Vol. 22, No.
4
Mechanism of Action of Acamprosate. Part I.
Characterization of Spermidine-Sensitive Acamprosate
Binding Site in Rat Brain
Mickael Naassila, Saloua Hammoumi, Elisabeth Legrand, Philippe Durbin, and Martine Daoust
It has been suggested that the anticraving drug, acamprosate, acts
via the glutamatergic system, but the exact mechanism of action is
still unknown. The aim of this study was to characterize PHlacam-
A has been successfully used for several years in clinical
prosate binding and establish whether this showed any relation to
sites on the NMDA receptor complex. We found saturable specific
binding of [3H]acamprosate to rat brain membranes with a KDof 120
pM and a B,,, of 450pmol/mg of protein. This acamprosate binding
site was sensitive to inhibition by spermidine (Ern: 13.32 k 1.1 pM;
Hill coefficient = 1.W), and arcaine and glutamate both potentiated
the inhibitory effect of spermidine. Acamprosate binding to the
acamprosate binding site was also sensitive to inhibition by divalent
cations (Ca2+,Mg2+, and S?’). Conversely, acamprosate displaced
[‘4C]spermidine binding from rat brain membranes with an C I , of
645 pM and a Hill coefficient = 1.74. This inhibitory effect of acam-
prosate was not affected by arcaine, and was associated with a
significant reduction in B,,, and binding affinity for spermidine, sug-
gesting an allosteric interaction between acamprosate and a sper-
midine binding site. These data are consistent with an effect of
acamprosateon the NMDA receptor protein complex, and acampro-
sate was also found to alter binding of [3H]dizocilpine to rat brain
membranes. When no agonists were present in vitro (minimal NMDA
receptor activation), acamprosate markedly potentiated rH]dizo-
cilpine binding at concentrationsin the 5 to 200 p M range. However,
under conditions of maximal receptor activation (100 p M glutamate,
30 pM glycine), acamprosate only inhibited rH]dizocilpine binding
(at concentrations concentrations >lo0 pM). When these binding
studies were performed in the presence of 1 p M spermidine, the
enhancing effects of acamprosate on PHIdirocilpine binding were
inhibited. The results show that acamprosate binds to a specific
spermidine-sensitive site that modulates the NMDA receptor in a
complex way. Together, with data from al Quatari et al. (see next
paper), this work suggests that acamprosate acts as “partial co-
agonist” at the NMDA receptor, so that low concentrationsenhance
activation when receptor activity is low, whereas higher concentra-
tions are inhibitory to high levels of receptor activation. This may be
relevant to the clinical effects of acamprosate in alcohol-dependent
patients during abstinence.
Key Words: Acamprosate, Acamprosate Binding Site (ASS), Sper-
midine, NMDA Receptor Complex, Craving.
From thi, INSERM U 29.5 (M.N., S.H.) UFR de Mddecine-Pharmacie de
Horieti, Saitri Elienne Roii,my3 Frutice; UFR de Phamiarie (E.L., M.D.),
Universite [ I c Picurdit. hikes Verne, Amiens, France; and Lipha Company
( PD.), Lyon. harice.
Received ,for piihlicaiioir May 28. 1997; accepied Janiiary 7, I998
This stimdy M”W .supported lJy Groiipe Lipha.
Reptint reqiiesis: Propssor Martitie Daoust, Universifi!de Picardie Jules
Vetne, Facrilid rle Phamiacie. 1 nie drs Louvels 8000 Amiens, France; Far:
.33-3-22-82-76; E-tnail: i i o o i i s t ~ ~ e a . ~ i r e t . ~ .
Copyrighi 0 I998 by T/le Research Sociep o n Alcoholi.rm.
802
June 1998
CAMPROSATE (CALCIUM acetylhomotaurinate)
practice to prevent relapse in weaned alcoholics. Acampro-
sate was marketed first in France and then in other Euro-
pean countries.
The success of multicenter clinical trials’ encourages pre-
clinical studies to understand the exact mechanisms by which
acamprosate treatment prevents alcohol abuse disorders.
Acamprosate also decreases alcohol craving in rats, in
various animal models (including free choice paradigm):
or animal models for alcohol ~ r a v i n gRecent.~ studies also
demonstrated that acamprosate reduced alcohol-induced
physical signs of ~ i t h d r a w a l In
. ~ a rat model of long-term
alcohol-drinking that represents an animal model of human
relapse behavior, acamprosate diminished the alcohol de-
privation effect, confirming the anticraving action of acam-
prosate treatment.3
On the other hand, biochemical studies have shown that
acamprosate could have an action on the glutamatergic
system. Acamprosate, in vitro, reduces the postsynaptic
efficacy of excitatory amino acid neurotransmitters and
neuronal excitability in the rat neocortex.’
Taken together, these results demonstrated that the effect
of acamprosate treatment on alcohol drinking could be due to
an action on brain excitatory amino acid transmission.’
It seems plausible to imagine an interaction between an
anticraving drug and the amino acid system; various data
indicate that glutamatergic and specifically NMDA recep-
tors may represent an important site of action for alcohol in
the CNS. Biochemical and electrophysiological studies
have shown that in vitro ethanol inhibits NMDA receptor-
mediated calcium currents6 and NMDA neurotoxicity.’ In
contrast, chronic alcohol abuse upregulates NMDA recep-
tor function’ and the density of NMDA receptor channel
binding sites, especially in the cortex and hippocampus.’
On the other hand, the NMDA receptor has been pro-
posed to be the postsynaptic receptor activated during
brain neuronal damages associated with chronic alcohol
exposure: loss of memory or learning. The regulation of
NMDA receptor plays a critical role in many pathological
conditions of neuronal degeneration observed during
and/or after chronic periods of alcohol abuse. Moreover,
the NMDA receptorhon channel complex contains a num-
ber of distinct sites for endogenous and exogenous ligands;
Alcohol Clin Erp Res, Vol22, No 4, 1998: pp 802-809
MECHANISM OF ACTION OF ACAMPROSATE. PART I 803

this includes binding sites for glutamate (or NMDA), gly- Table 1. Influence of Various Buffer Concentrations on Specific
rH]Acamprosate Binding
cine, Mg2+, Zn2+, and open channel blockers (such as
Buffer 0.1 mM Tris 1 mM Tris 10 mM Tris
ketamine, PCP, and dizocilpine). Glycine is considered to
be a “co-agonist” at the NMDA receptor complex, and Specific binding 202 t 12 (55)’ 103 It 19 (60)’ 66 2 3 (40)
(fmol/mg of protein)
endogenous polyamines (spermidine, spermine, and pu-
trescine) have been described as endogenous neuromodu- Buffer 0.1 mM HEPES 1 mM HEPES 10 mM HEPES

lators of NMDA receptors. They act at a distinct recogni- Specific binding 211 t 14 (45)’ 155 % 14 (45)’ 89 5 3 (35)
(fmol/mg protein)
tion site within the NMDA receptor complex. The
modulation of the NMDA receptor at one subtype of mod- 25 nM of rH]acamprosate were incubated in 1 ml of each buffer containing 0.5
mg of membrane protein. Nonspecific binding was determined using 5 mM of
ulatory site seems to be important in the involvement of unlabeled acamprosate. Results are expressed as means % S E M of four inde-
this complex in neuronal changes accompanying chronic pendent experiments.
alcohol exposure. Littleton” suggested that acamprosate * p < 0.001, compared with the respective lower values using Student‘s t test
(represents % of specific binding).
actas on a mechanism “downstream” from amino acid re-
ceptor activation to depress neuronal excitability.
The aim of this work was to answer two fundamental and represented 30% of total binding. For saturation studies, increased
questions concerning the mechanism of action of acampro- concentrations (up to 500 pM) of unlabeled spermidine were used.
sate: (1) Does acamprosate bind to a central site? (2) If yes,
is this binding site associated with the NMDA receptor [3H]Dizocilpine B i n d i n e
complex? [3H]Dizocilpine binding assays were performed using well-washed
We studied binding conditions of [3H]acamprosateon brain membranes, prepared as described for other binding studies. 0.1 mg of
membrane preparations to define the number of binding site protein was incubated in 1 ml of 10 mM Tris-HC1 buffer @H 7.4, 20°C)
containing 0.5 nM [3H]dizocilpine at room temperature for 2 hr. Tubes
populations and their affinity. We also defied displacing were rapidly filtered on Whatman glass fiber filters (45 pm pore size) and
agents and characterized the nature of [3H]acamprosatebind- rinsed with 2 X 5 ml of cold Tris buffer. Radioactivity was determined
ing site (ABS). using 5 ml of ACS fluid scintillation.
Nonspecific binding was defined using 100 FM dizocilpine. In some
experiments, saturating concentrations of glutamate (100 pM) and glycine
MATERIALS AND METHODS (30 pM) were added to the incubation buffer (see “Results”).
Membrane Preparation For all of these studies, radioactivity was counted in a LKB 1211
spectrometer [efficiency = 26% for 3H and 40% for 14C; each sample
Membranes were prepared using whole brains of adult male Sprague- (total and nonspecific) was determined in triplicate].
Dawley rats (250 g) obtained from Charles River (Saint Aubin 1Bs Elbeuf, Binding parameters (KD, Bmm)were evaluated using Ligand software
France). Brains were rapidly removed on ice and thawed in 5 volumes of (Biosoft, New York). Data for inhibition parameters were expressed as
0.32 M sucrose using an Elvejhem type potter. After the first centrifuga- IC,, values using Origin software. All displacing agents were dissolved in
tion (3,000 X g, 4°C 15 rnin), the supernatant was further centrifuged the assay buffer, and the pH was adjusted to 7.4.
(48,000 X g, 4°C. 15 rnin). The pellet was carefully rinsed five more times
using 5 volumes of 1 mM Tris-HCI buffer (pH 7.4, 20°C). The final pellet
was frozen (- 18°C) until use. Statistical Analysis
Means t SEM were compared using Student’s t test or ANOVA
[-’HIAcamprosate Binding analysis, as indicated in the text. In the case of results expressed as
percentage of control values, statistical analysis was performed using
Assays were performed using thawed membranes prepared in 1 mM
binding values.
Tris-HC1 buffer (pH 7.4,20”C) and rinsed twice in the same buffer. 100 p1
of membrane preparation containing [0.5 mg of protein (Lowry)] was
incubated in the presence of 2 nM [3HJacamprosate (s.a.: 48 Ciimmol) and Drugs
1 mM Tris-HC1 buffer (pH 7.4, 20°C). For saturation cuwe assays, in-
[3H]Acamprosate (specific activity = 48 Ci/mmol) and [’4C]spermidine
creasing concentrations (pM to mM) of unlabeled acamprosate were
(specific activity = 110 mCi/mmol) were obtained from Amersham
used. The final assay volume was 1 ml. Nonspecific binding was deter-
(France) and [3H]dizocilpine (specific activity = 23 Ci/mmol) was ob-
mined using 5 mM of acamprosate. Assays were incubated at room
tained from NEN (France). Spermidine, arcaine, and dizocilpine were
temperature (20°C) and terminated by rapid centrifugation (48,000 X g,
purchased from RBI. Acamprosate was a gift from Groupe Lipha (Lyon,
4”C, 10 min). The final pellet was resuspended in 250 111 of distilled water
France). All other drugs were obtained from commercial sources.
and 5 ml of ACS (Wallack) scintillation fluid for radioactivity determina-
tion. Under these conditions, specific binding represented 40% of total
binding.
RESULTS

[‘JC]Spemidine Binding Optimization of [3H]Acamprosate Binding

This binding has been described elsewhere.” Brain membrane prepa-
rations were thawed and rinsed as described. 5 nM spermidine was incu-
bated in the presence of membranes (0.5 mg of protein) in 10 mM
Tris-HC1 buffer (pH 7.4, 20”C), 15 rnin at 20°C. Bound and free spermi-
dine were separated by centrifugation (48,000 X g, 15 min, 4°C). Radio-
activity of the final pellet was determined using 5 ml of ACS fluid
scintillation. Nonspecific binding was evaluated with 10 mM spermidine
We studied the effect of various concentrations of Tris
and HEPES buffers on specific acamprosate binding. The
highest concentrations (10 mM) of the two buffers de-
creased specific binding of acamprosate. 0.1 mM Tris and
HEPES both significantly increased acamprosate binding
(Table 1) by 300%, compared with the 10 mM concentra-
804 NAASSIIA ET AL.

binding was 0.4 mg of proteidml. Maximum specific binding

was obtained at 20°C after a 1Imin incubation period.

Ca Characterization of ABS
Various pharmacological tools for the NMDA receptor
were tested at concentrations reported to be active at the
II 0 NMDA receptor. Only 10 pM of spermidine and CaCl,
0 were able to significantly decrease (by 40 and 31%, respec-
Fig. 1. Chemical structure of acamprosate.
tively) specific acamprosate binding. Other agents tested
Table 2. Effect of Various Pharmacological Agents on Specific had no effect at the same concentration: glycine (co-ago-
[3H]Acamprosate Binding (fmolhg of protein) nist), ketamine (antagonist), taurine (acamprosate analog),
Displacing agents (10 pM) NMDA, GABA, CGS 19755 (NMDA antagonist), gluta-
Control Glycine Ketamine Taurine GABA Glutarnate mate, and MgC1, (see Fig. 4, Table 2).
13.54 f 1.05 16.33 2 2.5 15 2 1.5 15.5 t 1.8 12.2 f 2.6 13.4 f 2.4 Figure 2 shows that acamprosate bound to a single site
MgCI, NMDA MK 801 CGS 19755 Spermidine CaCI,
*
(KD = 120 15 pM, B,, = 450 +- 50 pmol/mg of protein).
From analysis binding using acamprosate alone, it was
11.74? 0.52 12.5 -t 1.55 11.5 5 1.7 12.25 ? 1.8 8.5 f 0.52"* 9.42 ? 1.1'
impossible to resolve binding into more than one site. On
5 nM of [3H]acamprosatewas incubated in the presence of 10 pM of each drug the other hand, when 100 pM spermidine was used to
in 1 ml of 1 rnM Tris buffer (pH 7.4, 20°C) containing 0.5 mg membrane protein.
Nonspecific binding was determined using 1 mM of unlabeled acamprosate. Results define nonspecific binding, another population of binding
are expressed as the means 5 SEM of four or eight (control) independent experi- site (KD = 69 t 11 p,M and B,, = 218 5 16 pmol/mg of
mentations. Statistical analysis was performed using Student's t test. protein) was identified.
p values are as follows: * p < 0.05; ** p < 0.01 compared with control values.
The effect of spermidhe on ABS is shown in Fig. 3. Sper-
midine alone (Fig. 3a) decreased acamprosate binding with an
tion; but, these concentrations did not allow stable pH IC,, of 13.32 2 1.1pM and a Hill coefficient of 1.04.The IC,,
values during experiments. Moreover, as shown in Table 1, for glutamate alone at the ABS was higher than 100 pM, but
binding 1 mM Tris buffer represented higher specific bind- glutamate decreased IC,, values for spermidine (IC50 =
ing, compared with the HEPES buffer. Consequently, we 7.99 5 0.65 p M , p < 0.01 vs. spermidine alone).
chose 1 mM Tris (pH 7.4,ZoOC) for subsequent experiment. Arcaine, the competitive antagonist of spermidine, had a
Under these buffer conditions, specific bound values were slight, but nonsignificant effect on acamprosate binding
103 -C 19 fmol/mg of protein, using 25 nM ['Hlacampro- (Fig. 3b), but potentiated spermidine-induced acamprosate
sate, and 5 mM unlabeled acamprosate for nonspecific binding inhibition (p < 0.001 for the effect of arcaine +
binding determination. On the other hand, different incu- spermidine versus that of spermidine alone).
bation temperatures were realized. Specific binding at WC, Other displacing agents were tested (Fig. 4), and stron-
20"C, and 37°C represented 32 2 5 , 64 -C 6, and 39 2 5 % tium was found to be the most active with an IC,, of 76.4
of total binding (data not shown). Nonspecific binding was pM. Dizocilpine also decreased acamprosate binding with
evaluated using unlabeled 5 mM acamprosate and repre- an IC,, of 430.5 pM. Other IC,,'s were 1275.8 and 466.7
sented 40% of total binding (Fig. 1).A marked decrease of pM for glutamate and ifenprodil respectively; no detect-
specific binding (data not shown) was observed with the able effect was observed for ketamine (Fig. 4).
filtration procedure for bound/free separation, suggesting a
high rate dissociation; this was observed for aff incubation
temperatures (OOC, 2WC, and 37°C). Bound and free fractions ABS and Spermidine Binding
were obtained by centrifugation (48,000 X g, 10 min, 4°C). Acamprosate decreased ['4C]spermidine binding (Fig. 5)
The optimum protein concentration for [3H]acamprosate with an IC,, of 645 pM and a Hill coefficient = 1.74. The

Fig. 2. Scatchard representation of [3H]acarnprosate

-- 3
2$
4 40u

kl
binding. 0.4 rng of total brain membrane protein prepared as 4

t
300
2,s
described in "Materials and Methods" were incubated in 1 ml g 200
a
of 1 rnM Tris-HCI buffer (pH 7.4, 20°C). Nonspecific binding 5 2
$
s100
0
o O K

was determined using 5 mM acamprosate (squares) or 100 2 0 200 400 600 800 IOU0
pM sperrnidine (circles). In the case of acamprosate nonspe-
cific determination, Scatchard analysis revealed a single
5
Z
13141A~amprosatrpM

E l
population of binding sites (13"~ = 450 f 50 p m o l h g of 3
+Non specific binding:
protein; KO= 120 f 15 pM). When nonspecific binding was $ 0,5 acamprosate 1 mM
determined using 100 p M spermidine, another site was ob-
-+Non specific binding:
Six B,,,
69 pM.with
served independent
= 218 ? determinations
16 p m o l h g of protein
were performed,
and KO = 0 0 200 400 600 spermidine IOOpM
and the figure represents a typical experiment. Boiuid (pmoles/mg.prot)
MECHANISM OF ACTION OF ACAMPROSATE. PART I
1
-
’*O
- - Glutamate
-I-
_i
1 1000
***
rti +
effect of acamprosate was not arcaine-sensitive (IC50= 610
p M and a Hill coefficient = 1.76).
Binding parameters for [14C]spermidine binding were:
B,, = 10.73 5 0.05 pmoVmg of protein and KD = 4.5 2
0.3 pM. When 1 mM acamprosate was added, affinity and
number of binding sites were both significantly lowered
(BmW= 4.04 2 0.02 pmoVmg of protein, p < 0.01, and
*
KD = 10.3 0.15 p M , p < 0.01) (data not shown).
Acarnprosate and [3H]Dizocilpine Binding (Figs. 6 and 7)
Acamprosate and spermidine both increased, but not
significantly, dizocilpine binding in the presence of saturat-
ing concentrations of the co-agonists glutamate and glycine
(Fig. 6a). The increase observed was not significant. For
higher concentrations (higher than 100 pM), spermidine
and acamprosate both significantly decreased [3H]dizo-
cilpine binding.
805
Spermidine
Spermidine + Glu I00 pM
Fig. 3. 25 nM of rH1acamprosate was incu-
bated in Tris-HCI buffer as described in “Materials
and Methods” in the presence of each drug or
both drugs. Effect of spermidine, glutamate, and
spermidine + 100 p M glutamate on specific
rH]acamprosate binding (a). IC,o values were
13.32 5 1.1 p M and 7.99 z 0.65 pM for spermi-
dine and spermidine + 100 pM glutamate. re-
spectively. Hill coefficient for spermidine was
1.04. ICs0 for glutamate alone was higher than
100 pM. n = 6. Bindingvalue was 65 2 5 fmol/mg
of protein for control values. (b) The effect of 50
pM spermidine, 10 pM of arcaine, and both drugs
on specific rH1acamprosate binding. Spermidine
significantly decreased specific rH1acamprosate
binding ( p < 0.001,compared with control val-
ues). 10 p M of arcaine decreased specific rH]a-
camprosate binding, but not significantly. 50 p M
spermidine and 10 p M arcaine both had additive
effects on specific rH]acamprosate binding ( p c
0.001, compared with spermidine or arcaine
alone). Results are expressed as % of specific
rH]acamprosate binding (means 5 SEM of six
independent experiments).
These effects were also significant in the absence of
co-agonists (Fig. 6b), and both were antagonized by the
antagonist arcaine. 100 pM arcaine prevented the enhanc-
ing effect of spermidine or acamprosate over the concen-
tration range of 1 to 100 p M of these ligands.
When acamprosate and spermidine were added together
(Fig. 7), at nonsaturating concentrations (pM for spermi-
dine and less than pM for acamprosate), they significantly
decreased dizocilpine binding. For acamprosate concentra-
tions that significantly increased dizocilpine binding (high-
er than ,uM), spermidine significantly antagonized the
acamprosate effect on [3H]dizocilpine binding.
DISCUSSION
This study shows, for the first time, that acamprosate
binds to a specific site on brain membranes. This ABS is
806 NMSSILA ET AL.

sities of spermidine binding sites are consistent with a role

in intracellular a~tivities'~ for spermidine. The cellular lo-
calization of ABS remains to be determined. We also ver-
ified that [3H]acamprosate was not trapped by a membrane
carrier. In synaptosomal preparations using different meth-
ods [i.e., ouabaine or 0°C incubation (data not shown)], we
did not find any specific [3H]acamprosate uptake. This,
together with the Hill coefficient of 1.04 for spermidine
displacement of acamprosate, and the fact that glutamate
enhanced the effect of spermidine on the ABS, suggests a
0 /____- ~ 1- I
competitive interaction between acamprosate and spermi-
10 I00 1000 I0000 dine for a receptor site on the NMDA receptor complex.
Displacingagents ( pM) However, arcaine, which is a competitive antagonist of
spermidine at one site on the NMDA receptor, failed to
1 +Glutamate
~-
-=- lfeiiprodil *-Ketamiiie
~ ~- ~
- b M K 801 +Strontium
~~
1 significantly displace acamprosate and ifenprodil, which is
Fig. 4. Effect of ifenprodil. dizocilpine. and strontium on specific rH]acam- also suggested to interact with this site,14 displaced acam-
prosate binding. 25 nM of [3H]acamprosate was incubated in Tris-HCI buffer as prosate only at millimolar concentrations. There are sev-
described in "Materials and Methods" in the presence of various concentrations
of each drug. IC,,'s were 76.4 i 5.2, 430.5 -C 25, and 466.7 t- 32 p M for
eral potential explanations, but two alternatives seem most
stontium, dizocilpine, and ifenprodil, respectively. Binding value for control was likely: either ABS is a spermidine receptor on the NMDA
63 t- 4 fmol/mg of protein. Results are expressed as % of specific rH]acarnpro- complex (which is insensitive to arcaine or ifenprodil") or
sate binding (means -C SEM of six independent experiments).
the interactions between acamprosate and spermidine are
allosteric.
100 1 The cation-sensitive spermidine site on the NMDA re-
* ' a~ possible candidate for the ABS, and this
c e p t ~ r ' ~ is

c RO1-1 might explain the requirement for the Ca2+ moiety in the
acarnprosate molecule. On the other hand, allosteric inter-
actions could account for the results; it is still uncertain

a
Log [acamprosate] pM
8 6o whether or not the A B S is associated with the NMDA
receptor complex. The observation that arcaine, although

I
not displacing acamprosate, potentiated spermidine-in-
duced displacement, argues that the interactions are allo-
2o
0 - -+ + / -, t ,-,-I
[Acamprosate] pM steric and take place on the NMDA receptor complex. Sim-
10 I000 I00000 ilarly, the observation that acamprosate displaced
Fig. 5. Effect of acamprosate with or without 10 FM arcaine on [14C]spermi- spermidine binding with a Hill coefficient of 1.74 suggests
dine binding. 5 nM of [14C]spermidinewas incubated in the presence of the drugs an allosteric rather than a competitive interaction with a
(see "Materials and Methods" for details). C I , for acamprosate with and without
arcaine were 610 t- 15 and 645 i 22 FM, respectively, with a Hill coefficient of spermidine receptor site.
1.74. n = 6. Results are expressed as % of control values. Control value for To investigate whether acamprosate directly affects the
specific ['4C]spermidine binding was 150 ? 5 fmol/mg of protein. n = 6 inde- NMDA receptor, we studied the effects of acamprosate
pendent experiments.
and spermidine on [3H]dizocilpine binding to rat brain
membranes. Acamprosate potentiated dizocilpine binding
abundant within the CNS and has relatively low affinity for at concentrations in the micromolar range. This potentiat-
acamprosate. It is possible that two or more populations of ing effect became significant in the absence of co-agonists
ABS are present within the CNS, based on the effects of glutamate and glycine (i.e., submaximal activation of the
high concentrations of spermidine. The difference between NMDA receptor). As shown in Fig. 6a, acamprosate did
B,, values obtained with or without spermidine as the not significantly increase [3H]dizocilpine binding for the
determinant of nonspecific binding suggests that the sper- lower concentrations of acamprosate. Different results
midine-sensitive population of ABS represents a subgroup were obtained in case of the absence of both glutamate and
of the total ABS. The apparent KD values for spermidine glycine. In this case (Fig. 6b), acamprosate and spermidine
displacement of acamprosate from the ABS are in agree- increased [3H]dizocilpinebinding with a higher efficiency,
ment with the data of Yoneda and Ogita," who reported a suggesting that the modulation of NMDA receptor is im-
KD of 160 pM for spermidine to brain membranes. The portant for an acamprosate interaction. In the same way,
high concentration of ABS is similar to that of spermidine spermidine alone (Fig. 6b) is less effective to reduce [3H]di-
binding sites described by Yoneda and Ogita" or Gilad and zocilpine binding without glutamate or glycine. This result
Gilad,'* who found B,, values ranging between 837 and is in agreement with those of and it was
2661 pmolfmg of protein in all brain structures examined demonstrated that, for the highest concentrations of sper-
and 28.6 pmol/mg of protein, respectively. These high den- midine, [3H]dizocilpinebinding was decreased.
MECHANISM OF ACTiON OF ACAMPROSATE. PART i

_ _ ~ ~ -
MK 801 binding
a

-Spermidme

*** . fig. 6. Effect of acamprosate and spermidine on

[3H]dizocilpine specific binding with (a) or without (b)
100 p M glutamate (GLU) and 30 p M glycine (GLY).
Acamprosate and spermidine both increased rH]dizo-
cilplne binding. They both significantly p < 0.001,
compared with control values) decreased PHldizo-
I 10 I00 I000 10000 100000 cilpine binding for concentrationshigher than 100 p M in
case of maximal activation of NMDA receptor (a). Con-
trol value for rH1dizocilpinespecific binding was 153 lr
16 fmol/mg of protein (fl = 4). p M concentrations of
spermidine, but not acarnprosate, significantly de-
creased PHjdizocilpine. Binding in the presence of p M

I
spermidine was 45 2 6 fmoVmg of protein ("^p <
MK 801 binding b
0,001). In the case of submaximal activation of NMDA
without glutamate and glycine receptor (b), acamprosate and spermidine significantly
( " p < 0.001, compared with control values) increased
f'Hldizocilpine binding. For higher concentrations (mM
***
300 T '
range), spermidine significantly (" p < 0.01) decreased
[3H]dizocilpine, but acamprosate did not. Control value
for rHldizociipine binding was 53 5 2 fmol/mg of pro-
-Sperniidine tein (n = 4). Results are expressed as % of specific
pH]dizocilpine binding without any drug. See "Materials
and Methods" for experimental details.

This suggests a direct interaction between acamprosate However, at higher concentrations of acamprosate, en-
and the NMDA receptor complex tending to increase chan- hancement of dizocilpine binding was more marked than in
nel opening. Spermidine induced similar effects, but this the absence of spermidine. The significance of this latter
does not necessarily implicate a spermidine receptor in the effect is unclear; it may represent a shift of the acamprosate
effects of acamprosate. However, the effects of both acam- activation curve for dizocilpine binding (Fig. 7) or an inhi-
prosate and spermidine on dizocilpine binding were com- bition of the effects of high concentrations of acamprosate
pletely prevented by arcaine, and, if arcaine constitutes a on the displacement of dizocilpine binding (Fig. 6). In any
genuine receptor antag~nist,'~.'~ then this strongly suggests case, this effect occurs at acamprosate concentrations much
that the direct effects of acamprosate on the NMDA re- higher than those likely to be achieved in vivo.17 The im-
ceptor are influenced allosterically by ligand interactions portant interactions between acamprosate and spermidine
with a spermidine receptor on the NMDA receptor com- are probably those that occur in the range lower than
plex. micromolar; based on these findings, they may be mutually
This finding suggests that acamprosate binding to the inhibitory. In the presence of spermidine, acamprosate may
ABS will affect NMDA receptor function differently, de- therefore enhance NMDA receptor activity, whereas in the
pending on the availability of spermidine. This hypothesis is absence of spermidine, the enhancing effects of acampro-
supported by results obtained for [3H]dizocilpine binding sate and spermidine may both be inhibited. More extensive
when spermidine and acamprosate were both present to- work on these potentially very important interactions is
gether. At low concentrations of both acamprosate and clearly necessary.
spermidine (<1 kM), the effect on [3H]dizocilpinebinding Moreover, the presence of more ABS than spermidine
appeared to be competitive, so that enhancement of dizo- binding sites (-500 vs. 11 pmol/mg of protein) could sug-
cilpine binding associated with these concentrations of gest that acamprosate may bind on other modulatory bind-
acamprosate in the presence of spermidine were inhibited. ing sites. It has been suggested (Littleton, personal com-
808
Fig. 7. Effect of acarnprosate with (open cir-
70 r
cles) or without (full squares) 1 pM sperrnidine
(Sp) on [3H]dizocilpine binding. Results are ex-
pressed as means -t SEM (n = 4) of % of rH]di-
zocilpine binding enhancement, compared with
control values (35 f 3 frnolhg of protein). Sper-
rnidine significantly shifted acarnprosate action
for inactive acarnprosate concentrations (<lo0
pM). The effect of sperrnidine on acarnprosate-
induced MK 801 binding increase was statisti-
cally significant ( p = 0.006, ANOVA test). At
concentrations of acarnprosate that increased
[3H]dizocilpine, sperrnidine did not affect acarn-
prosate-induced rH1dizocilpine binding en-
hancement. Binding in the presence of pM sper-
rnidine alone was 39 f 2 frnol/rng of protein.
munication) that ABS could also be present on calcium
channels, modulating the opening of these channels. On
the other hand, McConnel et a1.I’ demonstrated that sper-
midine binding sites could interfere with diverse recogni-
tion sites.
Taken together, these results strongly suggest that acam-
prosate acts on the ABS to cause an allosteric modulation
of the NMDA receptor complex. Similarly, conformational
changes of the NMDA receptor complex, particularly those
produced at the spermidine site, influence the affinity of
the ABS. The consequence of the interactions between
acamprosate and the NMDA receptor under physiological
conditions in vivo are difficult to extrapolate from these
data, but acamprosate is probably best characterized as a
“partial agonist” at the receptor, exerting its major inter-
actions at the receptor complex on a spermidine sensitive
site. However, spermidine can have excitatory or inhibitory
effects at various sites of the NMDA receptor,” and it is far
from certain which of these sites is modulated by acampro-
sate. Overall, data predict that acamprosate should en-
hance NMDA receptor activity when this is low and inhibit
activity when it is high, particularly when high levels of
activity are a consequence of activation by polyamines.
Whether these effects of acamprosate have any relevance
to the clinical use of the drug in prevention of relapse in
alcohol-dependent patients remains uncertain. However,
the mechanism is at least consistent with an anticraving
effect in ethanol withdrawal.lo The intriguing possibility
exists that the ABS on the NMDA receptor complex is even
more directly concerned with the action of ethanol. Thus,
although it has been suggested that ethanol inhibits the
activation of NMDA receptors by modulation of the glycine
site,*” it is now acknowledged that this is not true for all
NAASSllA ET AL.
neurons.8 The accompanying paper by a1 Quatari et al.‘l
shows that administration of either acamprosate or ethanol
in vivo causes a loss of the enhancing effect on dizocilpine
binding of acamprosate in vitro. This suggests that acam-
prosate and ethanol cause very similar adaptative changes
in the NMDA receptor, and a potential mechanism consists
of an increase in NRlb splice variants reported to be
caused by both drugs.” These findings suggest that ethanol
and acamprosate may have similar targets within the
NMDA receptor complex, and it is now essential to inves-
tigate this possibility by radioligand binding and functional
studies. Close interactions of this type could certainly help
to explain some of the clinical effects of acamprosate in
preventing relapse in alcohol-dependent patients. The ef-
fects of ethanol dependence itself on the molecular effects
of acamprosate are also of importance herein, and some of
these are considered in the accompanying paper.
Moreover, additional studies have to be done to know
the exact role of acamprosate on the NMDA receptor
functioning.
In conclusion, at least one binding site of acamprosate on
brain membranes is spermidine-sensitive, and probably di-
rectly modulates the NMDA receptor complex. This bind-
ing site is probably not a “spermidine receptor” on the
NMDA receptor complex, but it modulates this site allos-
terically and can in turn be modulated by spermidine bind-
ing. The consequences for NMDA receptor function are
likely to be complex, with acamprosate being best described
as a “partial co-agonist” for the NMDA receptor. This
action may be relevant to its interactions with ethanol and
the NMDA receptor, and with its clinical use in relapse
prevention.
MECHANISM OF ACTION OF ACAMPROSATE. PART I
ACKNOWLEDGMENT
We thank John Littleton for his helpful discussion and Doctor
Bonhomme from Groupe Lipha.
REFERENCES
1. Sass H: Calcium acetylhomotaurinate: Results of multicenter stud-
ies. Pharmacopsychiatry 26:194, 1993
2. Boismarc F, Daoust M, Moore N, Saligaut C, Chadclaud M, Chre-
tien P, Durlach J, Lhuintrc JP: A homotaurine derivative reduces the
voluntary intake of cthanol by rats: Are cerebral GABA receptors in-
volved? Pharmacol Biochem Behavior 21:787-789, 1984
3. Spanagel R, Holtcr SM, Allingham K, Landgraf R, Ziegelgan-
sberger W: Acamprosate and alcohol. I. Effects on alcohol intake follow-
ing alcohol deprivation in the rat. Eur J Pharmacol 305:39-44, 1996
4. Spanagel R, Putzkc J, Stefferl A, Schobitz B, Ziegelgansberger W
Acamprosatc and alcohol. 11. Effects on alcohol withdrawal in the rat. Eur
J Pharmacol 305:45-50, 1996
5. Zeise ML, Kasparov S, Capogna M, Ziegelgansberger W: Acam-
prosatc (calcium acetylhomotaurinatc) decreases postsynaptic potentials
in the rat neocortcx: Possible involvement of excitatory amino acid recep-
tors. Eur J Pharmacol 231:47-52, 1993
6. Lovingcr DM, White G, Weight F F Ethanol inhibits NMDA-
activatcd ion current in hippocampal neurons. Science 243:1721-1724,
1989
7. Takedera T, Suzuki R, Mohri T Protection by cthanol of cortical
neurons from N-methyl-D-aspartate-inducedneurotoxicity is associated
with blocking calcium influx. Brain Res 537:109-114, 1990
8. Hue X J , Ticku M K Chronic ethanol treatment upregulates the
NMDA receptor function and binding in mammalian cortical neurons.
Mol Brain Res 30:347-356, 1995
9. Grant K, Snell L, Tabakoff B: NMDA receptor antagonists and the
suppression of ethanol withdrawal seizures. Alcohol Clin Exp Res 15:332,
1991
10. Littleton J: Acamprosate in alcohol depcndcnce: How does it
work? Addiction 90:1179-1188, 1995
809
11. Yoneda Y, Ogita K Novel fourth binding site of ['Hlspermidine
within the NMDA receptor complex, in Kino S, et al (eds): Neuroreceptor
Mechanisms in Brain. New York, Plenum Press, 1991, pp 455-475
12. Gilad GM, Gilad V A Polyamine uptake, binding and release in rat
brain. Eur J Pharmacol 193:41-46, 1991
13. Mantione CR, Demirgoren S, London ED: Specific binding of
[3H]spermidine to membranes of rat brain. Eur J Pharmacol 180:393-394,
1990
14. Schocmaker H, Allen J, Langer SZ: Binding of ["Hlifenprodil, a
novel NMDA antagonist, to a polyamhe-sensitive site in the rat cerebral
cortex. Eur J Pharmacol 176249-250, 1990
15. Reynolds IJ: Arcainc is a competitive antagonist of the polyamine
site on the NMDA receptor. Eur J Pharmacol 177:215-216, 1990
16. Reynolds IJ: Arcaine uncovers dual interactions of polyamines with
the N-methyl-D-aspartate receptor. J Pharmacol Exp Ther 255:lOOl-1007,
1990
17. Durbin PH, Belleville M, Chabac S: Evidence of acamprosate
penetration into the rat brain. Behav Pharmacol 6:620, 1995
18. McConnell SE, D'Souza SW, Slater P: A polyamine binding site
labelled with ['Hlspermidine in developing human cerebellum. Eur
J Pharmacol 262:293-296, 1994
19. Enomoto R, Ogita K, Han D, Yoneda Y: Differential potentiation
by spermidine of abilities of a variety of displacers for [3H]-MK 801
binding in hippocampal synaptic membranes. Neurosci Res 16:217-224,
1993
20. Chu B, Anantharam V, Treistman SN: Ethanol inhibition of re-
combinant heteromeric NMDA channels in the presence and absence of
modulators. J Neurochem 65:140-148, 1995
21. al Quatari M, Bouchenafa 0, Littleton J: Ethanol dependence
modifies effects of acamprosate on NMDA receptor binding in mem-
branes from rat cerebral cortex. Alcohol Exp Clin Res (in press).
22. Putzke J, Spanagel R, Toelle TR, Ziegelgansberger W: The novel
anti-craving drug acamprosate alters the expression of NMDA receptor
splice variant mRNAs in the rat brain. J Neural Transm 103:XLV-XLXI,
1996