Novo Pharmaceuticals, St. Andrews House, 59 St. Andrews Street, CB2 3DD Cambridge, UK, Drug Design Group,
Department of Pharmacology, University of Cambridge, Tennis Court Road, CB2 1QJ Cambridge, UK, and
Leiden/ Amsterdam Center for Drug Research (LACDR), Division of Medicinal Chemistry, Department of Pharmacology,
Faculty of Chemistry, Vrije Universiteit, De Boelelaan 1083, 1081 HV Amsterdam, The Netherlands
Received November 6, 2000
New molecular modeling tools were developed to construct a qualitative pharmacophore model
for histamine H3 receptor antagonists. The program SLATE superposes ligands assuming
optimum hydrogen bond geometry. One or two ligands are allowed to flex in the procedure,
thereby enabling the determination of the bioactive conformation of flexible H3 antagonists.
In the derived model, four hydrogen-bonding site points and two hydrophobic pockets available
for binding antagonists are revealed. The model results in a better understanding of the
structure-activity relationships of H3 antagonists. To validate the model, a series of new
antagonists was synthesized. The compounds were designed to interact with all four hydrogen-
bonding site points and the two hydrophobic pockets simultaneously. These ligands have high
H3 receptor affinity, thereby illustrating how the model can be used in the design of new classes
of H3 antagonists.
In tr od u ction
ular modeling studies can give a fresh impetus to the
design of new ligands. Recently, a pharmacophore model
for histamine H3 antagonists has been described.9 The
interaction of selected H3 ligands with an aspartate
residue of the receptor that is available for ligand
binding was investigated. As two distinct lipophilic
pockets available for ligand binding were revealed, the
derived model explains the differences in structure-
activity relationship (SAR) observed for the lipophilic
tails of different classes of antagonists. However, de-
tailed information about the relative location and shape
of the distinct lipophilic pockets could not be obtained
as the computational costs of the methods applied (a
density functional approach using parallel supercom-
puters) prompted the truncation of the lipophilic tails
of the antagonists to more manageable methyl groups.
Furthermore, several classes of H3 antagonists had to
be omitted in the preliminary model as they lack a basic
moiety that interacts with the aspartate residue.
Histamine mediates its actions via the stimulation of
three distinct receptor subtypes, H1, H2, and H3.1
Selective antagonists of H1 and H2 receptors have been
very successful in the treatment of allergic reactions and
gastric ulcers, respectively, but the therapeutic use of
H3 receptor-related drugs has yet to be established.
However, since the discovery of the H3 receptor in 1983
by Arrang and co-workers,2 considerable progress has
been made in understanding the role of the H3 receptor
in (patho)physiology. The receptor controls neuronal
synthesis of histamine3 and, in addition, regulates the
release of the neurotransmitter into the synaptic cleft.4
Furthermore, it acts as a heteroreceptor at, e.g., sero-
toninergic, noradrenergic, cholinergic, dopaminergic,
and peptidergic neurons.5 H3 receptors have been
identified in peripheral tissues,6 but the highest densi-
ties have been found in distinct areas of the central
nervous system, especially in the areas that are associ-
ated with cognition.7 Pharmacological studies have
provided clear indications for the clinical use of selective
H3 ligands, and centrally active H3 antagonists are
considered potential therapeutic agents to treat neuro-
logical disorders such as obesity, epilepsy, sleeping
disorders, and cognition and memory deficits.8 Molec-
Molecular similarity studies of the H3 antagonists can
be deceptive as several studies10 have indicated that
different classes of antagonists interact with different
sets of (receptor) site points, i.e., indicating partial
similarity.11,12 Furthermore, modeling studies are seri-
ously hampered by the flexibility of the H3 ligands (vide
infra).
* Corresponding author. Tel: +44 1223 488810. Fax: +44 1223
488899. E-mail: Iwan.deEsch@denovopharma.com.
§ De Novo Pharmaceuticals.
To overcome the described problems, a new molecular
modeling tool has been developed. The program SLATE
superposes two flexible ligands assuming optimum
hydrogen-bonding geometry. During the optimization,
the ligands are represented by specific points used for
† University of Cambridge.
‡ Vrije Universiteit.
#
Present address: Dipartimento di Scienze Farmaceutiche, Uni-
versita` di Catania, v.le A. Doria 6, Catania, Italy.
10.1021/jm001109k CCC: $20.00 © 2001 American Chemical Society
Published on Web 04/25/2001

Pharmacophore Model for Histamine H3 Antagonists
J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 11 1667
F igu r e 1. Schematic representation of the objective function calculated by the program SLATE. In this example, the points used
for superposition are the ligand hydrogen-bonding acceptor atoms (I), the receptor hydrogen-bonding acceptor atoms (i, ii, and
iii), and the two points on either side of the vector passing through aromatic centroids perpendicular to aromatic rings (1 and 2).
The distances between these points are expressed in the distance matrix, and a difference distance matrix is calculated by
subtraction of distance matrixes. The objective function in SLATE is the sum of the difference distance matrix elements. See text
for details.
superposition, i.e., ligand hydrogen-bonding acceptor
atoms, receptor hydrogen-bonding acceptor atoms (given
by projecting ligand donor hydrogen atoms toward the
optimum hydrogen-bonding position of the site), and
aromatic rings. For each ligand, the 3D arrangement
of the points used for superposition (the “pharmaco-
phore”) is expressed as a distance matrix (Figure 1).
After every conformational change the difference dis-
tance matrix is calculated by subtraction of the distance
matrixes. The objective function that is minimized by
SLATE is the sum of the difference distance matrix
elements, i.e., the score for the degree of similarity of
the “pharmacophores” described by the two ligands.
Simulated annealing is used for this minimization,
which enables the rapid identification of good solutions,
ideally the global minimum.13 The procedure involves
changing only the torsion angles and the relative or-
dering of corresponding points, thus being rapid enough
to allow one or both ligands to flex. The resulting,
optimized conformations are superposed (by rotation
and transformation, while keeping the derived confor-
mations fixed) using the MATFIT algorithm,14 a so-
called rigid fit procedure. Details and validation of
SLATE have been described elsewhere.15
Th e Liga n d s. Different classes of H3 antagonists are
known; for a thorough review, the reader is referred to
the literature.16,17 One highly potent representative of
the different classes is shown in Figure 2. The biological
data stem from different pharmacological assays and
are difficult to compare with respect to scale, error, and
additional influences, restricting molecular modeling
studies to more qualitative approaches. The most con-

1668 J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 11
De Esch et al.
F igu r e 2. H3 antagonists studied. The biological data are determined by evaluation of the influence of the compound on electrically
evoked, cholinergic contractions of guinea pig intestine preparations (pA2) or by evaluation of the influence of the compound on
K+-stimulated [3H]-histamine release on rat cortex (pKi).
served feature of H3 ligands is the 4(5)-substituted
imidazole ring that is essential for the activity. Whereas
additional substituents on the imidazole ring lead to a
dramatic reduction in activity, replacement of the
imidazole by other functional groups results in com-
pounds with no H3 activity at all. This crucial role of
the imidazole ring in most classes of H3 antagonists
indicates that at least this part of the ligands binds to
the same receptor site and that this interaction is
compulsory. It has to be noted that, very recently, new
classes of H3 antagonists have been described that lack
an imidazole ring.18 It has yet to be established whether
these compounds interact at the same binding site of
the H3 receptor as the imidazole-containing ligands.
Several classes of antagonists shown in Figure 2 have
an imidazole side chain that contains an N-H moiety
which can interact with a hydrogen-bonding acceptor
site (1, 2, 3, 7, 8, 9, 10, 11, 13). It has been shown that
these basic pharmacophoric elements can interact with
a hydrogen-bonding acceptor site point.9,10 Attachment
of a lipophilic residue (“tail”) to this basic moiety can
significantly influence the H3 activity. For many classes
of antagonists, cycloalkyl groups and (halogenated)
benzyl groups lead to the most potent compounds within
the given series.
Some classes of antagonists lack an N-H moiety in
their imidazole side chain (4, 5, 6, 12, 14). Such
compounds bind at the imidazole-binding site and a
lipophilic pocket, obviously without interacting with the
aforementioned hydrogen-bonding site point. The avail-
able data suggest that these ligands have a similar SAR
with respect to the terminal lipophilic moiety, i.e.,
cycloalkyl or substituted benzyl groups leading to the
most potent compounds. However, the SAR concerning
the terminal lipophilic moiety throughout the different
classes of H3 antagonists is far from unambiguous, and
many authors have suggested different binding sites for
H3 antagonists. For example, Stark and co-workers have
proposed a different, unique binding mode for 3 and its
derivatives, as substitution of the isothiourea group by
halogenated benzyl groups leads to far more potent
compounds than substitution with cycloalkyl groups.19
Several authors have suggested an alternative binding
mode for thioperamide (1) and its derivatives because
these compounds also have a peculiar SAR concerning
the lipophilic tail.20-22 Vollinga et al. have described a
class of antagonists (represented by 8) lacking a sig-
nificant influence of the type of lipophilic terminus on
the activity.23 None of these findings have been ratio-
nalized firmly.
Meth od s
The molecular coordinates of the ligands were con-
structed using MACROMODEL.24 The basic nitrogen
atoms in the side chain of the ligands were protonated
except the nitrogen atoms of the urea, thiourea, and
carbamate groups.25,26 Since, at present, no indication
exists about the bioactive tautomeric form of the imi-
dazole moiety,27 the Nt-H tautomer was arbitrarily
selected for all investigated compounds. Aromatic re-
gions were included in the matching procedure to ensure
a tight superposition of the imidazole rings. The torsion
angles of all single bonds were affected by the optimiza-
tion procedure. Furthermore, the bonds of the isothio-
urea, thiourea, and amidine moieties were allowed to
flip (180°) during the optimization to generate all
possible configurations. For the structures containing
a piperidine ring, all different ring conformers were
generated and used as starting geometries. To this end,
MACROMODEL24 was used for molecular mechanics
conformational analysis of the different possible ring
structures (using the ring closure bond option). By

Pharmacophore Model for Histamine H3 Antagonists
J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 11 1669
F igu r e 3. Construction of the pharmacophore by superimpos-
ing 1 and 2, letting both compounds flex. The fit with the
highest steric similarity is shown (stereoview). Carbon atoms
are shown in green, oxygen atoms in red, nitrogen atoms in
blue, sulfur atoms in yellow, and hydrogen atoms in white.
F igu r e 4. Superposition of ligands 1, 3, and 9 (stereoview)
illustrates the position of the four hydrogen-bonding site points
(A-D) and two lipophilic pockets (1 and 2) and gives an
indication about the steric requirements of the derived phar-
macophore. The most likely positions of the receptor site points
were calculated using the program DOH.31 Acceptor site points
are shown in magenta and donor site points in yellow.
rotating all bonds through 360° with increments of 10°,
a large number of conformations was generated. These
conformations were energy optimized using the Amber
force field28 with the program BATCHMIN 2.724 in order
to obtain low-energy conformations.
All trials using SLATE were run under the default
annealing conditions.15 Null correspondences were in-
troduced according to literature procedures.29,30 The so-
called difference-distance matrix distance threshold
parameter (see Mills and co-workers15 for more details)
was set equal to the high accuracy of 0.0001, to force
exact superposition of pharmacophore points where
possible. Fifty trials for each pair of molecules were run,
and the trials in each set were ranked according to their
steric similarity, calculated as the fraction of surface
volume overlap by the program PLM.31 Compounds 13
and 14 were fitted onto the pharmacophore using
PSEUDO.32 This program superposes a flexible ligand
on a template by maximizing the overlap of the molec-
ular skins. For each compound, 50 trials were run (using
the default simulated annealing settings15). The fits
with the best steric score were selected. For the super-
posed structures, the most likely positions of the comple-
mentary hydrogen-bonding atoms in the binding site
were predicted using the program DOH.30
Con str u ction of th e P h a r m a cop h or e. Initially, the
lipophilic tails of the antagonists used to construct the
pharmacophore were truncated to methyl groups to
facilitate steric evaluation of the fits and to circumvent
the anticipated partial similarity problems that may
accompany these substructures (vide supra). The ap-
proach of pairwise matching was applied, i.e., looking
for one unique conformation of a reference ligand that
is found in superposition trials with different ligands.
Identical results were obtained irrespective of the
particular ligand that was used as the reference struc-
ture. Here, the procedure will be described for thiop-
eramide (1) as the reference structure. First, 1 (with
the piperidine ring in the chair conformation) and
GT2331 (2) were superposed by running 50 trials of
SLATE, allowing both compounds to flex. The fit with
the best steric score is shown in Figure 3. In total, six
different conformations of 1, called set I, were found to
match different conformations of 2. In a similar experi-
ment, thioperamide (1) and clobenpropit (3) were su-
perposed, allowing both ligands to flex. Many different
conformations of the ligands within this set II were
found. However, comparison of the conformations of
thioperamide (1) in set I and set II revealed that only
one conformation was found in both sets. Furthermore,
this unique conformation of thioperamide was found in
the superpositions that had the highest steric similarity
score in both sets I (Figure 3) and II (not shown).
Therefore, this conformation of thioperamide, and the
corresponding conformations of 2 and 3, were accepted
as the bioactive conformations. A unique pharmaco-
phore could only be found when the piperidine ring of 1
was in the chair conformation. When the piperidine ring
was (fixed) in a different conformation, no identical
conformations of the reference ligand 1 in the different
sets could be found.
As can be seen in Figure 3, the methyl groups (that
substitute the lipophilic tails) of the antagonists 1 and
2 have a different position and orientation, indicating
that this procedure finds two lipophilic pockets available
for antagonist binding, thereby validating earlier find-
ings.9 Figure 3 also illustrates that it is not necessary
to force the basic nitrogen atoms of the ligands to occupy
the same position in space. A certain degree of positional
freedom for these substructures is allowed to obtain an
optimal position to interact with the complementary
hydrogen-bonding site points, hence taking into account
the directionality of the intermolecular hydrogen bond.
The lipophilic tail of 1 and 2, a cyclohexyl group in
both cases, was added using the program MacroModel.
This substructure was energy-optimized using the
Amber force field28 while the positions of all other atoms
of the template were frozen. The relative position of the
lipophilic tail of clobenpropit (3), a 4-Cl-benzyl group,
was determined using SLATE. Again, the approach of
pairwise matching was used. The complete flexible
antagonists 4, 5, and 6, were matched with clobenpropit
3 that was fixed in the bioactive conformation (as
described in a previous subsection) except for the
lipophilic tail and the bonds of the isothiourea group,
which were allowed to flip 180°. Comparing the three
sets as described before revealed the unique relative
position of the lipophilic tails.
The superposed structures describe the relative posi-
tion and orientation of the imidazole moieties, the basic
groups in the imidazole side chain, and the lipophilic
tails of these ligands (see Figure 4). Subsequent ligands

1670 J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 11
De Esch et al.
Ta ble 1. Histamine H3 Antagonistic Activity and Calculated
Torsion Angle æ of Two Selected Thioperamide Derivatives as
Published in a QSAR Study by Windhorst and Co-workers20
cmpd
R
pA2 ((sem)
æ, deg
15
16
2-Cl
4-Cl
8.21 (0.09)
7.21 (0.08)
-168.5
-122.2
F igu r e 5. Superposition of ligands 1-12 (stereoview). Hy-
drogen atoms attached to carbon atoms are omitted.
were fitted onto this pharmacophore. In all cases, the
fit with the highest steric score was selected. This
procedure reveals four hydrogen-bonding site points.
Obviously, the two site points that can interact with the
basic moiety in the imidazole side chains of the H3
ligands introduce another cause for partial similarity.
It has to be noted that the position of both side points
was confirmed by using different combinations of refer-
ence and subsequent structures and using SLATE to
determine the bioactive conformations by allowing all
structures to flex.
F igu r e 6. Detailed description of the shape of pocket 2 by
superposing 1, 8 (yellow), 15, and 16. See text for details.
In this QSAR study, using 15 thioperamide derivatives,
it was revealed that the latter effect was caused by the
unique character of the piperidine-1-carbothioic acid
amide moiety. A correlation was found between the
activity of the different analogues and the torsion angle
æ of these compounds (illustrated in Table 1).
Incorporation into our pharmacophore model of two
of these compounds (15 and 16), frozen in the conforma-
tion predicted by Windhorst et al,20 illustrates that the
2-Cl benzyl group of 15 uses the same plane in 3D space
as the methyl group of 8 and the cyclohexyl group of 1
(Figure 6). However, the 4-Cl benzyl group of the less
active compound 16 is twisted out of this plane. Thus,
the more active ortho-substituted derivative is in better
steric agreement with the pharmacophore for H3 an-
tagonists, as was already implied by the aforementioned
QSAR study.
Resu lts a n d Discu ssion
The binding site for histamine H3 receptor antagonists
can be described by four hydrogen-bonding site points
and two lipophilic pockets. The relative orientation of
these features is shown in Figure 4 by superposition of
ligands 1, 3, and 9, which also gives an excellent
illustration of the steric requirements of the derived
pharmacophore. The imidazole moiety of the ligands
interacts with site points A and B. Basic nitrogen atoms
of the imidazole side chain of the ligands can interact
with site points C and/or D. Only clobenpropit (3) is able
to interact with all four hydrogen-bonding site points
simultaneously. This might explain the high potency of
this antagonist.
Figure 5 shows the ligands 1-12 which were super-
posed using SLATE. Pocket 1 is occupied by antagonists
2, 3, 4, 5, 6, 7, 10, 11, and 12. These antagonists have
cycloalkyl- and (substituted-) benzyl- groups. The hy-
drophobic region is easily accessible, and (as is apparent
from the SAR) with the proper lipophilic tail, a high
increase in affinity is obtained. The lipophilic tails of 1
and 8 interact with hydrophobic pocket 2 (Figure 5). The
entrance to this pocket seems to be rather narrow, being
enclosed by the two site points C and D. The thiourea
moieties of ligands 1 and 8 are only partly overlapping,
so the orientation, and hence the SAR, of the lipophilic
tails are different.
Surprisingly, compound 13 could not be fitted into the
pharmacophore using SLATE. Using PSEUDO under
the default conditions,32 a conformation of ligand 13 was
derived (Figure 7) in which the amidine moiety does not
interact with the hydrogen-bonding acceptor site points
C or D, explaining why SLATE did not find this
conformation. No data for the role of the amidine group
are available, but the theoretical considerations pre-
sented here indicate that this group is not involved in
hydrogen-bonding interaction with the receptor. The
program PSEUDO was also used to fit compound 14.
No thorough structural variations of the lipophilic tail
of this antagonist (i.e., the tert-butyl group) have been
reported. Therefore, no experimental data are available
that could indicate whether the compound interacts
with pocket 1 or pocket 2. The program PSEUDO
indicates that the tert-butyl group of 14 occupies pocket
2 (Figure 8). One of the methyl groups of this moiety is
positioned just above the center of the cyclohexyl ring
Detailed information about the shape of pocket 2 was
obtained by incorporation of the results of a QSAR
study.20 It was reported that substitution of the cyclo-
hexyl group of 1 by substituted benzyl groups leads to
ligands with reduced activity. The position of the sub-
stituent on the benzyl group had a peculiar effect on
the activity, with ortho substituents being more favor-
able for H3 antagonistic activity than para substituents.

Pharmacophore Model for Histamine H3 Antagonists
J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 11 1671
F igu r e 9. Superposition of thioperamide (1) (in faint) and
clobenpropit (3).
F igu r e 7. Superposition of 1, 3, and 13 (the latter in bold).
Compound 13 cannot interact with site points C or D.
Hydrogen atoms attached to carbon atoms are omitted.
Sch em e 1a
F igu r e 8. Superposition of 1 and 14.
of thioperamide. This is in perfect agreement with
recently published data in which it was shown that
replacement of the cyclohexyl ring of thioperamide (1)
by an adamantyl group results in a slightly more potent
H3 antagonist.33 The chiral cyclopropyl units of 2 and
14 occupy exactly the same position in 3D space as
might be expected when considering the stereospecificity
of the H3 receptor.
a
(i) DCM; (ii) ethanol, ∆.
Sch em e 2a
Va lid a tion of th e P h a r m a cop h or e. Design a n d
Syn th esis. To validate the presence of two hydrophobic
pockets available for H3 antagonist binding, a small
series of compounds was designed and synthesized, and
the affinity of the ligands for the H3 receptor was
determined. For this study, clobenpropit (3) was an ideal
lead compound, as a (substitutable) hydrogen atom of
the isothiourea moiety is pointing directly toward the
second hydrophobic pocket. On the basis of the phar-
macophore model, the receptor can accommodate a
cyclohexyl group in this position, as is illustrated in
Figure 9 by the superposition of clobenpropit (3) and
thioperamide (1). In addition to the interaction with two
hydrophobic pockets, the compounds have interaction
with all four hydrogen-bonding site points.
a
(i) ethanol, HBr, ∆.
To investigate the necessity of the imidazole moiety,
the two nonimidazole compounds 26 and 27 were
synthesized, via an analogous route (Scheme 2).
P h a r m a cology. The affinity for the histamine H3
receptor was determined by displacement of the radio-
ligand [3H] NR-methylhistamine using membrane ho-
mogenates of rat cerebral cortex.
The ligands 22-25 were synthesized according to
Scheme 1. Intermediate 4-(3-bromo-propyl)-1H-imida-
zole hydrobromide (17) was treated with thiourea
compounds (18-21) to yield the target ligands (22-25).
The thiourea compounds (18-21) were prepared from
commercially available isothiocyanates and the corre-
sponding amines.
Resu lts
Substitution of the isothiourea group of clobenpropit
by a methyl or ethyl group results in a significant
decrease of H3 affinity (Table 2). However, with the

1672 J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 11
De Esch et al.
Ta ble 2. Histamine H3 Receptor Binding Affinity of the Designed Compounds
cmpd
name
R1
R2
R3
pKi
3
22
23
24
25
26
27
clobenpropit
VUF 5224
VUF 5225
VUF 5228
VUF 5642
VUF 5649
VUF 5650
4-Cl-benzyl
4-Cl-benzyl
4-Cl-benzyl
4-Cl-benzyl
cyclo-hexyl
4-Cl-benzyl
4-Cl-benzyl
H
4(5)-imidazolyl
4(5)-imidazolyl
4(5)-imidazolyl
4(5)-imidazolyl
4(5)-imidazolyl
H
9.6 ( 0.1
8.5 ( 0.1
8.1 ( 0.1
9.3 ( 0.2
7.9 ( 0.1
<4
5.3 ( 0.07
methyl
ethyl
cyclo-hexyl
cyclo-hexyl
methyl
cyclo-hexyl
H
cyclohexyl-substituted compound 24, the high H3 affin-
ity (pKi ) 9.3 ( 0.2) is regained. Attachment of two
cyclohexyl groups to the isothiourea (compound 25)
leads to a significant decrease in activity (pKi ) 7.9 (
0.1).
Exp er im en ta l Section
Ch em istr y. Gen er a l P r oced u r e. 1H and 13C NMR spectra
were recorded on a Bruker AC-200 spectrometer (unless
indicated otherwise) with tetramethylsilane as an internal
standard. Melting points were determined on an Electrother-
mal IA9200 apparatus and are uncorrected. Solvents were
purified and dried by standard procedures. 4-(3-Bromo-propyl)-
1H-imidazole hydrobromide (17) was prepared according to
literature procedure.34,35
This series of compounds reveals that the H3 receptor
can accommodate antagonists which have two lipophilic
moieties in their imidazole side chain, thereby validat-
ing the pharmacophore model. It is also shown that H3
affinity is influenced not only by the putative hydro-
phobic interaction of two isothiourea substituents and
the two different pockets of the receptor (compare 22
and 24). Additional factors (e.g., electronic effects) have
an influence on the interaction of the ligands with the
receptor (compare 22, 24, and 25, noting that the
electron-releasing substituents on the isothiourea group
may reduce the interaction of the isothiourea moiety and
the two hydrogen-bonding acceptor site points). A
thorough quantitative structure-activity relationship
study is needed to unravel the precise contribution of
the distinct physical properties of the isothiourea sub-
stituents on the H3 affinity.
N-Meth yl-N′-(4-ch lor oben zyl)th iou r ea (18). To a solu-
tion of methyl isothiocyanate (2.19 g, 30.0 mmol) in anhydrous
diethyl ether (25 mL) was added dropwise a solution of
4-chlorobenzylamine (4.25 g, 30.0 mmol) in anhydrous diethyl
ether (25 mL). After vigorously stirring the suspension for 2
h, the precipitated product was collected by filtration, washed
with anhydrous diethyl ether (20 mL), and dried. A white solid
was isolated (5.65 g, 88%) which was pure by TLC (eluent
CHCl3) and was used without further purification. An analyti-
cal sample was recrystallized from toluene. Mp: 109.5-110.5
1
°C. H NMR (CDCl3): δ 2.96 (d, J ) 5.3 Hz, 3H), 4.67 (d, J )
6.0 Hz, 2H), 6.18 (bs, 2H), 7.18-7.24 (m, 4H).
N-(4-Ch lor oben zyl)-N′-eth ylth iou r ea (19). Analogous to
the preparation of 18, using ethyl isothiocyanate. Recrystal-
lized from toluene, yield 90%. Mp: 100-101 °C. 1H NMR
(CDCl3): δ 1.19 (t, J ) 6.7 Hz, 3H), 3.28-3.48 (m, 2H), 4.65
(d, J ) 6.0 Hz, 2H), 5.95 (bs, 1H), 7.18-7.24 (m, 4H).
N-(4-Ch lor oben zyl)-N′-cycloh exylth iou r ea (20). Analo-
gous to the preparation of 18, using cyclohexyl isothiocyanate.
Recrystallized from toluene, yield 76%. Mp: 115.5-116 °C. 1H
NMR (CDCl3): δ 0.95-2.10 (m, 10H), 3.80 (m, 1H), 4.58 (d, J
) 6.0 Hz, 2H), 5.90 (bs, 1H), 6.28 (bs, 1H), 7.10-7.30 (m, 4H).
N,N′-Biscycloh exylth iou r ea (21). Analogous to the prepa-
ration of 18, using cyclohexylamine and cyclohexyl isothiocy-
anate. Recrystallized from toluene, yield 24%. Mp: 180-182
°C. 1H NMR (CDCl3): δ 1.02-2.10 (m, 20H), 3.80 (m, 2H), 5.64
(bs, 2H).
N-(4-Ch lor ob en zyl)-S-[3-(4(5)-im id a zolyl)p r op yl]-N′-
m eth yl-isoth iou r ea Dih yd r obr om id e (22). A solution of 18
(2.70 g, 10.0 mmol) and 4-(3-Bromo-propyl)-1H-imidazole
hydrobromide (17) in ethanol (25 mL) was refluxed for 48 h.
The reaction mixture was concentrated in vacuo, and the
residue was purified by column chromatography (acetate/
methanol, 3/1, v/v). The product was precipitated in acetone
to give a white solid (0.85 g, 18%). Mp: 163.0-164.0 °C.
1H NMR (DMSO-d6): δ 1.80-2.08 (m, 2H), 2.73 (t, J ) 7.5
Hz, 2H), 3.05 (s, 3H), 3.22-3.50 (m, 3.8 H), 4.64 (s, 2H),
7.30-7.52 (m, 5H), 9.04 (s, 1H). Anal. (C15H21Br2ClN4S)
C, H, N.
The deletion of the imidazole heterocycle proves to be
disastrous for H3 affinity (Table 2). These results again
illustrate the major influence of the imidazole nucleus
on the H3 receptor affinity of ligands at this binding
site.10 Interestingly, replacement of the methyl group
of 26 by a cyclohexyl group, leading to 27, results in a
better H3 ligand. A similar improvement can be seen
when comparing the imidazole-containing compounds
22 and 24. However, it has to be noted that straight-
forward comparison of non-imidazole ligands 26 and 27
and the imidazole-containing compounds 22 and 24 is
ambiguous as the non-imidazole compounds might
interact with a different binding site of the H3 receptor.
Con clu sion s
Using newly developed molecular modeling tech-
niques, a pharmacophore model for histamine H3 an-
tagonists has been derived. The binding site of the
histamine H3 receptor is characterized by four hydrogen-
bonding site points and two lipophilic pockets. Although
qualitative by necessity, this study gives an explanation
of the observed differences in SAR found for H3 receptor
antagonists. The new model has been used to design
new antagonists that interact with all four hydrogen-
bonding site points and the two distinct lipophilic
pockets simultaneously. These results illustrate that the
model can be used to design novel classes of antagonists.
N-(4-Ch lor oben zyl)-N′-eth yl-S-[3-(4(5)-im id a zolyl)p r o-
p yl]isot h iou r ea Dih yd r ob r om id e (23). Analogous to the
preparation of 22, using 19. The product was isolated as an
oil (25%). 1H NMR (DMSO-d6): δ 1.20 (t, J ) 7.3 Hz, 3H),
1.75-2.03 (m, 2H), 2.71 (t, J ) 7.5 Hz, 2H), 3.10-3.65 (m, 5.8
H), 4.66 (s, 2H), 7.20-7.58 (m, 5H), 8.98 (s, 1H). Anal.
(C16H23Br2ClN4S) C, H, N.
N-(4-Ch lor oben zyl)-N′-cycloh exyl-S-[3-(4(5)-im idazolyl)-
p r op yl]isoth iou r ea Dih yd r obr om id e (24). Analogous to

Pharmacophore Model for Histamine H3 Antagonists
J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 11 1673
(2) Arrang, J . M.; Garbarg, M.; Schwartz, J .-C. Auto-inhibition of
brain histamine release mediated by a novel class (H3) of
histamine receptor. Nature 1983, 302, 832-837.
(3) Arrang, J .-M.; Garbarg, M.; Schwartz, J .-C. Autoinhibition of
histamine synthesis mediated by presynaptic H3-receptors.
Neuroscience 1987, 13, 149-157.
(4) Arrang, J .-M.; Garbarg, M.; Schwartz, J .-C. Autoregulation of
histamine release in brain by presynaptic H3-receptors. Neuro-
science 1985, 15, 553-562.
the preparation of 22, using 20. After purification by column
chromatography and evaporation of a uniform fraction, the
semisolid residue was suspended in dry diethyl ether and
stirred until complete precipitation. The very hygroscopic
material was filtered and dried. Yield: 22%. Mp: 127.0-130.0
1
°C. H NMR (DMSO-d6): δ 1.00-2.08 (m, 12H), 2.67 (t, J )
7.5 Hz, 2H), 3.31 (t, J ) 7.2 Hz, 2H), 3.79 (m, 1H), 4.68 (s,
2H), 7.29-7.51 (m, 5H), 8.81 (s, 1H). Anal. (C20H29Br2ClN4S)
C, H, N.
(5) Schlicker, E.; Kathmann, M. Modulation of in vitro neurotrans-
mission in the CNS and in the retina via H3 heteroreceptors. In
The histamine H3 receptor; a target for new drugs, 1st ed.; Leurs,
R., Timmerman, H., Eds.; Elsevier Science B. V.: Amsterdam,
1998; pp 13-26.
N,N′-Biscycloh exyl-S-[3-(4(5)-im idazolyl)pr opyl]isoth io-
u r ea Dih yd r obr om id e (25). Analogous to the preparation
of 22, using 21. The product was isolated as a white solid
1
(18%). H NMR (DMSO-d6): δ 1.00-2.08 (m, 22H), 2.74 (t, J
(6) Bartaccini, G.; Coruzzi, G.; Poli, E. Functional role of histamine
) 7.5 Hz, 2H), 3.32 (t, J ) 7.0 Hz, 2H), 3.63-3.90 (m, 2H),
7.47 (s, 1H), 8.95 (s, 1H). Anal. (C19H34Br2N4S) C, N; H: calcd,
6.71; found, 5.73.
H
3 receptors in peripheral tissues. In The histamine H3 receptor;
a target for new drugs, 1st ed.; Leurs, R., Timmerman, H., Eds.;
Elsevier Science B. V.: Amsterdam, 1998; pp 59-112.
(7) Arrang, J .-M.; Garbarg, M.; Lancelot, J .-C.; Lecomte, J .-M.;
Pollard, H.; Robba, M.; Schunack, W.; Schwartz. J .-C. Highly
potent and selective ligands for histamine H3-receptors. Nature
1987, 327, 117-123.
(8) Leurs, R.; Blandina, P.; Tedford, C.; Timmerman, H. Therapeutic
potential of histamine H3 receptor agonists and antagonists.
Trends Pharmacol. Sci. 1998, 19, 177-183.
(9) De Esch, I. J . P.; Timmerman, H.; Menge, W. M. P. B.;
Nederkoorn, P. H. J . A qualitative model for the histamine H3
receptor explaining agonistic and antagonistic activity simul-
taneously. Arch. Pharm. Pharm. Med. Chem. 2000, 333, 254-
260.
(10) De Esch, I. J . P.; Gaffar, A.; Menge, W. M. P. B.; Timmerman,
H. Synthesis and histamine H3 receptor activity of 4-(n-alkyl)-
1H-imidazoles and 4-(ω-phenylalkyl)-1H-imidazoles. Bioorg.
Med. Chem. 1999, 7, 3003-3008.
N-(4-Ch lor oben zyl)-N′-m eth yl-S-pr opyl]isoth iou r ea Hy-
d r obr om id e (26). Analogous to the preparation of 22, using
18 and 1-bromopropane. The crude product was purified by
chromatography (CHCl3/methanol, 95/5, v/v). A white product
1
was isolated (43%). Mp: 107.5-108.5 °C. H NMR (CDCl3): δ
0.96 (t, J ) 7.5 Hz, 3H), 1.50-1.75 (m, 2H), 3.04 (s,3H), 3.47
(t, J ) 6.0 Hz, 2H), 4.59 (s, 2H), 7.60 (br s, 1H), 7.14-7.38 (m,
4H), 9.70 (bs, 1H). Anal. (C12H18BrClN2S) C, H, N.
N-(4-Ch lor ob en zyl)-N′-cycloh exyl-S-p r op yl]isot h io-
u r ea Hyd r obr om id e (27). Analogous to the preparation of
26, using 20. A white solid was obtained (43%). Mp: 146.0-
1
147.0 °C. H NMR (CDCl3): δ 0.94 (t, J ) 7.34 Hz, 3H), 1.03-
1.93 (m, 10H), 1.51-1.68 (m, 2H), 3.11 (t, J ) 5.34 Hz, 2H),
3.64-3.88 (m, 1H), 4.67 (s, 2H), 6.28 (bs, 1H), 7.12-7.41 (m,
4H). Anal. (C17H26BrClN2S) C, H, N.
(11) Mattos, C.; Ringe, D. Multiple binding modes. In 3D QSAR in
drug design: theory, methods and applications; Kubinyi, H., Ed.;
Escom: Leiden, 1993; pp 226-254.
P h a r m a cology. The histamine H3 receptor affinity was
determined on rat cortical membranes with [3H]-NR-methyl-
histamine (81.9 Ci/mmol, NEN life science products, Brussels,
Belgium) according to the method of West et al.36 with
modifications. Briefly, animals were killed by decapitation, and
the cerebral cortex rapidly removed. Rat cortices were homog-
enized in 15 volumes (wt/vol.) of ice-cold Tris/HCl buffer (50
mM Tris/HCl; 5 mM MgCl2, 145 mM NaCl; pH 7.4 at 4 °C)
using an Ultra-Turrax homogenizer (8 s) and a glass-Teflon
homogenizer (four strokes up and down) subsequently. All
subsequent steps were carried out at 0-4 °C. The homogenate
was centrifuged at 800g for 10 min. The pellets were discarded,
and the supernatant was centrifuged for 20 min at 40000g.
The resulting pellet was resuspended, and the last centrifuga-
tion step was repeated. The pellet was resuspended in 1.5
volume (wt/vol.) Tris/HCl buffer to give a final concentration
of ∼300 µg/100 µL and stored in aliquots at -80 °C. Protein
concentration was determined using Biorad protein assay (Bio-
Rad laboratories GmbH, Munich, Germany). Competition
binding experiments were carried out in polypropylene tubes
in a total volume of 400 µL of 50 mM Na+ phosphate buffer
pH 7.4 at 37 °C, containing 30 µg of protein, 1nM of [3H]-NR
methylhistamine, and 0.1 to 10 000 nM of the compound to
be tested. Samples were incubated for 40 min at 25 °C. The
incubation was started by the addition of 100 µL of membranes
(30 µg) and terminated by rapid filtration through polyethyl-
eneimine (0.3 wt %/vol.) pretreated Whatman GF/C filters
using a Brandel filtration apparatus. The filters were washed
twice with 3 mL of ice-cold Tris/HCl buffer (50 mM Tris/HCl;
5 mM MgCl2, 145 mM NaCl; pH 7.4 at 4 °C). The radioactivity
retained on the filters was measured using liquid scintillation
counting. Competition isotherms were analyzed with the
GraphPad Prism software (GraphPad, Intuitive Software for
Science, San Diego, CA). Ki values were determined with the
equation Ki ) IC50/(1 + ([ligand]/Kd)).
(12) Perkins, T. D. J .; Mills, J . E. J .; Dean, P. M. Molecular partial
similarity using surface-volume comparisons. In Computer-
assisted lead finding and optimization; van de Waterbeemd, H.,
Testa, B., Folkers, G., Eds.; VCH: Weinheim, 1997; pp 421-
432.
(13) Barakat, M. T.; Dean, P. M. Molecular structure matching by
simulated annealing. J . Comput.-Aided Mol. Des. 1990, 4, 295-
316.
(14) McLachlan, A. D. Gene duplication in the structural evolution
of chymotrypsin. J . Mol. Biol. 1979, 128, 49-79.
(15) Mills, J . E. J .; De Esch, I. J . P.; Perkins, T. D. J .; Dean, P. M.
SLATE: a method for the superposition of flexible ligands. J .
Comput.-Aided. Mol. Des. 2001, 15, 81-96.
(16) Leurs, R.; Vollinga, R. C.; Timmerman, H. The medicinal
chemistry and therapeutic potentials of ligands of the histamine
H3-receptor. Prog. Drug Res. 1995, 45, 107-165.
(17) Lipp, R.; Stark, H.; Schunack, W. Pharmacochemistry of H3
Receptors: The histamine receptor. In Receptor Biochemistry
and Methodology; Schwartz, J .-C., Haas, H. L., Eds.; Wiley-Liss
Inc.: New York, 1992; Vol. 16; pp 57-72.
(18) Ganellin, C. R.; Leurquin, F.; Piripitsi, A.; Arrang, J .-M.;
Garbarg, M.; Ligneau, X.; Schunack, W.; Schwartz, J .-C. Syn-
thesis of potent non-imidazole histamine H3-receptor antago-
nists. Arch. Pharm. Pharm. Med. Chem. 1998, 331, 395-404.
(19) Stark, H.; Purand, K.; Ligneau, X.; Rouleau, A.; Arrang, J .-M.;
Garbarg, M.; Schwartz, J .-C.; Schunack.; W. Novel carbamates
as potent histamine H3 receptor antagonists with high in
vitro and oral in vivo activity. J . Med. Chem. 1996, 39, 1157-
1163.
(20) Windhorst, A. D.; Timmerman, H.; Worthington, E. A.; Bijloo,
G. J .; Nederkoorn, P. H. J .; Menge, W. M. P. B.; Leurs, R.;
Herscheid, J . D. M. Characterization of the binding site of the
histamine H3 receptor. 3. Synthesis, in vitro pharmacology and
QSAR of a series of monosubstituted benzyl analogues of
thioperamide. J . Med. Chem. 2000, 43, 1754-1761.
(21) Ganellin, C. R.; Hosseini, S. K.; Khalaf, Y. S.; Tertiuk, W.;
Arrang, H.-M.; Garbarg, M.; Ligneau, X. Schwartz, J .-C. Design
of potent non-thiourea H3-receptor histamine antagonists. J .
Med. Chem. 1995, 38, 3342-3350.
(22) Plazzi, P. V.; Bordi, F.; Mor, M.; Silva, C.; Morini, G.; Caretta,
A.; Barocelli, E.; Vitali, T. Heteroarylaminoethyl and het-
eroarylthioethyl imidazoles. Synthesis and H3-receptor affinity.
Eur. J . Med. Chem. 1995, 30, 881-889.
Ackn owledgm en t. This research was in part (I.D.E.)
supported by the Dutch Technology Foundation (STW).
(23) Vollinga, R. C.; Menge, W. M. P. B.; Leurs, R.; Timmerman, H.
New analogs of burimamide as potent and selective histamine
H3 receptor antagonists, the effect of chain length variation of
the alkyl spacer and modifications of the N-thiourea substituent.
J . Med. Chem. 1995, 38, 2244-2250.
Refer en ces
(1) Leurs, R.; Smit, M. J .; Timmerman, H. Molecular pharmacologi-
cal aspects of histamine receptors. Pharmacol. Ther. 1995, 66,
413-463.

1674 J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 11
De Esch et al.
(24) Mohamadi, F.; Richards, N. G. J .; Guida, W. C.; Liskamp, R.;
Lipton, M.; Caufield, C.; Chang, G.; Hendrickson, T.; Still, W.
C. MacroModel -An integrated software system for modeling
organic and bioorganic molecules using molecular mechanics.
J . Comput. Chem. 1990, 11, 440-467.
(25) Smith, S. E.; Rawlins, M. D. Variability in Human Drug
Response; Butterworth: London, 1973; pp 154-165.
(26) Mor, M.; Bordi, F.; Silva, C.; Rivara, S.; Crivori, P.; Vincenzo
Plazzi, P.; Ballabeni, V.; Caretta, A.; Barocelli, E.; Imicciatore,
M.; Carrupt, P.-A.; Testa, B. H3-Receptor antagonists: synthesis
and structure-activity relationships of para- and meta-substi-
tuted 4(5)-phenyl-2-[[2-[4(5)-imidazolyl]ethyl]thio]imidazoles. J .
Med. Chem. 1997, 40, 2571-2578.
(27) De Esch, I. J . P.; Nederkoorn, P. H. J .; Timmerman, H. Molecular
modelling studies of histamine H3 receptor ligands. In The
histamine H3 receptor; a target for new drugs, 1st ed.; Leurs,
R., Timmerman, H., Eds.; Elsevier Science B. V.: Amsterdam,
1998; pp 59-112.
(28) Weiner, S. J .; Kollman, P. A.; Nguyen, D. T.; Case, D. A. An all
atom force field for simulations of proteins and nucleic acids. J .
Comput. Chem. 1986, 7, 230-252.
(30) Mills, J . E. J .; Perkins, T. D. J .; Dean, P. M. An automated
method for predicting the positions of hydrogen-bonding atoms
in binding sites. J . Comput.-Aided Mol. Des. 1997, 11, 229-242.
(31) Perkins, T. D. J .; Mills, J . E. J .; Dean, P. M. Molecular surface-
volume and property matching to superpose flexible and dis-
similar molecules. J . Comput.-Aided Mol. Des. 1995, 9, 479-490.
(32) The program PSEUDO is like the program in ref 12, but
conformational flexibility is allowed during the optimization.
(33) Goto, T.; Sakashita, H.; Murakami, K.; Sugiura, M.; Kondo, T.;
Fukaya, C. Novel histamine H3 receptor antagonists: synthesis
and evaluation of formamidine and S-methylisothiourea deriva-
tives. Chem. Pharm. Bull. 1997, 45, 305-311.
(34) Kivits, G.; Hora, J . J . Heterocycl. Chem. 1975, 12, 577.
(35) Bloemhoff, W.; Kerling, K. Recueil Trav. Chim. Pays Bas 1970,
89, 1181.
(36) West, R. E. J .; Zweig, A.; Shih, N. Y.; Siegel, M. I.; Egan, R. W.;
Clark, M. A. Identification of two receptor subtypes. Mol.
Pharmacol. 1990, 38, 610-613.
(29) Barakat, M. T.; Dean, P. M. Molecular structure matching by
simulated annealing. III. The incorporation of null correspon-
dences into the matching problem. J . Comput.-Aided Mol. Des.
1991, 5, 107-117.
J M001109K