Pharmacophore-based search, synthesis, and biological evaluation of anthranilic amides as novel blockers of the Kv1.5 channel

Article abstract of DOI:10.1016/j.bmcl.2004.03.057

The search for novel, potent Kv1.5 blockers based on an anthranilic amide scaffold employing a pharmacophore-based virtual screening approach is described. The synthesis and structure-activity relationships (SAR) with respect to inhibition of the Kv1.5 ch

Full text of DOI:10.1016/j.bmcl.2004.03.057

Bioorganic & Medicinal Chemistry Letters 14 (2004) 2823–2827
Pharmacophore-based search, synthesis, and biological evaluation
of anthranilic amides as novel blockers of the Kv1.5 channel
*
€
Stefan Peukert, Joachim Brendel, Bernard Pirard, Carsten Strubing,
€
Heinz-Werner Kleemann, Thomas Bohme and Horst Hemmerle
Aventis Pharma Deutschland GmbH, Medicinal Chemistry and DG Cardiovascular, D-65926 Frankfurt, Germany
Received 11 December 2003; revised 8 March 2004; accepted 19 March 2004
Abstract—The search for novel, potent Kv1.5 blockers based on an anthranilic amide scaffold employing a pharmacophore-based
virtual screening approach is described. The synthesis and structure–activity relationships (SAR) with respect to inhibition of the
Kv1.5 channel are discussed. The most potent compounds display sub-micromolar inhibition of Kv1.5 and no significant effect on
the HERG channel. In addition, good oral bioavailability is demonstrated for compound 3i in rats.
Ó 2004 Elsevier Ltd. All rights reserved.
1. Introduction
found only in human atrial cells and not in the ventri-
cles.6 Therefore it is regarded as a promising target for
the development of new atrial selective antiarrhythmics.
Since then, many companies have filed patents claiming
compounds as blockers of the Kv1.5 channel.7 In
addition, Aventis, Merck, and Nissan have published
animal trial data supporting the use of this new
approach but no clinical trials have yet been carried out
to our knowledge.7
Currently available drug treatment for atrial fibrillation
(AF) is less than satisfactory and fraught with severe
difficulties. The presently used antiarrhythmics of classes
I (sodium channel blockers) and III (potassium channel
blockers) can terminate AF and reduce its recurrence,
but may increase mortality due to a variety of adverse
effects, including the risk of potentially lethal ventricular
proarrhythmia (CAST-trial,1 SWORD-trial2). The gold
standard of anti-arrhythmic therapy, amiodarone, a
very unselective drug, is hampered by many noncardiac
side effects.3 Therefore, there is an unmet medical need
for the development of safer and more efficient drugs for
the treatment of atrial arrhythmias.4
In the present study, we report the discovery and sub-
sequent structure–activity investigation of novel
anthranilic amides with favorable pharmacokinetic
properties as blockers of the Kv1.5 channel.
Most of the available class III antiarrhythmics are
blockers of IKr (e.g., dofetilide, ibutilide). The new
antiarrhythmic agent azimilide is a combined blocker of
IKr and IKs.5 However, since IKr and IKs are both
present in the atrium and ventricle, undesired ventricu-
lar effects may be observed for these drugs, which limit
their use for the treatment of atrial arrhythmias. In
contrast to IKr and IKs, the ultra-rapid delayed rectifier
IKur, which is based on the Kv1.5 channel, has been
2. Pharmacophore-based lead identification
So far, our efforts to identify Kv1.5 blockers have
focused on two chemical series, namely the bisaryls and
the meta-substituted benzene sulfonamides (Fig. 1).8
Seven Kv1.5 blockers from the two series were selected
Cl
Cl
O
H
N
N
H
N
O
S
O
O
O
O
N
H
NH
Keywords: Kv1.5 blockers; Anthranilic amides; Atrial fibrillation.
* Corresponding author. Tel.: +49-69-305-20745; fax: +49-69-331399;
e-mail: stefan.peukert@aventis.com
(IC50 = 0.2 µM)
(IC50 = 2.8 µM)
1
2
Figure 1. Representatives of the two compound series used for Kv1.5
pharmacophore modeling.
Present address: Lilly Research Laboratories, Lilly Corporate
Center, Indianapolis, IN 46285, USA.
0960-894X/$ - see front matter Ó 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2004.03.057

2824
S. Peukert et al. / Bioorg. Med. Chem. Lett. 14 (2004) 2823–2827
Figure 3. Chemical structure of the anthranilic amide lead 3a.
Figure 2. (a) Pharmacophore model, (b) matching of compound 1 and
2 with pharmacophore model.
to identify a pharmacophore model.9 The derived
pharmacophore consists of three hydrophobic centers in
a triangular arrangement as depicted in Figure 2a. As
illustrated in Figure 2b, one phenyl of the bisaryl core of
compound 1 and the central phenyl moiety of com-
pound 2 matches the middle hydrophobic center, while
within each structure class, both ends of the side chains
correspond to the remaining hydrophobic centers of the
model. The model is consistent with the SARpreviously
observed for bisaryl Kv1.5 blockers.8a Furthermore,
the pharmacophore model served as a query to screen an
in-house data bank of 423 Kv1.5 blockers, which
resulted in the retrieval of 58% of the known Kv1.5
blockers. This performance was deemed satisfactory for
lead identification. Therefore, we ran a 3D search within
the Aventis compound collection using this model as a
query. This strategy resulted in 4234 virtual hits, 1975
compounds of them passed the sequence of filters
described in Ref. 10. These 1975 compounds were sub-
mitted to a hierarchical cluster analysis resulting in 27
clusters. Representatives from 18 clusters were actually
available for in vitro screening in Xenopus oocytes,
which yielded one active compound with an IC50 of
5.6 lMà (Fig. 3), belonging to a new class of Kv1.5
blockers. Although compound 3a is less potent than the
best representatives of the previously investigated
chemical classes of Kv1.5 blockers, it exhibits a favor-
able pharmacokinetic profile. Therefore, a main goal
was to improve in vitro potency of the anthranilic
amides while keeping the pharmacokinetic profile of 3a.
Scheme 1. Reagents and conditions: (a) R3SO2Cl, Na2CO3, water,
60 °C; (b) SOCl2, 60 °C; (c) HNR4R5, Et3N, DCM, 20 °C; (d) H2, Pd/
C, THF/MeOH, 1 bar, 20 °C; (e) HNR4R5, DMF, 60 °C; (f) R3SO2Cl,
pyridine, DCM, 20 °C.
and subsequent reaction with amines (route A). Requi-
site secondary amines, which were not commercially
available, were obtained by reductive amination with
NaCNBH4 in methanol.11 Whereas the route
A
described above, is ideal to probe the influence of the
amine substituents R4, R5 as it is introduced in the last
step, this is not true for R3. In order to rapidly evaluate
substituent R3 we relied on route B where we used nitro
benzoic acids 6 or anhydrides 8 and permutated the
sulfonyl group R3 in the last step. Anhydrides 8 were
opened in DMF with amines to yield anilines 9. These
intermediates could be also prepared by palladium-cat-
alyzed hydrogenation of the nitro compounds 7, which
for their part were formed by coupling amines to the
nitro benzoic acids 6 after activation with thionyl chlo-
ride. Finally, conversion of the anilines 9 with sulfonyl
chlorides in the presence of pyridine afforded 3 (route B).
3. Chemistry
To systematically evaluate the structure–activity rela-
tionships in the boxed regions shown in Figure 3, we
employed two different synthetic routes (Scheme 1).
Compounds 3 were prepared from anthranilic acids 4 by
conversion with sulfonyl chlorides, followed by activa-
tion of the acid functionality of 5 to the acid chloride
4. Biological results and discussion
à
All IC50 values in this manuscript are determined in Xenopus oocytes
expressing human Kv1.5 channels. Electrophysiology is described in
detail in Ref. 8a.
As a first step in our investigation of the structural
requirements for Kv1.5 inhibition we examined the role

S. Peukert et al. / Bioorg. Med. Chem. Lett. 14 (2004) 2823–2827
2825
Table 2. Inhibitory activity of anthranilic amides against the Kv1.5
channel
of the amide substituents (Table 1). In accordance with
the pharmacophore model at least one hydro-phobic
contact is needed. The distance of this hydrophobic
moiety does matter for potency (3b vs 3c) with an
optimum for the phenyl group at the benzylic position.
Further increase of activity is obtained by small a-sub-
stituents X (3d–g) with a preference for the (S)-stereo-
chemistry at this position. Next to these secondary
amides, tertiary amides were potent, which carried het-
erocyclic substituents (3i–j) or a cycloalkyl substituent
(e.g., 3k) in addition to the benzyl group. Smaller sub-
stituents such as the methyl in compounds 3l or an end
group with an additional linker as in 3m result in a
decrease of potency. The benzyl group can be readily
replaced by substituents containing heterocycles (3n,o)
without significant loss of activity, which can be used for
fine tuning physicochemical and ADME properties.
Compound
R3
IC50 [lM]
3p
p-Methoxyphenyl
0.6
3q
3r
0.6
3.5
1-Naphtyl
3-Pyridyl
3s
3t
9.0
36%@10lMa
We next turned our attention to the influence of the
sulfonyl substituent R3 on the activity (Table 2). In
addition to our p-toluyl probe from Table 1, the
p-methoxyphenyl (3p) and 8-quinolinyl (3q) substituents
turned out to be excellent moieties. Thereby, the nitro-
gen and its position proved critical as all other related
substituents R3 (3r–t) gave rise to compounds with
diminished activity. However, the aryl group can be
replaced successfully by an aliphatic group, isopentyl
(3u) being the among the best. Reduced steric bulk at
this position resulted in lower potency as seen by com-
parison of compounds 3u–w.
3u
1.7
3v
3w
Propyl
Ethyl
3.1
10
a Percent inhibition at 10 lM concentration.
incorporated into this site a further loss in activity is
observed (3af). On the other hand, R1 plays together
with substituent Z an electronic role by determining the
acidity of the sulfonamide functionality. Remarkably,
the inhibitory activity showed a correlation with the
pKa values of the compounds. A pKa difference of 1.3
(3x vs 3ab) results in an eightfold drop in potency.
Replacement of the acidic hydrogen by a methyl group
(3ag) resulted in further reduction of activity. These
results suggest that the hydrogen might be involved in
Finally, the findings, which were obtained with various
R1 substituents indicate that this residue influences the
inhibitory activity by steric and electronic factors:
Increasing its size from methoxy (3x) to acetyl (3ac) and
larger ether residues (3ad,ae) has a clearly detrimental
effect on the potency (Table 3). If a basic nitrogen is
Table 1. Inhibitory activity of anthranilic amides against the Kv1.5 channel
Compound
R4
X
Y
IC50 [lM]
3b
3c
3d
3e
3f
H
H
CH2Ph
Ph
6.7
4.1
3.2
0.7
2.7
1.6
0.5
0.7
0.7
1.9
5.0
8.5
2.2
1.2
H
H
rac-Me
H
Ph
Ph
H
S-Et
R-Et
H
Ph
Ph
3g
3h
3i
H
CH2Ph
rac-Cyclopropyl
H
H
H
H
H
H
H
H
3-Pyrazolyl
3-Pyridyl
CH2Ph
CH2Ph
3j
2-(N-Me)-imidazolyl
Cyclopropyl
Methyl
3k
3l
CH2Ph
CH2Ph
3m
3n
3o
CH2Ph
CH2(3-Me-pyridyl)
CH2(2-furanyl)
CH2CH2(1-imidazolyl)
3-Pyridyl
2-(N-Me)-imidazolyl

2826
S. Peukert et al. / Bioorg. Med. Chem. Lett. 14 (2004) 2823–2827
Table 3. Inhibitory activity of anthranilic amides against the Kv1.5 channel
Compound
R1
Z
pKa
IC50 [lM]
3x
3y
OMe
H
Me
Me
H
7.5
7.2
6.9
6.5
6.2
––
0.5
0.7
1.3
3.8
4.0
3.8
3z
H
3aa
3ab
3ac
F
Cl
H
Acetyl
CF3
Me
3ad
Me
––
46%@10lMa
3ae
Me
––
41%@10lMa
3af
Me
––
15%@10lMa
a Percent inhibition at 10 lM concentration.
3. Tomaselli, G. Heart Drug 2001, 1, 183.
an intramolecular hydrogen bond with the carbonyl
oxygen, which stabilizes a favorable active conformation.
4. Bril, A. Curr. Opin. Pharmacol. 2002, 2, 154.
5. Nishida, A.; Uemura, H.; Ogura, T.; Furusawa,
H.; Yabana, H.; Nakaya, H. Jpn. J. Pharmacol. 2000,
82, 1.
6. (a) Li, G. R.; Feng, J.; Yue, L.; Carrier, M.; Nattel, S.
Circ. Res. 1996, 78, 689; (b) Feng, J.; Wible, B.; Wang, Z.;
Nattel, S. Circ. Res. 1997, 80, 572; (c) Amos, G. J.;
Wettwer, E.; Metzger, F.; Li, Q.; Himmel, H. M.; Ravens,
U. J. Physiol. (Lond.) 1996, 491, 31.
To be considered as safe antiarrhythmics selectivity
toward the HERG channel is mandatory. As could be
shown with lead 3a and compounds 3i, 3n, 3q, and 3v,
there was no significant effect on the IKr current: 6%,
11%, 14%, 2%, and 1% inhibition at 10 lM concentra-
tion, respectively. Finally, for compound 3i oral bio-
availability in rats of 43% was found.
7. (a) Brendel, J.; Peukert, S. Expert. Opin. Ther. Patents
2002, 12, 1589; (b) Brendel, J.; Peukert, S. Curr. Med.
Chem.––CV & H Agents 2003, 1, 273.
In conclusion, we have discovered and optimized a new
class of potent Kv1.5 blockers based on a simple
anthranilic acid scaffold. The compounds show no sig-
nificant effect on the IKr current and are orally bio-
available, and are therefore promising drug substances
for a new and safe treatment of atrial fibrillation.
€
8. (a) Peukert, S.; Brendel, J.; Pirard, B.; Bruggemann, A.;
Below, P.; Kleemann, H.-W.; Hemmerle, H.; Schmidt, W.
J. Med. Chem. 2003, 46, 486; (b) Aventis Pharma
Deutschland GmbH, WO 0100573, 2001.
9. The conformational space of the seven Kv1.5 blockers
selected as training set for pharmacophore elucidation was
explored, using
a Monte Carlo Multiple Minimum
(MCMM) search protocol. This search protocol involved
the generation of maximum 5000 conformers by random
modifications of minimum 2 and maximum 7 torsion
angles at a time. Each generated conformer was sub-
mitted to 250 steps of conjugate gradient minimization.
Then, we discarded strained (more than 50 kJ molÀ1 above
the minimum) as well as duplicate conformations (distance
Acknowledgements
We wish to thank H. Plankenhorn, M. Schnierer, G.
Erhard, T. Spors, S. Peukert, and R. Weck for their
technical assistance in the synthesis of the compounds
€
and also A. Hertler and S. Muller for their assistance in
the pharmacological experiments.
ꢀ
cut off of 0.25 A between equivalent heavy atoms).
Conformers passing these filters were minimized until
convergence (convergence criterion 0.05 kJ AÀ1 molÀ1).
ꢀ
The energy was evaluated using the MacroModel 6.5
(www.schroedinger.com) implementation of the OPLS
All–Atom force field. All the calculations were carried
out in water, as modeled by the generalized Born/solvent
accessible surface solvation model. The next step was to
select representative sets of conformers, using hierarchical
clustering on distances between all the potential pharma-
cophoric centers. Within each cluster, we chose the
minimum energy conformation as input for the automated
References and notes
1. Pratt, C. M.; Moye, L. A. Am. J. Cardiol. 1990, 65, 20B.
2. Waldo, A. L.; Camm, A. J.; Deruyther, H. Lancet 1996,
348, 7.

S. Peukert et al. / Bioorg. Med. Chem. Lett. 14 (2004) 2823–2827
2827
pharmacophore elucidation tool, DISCO (www.tri-
pos.com). After visual inspection of the DISCO solutions,
one of them was selected as query for 3D database
searches with UNITY (www.tripos.com).
activity; (b) known Kv1.5 blockers; (c) molecular weight
outside the 200–500 range, hydrophilic solvent-accessible
surface area greater than 120 Az , calculated logPoctanol/
water outside the À2.0–5.0 range. QikProp 1.5 (www.sch-
roedinger.com) was used to compute the physical proper-
ties.
2
ꢀ
10. Compounds with the following features were removed
from the 3D search hit list: (a) compounds with reactive
moieties or groups known to be detrimental for Kv1.5
11. Lane, C. F. Synthesis 1975, 135.