4-oxystilbene compounds are selective ligands for neuronal nicotinic αBungarotoxin receptors

Article abstract of DOI:10.1038/sj.bjp.0701957

1. Starting from the structure of an old 4-oxystilbene derivate with ganglioplegic activity (MG624), we synthesized two further derivates (F2 and F3) and two stereoisomers of F3 (F3A and F3B), and studied their selective effect on neuronal nicotinic acetylcholine receptor (AChR) subtypes. 2. MG 624, F3, F3A and F3B inhibited of ¹²⁵-αBungarotoxin (αBgtx) binding to neuronal chick optic lobe (COL) membranes, with nM affinity, but inhibited ¹²⁵I-αBgtx binding to TE671 cell-expressed muscle-type AChR only at much higher concentrations. 3. We immobilized the α7, β2 and β4 containing chick neuronal nicotinic AChR subtypes using anti-subunit specific antibodies. MG 624, F3, F3A and F3B inhibited ¹²⁵I-αBgtx binding to the α7-containing receptors with nM affinity, but inhibited ³H-Epi binding to β2-containing receptors only at very high concentrations (more than 35 μM); their affinity for the β4-containing receptors was ten times more than for the β2-containing subtype. 4. Both MG624 and F3 compounds inhibited the ACh evoked currents in homomeric oocyte-expressed chick α7 receptors with an IC₅₀ of respectively 94 and 119 nM. 5. High doses of both MG 624 and F3 depressed the contractile response to vagus nerve stimulation in guinea pig nerve-stomach preparations although at different IC₅₀s (49.4 vs 166.2 μM) The effect of MG624 on rat nerve-hemidiaphragm preparations was 33 times less potent than that of F3 (IC₅₀ 486 vs 14.5 μM). 6. In conclusion, MG624 and F3 have a high degree of antagonist selectivity for neuronal nicotinic αBgtx receptors containing the α7 subunit.

Full text of DOI:10.1038/sj.bjp.0701957

British Journal of Phar
m
acology (1998) 124, 1197 ± 1206
1998 Stockton Press All rights reserved 0007 ± 1188/98 $12.00
http://www.stockton-press.co.uk/bjp
4-oxystilbene compounds are selective ligands for neuronal nicotinic
aBungarotoxin receptors
1,2,6C. Gotti, 1,2B. Balestra, 1,2M. Moretti, 3G.E. Rovati, 5L. Maggi, 3G. Rossoni, 2F. Berti, 4L. Villa,
4M. Pallavicini & 1,2F. Clementi
2
1CNR Cellular and Molecular Pharmacology Center, Milan, Italy; Department of Medical Pharmacology, University of Milan,
Milan, Italy; 3Institute of Pharmacological Sciences, Faculty of Pharmacy, University of Milan, Italy; 4Institute of Pharmaceutical
Chemistry and Toxicology, University of Milan, Italy; Biophysical Laboratory, Istituto Regina Elena, Experimental Research
Centre, Rome, Italy
5
1 Starting from the structure of an old 4-oxystilbene derivate with ganglioplegic activity (MG624), we
synthesized two further derivates (F2 and F3) and two stereoisomers of F3 (F3A and F3B), and studied
their selective e€ect on neuronal nicotinic acetylcholine receptor (AChR) subtypes.
MG 624, F3, F3A and F3B inhibited of 125I-aBungarotoxin (aBgtx) binding to neuronal chick optic
2
lobe (COL) membranes, with nM anity, but inhibited 125I-aBgtx binding to TE671 cell-expressed
muscle-type AChR only at much higher concentrations.
3
We immobilized the a7, b2 and b4 containing chick neuronal nicotinic AChR subtypes using anti-
subunit speci®c antibodies. MG 624, F3, F3A and F3B inhibited 125I-aBgtx binding to the a7-containing
receptors with nM anity, but inhibited 3H-Epi binding to b2-containing receptors only at very high
concentrations (more than 35 mM); their anity for the b4-containing receptors was ten times more than
for the b2-containing subtype.
4
Both MG624 and F3 compounds inhibited the ACh evoked currents in homomeric oocyte-expressed
chick a7 receptors with an IC50 of respectively 94 and 119 nM.
High doses of both MG 624 and F3 depressed the contractile response to vagus nerve stimulation in
5
guinea pig nerve-stomach preparations although at di€erent IC50s (49.4 vs 166.2 mM) The e€ect of
MG624 on rat nerve-hemidiaphragm preparations was 33 times less potent than that of F3 (IC50 486 vs
14.5 mM).
6
In conclusion, MG624 and F3 have a high degree of antagonist selectivity for neuronal nicotinic
aBgtx receptors containing the a7 subunit.
Keywords: 4-oxystilbene; neuronal nicotinic receptors; subtypes; aBungarotoxin receptors; aBungarotoxin; Epibatidine; chick
optic lobe
Introduction
Neuronal nicotinic acetylcholine receptors (AChRs) are widely
distributed in the peripheral and central nervous systems. They
are known to be involved in the control of ganglionic
transmission and complex brain processes such as attention,
memory, locomotor activity, nociception, temperature control
and excitability; furthermore, they are also involved in a
number of brain pathologies and in smoking addiction
(reviewed in Dani & Heinemann, 1996; Role & Berg, 1996;
Gotti et al., 1997a). They are localised in both postsynaptic
structures, where they are responsible for nicotinic transmis-
sion, as well as in presynaptic boutons, where they modulate
the release of a number of neurotransmitters (reviewed in
Wonnaccott, 1997).
Neuronal nicotinic AChRs are a family of ACh-gated
cation channels, which have a pentameric structure that is
similar to that of muscle AChRs. Eight alpha (a2 ± a9) and
three beta (b2 ± b4) subunits coding for neuronal nicotinic
AChR subunits have been cloned in vertebrates. Expression
studies in heterologous systems have demonstrated that at least
two classes of nicotinic receptors can be generated: one by only
expressing a single type of a subunit, as in the case of the a7, a8
and a9 subtypes (homomeric receptors), and the other by
means of the pairwise combination of one kind a subunit other
than that forming the homomeric receptors and one kind of b
subunit (heteromeric receptors). Binding studies have also
con®rmed the presence of the two classes: the ®rst includes the
receptors containing either a7 or a8 subunits, or both, and bind
aBungarotoxin (aBgtx) with high anity ; the second includes
the receptors that bind nicotinic agonists with very high anity
(pM-nM) but do not bind aBgtx. The various subtypes have
di€erent pharmacological and electrophysiological properties
(reviewed in Sargent, 1993; Papke, 1993; McGehee & Role,
1995; Gotti et al., 1997a).
The role and function of most of these neuronal nicotinic
AChR subtypes are dicult to establish because individual
neurons can coexpress various subunit combinations. Recent
immunoprecipitation and immunolocalization experiments
have demonstrated that the subunit composition of native
receptors is more complex than was ®rst thought, with
certain subtypes being made up of as many as three or four
di€erent subunits (Conroy
Kobrin, 1997).
& Berg, 1995; Forsayeth &
The most abundant subtype found in vertebrate brain is the
a4b2 subtype, whereas little or no a4 subunit is present in
autonomic neurons, whose major subtype contains the a3 and
b4 subunits. The a7-containing subtype accounts for most of
the high anity aBgtx binding sites present in both the central
6 Author for correspondence at: CNR Cellular and Molecular
Pharmacology Center, Via Vanvitelli 32, 20129 Milano, Italy.

1198
C. Gotti et al
Selective ligands for neuronal aBungarotoxin receptors
Methods
and peripheral nervous systems (reviewed in McGehee & Role,
1995).
One of the goals of recent research in the nicotinic ®eld
has therefore been to ®nd substances that speci®cally
recognise di€erent receptor subtypes in order to help in the
characterisation of neuronal nicotinic AChRs and, possibly,
in the development of drugs that may be useful in the
treatment of some of the pathologies in which neuronal
nicotinic AChRs are involved. A number of old and new
compounds have been investigated in both binding and
functional studies, but few of them show selective speci®city
for neuronal nicotinic AChR subtypes (recently reviewed by
Gotti et al., 1997a).
Recent advances in the search for novel neuronal nicotinic
AChRs ligands include the discovery of 2-(3-pyridyloxy-
methyl) azacyclic compounds I that have subnanomolar
anity for rat brain [3H]-(7)-cytisine binding sites (Abreo et
al., 1996), as well as their analogues II, in which the pyridine
ring has been bioisosterically replaced by a substituted aryl
moiety (Elliott et al., 1996) (Figure 1). The Ki values of the
latter compounds for rat brain [3H]-(7)-cytisine binding sites
ranges from 3 to 410,000 nM depending on the aromatic
substitution. Electron-withdrawing groups at the meta position
improve Ki values, whereas substitution at the orto and para
positions is generally unfavourable.
Chemical synthesis
The synthesis of N,N,N-triethyl-2-(4-trans-stilbenoxy)ethy-
lammonium iodide 1 (MG 624) was described by Cavallini
et al. (1953). N,N,N-trimethyl-2-(4-trans-stilbenoxy)propy-
lammonium iodide 2 (F2) and N,N,N-trimethyl-1-(4-trans-
stilbenoxy)-2-propylammonium iodide 3 (F3) were prepared
by treating the corresponding N,N-dimethyl derivatives with
iodomethane. A mixture of F2 and F3 was previously
obtained from trans-4-hydroxylstilbene and 2-dimethylami-
noisopropyl chloride and then resolved by means of
chromatography before conversion to the quaternary
ammonium salts 2 and 3.
Racemic
N,N-dimethyl-1-(4-trans-stilbenoxy)-2-propyla-
mine, the immediate precursor of 3, was subjected to optical
resolution with dibenzoyltartaric acid to yield the two
enantiomeric tertiary amines, that were subsequently trans-
formed into the quaternary ammonium salts (S)- and (R)- 3
(F3A and F3B) (see Figure 2).
Preparation of chick optic lobe and retina membranes
and extracts
On the basis of these results and the previously known
e€ects of 4-oxystilbene derivatives on ganglionic neuronal
nicotinic AChRs (Mantegazza & Tommasini, 1955), we
decided to explore compounds 1-3, (S)-3 and (R)-3 (Figure
2) as potentially subtype-selective neuronal nicotinic AChR
ligands.
Chick optic lobes (COLs) and retinas were dissected from 1-
day old chickens, immediately frozen in liquid nitrogen and
stored at 7808C for later use. No di€erences in the binding
properties of fresh and frozen tissues were observed (Gotti et
al., 1994, 1997b). For every experiment, the two types of
membranes were separately homogenized in an excess of
50 mM Na phosphate pH 7.4, 1 M NaCl, 2 mM EDTA, 2 mM
EGTA and 2 mM PMFS for 2 min in an ultraTurrax
homogenizer. The homogenate was then diluted and
centrifuged for 1 h at 60,000 g. This procedure of homo-
genization, dilution and centrifugation was done three times,
after which the pellets were collected, rapidly rinsed with
50 mM Na phosphate, 50 mM NaCl, 2 mM EDTA, 2 mM
EGTA, 2 mM PMFS and then resuspended in the same bu€er
containing a mixture of 5 mg/ml of each of the following
protease inhibitors: leupeptin, bestatin, pepstatin A and
aprotinin (Sigma). Triton X-100 at a ®nal concentration of
2% was added to the washed membranes and the membranes
were extracted for 2 h at 48C.The extract was then
centrifuged for 1.5 h at 60,000 g, and the clear supernatant
recovered.
Figure 1 Chemical structure of 2-(3-pyridyloxymethyl) azacyclic
compounds (I) (taken from Abreo et al., 1996) and their analogues
(II) (taken from Elliott et al., 1996).
TE671 cell membranes
TE671 cells obtained from the American Type Culture
Collection were grown at con¯uency for ten days in petri
dishes in RPMI medium containing 10% fetal calf serum,
2 mM glutamine, and 100 mg/ml penicillin and streptomycin
The cells were detached using phosphate bu€ered saline (PBS)
containing 1 mM PMFS, washed twice by centrifugation,
resuspended in 50 mM Tris HCl pH 7.4 containing 50 mM Na
Cl 2 mM EDTA, 2 mM EGTA mM and then homogenized. A
5 mg/ml mixture of the protease inhibitors leupeptin, bestatin,
pepstatin A, aprotinin and 1 mM PMFS was added to the
membrane homogenate.
Peptide synthesis and polyclonal antibody production
The peptides and antibodies (Abs) were produced as
previously described (Gotti et al., 1994). Anti-a7 and anti-b2
chick-speci®c antibodies (Abs) were produced in rabbits by
Figure 2 Chemical structure of 4-oxystilbene compounds.

C. Gotti et al
Selective ligands for neuronal aBungarotoxin receptors
1199
means of immunization with peptides chosen from the most
divergent region of the a7 and b2 subunits (the cytoplasmic
loop between M3 and M4) and against peptides located at the
COOH terminal for the b2 and b4 subunits.
The speci®city of the polyclonal Abs was tested in ELISA
for possible crossreactivity with other peptides obtained from
AChR subunits, as well as in immunoprecipitation experi-
ments and on Western blots of puri®ed receptors, and the Abs
were found to be speci®c for the subtypes containing the
corresponding subunit. The Abs raised against the a7, b2 and
b4 peptides were puri®ed on an anity column made by
coupling the corresponding peptide to CNBr-activated
Sepharose 4B (Pharmacia) according to the manufacturer's
instructions.
to block possible proteolysis during the long incubation time
of the assays.
Non-speci®c binding (averaging 25 ± 30% of total binding)
was determined in parallel by means of incubation in the
presence of 2 mM unlabeled aBgtx. At the end of the
incubation, the samples were ®ltered on GFC ®lters presoaked
in PBS +1% BSA through a Brandell-apparatus and the ®lters
counted in a g counter.
3
Saturation experiments with H-Epi were performed over-
3
night by incubating aliquots of COL membranes with H-Epi
concentrations ranging from 0.005 to 5 nM at 208C.
Non-speci®c binding (averaging 10 ± 15% of total binding)
was determined in parallel by means of incubation in the
presence of 100 nM ± 1 mM unlabeled Epi.
Receptor immobilization by subunit-speci®c antibodies
Immobilized subtypes The subtypes immobilized by the
corresponding subunit-speci®c Abs were incubated overnight
3
The anity-puri®ed Abs were bound to the wells by overnight
incubation at 48C at a concentration of 10 mg/ml in phosphate
bu€er pH 7.5. On the following day, the wells were washed in
order to remove the excess of unbound Abs, and then
incubated overnight at 48C with 200 ml of 2% Triton X-100
extract obtained from COL membranes in the case of the a7
and b2-containing receptors, and from chick retina extract in
the case of the b4-containing receptors. After overnight
incubation with the extract, the wells were washed and the
presence of immobilized receptor was revealed by means of
at 208C with 300 ml of H-Epi ranging from 0.005 ± 5 nM, or
with 300 ml 125I-aBgtx ranging from 0.01 ± 20 nM. All the
incubations were performed in a bu€er A plus 2 mg/ml BSA
and 0.05% Tween 20. Speci®c labeled ligand binding was
de®ned as total binding minus the binding in the presence of
3
100 nM cold Epi or 2 mM aBgtx. The inhibition of H-Epi and
125I-aBgtx binding to the immobilized subtypes by the tested
compounds, was measured by preincubating the indicated
concentrations of compounds for 30 min at RT, followed by
overnight incubation with 0.05 nM 3H-Epi or 0.25 nM 125I-
aBgtx at 208C.
3
125I-aBgtx or H-Epi binding.
The subtypes immobilized by the corresponding subunit-
speci®c Abs were incubated overnight at 208C with 300 ml of
Receptor subtype immunopuri®cation
3
3H-Epi or 125I-aBgtx. The binding experiments of H-Epi and
The chick optic lobe (COL) and retina extracts were prepared
as described above. The a7-containing receptors were
immunopuri®ed as previously described (Gotti et al., 1994).
The b2-containing receptors were puri®ed from the COL
extract and, in order to remove the b4-containing receptors,
the extract (20 ml for each round) was ®rst incubated with
5 ml of Sepharose-4B with bound anti-b4 Abs (1 mg/ml of
puri®ed Abs) and then the ¯owthrough of this column was
incubated twice with Sepharose 4B with bound anti-b2 Abs.
The b4-containing receptors were puri®ed from chick retina
extract, which was ®rst incubated with 5 ml of Sepharose-4B
and bound anti-b4 Abs (1 mg/ml of puri®ed Abs). The bound
receptors were eluted with 0.2 M glycine (pH 2.2) or with the
peptides used for the immunization.
125I-aBgtx were performed according to a mixed protocol
combining both saturation (the ®rst 7 ± 8 concentrations of
respectively 0.005 ± 5 nM and 0.01 ± 20 nM) and displacement
curves (the last 3 ± 4 concentrations of 50 ± 1000 nM for both
ligands) (Rovati et al., 1991).
By e€ectively combining both saturation and competition
protocols in a single curve, high ligand concentrations can be
reached without using excessive amounts of labeled ligand (the
competition part of the curve), while retaining adequate
radioactivity in the lower concentration range (the saturation
part of the curve). To study the interaction of the heterologous
3
ligands, 0.05 nM of H-Epi or 0.25 nM 125I-aBgtx were used in
displacement studies: the non-speci®c binding was respectively
10 ± 15% and 25 ± 30% of the total binding for 3H-Epi and 125I-
aBgtx, (calculated by LIGAND as one of the unknown
parameters of the model).
After incubation, the wells were washed seven times with
ice-cold PBS containing 0.05% Tween 20 and the bound
radioactivity recovered by incubation with 200 ml of 2 N
NaOH for 2 h. The bound radioactivity was then
determined by means of liquid scintillation counting in a b
counter for 3H-Epi, or directly counted in a g counter for
125I-aBgtx.
After elution of receptors with either glycine or peptides, the
receptors were desalted and concentrated on a Centricon 10
microconcentrator (Amicon).
The recovery at each immunoprecipitation step was
3
determined by means of 125I-aBgtx or H-Epi binding and by
quantitative immunoprecipitation of the receptors present in
the solution before and after each immunopuri®cation as
previously described (Gotti et al., 1994).
3H-Epibatidine and 125I-aBungarotoxin binding
Data analysis The experimental data obtained from the
saturation binding experiments to membranes or immunoim-
mobilized subtypes were analyzed by means of a non-linear least
square procedure using the LIGAND program as described by
Munson and Rodbard (1980). The calculated binding para-
meters were obtained by simultaneously ®tting two independent
experiments for the membranes and three independent
experiments. An `extra sum of squares' F-test was performed
by the LIGAND program to evaluate the di€erent binding
models statistically (i.e., one site vs two site models, comparison
of the binding parameters, etc.) (Munson & Rodbard, 1980).
Membranes All of the membranes used for the binding
assays were previously washed by centrifugation with a
bu€er containing 50 mM Tris-HCl pH 7, 150 mM NaCl,
5 mM KCl, 1 mM MgCl2, 2.5 mM CaCl2 (bu€er A).
Saturation experiments were performed by incubating
aliquots of COL, and TE671 membranes with 125I-aBgtx at
48C for 24 and 48 h respectively in the same bu€er A plus
2 mg/ml BSA. To the incubation mixture a ®nal concentra-
tion of 5 mg/ml of the protease inhibitors leupeptin, bestatin,
pepstatin A, aprotinin and 2 mM PMFS was added in order

1200
C. Gotti et al
Selective ligands for neuronal aBungarotoxin receptors
The Ki values of all of the tested drugs were determined by
means of LIGAND, using the data obtained from 3 ± 4
independent experiments.
The equilibrium ligand binding data were analysed by
means of the LIGAND computer program, (Munson &
Rodbard, 1980). Models of increasing complexity were
considered. The selection of the best ®tting model and
evaluation of the statistical signi®cance of the parameter
di€erence was based on the F-test for the extra sum of square
principle (Draper & Smith, 1966).
of 0.1 Hz, and a voltage three times greater than that necessary
to produce maximal twitches. Direct stimulation was elicited
using stimuli of the same type and frequency delivered via a
®eld electrode. The contraction of the muscle was recorded
using an isometric transducer (HP-FTA 100-1).
The data for both preparations were obtained in seven
separate experiments. For both tissues, the IC50 and 95%
con®dence limits were calculated from the dose-response
curves of MG 624 and F3.
A statistical level of signi®cance of P50.05 was accepted.
The calculated binding parameters were obtained by simulta-
neously ®tting three independent experiments (six curves, three
homologous and three heterologous curves). Binding is
expressed as the ratio of the bound concentration over total
ligand concentration (B/T) vs the logarithm of total
concentration. The total concentration is the sum of `hot' and
`cold' ligand. All of the curves shown were computer
generated.
Results
Antibody speci®city
Since the validity of our pharmacological experiments on the
immobilized subtypes is completely dependent on the Abs used
to immunoimmobilize the subtypes, special care was taken to
characterize the speci®city of the Abs used by means of
immunoprecipitation and Western blotting experiments.
Oocyte injection and electrophysiological recordings
Anti-a7 Abs When tested by immunoprecipitation, the anti
a7-Abs (mean+sem; n=3) were capable of immunoprecipitat-
ing 95% of the 125IaBgtx receptors present in the COL extract:
70+2% of receptors contain the a7 subunit and 25+0.8%
contain both the a7 and a8 subunit, as already reported by
Gotti et al. (1994).
The cDNA encoding the chick neuronal nicotinic a7 subunit
was kindly provided by Dr Marc Ballivet. Full-length cDNA
encoding wild type a7 was expressed as previously described
(Palma et al., 1996). Stage VI oocytes were injected
intranuclearly (10 nl) with cDNA clones (0.2 mg/ml) using a
pressure micro injector (Eppendorf, Germany)
Membrane currents were recorded 2 ± 4 days after injection,
using a voltage-clamp technique with two microelectrodes
®lled with 3 M KCl. The oocytes were placed in a recording
chamber (volume, 0.1 ml) continuously perfused with oocyte
Ringer's medium (82.5 mM NaCl/2.5 mM KCl/2.5 mM CaCl2/
1 mM MgCl2/5 mM HEPES/adjusted to pH 7.4 with NaOH) at
controlled room temperature (20 ± 228C), usually in the
presence of atropine (0.5 mM) (Miledi et al., 1989).
In order to construct dose/response relationships, the
oocyte membrane potential was held at 760 mV and the
drugs were applied at 3 min intervals.
To estimate the half-inhibitory concentration (IC50) of MG
624 and F3, the data were ®tted to Hill equations using least-
square routines (included in Sigma Plot, Jandel, Germany):
I/Imax=IC50 nH/([MG 624 or F3]nH+IC50 nH) (1) where
[MG 624] is the drug dose, nH is the Hill coecient, and Imax
the maximum response. For more details, see Miledi et al.
(1989) and Palma et al. (1996).
Since COL extract contains a large number of receptors that
3
bind H-Epi with high anity, we tested our anti-a7 Abs for
their capacity to immunoprecipitate the high anity H-Epi
3
labeled receptors, and found that they could not immunopre-
3
cipitate any of the H-Epi labeled receptors. The anti-a7 Abs
were used to immunopurify the a7-containing receptors from
COL extract and to study the subunit composition of these
receptors. Figure 3 (left) shows the immunopuri®ed a7
receptors run on SDS ± PAGE probed with the anti-a7, anti-
a8, anti-b2 and anti-b4 Abs. In this immunopuri®ed a7
containing receptor, both anti-a7 and anti-a8 Abs were capable
of recognizing a single band of Mr 57 KD (albeit at di€erent
intensities) whereas the anti-b2 and anti-b4 Abs could not
recognize any peptide.
The recognition by the anti-a8 Ab is not due to cross-
reactivity, but to the presence of the a7-a8 subtype that
contains both the a7 and a8 subunits (Gotti et al., 1994,
1997b).
Anti-b2 Abs The anti-b2 Abs, directed against the cytoplas-
mic peptide, as well as those directed against the COOH
peptide, were capable of immunoprecipitating 95+2% of the
3H Epi high anity labeled receptors. The amounts of receptor
immunoprecipitated by these polyclonal Abs were almost
identical to those immunoprecipitated by mAb 270, which
speci®cally recognizes the b2 subunit (Whiting et al., 1987)
and, in our hands, immunoprecipitated 90% of the 3H-Epi
high anity site.
In order to identify the other subunits present in the COL
extract together with the b2 subunit, we performed immuno-
precipitation experiments using subunit speci®c mAbs. To this
end, we used anti-a3 (mAb 313), anti-a4 (mAb 299), anti-a5
(mAb 35), and anti-b4 Abs, and found that they respectively
immunoprecipitated 15+2, 70+4, 28+2 and 10+1% of the
3H-Epi high anity binding sites. The results of these
immunprecipitation experiments together with those of
immunodepletion experiments not reported here, suggest that
the b2-containing receptors represent an heterogeneous
population of receptor subtypes containing the a3b2b4, a3b2,
In vitro preparations
To reveal the selectivity and functional activity of the
compounds on ganglionic and muscle type AChRs, we studied
their e€ects on guinea pig vagus nerve-stomach and rat phrenic
nerve hemidiaphragm preparations. The rats and guinea pigs
were killed and exsanguinated under pentobarbital anaesthe-
sia. The guinea pig stomachs and attached vagus nerves were
removed and mounted in a 100 ml organ bath according to the
method described by Paton and Vane (1963). Using insulated
platinum ring electrodes, the nerves were stimulated with
rectangular 0.2 msec pulses at a frequency of 20 Hz and a
voltage of 0.2 V for a period of 10 s every 2 min. The
contractions of the stomach were recorded by means of a
pressure transducer (HP267BC).
The isolated left phrenic nerve-hemidiaphragm was pre-
pared as described by Bulbring (1946). Indirect stimulation of
the hemidiaphragm was elicited via a bipolar silver wire
electrode on the phrenic nerve using 1 ms pulses at a frequency

C. Gotti et al
Selective ligands for neuronal aBungarotoxin receptors
1201
Table 1 Comparison of MG 624, F2, F3, F3A, and F3B
binding anities at the COL and TE671 membrane bound
receptors
Tissue:
Ki
Col
mM
TE671
mM
MG 624
F2
F3
F3A
F3B
0.055 (14)
26.5 (33)
0.109 (29)
0.092 (46)
0.094 (39)
70 (20)
196 (16)
77 (19)
59 (22)
38.2 (21)
125I-aBgtx anity (Kd)
2 nM (27)
0.5 nM (20)
Figure 3 Western blot analysis of the immunoisolated receptor
subtypes. (Left) a7 containing receptors puri®ed from chick optic
lobe extract using anti-a7 Abs, (Center) b2 containing receptors
puri®ed from chick optic lobe extract using anti-b2 Abs and (Right)
b4 containing receptors puri®ed from retina extract using anti-b4
Abs. The immunopuri®ed receptors were separated on 9%
acrylamide SDS gels, electrotransferred to nitrocellulose and probed
with the indicated subunit-speci®c Abs or mAbs used at concentra-
tions of 5 ± 10 mg/ml. The anti a7, a8, b2 and b4 Abs were those
produced by us and described in Methods section, whereas the anti
a3(mAb 313), anti a4(mAb 299) and anti-a5 (mAb35) were the Abs
commercially available. The bound Abs were revealed by means of
125I-Protein A. When mAbs were used, the blots were incubated for a
further 2 h with anti-rat IgG diluted 1 : 1000. The molecular weight
markers (top to bottom) are 97 kD, 66 kD, 45 kD, 31 kD and
21 kD.
The Kd and Ki values were derived from 125I-aBgtx
saturation and competition binding curves to the COL and
TE671 membranes. The curves obtained from two separate
experiments were ®tted using a non-linear least squares
analysis program and the F test (Munson and Rodboard).
The numbers in brackets represent the % of CV.
polyclonal Abs raised against the peptides of the a7, b2 and
b4 subunits speci®cally recognize the native receptors contain-
ing those subunits, both in immunoprecipitation and after
SDS ± PAGE.
Ligand binding to COL and TE671 membranes
a4a5b2 or a4b2 subunits; in this population, the pure a4b2
subtype certainly represents 65% of the 3H-Epi-binding, the a4
a5b2 represents 20+4% and the other subtypes together
represent no more than 15%.
The binding of 125I-aBgtx to COL membranes has a Kd value
of 2 nM [coecient of variation (CV)=27%] and a Bmax of
362+35 fmol/mg of protein. MG 624, F2 and F3, and the two
stereoisomers of F3 (F3A and F3B) competitively displaced
Our immunoprecipitation experiments using anti-subunit
speci®c Abs detected the presence of a small amount of b4
containing receptors (10+3%) in COL extract. In order to
check the real speci®city of our anti-b2 Abs, we ®rst
immunodepleted the COL extract of the b4 containing
receptors and then used the anti-b2 Abs to immunopurify the
b2-containing receptor. The immunopuri®ed receptor was then
probed with the anti-a3 (mAb 313), anti-a4 (mAb 299), anti-a5
(mAb35), anti-a7, anti-b2 and anti-b4 Abs. The results are
shown in Figure 3 (middle). mAb 313, mAb 299, mAb35 and
anti-b2 Abs recognized only a single peptide of respectively Mr
57+1, Mr 66, Mr 52+1 and 54+1 kD; the anti-a7 and anti-b4
Abs did not recognize any peptide.
125I-aBgtx binding in
a concentration-dependent manner
although with di€erent potencies; the Ki values reported in
Table 1, which were obtained from the competition experi-
ments, were respectively: 0.055, 26.5, 0.109, 0.092 and
0.094 mM.
The least potent of the ®ve compounds was F2 (481 and 243
times less potent than MG 624 and F3 respectively); the two
F3 enantiomers had very similar potencies, thus indicating
their lack of stereoselectivity.
In addition to the receptor class that binds aBgtx with high
anity, chick brain membranes also express another class of
nicotinic receptors that do not bind aBgtx but bind nicotinic
agonists with high anity. This is composed of several
subtypes (a4b2, a4b2a5, a3b2) and is labeled with high anity
by the newly discovered nicotinic agonist Epi (Gerzanich et al.,
1995).
Anti-b4 Abs The anti-b4 Abs in chick retina extract were
tested in 3 separate experiments and found to immunopreci-
pitate 95+2% of the 3H-Epi high anity binding, whereas
mAb 313, mAb 299, mAb 35 and anti-b2 Abs respectively
immunoprecipitated 66+2, 55+3, 17+1,5 and 40+0.8%,
thus suggesting the presence of several neuronal nicotinic
AChR subtypes also in this tissue, although those containing
both the a3 and b4 subunits represent the large majority
(&70%).
Using anti-b4 Abs, we immunopuri®ed the b4-containing
chick retina receptors, ran them on SDS ± PAGE and probed
them with subunit-speci®c mAbs and Abs. The results are
shown in Figure 3 (right). mAb 313, mAb 299, mAb35, anti-b2
and anti-b4 recognized a single peptide of respectively 57, 66,
52, 54 and 54 kD, whereas the anti-a7 Abs did not recognize
any peptide.
3
We used H-Epi as a ligand of this class of receptors and
found that it binds to chick optic lobe membranes with a Kd of
64 pM (CV=25%) and a Bmax of 86+12 fmol/mg of protein.
When tested in competition experiments using compound
concentrations, ranging from 1 nM to 50 mM, none of the ®ve
3
compounds could inhibit H-Epi binding to COL membrane.
In order to obtain a complete pharmacological pro®le of
these compounds, we also tested them on membranes obtained
from TE671 cells, which are known to express a muscle type
nicotinic AChR. We found that 125I-aBgtx binds to this
receptor with a Kd of 0.52 nM (CV=20%) and a Bmax of
133+10 fmol/mg of protein; all ®ve compounds inhibited
binding only at very high concentrations, with Ki values
respectively of 70, 196, 77, 59 and 38.2 mM, and with a very
similar potencies (see Table 1).
The molecular masses of the a3, a5, a7, a8, b2 and b4
subunits, determined by Western blot corresponded to the
expected sizes deduced from their cDNA sequences, whereas
the molecular mass of the a4 subunit was slightly lower.
In conclusion, on the basis of the results of our previous
(Gotti et al., 1994, 1997b) and present experiments, the
Ligands binding to immobilized subtypes
Previous experiments (Gotti et al., 1994) as well as those
reported above for the anti-a7 and anti-b2 Abs, showed that

1202
C. Gotti et al
Selective ligands for neuronal aBungarotoxin receptors
anti-a7 Abs immunoprecipitate 90% of the 125I-aBgtx binding
sites and that anti-b2 Abs immunoprecipitate more then 90%
3
of the high anity H-Epi sites present in COL extract. This
indicates that the binding of 125I-aBgtx is predominantly to an
a7-containing subtype, whereas that of the high anity 3H-Epi
is to a b2-containing subtype. In the competition experiments
performed on COL membranes, we found that MG 624 and
F3 were selective in inhibiting 125I-aBgtx binding, whereas they
did not have any e€ect on the inhibition of 3H-Epi. In order to
identify the subtype target of these compounds, we decided to
test them on subtypes speci®cally immuno-immobilized by the
corresponding Abs. Furthermore, since these compounds are
reported to have ganglioplegic e€ects, we also decided to test
them on the a3b4 containing subtype, a ganglionic type of
nicotinic receptor which is present in high concentrations in the
chick retina.
Using subunit-speci®c antibodies, we immobilized the a7-
and b2-containing receptors present in the Triton X-100
extract obtained from COL, and the b4-containing receptors
present in the Triton X-100 extract obtained from chick retina.
The receptors immobilized on the anti-a7 speci®c Abs have a
Kd for 125I-aBgtx of 0.5 nM (CV=27%). The Ki of the
receptors for the compounds are shown in Table 2. Figure 4
shows the inhibition curves of MG 624, F2 and F3 on the
binding of 125I-aBgtx. The most potent compound in inhibiting
125I-aBgtx binding to the a7 subtype was MG 624: the order of
potency of MG 6244F34F2.
Figure
4
Binding of 125I-aBgtx to immobilized a7-containing
receptors and its inhibition by MG 624, F2, F3, F3A and F3B. The
binding of 125I-aBgtx to immobilized a7-containing receptors and its
inhibition by MG 624, F2 F3, F3A and F3B were performed according
to
a mixed protocol combining both saturation (the ®rst 7 ± 8
concentrations of respectively 0.01 ± 20 nM) and displacement curves
(the last 3 ± 4 concentrations of 50 ± 2000 nM). The a7-containing
receptors (immunoimmobilized as described in Methods section) were
incubated for 30 min at 208C with the indicated compound
concentrations. 125I-aBgtx (at a ®nal concentration of 0.25 nM) was
then added and left overnight. The curves were obtained by ®tting the
data from three separate experiments each performed in triplicate using
the LIGAND programme. The dotted lines represent the 95% c.l.: for
the sake of clarity, these are only shown for the 125I-aBgtx curve.
The immobilized b2-containing receptors gave a Kd for 3H-
Epi of 36 pM (CV=37%). The Ki values of the compounds are
reported in Table 2. Figure 5 shows the inhibition curves of
MG 624, F2 and F3 on the binding of 3H-Epi to the b2-
containing subtypes. Con®rming the results obtained on
membranes, for MG 624 and F3, Ki values of respectively
69.5 and 39 mM were more than three (for MG624) and almost
three (for F3) orders of magnitude less potent in displacing 3H-
Epi to the b2-containing subtype than in displacing 125I-aBgtx
to the a7 subtype. F2 also had a very low potency on this
subtype, with a Ki=374 mM.
In order to obtain a complete pharmacological pro®le of the
e€ects of the compounds on the di€erent neuronal nicotinic
AChR subtypes, we tested them on the b4-containing subtypes
immunoimmobilized on the anti-b4 Abs. 3H-Epi binds to these
receptors with a Kd of 20 pM (CV=27); Figure 6 shows the
inhibition curves of MG 624, F2 and F3 on the binding of 3H-
Epi to the b4 containing subtypes (Ki values in Table 2).
Table 2 Comparison of MG 624, F2, F3, F3A and F3B
binding anities to the immunoimmobilized subtypes
Subtype:
Ki
a7
mM
b4
mM
b2
mM
MG 624
F2
F3
F3A
F3B
0.027 (31)
8.3 (20)
14.3 (21) 374
69.5 (25)
27.4
(25)
0.050 (29)
0.070 (40)
0.081 (35)
(38)
(25)
(33)
(31)
4.01 (26)
3.84 (33)
39
75
45
Figure 5 Binding of 3H-Epi to immobilized b2 containing receptors
and its inhibition by MG 624, F2, F3, F3A and F3B. The binding of
3H-Epi to immobilized b2 containing receptors and the inhibition of
the binding by MG 624, F2, F3, F3A and F3B were performed
according to a mixed protocol combining both saturation (the ®rst
7 ± 8 concentrations of respectively 0.005 ± 5 nM) and displacement
curves (the last 3 ± 4 concentrations of 50 ± 1000 nM). The immu-
noimmobilized b2-containing receptors were incubated for 30 min at
208C with the indicated concentration of compounds. 3H-Epi (at a
®nal concentration of 50 pM) was then added and left overnight. The
curves were obtained and processed as described in Figure 4. The
dotted lines represent the 95% c.l.: for the sake of clarity, only the
3H-Epi curve is shown.
4
(38)
Ligand anity 0.5 nM (27)
(Kd):
125I-aBgtx
20 pM (27)
3H-Epi
36 pM (37)
3H-Epi
The Kd and Ki values were derived from 125I-aBgtx
saturation and competition binding curves to the a7 subtype
and 3H-Epi saturation and competition binding curves to b2
and b4 containing subtypes. The curves obtained from three
separate experiments were ®tted using a non-linear least
squares analysis program and the F test (Munson and
Rodboard). The numbers in brackets represent the % of CV.

C. Gotti et al
Selective ligands for neuronal aBungarotoxin receptors
1203
MG 624, F2 and F3 were approximately 10 times more
potent in inhibiting the binding of 3H-Epi to the b4-containing
than to the b2-containing subtypes. None of the compounds
showed any selectivity toward the b2 or b4-containing receptor
subtypes.
Neither of the stereoisomers of F3, which was the most
active compound on b2 and b4-containing receptors, showed
any increase in speci®city for either of the receptor subtypes.
The nature of the inhibition of MG624 and F3 on the
binding of 125I-aBgtx to the immobilized a7-containing
receptors was investigated by performing two saturation
binding experiments in the presence and absence of 0.3 mM
MG624 or F3. Scatchard analysis of the 125I-aBgtx binding in
the absence and presence of MG624 or F3 indicated that
competitive inhibition occurred, as the slope decreased in the
presence of each of the compounds whereas the Bmax was not
changed.
increasing concentrations of MG624 or F3, 30 s before
applying ACh at about its EC50 value. Both compounds
completely blocked the ACh-evoked currents at a concentra-
tion of 1 mM; the IC50 and nH values of MG624 for the a7
receptor were 94 nM and 1.2 (n=6), those of F3 were 119 nM
and 1.3 (n=3).
In vitro experiments
The biological activity of MG624 and F3 compounds was
further characterized by investigating their e€ects on gang-
lionic and muscle-type synaptic transmission. It was already
known that MG 624 had very strong ganglioplegic activity
(Cavallini et al., 1953; Mantegazza & Tommasini, 1955) and
very weak antimuscarinic property in vivo; furthermore, it has
been reported that it has no activity on the neuromuscular
junction. In order to con®rm these results and to compare
them with those of F3, we tested the in vitro e€ects of MG 624
and F3 on two di€erent tissue preparations: the guinea pig
vagus nerve-stomach and the rat phrenic nerve-hemidia-
phragm preparations.
E€ects of MG624 and F3 on oocyte-expressed chick a7
subtype
As the binding experiments showed that MG 624 and F3 were
the most interesting compounds, we tested their biological
e€ects on the functional activity of the chick homomeric a7
subtype expressed in oocyte.
Oocytes injected with a7 subunit cDNA responded to ACh
with an inward current (IACh) whose peak amplitude
depended on the ACh concentration; this current was blocked
by the a7-selective antagonist aBgtx. With the EC50 concentra-
tion of 100 mM ACh, applied to oocytes held at 760 mV, the
mean current amplitude was 594+127 nA (mean+sem) (9
oocytes, 2 donors).
Neither the non-injected oocytes nor those expressing a7
neuronal nicotinic AChRs, responded to MG 624 or F3
(10 nM ± 1 mM) when applied alone. To compare the potency of
MG624 and F3 in inhibiting a7 subtype ACh elicited currents,
experiments were performed in a7-injected oocytes exposed to
Vagus nerve-stomach preparation The guinea-pig vagus nerve
contains axons that synapse with the post-postganglionic
cholinergic neurons which mediate the contractile phase of
the smooth muscle of the stomach. The typical responses
elicited by nerve stimulation, and those elicited by the addition
of ACh, before and after the addition of the compounds are
shown in Figure 7 and Table 3. Both compounds depressed the
contractile response obtained by electrical ®eld stimulation of
the vagus, although with di€erent IC50 values in mM, of 49.4
(40.4 ± 60.3) for MG 624 and 166.2 for F3 (129.6 ± 213.1), but
were totally ine€ective on the stomach contractions obtained
by direct stimulation of the muscarinic receptors by
exogenously added ACh.
Phrenic nerve-hemidiaphragm To verify whether MG 624 and
F3 can have an e€ect on muscle-type AChRs, we tested them
on the phrenic nerve-hemidiaphragm preparation. (Figure 8
and Table 3). We found that, MG 624 had very little activity
Figure 6 Binding of 3H-Epi to immobilized b4 containing receptors
and its inhibition by MG 624, F2, F3, F3A and F3B. The binding of
3H-Epi to the immobilized b4 containing subtype and the inhibition
of the binding by MG 624, F2, F3, F3A and F3B were performed as
described in Figure 5. The curves were obtained as described in
Figure 4. The dotted lines represent the 95% c.l.: for the sake of
clarity, only the 3H-Epi curve is shown.
Figure 7 Typical tracings in two di€erent experiments showing the
e€ect of MG624 and F3 on guinea-pig stomach contractions induced
by electrical stimulation (ES) of the vagus nerve and acetylcholine
(ACh=5 mM). ES: 0.2 ms; 20 Hz, 0.2 V for 10 s every 2 min.

1204
C. Gotti et al
Selective ligands for neuronal aBungarotoxin receptors
on the evoked contractions (IC50=485.9 mM) whereas F3
(IC50=14.4 mM) was more than 33 times more potent; neither
compound in¯uenced the muscle contraction obtained by
means of direct muscle stimulation.
These in vitro data show that MG 624 has inhibitory activity
on guinea pig stomach parasympathetic ganglia but very weak
activity on the rat hemidiaphragm neuromuscular junction. F3
had the opposite e€ects: a good neuromuscular blocking
activity and a ganglioplegic e€ect only at 10 times higher
concentrations.
the a7 and a8 neuronal nicotinic AChR subtypes in
vertebrates) and 3H-Epi (which binds to the b2 and b4-
containing neuronal nicotinic AChR subtypes with pM
anity). We ®rst investigated the binding of these compounds
to crude neuronal membranes and then performed binding
experiments on de®ned immunoimmobilized neuronal nico-
tinic AChR subtypes.
Both MG624 and F3 compounds inhibited the binding of
125I-aBgtx to the a7-containing receptors with nanomolar
anity (27 and 50 nM respectively), but inhibited the binding
3
of H-Epi to the b4-containing and b2-containing subtypes at
concentrations that were two to three orders of magnitude
higher. The same compounds inhibited the binding of 125I-
aBgtx to the muscle-type AChR with a potency similar to that
found at the b2-containing subtype. These compounds a€ected
the subtypes with the following order of potency: a7-
containing 44b4-containing 4b2-containing=muscle-type
receptors.
Discussion
The data reported here indicate that the two 4-oxystilbene
derivatives, the old drug MG624 and the newly synthesized F3,
are potent nicotinic antagonists that are apparently selective
towards neuronal aBgtx receptors.
The results showing the selectivity towards the a7 subtype
were obtained in chick optic lobe, a neuronal tissue consisting
of 60 ± 70% of receptors that only contain the a7 subunit and
the other 20 ± 25% of receptors containing both the a7 and a8
subunits (both subtypes may also contain other unidenti®ed
subunits) (Gotti et al., 1994). Our binding studies clearly show
that these drugs are active on a7 and a7-a8 containing
receptors, but we cannot exclude the possibility that they are
also active on a8-containing receptors.
Binding studies
To investigate whether MG 624, F2, F3, F3A and F3B have a
selective activity towards the di€erent neuronal nicotinic
AChR subtypes present in ganglia and in the CNS, we studied
their inhibition of the binding of 125I-aBgtx (a toxin that binds
Only very high concentrations of our compounds a€ect b2
containing receptors, which we have shown here to be a
heterogeneous class of receptors. Although with approxi-
mately one-hundredth of the potency shown towards the a7
subtype, both MG624 and F3 showed an anity (in the low
mM range) for the receptors containing the b4 subunit (see
Table 2), which also contain the a3 subunit, as revealed by
Western blot and immunoprecipitation analyses using subunit
speci®c antibodies.
Table
3 Estimation of MG 624 and F3 IC50 values
obtaining from dose-response curves of the vagus nerve-
stomach and phrenic nerve hemidiaphragm
Vagus nerve-
stomach
Ic50 (mM)
Phrenic nerve-
hemidiaphragm
IC50 (mM)
MG 624
F3
49.9 (40.5 ± 60.3)
166.2 (129.6 ± 213.1)
485.9 (405.4 ± 582.3)
14.4 (12.0 ± 17.4)
Although F3 is twice as active as MG624 on a3b4-
containing receptors it is not selective, since it blocks both
muscle and a7 receptors (see Table 3). However, its relatively
low Ki value may make it useful in the study of the biophysical
and functional characteristics of a3b4 containing receptors.
The data for both preparations were obtained in seven
separate experiments. For both tissues, the IC50 and 95%
con®dence limits (in brackets) were calculated from the dose-
response curves of MG 624 and F3.
Functional studies
The data in the literature, and our own in vitro results and
preliminary electrophysiological data from oocyte-expressed
subtypes, showed that MG 624 and F3 are nicotinic
antagonists that bind to the a7 subtype with nM anity. The
MG 624 and F3 concentrations required to block the response
of the ACh-activated homomeric chick a7 subtype expressed in
oocytes (94 and 119 nM respectively) are in close agreement
with those required to inhibit the binding of 125I-aBgtx to the
a7-containing subtype (27 and 50 nM respectively).
Although MG624 and F3 show similar potency in binding
studies towards the a7-containing subtype, and also towards
the muscle-type AChR, our `in vitro' studies clearly
demonstrate that the two compounds inhibit the contractile
responses in vagus nerve-stomach and phrenic nerve-hemi-
diaphragm preparations with opposite selectivities: F3 had
greater neuromuscular blocking activity (IC50 of 14.43 mM)
than MG624 (IC50 485 mM).
Given the fact that MG624 has a selective e€ect on a7-
containing receptors and only a€ects muscle-type AChR at
very high concentrations it could be a very useful tool not only
for the in vitro study of the a7 subtype, but also and especially
for in vivo studies because of its good therapeutic index (LD
Figure
8 Typical mechanograms of two di€erent experiments
showing the e€ect of MG624 and F3 on rat hemidiaphragm twitches
indirectly evoked by electrically stimulating the phrenic nerve.
Indirect electrical stimulation: 0.1 Hz, 1 ms, 0.5 V. Direct electrical
stimulation (D): 0.1 Hz, 1 ms, 2 V.

C. Gotti et al
Selective ligands for neuronal aBungarotoxin receptors
1205
50=30 mg/kg versus 0.5 mg/kg to obtain long-lasting gang-
lioplegic activity) (Mantegazza & Tommasini, 1955). Its
therapeutic index could be even better if the a7 subtype is
considered as its target, towards which its anity is 307 times
higher than towards the a3b4 subtype. The a7 receptor has
recently aroused great interest since it has been found to be
present in discrete and important areas of the brain, such as the
hippocampus and caudate and motor cortex (Rubboli et al.,
1994); furthermore, in vitro experiments suggest that its Ca2+
permeability may lead to pleiotropic e€ects on neuronal
communication and development (Role & Berg, 1996), and it
has also been suggested that it may be involved in several brain
pathologies (reviewed in Gotti et al., 1997a). However, its role
in brain function is not yet well known partly because of the
lack of appropriate tools for studying it.
although it is not sucient (F2 has a rather low anity for
neuronal nicotinic AChRs). Our results obtained using a small
number of compounds allow a tentative suggestion that a trans
conjugated, approximately coplanar structure might favor
receptor-bonding interactions with the a7 subtype, involving
Van der Waals and hydrophobic bonding and possibly charge-
transfer complexing.
The 4-oxystilbene derivative F3 can be considered an open
chain congener of compounds II (Figure 1), formally
obtainable by the cleavage of the 3,4 bond of the azacycle.
The binding data for compounds II indicate that some of them
possess nanomolar anity for the a4b2 subtype. On the
contrary, the open chain congener N,N,N-trimethyl-1-(4-trans-
stilbenoxy)-2-propylammonium, F3, has completely lost the
high anity towards the a4b2 subtype but has acquired
nanomolar anity for the a7 subtype and low anity for the
receptor containing the b4 subunit. Moreover, the enantiomers
of this compound, F3A and F3B, show little stereoselective
binding. Two structural variations of F3 were considered: (i)
the alteration of the quaternary ammonium head and the
removal of the methyl group from the alkyl bridge (see MG
624); (ii) the reverse position of the ammonium head and the
stilbenoxy residue (see F2). In the ®rst case, a favourable
in¯uence of the N,N,N-triethyl ammonium moiety on anity
can be observed. The di€erences in Ki values between MG 624
and F3 may be due to better exploitation of the binding pocket
as a result of the ammonium volume. A rather low anity for
neuronal nicotinic AChRs is shown by F2, which carries the
methyl group near the oxygen. This result leads to the
hypothesis that the methyl group has an unfavourable e€ect
on the modulation of the sterical ®t to the binding site of the
receptor. In this regard, it is supposed that the methyl group
interferes with a hypothetical region of steric sensitivity or with
free access to the H-bond donor group.
These results clearly show that the 4-oxystilbene moiety, an
ammonium head and an alkyl bridge with a methylene group
close to the phenoxy oxygen are necessary requirements for
achieving neuronal nicotinic AChR a7 subtype selectivity.
In conclusion, we have found that two compound belonging
to oxystilbene derivates show a strong selectivity towards
neuronal nicotinic receptors, particularly the a7 subtype. We
suggest that further investigations of this class of compounds
could be important for ®nding compounds with neuronal
nicotinic AChR selectivity, and for understanding the
structure of the ligand binding site of these receptors.
Although other selective a7 antagonists are currently
available and act at lower concentrations than MG624, such
as the Bungarus and Conus a-toxins, and methyllycaconitine
(reviewed in Gotti et al., 1997a), we believe that they could be
rather toxic, expensive and more dicult to use in in vivo
studies.
The nicotinic agonists so far known are not selective and
even the recently discovered nicotinic ligands (Epi, ABT 418,
A-35380 and GTS 21) have a lower anity for the a7 than for
the a4b2 subtype (reviewed in Gotti et al., 1997a). The only a7
agonist so far reported is the very recently discovered
anabaseine (Kem et al., 1997), a toxin that has a Ki of 58 nM
in inhibiting the binding of 125I-aBgtx to the a7 receptor
present in rat brain membranes, a value that is very similar to
that found for MG624 and F3; the drawback of this agonist is
that it also inhibited the binding of 125I-aBgtx to the muscle-
3
type AChR and the binding of H-methylcarbamylcholine to
the rat a4b2 subtype with very similar potency.
For all of these reasons, MG 624 and F3 could be very
important for understanding the function of the a7 receptor in
vivo. It is also conceivable that they could be used for clinical
purposes and possibly for other pathologies in which
experimental data suggest that a7 may be involved, such as
the control of nicotine addiction, the control of cell
proliferation in some lung tumours (Codignola et al., 1994;
Quik et al., 1994) or the control of brain cell excitability
(Marks et al., 1989).
Chemical characteristics
In our compounds, three functional groups seem to be
important to explain their anity and selectivity towards
neuronal nicotinic AChR subtypes: (i) the styryl substituent;
(ii) the alkyl bridge connecting the quaternary ammonium
group to the phenoxy oxygen; (iii) the ammonium head. It is
very interesting to observe that the styryl moiety at the para
position of the phenyl ring confers a preferential predisposition
for neuronal nicotinic AChRs, (especially for the a7 subtype),
We would like to thank Mr Kevin Smart, Mr Paolo Tinelli and Ms
Ida Ru€oni for their aid with the manuscript. This work was
Â
supported in part by grants from Fabriques de Tabac Reunies,
Neuchatel, Switzerland, the Italian Ministry of University and
Ã
Scienti®c and Technological Research, the European Programme
`Training and Mobility of Researchers', Contract n8ERB4061PL97-
0790 and the Telethon grant n81047 to Cecilia Gotti.
References
ABREO, M., LIN N-H., GARVEY D., GUNN, D., HETTINGER A.M.,
VASICAK, J., PAVLICK, P., MARTIN, Y., DONELLY-ROBERTS, D.,
ANDERSON, D, SULLIVAN, J., WILLIAMS, M., ARNERIC, S. &
HOLLADAY, M. (1996). Novel 3-pyridyl ethers with subnano-
molar anity for central neuronal nicotinic acetylcholine
receptors. J. Med. Chem., 39, 817 ± 825.
BULBRING E. (1946). Observation on the isolated phrenic nerve
hemidiaphragm preparation of the rat. Br. J. Pharmac.
Chemother., 1, 38 ± 61.
CAVALLINI, G., MANTEGAZZA, P., MASSARINI, E. & TOMMASINI,
R. (1953). Sull'attivita ganglioplegica di alcuni derivati alchila-
Á
minici dello stilbene e del difenile. Il Farmaco, 6, 317 ± 331.
CODIGNOLA, A., TARRONI, P., CATTANEO, M.G., VICENTINI, L.M.,
CLEMENTI, F. & SHER, E. (1994). Serotonin release and cell
proliferation are under the control of a-bungarotoxin-sensitive
nicotinic receptors in small-cell lung carcinoma cell lines. FEBS
Lett., 342, 286 ± 290.

1206
C. Gotti et al
Selective ligands for neuronal aBungarotoxin receptors
CONROY, W.G. & BERG, D.K. (1995). Neurons can maintain multiple
classes of nicotinic acetylcholine receptors distinguished by
di€erent subunit compositions. J. Biol. Chem., 270, 4424 ± 4431.
DANI, J.A. & HEINEMANN, S. (1996). Molecular and cellular aspects
of nicotine abuse. Neuron, 16, 905 ± 908.
MCGEHEE, D.S. & ROLE, L.W. (1995). Physiological diversity of
nicotinic acetylcholine receptors expressed by vertebrate neu-
rons. Annu. Rev. Physiol., 57, 521 ± 546.
MILEDI, R., PARKER I. & SUMIKAWA, K. (1989). Fidia research
foundation Neuroscience Award Lectures, 1987 ± 1988 Raven:
New York, 3, 57 ± 90.
DRAPER, N.R. & SMITH, H. (1966). Applied regression analysis. New
York: Wiley.
MUNSON, P.J.
& RODBARD, D. (1980). LIGAND: a versatile
ELLIOTT, R., KOPECKA, H., GUNN, D., LIN, N., GARVEY, D.,
RYTHER, K., HOLLADAY, M., ANDERSON, D., CAMPBELL, J.,
SULLIVAN, J., BUCKLEY, M., GUNTHER, K., O'NEILL A.,
DECKER, M. & ARNERIC, S. (1996). 2-(Aryloxymethyl) azacyclic
analogues as novel nicotinic acetylcholine receptor (neuronal
nicotinic AChR) ligands. Bioorganic & Medical Chemistry letters,
19, 2283 ± 2288.
FORSAYETH, J.R. & KOBRIN, E. (1997). Formation of oligomers
containing the b3 and b4 subunits of the rat nicotinic receptor. J.
Neurosci., 17, 1531 ± 1538.
GERZANICH, V., PENG X., WANG, F., WELLS, G., ANAND, R.,
FLETCHER, S. & LINDSTROM J. (1995). Comparative pharma-
cology of epibatidine: a potent agonist for neuronal nicotinic
acetylcholine receptors. Mol. Pharmacol., 48, 774 ± 782.
GOTTI, C., HANKE, W., MORETTI, M., BALLIVET, M., CLEMENTI, F.
& BERTRAND, D. (1994). Pharmacology and biophysical proper-
ties of a7 and a7-a8 a-Bungarotoxin receptor subtypes immuno-
puri®ed from the chick optic lobe. Eur. J. Neurosci., 6, 1281 ±
1291.
computerized approach for characterization of ligand-binding
systems. Anal. Biochem., 107, 220 ± 239.
PALMA, E., BERTRAND, S., BINZONI, T. & BERTRAND D. (1996).
Neuronal nicotinic a7 receptor expressed in Xenopus oocytes
presents ®ve putative binding sites for methyllycaconitine. J.
Physiol (London), 491.1, 151 ± 161.
PAPKE, R.L. (1993). The kinetic properties of neuronal nicotinic
receptors: genetic basis of functional diversity. Progr. Neurobiol.,
41, 509 ± 531.
PATON, W. & VANE J. (1963). An analysis of the responses of the
isolated stomach to electrical stimulation and to drug. J. Physiol.,
165, 10 ± 46.
QUIK, M., CHAN, J. & PATRICK, J. (1994). a-Bungarotoxin blocks the
nicotinic receptor mediated increase in cell number in
neuroendocrine cell line. Brain Res., 655, 161 ± 167.
a
ROLE, L.W.
& BERG, K.D. (1996). Nicotinic receptors in the
development and modulation of CNS synapses. Neuron, 16,
1077 ± 1085.
ROVATI, G.E., RABIN, D. & MUNSON, P.J. (1991). Analysis, design
and optimization of ligand binding experiments. In Horizon in
Endocrinology (Vol II). ed. Maggi, M. & Geenen, E.V. pp. 155 ±
167. New York: Serono Symposia Publication from Raven Press.
RUBBOLI, F., COURT, J.A., SALA, C., MORRIS, C., CHINI, B., PERRY,
E. & CLEMENTI, F. (1994). Distribution of nicotinic receptors in
the human hippocampus and thalamus. Eur. J. Neurosci., 6,
1596 ± 1604.
GOTTI, C., FORNASARI, D. & CLEMENTI F. (1997a). Human
neuronal nicotinic receptors. Progr. Neurobiol., 53, 199 ± 237.
GOTTI, C., MORETTI, M., MAGGI, R., LONGHI, R., HANKE, W.,
KLINKE, N.
& CLEMENTI, F. (1997b). a7 and a8 nicotinic
receptor subtypes immunopuri®ed from chick retina have
di€erent immunological, pharmacological and functional prop-
erties. Eur. J. Neurosci., 5, 1201 ± 1211.
KEM, W., MAHNIR, V., PAPKE, R., & LINGLE, C. (1997). Anabaseine
is a potent agonist on muscle and neuronal alpha-bungarotoxin-
sensitive nicotinic receptors. J. Pharmacol. & Exp. Therap., 283,
979 ± 992.
MANTEGAZZA, P. & TOMMASINI, R. (1955). Central antinicotinic
activity of 4-oxystilbene and 4-oxydiphenylethane derivatives.
Arch. Int. Pharmacodyn., 4, 371 ± 403.
SARGENT, P.B. (1993). The diversity of neuronal nicotinic acetylcho-
line receptors. Annu. Rev. Neurosci., 16, 403 ± 443.
WHITING, P.J., LIU, R., MORLEY, B.J. & LINDSTROM, J.M. (1987).
Structurally di€erent neuronal nicotinic acetylcholine receptor
subtypes puri®ed and characterized using monoclonal antibo-
dies. J. Neurosci., 7, 4005 ± 4016.
WONNACOTT, S. (1997). Presynaptic nicotinic ACh receptors.
Trends Neurosci., 20, 92 ± 98.
MARKS, M.J., STITZEL, J.A.
& COLLINS, A.C. (1989). Genetic
in¯uences on nicotine responses. Pharmacol. Biochem. Behav.,
33, 667 ± 678.
(Received December 15, 1998
Revised April 23, 1998
Accepted April 24, 1998)