3,8-Diazabicyclo-[3.2.1]-octane derivatives as analogues of ambasilide, a Class III antiarrhythmic agent

Article abstract of DOI:10.1016/S0223-5234(01)01246-6

Ambasilide, a representative of Class III antiarrhythmics, was reported to prolong the cardiac action potential duration in the dog, with little or no effect on Ca and Na currents. We synthesised a series of ambasilide analogues, having the 3,8-diazabicyclo-[3.2.1]-octane moiety instead of the 3,7-diazabicyclo-[3.3.1]-nonane present in ambasilide. The compounds were tested both in vitro extracellular electrophysiological assays and by the conventional microelectrode technique. Most of them lengthened the effective refractory period (ERP) with no change or slight increase on the impulse conduction time (ICT). Similarly some of the tested compounds lengthened the action potential duration (APD), a typical Class III feature, without exerting any significant effect on the maximal rate of depolarization, therefore apparently lacking Class I antiarrythmic activity.

Full text of DOI:10.1016/S0223-5234(01)01246-6

Eur. J. Med. Chem. 36 (2001) 495–506
´
© 2001 Editions scientifiques et me´dicales Elsevier SAS. All rights reserved
PII: S0223-5234(01)01246-6/FLA
Original article
3,8-Diazabicyclo-[3.2.1]-octane derivatives as analogues of ambasilide, a Class
III antiarrhythmic agent
Stefania Villaa, Daniela Barloccoa,*, Giorgio Cignarellaa, Gyula Julius Pappb, Bea´ta Bala´tib,
Ja´nos Taka´csb, Andra´s Varro´b, Andra´s Borosyc, Katalin Keseruˆc, Pe´ter Ma´tyusd
aIstituto di Chimica Farmaceutica, Universita´ di Milano, Viale Abruzzi 42, 20131 Milan, Italy
bDepartment of Pharmacology and Pharmacotherapy, University Szeged, Szeged H-6701, Hungary
cInstitute for Drug Research Ltd., Budapest H-1045, Hungary
dInstitute of Organic Chemistry, Semmelweis University, Budapest H-1092, Hungary
Received 17 July 2000; revised 13 March 2001; accepted 8 May 2001
Abstract – Ambasilide, a representative of Class III antiarrhythmics, was reported to prolong the cardiac action potential duration
in the dog, with little or no effect on Ca and Na currents. We synthesised a series of ambasilide analogues, having the
3,8-diazabicyclo-[3.2.1]-octane moiety instead of the 3,7-diazabicyclo-[3.3.1]-nonane present in ambasilide. The compounds were
tested both in vitro extracellular electrophysiological assays and by the conventional microelectrode technique. Most of them
lengthened the effective refractory period (ERP) with no change or slight increase on the impulse conduction time (ICT). Similarly
some of the tested compounds lengthened the action potential duration (APD), a typical Class III feature, without exerting any
´
significant effect on the maximal rate of depolarization, therefore apparently lacking Class I antiarrythmic activity. © 2001 Editions
scientifiques et me´dicales Elsevier SAS
antiarrhythmic activity / ambasilide / diazabicyclo-[3.2.1]-octane
1. Introduction
talol, E-4031 and dofetilide inhibit effectively, the
rapid component of the delayed rectifier outward
potassium current (IKr), thereby they prolong the ac-
tion potential duration (APD) [2]. The APD-length-
ening effect is more pronounced at slow heart rate
than during tachycardia. This reverse frequency-de-
pendent effect could be dangerous during bradycar-
dia. A recent clinical trial with the selective IKr
The results of the cardiac arrhythmia suppression
trial (CAST) unexpectedly showed that the sodium
channel blocker (Class I) antiarrhythmic drugs
flecainide and encainide did not decrease but actually
increased sudden cardiac death in postinfarction pa-
tients [1]. As a consequence, development of almost
all Class I antiarrhythmics was halted and, parallel
with this, interest was shifted toward Class III an-
tiarrhythmic agents. Compounds belonging to the
blocker
D
-sotalol (survival with oral
D
-stall,
SWORD) [3] proved that this agent also increased the
mortality in patients suffering from myocardial in-
farction. This negative result has been explained by
proarrhythmic potential of the drug as a possible
consequence of its reverse frequency dependent repo-
larization lengthening effect. Recently it was also re-
ported that ambasilide (1) (see figure 1), a new Class
III antiarrhythmic drug, produced less reverse fre-
quency dependent APD prolongation [4] in compari-
son to the effect of other Class III antiarrhythmic
agents [5].
latter class, such as
D
-sotalol, E-4031 and dofetilide,
lengthen repolarization and the effective refractory
period (ERP) with no influence on the sodium current
and the impulse conduction. The currently available
selective Class III antiarrhythmic agents such as
D
-so-
* Correspondence and reprints
E-mail address: daniela.barlocco@unimi.it (D. Barlocco).

496
S. Villa et al. / European Journal of Medicinal Chemistry 36 (2001) 495–506
with side chains previously investigated in Class III
antiarrhythmics [6–8] were synthesised. Finally, the
intermediate nitro derivative 2d and its open analogue
4 were also considered.
2. Chemistry
Figure 1. Ambasilide.
As shown in figure 2, the 3-benzyl-3,8-diazabicyclo-
[3.2.1]-octane [9] was condensed either with the ap-
propriate acyl chloride to give 2a,d or with the
Therefore, the goal of the present work was to
design and synthesise new ambasilide analogues with
a frequency independent APD lengthening effect;
such compounds may be expected to exert more ad-
vantageous antiarrhythmic activity. Accordingly,
compounds 2a and 3 were prepared, in which the
diazabicyclo-[3.3.1]-nonane nucleus of the model was
substituted by the lower homologue diazabicyclo-
[3.2.1]-octane. In addition, several derivatives (2b, c,
e–g) which combine the diazabicyclooctane moiety
required
2-(4-substituted-phenoxy)-1-chloroethane
(5a,b) [10, 11] to provide 2e,f. Treatment of the amino
derivative 2a with methanesulfonyl chloride in pyri-
dine at 0 °C gave 2b. However, if the reaction was
carried out in CH2Cl2 at reflux, 2c was obtained. The
reverted isomer 3-(4%-aminobenzoyl)-8-benzyl-3,8-di-
azabicyclo-[3.2.1]-octane (3) was synthesised from 8-
Boc-3,8-diazabicyclo-[3.2.1]-octane [12], that was
condensed with p-amino-benzoyl-chloride in CH2Cl2
Figure 2. Synthetic route to compounds 2a–g.

S. Villa et al. / European Journal of Medicinal Chemistry 36 (2001) 495–506
497
Figure 3. Synthetic route to compound 3.
in the presence of triethylamine (TEA). Deprotection
of the amino group by an ether solution of hydro-
chloride acid and subsequent condensation with ben-
zyl chloride gave 3 (see figure 3). Finally, (figure 4)
compound 4 was prepared from the 2,6-dimethyl
piperazine by condensing first with benzyl chloride in
toluene at reflux overnight and subsequently with
p-nitrobenzoyl chloride (see table I and Section 6 for
data).
impulse conduction time (ICT) and effective refractory
(ERP) period in isolated dog cardiac right ventricular
trabecular muscles. As table II shows, all the tested
compounds with the exception of 2g, lengthened the
ERP with either only slightly or not changing the ICT.
3. Pharmacology
The compounds were tested both by in vitro extra-
cellular electrophysiological assays and by the con-
ventional microelectrode technique (see Section 6 for
details).
4. Results and discussion
4.1. Biological data
The compounds were first tested by in vitro
extracellular electrophysiological methods measuring the
Figure 4. Synthetic route to compound 4.

Table I. Physicochemical properties of compounds 2a–g.
Comp.
X
R
%Yield Mp/°Ca Formulab
1H-NMR (DMSO), l (ppm)c
2a
CO
NH2
37
186–189 C20H23N3O
1.9–2.4 (m, 4H H-6,7); 3.3–3.7 (m, 4H H-2,4); 4.2–4.3 (m, 1H H-1);
4.4 (s, 2H CH2ꢀPh); 4.69 (bs., 2H NH2 exch. with D2O); 4.9–5.0 (m,
1H H-5); 6.6 (d, 2H H-3%,5%); 6.7 (d, 2H H-2%,6%); 7.5–7.7 (m, 5H
HCl
’
HAr ꢀPh).
2b
CO
NHSO2CH3 35
N(SO2CH3)2 50
157–159 C21H25N3O3S 1.9–2.4 (m, 4H H-6,7); 3.2 (s, 3H SO2CH3); 3.3–3.7 (m, 4H H-2,4);
HCl
4.2–4.3 (m, 1H H-1); 4.4 (s, 2H CH2ꢀPh); 4.3–4.6 (m, 1H H-5);
’
4.9–5.1 (m, 1H exch. with D2O NH); 7.5–7.7 (m, 7H HAr ꢀPh H-3%,5%);
7.4 (d, 2H H-2%,6%).
2c
2d
2e
CO
CO
210–212 C22H27N3O5S2 1.9–2.4 (m, 4H H-6,7); 3.3–3.5 (m, 4H H-2,4); 3.6 (s, 6H (SO2CH3)2);
HCl
4.3–4.5 (m, 1H H-1); 4.4 (s, 2H CH2ꢀPh); 4.9–5.1 (m, 1H H-5);
’
7.5–7.6 (m, 5H HAr ꢀPh); 7.6–7.8 (m, 4H H-2%,3%,5%,6%).
NO2
45
35
190–192 C20H21N3O3
1.9–2.4 (m, 4H H-6,7); 3.2–3.8 (m, 4H H-2,4); 4.3–4.4 (m, 1H H-1);
’
HCl
4.5 (s, 2H CH2ꢀPh); 4.9–5.1 (m, 1H H-5); 7.5–7.7 (m, 5H HAr ꢀPh);
7.7 (d, 2H H-2%,6%); 8.4 (d, 2H H-3%,5%).
OCH2CH2 NH2
180–181 C21H27N3O
1.8–2.0 (m, 4H H-6,7); 2.4 (m, 2H CH2-N ); 2.5–2.7 (m, 2H H-2,4);
2.7–2.9 (m, 2H H-2,4); 3.2-3.3 (m, 2H CH2-O); 3.3 (s all., 2H exch.
with D2O NH2); 3.5 (s, 2H CH2ꢀPh); 4.0–4.1 (m, 2H H-1,5); 6.6 (d,
HCl
’
2H H-3%,5%); 6.7 (d, 2H H-2%,6%); 7.5–7.7 (m, 5H HAr ꢀPh).
2f
OCH2CH2 NHCOCH3 12
OCH2CH2 NHSO2CH3 28
95–97c
C23H29N3O2
1.8–2.0 (m, 4H H-6,7); 2.1 (s, 3H COCH3); 2.3 (d, 2H CH2-N);
2.5–2.7 (m, 2H H-2,4); 2.7–2.9 (m, 2H H-2,4); 3.1–3.3 (m, 2H
CH2-O); 3.5 (s, 2H CH2ꢀPh); 4.0–4.2 (m, 2H H-1,5); 6.8 (d, 2H
’
H-2%,6%); 7.1–7.3 (m, 5H HAr ꢀPh); 7.4 (d, 2H H-3%,5%); 8.1–8.4 (m, 1H
exch. with D2O)d.
2g
198–199 C22H29N3O3S 1.8–2.0 (m, 4H H-6,7); 2.3–2.7 (m, 4H H-2,4); 2.7–2.9 (m, 2H
HCl
CH2-N); 2.9 (s, 3H SO2CH3); 3.2–3.4 (m, 2H CH2-O); 3.5 (s , 2H
CH2ꢀPh); 4.0–4.2 (m, 2H H-1,5); 5.8–6.4 (m, 1H exch. with D2O
NH); 6.8 (d, 2H H-3%,5%); 7.2 (d, 2H H-2%,6%); 7.2–7.4 (m, 5H
’
HAr ꢀPh);
a As the hydrochloride.
b Analyses were within 9 0.4% of the theoretical values.
c As the free base.
d CDCl3 as the solvent.

S. Villa et al. / European Journal of Medicinal Chemistry 36 (2001) 495–506
499
Table II. Effects of 2a–g, 3, 4 and reference compounds on the
impulse conduction time and effective refractory period in
isolated dog right ventricular trabecular musclea.
were studied in dog right ventricular papillary muscles
by the conventional microelectrode technique. The re-
sults are summarised in table IV. Moreover, the most
significant results are illustrated in the figure 5. Com-
pounds 2b, 2d and 4 but not 2a, 2e, 2f, 2g lengthened
the action potential duration (APD) without apparently
influencing the action potential amplitude (APA) and
maximal rate of depolarisation (Vmax). The repolarisa-
tion lengthening — so called Class III antiarrhythmic
effect — observed with most of the available Class III
antiarrhythmic drugs strongly depends on the heart rate
or stimulation frequency. The slower the rate the more
pronounced APD lengthening was observed, which phe-
nomenon was called reverse use or rate dependency [14].
This is considered as a disadvantageous feature since the
excessive APD prolongation at slow rate may induce
early after depolarisation (EAD) and consequently tor-
sade de pointes tachycardia. In addition, at high rate,
i.e. at tachycardia, where possible APD lengthening
would be especially important to prevent re-entry type
tachycardia, the less pronounced APD prolongation
may diminish antiarrhythmic effectiveness. Therefore we
b
Compounds (10 mM)
n
ICT c (ms)
ERP d (ms)
Control
2a
Control
2b
Control
2c
Control
2d
Control
2e
Control
2f
Control
2g
Control
3
8
8
8
8
15.691.7
17.591.6 e
15.691.2
18.091.2 e
238.793.5
264.495.5
251.294.8
e
275.095.9 e
250.893.8
279.695.2 e
246.795.7
261.296.0 e
253.795.0
267.594.1 e
240.092.1
257.592.8 e
238.794.1
243.794.2
251.294.8
268.794.5 e
230.894.2
247.194.2 e
249.293.1
275.093.8 e
12 14.791.3
12 14.891.2
12 13.390.5
12 14.290.5 e
8
8
14,591.3
17.091.2 e
12 17.490.9
12 19.391.1 e
8
8
8
8
13.190.8
15.091.1 e
17.591.4
20.991.5 e
Control
4
12 11.890.6
12 12.790.6 e
12 14.591.3
12 18.792.2 e
Control
Ambasilide
(5 mM)
Control
E-4031
(5 mM)
12 13.291.2
12 14.890.9 e
245.095.1
266.293.9 e
Table III. Effects of 2a–g and 4 on the 2 mM Cromakalim
(CR) evoked ERP shortening in dog cardiac right ventricular
trabecular musclea.
a Stimulation frequency 1 Hz.
b n=number of preparation.
c ICT=impulse conduction time.
d ERP=effective refractory period.
e PB0.05.
b
Compounds
(10 mM)
n
ERP c (ms) Change
(%)
Before CR d After CR
These changes suggested that these compounds exert
repolarisation lengthening (action potential duration
lengthening) effect in the dog ventricle. This possibility
was further tested by the conventional microelectrode
technique (see later). The possible effect of compounds
on the ATP-sensitive potassium current was also tested
with extracellular in vitro electrophysiological technique
measuring the ERP before and after the application of 2
mM Cromakalim. Cromakalim [13] is a known opener of
the ATP-sensitive potassium channels and thereby
shortens cardiac repolarisation and consequently the
ERP. The attenuation or prevention of the Cromakalim
induced ERP shortening was considered as inhibition of
the ATP sensitive potassium current. As table III shows
2a, 2e and 2f significantly attenuated the Cromakalim
evoked ERP shortening which suggests that these com-
pounds may inhibit the ATP sensitive potassium current
in some degree. The effects of the 2a,b,d–g, 4 and
reference compounds on the action potential parameters
Control
2a
Control
2b
Control
2c
Control
2d
Control
2e
Control
2f
Control
2g
Control
4
12 233.394.8
12 249.693.8
12 230.897.5
12 256.298.0
12 253.394.1
12 252.994.0
12 236.792.6
12 242.593.1
12 255.093.8
12 270.094.4
12 249.293.1
12 267.193.4
12 246.794.7
12 261.793.7
12 241.793.2
12 260.495.2
176.293.8 −24.391.5
212.595.3 −14.991.6 e
160.496.5 −30.690.9
166.796.7 −35.191.0
190.094.1 −25.091.0
185.092.3 −26.791.2
147.597.3 −37.892.8
157.599.4 −34.994.1
160.096.1 −37.292.4
229.296.0 −14.792.9 e
172.593.7 −30.791.7
233.394.5 −12.591.7 e
165.095.4 −32.692.5
190.094.1 −27.391.8
161.795.9 −33.391.7
179.695.2 −31.092.0
a Stimulation frequency 1 Hz.
b n=number of preparation.
c ERP=effective refractory period.
d CR=cromakalin (2 mM).
e PB0.05.

500
S. Villa et al. / European Journal of Medicinal Chemistry 36 (2001) 495–506
Table IV. Effects of 2a,b,d–g, 4 and reference compounds on
the action potential parameters in isolated dog right papillary
muscle.
search to those solutions whose numbers of pharma-
cophore points are as high as possible (because more
common pharmacophore points will present a better
overall fit) and whose superposition is still acceptable.
On the other hand, consideration based on X-ray crys-
tallographic evidence suggests that a rigid superposition
of pharmacophore points is unnecessary. Therefore one
Compounds N APA a (mV) APD90 (ms) Vmax (V/s)
(10 mM)
b
c
Control
2a
Control
2b
Control
2d
Control
2e
Control
2f
Control
2g
Control
4
Control
Ambasilide
(10 mM)
Control
E-4031 (1
mM)
3
3
4
4
4
4
3
3
3
3
3
3
3
3
4
4
107.092.1 263.7925.3
10993.2 261.0928.7
107.291.2 235.592.0
198.3926.9
203.7922.4
208.5922.4
,
generally considers a 1 A tolerance between pharma-
111.292.6 280.298.2 d 206.7915.37
,
cophoric elements to be very good and tolerance of 2 A
,
109.091.3 237.5910.4
245.7921.1
to be acceptable. Therefore 1.5 A tolerance, with the
108.791.9 260.7911.5 d 240.2923.2
maximum number of pharmacophoric elements seems to
104.391.5 245.394.9
105.292.0 245.3916.5
109.790.9 242.0914.5
111.790.3 229.0917.4
110.092.9 257.795.5
110.392.0 257.097.0
106.093.0 241.796.7
196.7930.1
176.3922.1
285.3921.7
283.0920.6
186.7915.6
187.0918.0
199.399.3
106.093.1 254.793.7 d 202.0910.2
111.091.7 247.7913.7
220.5912.9
111.792.6 299.7917.5 d 210.095.0
4
4
113.090.7 258.5910.4
174.597.5
113.590.6 301.294.2 d 179.791.7
Stimulation frequency 1 Hz.
a APA=action potential amplitude.
b APD90=90% repolarisation.
c Vmax=maximal rate of depolarisation.
d PB0.05.
Figure 5. The frequency dependent effect of 2b, 2d, ambasilide
(1) and E-4031 on the action potential duration at 90% of
repolarization (APD90) in dog right ventricular papillary mus-
cle is represented. The abscissa shows the stimulation fre-
quency expressed in cycle lengths, the ordinate indicates the
normalized APD lengthening expressed in percentage. Values
represent the mean of four successful experiments, bars indi-
cate standard error.
investigated the frequency dependent effect of 2b, 2d,
E-4031 and ambasilide on the APD in dog right papil-
lary muscle. As figure 5 shows, E-4031 exerted the
unfavourable reverse rate dependent APD prolongation
while after 2b, 2d and ambasilide application the APD
lengthening effect was not apparently dependent on the
stimulation frequency.
Table V. Minimal and maximal conformational energy values
(Kcal/mol) of compounds 2a–g and ambasilide.
4.2. Modeling
4.2.1. Conformation analysis
Compound
Emin
Emax
Results are summarised in table V, where minimal
and maximal conformational energy values for each
compound are listed. All generated structures were used
in DISCO computations.
Ambasilide
66.08
19.48
19.63
19.20
21.67
14.64
17.73
13.58
70.41
27.72
28.58
28.68
29.90
25.87
30.21
27.04
2a
2b
2c
2d
2e
2f
4.2.2. DISCO analysis
The calculation forced to find at least one donor
atom, one acceptor atom and two hydrophobic centres
resulted in 119 models. The user must constrain the
2g

S. Villa et al. / European Journal of Medicinal Chemistry 36 (2001) 495–506
501
of the final model are displayed in table VI. The dis-
tances of the most important pharmacophoric elements
are shown in table VII and VIII. Relative energy (Erel) is
the increase in conformational energy of a conformer
over the lowest energy conformer of the compound.
Reference structure ambasilide with pharmacophoric el-
ements found by DISCO is displayed by figure 7. The
key features of this pharmacophore (DA, HYD5 and
HYD6) are in appropriate agreement with the most
important elements of the model defined in a previous
study [15]. The pharmacophoric model defined could be
useful for designing novel compounds with Class III
antiarrhythmic activity.
5. Conclusions
Figure 6. The compounds as aligned by DISCO. The reference
molecule, ambasilide, is black.
A new series of ambasilide analogues were synthe-
sised which have the 3,8-diazabicyclo-[3.2.1]-octane
Table VI. Scores of the final model for compounds 2a–g and
ambasilide.
,
Table VII. Distances (in A) between pharmacophoric elements
of the final model.
a
c
Compound
FIT
Overlap b DE
Tolerance d
,
Distance (A)
TOL
1
2
Ambasilide
0.00
0.82
0.83
1.74
1.73
1.28
1.09
1.55
273.12
170.37
167.75
144.00
144.00
157.25
177.75
156.37
0.00
0.00
1.19
1.19
1.19
1.19
1.00
1.46
1.12
5.53
3.59
4.87
3.63
6.44
2.87
1.52
6.94
3.81
1.58
5.02
5.23
6.28
4.72
6.31
1.50
1.50
1.50
1.50
1.50
1.50
1.50
1.50
1.50
1.50
1.50
1.50
1.50
1.50
1.50
AA1
AA1
AA1
AA1
AA1
DA2
DA2
DA2
DA2
2a
2b
2c
2d
2e
2f
0.42
0.80
0.72
0.40
2.15
4.08
7.90
HYD3
HYD4
HYD5
HYD6
HYD3
HYD4
HYD5
HYD6
HYD4
HYD5
HYD6
HYD5
HYD6
HYD6
2g
DA2
a FIT, RMS deviation of feature atom coordinates from those
HYD3
HYD3
HYD3
HYD4
HYD4
HYD5
,
of the corresponding reference features (A).
b OVERLAP, intersection of the van der Waals volumes of the
3
,
reference and the superimposed molecule (A ).
c DE, the increase in conformational energy of the conformer
over the minimal energy conformation of the compound.
d Tolerance, maximal distances between the corresponding fae-
tures of the reference and the superimposed molecule.
TOL=tolerance of the distances. AA1=H-bond acceptor
atom, DA2=H-bond donor atom, HYD3,4,5,6=Hydropho-
bic centers.
be a good compromise. The final model has been se-
lected from the set of models mentioned above by the
energy values of the reference conformers, since the
bioactive conformation probably is one of the low-en-
ergy conformers. Moreover a visual inspection of the
alignment of this model, also revealed a favorable over-
lap of molecular parts other than those identified for the
pharmacophore. The final model is displayed in figure 6.
It contains one donor atom, (protonated N), one accep-
tor oxygen atom and four hydrophobic centers. Scores
(fit, overlap, relative energy of the selected conformer)
Table VIII. Atomic coordinates of the final model.
Atom
X
Y
Z
AA1
DA2
HYD3
HYD4
HYD5
HYD6
−5.764856
−9.541248
−7.804212
−9.055249
−2.912106
−7.024588
−1.590058
−4.035973
−4.352076
−4.621854
−3.460097
−3.495673
−3.191282
0.026391
−2.264429
−1.237577
−1.956332
2.834229

502
S. Villa et al. / European Journal of Medicinal Chemistry 36 (2001) 495–506
ical values. The structures of all compounds were
consistent with their analytical and spectroscopic data.
6.1.1. 3-Benzyl-8-substituted
benzoyl-3,8-diazabicyclo-[3.2.1]-octanes (2a,d)
A mixture of the appropriate benzoyl chloride (0.01
mol), 3-benzyl-3,8-diazabicyclo-[3.2.1]-octane [9] (0.01
mol) and triethylamine (0.01 mol) in dichloromethane
(20 mL) was refluxed overnight. After cooling, the inor-
ganic salts were filtered off, the solvent evaporated and
the crude product purified by silica-gel chromatography,
eluting with dichloromethane–ethyl acetate (9:1) (see
table I for data).
Figure 7. Pharmacophoric elements of ambasilide found by
DISCO.
6.1.2. 3-Benzyl-8-(4%-methanesulfonanilide)-
carbonyl-3,8-diazabicyclo-[3.2.1]-octane (2b)
To a vigorously stirred solution of 2a (1 g, 3.1 mmol)
in pyridine (20 mL) cooled at 0–5 °C, methanesulfonyl
chloride (0.25 mL, 3.1 mmol) was added dropwise. The
reaction mixture was poured into water (80 mL) and
extracted with ethyl acetate (3×30mL). The organic
layers were collected and dried (Na2SO4) and the solvent
evaporated to give 2b, which was purified by silica gel
chromatography, eluting with dichloromethane–
methanol (49:1) (see table I for data).
moiety instead of the 3,7-diazabicyclo-[3.3.1]-nonane
present in the reference compound. The compounds
were tested to verify their capability to exert Class III
antiarrhythmic effects, both in vitro extracellular elec-
trophysiological assays and by the conventional mi-
croelectrode technique. Most of them lengthened the
effective refractory period (ERP) with significant in-
crease on the impulse conduction time (ICT). More-
over, they lengthened the action potential duration
(APD), a typical Class III feature, without exerting
any significant Class I antiarrythmic activity.
6.1.3. 3-Benzyl-8-(4%-bis-methansulfonanilide)-
Compounds reported in this paper are expected to
possess significant Class III type antiarrhythmic activ-
ity. Further work is now in progress to investigate the
in vivo effects of these substances.
carbonyl-3,8-diazabicyclo-[3.2.1]-octane (2c)
To a vigorously stirred solution of 2a (1 g, 3.1 mmol)
in dichloromethane (20 mL), cooled at 0–5 °C,
methanesulfonyl chloride (0.5 mL, 6.2 mmol) was added
dropwise and the reaction refluxed for 1 h. After cool-
ing, the reaction mixture was poured into water (80 mL)
and extracted with ethyl acetate (3×30 mL). The or-
ganic layers were collected and dried (Na2SO4) and the
solvent evaporated to give 2c, which was purified by
silica gel chromatography, eluting with dichloro-
methane–methanol (19:1) (see table I for data).
6. Experimental
6.1. Chemistry
Melting points were determined with a Bu¨chi 510
1
capillary apparatus and are uncorrected. H NMR spec-
tra were recorded on a Brucker AC200 spectrometer;
chemical shifts are reported as l (ppm) relative to
tetramethylsilane as internal standard; dimethyl
sulphoxide-d6 was used as the solvent, unless otherwise
noted. TLC on silica-gel plates was used to check
product purity. Silica gel 60 (Merck 70–23 mesh) was
used for column chromatography while Silica gel 60
(Merck 230–400 mesh) for column flash chromatogra-
phy. Elemental analyses were within 0.4 of the theoret-
6.1.4. 3-Benzyl-8-aryloxy-ethylen-3,8-
diazabicyclo-[3.2.1]-octane (2e,f)
A mixture of the appropriate chloride (5a,b) (0.01
mol), 3-benzyl-3,8-diazabicyclo-[3.2.1]-octane [9] (0.01
mol) and triethylamine (0.01 mol) was heated at 180 °C
for 0.5 h. After cooling the crude product was purified
by silica gel chromatography, eluting with dichloro-
methane–methanol (19:1) (see table I for data).

S. Villa et al. / European Journal of Medicinal Chemistry 36 (2001) 495–506
503
6.1.5. 3-Benzyl-8-(4%-methanesulfonanilide)-
ethylen-3,8-diazabicyclo-[3.2.1]-octane (2g)
Compound 2g was prepared as above reported for 2b
starting from 2e (see table I for data).
chromatography, eluting with dichloromethane–ethyl
acetate (19:1).
Yield 10% oil. 1H NMR: (CDCl3) 1.3 (s, 3H CH3); 1.4
(s, 3H CH3); 2.1–2.3 (m, 2H H-3,5); 2.7 (m, 2H H-3,5);
3.5 (s, 2H CH2ꢀPh); 3.8–4.4 (m, 2H H-2,6); 7.2–7.4 (m,
5H HArꢀPh); 7.5 (d, 2H H-2%,6%); 8.2 (d, 2H H-3%,4%).
6.1.6. 3-(4%-Aminobenzoyl)-8-benzyl-3,8-diazabicyclo-
[3.2.1]-octane (3)
6.1.8. 2-(4-acetamidophenoxy)-1-ethanol (5a) and
2-(4-aminophenoxy)-1-chloroethane (5b)
The 8-Boc-3,8-diazabicyclo-[3.2.1]-octane [12] was
condensed with p-amino-benzoyl chloride as above re-
ported for 2a to give the 3-(4%-aminobenzoyl)-8-Boc-3,8-
diazabicyclo-[3.2.1]-octane. Yield 60%, oil.
To a well stirred solution of 4-acetamidophenol (33.75
g, 0.222 mol) and sodium hydroxide (9.38 g, 0.235 mol)
in water (118 mL), ethylene chlorohydrine (18.75 g,
0.232 mol) was added, and the reaction mixture heated
at 60–70 °C for 8 h. After cooling, the solution was
kept in the refrigerator overnight. The so formed crys-
tals were filtered, washed with water (2×50 mL) and
dried to give 2-(4-acetamidophenoxy)-1-ethanol (5a) [10]
1H NMR: (CDCl3) 1.5 (s, 9H 3×CH3); 1.7–2.0 (m,
4H H-6,7); 3.0–3.4 (m, 4H H-2,4); 3.9 (bs, 2H exch.
with D2O NH2); 4.1–4.4 (m, 2H H-1,5); 6.7 (d, 2H
H-3%,5%); 7.2 (d, 2H H-2%,6%).
To a cooled solution of 3-(4%-aminobenzoyl)-8-Boc-
3,8-diazabicyclo-[3.2.1]-octane (0.01 mol) in diethyl
ether (10 mL) a solution of HCl in diethyl ether was
added until the pH was about 1 and the mixture stirred
overnight at room temperature (r.t.). Sodium hydroxide
(6 N) was then added until pH 9, the layers separated
and the aq phase extracted twice by diethyl ether (2×20
mL) The reunited organic layers were dried over sodium
sulfate and the solvent evaporated to give 3-(4%-amino-
benzoyl)-3,8-diazabicyclo-[3.2.1]-octane (90%) which
was directly condensed with equimolar benzyl chloride
in CH2Cl2 at reflux in the presence of TEA to give 3.
Yield 65%, oil.
1
(36.68 g; 84.6%). mp 124–125 °C H NMR (CDCl3):
2.1 (s, 3H CH3); 2.2 (bs, 1H exch. with D2O OH); 3.8
(m, 2H CH2ꢀN); 4.2 (t,2H CH2ꢀO); 6.8 (d, 2H H-2%,6%);
7.2 (bs, 1H exch. with D2O NH); 7.4 (d, 2H H-3%,5%).
To a well stirred solution of 2-(4-acetamidophenoxy)-
1-ethanol (13.6 g, 0.07 mol) in chloroform (26 mL) and
N,N-dimethyl-formamide, at 0–5 °C, a cold (0–5 °C)
solution of thionyl chloride (6 mL) in chloroform (9
mL) was added dropwise during 30 min. The mixture
was refluxed for 2.5 h, then evaporated to dryness under
vacuum to give an orange oil, which was added of water
(70 mL) and conc. HCl (68 mL), refluxed for 0.5 h and
filtered while still hot. After cooling, the so formed
crystals were separated to give 2-(4-aminophenoxy)-1-
chloroethane hydrochloride [10], which was dissolved in
water (74 mL) at 60 °C and the pH brought to 7 by
dropwise addition of conc. ammonium hydroxide (5.2
mL). The mixture was cooled and filtered, the crystals
washed with water (2×50 mL) and dried to give 5b [11]
1H NMR: (CDCl3) 1.4–2.0 (m, 4H H-6,7); 2.2–2.5
(m, 1H H-2); 2.6–2.8 (m, 1H H-4); 3.0-3.4 (m, 2H exch.
with D2O NH2); 3.4–3.5 (m, 2H H-2,4); 3.8–4.0 (m, 1H
H-1); 4.3 (s, 2H CH2ꢀPh); 4.6–4.8 (m, 1H H-5); 6.5–6.7
(m, 2H H-3%,5%); 7.2–7.5 (m, 7H 5HArꢀPh., H-2%,6%).
6.1.7. 4-Benzyl-1-(4%-nitro-phenyl)-
carbonyl-2,6-dimethyl-piperazine (4)
1
(10.1 g, 85%), m.p.: 88–90 °C. H NMR (CDCl3): 3.8
A mixture of 2,6-dimethyl-piperazine (1 g, 8.75
mmol), benzyl chloride (1.1 g, 8.75 mmol) and triethy-
lamine (0.85 mL, 8.75 mmol) in toluene (10 mL) was
refluxed overnight. After cooling, the inorganic salts
were filtered off and the solvent evaporated to give
4-benzyl-2,6-dimethyl-piperazine which was used with-
out further purification.
A mixture of 4-benzyl-2,6-dimethyl-piperazine (0.74 g,
3.6 mmol), p-nitro-benzoyl chloride (0.67 g, 3.6 mmol)
and triethylamine (0.35 mL, 3.6 mmol) in toluene (10
mL) was refluxed overnight. After cooling, the inorganic
salts were filtered off and the solvent evaporated to give
a crude product (0.12 g), which was purified by silica gel
(t, 2H CH2ꢀCl); 4.2 (t, 2H CH2ꢀO); 6.8 (d, 2H H-3%,5%);
7.2 (bs, 1H exch. with D2O); 7.4 (d, 2H H-2%,6%).
6.2. Biological section
6.2.1. Extracellular electrophysiological recordings [16]
Adult mongrel dogs (5–12 kg) were anaesthetized by
30 mg/kg intravenous pentobarbital. Their hearts were
removed and small pieces of right ventricular papillary
and trabecular muscle muscles were prepared and placed
into the tissue bath (50 mL) which contained modified
Tyrode’s solution (Na+ 140 mM, K+ 4 mM, Ca 1.8 mM,
Mg2+ 1.0 mM, Cl− 129.6 mM, HCO−3 20 mM and

504
S. Villa et al. / European Journal of Medicinal Chemistry 36 (2001) 495–506
glucose 11 mM) and was oxygenated with 95% O2 and
5% CO2. The temperature was kept constant at 36
0.5 °C. The pH was 7.4 0.05.
(frequency 1 Hz), using 2 ms rectangular constant cur-
rent pulses isolated from the ground across bipolar
platinum electrodes in contact with the preparation. At
least 1 h was allowed for each preparation to equilibrate
while it was continuously superfused with Tyrode’s solu-
tion. The temperature was kept constant at 37 °C.
Transmembrane potentials were recorded by conven-
tional microelectrode techniques. Microelectrodes filled
with 3 M KCl with tip resistances of 5–20 MV were
connected to the input of a high-impedance electrometer
(HSE microelelctrode amplifier type 309) referenced to
the ground. The first derivative of transmembrane
voltage with respect to time (Vmax) was electronically
derived by (HSE differentiator type 309) with linear
response in the range of 20–1000 V/s. Voltage outputs
from all amplifiers were displayed on a dual-beam mem-
ory oscilloscope (Tektronix 2230 100 MHz digital stor-
age oscilloscope).
Resting membrane potential (RP), action potential
amplitude (APA), and action potential duration at 90%
of repolarization (APD90) were obtained with software
(developed in our institution) with an 386 IBM-compat-
ible computer connected to the digital output of the
oscilloscope. When different steady-state stimulation cy-
cle lengths were applied, action potential parameters
were measured at each cycle length after adaptation to
the new pacing cycle length (usually after 2 min).
The preparations were superfused for 30 min with the
tested compounds before measurements started. If possi-
ble, the same impalement was maintained throughout
the experiment. When impalement was lost during mea-
surement, readjustment was attempted. If the readjusted
parameters deviated more than 10% from the previous
parameters experiments were terminated.
During the equilibration period (60 min) the right
ventricular trabecular muscle preparations were driven
with stimuli of 2 ms duration and twice the diastolic
threshold strength at 1000 ms of cycle length through
bipolar platinum electrodes. To measure conduction
time, bipolar extracellular platinum electrodes (diame-
ter=0.1 mm) were placed on the surface of the right
ventricular wall along the direction of the trabecular
muscle fibers, 10–12 mm away from the stimulating
electrodes, and the propagated biphasic action poten-
tials were recorded extracellularly. The propagated ex-
tracellular action potentials appeared at the recording
electrode, and the time difference between the sign of
the stimulus artifact and the first inflexion point of the
extracellular potential provided a measure of conduction
time. The extracellular action potentials were amplified
with an amplifier (Eltron GMK) and were displayed on
the screen of an oscilloscope (Medicor VM 62 A).
The effective refractory period (ERP) was determined
at three times the threshold strength using twin impulses
with gradually increasing the coupling intervals.
The stimulation frequency was varied between 3 and
0.5 Hz. Measurements were taken after sufficient adap-
tation of the fibers to the new cycle length.
Drugs were diluted directly in the tissue bath from 10
mM stock solution dissolved in DMSO to reach the
appropriate final concentrations. The solvent at 0.1%
did not evoke significant changes in the measured
parameters.
The drug effects were determined after 30–45 min
equilibration period after which steady-state drug effects
were achieved. Student’s t-test for paired data was used
to determine statistical significance of the results. The
results were considered significant when P was <0.05.
6.2.2. Conventional microelectrode technique [17]
Adult mongrel dogs (5–12 kg) of both sexes were
used. After animals were anaesthetized with sodium
pentobarbital 30 mg/kg intravenously, their hearts were
rapidly removed through right lateral thoracotomy and
immediately rinsed in oxygenated Tyrode’s solution con-
taining (in mM): Na+ 140, K+ 4, Ca2+ 1.8, Mg2+ 1, Cl−
129.6, HCO−3 20, and glucose 11. The pH of this solution
was 7.35–7.45 when gassed with 95% O2/5% CO2 at
37 °C. Right ventricular papillary muscles were individ-
ually mounted in a tissue chamber (ꢀ50 mL vol). Each
preparation was initially stimulated (HSE stimulator
type 215/II.) at a basic cycle length (BSL) of 1000 ms
6.3. Molecular modeling methods
Mapping the conformational spaces of ligands was
carried out with an energy window high enough to
include the putative bioactive conformation. Each con-
former set was next provided for a DISCO analysis. All
calculations were performed on a Silicon Graphics
INDY R4400 workstation by utilizing Tripos’ SYBYL
molecular modeling software.
6.3.1. Molecular conformations
Starting structures were drawn by Sketch Molecule
option, and geometry optimizations were carried out
with Tripos force field.

S. Villa et al. / European Journal of Medicinal Chemistry 36 (2001) 495–506
505
Sets of conformations of all compounds were gener-
ated by Multisearch option in DISCO interface. This
conformer generator, like Randomsearch, searches vari-
ous energy minima available to a molecule by randomly
perturbing torsions, followed by minimisation and,
finally eliminating duplicates. In Multisearch method,
torsions are limited to those which can change the
internal geometry of atoms that determine DISCO fea-
tures. Therefore, in connection with a DISCO confor-
mation is not necessarily identical with the lowest energy
conformation, energy cutoff was set relatively high value
(15 kcal/mol). Searches were done with the following
parameters:
least number of conformations.) These class features
were found common in all structures.
The analysis was done with the following parameters:
STRUCTURE REQUIREMENTS:
Match All
–
FEATURE REQUIREMENTS:
Do 3 To 8
DISTANCE TOLERANCE
Coarse Fit (range of tolerance was varied between 0.5
A- 5.0 A)
Analysis, the Multisearch method seems to be supe-
rior over Randomsearch that does not check distances
of features.
Acknowledgements
In our cases, computations were always stopped after
100 conformations have been found, since DISCO was
limited to study less than 100 conformations of each
compound. Since bioactive conformation is not neces-
sarily identical with the lowest energy conformation,
energy cutoff was set relatively high. Searches were done
with the following parameters:
This work was supported by grants obtained from the
Hungarian National Research Foundation (OTKA T
032558 and T031910), Hungarian Ministry of Health
(ETT 536/2000) and from the Hungarian Ministry of
Education FKFP (1025/1997). Financial support from
CNR and Ministero dell’Universita’ e della Ricerca
Scientifica is also acknowledged.
PRESCREENING:
Minimizer: Maxmin
References
Cutoff: 15 kcal/mol
Maxmin RMS Gradient: 3.0
[1] The Cardiac Arrhythmia Suppression Trial (CAST) Investiga-
tors: Preliminary report. Effect of encainide and flecainide on
mortality in a randomized trial of arrhythmia suppression after
myocardial infarction, N. Engl. J. Med. 321 (1989) 406–412.
TERMINATION:
[2] Ohler A., Amos G.J., Wettwer E., Ravens U., Naunyn
Max. Cycles: 3000
Schmiedeberg’s Arch. Pharmacol. 349 (1994) 603–610.
[3] Waldo A.L., Camm A.J., deRuyter H., Freidman P.L., Mac-
Neil D.J., Pitt B., Pratt C.M., Rodda B.E., Schwartz P.J., Am.
J. Cardiol. 75 (1995) 1023–1027.
Max. Conformers: 100
Maxmin RMS Gradient: 0.05
CONFORMER COMPARISON:
Method: Fit
[4] Varro´ A., Lathrop D.A., J. Cardiovasc. Pharmacol. 16 (1990)
557–567.
[5] Bala´ti B., Iost N., Simon J., Varro´ A, Papp J.G., Gen. Pharma-
col. 34 (2000) 85–93.
[6] Phillips G.B., Morgan T.K., Nickisch K., Lind J.M., Gomez
R.P., Wolh R.A., Argentieri T.M., Sullivan M.E., J. Med.
Chem. 33 (1990) 627–633.
Criteria: Max. Distance
Deviation Limit: 0.40.
[7] Connors S.P., Dennis P.D., Gill E.W., Terrar D.A., J. Med.
Chem. 34 (1991) 1570–1577.
6.3.2. DISCO analysis
[8] Morgan T.K., Lis R., Lumma W.C., Nickisch K., Wolh R.A.,
Phillips G.B., Gomez R.P., Lampe J.W., Di Meo S.V., Marisca
A.J., Forst J., J. Med. Chem. 34 (1990) 1091–1097.
Using scan rows for reference option of DISCO,
Ambasilide was offered as reference with the summary
of possible model components: one donor atom (DA),
one acceptor site (AS), three donor sites (DS), two
acceptor atoms (AA) and four hydrophobe rings (Hyd).
(DISCO as a rule, selects the reference molecule with the
least number of possible points for superposition and
[9] Cignarella G., Nathansohn G.G., J. Org. Chem. 26 (1961)
2747.
[10] Herbst S., J. Org. Chem. 17 (1952) 595–598.
[11] Sharpe C.J., Palmer P.J., Evans D.E., Brown G.R., King G.,
Shadbolt R.S., Trigg R.B., Ward R.J., Ashford A., Ross J.W.,
J. Med. Chem. 15 (1972) 523–529.

506
S. Villa et al. / European Journal of Medicinal Chemistry 36 (2001) 495–506
[12] Cignarella G., Barlocco D., Vianello P., Villa S., Pinna G.A.,
Fadda P., Fratta W., Toma L., Gessi S., Il Farmaco 53 (1998)
667–674.
[15] Matyus P., Borosy A.P., Varro A., Papp J.G., Barlocco D.,
Cignarella G., Int. J. Quant. Chem. 69 (1998) 21–30.
[16] Varro´ A., Bo´di I., Jedna´kovits A., Rabloczky G., Arch. Int.
Pharmacodyn. Ther. 292 (1988) 157–165.
[13] Sanguinetti M.C., Scott A.L., Zingaro G., Siegl P.K.S., Proc.
Natl. Acad. Sci. USA 85 (1988) 8360–8364.
[17] Ha´la O., Ne´meth M., Varro´ A., Papp J.G., J. Cardiovasc. Phar-
macol. 24 (1994) 559–565.
[14] Hondeghem L.M., Snyders D., J. Circulation 81 (1990) 686–690.
.