Cannabidiol is an allosteric modulator at mu- and delta-opioid receptors

Abstract

The mechanism of action of cannabidiol, one of the major constituents of cannabis, is not well understood but a noncompetitive interaction with mu opioid receptors has been suggested on the basis of saturation binding experiments. The aim of the present study was to examine whether cannabidiol is an allosteric modulator at this receptor, using kinetic binding studies, which are particularly sensitive for the measurement of allosteric interactions at G protein-coupled receptors. In addition, we studied whether such a mechanism also extends to the delta opioid receptor. For comparison, (-)-Δ9- tetrahydrocannabinol (THC; another major constituent of cannabis) and rimonabant (a cannabinoid CB1 receptor antagonist) were studied. In mu opioid receptor binding studies on rat cerebral cortex membrane homogenates, the agonist 3H-DAMGO bound to a homogeneous class of binding sites with a KD of 0.68± 0.02 nM and a Bmax of 203±7 fmol/mg protein. The dissociation of H-DAMGO induced by naloxone 10 μM (half life time of 7±1 min) was accelerated by cannabidiol and THC (at 100 μM, each) by a factor of 12 and 2, respectively. The respective pEC 50 values for a half-maximum elevation of the dissociation rate constant koff were 4.38 and 4.67; 3H-DAMGO dissociation was not affected by rimonabant 10 μM. In delta opioid receptor binding studies on rat cerebral cortex membrane homogenates, the antagonist H-naltrindole bound to a homogeneous class of binding sites with a KD of 0.24±0.02 nM and a Bmax of 352± 22 fmol/mg protein. The dissociation of 3H-naltrindole induced by naltrindole 10 μM (half life time of 119±3 min) was accelerated by cannabidiol and THC (at 100 μM, each) by a factor of 2, each. The respective pEC50 values were 4.10 and 5.00; 3H-naltrindole dissociation was not affected by rimonabant 10 μM. The present study shows that cannabidiol is an allosteric modulator at mu and delta opioid receptors. This property is shared by THC but not by rimonabant.

Full text

Naunyn-Schmiedeberg’s Arch Pharmacol
DOI 10.1007/s00210-006-0033-x
ORIGINAL ARTICLE
Markus Kathmann
.
Karsten Flau
.
Agnes Redmer
.
Christian Tränkle
.
Eberhard Schlicker
Cannabidiol is an allosteric modulator at mu- and delta-opioid
receptors
Received: 4 October 2005 / Accepted: 4 January 2006
# Springer-Verlag 2006
Abstract The mechanism of action of cannabidiol, one of
the major constituents of cannabis, is not well understood
but a noncompetitive interaction with mu opioid receptors
has been suggested on the basis of saturation binding
experiments. The aim of the present study was to examine
whether cannabidiol is an allosteric modulator at this
receptor, using kinetic binding studies, which are particu-
larly sensitive for the measurement of allosteric interactions
at G protein-coupled receptors. In addition, we studied
whether such a mechanism also extends to the delta opioid
receptor. For comparison, (-)-Δ
9
-tetrahydrocannabinol
(THC; another major constituent of cannabis) and rimona-
bant (a cannabinoid CB
1
receptor antagonist) were studied.
In mu opioid receptor binding studies on rat cerebral cortex
membrane homogenates, the agonist
3
H-DAMGO bound to
a homogeneous class of binding sites with a K
D
of 0.68±
0.02 nM and a B
max
of 203±7 fmol/mg protein. The
dissociation of
3
H-DAMGO induced by naloxone 10 μM
(half life time of 7±1 min) was accelerated by cannabidiol
and THC (at 100 μM, each) by a factor of 12 and 2,
respectively. The respective pEC
50
values for a half-
maximum elevation of the dissociation rate constant k
off
were 4.38 and 4.67;
3
H-DAMGO dissociation was not
affected by rimonabant 10 μM. In delta opioid receptor
binding studies on rat cerebral cortex membrane homoge-
nates, the antagonist
3
H-naltrindole bound to a homoge-
neous class of binding sites with a K
D
of 0.24±0.02 nM and
aB
max
of 352± 22 fmol/mg protein. The dissociation of
3
H-
naltrindole induced by naltrindole 10 μM (half life time of
119±3 min) was accelerated by cannabidiol and THC (at
100 μM, each) by a factor of 2, each. The respective pEC
50
values were 4.10 and 5.00;
3
H-naltrindole dissociation was
not affected by rimonabant 10 μM. The present study
shows that cannabidiol is an allosteric modulator at mu and
delta opioid receptors. This property is shared by THC but
not by rimonabant.
Keywords
3
H-DAMGO binding
.
3
H-Naltrindole
binding
.
(-)-Δ
9
-Tetrahydrocannabinol
.
Cannabinoid CB
1
receptor
.
Rat cerebral cortex
.
Rimonabant
Abbreviations AEBSF: 4-(2-aminoethyl)
benzenesulfonyl fluoride
.
3
H-DAMGO:
3
H-Tyr-D-Ala-
Gly-N-methyl-Phe-Gly-ol
.
3
H-NTI:
3
H-naltrindole
.
CBD: (-)-cannabidiol
.
THC: (-)-Δ
9
-tetrahydrocannabinol
Introduction
(-)-Δ
9
-Tetrahydrocannabinol (THC) is the main agent of the
pharmacological effects caused by the consumption of
cannabis and plays a major role both in its use for rec-
reational and medicinal purposes. However, the non-
psychotropic cannabidiol (CBD), several cannabinoid
analogues and newly discovered modulators of the endog-
enous cannabinoid system, so-called endocannabinoids (e.g.
anandamide; for review, see Pertwee 1999;Howlettet
al. 2002), are also promising candidates for clinical research
and therapeutic uses. Cannabinoids exert many effects
through activation of G-protein-coupled cannabinoid recep-
tors in the brain and peripheral tissues (for review, see
Ameri 1999; Schlicker and Kathmann 2001;Howlettet
al. 2002). To date, two types of cannabinoid receptors, CB
1
and CB
2
, have been identified, for both of which THC has
marked affinity. On the other hand, cannabidiol has a very
low affinity for either cannabinoid receptor and little is
M. Kathmann (*)
.
K. Flau
.
A. Redmer
.
E. Schlicker
Department of Pharmacology and Toxicology,
School of Medicine,
University of Bonn,
Reuterstr. 2b,
53113 Bonn, Germany
e-mail: m.kathmann@uni-bonn.de
Fax: +49-228-735404
C. Tränkle
Department of Pharmacology and Toxicology,
Institute of Pharmacy,
University of Bonn,
Gerhard-Domagk-Str. 3,
53121 Bonn, Germany

known so far about its mechanism of action (for review, see
Pertwee 1999; Howlett et al. 2002).
Earlier findings indicate an antianxiety (Guimaraes et
al. 1994), neuroprotective and anticonvulsant effect of
CBD (Consroe et al. 1981; Martin et al. 1987); CBD also
inhibits the release of inflammatory cytokines from blood
cells (Srivastava et al. 1998; Malfait et al. 2000). There is
evidence for receptor-independent mechanisms of CBD
including its antioxidant properties or its direct interaction
with cytochrome P450 enzymes (Hampson et al. 1998;
Bornheim and Correia 1989). The effect of CBD may also
involve stimulation of VR1 receptors or an increase in the
level of endogenous anandamide (Bisogno et al. 2001)or
may be mediated through an unknown specific receptor.
Very recently, Health Canada has approved Sativex
(Cannabis sativa L. extract; the ratio of THC to CBD is
2.7 mg : 2.5 mg per spray), a new drug developed by GW
Pharmaceuticals; this drug proved successful as adjunctive
treatment for the symptomatic relief of neuropathic pain in
adults with multiple sclerosis (Rog et al. 2005).
In the present study, the possibility that CBD possesses
an allosteric effect on ligand binding to mu and delta opioid
receptors (Vaysse et al. 1987) has been further examined,
using kinetic binding studies with
3
H-DAMGO and
3
H-
naltrindole, respectively. In such experiments, the dissocia-
tion of the radioligand from the orthosteric binding site is
induced by a high concentration of an orthosteric ligand
and a putative allosteric ligand alters the velocity of
dissociation (Christopoulos and Kenakin 2002). For the
sake of comparison, THC and the selective CB
1
antagonist
rimonabant (previous name SR 141716) were also included
in our study.
Materials and methods
Cell membrane preparation
Cerebral cortical tissue from male Wistar rats was homo-
genized (Potter-Elvehjem) in 25 volumes of ice-cold Tris-
HCl-EDTA buffer (TE-buffer: Tris 50 mM; EDTA 5 mM;
pH 7.5) containing 10.27% sucrose and centrifuged at
1,000×g for 10 min (4°C). The supernatant was centrifuged
at 35,000×g for 10 min (4°C) and the pellet was washed
twice with TE-buffer. Finally the pellet was resuspended in
TRIS-buffer (Tris 50 mM; pH 7.4) and frozen at −80°C.
Protein concentration was assayed by the method described
by Bradford (1976).
3
H-DAMGO binding assay
To determine mu opioid receptor binding, we used the
agonist radioligand
3
H-DAMGO (Zhao et al. 2003). In
saturation binding experiments, seven concentrations of
3
H-DAMGO (0.1 nM to 10 nM) were used in a final
volume of 0.5 ml TRIS-buffer containing 10 μM AEBSF
(serine proteinase inhibitor; AEBSF was also used in the
other types of binding experiments with
3
H-DAMGO).
Incubation was performed at 25°C and terminated after
60 min by rapid filtration through polyethylenimine
(0.3%)-pretreated Whatman GF/C filters. Naloxone
(10 μM) was used to determine the non-specific binding
(here and also in pseudo-competition and association
binding experiments; 16% at 0.5 nM).
Displacement experiments (pseudo-competition experi-
ments) with
3
H-DAMGO (0.5 nM) were performed in
TRIS-buffer in a final volume of 0.5 ml containing various
concentrations of the drugs under study. Incubation was
performed at 25°C and terminated after 120 min by rapid
filtration through polyethylenimine (0.3%)-pretreated
Whatman GF/C filters.
Kinetic experiments were performed in TRIS-buffer
containing
3
H-DAMGO (0.5 nM). To study
3
H-DAMGO
association kinetics, assays were prepared in a larger volume
of TRIS-buffer allowing withdrawal of several aliquots of
0.5 ml at adequate time intervals over a period of up to
120 min.
3
H-DAMGO dissociation kinetics: Radioligand
and membranes were incubated for 45 min at 25°C and then
naloxone (10 μM) or a combination of naloxone and the
drugs under study was added. Assays were carried out in a
larger volume of TRIS-buffer allowing withdrawal of several
aliquots of 0.5 ml at adequate time intervals over a period of
up to 120 min.
3
H-NTI binding assay
To determine delta opioid receptor binding, we used the
antagonist radioligand
3
H-naltrindole (
3
H-NTI) (Contreras et
al. 1993). In saturation binding experiments, eight concen-
trations of
3
H-NTI (0.05 nM to 5 nM) were used in a final
volume of 0.5 ml TRIS-buffer. Incubation was performed at
37°C and terminated after 120 min by rapid filtration through
polyethylenimine (0.3%)-pretreated Whatman GF/C filters.
Naltrindole (10 μM) was used to determine the non-specific
binding (here and also in pseudo-competition and associa-
tion binding studies; 22% at 0.1 nM).
Displacement experiments (pseudo-competition experi-
ments) with
3
H-NTI (0.1 nM) were performed in TRIS-buffer
in a final volume of 0.5 ml containing various concentrations
of the drugs under study. Incubation was performed at 37°C
and terminated after 120 min by rapid filtration through
polyethylenimine (0.3%)-pretreated Whatman GF/C filters.
Kinetic experiments were performed in TRIS-buffer
containing
3
H-NTI (0.1 nM). To study
3
H-NTI association
kinetics, assays were prepared in a larger volume of TRIS-
buffer allowing withdrawal of several aliquots of 0.5 ml at
adequate time intervals over a period of up to 180 min.
3
H-
NTI dissociation kinetics: Radioligand and membranes
were incubated for 120 min at 37°C and then naltrindole
(10 μM) or a combination of naltrindole and the drugs
under study was added. Assays were carried out in a larger
volume of TRIS-buffer allowing withdrawal of several

aliquots of 0.5 ml at adequate time intervals over a period
of up to 300 min.
Statistics and calculations
Results are given as means±SEM of n experiments.
Experimental data from the individual binding experiments
were analyzed by computer-aided, nonlinear regression
analysis using Prism software (Vers. 3.0, Gr aph Pad, San
Diego, Calif., USA). Effects of the allosteric agents on
3
H-DAMGO and
3
H-NTI equilibrium binding were
analyzed by nonlinear regression analysis applying a
four parameter logistic equation including a slope factor
n
H
.TheF-test wa s applied in order to evaluate
successively whether the inhibition data of
3
H-DAMGO
or
3
H-NTI binding by the test drugs (1) occurred with
slopes different from unity and (2) if n
H
<1.00, were better
fitted by a c ompetition one-site or a two-site model.
Dissociation data were fitted applying a monoexponential
compared to a double exponential decay function (F-test).
P<0.05 was taken as criterion for statistical significance.
Drugs used
[Tyrosyl-3,5-
3
H(N)]-DAMGO (Tyr*-D-Ala-Gly-N-meth-
yl-Phe-Gly-ol;
3
H-DAMGO, specific activity 51 Ci/
mmol), [3′,7′-
3
H]-naltrindole (
3
H-NTI, specific activity
20 Ci/ mmol) (PerkinElmer, Boston, MA, USA); AEBSF
(4-(2-aminoethyl)benzenesulfonyl fluoride), (-)-Δ
9
-tetrahy-
drocannabinol (Sigma, München, Germany); (-)-cannabid-
iol (Sigma, München, Germany or GW Pharmaceuticals,
Salisbury, England); DAMGO, naltrindole hydrochloride
(Bachem, Weil am Rhein, Germany); rimonabant (N-
piperidino-5-(4-chlorophenyl)-1-(2,4-dichlorophenyl)-4-meth-
yl-3-pyrazole-carboxamide; SR141716, Sanofi-Aventis,
Montpellier, France). Drugs were dissolved in DMSO
(cannabidiol, rimonabant, THC) or water (other drugs) and
diluted with water to obtain the concentration required.
Results
3
H-DAMGO binding
In saturation binding experiments (25°C) on rat brain cortex
membranes using seven concentrations of the radioligand
3
H-DAMGO and naloxone (10 μM) (to determine non-
specific binding), a K
D
value of 0.68±0.02 nM with a
maximum number of binding sites (B
max
) of 203±7 fmol/mg
protein was determined (Fig. 1a). Scatchard analysis revealed
a straight line (Fig. 1a, inset); the coefficient (n
H
) obtained by
Hill analysis was not different from unity (not shown).
To determine association kinetics, rat brain cortical
membranes were exposed to 0.5 nM
3
H-DAMGO over a
period of up to 120 min, yielding a monoexponential
association time course with a half life time t
1/2on
of 2.69±
0.09 min and an apparent rate constant k
appon
of 0.26±
0.01 min
−1
(Fig. 2a).
To study dissociation kinetics, rat brain cortical mem-
branes were incubated with 0.5 nM
3
H-DAMGO for 45
min and then naloxone (10 μM) was added to visualize
radioligand dissociation yielding a monoexponential time
course with a half life time t
1/2off
of 7.0±0.9 min and a
dissociation rate constant k
off
of 0.11±0.02 min
−1
(Fig. 2b).
To evaluate the effects of the test compounds, naloxone
10 μM was added together with 30 μ M THC, 30 μM
cannabidiol or 10 μM rimonabant, respectively (Fig. 3).
Both cannabidiol 30 μM and THC 30 μM accelerated
3
H-
DAMGO dissociation, whereas rimonabant 10 μM had no
effect. The accelerating effects of cannabidiol and THC on
the dissociation of
3
H-DAMGO were further investigated
using different concentrations. The dissociation of
3
H-
DAMGO induced by naloxone 10 μM was accelerated by
cannabidiol and THC in a concentration-dependent manner
(Fig. 3, inset). The latter drug increased the rate constant of
3
H-DAMGO dissociation maximally by a factor of 2; the
respective pEC
50
value was 4.67. In the case of cannabi-
diol, 100 μM, the highest concentration under study,
accelerated
3
H-DAMGO control dissociation about 12 fold
but did not describe the maximum increase. The pEC
50
value of 4.38 may, therefore, overestimate the potency of
the drug to accelerate dissociation.
In displacement experiments (pseudo-competition ex-
periments), rat brain cortex membranes were incubated for
120 min with medium containing 0.5 nM
3
H-DAMGO and
various concentrations of the drugs under study. Specific
3
H-DAMGO binding was inhibited monophasically by
cannabidiol, THC and rimonabant with pIC
50
values of vav
5.01±0.03 (n
H
=−1.46), 4.49±0.04 (n
H
=−1.60) and 5.39±
0.07 (n
H
=−0.73), respectively (Fig. 4a). All slope factors
were significantly different from unity (F-test, P<0.05).
Though inhibition of
3
H-DAMGO binding by rimonabant
occurred with a shallow slope, a two site competition
model did not yield a better curve fit than a one site model
(F-test, P>0.05). Note that cannabidiol was able to fully
suppress
3
H-DAMGO binding; for THC and rimonabant,
the maximum inhibitory effects could not be determined.
In an additional series of experiments, the alteration of
equilibrium binding by cannabidiol, THC and rimonabant
was studied. The experimental design was similar to that
used for the experiments in which dissociation kinetics
were examined, however, naloxone was omitted. The
dissociation curve for naloxone 10 μM (obtained from
Fig. 3) is depicted once again in Fig. 5a. Compared to this
curve, the curve for cannabidiol and THC, 100 μM each
was shifted to the left but showed the same plateau
(Fig. 5a). The curve for rimonabant 10 μM was shifted to
the right and the maximum level of binding inhibition was
lower than that for naltrindole (Fig. 5a).
3
H-NTI binding
In saturation binding experiments (37°C) on rat brain
cortex membranes using eight concentrations of the

radioligand
3
H-NTI and naltrindole (10 μM) (to determine
non-specific binding), a K
D
value of 0.24±0.02 nM with a
maximum number of binding sites (B
max
) of 352±22 fmol/
mg protein was determined (Fig. 1b). Scatchard analysis
revealed a straight line (Fig. 1b, inset), Hill analysis yielded
a Hill coefficient (n
H
) not different from unity (not shown).
To study association kinetics, rat brain cortical mem-
branes were exposed to 0.1 nM
3
H-NTI over a period of up
to 180 min, yielding a monoexponential association time
course with a half life time t
1/2on
of 27.8±1.2 min and an
apparent rate constant k
appon
of 0.025±0.001 min
−1
(Fig. 2c).
To determine dissociation kinetics, rat brain cortical
membranes were incubated with 0.1 nM
3
H-NTI for
120 min and then naltrindole (10 μM) was added to
visualize radioligand dissociation yielding a monoexpo-
nential time course with a dissociation half life time t
1/2off
of 118.6±3.1 min and a dissociation rate constant k
off
of
0.0059±0.0002 min
−1
(Fig. 2d). In additional experiments,
naltrindole (10 μM) was added together with 100 μM
THC, 100 μM cannabidiol or 10 μM rimonabant (Fig. 6).
Both cannabidiol 100 μM and THC 100 μM accelerated
the dissociation of
3
H-NTI, whereas rimonabant 10 μM
had no effect on the dissociation kinetics. The effects of
cannabidiol and THC on the increase in
3
H-NTI dissocia-
tion kinetics were further investigated using different
concentrations. The dissociation of
3
H-NTI induced by
naltrindole was increased by either drug (Fig. 6, inset).
THC increased the k
off
value maximally by 70% (pEC
50
value 5.00). The k
off
value was increased by cannabidiol
100 μM by about the same amount, however, this was not
Fig. 2 Association (a, c) and dissociation (b, d) kinetics of
3
H-
DAMGO (a, b) and
3
H-NTI (c, d) binding to rat brain cortical
membranes. (a, b) For
3
H-DAMGO association kinetics, incubation
was performed at 25°C for up to 120 min (naloxone 10 μM was
used to determine non-specific binding); for
3
H-DAMGO dissocia-
tion kinetics, incubation was performed for 45 min at 25°C and then
naloxone (10 μM) was added to visualize dissociation. (c, d) For
3
H-NTI association kinetics, incubation was performed at 37°C for
up to 180 min (naltrindole 10 μM was used to determine non-
specific binding); for
3
H-NTI dissociation kinetics, incubation was
performed for 120 min at 37°C and then naltrindole (10 μM) was
added to induce dissociation. Means from four experiments (in
triplicate) are shown (for some data points, SEM is contained within
the symbol)
Fig. 1 Saturation of specific
3
H-DAMGO (a) and specific
3
H-NTI
(b) binding to rat brain cortex membranes. (a) Incubation was
performed at 25°C and lasted for 60 min and naloxone (10 μM) was
used to determine non-specific binding. (b) Incubation was per-
formed at 37°C and lasted for 120 min and naltrindole (10 μM) was
used to determine non-specific binding. Scatchard analysis of the
saturation data is presented in the respective insets. Means from
three experiments (in triplicate) are shown (for some data points,
SEM is contained within the symbol)

yet the maximum effect. For this reason, the calculated
pEC
50
value (4.10) should represent an overestimation.
In displacement experiments (pseudo-competition ex-
periments), rat brain cortical membranes were incubated
for 120 min with medium containing 0.1 nM
3
H-NTI and
various concentrations of the drugs under study. Specific
3
H-NTI binding was inhibited by cannabidiol at a single
site with a pIC
50
of 4.97±0.10 (n
H
=-0.83). THC (up to
100 μM) and rimonabant (up to 10 μM) inhibited
3
H-NTI
binding by no more than 20% (Fig. 4b).
In the final series of experiments, the alteration of
equilibrium binding by cannabidiol and THC was studied.
The experimental design was similar to that used for the
experiments in which dissociation kinetics were examined,
however , (unlabelled) naltrindole was omitted. The dissocia-
tion curve for naltrindole 10 μM (obtained from Fig. 6)is
depicted once again in Fig. 5b. Compared to this curve, the
Fig. 4 Effects of (-)-Δ
9
-tetrahydrocannabinol (THC), cannabidiol
(CBD) and rimonabant (SR) on
3
H-DAMGO (a)or
3
H-NTI (b)
equilibrium binding to rat cerebral cortical membranes. Unspecific
binding was determined using naloxone 10 μM(a) and naltrindole
10 μM(b), respectively. Means±SEM from 4 experiments in
triplicate. Error bars are not shown when they are smaller than the
symbols
Fig. 5 Time-dependent alteration of equilibrium binding of
3
H-
DAMGO (a) and
3
H-NTI (b) to rat brain cortical membranes by
rimonabant (SR), (-)-Δ
9
-tetrahydrocannabinol (THC) and/or canna-
bidiol (CBD). Incubation was performed for 45 min at 25°C (a) and
for 120 min at 37°C (b) and then SR, THC or CBD was added. For
the sake of comparison, the dissociation curves for naloxone 10 μM
(a; obtained from Fig. 3) and for naltrindole 10 μM(b; obtained
from Fig. 6) are shown here again. Means±SEM from four
experiments in triplicate. Error bars are not shown when they are
smaller than the symbols
Fig. 3 Dissociation kinetics of
3
H-DAMGO binding to rat brain
cortical membranes. Incubation was performed for 45 min at 25°C
and then naloxone (10 μM) was added alone or in the presence of
30 μM (-)-Δ
9
-tetrahydrocannabinol (THC), 30 μM cannabidiol
(CBD) or 10 μM rimonabant (SR). Inset: Increase in
3
H–DAMGO
dissociation rate constant K
off
by various concentrations of THC and
CBD. Means±SEM from four experiments in triplicate. Error bars
are not shown when they are smaller than the symbols

curve for cannabidiol 100 μM was slightly shifted to the left
but showed the same plateau (Fig. 5b). The curve for THC
50 μM was shifted to the right and the maximum level of
binding inhibition was lower than that for naltrindole (Fig. 5b).
Discussion
The aim of the present study was to further examine the
postulated allosteric effect of CBD at mu opioid receptors
(Vaysse et al. 1987) and to check whether this drug
possesses an allosteric effect also at delta opioid receptors.
THC and rimonabant (previous name SR 141716) were
also included in this study. The latter drug was chosen since
it is chemically unrelated to CBD and is well-characterized
as a selective CB
1
receptor antagonist or, more precisely,
inverse agonist (Pertwee 2005). THC was used since it
showed a noncompetitive interaction with mu and delta
(but not with kappa) opioid receptors in saturation
experiments (Vaysse et al. 1987). In detail, the latter
authors found that THC did not affect the affinity but
decreased the density of mu opioid receptors (labelled by
3
H-dihydromorphine) and delta opioid receptors (labelled
by
3
H-D-Pen
2
, D-pen
5
-enkephalin (
3
H-DPDPE)). We
wanted to further elucidate possible allosteric effects in
kinetic binding experiments. Unlike Vaysse et al. (1987),
who used the whole rat brain, we employed membranes
from the cerebral cortex since both mu and delta opioid
receptors exhibit a higher density in the cortex when
compared to many other brain regions (Dhawan et
al. 1996). For labelling of mu opioid receptors we used
the agonist radioligand
3
H-DAMGO, which possesses a
threefold higher affinity at this receptor than
3
H-dihydro-
morphine (Ulibarri et al. 1987). Although we would have
preferred to use
3
H-DPDPE for labelling of delta opioid
receptors, its extremely high unspecific binding (found in
initial experiments and also described by Akiyama et
al. 1985) led us to use the antagonist radioligand
3
H-
naltrindole (
3
H-NTI) instead.
Saturation binding studies revealed that both radioli-
gands bound to a single class of receptors with K
D
and
B
max
values comparable to results obtained from other
groups (Zhao et al. 2003; Contreras et al. 1993). Much
emphasis was put in the present study on kinetic binding
experiments. Association studies were carried out, which,
however (although of interest for the design of other types
of binding studies), are less suited for the identification of
allosteric ligands since a binding of the drug under
consideration to the orthosteric binding site cannot be
excluded. In contrast, in dissociation studies, the orthos-
teric binding site is occupied by a very high concentration
of a competitive antagonist and the drug under study has to
bind to another (i.e., allosteric) binding site (Christopoulos
and Kenakin 2002).
Dissociation of
3
H-DAMGO from the mu opioid
receptor, visualized by naloxone 10 μM, was not affected
by rimonabant but accelerated by cannabidiol and THC.
Our data demonstrate that the latter two drugs are allosteric
modulators of ligand binding kinetics at mu opioid
receptors, thus extending the findings obtained by Vaysse
et al. (1987) using a different experimental approach. THC
increased the dissociation of
3
H-DAMGO by a factor of 2;
cannabidiol increased the dissociation markedly at least by
a factor of 12 (the highest concentration feasible for
investigation in this study). The allosteric effect of CBD
might explain that this drug at 10 μM slightly shifted to the
right the concentration-response curve of DAMGO for its
inhibitory effect on the electrically induced twitch response
in the mouse vas deferens (Pertwee et al. 2002).
In addition, cannabidiol and THC accelerated dissocia-
tion of
3
H-NTI from delta opioid receptors, visualized by
NTI 10 μM, whereas rimonabant failed to do so. These
results show that cannabidiol and THC are allosteric
modulators also of ligand binding to the delta opioid
receptor. Thus, a noncompetitive interaction with the delta
opioid receptor as postulated earlier for THC by Vaysse et
al. (1987) was demonstrated here for the first time both for
THC and cannabidiol. Both drugs increased the dissocia-
tion about twofold although for cannabidiol again the
maximum could not be determined precisely in the con-
centration range feasible under study. An acceleration of
radioligand dissociation is also known from e.g. amiloride
Fig. 6 Dissociation kinetics of
3
H-NTI binding to rat brain cortical
membranes. Incubation was performed for 120 min at 37°C and then
naltrindole (10 μM) was added alone or in the presence of 100 μM
(-)-Δ
9
-tetrahydrocannabinol (THC), 100 μM cannabidiol (CBD) or
10 μM rimonabant (SR). Inset: Increase in
3
H-NTI dissociation rate
constant k
off
by various concentrations of THC and CBD. Means±
SEM from four experiments in triplicate. Error bars are not shown
when they are smaller than the symbols

in α
1A
-adrenoceptors (Leppik et al. 1998) and in α
2A
-
adrenoceptors (Leppik et al. 2000).
Next, the question was addressed how THC, cannabidiol
and rimonabant behave in pseudo-competition experi-
ments.
3
H-DAMGO binding was monophasically inhibited
by cannabidiol and THC. The fact that the Hill coefficients
were markedly higher than unity may be interpreted as a
positive homotropic cooperativity between the respective
inhibitor molecules. This could imply that two molecules of
cannabidiol or THC bind to an allosteric site thereby
inhibiting the binding of
3
H-DAMGO to the orthosteric
(= agonist) binding site. Such a behaviour is also known
from the action of tacrine on the binding of orthosteric
ligands in muscarinic receptors (e.g., Potter et al. 1989;
Tränkle et al. 2003). Unexpectedly,
3
H-DAMGO binding
was also inhibited by rimonabant and the possibility has to
be considered that this drug is an agonist at the mu opioid
receptor. A flat inhibition curve with a Hill coefficient lower
than unity, as found in our study, frequently occurs when an
agonist binds to a G protein-coupled receptor labelled by an
agonist radioligand (for review, see Kenakin 1993). The
low affinity of rimonabant (pIC
50
of 5.39) would not
contradict the findings by Rinaldi-Carmona et al. (1994),
who found that the IC
50
value of rimonabant at opiate
receptors is lower than 1 μM.
The effects of rimonabant, THC and cannabidiol were
also studied in binding experiments in which the
membranes were exposed to
3
H-DAMGO until equilibri-
um was reached before one of the three drugs was added.
These experiments, in which the alteration of equilibrium
binding is studied kinetically, differ from the dissociation
binding experiments inasmuch as naloxone, which serves
to visualize dissociation (= prevention of radioligand re-
association) of binding, is not used. Such experiments
allow to examine kinetically the inhibition of equilibrium
binding following the administration of a test drug. The
fact that rimonabant inhibited
3
H-DAMGO binding and
that the half life times of the curves for rimonabant and
naloxone are virtually identical is compatible with the view
that rimonabant indeed competitively interacts with
3
H-
DAMGO (Fig. 5a). On the other hand, the dissociation
curves for THC and cannabidiol were much steeper than
that for naloxone, again suggesting that the former two
drugs accelerate the dissociation of
3
H-DAMGO (Fig. 5a).
With respect to the effects of drugs on
3
H-NTI binding in
pseudo-competition experiments, the marked difference
between the curves of THC and cannabidiol is remarkable.
The fact that THC inhibited
3
H-DPDPE binding in the
study of Vaysse et al. (1987), but almost failed to inhibit
3
H-NTI binding in the present one, is, however, not so
surprising since the identification of allosteric phenomena
critically depends on the choice of the orthosteric radio-
ligand (Christopoulos and Kenakin 2002). To estimate
whether THC may also have an influence on the associ-
ation of
3
H-NTI, experiments were performed in which the
alteration of equilibrium binding is studied kinetically. In
this additional set of binding experiments (Fig. 5b),
cannabidiol (100 μM) and THC (50 μM) were used at
concentrations that accelerate the dissociation of
3
H-NTI
binding induced by naltrindole to the same extent (Fig. 6),
equivalent to an increase in k
off
of about 70% (Fig. 6,
inset). The experiments of Fig. 5b show that the time
course for cannabidiol 100 μM is similar to the curve for
naltrindole 10 μM whereas the time course for THC 50 μM
is slower (and shows a lower maximum level of inhibition).
The data are compatible with the conclusion that THC not
only increases dissociation but, in addition, increases as-
sociation and thereby weakens its inhibitory effect on
equilibrium binding.
Finally, the question has to be addressed whether the
allosteric effect of cannabidiol at the mu and delta opioid
receptor may explain its in vivo activity. This is unlikely
since a single dose oral preparation containing 100 mg of
cannabidiol yielded a plasma concentration of about 36 ng/
ml or 100 nM in man (Dr. S. Wright, GW Pharmaceuticals,
Salisbury, England, personal communication), which is
lower by a factor of about 100 than the EC
50
values
determined in the present study. In addition, it is also very
unlikely that the effects of THC at mu and delta receptors
and the effect of rimonabant at mu receptors contribute to
the overall effects of both drugs in vivo. Thus, the K
i
values
of THC and rimonabant at the cannabinoid CB
1
receptors,
which represent the molecular target of the effects of both
drugs, is about 50 nM (Pertwee 1999) and 2 nM (Rinaldi-
Carmona et al. 1994), respectively. The allosteric effects of
THC at mu and delta opioid receptors occur at a 200-fold
and the effect of rimonabant at mu opioid receptors at a
1,400-fold higher concentration.
In conclusion, our study shows that cannabidiol is an
allosteric modulator of ligand binding to mu and delta
opioid receptors. This property is shared by THC but not by
rimonabant. All effects occur at very high concentrations
and cannot be expected to contribute to the in vivo action of
the three drugs.
Acknowledgements This study was supported by grants from the
Deutsche Forschungsgemeinschaft (Schl 266/5-5 and Graduierten-
kolleg 246 TP 01). We are also indebted to Mrs. P. Zeidler for her
skilled technical assistance and to GW Pharmaceuticals and Sanofi-
Aventis for gifts of drugs.
References
Akiyama K, Gee KW, Mosberg HI, Hruby VJ, Yamamura HI (1985)
Characterization of [
3
H][2-D-penicillamine, 5-D-penicilla-
mine]-enkephalin binding to δ opiate receptors in the rat
brain and neuroblastoma-glioma hybrid cell line (NG 108-15).
Proc Natl Acad Sci USA 82:2543–2547
Ameri A (1999) The effects of cannabinoids on the brain. Prog
Neurobiol 58:315–348
Bisogno T, Hanuš L, De Petrocellis L, Tchilibon S, Ponde DE, Brandi
I, Moriello AS, Davis JB, Mechoulam R, Di Marzo V (2001)
Molecular targets for cannabidiol and its synthetic analogues:
effect on vanilloid VR1 receptors and on the cellular uptake and
enzymatic hydrolysis of anandamide. Br J Pharmacol 134:
845–852
Bornheim LM, Correia MA (1989) Effect of cannbidiol on cytochrome
P-450 isoenzymes. Biochem Pharmacol 38:2789–2794

Bradford MM (1976) A rapid and sensitive method for the
quantitation of microgram quantities of protein utilizing the
principle of protein-dye binding. Anal Biochem 72:248–254
Christopoulos A, Kenakin T (2002) G protein-coupled receptor
allosterism and complexing. Pharmacol Rev 54:323–374
Consroe P, Martin A, Singh V (1981) Antiepileptic potential of
cannabidiol analogs. J Clin Pharmacol 21:S428–S436
Contreras PC, Tam L, Drower E, Rafferty MF (1993) [3H]
naltrindole: a potent and selective ligand for labeling delta-
opioid receptors. Brain Res 604:160–164
Dhawan BN, Cesselin F, Raghub ir R, Reisine T, Bradley PB,
Portoghese PS, Hamon M (1996) International Union of Phar-
macologists. XII. Classification of opioid receptors. Pharmacol
Rev 48:567–592
Guimaraes FS, De Aguiar JC, Mechoulam R, Breuer A (1994)
Anxiolytic effect of cannabidiol derivatives in the elevated
plus-maze. Gen Pharmacol 25:161–164
Hampson AJ, Grimaldi M, Axelrod J, Wink D (1998) Cannabidiol
and (-)Δ
9
-tetrahydrocannabinol are neuroprotective. Proc Natl
Acad Sci USA 95:8268–8273
Howlett AC, Barth F, Bonner TI, Cabral G, Casellas P, Devane WA,
Felder CC, Herkenham M, Mackie K, Martin BR, Mechoulam
R, Pertwee RG (2002) International Union of Pharmacology.
XXVII. Classification of cannabinoid receptors. Pharmacol Rev
54:161–202
Kenakin T (1993) Pharmacological analysis of drug-receptor
interaction. Raven Press, New York
Leppik RA, Lazareno S, Mynett A, Birdsall NJ (1998) Character-
ization of the allosteric interactions between antagonists and
amiloride analogues at the human alpha
2A
-adrenergic receptor.
Mol Pharmacol 53:916–925
Leppik RA, Mynett A, Lazareno S, Birdsall NJ (2000) Allosteric
interactions between the antagonist prazosin and amiloride
analogs at the human alpha
1A
-adrenergic receptor. Mol Phar-
macol 57:436–445
Malfait AM, Gallily R, Sumariwalla PF, Malik AS, Andreakos E,
Mechoulam R, Feldmann M (2000) The nonpsychoactive
cannabis constituent cannabidiol is an oral anti-arthritic ther-
apeutic in murine collagen-induced arthritis. Proc Natl Acad Sci
USA 97:9561–9566
Martin AR, Consroe P, Kane VV, Shah V, Singh V, Lander N,
Mechoulam R, Srebnik M (1987) Structure-anticonvulsant
activity relationships of cannabidiol analogs. NIDA Res
Monogr 79:48–58
Pertwee RG (1999) Pharmacology of cannabinoid receptor ligands.
Curr Med Chem 6:635–664
Pertwee RG (2005) Inverse agonism and neutral antagonism at
cannabinoid CB
1
receptors. Life Sci 76:1307–1324
Pertwee RG, Ross RA, Craib SA, Thomas A (2002) (-)-Cannabidiol
antagonizes cannabinoid receptor agonists and noradrenaline in
the mouse vas deferens. Eur J Pharmacol 456:99–106
Potter LT, Ferrendelli CA, Hanchett HE, Holliefield MA, Lorenzi
MV (1989) Tetrahydroaminoacridine and other allosteric
antagonists of hippocampal M1 muscarine receptors. Mol
Pharmacol 35:652–660
Rinaldi-Carmona M, Barth F, Heaulme M, Shire D, Calandra B,
Congy C, Martinez S, Maruani J, Neliat G, Caput D, Ferrara P,
Soubrie P, Breliere JC, Le Fur G (1994) SR 141716A, a potent
and selective antagonist of the brain cannabinoid receptor.
FEBS Lett 350:240–244
Rog DJ, Nurmikko TJ, Friede T, Young CA (2005) Randomized,
controlled trial of cannabis-based medicine in central pain in
multiple sclerosis. Neurology 65:812–819
Schlicker E, Kathmann M (2001) Modulation of transmitter release
via presynaptic cannabinoid receptors. Trends Pharmacol Sci
22:565–572
Srivastava MD, Srivastava BI, Brouhard B (1998) Delta9 tetrahy-
drocannabinol and cannabidiol alter cytokine production by
human immune cells. Immunopharmacology 40:179–185
Tränkle C, Weyand O, Voigtländer U, Mynett A, Lazareno S,
Birdsall NJ, Mohr K (2003) Interactions of orthosteric and
allosteric ligands with [
3
H]dimethyl-W84 at the common
allosteric site of muscarinic M
2
receptors. Mol Pharmacol
64:180–190
Ulibarri I, Garcia-Sevilla JA, Ugedo L (1987) Modulation of brain
α
2
-adrenoceptor and μ-opioid receptor densities during mor-
phine dependence and spontaneous withdrawal in rats. Naunyn-
Schmiedeberg’s Arch Pharmacol 336:530–537
Vaysse PJJ, Gardener EL, Zukin RS (1987) Modulation of rat brain
opioid receptors by cannabinoids. J Pharm Exp Ther 241:534–539
Zhao GM, Qian X, Schiller PW, Szeto HH (2003) Comparison of
[Dmt1]DALDA and DAMGO in binding and G protein
activation at μ, δ and κ opioid receptors. J Pharm Exp Ther
307:947–954