Probing the pharmacophore for allosteric ligands of muscarinic M 2 receptors: SAR and QSAR studies in a series of bisquaternary salts of caracurine V and related ring systems

Article abstract of DOI:10.1021/jm0311341

Allosteric effects on muscarinic acetylcholine M₂ receptors were examined in a series of bisquaternary salts of the Strychnos alkaloid caracurine V (6) and related iso-caracurine V, tetrahydrocaracurine V, and bisnortoxiferine ring systems. The

Full text of DOI:10.1021/jm0311341

J . Med. Chem. 2004, 47, 3561-3571
3561
P r obin g th e P h a r m a cop h or e for Alloster ic Liga n d s of Mu sca r in ic M2 Recep tor s:
SAR a n d QSAR Stu d ies in a Ser ies of Bisqu a ter n a r y Sa lts of Ca r a cu r in e V a n d
Rela ted Rin g System s
Darius P. Zlotos,*,† Stefan Buller,‡ Nikolaus Stiefl,† Knut Baumann,† and Klaus Mohr‡
Pharmaceutical Institute, University of Wu¨rzburg, Am Hubland, 97074 Wu¨rzburg, Germany, and Institute of Pharmacy,
Department of Pharmacology and Toxicology, University of Bonn, An der Immenburg 4, 53121 Bonn, Germany
Received December 11, 2003
Allosteric effects on muscarinic acetylcholine M2 receptors were examined in a series of
bisquaternary salts of the Strychnos alkaloid caracurine V (6) and related iso-caracurine V,
tetrahydrocaracurine V, and bisnortoxiferine ring systems. The compounds inhibited dissocia-
tion of the orthosteric antagonist [3H]N-methylscopolamine (NMS) from porcine cardiac M2
receptors with EC0.5,diss values from 4 to 3270 nM. The majority of compounds hardly changed
[3H]NMS equilibrium binding, indicating similar binding affinities in free and NMS-occupied
M2 receptors. The most potent agents were found in the caracurine V, iso-caracurine V, and
tetrahydrocaracurine V series and carried nonpolar alkyl groups with a maximal chain length
of three carbon atoms. 3D QSAR (CoMSIA) analysis explained the wide range of binding
affinities by steric and electrostatic properties of the side chains. Furthermore, the findings
suggest that the spatial orientation of the “caracurine” aromatic rings compared with the
bisnortoxiferine ring skeleton is favorable to optimal allostere-receptor interactions.
In tr od u ction
systems.26-28 These pharmacophoric elements are fixed
in the very rigid ring skeleton of the neuromuscular
blockers toxiferine I 1 and alcuronium 2.
Apart from the conventional orthosteric binding site,
to which the endogenous ligand acetylcholine and
competitive antagonists such as N-methylscopol-
amine and atropine bind, all five muscarinic receptor
subtypes (M1-M5) contain a second, allosteric binding
site.1-7 Due to greater sequence differences between the
amino acids within the allosteric sites on different
muscarinic subtypes, ligands for the allosteric site
appear to have a higher potential for subtype selectivity
than orthosteric ligands. Furthermore, allosteric agents
may achieve subtype selectivity not only by different
binding affinities to the allosteric sites but also by a
subtype selective cooperativity with the orthosteric
ligand, i.e., the equilibrium binding of an orthosteric
ligand can be either enhanced or reduced or left un-
affected by the allosteric modulator in a subtype-
dependent fashion. The cardiac M2 receptor appears to
be especially sensitive to allosteric modulation.8-10 Its
allosteric site is well-defined, 9,11,12 and affinity data for
a number of structurally different allosteric modulators
have been reported for M2 receptors occupied by [3H]-
N-methylscopolamine ([3H]NMS).13 So far, SAR studies
have been carried out in several series of compounds
derived from hexamethonium,14 bispyridinium TMB-4,15
W84,16-18 WDUO and IWDUO,19 strychnine and bru-
cine,20 truxillic acid,21,22 thiazoloandrostane,23 penta-
cyclic carbazolones,24 and bisquaternary dimers of strych-
nine and brucine.25 Molecular modeling studies led to
a hypothesis of the pharmacophore consisting of two
positively charged nitrogen atoms at a distance of
approximately 10 Å surrounded by two aromatic ring
Alcuronium is a well-known, very potent classical
enhancer of NMS binding to muscarinic M2 receptors.29
Its cyclization product, diallylcaracurinium salt V 3, was
identified as an allosteric agent with a similar high
binding affinity for the M2 receptor.30
* To whom correspondence should be addressed. Tel. +49 931-888-
5489; fax +49 931-888-5494; e-mail: zlotos@pzlc.uni-wuerzburg.de.
† University of Wu¨rzburg.
Studies at M2/M5 chimeric receptors and site-directed
mutagenesis in wild-type and chimeric receptors allowed
‡ University of Bonn.
10.1021/jm0311341 CCC: $27.50 © 2004 American Chemical Society
Published on Web 05/29/2004

3562 J ournal of Medicinal Chemistry, 2004, Vol. 47, No. 14
Zlotos et al.
identification of the M2 receptor epitopes responsible for
the high M2/M5 selectivity of 2 and 3.12,31 3 derives M2/
M5 selectivity in NMS-occupied receptors from two
amino acids: tyrosine-177 at the second outer loop of
M2 and threonine-423 at the top of the seventh trans-
membrane domain. In case of 2, only an epitope con-
tained in the second outer loop is responsible for high
M2/M5 selectivity. Molecular modeling studies revealed
the π-π interaction between one of the aromatic rings
of 3 and the benzene ring of tyrosine-177 to be essential
in the ligand-receptor complex.31 The divergent binding
properties of 2 and 3 could be due to a different relative
spatial arrangement of the pharmacophoric elements in
both ring systems. The stereochemistry of caracurine
V,32 toxiferine I, and the related iso-caracurine V33 and
tetrahydrocaracurine V ring systems was recently de-
termined by means of NMR spectroscopy and semi-
empirical calculations.34 While the relative orientation
of the aromatic indole rings is similar in all three
“caracurine“ ring systems, opening of both tetrahy-
drooxepine rings of caracurine V to give the bisnor-
toxiferine ring skeleton leads to the conformational
change of the central eight-membered ring from a crown
form to a boat form, resulting in a considerably changed
geometry of the whole ring skeleton. Thus, 3 and related
agents appear to be useful tools to study the molecular
mechanisms involved in ligand recognition by the mus-
carinic allosteric site.
The purpose of this study is to examine how the
stereochemical difference in the above-mentioned ring
systems affects their affinity for the allosteric site of
muscarinic M2 receptors. Our preliminary report on the
bisquaternary caracurine V series30 as well as studies
in the monoquaternary strychnine and brucine series20
indicated that the affinity for the allosteric site is very
sensitive to the properties of the N-substituent. To find
the best affinity-increasing groups, caracurine V was
first quaternized with a great number of N-substituents
of different size, lipophilicity, and electronic properties.
The affinities of the resulting compounds for the allo-
steric site of NMS-occupied M2-receptors were analyzed
by means of 3D-QSAR CoMSIA study. The findings
should result in an improved understanding of the
allosteric pharmacophore and help develop better tools
for probing the M2 allosteric binding site.
droxyl group.37 The E-configuration of the major isomer
was confirmed by the chemical shift of C-17 at δ 74.9
ppm, upfield by 4.3 ppm from the corresponding signal
of the Z-isomer owing to γ-interaction between the
hydroxyl group and C-17 in E-4.38 The E-configuration
of the major oxime is opposite to the stereochemistry
proposed previously by Hyman et al.36 Only E-4 was
used as a starting material for the synthesis of the key
intermediate Wieland-Gumlich aldehyde (5).36 5 exists
in a DMSO solution as an approximately 1:1 mixture
of two C-17 epimers as indicated by two sets of signals
1
in both H and 13C NMR spectra (400 MHz, DMSO-d6).
Tetramethoxycaracurine V (6a ) was prepared simi-
larly to 6 (Scheme 1). Nitrosation of brucine, recently
decribed by Zlotos et al.,25 gave two isomeric oximes,
E-4a (89%) and Z-4a (2%). Interestingly, treatment of
E-4a with thionyl chloride and subsequent heating in
aqueous HCl provided the expected dimethoxy-Wieland-
Gumlich-aldehyde (5a ) in only 17%. The very low yield
of the desired 5a can be explained by a different course
of the second-order Beckmann rearrangement of E-4a
compared to E-4. While in E-4 the C17-C22 bond was
preferentially cleaved, cleavage of the C22-C23 bond,
probably induced by the electron-donating methoxy
group in position 10 of the aromatic ring, was the major
reaction of E-4a . The resulting cyano group in position
17 was subsequently hydrolyzed by aqueous HCl, giving
a water-soluble amino acid which could not be isolated
from the reaction mixture. Treatment of the reaction
mixture upon evaporation of thionyl chloride with steam
at pH 3 instead of aqueous HCl yielded the unchanged
nitrile 7 (Scheme 2).
The absolute configuration of C-17 was unchanged in
the course of the rearrangement as indicated by the
coupling constant between H-17 and H-16 (J ) 2.3 Hz)
which reflects the cis-orientation of the hydrogen atoms.
19,20,19′,20′-Tetrahydrocaracurine V (8) was obtained
by catalytic hydrogenation of 6 in ethanol using PtO2
as catalyst as described in our preliminary report.34
Since quaternization attempts of 8 with methyl iodide
and allyl bromide led to decomposition of the tetrahy-
drocaracurine V ring system, dimethyltetrahydrocara-
curinium V dichloride (10) was prepared by the catalytic
hydrogenation of dimethylcaracurinium V dichloride
(11) with PtO2 in water. The double quaternization of
caracurine V (6) using 2.5-fold excess of methyl iodide
and a number of alkyl bromides readily proceeded in a
chloroform solution at room temperature. Introduction
of propyl and phthalimidopropyl substituents required
the corresponding alkyl iodides and higher tempera-
tures. The resulting bisquaternary caracurinium V salts
3 and 11-30 are displayed in Table 1.
Ch em istr y
Caracurine V (6) was obtained from the dimerization
of the Wieland-Gumlich aldehyde (5) using pivalic acid
according to the modified procedure of Battersby and
Hodson.35 5 was prepared from strychnine by nitrosa-
tion at C-22 using isoamyl nitrite/sodium ethoxide and
subsequent second-order Beckmann rearrangement of
the resulting E-22-oximinostrychnine (E-4) with thionyl
chloride as previously described by Hyman et al.36
However, a much more convenient method for preparing
E-4 involved using tert-butyl nitrite/t-BuOK as a nit-
rosation agent. Apart from the major E-stereoisomer of
4 (64%), a small amount of Z-4 (17%) was obtained. The
stereochemistry was determined by NMR spectroscopy.
The H-17 resonance signal of the major E-4 isomer
appeared at δ 5.03 ppm, downfield by 0.32 ppm from
the corresponding signal of the minor Z-4 isomer due
to the anisotropic deshielding effect of the oxime hy-
N-Allylcaracurinium V bromide (31), the only mono-
quaternary compound in this series, was obtained by
treatment of allyl bromide with a large excess of 6.
Quaternization of tetramethoxycaracurine V (6a ) with
allyl bromide in chloroform gave the only tetramethoxy-
caracurinium V salt (32).
Caracurine V dioxide (33) was prepared by N-oxida-
tion of 6 using hydrogen peroxide as previously de-
scribed by Verpoorte.39 Iso-caracurine V (34) was ob-
tained from decomposition of 6 with TFA according to
Zlotos.33 It could be readily double-alkylated with meth-
yl iodide, allyl bromide, and 3-bromocyclohexene in

Bisquaternary Salts of Caracurine V
J ournal of Medicinal Chemistry, 2004, Vol. 47, No. 14 3563
Sch em e 1a
a
Reagents and conditions: (i) tert-butyl nitrite/t-BuOK/toluene, 50 °C, 2 h; (ii) SOCl2, rt, 2.5 h, (iii) 2 M HCl, 100 °C, 3 h; (iv) pivalic
acid, 120 °C, 14 h; (v) 2 R’-Hal, CHCl3, rt, 30 min.; (vi) TFA, 15 h; (vii) 2 R-Br, CHCl3, rt, 30 min; (viii) 1. MeI, CHCl3, rt, 30 min; 2.
IRA-400 resin, chloride phase; 3. H2 (50 bar)/PtO2, H2O, 16 h.
Sch em e 2a
Kinetic dissociation binding experiments were con-
ducted, and the allosteric effect of the test compounds
on the dissociation of [3H]NMS was determined (two to
four experiments for each compound). The concentration
of the test compounds that inhibits [3H]NMS dissocia-
tion half-maximally, EC0.5,diss, can be taken as a mea-
sure of the equilibrium dissociation constant of allosteric
agent binding to [3H]NMS-occupied receptors.40,9 The
EC0.5,diss values are compiled in Table 1.
a
Reagents and conditions: (i) SOCl2, rt, 2.5 h; (ii) H2O; 1 M
Na2CO3 to pH 3; (iii) steam, 90 °C, 1.5 h.
Equilibrium binding experiments were performed to
get insight in affinities of the test compounds to NMS-
free receptors and to obtain information on cooperative
effects between the allosteric and the orthosteric ligand.
The effect of the test compounds on [3H]NMS equilib-
rium binding was investigated at a selected concentra-
tion of allosteric modulator (EC0.25,diss) that reduced the
rate of [3H]NMS dissociation by about 75% down to a
level of 25% of the control. The equilibrium binding data
(from two to five independent experiments with qua-
druplicated values) were expressed relative to the
control binding in the absence of modulator, which was
set as 100%. These data are included in Table 1. [3H]-
NMS equilibrium binding of approximately 100% indi-
cates neutral cooperativity, which means that [3H]NMS
chloroform to give the bisquaternary analogues 35-37
(Scheme 1).
The Calabash-curare alkaloid toxiferine I (1) was
prepared from the dimerization of Wieland-Gumlich
aldehyde methochloride using pivalic acid as described
by Battersby and Hodson.35
P h a r m a cologica l Stu d ies
The allosteric interaction of the test compounds with
muscarinic M2 receptors was studied in homogenates
of porcine heart ventricles with [3H]N-methylscopol-
amine ([3H]NMS) as the ligand for the orthosteric
receptor site.

3564 J ournal of Medicinal Chemistry, 2004, Vol. 47, No. 14
Zlotos et al.
Ta ble 1. Pharmacological Data for All Investigated
Compounds (for details, see text)
to have a higher affinity for free than for NMS-liganded
receptors. For the majority of compounds it can be
concluded that structure-activity-relationships derived
for NMS-occupied receptors also refer to the interaction
with NMS-free receptors.
Resu lts a n d Discu ssion
SAR in th e Ca r a cu r in e V Ser ies. Except caracu-
rine V dioxide (33), all investigated compounds in the
caracurine V series were able to retard the dissociation
of [3H]NMS from porcine cardiac membranes. Since 33
is the only compound with positive charges on nitrogen
atoms fully compensated by the negative charged oxy-
gen atoms, ammonium cations seem to be important for
the receptor-ligand interaction. In the series of 22
bisquaternary caracurinium V salts a wide range of
allosteric potency with EC0.5,diss values from 4 nM to
3270 nM was observed. This indicates that the allosteric
effect on [3H]NMS binding to M2 receptors is very
sensitive to the properties of N-substituents at the
caracurine V ring system. The most potent caracurine
V analogues carry propargyl (12, 4 nM) and allyl (3, 10
nM) substituents. Saturation of the propargyl triple
bonds of 12 to give the dipropyl analogue 16 resulted
in a 8-fold lower binding affinity. Thus, a triple or double
bond is essential for high binding. Increase in bulk,
generated by cyclopropylmethyl (17, 14 nM) and cyclo-
hexenyl (15, 20 nM) substituents, did not affect the high
binding affinity remarkably. Substitution of the cyclo-
propane ring of 17 by an oxirane ring (18, 65 nM)
decreased affinity (4.5-fold). Exchange of ethinyl groups
of 12 with isosteric cyano groups (19) and with acet-
amide moieties (20) caused a dramatic decrease of
allosteric potency (50-fold for 19 and 70-fold for 20).
While extension of propargyl and allyl substituents by
one methyl group to give 13 and 14, respectively, led to
considerably lower binding affinity (11-fold for 13 and
7-fold for 14), reduction of the chain length to only one
methyl group resulted in only marginal change in
allosteric potency (11, 8 nM). The absence of both
N-substituents as given for the caracurine V base 6 (442
nM) produced a dramatic decrease in affinity relative
to the most potent bisquaternary analogues 12 (110-
fold), 3 (44-fold), and 11 (55-fold). In contrast, removal
of only one N-allyl group of 3 to give the monoquater-
nary compound 31 (18 nM) hardly affected the high
allosteric potency. Introduction of longer and more bulky
substituents, such as different substituted benzyl groups,
resulted in a strong reduction of binding affinity. For
example, dibenzylcaracurinium V dibromide (22) (69
nM) is about 7 times less potent than the diallyl
analogue 3. A further increase in bulk generated by
different benzyl substituents such as methoxy (25, 26),
nitro (27), bromo (28), and trifluoromethyl (29) groups,
as well as by an expansion to the naphthyl group (23),
resulted in an 5-25-fold decrease of allosteric potency.
The only exception is the pentafluorobenzyl derivative
24 (54 nM) with similar binding affinity as the benzyl
analogue 22 (69 nM), probably due to a small volume
of the fluorine atoms. The electrical effect of the benzyl
substituent does not seem to be crucial for the receptor-
ligand interaction as indicated by similar binding af-
finities of the nitro (27), bromo (28), and dimethoxy (26)
analogues. The compound with the lowest binding
a
b
iso-Caracurine V ring system. N-Allylcaracurinium V bro-
mide. c Tetramethoxycaracurine V ring system. Tetrahydrocara-
curine V ring system. e nH significantly different from unity (F-
test, p < 0.05).
d
is unchanged despite allosteric agent binding to the
receptor. In this case, the affinity of the test compound
to the free receptor equals its affinity to the [3H]NMS-
occupied receptor. Values greater or lower than 100%
indicate positive or negative cooperativity, respectively,
between the allosteric and the orthosteric ligand. As can
be seen from Table 1, the test compounds except
diphenacylcaracurinium V salt (30) hardly affected [3H]-
NMS equilibrium binding. Thus, the test compounds
appear to have a very similar affinity for free receptors
as compared to NMS-occupied receptors. 30, however,
that is clearly negatively cooperative with NMS, seems

Bisquaternary Salts of Caracurine V
J ournal of Medicinal Chemistry, 2004, Vol. 47, No. 14 3565
affinity in this series is the very long and bulky phenacyl
analogue (30).
Rin g System s. The stereochemistry of caracurine V,
toxiferine I, iso-caracurine V, and tetrahydrocaracurine
V was recently determined by means of NMR spectros-
copy and semiempirical calculations.34 In all four ring
systems, the pharmacophore is fixed in a highly fused
and rigid heterocyclic skeleton. In caracurine V and iso-
caracurine V, the pharmacophoric elements (two posi-
tively charged nitrogens surrounded by two aromatic
systems) adopt a nearly identical spatial arrangment
in terms of the internitrogen distance (dimethylcara-
curinium V dichloride 11: 9.67 Å; dimethyl-iso-cara-
curinium V diiodide 35: 9.54 Å) and of the relative
position of the indole rings. In the bisnortoxiferine ring
system, the N+-N+-distance remained the same (toxi-
ferine 1: 9.68 Å), while the relative spatial position of
the aromatic rings dramatically changed. In dimeth-
yltetrahydrocaracurinium V dichloride (10), the N+-
N+-distance is reduced to 8.80 Å, while the orientation
of the aromatic rings is similar to that observed for 11
and 35. The reduced internitrogen distance and the
absence of both double bonds in 10 has no significant
influence on the allosteric potency as indicated by an
EC0.5,diss value of 10 (4 nM) which is in the same range
as the binding affinities of 11 (8 nM) and 35 (14 nM).
Interestingly, the allosteric potency of toxiferine I (1)
(96 nM) is much lower than those observed for other
N-methyl-substituted compounds suggesting that the
relative orientation of the aromatic rings as given in the
caracurine V and iso-caracurine V ring systems is
required for an optimal ligand-receptor interaction.
This assumption could not be confirmed in the allyl
series. Alcuronium (diallylbisnortoxiferine) 2 (5 nM) is
nearly as potent as the other allyl-substituted caracu-
rine V 3 (10 nM) and iso-caracurine V 36 (3 nM)
compounds. However, because of differences in receptor
epitopes involved in binding of alcuronium (2) and
diallylcaracurinium V (3),12 the bisnortoxiferine and
caracurin V ring systems should not be considered in
the same (Q)SAR analysis.
Introduction of two methoxy goups at both aromatic
rings of diallylcaracurinium V dibromide (3) resulted
in about 20-fold decrease of allosteric potency (32, 204
nM). This is in agreement with our findings observed
in a series of bisquaternary dimers of strychnine and
brucine25 and suggests that the allosteric binding site
does not tolerate additional methoxy groups at the
lateral heterocyclic sides of the caracurine V ring
system.
In summary, alkylation of caracurine V with simple
alkyl groups resulted in a dramatic increase in affinity
at the NMS-liganded M2 receptors. The allosteric bind-
ing site on muscarinic M2 receptors seems to best
accommodate aliphatic N-substituents with a maximal
length corresponding to three carbon atoms. Double or
triple bonds are favorable for best allosteric potency.
Increase in bulk not exceeding the substituent’s length
of three carbon atoms does not affect binding affinity.
Introduction of longer and more bulky substituents
leads to reduced allosteric potency.
Com p a r ison w ith th e SAR in a Qu a ter n a r y
Str ych n in e Ser ies. The affinity and allosteric proper-
ties at the M1-M4 subtypes of muscarinic receptors of
15 quaternary strychnine derivatives with a number of
substituents also used in this study was recently
published.20 However, the SAR of binding affinities at
NMS-liganded M2 receptors found by Gharagozloo et al.
is different from those observed in our bisquaternary
caracurine V series. Strychnine and caracurine V (6),
which in simplest terms can be regarded as a dimeric
strychnine, show similar binding affinities (strych-
nine: 525 nM,25 caracurine V: 442 nM). Whereas
N-alkylation of strychnine with simple alkyl groups such
as methyl, ethyl, propyl, allyl, and propargyl was
reported to result in marginal changes in affinity except
for the allyl substitution (4-fold increase),20 we observed
a dramatic (15-110-fold) increase of binding affinity
after double quaternization of caracurine V with identi-
cal substituents. Moreover, while CH2CN substitution
of strychnine slightly reduced binding, we observed an
about 2-fold increase of allosteric potency after the same
substitution of caracurine V. Finally, N-substitution of
strychnine with a very bulky naphthyl group caused a
6-fold increase in binding affinity. In contrast, the
corresponding dinaphthyl caracurinium V analogue (23)
is 3.5 times less potent than caracurine V itself.
The considerably higher affinity of the most potent
bisquaternary caracurine V analogues relative to the
corresponding strychnine derivatives can be explained
based on the pharmacophore hypothesis consisting of
two positively charged groups surrounded by two aro-
matic ring systems.27 The presence of only two of these
pharmacophoric elements, i.e., one positive charge and
one indole ring, gives rise to limited interactions of the
monoquaternary strychnines with the receptor protein
and, in consequence, to reduced binding affinity. On the
other hand, expansion of the caracurine V ring system
by double alkylation with very bulky substituents
provides large-sized compounds which probably no
longer fit into the receptor binding pocket.
In summary, the three highly potent iso-caracurine
V analogues 35-37, as well as the dimethyltetrahydro-
caracurinium V salt (10), derive affinity from nearly the
same relative spatial arrangement of indole rings as
given for the caracurine V ring system. Reduction of the
internitrogen distance from 9.7 Å (caracurine V salts)
to 8.8 Å (10) did not change the high allosteric potency.
CoMSIA Stu d ies. To supplement and validate the
SARs found so far, and to obtain a model that can be
used for the prediction of unsynthesized compounds, a
CoMSIA41 study was performed (for geometry optimiza-
tion and alignment, see Experimental Section). Since
alcuronium (2) was identified to bind in a different mode
to the allosteric binding site,12 the bisnortoxiferine
analogues 1 and 2 were excluded from the analysis. In
addition to that, 33 was excluded from the data set since
no biological activity could be measured. For outlier
detection leave-one-out cross-validation (LOO-CV) re-
siduals were visually inspected and caracurine V (6) was
identified as outlier.
The results of the CoMSIA analysis are given in Table
2. Overall model quality in terms of the cross-validated
squared multiple correlation coefficient R2CV-1, often
referred to as q2, was good (q2 ) 0.70). To assess the
robustness of the model, a repetitive 5-fold cross-
Com p a r ison of Ca r a cu r in e V, iso-Ca r a cu r in e V,
Bisn or toxifer in e, a n d Tetr a h yd r oca r a cu r in e V

3566 J ournal of Medicinal Chemistry, 2004, Vol. 47, No. 14
Zlotos et al.
Ta ble 2. Results of the CoMSIA Study
b
c
e
f
model
ka
R2
RMSEPCV-k
R2 d
RMSEC
n/nsel
LVg
CV-k
CoMSIA-LOO
CoMSIA-5-fold
1
20%
0.70
0.57h
0.48
0.91
0.91
0.28
0.28
8159/n.a.
8159/n.a.
4
0.67h
3.94h
Abbreviations: LOO: leave-one-out cross-validation; 5-fold: repetitive 5-fold cross-validation; LMO: leave-multiple-out cross-validation.
a
b
k: number of objects left out. R2CV-k: leave-k-out cross-validated coefficient of determination. c RMSEPCV-k: leave-k-out cross-validated
d
root-mean-squared error of prediction. R2: coefficient of determination. e RMSEC: root-mean-squared error of calibration. f n/ nsel: number
g
h
of variables/number of selected variables. LV: number of latent variables. Mean value of 81 repetitions.
validation42 was performed (81 repetitions). The result-
ing mean R2CV-20% of 0.57 (minimum: 0.42, maximum:
0.66) indicates that the model is quite robust since 20%
of the data can be removed from the training data and
can in turn be predicted with sufficient accuracy.
Interpretation of the CoMSIA plots highlighted two
structural features as important for biological activity.
Combined interpretation of the steric and the electro-
static field allows differentiation of the optimal sub-
stituent size at the quaternary nitrogen. Additionally,
the hydrogen-bond acceptor field emphasizes the det-
rimental effect of atoms which are capable of hydrogen-
bonding close to the quaternary nitrogen. The described
characteristics are best explained with the help of
examples.
Figure 1 displays the low active compound 23 (large
N-substituent) and the two highly active compounds 3
and 16 (rather small N-substituents). The yellow con-
tours of the CoMSIA plot highlight regions in space
which, when occupied by a substituent, reduce biological
activity. It can easily be seen that this is the case for
compound 23. When comparing compounds 3 and 16,
it can be concluded that the maximum length of the
substituent lies between these two substructures, since
the biological activity of 3 which is not extending inside
the yellow contour is slightly higher than for 16 which
is extending inside the contour. The nitrogen substitu-
tion is further described by the electrostatic field. Here,
substituents need to be inspected in different regions
(see Figure 2).
On one hand, the partial charge of the carbon atom
directly connected to the quaternary nitrogen should be
negative (red regions). This is the case when no electron-
withdrawing groups (e.g. carbonyl, benzyl) are con-
nected to this carbon atom. On the other hand, electron
rich parts of the substituent should not extend within
the blue colored regions of the CoMSIA plot. If this is
the case (e.g. for benzyl substituents), allosteric potency
F igu r e 1. CoMSIA steric contour map with sterically disfa-
is drastically decreased. One reason for this phenom-
enon could be that the ionic interaction of the quater-
nary nitrogen atom with the receptor is reduced owing
to a compensation of the respective positive charge with
partial negative charges of the substituent. Another
explanation for this observation could be drawn from
the hydrogen-bond acceptor contour map. This map
shows that atoms with hydrogen-bond capabilities (e.g.
the carbonyl oxygen of compound 30, see Figure 2)
which are located close to the quaternary nitrogen are
detrimental for a high biological activity. Since hydrogen-
bond acceptor atoms are usually incorporated in par-
tially negative charged groups, this plot is in accordance
with the electrostatic contour map.
vored regions colored yellow. For reasons of clarity, hydrogen
atoms are not shown. Substituent size of the quaternary
nitrogen is crucial for biological activity. The maximum
substituent size is located between an allyl (3, EC50 ) 10 nM,
top) and a propyl group (16, EC50 ) 30 nM, middle). Com-
pounds with larger substiuents (e.g. 23, EC50 ) 1558 nM,
bottom) show a decreased biological activity.
the oxygen atoms were identified as relevant for binding
affinity (magenta colored regions in the center of Figure
3). One reason for this might be the cluster in activity-
space of compounds with the iso-caracurine skeleton,
i.e., all such compounds (35, 36, 37) included in the
analysis show high allosteric potency. It should be
mentioned that a 2D-QSAR analysis using a topological
SESP descriptor43 obtained a QSAR model of a compa-
rable quality (see Supporting Information for details)
Unexpectedly, the hydrogen-bond acceptor map, did
not highlight differences in the skeleton of the struc-
tures (open ring and closed ring systems), even though

Bisquaternary Salts of Caracurine V
J ournal of Medicinal Chemistry, 2004, Vol. 47, No. 14 3567
allosteric potency. This finding can be explained by the
nearly identical spatial arrangement of the aromatic
indole rings in all three “caracurine V” ring systems.
In contrast, opening of both tetetrahydrooxepine rings
of dimethylcaracurinium V dichloride (11) to give
toxiferine I (1) caused a 10-fold decrease of binding
affinity due to a considerable conformational change of
the whole ring system. The findings suggest that the
relative orientation of the aromatic rings as given in the
“caracurine V” ring systems is required for an optimal
ligand-receptor interaction. The SAR results could be
confirmed by 3D-QSAR (CoMSIA) anylysis. The wide
range of binding affinities observed for all investigated
compounds could be explained by steric and electrostatic
properties of the N-substituents.
All of the agents except for the negatively cooperative
diphenacylcaracurinium V analogue 30 are nearly neu-
traly cooperative with the antagonist [3H]NMS, i.e., they
do not change the [3H]NMS equilibrium binding at the
M2 orthosteric binding site. Thus, any effects of struc-
tural modifications on the allosteric potency are only
due to changes in receptor binding affinity and not in
cooperativity with [3H]NMS. However, one has to keep
in mind that the cooperativity between the allosteric
modulators included in our study and other orthosteric
ligands, especially the endogenous agonist acetylcholine,
can be quite different.
F igu r e 2. CoMSIA electrostatic contour map (hydrogens
suppressed). Regions where partially positive and negative
charge increases biological activity are colored blue and red,
respectively. Electron-withdrawing substituents such as phen-
acyl (30, EC0.5,diss ) 3270 nM) reduce biological activity in two
ways. On one hand they increase the partial positive charge
in the red colored region, and on the other hand they extend
into regions which favor a positive charged group.
Owing to the close structural relationship of caracu-
rine V and iso-caracurine V analogues to the strong
muscle relaxants toxiferine I and alcuronium, all in-
vestigated compounds are likely to exhibit neuromus-
cular-blocking activity, which would limit their useful-
ness as research tools and make their therapeutical use
impossible. However, recent radioligand binding studies
at the muscle type of nicotinic acetylcholine receptors
indicated that from all ring systems under investigation,
the caracurine V ring skeleton is likely to have the
lowest neuromuscular blocking activity.44 Consequently,
reduction of the caracurine V ring skeleton to structural
features responsible for good allosteric potency could
possibly lead to potent allosteric muscarinic compounds
with negligible neuromuscular blocking activity. There-
fore, synthetic work on simpler heterocyclic ring systems
derived from caracurine V is continued in our labora-
tory.45,46
F igu r e 3. CoMSIA hydrogen-bond acceptor contour map
(hydrogens suppressed). Corresponding regions to ligand-
acceptor sites that increase or decrease biological activity are
colored magenta and orange, respectively. Substituents which
possess a hydrogen-bond acceptor close to the orange colored
region show a low biological activity. This observation is in
accordance with the electrostatic map and is found not only
for the displayed compound (30, EC0.5,diss ) 3270 nM) but also
for compounds with smaller substituents (e.g. 18, 19, 20).
Su m m a r y a n d Ou tlook
In this paper we have been able to better define the
pharmacophore for allosteric ligands of muscarinic M2
receptors in terms of the relative spatial arrangement
of the aromatic rings and the properties of the N-
substituents using bisquaternary analogues of the
alkaloids caracurine V, iso-caracurine V, tetrahydro-
caracurine V, and bisnortoxiferine. Double quaterniza-
tion of the Strychnos alkaloid caracurine V with non-
polar alkyl groups of a maximal chain length of three
carbon atoms yielded compounds with up to 100-fold
higher binding affinity for the allosteric binding site of
NMS-occupied muscarinic M2 receptors. Notable com-
pounds are dimethyl (3), diallyl (11), and dipropargyl
(12) caracurine V analogues. These findings are in
contrast to SARs found in the monoquaternary strych-
nine series and confirm the pharmacophore hypothesis
of two positively charged nitrogen atoms in a distance
of about 10 Å surrounded by two aromatic ring systems.
Two chemical modifications of the caracurine V ring
system, i.e., reduction of both double bonds yielding
dimethyltetrahydrocaracurinium V salt (10) and open-
ing of one tetrahydrooxepine ring to give the iso-
caracurinium V salts (35-37), maintained the high
Exp er im en ta l Section
Gen er a l Meth od s. Starting materials were prepared ac-
cording to literature procedures as indicated, and if no refer-
ence is quoted, they are available commercially. Alcuronium
(2) was a generous gift from Roche, Basel. Melting points were
determined with
a Gallenkamp melting point apparatus
(Sanyo) and were not corrected. 1H and 13C NMR spectra were
recorded on Varian XL 300 and Bruker AV 400 instruments.
IR spectra were obtained using a Biorad PharmalyzIR FT-IR
spectrometer. IR and NMR data of all investigated compounds
are given in the Supporting Information.
TLC was carried out on Merck silica gel 60 F254 aluminum
sheets. Mass spectra of the tertiary amines were recorded on
a Finnigan MAT 8200 spectrometer (70 eV). Mass spectra of
the quaternary compounds were run by fast-atom-bombard-
ment on Kratos Concept 1H and Finnigan MAT 90 mass
spectrometers in 3-nitrobenzyl alcohol as matrix. Elemental
analyses were performed by the microanalytical section of the
Institute of Inorganic Chemistry, University of Wu¨rzburg. All
reactions were carried out under an argon atmosphere.
E-22-Oxim in ostr ych n in e (E-4) an d Z-22-Oxim in ostr ych -
n in e (Z-4). tert-Butyl nitrite (80 mL) and tert-BuOK (8 g) were

3568 J ournal of Medicinal Chemistry, 2004, Vol. 47, No. 14
Zlotos et al.
added to the solution of strychnine (10.0 g, 0.03 mol) in dry
toluene (500 mL). The reaction mixture was stirred at 50 °C
for 2 h. After the mixture was cooled to room temp, 10%
aqueous NH4Cl solution (100 mL) was added and vigorous
stirring was continued for 15 min. Water and toluene were
removed in vacuo, the yellow residue was suspended in CHCl3/
MeOH 4:1 (300 mL), and the mixture was stirred for 30 min.
The residue obtained upon filtration and evaporation of CHCl3
and MeOH was purified on a silica gel column, eluting with
CHCl3/MeOH/25%NH3 130:10:1 to afford two oximes.
Evaporation of the solvent afforded NMR pure colorless
crystals of 10 in quantitative yield, mp > 300 °C (dec); [R]22
D
-117° (c 0.46, DMSO); MS-FAB m/e 655 and 653 (M2+Cl-),
617 (M2+ - H+). Anal. (C40H50N4O2Cl2 H2O) C, H, N.
Gen er a l Dou ble Qu a ter n iza tion P r oced u r e of Ca r a -
cu r in e V (6), iso-Ca r a cu r in e V (34), a n d Tetr a m eth oxy-
ca r a cu r in e V (6a ). The solution of the respective halide (0.85
mmol) in chloroform (5 mL) was added to a solution of the
alkaloid (100 mg, 0.17 mmol) in chloroform (5 mL). After being
stirred at room temperature for 30 min, the crystallized
ammonium salt was isolated by filtration. If no crystallization
occurred, the product was precipitated by adding diethyl ether.
The collected ammonium salt was washed with chloroform or
with a chloroform/diethyl ether mixture 1:1 and dried in a
vacuum at 80 °C. No further purification was necessary as
indicated by TLC (silica gel, mobile phase MeOH/2 M NH3/2
M aqueous NH4NO3 84:24:12) and 1H NMR spectra.
Melting points of all caracurinium V and iso-caracurinium
V salts are higher than 300 °C (dec).
(Z-4) (1.9 g, 17%), yellow crystals; mp > 320 °C (dec); Rf )
0.44 (CHCl3/MeOH/25% NH3 130:10:1); [R]22 -471° (c 0.57,
D
DMSO); MS m/e 363 (M+, 100), 346 (87), 318 (57). Anal.
(C21H21N3O3) C, H, N.
(E-4) (6.8 g, 62%), yellow crystals; mp > 320 °C (dec); Rf )
0.22 (CHCl3/MeOH/25% NH3 130:10:1); [R]22 -265° (c 0.58,
D
DMSO); MS m/e 363 (M+, 100), 346 (90), 318 (38).
Gen er a l P r oced u r e for th e Syn th esis of Wiela n d -
Gu m lich Ald eh yd e (5) a n d 10,11-Dim eth oxy-Wiela n d -
Gu m lich Ald eh yd e (5a ). The appropriate oxime was added
to thionyl chloride (100 mL) under ice-cooling within 20 min,
and the mixture was stirred at room temperature for 2.5 h.
Thionyl chloride was removed under reduced pressure at 40
°C, and the residue was treated with 2 M HCl (200 mL)
(ca u tion !, HCN evolu tion ). The resulting mixture was
stirred for 3 h at 100 °C. The cold solution was made alkaline
with 25% ammonia and extracted with CHCl3/n-butanol 10:1
(5 × 100 mL). The combined organic layers were washed with
water (50 mL), dried over MgSO4, and evaporated to give a
brown foam which was chromatographed on neutral alumina
containing 10% H2O, eluting with CHCl3/toluene 5:1.
Wiela n d -Gu m lich Ald eh yd e (5). 5 was prepared from
the reaction of E-4 (6 g, 16.5 mmol) according to the general
procedure. Yield 4.1 g (80.0%); mp 215 °C (Lit.36 208-210 °C).
5 exists in DMSO solution as a mixture of 17-(S) and 17-(R)
epimers.
4,4′-Dia llylca r a cu r in iu m V Dibr om id e (3). Obtained
from 6 and allyl bromide (102 mg) as a white solid; yield 130
mg (92%); [R]22D +95° (c 0.54, DMSO); FAB-MS m/e 747.2 and
745.2 (M2+Br-), 665.3 (M2+ - H+), 625.1. Anal. (C44H50N4O2-
Br2) C, H, N.
4,4′-Dim eth ylca r a cu r in iu m V Dich lor id e (11).34 A solu-
tion of the precipitated iodide in water was passed through
“Amberlite” IRA-400 resin, chloride phase and water was
evaporated under reduced pressure; yield 80 mg (68%); FAB-
MS m/e 651.3 and 649.3 (M2+Cl-), 613.2 (M2+ - H+), 556.1.
C
40H46N4O2Cl2.
4,4′-Dip r op a r gylca r a cu r in iu m V Dibr om id e (12). Ob-
tained from 6 and propargyl bromide (101 mg) as a white solid;
yield 135 mg (96%); [R]22 +47° (c 0.44, DMSO); FAB-MS m/e
D
743.2 and 741.2 (M2+Br-), 661.2 (M2+ - H+), 623.3. Anal.
(C44H46N4O2Br2) C, H, N.
4,4′-Bis(2-bu tyn -1-yl)ca r a cu r in iu m V Dibr om id e (13).
Obtained from 6 and 1-bromo-2-butyne (113 mg) as a white
10,11-Dim eth oxy-Wiela n d -Gu m lich Ald eh yd e (5a ). 5a
was prepared from the reaction of E-4a 25(6 g, 14 mmol)
according to the general procedure. Yield 0.9 g (17.0%); mp
210 °C (dec); MS m/e 370, (M+,100), 204 (54), 180 (53). Anal.
(C21H26N2O4) C, H, N.
solid; yield 115 mg (79%); [R]22 +73° (c 0.55, DMSO); FAB-
D
MS m/z 771.2 and 769.2 (M2+Br-), 689.3 (M2+ - H+), 637.3.
Anal. (C46H50N4O2Br2), C, H, N.
4,4’-Bis(tr a n s-2-bu ten -1-yl)ca r a cu r in iu m V Dibr om id e
(14). Obtained from 6 and trans-1-bromo-2-butene (115 mg)
10,11,10′,11′-Tetr a m eth oxyca r a cu r in e V (6a ). 5a (600
mg, 1.6 mmol) was heated for 14 h with 5 g of pivalic acid at
105 °C in an evacuated sealed tube. Water (50 mL) was added,
and the cool reaction mixture was made basic with 25%
ammonia. Extraction with CHCl3, drying over MgSO4, and
evaporation of the extract gave a brown foam which was
chromatographed on neutral alumina containing 10% H2O,
eluting with CHCl3/toluene 3:1 to give 6a as colorless crystal-
line solid (105 mg, 18%). mp > 320 °C (dec); MS m/e 704 (M+,
45), 703 (M+-1, 100), 685 (27). Anal. (C42H48N4O6) C, H, N.
17-(R)-Cya n o-19,20-d id eh yd r o-11,12-d im eth oxy-17,18-
ep oxycu r a n e (7). E-4a (8.5 g, 20 mmol) was added to thionyl
chloride (100 mL) under ice-cooling within 20 min, and the
mixture was stirred at room temperature for 2.5 h. Thionyl
chloride was removed under reduced pressure at 40 °C, and
water (100 mL) was added, followed by 1 M Na2CO3 until pH
3. Steam was passed through the reaction mixture for 1.5 h.
(ca u tion !, HCN evolu tion ). The cold solution was made
alkaline with 25% ammonia and extracted with CHCl3 (3 ×
100 mL). The combined organic layers were washed with water
(50 mL), dried over MgSO4, and evaporated to give a brown
foam which was chromatographed on neutral alumina con-
taining 10% H2O, eluting with CHCl3/toluene 1:1 to give 7 as
colorless crystalline solid (1.7 g, 23%). mp 188 °C; [R]22D -197°
(c 0.79, CHCl3); MS m/e 379 (M+, 100) (11), 364 (42), 204 (38),
189 (82), HRMS (C22H25N3O3) found m/e 379.1900, calcd
379.1896. Anal. (C22H25N3O3) C, H, N.
as a white solid; yield 110 mg (75%); [R]22 +91° (c 0.30,
D
DMSO); FAB-MS m/e 775.4 and 773.4 (M2+Br-), 693.5 (M2+
-
H+), 639.3. Anal. (C46H54N4O2Br2), C, H, N.
4,4′-Bis(cycloh exen e-3-yl)ca r a cu r in iu m V Dibr om id e
(15). The mixture of three diastereomers was obtained from
6 and 3-bromo-1-cyclohexene (137 mg) as a white solid; yield
140 mg (90%); [R]22D +37° (c 0.58, DMSO); FAB-MS m/e 827.4
and 825.4 (M2+Br-), 745.4 (M2+ - H+), 665.4. Anal. (C50H58N4O2-
Br2) C, H, N.
4,4′-Bis(cyclop r op ylm et h yl)ca r a cu r in iu m V Dib r o-
m id e (17). Obtained from 6 and (bromomethyl)cyclopropane
(115 mg) as a white solid; yield 80 mg (55%); [R]22 +53° (c
D
0.62, DMSO); FAB-MS m/e 775.3 and 773.3 (M2+Br-), 693.4
(M2+ - H+), 655.3. Anal. (C46H54N4O2Br2) C, H, N.
4,4′-Bis(1-oxir a n ylm eth yl)ca r a cu r in iu m V Dibr om id e
(18). The mixture of three diastereomers was obtained from
6 and epibromohydrine (117 mg) as a white solid; yield 70 mg
(48%); [R]22 +63° (c 0.45, DMSO); FAB-MS m/e 779.4 and
D
777.4 (M2+Br-), 697.4 (M2+ - H+), 641.3. Anal. (C44H50N4O4-
Br2) C, H, N.
4,4′-Bis(cya n om eth yl)ca r a cu r in iu m V Dibr om id e (19).
Obtained from 6 and bromoacetonitrile (117 mg) as a white
solid; yield 120 mg (85%); [R]22 +69° (c 0.56, DMSO); FAB-
D
MS m/e 745.3 and 743.3 4 (M2+Br-), 663.3 (M2+ - H+). Anal.
(C42H44N6O2Br2) C, H, N.
4,4′-Bis(ca r ba m oylm eth yl)ca r a cu r in iu m V Dibr om id e
(20). Obtained from 6 and 2-bromoacetamide (117 mg) as a
4,4′-Dim et h yl-(20S,20’S)-19,20,19’,20’-t et r a h yd r oca r a -
cu r in iu m V Dich lor id e (10). PtO2 (15 mg) was added to an
aqueous solution (20 mL) of 11 (75 mg, 0.1 mmol). The reaction
mixture was hydrogenated under the hydrogen pressure of 50
bar for 24 h. The catalyst was removed by filtration through
Celite, and the filter pad was washed with water (2 × 10 mL).
white solid; yield 80 mg (54%); [R]22 +70° (c 0.45, DMSO);
D
FAB-MS m/e 781.3 and 779.3 (M2+ - Br-), 699.3 (M2+ - H+).
Anal. (C42H48N6O4Br2) H, N, C: calcd, 58.61; found, 58.10.
4,4′-Diben zylca r a cu r in iu m V Dibr om id e (22). Obtained
from 6 and benzyl bromide (145 mg) as a white solid; yield

Bisquaternary Salts of Caracurine V
J ournal of Medicinal Chemistry, 2004, Vol. 47, No. 14 3569
140 mg (88%); [R]22D +82° (c 0.51, DMSO); FAB-MS m/e 847.3
and 845.3 (M2+Br-), 765.4 (M2+ - H+), 675.4. Anal. (C52H54N4O2-
Br2) C, H, N.
yield 97 mg (81%); [R]22 +39° (c 0.56, DMSO); FAB-MS m/e
D
625.2 (M+-Br-) Anal. (C41H45N4O2Br) C, H, N.
Gen er a l P r oced u r e for Syn th esis of 4, 4′-Dip r op yl-
ca r a cu r in iu m V Dich lor id e (16) a n d 4,4′-Bis[3-(1,3-d ioxo-
1,3-dih ydr o-isoin dol-2-yl)-pr opyl]car acu r in iu m V Dich lo-
r id e (21). 6 (100 mg, 0.17 mmol) was added to the solution of
propyl iodide (146 mg, 0.86 mmol) and N-(3-iodopropyl)-
phthalimide (268 mg, 0.86 mmol),48 respectively, in acetone
and CHCl3, respectively. After being heated under reflux for
30 min, the crystallized ammonium salts were isolated by
filtration and washed with acetone and CHCl3, respectively.
Aqueous solutions of the iodides were passed through “Am-
berlite” IRA-400 resin, chloride phase, and water was evapo-
rated under reduced pressure. No further purification was
necessary as indicated by TLC (silica gel, mobile phase
4,4′-Bis(2-n a p h th yl)ca r a cu r in iu m V Dibr om id e (23).
Obtained from 6 and 2-naphthyl bromide (188 mg) as a white
solid; yield 140 mg (80%); [R]22 +52° (c 0.50, DMSO); FAB-
D
MS m/e 947.4 and 945.4 (M2+Br-), 865.4 (M2+ - H+). Anal.
(C60H58N4O2Br2) C, H, N.
4,4′-Bis(p en ta flu or oben zyl)ca r a cu r in iu m V Dibr om id e
(24). Obtained from 6 and pentafluorobenzyl bromide (222 mg)
as a white solid; yield 110 mg (58%); FAB-MS m/e 1027.9 and
1025.9 (M2+Br-), 945.9 (M2+ - H+). Anal. (C52H44N4O2Br2F10
C, H, N.
)
4,4′-Bis(4-m eth oxyben zyl)ca r a cu r in iu m V Dich lor id e
(25). Obtained from 6 and 4-methoxybenzyl chloride (133 mg)
1
as a white solid; yield 115 mg (75%); [R]22 +99° (c 0.11,
MeOH/2 M NH3/2 M aqueous NH4NO3 84:24:12) and H NMR
D
DMSO); FAB-MS m/e 864.3 and 862.3 (M2+Cl-), 826.4 (M2+
-
spectra.
H+), 705.3. Anal. (C54H58N4O4Cl2) C, H, N.
16: Obtained as a white solid; yield 84 mg (65%);[R]22
D
+129° (c 0.26, DMSO); FAB-MS m/e 707.4 and 705.4 (M2+Cl-),
4,4′-Bis(3,4-d im eth oxyben zyl)ca r a cu r in iu m V Dich lo-
r id e (26). Obtained from 6 and 3,4-dimethoxybenzyl chloride47
669.4 (M2+ - H+). Anal. (C44H54N4O2Cl2 H2O) C, H, N.
(160 mg) as a white solid; yield 130 mg (79%); [R]22 +89° (c
21: Obtained as a white solid; yield 115 mg (64%); [R]22
D
D
0.56, DMSO); FAB-MS m/e 924.2 and 922.2 (M2+Cl-), 886.2
+39° (c 0.56, DMSO); FAB-MS m/e 997.4 and 995.4 (M2+Cl-)
(M2+ - H+). Anal. (C56H62N4O6Cl2) C, H, N.
959.3 (M2+ - H+). Anal. (C60H60N6O6Cl2 H2O) C, H, N.
4,4′-Bis(4-n itr oben zyl)ca r a cu r in iu m V Dibr om id e (27).
Obtained from 6 and 4-nitrobenzyl bromide (184 mg) as a
P h a r m a cology. Porcine cardiac homogenates were pre-
pared as described previously.11 Binding of [3H]N-methylsco-
polamine ([3H]NMS) (0.2 nM; specific activity 70-83.5 Ci/
mmol; Perkin-Elmer Life Sciences, Boston, MA) was measured
in a buffer composed of 4 mM Na2HPO4 and 1 mM KH2PO4
(pH 7.4) at 23 °C. Nonspecific [3H]NMS binding was deter-
mined in the presence of 10-6 M atropine and was less than
10% of the total binding. Membranes were separated by rapid
vacuum filtration through glass fiber filters (Schleicher and
Schu¨ll, No. 6; Dassel, Germany). Filters were washed twice
with 5 mL of ice-cold distilled water and placed into scintil-
lation vials containing 5 mL of scintillation fluid (Ready
protein, Beckman, Palo Alto, CA), and membrane bound
radioactivity was determined by liquid scintillation counting
in a Beckman LS 6000 counter. The pKD of NMS binding
amounted to 9.52 ( 0.04 (mean ( SEM, n ) 3).
Two types of experiments were applied to evaluate the effect
of the test compounds on the dissociation for [3H]NMS. In the
case of “complete dissociation experiments” cardiac membranes
were preincubated with 0.2 nM [3H]NMS for 30 min; radio-
ligand dissociation was then revealed by the addition of 1 µM
atropine in the presence or absence of allosteric test compound.
The time course of dissociation was observed by withdrawing
aliquots at appropriate times over a period of 120 min. Data
were analyzed by means of nonlinear regression analysis. [3H]-
NMS dissociation proceeded monophasically in the absence
and in the presence of test compound (t1/2,control ) 4.35 ( 0.09
min; mean ( SE, n ) 52) and was characterized by the
apparent rate constant of dissociation k-1. In “two-point kinetic
experiments”49 membranes were incubated with [3H]NMS for
30 min to obtain binding equilibrium. Specific [3H]NMS
binding was measured before (t ) 0 min) and after (t ) 10
min) the addition of 1 µM atropine alone or in combination
with a test compound. Specific [3H]NMS binding at t ) 0 min
and t ) 10 min served to characterize the monoexponential
time course of [3H]NMS dissociation and to obtain the apparent
rate constant of dissociation k-1. To generate concentration-
effect curves for the allosteric retardation of radioligand
dissociation, the apparent rate constant of dissociation k-1 was
expressed as a percentage of the value under control condi-
tions. Curve fitting was based on a four parameter logistic
function (GraphPad Prism Version 3.02; GraphPad, San Diego,
CA).
yellow solid; yield 170 mg (96%); [R]22 +54° (c 0.58, DMSO);
D
FAB-MS m/e, 937.1 and 935.1 (M2+Br-), 855.4 (M2+ - H+).
Anal. (C54H52N6O6Br2) C, H, N.
4,4′-Bis(4-br om oben zyl)car acu r in iu m V Dibr om ide (28).
Obtained from 6 and 4-bromobenzyl bromide (213 mg) as a
white solid; yield 145 mg (78%); [R]22 +63° (c 0.55, DMSO);
D
FAB-MS m/e 1007.1, 1005.1, 1003.1 and 1001.1 (M2+Br-), 921.2
and 923.2 (M2+ - H+). Anal. (C52H52N4O2Br4) C, H, N.
4,4′-Bis-[4-(t r iflu or om et h yl)-b en zyl]-ca r a cu r in iu m V
Dibr om id e (29). Obtained from 6 and 4-(trifluoromethyl)-
benzyl bromide (203 mg) as a white solid; yield 155 mg (85%);
[R]22 +55° (c 0.63, DMSO); FAB-MS m/e 984.3 and 982.3
D
(M2+Br-) 902.3 (M2+ - H+). Anal. (C54H52N4O2Br2F6) C, H, N.
4,4′-Dip h en a cylca r a cu r in iu m V Dibr om id e (30). Ob-
tained from 6 and phenacyl bromide (170 mg) as a white solid;
yield 140 mg (81%); [R]22 +88° (c 0.18, DMSO); FAB-MS m/e
D
936.3 and 934.3 (M2+Br-) 854.3 (M2+ - H+). Anal. (C54H54N4O6-
Br2) H, N, C: calcd, 63.91; found, 63.46.
4,4′-Dim eth yl-iso-ca r a cu r in iu m V Diiod id e (35). Ob-
tained from 34 and methyl iodide (120 mg) as a white solid;
yield 130 mg (88%); FAB-MS m/e 741.1 (M2+I-), 613.2 (M2+
-
H+). Anal.(C40H46N4O2I2) C, H, N.
4,4′-Dia llyl-iso-ca r a cu r in iu m V Dibr om id e (36). Ob-
tained from 34 and allylbromide (103 mg) as a white solid;
yield 125 mg (88%); [R]22 +77° (c 0.34, DMSO); FAB-MS m/e
D
747.1 and 745.1 (M2+Br-), 665.2 (M2+ - H+), 625.2. Anal.
(C44H50N4O2Br2) C, H, N.
4,4′-Bis(cycloh exen e-3-yl)-iso-ca r a cu r in iu m V Dibr o-
m id e (37). The mixture of three diastereomers was obtained
from 34 and 3-bromocyclohexene as a white solid; yield 115
mg (74%); [R]22 +101° (c 0.65, DMSO); FAB-MS m/e 827.3
D
and 825.3 (M2+Br-), 745.3 (M2+ - H+), 665.3. Anal. (C50H58N4O2-
Br2) C, H, N.
4, 4′-Bisa llyl-(10,11,10′,11′-tetr a m eth oxyca r a cu r in iu m
V Dibr om id e (32). Obtained from 6a (60 mg, 0.09 mmol) and
allylbromide (54 mg, 0.45 mmol) as a white solid; yield 60 mg
(85%); [R]22 +52° (c 0.53, DMSO); FAB-MS m/e 868.2 and
D
866.1 (M2+Br-), 786.3 (M2+ - H+), 745.3. Anal. (C44H50N4O2-
Br2) C, H, N.
4-Allylca r a cu r in iu m V Br om id e (31). The solution of
diallyl bromide (20 mg, 0.17 mmol) in chloroform (5 mL) was
added to a solution of 6 (500 mg, 0.86 mmol) in chloroform (5
mL). After being stirred at room temperature for 30 min, the
product was precipitated by adding diethyl ether. The collected
ammonium salt was washed with a chloroform/diethyl ether
mixture 1:1 and dried in a vacuum at 80 °C. No further
purification was necessary as indicated by TLC (silica gel,
mobile phase MeOH/2 M NH3/2 M aqueous NH4NO3 84:24:
12) and 1H NMR spectrum. 31 was obtainedas a white solid;
Equilibrium binding expermints were conducted using 0.2
nM [3H]NMS. The effect of the test compounds on [3H]NMS
equilibrium binding was investigated at concentrations of
allosteric modulator that were shown to effectively reduce the
rate of [3H]NMS dissociation to 25% of the control. The
appropriate time of incubation was determined according to
Lazareno and Birdsall40 (eq 31 therein). Five half-lives were
taken to ensure equilibrium binding conditions. Equilibrium

3570 J ournal of Medicinal Chemistry, 2004, Vol. 47, No. 14
Zlotos et al.
(6) Christopoulos, A.; Kenakin, T. G-Protein-coupled receptor al-
losterism and complexing. Pharmacol. Rev. 2002, 54, 323-374.
(7) Christopoulos, A. Allosteric binding sites on cell-surface recep-
tors: novel targets for drug discovery. Nat. Rev. Drug Discuss.
2002, 1, 198-210.
(8) Ellis, J .; Huyler, J .; Brann, M. R. Allosteric regulations of cloned
m1-m5 muscarinic receptor subtypes. Biochem. Pharmacol.
1991, 42, 1927.
(9) Tra¨nkle, C.; Mies-Klomfass, E.; Botero Cid, M. H.; Holzgrabe,
U.; Mohr, K. Identification of a [3H]ligand for the common
allosteric site of muscarinic acetylcholine M2 receptors. Mol.
Pharmacol. 1998, 54, 139-145.
(10) Ellis, J .; Seidenberg, M. Two allosteric modulators interact at a
common site on cardiac muscarinic receptors. Mol. Pharmacol.
1992, 42, 638-641.
binding data (from two to five independent experiments with
quadruplicated values) in the presence of allosteric modulator
were expressed in percent relative to the control binding in
the absence of modulator, which was set as 100%.
Molecu la r Mod elin g. Geom et r y Op t im iza t ion a n d
Align m en t. The three different ring skeletons of 6, 10, and
34
34 were selected as template structures for the alignment
procedure necessary for CoMSIA studies. As a first step, the
structures were aligned using SYBYL’s multifit procedure.50
For this alignment, the torsional angle between atoms C12-
C13-N-C17′ found in the AM1 calculations34 were con-
strained to their current value. Then the corresponding
quaternary nitrogen atoms and the centroids of the aromatic
rings C8-C13 and C8′-C13′ were aligned with spring con-
stants of 20 and 10, respectively. The resulting alignment was
very good. The remaining compounds of the data set were built
based on these template structures. For the latter step, the
backbone of the template structures was kept fixed, and only
the N-substituents were geometry optimized. For all optimiza-
tion steps, the Tripos force field with a Powell minimizer was
applied for 300 iterations. All other options were set to their
respective default values. Compounds with unsubstituted
nitrogen were ionized. After the alignment step, AM1 charges
were calculated for all structures (MOPAC keywords: 1SCF
AM1 MMOK).
3D-QSAR. CoMSIA fields were generated using SYBYL.50
The CoMSIA region was automatically defined and all avail-
able fields (steric, electrostatic, acceptor, donor, and hydro-
phobic) were calculated. Coefficient contour maps showing the
product of standard deviation and coefficient (“StdDev*Coeff”)
were used for interpretation purposes. To identify appropriate
contour levels, field value histograms for each feature were
analyzed, and levels that gave meaningful results were ap-
plied. It should be noted that contour maps showing steric and
electrostatic fields highlight those regions of the ligand which
enhance or decrease activity. In contrast to that, CoMSIA plots
for hydrogen bonding properties highlight areas where hydro-
gen-bond donor and hydrogen-bond acceptor features could be
located within a hypothetical receptor site.
(11) Tra¨nkle, C.; Mohr, K. Divergent modes of action among cationic
allosteric modulators of muscarinic M2 receptors. Mol. Pharma-
col. 1997, 51, 674-682.
(12) Buller, S.; Zlotos, D. P.; Mohr, K.; Ellis, J . Allosteric site on
muscarinic acetylcholine receptors:
a single amino acid in
transmembrane region 7 is critical to the subtype selectivities
of caracurine V derivatives and alkane-bisammonium ligands.
Mol. Pharmacol. 2002, 61, 160-168.
(13) Tra¨nkle, C.; Kostenis, E.; Burgmer, U.; Mohr, K. Search for lead
structures to develop new allosteric modulators of muscarinic
receptors. J . Pharmacol. Exp. Ther. 1996, 279, 926-933.
(14) Bejeuhr, G.; Blaschke, G.; Holzgrabe, U.; Mohr, K.; Su¨rig, U.;
Terfloth, G. Molecular modelling and synthesis of potent stabi-
lizers of antagonist binding to M2-cholinoceptors. J . Pharm.
Pharmacol. 1994, 46, 108-112.
(15) Botero Cid, M. H.; Holzgrabe, U.; Kostenis, E.; Mohr, K.; Tra¨nkle,
C. Search for the pharmacophore of bispyridinium-type allosteric
modulators of muscarinic receptors. J . Med. Chem. 1994, 37,
1439-1445.
(16) Botero Cid, H. M.; Tra¨nkle, C.; Baumann, K.; Pick, R.; Mies-
Klomfass, E.; Kostenis E.; Mohr, K.; Holzgrabe, U. Structure-
activity relationships in a series of bisquaternary phthalimidine
derivatives modulating the muscarinic M2-receptor allosterically.
J . Med. Chem. 2000, 43, 2155-2164.
(17) Raasch, A.; Scharfenstein, O.; Tra¨nkle, C.; Holzgrabe, U.; Mohr,
K. Elevation of ligand binding to muscarinic M2 acetylcholine
receptors by bis(ammonio)alkane-type allosteric modulators. J .
Med. Chem. 2002, 45, 3809-3812.
(18) Muth, M.; Bender, W.; Scharfenstein, O.; Holzgrabe, U.; Balat-
kova, E.; Tra¨nkle, C.; Mohr, K. Systematic development of high
affinity bis(ammonio)alkane-type allosteric enhancers of mus-
carinic ligand binding. J . Med. Chem. 2003, 46, 1031-1040.
(19) Nassif-Makki, T.; Tra¨nkle, C.; Zlotos, D.; Bejeuhr, G.; Cambareri,
A.; Pfletschinger, C.; Kostenis, E.; Mohr, K.; Holzgrabe, U.
Bisquaternary ligands of the common allosteric site of M2
acetylcholine receptors: optimization of the distances between
the pharmacophoric elements. J . Med. Chem. 1999, 42, 849-
858.
(20) Gharagozloo, P.; Lazareno, S.; Popham, A.; Birdsall, N. J . M.
Allosteric interactions of quaternary strychnine and brucine
derivatives with muscarinic acetylcholine receptors. J . Med.
Chem. 1999, 42, 438-445.
(21) Urbansky´, M.; Prosˇka, J .; Ricˇny, J .; Drasˇar, P. Truxillic acid
derivatives, neuromuscular blocking agents with very high
affinity for the allosteric binding site of muscarinic acetylcholine
receptors. Collect. Czech, Chem. Commun. 1999, 64, 1980-1992.
(22) Lysikova´, M.; Fuksova´, K.; Elbert, T.; J akubik, J .; Tucˇek, S.
Subtype-selective inhibition of [methyl-3H]-N-methylscopola-
mine binding to muscarinic receptors by alpha-truxillic acid
esters. Br. J . Pharmacol. 1999, 127, 1240-1246.
(23) Urbansky´, M.; Prosˇka, J .; Drasˇar, P. On steroids. CDII. Fused
thiazoloandrostanes and their quaternary salts, synthesis and
cooperative ligand binding to muscarinic acetylcholine receptor.
Collect. Czech, Chem. Commun. 1999, 64, 1457-1470.
(24) Gharagozloo, P.; Lazareno, S.; Miyauchi, M.; Popham, A.;
Birdsall, N. J . M. Substituted pentacyclic carbazoles as novel
muscarinic allosteric agents: synthesis and structure-affinity
and cooperativity relationships. J . Med. Chem. 2002, 45, 1259-
1274.
Only the steric, electrostatic, and hydrogen-bond acceptor
CoMSIA field were applied in this study since preliminary
knowledge gained from the SAR study motivated their use (see
Results and Discussion). The first two fields were thought to
be useful in identifying the patterns of the N-substitution,
whereas the hydrogen-bond acceptor field was thought to be
useful for identifying the different skeletons (open and closed
ring systems).
Ack n ow led gm en t. Thanks are due to the Deutsche
Forschungsgemeinschaft (MO 821/1) and the Fonds der
Chemischen Industrie, Deutschland, for financial sup-
port, as well as to Mechthild Kepe, Institute of Phar-
macy, University of Bonn, for her skillful technical
assistance.
Su p p or tin g In for m a tion Ava ila ble: IR and 1H and 13C
NMR data of all compounds and a 2D-QSAR model. This
material is available free of charge via the Internet at http://
pubs.acs.org.
Refer en ces
(1) Stockton, J . M.; Birdsall, N. J . M.; Burgen, A. S. V.; Hulme, E.
C. Modification of the binding properties of the muscarinic
receptors by gallamine. Mol. Pharmacol. 1983, 23, 551-557.
(2) Lee, N. H.; El-Fakahany, E. E. Allosteric antagonists of the
muscarinic acetylcholine receptor Biochem. Pharmacol. 1991, 42,
199-205.
(25) Zlotos, D. P.; Buller, S.; Holzgrabe, U.; Mohr, K. Bisquaternary
dimers of strychnine and brucine. A new class of potent enhanc-
ers of antagonist binding to muscarinic M2 receptors. Bioorg.
Med. Chem. 2003, 11, 2627-2634.
(26) Holzgrabe, U.; Wagener, M.; Gasteiger, J . Comparison of struc-
turally different allosteric modulators of muscarinic receptors
by self-organizing neural networks. J . Mol. Graphics 1996, 14,
185-193.
(27) Holzgrabe, U.; Hopfinger, A. J . Conformational analysis, mo-
lecular shape comparison, and pharmacophore identification of
different allosteric modulators of muscarinic receptors. J . Chem.
Inf. Comput. Sci. 1996, 36, 1018-1024.
(3) Ellis, J . Allosteric binding site on muscarinic receptors. Drug
Dev. Res. 1997, 40, 193-204.
(4) Christopoulos, A.; Lanzafame, A.; Mitchelson, F. Allosteric
interactions at muscarinic cholinoceptors. Clin. Exp. Pharmacol.
Phys. 1998, 25, 185-194.
(5) Holzgrabe, U.; Mohr, K. Allosteric modulation of antagonist
binding to muscarinic receptors. Drugs Discovery Today 1998,
3, 214-222.

Bisquaternary Salts of Caracurine V
J ournal of Medicinal Chemistry, 2004, Vol. 47, No. 14 3571
(28) Stiefl, N.; Baumann, K. Mapping property distributions of
molecular surfaces: algorithm and evaluation of a novel 3D
quantitative structure-activity relationship technique. J . Med.
Chem. 2003, 46, 1390-1407.
(29) Tucˇek, S.; Mus´ılkova´, J .; Nedoma, J .; Prosˇka, J .; Shelkovnikov,
S.; Vorl´ıcˇek, J . Positive cooperativity in the binding of alcuro-
nium and N-methylscopolamine to muscarinic acetylcholine
receptors J . Mol. Pharmacol. 1990, 38, 674-680.
(30) Zlotos, D. P.; Buller, S.; Tra¨nkle, C., Mohr, K. Bisquaternary
caracurine V derivatives as allosteric modulators of ligand
binding to M2 acetylcholine receptors. Bioorg. Med. Chem. Lett.
2000, 10, 2529-2532.
(31) Voigtlander, U.; J o¨hren, K.; Mohr, M.; Raasch, A.; Tra¨nkle, C.;
Buller, S.; Ellis, J .; Ho¨ltje, H.-D.; Mohr, K. Allosteric site on
muscarinic acetylcholine receptors: identification of two amino
acids in the muscarinic M2 receptor that account entirely for The
M2/M5 subtype selectivities of some structurally diverse allosteric
ligands in N-methylscopolamine-occupied receptors Mol. Phar-
macol. 2003, 64 (1) 21-31.
(32) Zlotos, D. P. Stereochemistry of caracurine V. J . Nat. Prod. 2000,
63, 864-865.
(33) Zlotos, D. P. iso-Caracurine V, a novel unexpected decomposition
product of caracurine V. J . Nat. Prod. 2003, 66, 119-120.
(34) Zlotos, D. P. Stereochemistry of caracurine V, iso-caracurine V,
bisnortoxiferine and tetrahydrocaracurine V ring systems. Eur.
J . Org. Chem. 2004, 11, 2375-2380.
dolichothyrsa Gilg ex Onochie et Hepper. J . Pharm. Sci. 1978,
67, 171-174.
(40) Lazareno, S.; Birdsall, N. J . M. Detection, quantitation, and
verification of allosteric interactions of agents with labeled and
unlabeled ligands at G protein-coupled receptors: interactions
of strychnine and acetylcholine at muscarinic receptors Mol.
Pharmacol. 1995, 48, 362-378.
(41) Klebe, G.; Abraham, U,: Mietzner, T. Molecular similarity
indices in a comparative analysis (CoMSIA) of drug molecules
to correlate and predict their biological activity. J . Med. Chem.
1994, 37, 4130-4146.
(42) Baumann, K.; Albert, H.; von Korff, M. A systematic evaluation
of the benefits and hazards of variable selection in latent variable
regression. Part I: search algorithm, theory and simulations.
J . Chemom. 2002, 16, 339-350.
(43) Baumann, K. An alignment-independent versatile structure
descriptor for QSAR and QSPR based on the distribution of
molecular features. J . Chem. Inf. Comput. Sci. 2002, 42, 26-
35.
(44) Zlotos, D. P.; Gu¨ndisch, D.; Ferraro, S.; Tilotta, B. Unpublished
results.
(45) Sripha, K., Zlotos D. P. 6,7,14,15-Tetrahydro[1,5]diazocino[1,2-
a:6,5-a′]diindole. Synthesis of a novel pentacyclic ring system.
Tetrahedron 2003, 59, 391-394.
(46) Sripha, K.; Zlotos, D. P.; Buller, S.; Mohr, K. 6,7,14,15-Tetrahy-
dro-15aH-azocino[1,2-a:6,5-b′]diindole. Synthesis of a novel pen-
tacyclic ring system. Tetrahedron Lett. 2003, 44, 7183-7186.
(47) Zlotos D. P., Meise W.: Synthesis and conformational analysis
of hexahydroisoquino-[3,2-b][3]benzazepines (iso-B-homober-
bines). Heterocycles 1997, 45, 2137-2157.
(35) Battersby, A. R.; Hodson, H. F. Alkaloids of Calabash Curare
and Strychnos species. Part I. Chemistry and structure of
hemitoxiferine-I and toxiferineI, including the preparation of
toxiferine-I. J . Chem. Soc. 1960, 736-739.
(48) Ollivier, C.; Renaud, P. A novel approach to formation of carbon-
nitrogen bonds: azidation of alkyl radicals with sulfonyl azides.
J . Am. Chem. Soc. 2001, 123, 4717-4727.
(36) Hymon, J . R.; Schmid, H.; Karrer; P.; Boller, A.; Els, H.; Fahrni,
P.; Fu¨rst, A. Die Chemie des Wieland-Gumlich-Aldehyds und
seiner Derivate. Helv. Chim. Acta 1969, 52, 1564-1602.
(37) J ackmann, J . M.; Sternhall, S. In Application of nuclear NMR
spectroscopy in organic chemistry; Pergamon Press Inc.: New
York, 1969; p 226.
(49) Kostenis, E.; Mohr, K. Two-point kinetic experiments to quantify
allosteric effects on radioligand dissociation Trends Pharmacol.
Sci. 1996, 17, 280-283.
(50) SYBYL 6.8 Tripos Inc., 1699 South Hanley Rd., St. Louis, MO
63144, 2000.
(38) Kalinowski, H. O.; Berger, S.; Braun, S. In 13C NMR-Spektrosko-
pie; Georg Thieme Verlag: Stuttgart, New York, 1984; p 220.
(39) Verpoorte, R.; Baerheim Svendsen, A. Alkaloids of Strychnos
J M0311341