Optimization of a pharmacophore model for 5-HT4 agonists using CoMFA and receptor based alignment

Article abstract of DOI:10.1016/j.ejmech.2005.07.017

Twenty two 5-HT₄ agonists obtained from our laboratory and the recent literature were used to develop a CoMFA model to predict 5-HT₄ agonist activity. Two models were produced and compared for predictivity, the first by alignments ba

Full text of DOI:10.1016/j.ejmech.2005.07.017

European Journal of Medicinal Chemistry 41 (2006) 16–26
http://france.elsevier.com/direct/ejmech
Original article
Optimization of a pharmacophore model for 5-HT4
agonists using CoMFA and receptor based alignment
Magdy N. Iskander a,*, Lok M. Leung a, Trevor Buley a, Fadi Ayad a, Juliana Di Iulio b,
Yean Y. Tan b, Ian M. Coupar b
a The Department of Medicinal Chemistry, Victorian College of Pharmacy, Monash University, 381 Royal Parade, Parkville, Vic. 3052, Australia
b The Department of Pharmaceutical Biology and Pharmacology, Victorian College of Pharmacy,
Monash University, 381 Royal Parade, Parkville, Vic. 3052, Australia
Received 5 April 2005; accepted 13 July 2005
Available online 15 November 2005
Abstract
Twenty two 5-HT4 agonists obtained from our laboratory and the recent literature were used to develop a CoMFA model to predict 5-HT4
agonist activity. Two models were produced and compared for predictivity, the first by alignments based on atom overlapping (model A) and the
second by adding agonist binding site interacting points of the 5-HT4 receptor (model B). Comparison of the two models showed that the q2
value for model A was 0.564 vs. 0.582 for model B. Model B indicated that the predictive power model stems from far lower steric contributions,
0.270 compared to model A′s 0.502. The dominant defining features were the electrostatic contributions for model B, 0.664 up from 0.477 in
model A. The contributions from the LogP factor were minimal, 0.085 in both models. The synthesized compounds showed agonist activity at
μmol level.
© 2005 Elsevier SAS. All rights reserved.
Keywords: 5-HT4; Agonist; Receptor; CoMFA; 5-Hydroxytryptamine
1. Introduction
endings to facilitate colonic propulsion [5] and on sensory
nerve endings where their function is thought to enhance sen-
sory perceptions arising from the abdomen. Therefore, the 5-
HT4 receptor has been identified as a therapeutic target for
treating irritable bowel syndrome (IBS) using selective ago-
nists with limited success to date [6,7]. The functional charac-
terization of 5-HT4 receptors is based on both agonist and an-
tagonist activities and has been extensively reviewed [8]. 5-
HT4 receptor agonists include indoleamine derivatives such as
5-methoxytryptamine (5-MeOT) and 5-carboxyamidotrypta-
mine (5-CT); the substituted benzamide derivatives such as ci-
sapride, renzapride and zacopride; the benzimidazolone deriva-
tives BIMU 1, BIMU 8 and DAU 6236 and the substituted
pyrrolizidine SC 53116. Only a small number of 5-HT4 ago-
nists are used clinically. Examples are the relatively non-selec-
tive agonists cisapride and metoclopramide and the recently
introduced selective partial agonist tegaserod for treating con-
stipation predominant IBS in female patients. These drugs have
relatively low potency when compared to the natural ligand.
Therefore, it is important to develop more potent and selective
5-Hydroxytryptamine (5-HT) is involved in controlling an
array of physiological functions and in promoting pathophysio-
logical conditions in states of excess or deficiency. Fourteen
receptors for 5-HT have been cloned. All, with the exception
of 5-HT3 (an ion channel) and 5-ht5B (unknown coupling), are
members of the GPCR superfamily of membrane-bound recep-
tors [1]. A major relevance of the 5-HT4 receptor is its involve-
ment in controlling several functions of the human large intes-
tine. A population of 5-HT4 receptors is localized in the
circular smooth muscle layer where their stimulation by endo-
genous 5-HT or 5-HT4 agonists leads to relaxation [2] utilizing
the adenylyl cyclase signal transduction system [3]. 5-HT4 re-
ceptors are also located in the mucosa of the colon where their
function is to promote Cl– secretion with associated fluid loss
from the lumen [4]. They are also present on cholinergic nerve
* Corresponding author. Tel.: +61 3 9903 9679; fax: +61 3 9903 9545.
E-mail address: magdy.iskander@vcp.monash.edu.au (M.N. Iskander).
0223-5234/$ - see front matter © 2005 Elsevier SAS. All rights reserved.
doi:10.1016/j.ejmech.2005.07.017

M.N. Iskander et al. / European Journal of Medicinal Chemistry 41 (2006) 16–26
17
agents with better pharmacological and side effect profiles than
those currently available.
template (Fig. 1c). Relative potencies and log P values were
assembled from literature values and from our experimental
results. The steric and electrostatic CoMFA fields were gener-
ated separately, with truncated energy at 30 kcal/mol and
scaled to equalize their weighting in the CoMFA models; stan-
dard CoMFA scaling. A region was created with boundaries
extending 4 Å beyond the largest structure in all directions
using a sp3 probe carrying a unit positive charge. One Å grid
spacing was used with 1728 lattice points to compare the mo-
lecular details.
PLS, leave-one-out (LOO) method regression analysis [20]
was used in conjunction with cross-validation to obtain the op-
timal number of components to be used in the subsequent ana-
lyses. A subset of CoMFA field sample points falling with a
standard deviation of ≤ 2.0 kcal/mol was used to perform PLS
regression analysis. The optimum number of principle compo-
nents in the final non-cross-validated QSAR equations was de-
termined to be that leading to the highest correlation (q2) and
the lowest standard error (Scv) in the LOO cross and validated
predictions. The non-cross-validation was used in the analysis
of the CoMFA results and the predictions.
Previous studies in our laboratories have resulted in a phar-
macophore model based on the molecular structural and phar-
macological properties of known agonists [9]. The pharmaco-
phore model was developed using seven known active
agonists, indolecarbazimidamide, 3-N-isopropylbenzimidazo-
lone amide, 3-N-ethylbenzimidazolone amide and benzamide,
R-zacopride, 5-CT and metoclopramide. Extensive study of the
X-ray structures of these agonists and closely-related com-
pounds, identified important interactions constraining the con-
formational flexibility of the side chains of the agonists. This
knowledge, together with molecular mechanics optimization
methods, allowed us to deduce the likely binding conforma-
tions of the agonists at the 5-HT4 agonist binding site. Super-
imposition of the agonists was carried out using atom-by-atom
fit with respect to 5-HT. This alignment was used to develop
CoMFA models of the likely 5-HT4 agonist binding site. The
models were predictive with cross-validated q2 values of 0.26–
0.51 and final r2 values of 0.96–0.99. Steric, electrostatic and
lipophilic properties of the agonist were all found to be impor-
tant. The final model identified regions in 3D property space
which were important for modulating the activity of agonists.
Consequently, several agonist compounds were designed and
their structures were based on the developed pharmacophore
recommendations. The pharmacological evaluation of these
agonists was carried out using the guinea pig ileum, which is
a standard bioassay for 5-HT4 ligands. The present CoMFA
study included these compounds together with 5-HT and the 5-
HT4-preferring agonists (1–6, 10–15, 17 and 18, Table 1) to
achieve more defined agonist site characteristics.
4. Results and discussion
CoMFA is a 3D-QSAR method that operates on a set of
ligands that have been superimposed to reflect their anticipated
common binding orientation. A bioactive conformation must
be energetically favorable and common among all active com-
pounds having the same binding mode. In this investigation
two CoMFA analysis were developed and compared in order
to achieve a predictive model to improve our earlier pharmaco-
phore [9] for agonist activity. The methods of fitting for the
two separate models vary with a few key points.
2. Experimental methods
4.1. CoMFA model A (alignment based on atom overlapping)
CoMFA [15,16] has been used to develop predictive models
to describe the 3D structure–activity relationship for the com-
pounds of interest based on their steric and electrostatic varia-
tions. The training and test sets used to generate the CoMFA
models consisted of 5-HT, literature compounds (1–6, 10–15
and 17 and 18) and our compounds (7–9, 16, and 19–22), Ta-
ble 1. The design of our synthesized compounds was based on
our early model [9]. Their biological evaluations were also car-
ried out in our laboratory.
In this CoMFA model a structural superimposition template
was used with the positions of the 5-hydroxyl oxygen, the cen-
troid of the adjacent aromatic ring and the charged indolyl
amine ~8 Å from the hydroxyl as the key points for fitting.
The molecules were superimposed using the SYBYL fit atom
routine, using the key molecular points listed above.
This template worked well for all compounds with the 5-HT
scaffold, however it was difficult to properly fit agonists
with different scaffold e.g. BIMU 8, ML 10302 and 2-[(4-acet-
amidopiperidinyl)ethyl]-4-amino-5-chloro-2-methylbenzoate.
These agonists do not have a hydroxyl group adjacent to the
aromatic ring and the distant nitrogens (atom equivalent to the
side chain indolyl N) and they vary significantly in their co-
ordinates relative to the central aromatic ring. However, mo-
lecules that did not contain one or more of the key atoms were
superimposed using the remaining key atoms and alternate
features of similar conformation to the 5-HT template. Addi-
tionally, compounds 1 and 2 were torsionally adjusted away
from their global minima conformation to one of the local
minima in order to fit the 5-HT template optimally (Fig. 1).
3. CoMFA model generation
All computation tasks were carried out using the SYBYL
6.9 molecular modeling package [17] running on a SGI
R5000 workstation. Structures were minimized using the
MMFF94s Force Field and MMFF94 [18,19] charges to a con-
vergence criterion of 0.01 kcal/mol. All valencies were filled
with hydrogens or lone pairs.
Pharmacophore sites were identified and aligned using the
oxygen atom of the 5-hydroxyl group, the centroid of the aro-
matic ring and the nitrogen atom of the indolyl group as the

18
M.N. Iskander et al. / European Journal of Medicinal Chemistry 41 (2006) 16–26
Table 1
Structures of 5-HT4 agonists used in the current studies. They consist of literature compounds (1–6, 10–15, 17, 18, 21 and 5-HT) and compounds designed and
synthesized in our laboratory (7–8, 16, 19, 20 and 22).
Compound
Structure
Relative
potency
Emax
log P
#N-pentyl-3-[(5-hydroxy-1H-indol-3-yl)methylene]
carbazimadamide (1) [10]
7.96
5.01
0.90
2.47
N-phenylethyl-3-[(5-hydroxy-1H-indol-3-yl)methylene]
1.00
2.89
carbazimadamide (2) [10]
3-[(5-Hydroxy-1H-indol-3-yl)methylene] carbazimadamide
(3) [10]
2.51
1.00
1.50
1.00
0.62
0.05
5-HT
ML 10302 (4) [11]
1.00
0.80
1.96
YM-36912 (5) [12]
BIMU 8 (6) [13]
0.91
0.71
0.90
1.00
3.23
3.65
(7)
(8)
(9)
0.69
0.65
0.62
0.72
0.59
0.91
3.35
3.38
3.42
(continued)

M.N. Iskander et al. / European Journal of Medicinal Chemistry 41 (2006) 16–26
19
Table 1 (continued)
Compound
Structure
Relative
potency
Emax
log P
2-[(4-Acetamido piperidinyl) ethyl]-4-amino-5-chloro-2-
0.38
0.59
0.60
methoxybenzoate (10) [11]
5-MeOT (11)
0.25
0.21
1.00
1.00
0.53
3.55
DAU 6236 (12) [13]
Zacopride (13) [14]
0.01
0.80
1.23
5-CT (14) [14]
0.01
1.00
0.70
–0.21
Cisapride (15) [14]
0.008
2.58
(16)
0.005
0.002
0.50
1.00
1.03
1.61
Metoclopramide (17) [10]
YM-47813 (18) [12]
0.001
1.00
0.42
2.29
0.85
(19)
0.0005
(continued)

20
M.N. Iskander et al. / European Journal of Medicinal Chemistry 41 (2006) 16–26
Table 1 (continued)
Compound
Structure
Relative
potency
Emax
log P
(20)
0.0003
1.00
2.69
5-Benzyloxytryptamine (21)
0.0003
0.0003
1.00
0.60
2.69
1.08
(22)
#N-pentyl-3-[(5-hydroxy-1H-indol-3-yl)methylene] carbazimadamide (1) has been chosen for this study because of its known activity in the guinea pig myenteric
plexus, however the actual drug compound is N-pentyl-3-[(5-methoxy-1H-indol-3-yl)methylene] carbazimadamide.
Fig. 1. Ligands 1 and 2 are torsionally adjusted away from their global minima conformation to fit the 5-HT template optimally. (a) Compound 1 is showing the
global minima planar with the guanidine group, ΔE = –9.48. (b) After 60° anticlockwise rotation of the N–C, ΔE = –9.17. and (c) The template used for CoMFA
analysis.
Thus, when an alignment was performed, groups beyond the
atoms used to define the alignment spread greatly in model A.
The pharmacophore features closest to the points of fitting are
a clearly defined region beyond the 5-hydroxyl where steric
bulk is unfavorable, an area above the 5-hydroxyl favoring
positive electrostatic charge and a negative charge around
the distant nitrogen (Fig. 2).
However, the less adequate fitting can be seen clearly in
Fig. 3, where three agonists, two of which fit well based on
the 5-HT scaffold (compounds 1 and 7) and one which does
not (compound 4). A wide spread of side chains shows up in
the features of the pharmacophore model; beyond the distant
nitrogen however, the wide spread fit of the side chains is ex-
pressed in the pharmacophore model as an virtually random
grouping of CoMFA regions, favored and unfavored steric
bulk overlapping along with a scattering of favored and unfa-
vored electrostatic regions.
Such a diversity of fields affects the predictive power of the
pharmacophore model A for the region beyond the distant ni-
trogen where all the variable side chains exist. Thus it was
necessary to create a new agonist alignment and pharmaco-
phore model to accommodate a variety of structures and im-
prove the predictive ability of the model. The obtained data
for model A are in Table 2 column A.
4.2. CoMFA model B (alignment based on receptor binding
site)
The second CoMFA model (B) was developed by combin-
ing the first alignment of model A and the alignment based on

M.N. Iskander et al. / European Journal of Medicinal Chemistry 41 (2006) 16–26
21
Fig. 2. The pharmacophore features developed from CoMFA model A using the complete dataset; all the 23 compounds from Table 1: blue regions favor negative
electrostatic charge, red regions favor positive electrostatic charge, yellow regions where steric bulk is unfavorable and regions of favored steric bulk are colored
green.
the model of the agonist site of the 5-HT4 receptor [21]. This
approach was adopted to accommodate a variety of agonist
structures in one pharmacophore. The molecular model of the
5-HT4 agonist site was created from the crystal structure of
bovine rhodopsin [22,23]. The 5-HT4 receptor model indicated
that hydrophobic interactions of the central aromatic ring were
of less significance when compared to the electrostatic interac-
tions between the 5-hydroxyl and serine 197 and the charged
indolyl amine and aspartic acid 100 (Fig. 4).
Test compounds that did not share a common structure were
superimposed via the strongest match of electrostatic points,
such as the carboxyl oxygen and secondary amine in the com-
pounds zacopride, cisapride and metoclopramide (Fig. 5).
While some regions of the alignment model B may appear
more complex, they play no direct part in predictive agonist
activity. The regions in model B are clearly defined areas of
favorable negative charge (Fig. 5) and an extremely well de-
fined narrow region of favorable steric bulk (Fig. 6).
Fig. 3. CoMFA model A pharmacophore. Reduced dataset shown for
simplicity, compounds shown are 1, 4, 7 and 5-HT.
Table 2
CoMFA analysis data of our early work [9], model A and model B
Model
Data from
Ref. [9]
A
B
LOO cross validated q2 value 0.570
0.564
0.974
6
0.582
0.994
8
Correlation coefficient r2
Principal components n
Contributions
0.994
3
The region to the left and below of the 5-hydroxy anchor
point were shown to have no predictive power in model B
pharmacophore (Figs. 5 and 6). In Figs. 5 and 6 all fields are
much better refined with better alignment, larger in most cases,
better shape definition (circular or spherical) and the region
Steric
Electrostatic
Lipophilic (log P)
0.352
0.419
0.228
0.502
0.477
0.052
0.270
0.664
0.085
Fig. 4. 5-HT (green) bound in the receptor site with the two anchoring points; SER 197 and ASP 100.

22
M.N. Iskander et al. / European Journal of Medicinal Chemistry 41 (2006) 16–26
Fig. 5. Pharmacophore features developed from CoMFA Model B using the complete dataset; all the 23 compounds from Table 1: modified model using electrostatic
anchor points: blue regions favor negative electrostatic charge; red regions favor positive electrostatic charge.
Fig. 6. Pharmacophore features developed from CoMFA Model B, modified model using electrostatic anchor points and the complete dataset from Table 1: yellow
regions where steric bulk is unfavorable. Regions of favored steric bulk are colored green.
beyond the distant nitrogen anchor point is no longer a disor-
dered clustering of various fields.
Model B indicated that the predictive power stems from far
lower steric contributions, 0.270 compared to model A at
0.502. The dominant defining feature was the increase of elec-
trostatic contributions for model B at 0.664 from 0.477 in mod-
el A. In both models the contributions from the LogP factor
were minimal, 0.085 in model B.
The increase in contributions from the electrostatic term,
both negative and positive, agrees well with the constructed 5-
HT4 binding site. In the binding site the residue serine 197
forms a strong hydrogen bond with the oxygen of the 5-hydro-
xyl group, the first alignment point, and the negatively charged
aspartic acid 100 interacts with the positively charged trypta-
mine nitrogen. Agonists contain a secondary N rather than a
positively charged nitrogen in the same location, a weaker
but still significant interaction occurs as the secondary nitrogen
still carries a lone pair of electrons and a slightly positive net
charge.
Pharmacophore model B (Fig. 7) shows the refined molecu-
lar alignment compared to model A alignment (Fig. 3). It posi-
tioned the aromatic ring system away from the traditional cen-
tral location of the aromatic ring in model A. The CoMFA
analysis indicated that this conformation was indeed favorable
as there were no significant steric or electrostatic problems
with accommodating the position of the tri-substituted aromatic
ring of compounds 8, 9, 10, 16, 17, 18 and 19. The obtained
data are shown in Table 2 column B.
4.3. Comparative analysis of models A and B
Comparative statistics and predicted values for 5-HT4 ago-
nists from the CoMFA analyses are shown in Table 2. Both A
and B models are more complex than the initial one [9], with
the number of components significantly increased due to a
wider range of compounds used. Of interest, despite the mod-
ified positions of the aromatic rings of some of the ligands (8,
9, 10, 16, 17, 18 and 19), their new locations are still sterically
allowed, except for those pointing outwards from the hydroxyl
group region (Fig. 6).
The refined fields of the pharmacophore model B also allow
us to see with greater confidence the groups and orientation of
groups that are critical to high 5-HT4 agonist activity, namely
the hydrogen bond acceptor 5-hydroxyl group, the positive
charge on the tryptamine nitrogen and for the first time the
specific orientation of the lipophilic side chain beyond the

M.N. Iskander et al. / European Journal of Medicinal Chemistry 41 (2006) 16–26
23
pride, cisapride and metoclopramide. Instead the negatively
charged carboxyl oxygen and the nitrogen approximately 7–
9 Å away were the key fitting points. This model placed the
aromatic ring system away from the traditional fitting points,
however the CoMFA analysis showed that there were no sig-
nificant steric or electrostatic problems with accommodating
the position of the tri-substituted aromatic ring. This finding
was consistent with the analysis of the 5-HT4 binding site
and mutagenesis investigations [22,24]. A significant region
below the 5-hydroxyl in the second model (B) is suitable for
aromatic ring fitting as found in compounds like cisapride or
zacopride where the steric bulk is unfavorable (colored yel-
low), regions of favored steric bulk are colored green
(Fig. 6). Utilizing this new refined model one can better see
the interactions involved of CoMFA contributions Table 3,
where the electrostatic interactions were the dominant defining
feature with the steric and lipophilic interactions being of less
importance.
Fig. 7. CoMFA model B, a refined pharmacophore. Reduced dataset shown for
simplicity, compounds shown are 1, 4, 7 and 5-HT.
charged nitrogen, angled upwards at about 60° out of the plane
of the tryptamine ring system. The central aromatic ring was
shown to be less significant, acting more as a spacer and scaf-
fold to keep the two electrostatic groups at the correct distance
and orientation from each other as well as steric bulk required
by the binding site.
5. Conclusions
The agonists which did not contain a central tryptamine
ring, such as the methoxybenzoate series, had their aromatic
ring off the side below the 5-hydroxyl. In the constructed bind-
ing site the location of these aromatic rings was conveniently
between two of the helices, forming further lipophilic interac-
tions without being sterically constrictive. Utilizing the refined
CoMFA model B which incorporate points from the binding
site it is expected to provide a more defined pharmacophore
for designing better 5-HT4 agonists.
The two models (A and B) shared common features; the
unfavorable steric region close to the 5-hydroxyl in yellow,
the specific electrostatic favorable region close to the charged
nitrogen in blue, the electrostatically unfavorable region be-
yond it in red and the steric favored region beyond it colored
green Figs. 2,3 and 6. The steric, hydrophobic and aromatic
rings stacking played a minor role in activity contributions.
CoMFA analysis molecular alignment of model A led to
significantly different positioning of all of the compounds that
contain the central tri-substituted aromatic ring such as zaco-
The modified CoMFA analysis (model B) agrees well with
the constructed 5-HT4 binding site. The CoMFA predictions of
this modified model have confirmed the importance of electro-
static contributions for 5-HT4 agonist activity. We have been
able to use these recommendations for a new pharmacophore
to design analogues with higher affinity and specificity. It must
be noted that the series of compounds studied is limited and
focused on developing a future robust model to discriminate
between closely related receptors of the 5-HT4 isoforms.
6. Pharmacological evaluation
Compounds were screened for 5-HT4 agonist activity using
a modification to the method described by Craig and Clarke
[14]. Briefly, this involved measuring the ability of the com-
pounds to increase the twitch responses of the guinea pig ileum
myenteric plexus preparation, stimulated with single pulses of
1 ms duration at a frequency of 0.1 Hz with approximately
Table 3
CoMFA contributions of model B
Steric
Steric
Steric
Electrostatic
Lipophilic
LogP
0.582
1.304
Electrostatic
Lipophilic
Electrostatic
LogP
Steric
Electrostatic
Steric
LogP
Steric
Lipophilic
Electrostatic
Lipophilic
Electrostatic
LogP
q2
SEP
0.561
1.336
2
0.823
0.848
46.613
0.419
1.578
3
0.786
0.958
23.269
0.440
1.510
2
0.750
1.009
29.983
0.481
1.532
4
0.889
0.709
35.975
0.654
1.186
2
0.858
0.761
60.295
0.533
1.379
2
0.803
0.895
40.866
0.326
1.700
3
0.706
1.122
15.221
n
2
r2
0.821
0.854
45.856
SEE
F
Contributions
Steric
Electrostatic
Lipophilic
LogP
0.151
0.380
0.444
0.025
0.153
0.385
0.462
–
0.270
0.644
–
0.294
0.706
–
0.898
–
–
0.243
–
0.756
–
–
–
0.452
0.548
–
0.906
–
0.094
0.085
–
0.102
Statistical properties of PLS models using various CoMFA fields and LogP. The LOO cross-validated r2 value is denoted q2. SEP is the standard error of predic-
tion. n is the number of components used in the PLS analysis. r2 is the non-cross-validated regression value. SEE is the standard error of estimation. F is the F-
statistic for the analysis and contributions give the relative contributions for each of the fields used.

24
M.N. Iskander et al. / European Journal of Medicinal Chemistry 41 (2006) 16–26
50% effective voltage. The tissues were bathed in Krebs Hen-
seilt solution at 37 °C and bubbled continuously with a mixture
of 95% O2 and 5% CO2. The modification involved adding
ketanserin (10 μM) and tropisetron (10 μM) to the bathing so-
lution to block 5-HT2A and 5-HT3 receptors, respectively. The
preparations were loaded under a constant tension of 0.5 g and
twitch responses were recorded isotonically using force–displa-
cement transducers (four-channel pen recorder, model 79D,
Grass Instruments Co., Quincy, MA, USA).
The concentration–response curves of the compounds were
constructed by exposing the strips to cumulative additions of
the compounds at 0.5–1.0 min intervals. Each preparation was
exposed to no more than six administrations to avoid desensi-
tization.
Gel 60 F254 sheets and the mobile phase was chloroform/
methanol (95:5) unless otherwise stated. Nuclear magnetic re-
sonance spectra were recorded on a Brüker AM-300 spectro-
meter at a frequency of 300 MHz. Chemical shifts (δ) are re-
ported relative to tetramethylsilane at δ ppm. Mass spectra
were recorded with a Micromass Platform spectrometer in elec-
trospray mode at cone voltages 30, 70 and 100 eV. The follow-
ing chemicals were purchased from Sigma-Aldrich Pty. Co.:
5-MeOT, 5-benzyloxytryptamine, thiophene-2-carboxalde-
hyde, pyrrole-2-carboxaldehyde, 2-thiophenecarbonyl chloride,
2-thiophenesulfonyl chloride, 2-thiopheneacetyl chloride, bor-
ane-tetrahydrofuran complex as 1.5 M solution in tetrahydro-
furan and ether.
7.2. General methods
6.1. Statistical analysis and reagents
Three main methods (a, b and c) were used for the synthesis
of the current compounds; aldehyde/amine, acid chloride/amine
condensation and the products from these reaction were reduced
with BH3 (Schemes 1 and 2). These methods have generated
compounds 7, 8, 9, 16, 17, 18, 19, 20 and 22 in good yields.
Potencies of the agonists are expressed as relative to 5-HT
(1.00) and intrinsic efficacy as Emax values (full agonists, viz 5-
HT = 1.00). These values were calculated by fitting the data
points to a sigmoidal concentration–response curve of variable
slope using non-linear regression using the program Graph Pad
Prism (4.0). The following drugs were used: 5-HT creatine sul-
fate complex (Sigma-Aldrich Pty. Co., Castle Hill, Australia);
ketanserin (Janssen-Cilag, Sydney, Australia) and tropisetron
(Sandoz Ltd., Basle, Switzerland).
7.2.1. Aldehyde and amine condensation
The appropriate aldehyde (1 mol equivalent) was added to
dry methanol (10 ml) containing 5-MeOT (1 mol) and the mix-
ture was refluxed under a nitrogen atmosphere for 3 h. The
reaction mixture was then cooled to 0 °C, and the imine inter-
mediate was reduced by the addition of NaBH4 (3 mol) slowly
and stirred for 30 min. The methanol was removed under re-
duced pressure and the organic crude was extracted with CDM,
washed with 10% citric acid (10 ml), saturated solution of
NaHCO3 (10 ml), distilled water (20 ml), dried over anhydrous
Na2SO4 and the organic solvent removed to yield the desired
compound.
7. Chemistry
7.1. General
Reactions were monitored using thin layer chromatography
(TLC) at 30 min intervals from the start to completion of reac-
tion. The stationary phase was 0.2 mm aluminum backed Silica
Scheme 1.
Scheme 2.

M.N. Iskander et al. / European Journal of Medicinal Chemistry 41 (2006) 16–26
25
7.2.2. Acid chloride and amine condensation
1H NMR (300 MHz; CDCl3): δ 3.22–3.49 (m, 4H,
NCH2CH2), 3.74 (s, 3H, OCH3), 6.68 (d, 1H, J = 8.40 Hz,
H-6), 7.02 (s, 1H, H-2), 7.11 (s, 1H, H-4), 7.21 (d, 1H,
J = 8.70 Hz, H-7), 7.47 (m, 1H, H-3′), 8.13 (d, J = 7.50 Hz,
1H, H-2′), 8.66 (s, 1H, H-4′), 8.91 (s, 1H, H-6′), and 8.93 (s,
1H, ArNH).
The appropriate chloride compound (0.6 mmol) was added
to dry CDM (10 ml) containing 5-MeOT (114 mg, 0.6) and
triethylamine (83 mg, 0.8 mmol) at room temperature. The
mixture was then refluxed under an atmosphere of nitrogen
for 3 h. The solution was then cooled and washed with 10%
citric acid saturated solution of sodium bicarbonate and dis-
tilled water. The CDM extract was dried over anhydrous
Na2SO4 and the solvent removed under reduced pressure to
give the desired product.
ESMS m/z 282 [(M + 1)/1].
7.3.1. N-[2-(5-Methoxy-1H-indol-3-yl)ethyl]-2-
thiophenemethaneamine (16)
Prepared in 82% yield from 2-thiophenecarboxaldehyde and
5-MeOT and the reduction with BH3.
7.2.3. Amide reduction using BH3
Borane (BH3) in THF (2 ml, 1 M) was added slowly to dry
THF (5 ml) containing the appropriate amide (0.30 mmol) at
0 °C under a nitrogen atmosphere. After the addition of BH3,
the reaction mixture was refluxed for 4 h. THF was removed
under reduced pressure and then NaOH (2 ml; 2.0 M) and
CDM (20 ml) were added to the remaining residue. The organ-
ic layer was then washed with distilled water (20 ml × 3) and
dried over anhydrous Na2SO4 and concentrated down to give
an oily brown product. The crude material was purified by
using preparative TLC plates (CHCl3/MeOH 95:5).
Specific preparations are explained thereafter under indivi-
dual compounds.
1H NMR (300 MHz; CDCl3): δ 1.86 (s, 1H, NH), 2.95–3.06
(m, 6H, 2 × NCH2 andCH2), 3.86 (s, 3H, OCH3), 6.85 (s, 1H,
H-2), 6.88–6.92 (m, 1H, H-6), 6.94 (d, J = 3.3 Hz 1H, H-4),
7.00 (s, 1H, H-5′), 7.05 (s, 1H, H-4′), 7.20 (s, 1H, H-3′), 7.22
(d, J = 9.0 Hz, H-7), 7.97 (s, 1H, ArNH).
ESMS m/z 287 [(M + 1)/1].
7.3.2. N-[2-(5-Methoxy-1H-indol-3-yl)ethyl]-2-
thiopheneacetamide (19)
Prepared in 60% yield from 2-thiopheneacetylchloride and
5-MeOT.
1H NMR (300 MHz; CDCl3): δ 2.89 (t, J = 6.6 Hz, 2H,
CH2), 3.52–3.59 (m, 4H, NCH2 and COCH2), 3.86 (s, 3H,
OCH3), 5.68 (s, 1H, NHCO), 6.80 (s, 1H, H-2), 6.84–6.86
(m, 1H, H-6), 6.89 (d, J = 2.4 Hz, 1H, H-4), 6.91–6.94 (m,
1H, H-4′), 7.20 (d, J = 5.1 Hz, 1H, H-3′), 7.26 (d, J = 9.3 Hz,
H-7), 7.91 (s, 1H, ArNH).
7.2.3.1. N-[2-(5-Methoxy-1H-indol-3-yl)ethyl],(N-methyl-1H-
indol-2-yl)methaneamine (7). This compound was prepared in
86% yield from 1-methylindole-2-carboxyaldehyde and 5-
MeOT and reduction with BH3.
1H NMR (300 MHz; DMSO): δ 2.81–2.89 (m, 4H,
NCH2CH2), 3.270 (s, 3H, NCH3), 3.70 (s, 3H, OCH3), 5.68
(s, 2H, HNCH2), 6.30 (s, 1H, H-3′), 6.68 (d, J = 8.7 Hz, 1H,
H-6), 6.95–6.06 (m, 5H, H-2, -4, -3′, -4′ and H-6′), 7.19 (d,
J = 8.7 Hz, 1H, H-7), 7.36 (d, J = 7.8 Hz, 1H, H-5′), 7.42 (d,
J = 8.0 Hz, 1H, H-7′), 10.53 (s, 1H, ArNH).
ESMS m/z 315 [(M + 1)/1].
7.3.3. N-[2-(5-benzyloxy-1H-3-yl)ethyl)]-2-
thiophenecarboxamide (20)
ESMS m/z 334 [(M + 1)/1].
Prepared in 65% yield from 2-thiophenecarbonylchloride
benzyloxytryptamine.
1H NMR (300 MHz; CDCl3): δ 2.89 (t, J = 6.9 Hz, 2H,
CH2), 3.52–3.59 (m, 2H, NCH2), 3.85 (s, 3H, OCH3), 5.70
(s, 1H, NHCO), 6.80–6.88 (m, indolyl and phenyl-H), 6.91–
6.93 (m, 1H, H-4′), 7.00 (s, 1H, H-5′), 7.20 (d, J = 5.1 Hz,
1H, H-3′), 7.22–7.26 (m, indolyl and phenyl-H), 7.97 (s, 1H,
ArNH).
7.2.3.2. N-[2-(5-Methoxy-1H-indol-3-yl)ethyl],(1,3-benzodiox-
ol-5-yl)methaneamine (8). This compound was prepared from
1,3-benzodioxole-5-carboxaldehyde and 5-MeOT and reduc-
1
tion with BH3. in 52% yield, H NMR (300 MHz; DMSO): δ
2.86–2.91 (m, 2H, NCH2), 3.45–3.52 (m, 2H, NCH2CH2), 3.71
(s, 3H, OCH3), 6.07 (s, 2H, (CH2O)2–CH2), 6.70 (d,
J = 6.6 Hz, 1H, H-6), 6.96 (d, J = 8.7 Hz, 1H, H-7), 7.03 (s,
1H, H-4′), 7.11 (s, 1H, H-2), 7.21 (d, J = 8.8 Hz, 1H, H-6′),
7.37 (s, 1H, H-4), 7.43 (d, J = 6.6 Hz, 1H, H-1′), 8.39 (s, 1H,
ArNH).
ESMS m/z 377 [(M + 1)/1].
7.3.4. N-[2-(5-Methoxy-1H-indol-3-yl)ethyl]-2-
thiopheneethaneamine (22)
ESMS m/z 313 [(M + 1)/1].
Prepared in 56% yield by the BH3 reduction of compound
19.
7.3. Preparation of N-[2-(methoxy-1H-indol-3-yl)ethyl]-N-(3-
pyridinylmethyl)-1-ethanamine (9)
1H NMR (300 MHz; CDCl3): δ 3.02–3.06 (m, 4H, 2 ×
CH2), 3.71–3.78 (m, 7H, 2 × NCH2 and OCH3), 6.28 (s, 3H,
NH), 6.85 (d, J = 8.7 Hz 1H, H-5′), 6.98–7.01 (m, 2H, H-6 and
H-4), 7.05 (s, 1H, H-2), 7.24 (d, J = 8.7 Hz, 1H, H-7), 7.34–
7.42 (m, 2H, H-3′ and H-4′), 8.32 (s, 1H, ArNH).
Prepared from nicotinoyl chloride hydrochloride and 5-
methoxtryptamine and then the reduction of the product with
BH3 gave compound 9 as a light brown semi-solid product in
90% yield.
ESMS m/z 301 [(M + 1)/1].

26
M.N. Iskander et al. / European Journal of Medicinal Chemistry 41 (2006) 16–26
[12] T. Suzuki, N. Imanishi, H. Itahana, S. Watanuki, K. Miyata, M. Ohta, H.
Acknowledgements
Nakahara, Y. Yamagiwa, T. Mase, Chem. Pharm. Bull. (Tokyo) 46
(1998) 1116–1124.
This work was funded by project grant from the National
Health and Medical Research Council of Australia (NHMRC)
and a scholarship to Mr. Ayad from the Faculty of Pharmacy,
Monash University.
[13] C.A. Rizzi, T. Coccini, L. Onori, L. Manzo, M. Tonini, J. Pharmacol.
Exp. Ther. 261 (1992) 412–419.
[14] D.A. Craig, D.E. Clarke, J. Pharmacol. Exp. Ther. 252 (1990) 1378–
1386.
[15] R.D. Cramer III, D.E. Patterson, J.D. Bunce, J. Am. Chem. Soc. 110
References
(1988) 5959–5967.
[16] R.S. Bohacek, C. McMartin, J. Med. Chem. 35 (1992) 1671–1684.
[1] IUPHAR receptor data base, 2002. http://iuphar-db.org/iuphar-rd/index.
html.
[2] P.G. McLean, I.M. Coupar, P. Molenaar, Br. J. Pharmacol. 115 (1995)
47–56.
[3] P.G. McLean, I.M. Coupar, Br. J. Pharmacol. 118 (1996) 1058–1064.
[4] R.A. Borman, D.E. Burleigh, Eur. J. Pharmacol. 309 (1996) 271–274.
[5] N.H. Prins, L.M.A. Akkermans, R.A. Lefebvre, J.A.J. Schuurkes, Br. J.
Pharmacol. 131 (2000) 927–932.
[6] S.A. Muller-Lissner, I. Fumagalli, K.D. Bardhan, Alimentary Pharmacol
& Ther 15 (2001) 1655–1666.
[7] M.J. Callahan, J. Clin. Gastroenterol. 35 (2002) S58–S67.
[8] D. Hoyer, D.E. Clarke, J.R. Fozard, P.R. Hartig, G.R. Martin, E.
J. Mylecharane, P.R. Saxena, P.P.A. Humphrey, Pharmacol. Rev. 46
(1994) 157–204.
[9] M.N. Iskander, I.M. Coupar, D.A. Winkler, J. Chem. Soc. Perkin Trans.
2 (1999) 153–158.
[10] K.H. Buchheit, R. Gamse, R. Giger, D. Hoyer, F. Klein, E. Kloppner, H.
J. Pfannkuche, H. Mattes, J. Med. Chem. 38 (1995) 2331–2338.
[11] D. Yang, J.L. Soulier, S. Sicsic, M. Mathe-Allainmat, B. Bremont, T.
Croci, R. Cardamone, G. Aureggi, M. Langlois, J. Med. Chem. 40
(1997) 608–621.
[17] SYBYL 6.9 Tripos Inc. 1699 South Hanley Road St. Louis, MO 63144-
2319 www.tripos.com.
[18] T. Halgren, J. Am. Chem. Soc. 112 (1990) 4710–4723.
[19] T. Halgren, J. Comp. Chem. 20 (1999) 720.
[20] S. Wold, C. Albano, W.J. Dunn III, U. Edlund, K. Esbensen, P. Geladi,
S. Hellberg, E. Johansson, W. Lindberg, M. Sjostrom, “Multivariate data
analysis in chemistry”, in: CHEMOMETRICS: Mathematics and Statis-
tics in Chemistry, B. Kowalski, Reidel, Dordrecht, Netherlands, 1984.
[21] G.R. Marshall, Binding-site modelling of unknown receptors. In 3D
QSAR in Drug Design. Theory, Methods and Applications, in: H. Kubi-
nyi (Ed.), Escom, Leiden, 1993, pp. 80–116.
[22] M.L. Lopez-Rodriguez, B. Murcia, M. Benhamu, M. Olivella, M. Cam-
pillo, L. Pardo, J. Comput. Aided. Mol. Des 15 (11) (2001) 1025–1033
(Nov).
[23] G. Wishart, D.H. Bremner, K.R. Sturrock, Receptor Channels 6 (4)
(1999) 317–323.
[24] M.L. Lopez-Rodriguez, B. Murcia, M. Benhamu, M. Olivella, M. Cam-
pillo, L. Pardo, J. Comput. Aided. Mol. Des. 15 (11) (2001) 1025–1033
(Nov.).