An aromatic microdomain at the cannabinoid CB(1) receptor constitutes an agonist/inverse agonist binding region.

Abstract

The cannabinoid CB(1) receptor transmembrane helix (TMH) 3-4-5-6 region includes an aromatic microdomain comprised of residues F3.25, F3.36, W4.64, Y5.39, W5.43, and W6.48. In previous work, we have demonstrated that aromaticity at position 5.39 in CB(1) is crucial for proper function of CB(1). Modeling studies reported here suggest that in the inactive state of CB(1), the binding site of the CB(1) inverse agonist/antagonist SR141716A is within the TMH3-4-5-6 aromatic microdomain and involves direct aromatic stacking interactions with F3.36, Y5.39, and W5.43, as well as hydrogen bonding with K3.28. Further, modeling studies suggest that in the active state of CB(1), the CB agonist WIN55,212-2 binds in this same aromatic microdomain, with direct aromatic stacking interactions with F3.36, W5.43, and W6.48. In contrast, in the binding pocket model, the CB agonist anandamide binds in the TMH2-3-6-7 region in which hydrogen bonding and C-H.pi interactions appear to be important. Only one TMH3 aromatic residue, F3.25, was found to be part of the anandamide binding pocket. To probe the importance of the TMH3-4-5-6 aromatic microdomain to ligand binding, stable transfected cell lines were created for single-point mutations of each aromatic microdomain residue to alanine. Improper cellular expression of the W4.64A was observed and precluded further characterization of this mutation. The affinity of the cannabinoid agonist CP55,940 was unaffected by the F3.25A, F3.36A, W5.43A, or W6.48A mutations, making CP55,940 an appropriate choice as the radioligand for binding studies. The binding of SR141716A and WIN55,212-2 were found to be affected by the F3.36A, W5.43A, and W6.48A mutations, suggesting that these residues are part of the binding site for these two ligands. Only the F3.25A mutation was found to affect the binding of anandamide, suggesting a divergence in binding site regions for anandamide from WIN55,212-2, as well as SR141716A. Taken together, these results support modeling studies that identify the TMH3-4-5-6 aromatic microdomain as the binding region of SR141716A and WIN55,212-2, but not of anandamide.

Full text

Page 1

An Aromatic Microdomain at the Cannabinoid CB1R eceptor Constitutes an
Agonist/Inverse Agonist Binding R egion
Sean D. McAllister,†Gulrukh Rizvi,†Sharon Anavi-Goffer,†Dow P. Hurst,‡J udy Barnett-Norris,‡
Diane L. Lynch,‡Patricia H. Reggio,‡and Mary E. Abood*,†
Forbes Norris ALS/ MDA Research Center, California Pacific Medical Center, San Francisco, California 94115, and
Department of Chemistry and Biochemistry, Kennesaw State University, 1000 Chastain Rd, Kennesaw, Georgia 30144
Received May 30, 2003
The cannabinoid CB1 receptor transmembrane helix (TMH) 3-4-5-6 region includes an
aromatic microdomain comprised of residues F3.25, F3.36, W4.64, Y5.39, W5.43, and W6.48.
In previous work, we have demonstrated that aromaticity at position 5.39 in CB1is crucial for
proper function of CB1. Modeling studies reported here suggest that in the inactive state of
CB1, the binding site of the CB1inverse agonist/antagonist SR141716A is within the TMH3-
4-5-6 aromatic microdomain and involves direct aromatic stacking interactions with F3.36,
Y5.39, and W5.43, as well as hydrogen bonding with K3.28. Further, modeling studies suggest
that in the active state of CB1, the CB agonist WIN55,212-2 binds in this same aromatic
microdomain, with direct aromatic stacking interactions with F3.36, W5.43, and W6.48. In
contrast, in the binding pocket model, the CB agonist anandamide binds in the TMH2-3-
6-7 region in which hydrogen bonding and C-H‚‚‚π interactions appear tobe important. Only
one TMH3 aromatic residue, F3.25, was found to be part of the anandamide binding pocket.
Toprobe the importance of the TMH3-4-5-6 aromatic microdomain toligand binding, stable
transfected cell lines were created for single-point mutations of each aromatic microdomain
residue to alanine. Improper cellular expression of the W4.64A was observed and precluded
further characterization of this mutation. The affinity of the cannabinoid agonist CP55,940
was unaffected by the F3.25A, F3.36A, W5.43A, or W6.48A mutations, making CP55,940 an
appropriate choice as the radioligand for binding studies. The binding of SR141716A and
WIN55,212-2 were found to be affected by the F3.36A, W5.43A, and W6.48A mutations,
suggesting that these residues are part of the binding site for these two ligands. Only the
F3.25A mutation was found to affect the binding of anandamide, suggesting a divergence in
binding siteregions for anandamidefrom WIN55,212-2, as well as SR141716A. Taken together,
these results support modeling studies that identify the TMH3-4-5-6 aromatic microdomain
as the binding region of SR141716A and WIN55,212-2, but not of anandamide.
Introduction
The cannabinoid receptor (CB1), a member of the
G-protein coupled receptor (GPCR) family, can interact
with five structurally distinct classes of compounds.
These include CB1classical cannabinoid agonists [such
as (-)-trans-∆-9-tetrahydrocannabinol, ∆9-THC], non-
classical cannabinoid agonists [such as (1R,3R,4R)-3-
[2-hydroxy-4-(1,1-dimethylheptyl)phenyl]-4-(3-hydroxy-
propyl)cyclohexan-1-ol, CP55,940], endogenous cannabi-
noid agonists (such as N-arachidonoylethanolamine,
anandamide), aminoalkylindole (AAI) agonists [such as
[2,3-dihydro-5-methyl-3-[(4-morpholinyl)methyl]pyrrolo-
[1,2,3-de]-1,4-benzoxazin -6-yl](1-naphthyl)methanone,
WIN55,212-2], and diaryl pyrazole antagonist/inverse
agonists [such as N-(piperidin-1-yl)-5-(4-chlorophenyl)-
1-(2,4-dichlorophenyl)-4-methyl-1H-pyrazole-3-carboxam-
ide, SR141716A] (see Chart 1). There is a rapidly
growing understanding of thecannabinoid system as an
important endogenous regulator of neuronal, cardiac,
reproductive, and immune function.1Therefore, there
is considerable value in understanding the specific
ligand-receptor interactions; this is particularly rele-
vant as compounds are developed for medical interven-
tion.
The CB2receptor, whose overall sequence similarity
in thetransmembranehelices (TMH) is only 68% of CB1,
has been used in chimera studies with CB1toprobe for
receptor-specific ligand interactions.2,3These studies
suggested that the TMH4-E2-TMH5 region of CB1
contains residues critical for the binding of the CB1/CB2
aminoalkylindole agonist WIN55,212-2 and the CB1-
selective diaryl pyrazole antagonist/inverse agonist
SR141716A. Mutation studies have also shown that
CB1/CB2 subtype differences in TMH34and TMH55
contribute to the CB2 selectivity of WIN55,212-2. A
recent mutant cycle study identified K 3.28 as an
important interaction sitefor SR141716A at CB1,6while
CB1Y5.39F/Y5.39I mutation studies have underscored
the structural importance of aromaticity at position
5.39.7Surprisingly, however, further elaboration of the
CB1binding site of either SR141716A or WIN55,212-2
through mutation studies has not been reported in the
literature.
In the ?2-adrenergic receptor and related aminergic
receptors, a highly conserved cluster of aromatic amino
acids is found on TMH6 that faces the binding site
* Corresponding author. Phone: (415)-923-3607. Fax: (415)-563-
7325. E-mail: mabood@cooper.cpmc.org.
†Forbes Norris ALS/MDA Research Center.
‡Kennesaw State University.
5139 J . Med. Chem. 2003, 46, 5139-5152
10.1021/jm0302647 CCC: $25.00© 2003 American Chemical Society
Published on Web 10/23/2003

Page 2

crevice (F6.44, W6.48, F6.51, and F6.52).8The CB1
receptor lacks this cluster on TMH6 (L6.44, W6.48,
L6.51, and L6.52) but contains instead a microdomain
of aromatic residues that face into the ligand binding
pocket in the TMH3-4-5-6 region, including F3.25,
F3.36, W4.64, Y5.39, W5.43, and W6.48 (Figure 1). The
importance of Y5.39 for receptor ligand recognition and
function in both CB1and CB2has been recently reported
by our group.7In this investigation, a Monte Carlo/
stochastic dynamics study of CB1in its inactive state
identified two aromatic subclusters in the TMH3-4-
5-6 region of CB1. The first cluster was a triad (F3.36/
W5.43/W6.48) involved in aromatic stacking interactions
that appeared tobe open for additional interaction with
ligand, while the second was Y5.39 and W4.64. This
latter cluster appeared to be more important for stabi-
lizing the positions of TMHs 4 and 5 in the TMH bundle
on the extracellular side.
We hypothesized that since SR141716A and WIN55,-
212-2 are both highly aromatic compounds, aromatic
Chart 1
F igure 1. A helix net representation of the sequence of the mouse CB1 receptor. The amino acids mutated in this study are
highlighted in bold.
5140J ournal of Medicinal Chemistry, 2003, Vol. 46, No. 24McAllister et al.

Page 3

stacking interactions may be important to the affinity
of each compound for the CB1receptor. Therefore, the
goal of the work presented here was toidentify through
modeling studies thosearomatic residues in theTMH3-
4-5-6 aromatic microdomain of CB1that may contrib-
ute toWIN55,212-2 and SR141716A binding and then
to experimentally test the results of these modeling
studies through mutation. Modeling and mutation
results reported here support the hypothesis that both
SR141716A and WIN55,212-2 bind within the TMH3-
4-5-6 aromatic microdomain region of CB1. These
results also reveal a divergence in the binding sites of
anandamide and CP55,940 from those of SR141716A
and WIN55,212-2.
R esults
Comparison of R and R * State Models in the
Absence of L igand. T he CB1 R State. Figure 2a
illustrates key features of the model of the CB1TMH
bundle in the inactive (R) state in the TMH3-4-5-6
region. One of the significant features of the model of
the CB1R TMH bundle is a salt bridge between K3.28-
(193) and D6.58(367) (N-O distance ) 2.6 Å; N-H-O
angle ) 159°).6This salt bridge is made possible by the
profound flexibility in TMH6 due to the presence of
G6.49 in the CWXP motif of TMH6.9The TMH3-4-
5-6 region of the R bundle in the absence of ligand is
characterized by a W6.48(357)/F3.36(201)/W5.43(280)/
Y5.39(276)/W4.64(256) aromatic cluster in which W6.48
stacks with F3.36 (d ) 5.3 Å, R ) 90°), while F3.36
stacks with W5.43 (d ) 5.6 Å, R ) 40°). W5.43 alsohas
an off-set parallel stack with Y5.39 (d ) 5.9 Å, R ) 0°),
while Y5.39 stacks with W4.64 (d ) 6.5 Å, R ) 90°) (see
the Experimental Section for definitions of d and R).
The CB1 R *State. Figure2b illustrates key features
of the model CB1TMH bundle in the active (R*) state
in theTMH3-4-5-6 region. Theconformational changes
that occur upon receptor activation result in rotations
of TMHs 3 and 6, as well as a change in the conforma-
tion of TMH6 (by moderation of its proline kink
angle).10-13In our models, both W6.48(357) and F3.36-
(201) undergoa change in their ?1values from R toR*.
?1 in W6.48(357) changes from g+ to trans and ?1 of
F3.36(201) changes from trans to g+ (see the Experi-
mental Section for the definition of ?1). In the R* TMH
bundle, the K 3.28(193) and D6.58(367) salt bridge is
broken (N-O distance ) 16.8 Å) because TMHs 3 and
6 rotate (counterclockwise from extracellular view)
during theR toR*transition (i.e. activation).14,15K3.28-
(193) has rotated away from D6.58(367) toward TMH7,
and D6.58(367) has rotated toward the TMH5-6 inter-
faceand is raised higher abovetheligand binding pocket
due tothe moderation of the TMH6 proline kink angle.
TheW6.48(357)/F3.36(201)/W5.43(280)/Y5.39(276)/W4.64-
(256) aromatic cluster present in the inactive state in
the absence of ligand also undergoes rearrangement,
with F3.36 nolonger part of this cluster. In the TMH3-
4-5-6 region of R*in the absence of ligand, W6.48 and
W5.43 form an off-set parallel aromatic stacking inter-
action with each other (d ) 4.9 Å, R ) 30°). W5.43 also
stacks with Y5.39 (d ) 6.6 Å, R ) 60°), while Y5.39
stacks with W4.64 (d ) 5.7 Å, R ) 90°). This series of
aromatic stacking interactions results in a large aro-
matic stack in R*comprised of W6.48(357)/W5.43(280)/
Y5.39(276)/W4.64(256). F3.36 (?1) g+) is not located
near an aromatic residue in the R* bundle; instead
F3.36 is bounded by V3.40(205), V3.32(197), and L6.44-
(353).
R eceptor Docking: L igand Binding Sites. In the
traditional two-statemodel for agonist action at GPCRs,
a receptor is thought to exist in two states that are in
equilibrium with each other, the ground or inactive (R)
state and the active (R*) state.16The binding of an
agonist ligand is thought toshift theequilibrium toward
R*, resulting in an increase in GDP/GTP exchange,
which initiates thesignaling cascade. In themost widely
discussed mechanism of inverse agonism, the inverse
agonist preferentially binds to the R over R* state,17
thus suppressing ligand-independent (constitutive) ac-
tivation. Figures 3, 5, and 6 illustrate the binding sites
F igure 2. Models of the CB1R and R* states in the TMH3-4-5-6 region (in the absence of ligand) are illustrated here. (a) In
the R state model, a salt bridge can form between K3.28 and D6.58 (shown in green). A microdomain of clustered aromatic residues
(shown in yellow), W6.48/F3.36/W5.43/Y5.39/W4.64 exists in the CB1TMH3-4-5-6 region. (b) The conformational changes that
occur upon receptor activation (R f R* transition) result in rotations of TMHs 3 and 6, as well as a change in the conformation
of TMH6 (by moderation of its proline kink angle).10-13The K3.28-D6.58 salt bridge is broken (residues illustrated in green).
The aromatic cluster present in the inactive state model also undergoes rearrangement, with W6.48/W5.43/Y5.39/W4.64 (shown
in yellow) maintaining an extended cluster, while F3.36 (yellow) is no longer part of this cluster.
Aromatic Microdomain at the Cannabinoid CB1ReceptorJ ournal of Medicinal Chemistry, 2003, Vol. 46, No. 24 5141

Page 4

identified for the CB1inverse agonist SR141716A in the
R state6and for the CB agonists WIN55,212-2 and
anandamide, each in the R* state.
Aromatic stacking interactions were identified for
both SR141716A and WIN55,212-2. Table 1 compares
modeling results for SR141716A6with thosefor WIN55,-
212-2. In this table, distances (d) and angles (R) for
aromatic systems that meet the criteria for aromatic
stacking interactions18are highlighted in bold.
SR 141716A in the R State. SR141716A has been
shown to act as a competitive antagonist and inverse
agonist in host cells transfected with exogenous CB1
receptor, as well as in biological preparations endog-
enously expressing CB1.19-21The SR141716A-CB1 in-
teraction presented here focuses on the R state of CB1.6
SR141716A was found tohave a single, direct aromatic
stacking interaction with F3.36(201) and with Y5.39-
(276) and two direct stacking interactions with W5.43-
(280) in CB1R (Table 1). These direct interaction sites
are colored yellow in Figure 3. The SR141716A/W5.43
interaction is the strongest, as W5.43 has aromatic
stacking interactions with both the monochloro (MC)
and dichlorophenyl (DC) rings of SR141716A. F3.36 has
a single tilted-T aromatic stacking interaction with the
DC ring of SR141716A at a close distance between ring
centroids (5.0 Å), while the single aromatic stacking
interaction of the MC ring of SR141716A with Y5.39 is
weaker, because of the longer distance between ring
centroids (6.0 Å). In binding, SR141716A becomes part
of the aromatic cluster in the TMH3-4-5-6 region. In
one of these clusters, F3.36(201) directly stacks with
W6.48 (d ) 4.9 Å, R ) 50°) and with W5.43 (d ) 5.9 Å,
R ) 60°). In a second cluster, Y5.39(276) directly stacks
with W4.64 (d ) 6.5 Å, R ) 80°). Aromatic residues that
are not direct interaction sites for SR141716A, but are
part of this extended cluster, are illustrated in green in
Figure 3.6F3.25(190) (colored blue in Figure 3) was not
found to be part of this extended cluster. Modeling
studies also suggested that in the SR141716A/CB1 R
complex, the carboxamide oxygen of SR141716A forms
a hydrogen bond with K3.28 (N-O distance ) 2.7 Å;
O-H-N angle ) 163°). We have recently shown using
a mutant cycle that K3.28 is, in fact, a direct interaction
site for the C-3 substituent of SR141716A.6
The docking position identified for SR141716A is
consistent with SR141716A SAR results reported by
Thomas and co-workers for substitution on themonochlo-
rophenyl (MC) and dichlorophenyl (DC) rings.22Thomas
reported that substitution at the 3 or 6 position on the
DC ring with iodine resulted in a decrease in CB1
affinity. In the SR141716A docked position reported
here, substitution of an iodine at the 3 or 6 position on
the DC ring would result in steric clashes with F3.36
for the 3 position and steric clashes with V6.56 and
V6.59 for the 6 position (see Figure 4A). These substitu-
tions would thereforebeexpected toreduceCB1affinity.
Thomas also reported that substituent enlargement at
the 4 position on the MC ring of SR141716A from chloro
to bromo or iodo did not result in a significant loss of
CB1affinity. Katoch-Rouseand co-workers reported that
substitution of a methoxy group at the 4 position on the
MC ring did not significantly affect CB1 affinity as
well.23In theSR141716A docked position reported here,
substitution of bromo, iodo, or methoxy at the 4 position
on theMC ring would not producea steric clash, because
this substituent resides in an area with adjacent unoc-
cupied space (see Figure 4B).
T able 1. Ligand-Aromatic Stacking Interactions in CB1R and R*
SR141716Aain RWIN55,212-2 in R*
MCb
DCc
NAPd
INDe
df(Å)
Rg(deg)
NAh
25
50
NA
d (Å)
5.0i
10.6
4.8
7.6
R (deg)d (Å)
R (deg)d (Å)
R (deg)
F3.36
Y5.39
W5.43
W6.48
7.9
6.5
4.8
11.7
80
NA
90
NA
6.4
10.7
4.5
4.2
20
NA
90
60
6.3
9.7
6.1
8.1
50
NA
NA
NA
aSee ref 6.bMC ) monochlorophenyl ring.cDC ) dichlorophenyl ring.dNAP ) naphthyl ring.eIND ) indole ring system.fd )
distancebetween aromatic ring centroids.gR ) anglebetween normal vectors of interacting rings.hNA, intervening aminoacid(s) prevent(s)
a stacking interaction.iDistances (d) and angles (R) for aromatic systems that meet the criteria for aromatic stacking interactions18are
highlighted in bold.
F igure 3. The model of the SR141716A/CB1R complex in the
TMH3-4-5-6 region of CB1is illustrated here. SR141716A
is in a minimum-energy conformation that produces the most
negative electrostatic potential in the C3 substituent region.
In this conformation (0.92 kcal/mol abovetheglobal minimum),
thepiperidinering is in a chair conformation with thenitrogen
lone pair of electrons pointing in the same direction as the
carboxamide oxygen.6The carboxamide oxygen of SR141716A
is involved in a hydrogen-bonding interaction with K3.28
(shown in yellow). K3.28 is alsoinvolved in a salt bridge with
D6.58 (shown in green). Aromatic residues with which
SR141716A stacks directly are shown in yellow. Aromatic
residues with which SR141716A does not stack directly, but
that are part of the extended aromatic cluster formed by
SR141716A binding, are illustrated in green.
5142 J ournal of Medicinal Chemistry, 2003, Vol. 46, No. 24 McAllister et al.

Page 5

WIN55,212-2 in the R * State. Figure 5 and Table
1 illustratemodeling results for theCB agonist WIN55,-
212-2 in the CB1R* state. WIN55,212-2 was docked
in a model of the CB1active (R*) state, because agonists
are thought to have higher affinity for the R* form of a
GPCR versus the R form. In our CB1model, the R f
R* transition results in the loss of direct interactions
between W5.43(280), F3.36(201), and W6.48(357). In the
model, this triad is once again linked with each other
via WIN55,212-2, which bridges thegap between these
residues in the ligand/R*complex. WIN55,212-2 has a
direct aromatic stacking interaction with W5.43(280)
and W6.48(357) and two direct stacking interactions
with F3.36(201). The distances (d) and angles (R) for
these interactions are given in bold in Table 1, and
residues with which WIN55,212-2 interacts directly are
illustrated in yellow in Figure 5. Aromatic residues in
the TMH3-4-5-6 region form a network or cluster
with which WIN55,212-2 interacts. W5.43(280) directly
stacks with Y5.39 (d ) 6.1 Å, R ) 60°), and Y5.39(276)
stacks with W4.64 (d ) 5.3 Å, R ) 70°). In binding,
WIN55,212-2 becomes part of an aromatic cluster that
includes F3.36(201)/W6.48(357)/W5.43(280)/Y5.39(276)/
W4.64(256) in the minimized complex. No hydrogen-
bonding interactions were identified for WIN55,212-2
in CB1R*. This result is consistent with recent reports
by Huffman and co-workers that cannabimimetic in-
doles retain high CB1affinity even when all hydrogen-
bonding potential has been removed (see compound 26
in ref 46).
The binding site identified here for WIN55,212-2 is
consistent with AAI SAR data that demonstrated that
the R stereoisomer of WIN55,212 (called WIN55,212-
2), but not the S stereoisomer of WIN55,212 (called
WIN55,212-3), produced effects at the cannabinoid
receptor.24In the docking position illustrated for WIN55,-
212-2 here, WIN55,212-3 would not be able to fit
because the morpholino alkyl tail would be sterically
blocked by theTMH6 backboneand residues V6.59(368)
and M6.55(364). Early results reported by D’Ambra and
co-workers24for WIN55,212 (a mix of R + S isomers,
C-2 Me) and its C-2 H and C-2 Et congeners showed
that when the C-2 substituent was a hydrogen atom or
a methyl group, compounds with good CB1affinity and
efficacy wereproduced. However, theanaloguein which
the C-2 substituent was an ethyl group was essentially
an inactive compound. Our docking studies indicated
that if the C-2 substituent was enlarged from methyl
F igure 4. This figure illustrates the consistency of the SR141716A binding site model identified here with published SAR.22,23
(A) SR141716A would have steric overlaps with CB1if an iodo substituent was located at the 3 or 6 position on the dichloro (DC)
ring. (B) SR141716A would not have steric overlaps with CB1if a substituent as large as a methoxy were located at the 4 position
on the monochloro (MC) ring.
F igure 5. The model of the WIN55,212-2/R* complex in the
TMH3-4-5-6 region of CB1is illustrated here. WIN55,212-2
was docked in its lowest energy s-trans conformation.45
Aromatic residues with which WIN55,212-2 stacks directly
are shown in yellow. Aromatic residues with which WIN55,-
212-2 does not stack directly, but that arepart of theextended
aromatic cluster formed by WIN55,212-2 binding, are il-
lustrated in green.
Aromatic Microdomain at the Cannabinoid CB1ReceptorJ ournal of Medicinal Chemistry, 2003, Vol. 46, No. 245143

Page 6

to ethyl, the naphthyl ring would have to adjust its tilt
and make the ring more perpendicular to the plane of
the indole due to steric constraints. This adjustment,
however, results in a steric clash with W5.43(280) in
the receptor binding pocket, such that the ligand has
to back away from the aromatic cluster, losing most of
the aromatic stacking interactions illustrated in Figure
5 for WIN55,212-2.
Anandamide in R *. While anandamide is not ca-
pable of engaging in aromatic (π-π) stacking interac-
tions, it can engage in C-H‚‚‚π interactions. Figure 6
illustrates results of docking studies for the endogenous
agonist anandamide in CB1R*. Anandamide was em-
ployed in this study as a control, since a docking study
of several dimethylanandamides25suggested that the
TMH2-3-7 region, not the TMH3-4-5-6 aromatic
residuerich region of CB1, was thebinding site. Docking
studies reported here suggest that in its binding site in
the CB1TMH2-3-6-7 region, anandamide is engaged
in a hydrogen-bonding interaction with K3.28(193). In
this interaction, K3.28(193) forms a hydrogen bond with
theamideoxygen of anandamide(N toamideO distance
) 2.6 Å, N-H-O angle ) 158°). At the same time, the
headgroup hydroxyl of anandamide is engaged in an
intramolecular hydrogen bond with the amide oxygen
(O to O distance ) 2.7 Å O-H-O angle ) 130°). The
formation of such an intramolecular hydrogen bond in
anandamide helps the hydroxyl exist in a hydrophobic
region and still satisfies a hydrogen bond.
Theresidues that linetheanandamidebinding pocket
are largely hydrophobic, including F2.57(171), F3.25-
(190), L3.29(194), V3.32(197), F6.60(369), F7.35(380),
A7.36(381), Y6.57(366), S7.39(384) (hydrogen bonded
back to its own backbone carbonyl oxygen), and L7.43-
(388). The interaction with F3.25(190) is a C-H‚‚‚π
interaction with theC5-C6 doublebond in anandamide.
In the R* bundle, F2.57(171) has an interaction with
the amide oxygen. Here, an aromatic ring hydrogen
interacts with one of the lone pairs of electrons of the
amide oxygen (C-O distance ) 3.7 Å, C-H-O angle )
168°, ring center to O distance ) 5.1 Å). A similar
interaction is evident in the X-ray structure of bovine
rhodopsin26in which F5.38(203) points its positive edge
into the backbone carbonyl oxygen of P4.60 (C-O
distance ) 3.3 Å, C-H-O ) 147°, ring center to O
distance ) 4.6 Å).
We have previously shown that the arrangement of
homoallylic double bonds in the acyl chain portion of
anandamide makes anandamide a very flexible mol-
ecule.25,27While there are no crystal structures of
anandamide bound to CB1available in the literature,
there is an X-ray crystal structure of anandamide’s
parent acid, arachidonic acid (20:4, n - 6), complexed
with adipocytelipid binding protein.28In this structure,
arachidonic acid clearly adopts a curved/U-shaped
conformation that is consistent with the conformations
identified for anandamide (20:4, n - 6) in our confor-
mational memories calculations25,27and is consistent
with the conformation of anandamide in CB1illustrated
in Figure 6. The importance of K3.28(193) as a direct
interaction site for anandamide is supported by the
work of Song and Bonner,29who reported that anand-
amide was unable to compete for [3H]WIN55,212-2
binding in a human CB1K3.28(192)A mutant and that
the potency of anandamide in inhibiting cAMP ac-
cumulation was reduced >100-fold in this mutant. This
loss of affinity could occur with the loss of a strong
hydrogen-bonding interaction.
The binding-site interactions illustrated for anand-
amide in Figure 6 agree with results first reported by
Pintoand co-workers,30which showed that thehydroxyl
group in the headgroup region of anandamide could be
replaced by a methyl group without a loss in CB1
affinity. This result suggested that the hydroxyl group
is not essential for anandamide binding and also that
this hydroxyl may exist in a hydrophobic region of CB1.
This result has been echoed in later endocannabinoid
structure-activity relationship studies that showed, for
example, that a cyclopropyl headgroup results in a very
high CB1affinity ligand.31In the binding site, identified
for anandamide in our model, the headgroup hydroxyl
is located in a hydrophobic pocket and satisfies its
hydrogen-bonding potential by forming an intramolecu-
lar hydrogen bond with the amide oxygen. This result
is consistent with recent NMR solution studies of
anandamide reported by Bonechi and co-workers,32who
found that this intramolecular hydrogen bond in anand-
amide persists in solution.
We recently reported that the TMH2-3-7 region of
CB1was the binding site for a series of dimethylanand-
F igure 6. Theanandamide/R*complex in theTMH2-3-6-7
region of CB1 R* is illustrated here. K3.28(193) forms a
hydrogen bond with the amide oxygen of anandamide. At the
same time, the headgroup hydroxyl of anandamide is engaged
in an intramolecular hydrogen bond with its amide oxygen.
The anandamide binding pocket is lined with residues (shown
in gray) that are largely hydrophobic, including L3.29, V3.32,
F6.60, F7.35, A7.36, Y6.57, S7.39 (hydrogen bonded back to
its own backbone carbonyl oxygen), and L7.43. F3.25 (shown
in yellow) has a C-H‚‚‚π interaction with the C5-C6 double
bond of anandamide, while F2.57 (shown in yellow) has an
interaction with the amide oxygen of anandamide.
5144J ournal of Medicinal Chemistry, 2003, Vol. 46, No. 24 McAllister et al.

Page 7

amide analogues.25Results of docking studies here
indicatethat anandamidecan occupy a slightly different
position in CB1(the TMH2-3-6-7 region) due to the
absence of the two methyl groups at C2 and C1′ found
in the dimethylanandamide analogues previously stud-
ied.
CP55,940 in CB1 R *. CP55,940 is used here as the
radioligand for binding studies because its binding site
does not involve interaction with the residues of the
aromatic microdomain studied here. In our revised CB1
model based upon the rhodopsin crystal structure,26
CP55,940 binds higher up in theCB1R*binding pocket,
above F3.36, W5.43, and W6.48 and tothe side of F3.25
(ref 33 and manuscript in preparation). This CP55,940
binding siteis extracellular tothebinding siteidentified
for WIN55,212-2 and involves hydrogen-bonding in-
teractions of the southern aliphatic hydroxyl (SAH) of
CP55940 with K3.28, the northern aliphatic hydroxyl
of CP55940 (NAH) with K260 in the CB1EC-2 loop, and
the phenolic hydroxyl of CP55940 with D6.58.
R eceptor R ecognition of Cannabinoid L igands.
To test the binding-site hypotheses generated by the
receptor model, selectivemutations of aromatic residues
in TMHs 3, 4, 5, and 6 of the CB1 receptor were
generated and stable cell lines were established. Stable
cell lines were created in HEK-293 cells expressing
either mouse wild-type CB1 (WT CB1), F3.25(190)A,
F3.36(201)A, Y4.64(256)A, W5.43(280)A, or W6.48-
(357)A, and Scatchard analyses were performed using
[3H]CP55,940 as the radioligand (Table 2). No specific
[3H]CP55,940 binding toHEK 293 cells was found prior
to transfection (data not shown). The affinity values
obtained for all the mutant receptors were not signifi-
cantly different from WT CB1receptor (Table 2). These
results are in agreement with the receptor model that
predicted the aromatic residues chosen for mutational
analysis would not be interaction sites for the bicyclic
cannabinoid, CP55,940 (see above). Since significant
changes in the affinity of [3H]CP55,940 were not ob-
served, this tritiated compound served as an excellent
tool for ligand competition studies. The mutation of
human W4.64A has been previously reported to result
in receptor sequestration in a mammalian cell line.3In
agreement with this previous finding, we found that
mouse W4.64(256)A was not properly expressed on the
cell surface in HEK cells as assessed by immunofluo-
rescence (Figure 7). Compared to WT CB1, few cells
showed surface expression of the mutant protein. We
also tested this mutation in an amphibian cell (oocyte)
background versus a mammalian to determine if Xeno-
pus oocytes might process the mutant protein correctly,
but functional expression was not observed (data not
shown). Further characterization of W4.64(256)A was
not carried out because of these findings. These results
for the W4.64A mutant are similar to those reported
by Rhee and co-workers for the W4.64A mutation in
CB2, which exhibited nobinding of any ligand, suggest-
ing improper folding or lack of cell surface expression.34
L igand Displacement Studies. F 3.25(190)A R e-
ceptor Affinity. The F3.25(190)A mutation had no
statistically significant effect on WIN55,212-2 or
SR141716A binding, but it resulted in a 6-fold loss in
affinity for anandamide (Table 3, Figure 8). This result
is consistent with modeling studies (see Figures 3, 5,
and 6) that show that F3.25(190) is not part of the
binding site of WIN55,212-2 (in R*) or SR141716A (in
R) but is a part of the anandamide binding pocket (in
R*).
F 3.36(201)A R eceptor Affinity. The affinity of
anandamide was unaffected by the F3.36(201)A muta-
tion, whereas the affinities of both WIN55,212-2 (9-
fold loss) and SR141716A (3-fold loss) were affected by
this mutation (Table 3, Figure 8). The binding results
for anandamide are consistent with the modeling stud-
T able 2. Cannabinoid Receptor Radioligand Binding
Determinations in Membranes Prepared from Wild Type and
Mutant Cell Linesa
cell lineKd(nM)
1.5 (0.53-2.4)
5.5 (0.3-10)
2.3 (0.6-4.1)
ND
5.4 (1.2-9.5)
3.2 (2.0-4.4)
Bmax(pmol/mg)
4.4 (3.5-5.3)
3.2 (1.9-4.5)
5.2 (3.6-6.9)
ND
1.9*(1.1-2.6)
4.4 (3.8-5.2)
WT CB1
F3.25(190)A
F3.36(201)A
W4.64(256)A
W5.43(280)A
W6.48(357)A
aExperiments using [3H]CP55,940 were performed on stably
transfected HEK 293 cells to evaluate binding affinities and
relative levels of receptor expression in the wild-type and mutant
receptors. Nonspecific binding was determined in the presence of
excess CP55,940 (see Experimental section). Data are the means
and corresponding 95% confidence limits of three independent
experiments each performed in triplicate. Theasterisk (*) indicates
statistically significant differences from wild type (p < 0.05). ND,
not determined.
F igure 7. Immunolabeling of HEK 293 cells stably expressing
wild type (WT) CB1 receptor (A) or transiently expressing
W4.64A mutant CB1receptor (B). Bright cell surface labeling
was observed around cells expressing the wild-type CB1
receptor. The majority of cells that were transfected with the
W4.64A mutant CB1receptor did not express the protein on
their cell surface.
T able 3. Ability of Three Cannabinoid Receptor Ligands to
Displace [3H]CP55,940 from Membranes Prepared from Wild
Type and Mutant Cell Linesa
Ki
cell lineWIN55,212-2banandamidec
12 (7.0-20)
15 (3.7-49)
107*(44-261)
W4.64(256)A ND
W5.43(280)A 199* (43-914) 0.3 (0.2-0.7)
SR141716Ab
4.8 (2.2-10)WT CB1
F3.25(190)A
F3.36(201)A
0.3 (0.1-0.6)
1.8* (0.6-5.6) 9.6 (4.6-20)
0.3 (0.1-1.1)
ND
14* (7.3-29)
ND
46%* displacement
at 5 µM
33* (23-45)W6.48(357)A 45* (22-92) 0.3 (0.1-0.5)
aInhibition constants were obtained from competition experi-
ments (see Experimental section). Data are the means and
corresponding 95% confidence limits of three independent experi-
ments each performed in triplicate. The asterisk (*) indicates
statistically significant differences from wild type (p < 0.05). ND,
not determined.bIn nM.cIn µM.
Aromatic Microdomain at the Cannabinoid CB1ReceptorJ ournal of Medicinal Chemistry, 2003, Vol. 46, No. 245145

Page 8

ies that did not identify F3.36 as a direct interaction
site for anandamide. Results for WIN55,212-2 and
SR141716A areconsistent with receptor docking studies
reported above, which show that F3.36(201) is a direct
interaction site for both WIN55,212-2 (in R*) and
SR141716A (in R). As illustrated in Figure 5, F3.36-
(201) forms twostacking interactions with WIN55,212-
2, as this residue can interact with both the naphthyl
(NAP) and the indole (IND) rings of WIN55,212-2.
Figure 3 illustrates that F3.36(201) has a single aro-
matic stacking interaction with thedichlorophenyl (DC)
ring of SR141716A in the R state of WT CB1. It is likely
that SR141716A can partially compensate for the loss
of aromaticity at 3.36(201) in the F3.36(201)A mutant
through a strengthened aromatic stacking interaction
with W6.48(357), which is located just beneath F3.36-
(201) in the WT CB1R state (Figure 3).
W5.43(280)A R eceptor Affinity. The W5.43(280)A
mutation produced an 8-fold loss in affinity for WIN55,-
212-2, but it did not affect anandamide binding (Table
3, Figure 8). This result is consistent with the modeling
studies that suggest that W5.43(280) is not part of the
anandamide binding pocket but suggest that WIN55,-
212-2 has a direct aromatic stacking interaction be-
tween W5.43(280) and its NAP ring. This aromatic
stacking interaction can be expected to be particularly
strong as the centroid to centroid distances between
interacting rings is at close distance (4.5 Å) and the
angle of the interacting ring planes is 90°, forming a
perfect tilted-T arrangement (see Table 1).18
The W5.43(280)A mutation had the most profound
effect on SR141716A binding compared to any of the
mutations reported here. Even at 5 µM, SR141716A
could only displace 46% of tritiated CP55,940. The
nonspecific effects of cannabinoids produced beyond 5
µM precluded us from testing higher concentrations.
This result is consistent with the modeling studies that
suggested that W5.43(280) is central in the aromatic
cluster interactions with SR141716A. As detailed in
Table 1 and illustrated in Figure 3, the model suggests
that W5.43 has direct stacking interactions with both
the MC and DC rings of SR141716A. It is likely that
W5.43 helps orient SR141716A in the binding pocket,
so loss of aromaticity at W5.43(280) would be expected
tohave a particularly deleterious effect upon SR141716A
binding.
W6.48(357)A R eceptor Affinity. The binding of
anandamide was unaffected by the W6.48(357)A muta-
tion, whereas the binding of both WIN55,212-2 (4-fold
loss) and SR141716A (7-fold loss) was affected by this
mutation (Table 3, Figure 8). These results are also
consistent with modeling results for anandamide, which
suggest that W6.48(357) is not part of the anandamide
binding site, and with results for WIN55,212-2, which
suggest a direct stacking interaction between W6.48-
(357) and the NAP ring of WIN55,212-2. The magni-
tude of the effect of the W6.48(357)A mutation on
SR141716A binding (7-fold loss) was, however, more
than would be expected from a cursory inspection of the
modeling study illustrated in Figure 3, as W6.48(357)
does not stack directly with SR141716A but stacks with
F3.36(201), which, in turn, stacks with SR141716A. It
is possible that the decrease in affinity upon mutation
of W6.48 to alanine is due to the loss of the extended
aromatic stacking interaction between SR141716A,
F3.36, and W6.48, as only one stacking interaction
would remain (i.e. between SR141716A and F3.36).
Discussion
In the absence of a crystal structure for CB1, the
combination of site-directed mutagenesis and molecular
modeling is a powerful tool that allows us to study
specific ligand-receptor interactions. In this study, we
wanted todetermine if aromatic residues in the binding
pocket of CB1 form specific interaction sites for ami-
noalkylindole agonist (WIN55,212-2) and diaryl pyra-
zole inverse agonist/antagonist (SR141716A) ligands,
but not for endogenous (anandamide) and nonclassical
cannabinoid (CP55,940) agonists.
Importance of CB1 TMH3-4-5-6 R egion. Muta-
tion results reported here in Table 3 suggest that
TMH3-4-5-6 aromatic microdomain residues F3.36,
W5.43, and W6.48 arepart of thebinding pocket for both
SR141716A and WIN55,212-2. One important proviso
associated with any mutation study is the possibility
that a gross structural changehas occurred in theligand
binding pocket as the result of a mutation. In this case,
it is possible that a change in a residue distant from
the ligand binding pocket could have an effect on ligand
F igure 8. Displacement of [3H]CP55,940 by SR141716A (A),
WIN55,212-2 (B), or anandamide (C) in membranes prepared
from HEK 293 cells transfected with wild-type or mutant CB1
receptors. Displacement curves were obtained in stable cell
lines expressing [0] WT CB1, [4] F3.25(190)A, [b] F3.36(201)A,
[*] W5.43(280)A, and [[] W6.48(357)A. Each data point shown
is the mean ( SE of at least three independent experiments
performed in triplicate.
5146 J ournal of Medicinal Chemistry, 2003, Vol. 46, No. 24 McAllister et al.

Page 9

binding. Such a structural change is unlikely here for
three reasons: (1) the aminoacid residue chosen as the
replacement is smaller than the WT residue, as recom-
mended by Ward et al.,35(2) the replacement residue is
not a helix breaker, such as Pro or Gly,36and is not a
residuethat has been documented toinducehelix bends,
such as Ser or Thr,37and (3) both CP55,940 and
anandamideretain WT affinity for theF3.36A, W5.43A,
and W6.48A mutants, providing strong evidence that a
structural rearrangement in the binding pocket has not
occurred here.35
The identification of the TMH3-4-5-6 region as the
binding region of SR141716A and WIN55,212-2 is
supported by previous mutation/chimera studies. A
recent CB1mutant cycle study indicated that K3.28 is
a direct interaction site for the C-3 substituent of
SR141716A.6Shire and co-workers have shown in CB1/
CB2chimera studies that the TMH4-E2-TMH5 region
of CB1 contains residues critical for the binding of
SR141716A and WIN55,212-2.3Mutation studies per-
formed by Song and co-workers have supported the
importance of the TMH3-4-5 region of CB1and CB2
for the binding of WIN55,212-2.5These studies showed
that the 15-20-fold higher affinity of WIN 55,212-2 for
theCB2receptor may beduein part toa direct aromatic
stacking interaction with F5.46, a residue that is
aromatic only in CB2.
Aromatic (π-π) Stacking Interactions in CB1
L igand R ecognition. Another hypothesis tested here
is that aromatic stacking interactions are important for
the binding of aminoalkylindole agonists (such as
WIN55,212-2) and diaryl pyrazole inverse agonist/
antagonists (such as SR141716A) at CB1. Attractive
π-π interactions areoneof themajor noncovalent forces
governing molecular recognition. Burley and Petsko18
have reported that aromatic-aromatic stacking interac-
tions in proteins operate at distances (d) of 4.5-7.0 Å
between ring centroids. The angle (R) between normal
vectors of interacting aromatic rings typically is between
30° and 90°, producing a “tilted-T” or “edge-to-face”
arrangement of interacting rings. Hunter and co-work-
ers38have reported that π-π parallel stacking interac-
tions (R < 30°) between phenylalanine residues in
proteins are also favorable if the rings are offset from
each other. Recent ab initio calculations on benzene
dimers have estimated that tilted-T and off-set parallel
π-π (aromatic) stacking interactions have stabilization
energies of 2.7 and 2.8 kcal/mol respectively, suggesting
that these may be very important noncovalent interac-
tions.39
In addition to W4.64 and Y5.39, docking studies
reported here identified F3.36, W5.43, and W6.48 as
part of the binding pockets of SR141716A and WIN55,-
212-2 (Table 1, Figures 3 and 5). These docking results
suggest that aromatic stacking interactions may be
important for AAI and diaryl pryrazole ligand binding.
Consistent with the docking results, we have reported
here that mutation of F3.36, W5.43, or W6.48 to a
nonaromatic residue(alanine) results in reduced affinity
for SR141716A and WIN55,212-2. If such an affinity
decrease is the direct result of loss of aromatic stacking
interactions via substitution of a nonaromatic for an
aromatic residue within the ligand binding pocket, then
a comparable change in the ligand should also lead to
a decrease in affinity, i.e., in each ligand structural
class, substitution of nonaromatic for aromatic moieties
of equivalent or smaller size (with no change in the
receptor binding site) should also result in reduced
receptor affinity. Consistent with this premise, SAR
studies of the C-5 substituent of SR141716A (monochlo-
rophenyl (MC) ring) have indicated that loss of aroma-
ticity in the C5 substituent results in ligands with
dramatically reduced CB1affinities.40,41Loss of aroma-
ticity in the N1 substituent of SR141716A (the dichlo-
rophenyl (DC) ring) by a change to a cyclohexyl N1
substituent has been reported to result in a significant
loss in CB1affinity as well.42
The importance of aromaticity for proper AAI interac-
tion with CB1has been illustrated by Huffman and co-
workers.43These investigators reported that replace-
ment of the naphthyl (NAP) ring of WIN55,212-2 with
an alkyl (CH3) or alkenyl [(CH3)2CdCH] group resulted
in complete loss of CB1affinity (Ki> 10 000 nM) in both
cases. Eissenstat and co-workers44reported that re-
placement of the naphthyl ring of WIN55,212-2 with
a hydrogen resulted in a compound with a very high
IC50at CB1. These findings are significant because they
underscore that an aromatic system as part of the C-3
substituent of WIN55,212-2 may be important and
becauseit suggests that therequirement at this position
is not simply a requirement for any hydrophobic moiety.
While recent CB1 K3.28A mutation studies6have
indicated that SR141716A binding to CB1 involves
a hydrogen bond between the C-3 substituent of
SR141716A and K3.28 (seeFigure3), hydrogen bonding
does not appear to be important for WIN55,212-2
interaction with CB1. Song and Bonner reported that a
CB1K3.28A mutation had no effect on the affinity of
WIN55,212-2 for CB1.29The carbonyl oxygen of WIN55,-
212-2 would appear to be a likely hydrogen-bond
acceptor; however, rigid AAI analogues in which the
carbonyl group was replaced with a methylene group,
theE naphthylideneindenes, havebeen shown toretain
high affinity at both the CB1 and CB2 receptors.45
Huffman and co-workers46recently synthesized an
indole analogue of WIN55,212-2 for which all hydrogen-
bonding potential was removed. This compound was
found toretain high CB1affinity. These results support
themodeling reported herethat indicates that hydrogen
bonding is not important for interaction of the AAI
WIN55,212-2 with CB1.
C-H‚‚‚π Interactions in CB1/E ndocannabinoid
R ecognition. C-H‚‚‚π interactions are moderate, but
nevertheless important, interactions that contribute
to protein stability.47Mutation studies reported here
suggest that F3.25 is part of the binding site of anan-
damide, while modeling studies suggest that F3.25 has
a C-H‚‚‚π interaction with the C5-C6 bond of anan-
damide. This interaction is consistent with the recent
crystal structure of fatty acid amide hydrolase (FAAH)
in which the arachidonyl inhibitor methoxy arachidonyl
phosphonate (MAP) is bound.48Here several aromatic
aminoacids (F194, F244, Y335, F381, F432, and W531)
line the substrate binding pocket surrounding the
arachidonyl chain and F194, F381, and F432 are
engaged in C-H‚‚‚π interactions with the first through
third doublebonds, respectively, of theMAP arachidonyl
acyl chain. Binding-site interactions identified here for
Aromatic Microdomain at the Cannabinoid CB1Receptor J ournal of Medicinal Chemistry, 2003, Vol. 46, No. 245147

Page 10

anandamidealsoagreewith therecent crystal structure
of arachidonic acid bound to prostaglandin synthase.49
This crystal structure shows C-H‚‚‚π interactions
between arachidonic acid acyl chain double bonds and
aromatic residues.
Conclusions
Modeling and mutation studies reported here suggest
that the aromatic microdomain in the TMH3-4-5-6
region comprised of F3.36, W4.64, Y5.39, W5.43, and
W6.48 is the binding-site region for both SR141716A
and WIN55,212-2, but not for anandamide. Only F3.25
was found to be important for the binding of anand-
amide at CB1. These results indicate that the endocan-
nabinoid binding pocket at CB1 may only partially
overlap the binding pockets of SR141716A and WIN55,-
212-2.
Results reported here alsosuggest that the set of key
noncovalent forces governing molecular recognition of
theaminoalkylindoles (WIN55,212-2), diaryl pryrazoles
(SR141716A), and the endocannabinoids (anandamide)
at CB1 may be different. For the antagonist/inverse
agonist, SR141716A, a combination of aromatic stacking
interactions and hydrogen bonding6are key, while
aromatic stacking interactions alone appear to be the
principal intermolecular forces governing the binding
of the aminoalkylindole WIN55,212-2 at CB1. For the
endocannabinoid anandamide, C-H‚‚‚π and hydrogen-
bonding interactions appear to be key intermolecular
forces for CB1recognition.
E xperimental Section
R eceptor Model Construction. Amino Acid Number-
ing System. In thediscussion of receptor residues that follows,
theaminoacid numbering schemeproposed by Ballesteros and
Weinstein50is used. In this numbering system, themost highly
conserved residue in each transmembrane helix (TMH) is
assigned a locant of 0.50. This number is preceded by theTMH
number and followed in parentheses by the sequence number.
All other residues in a TMH are numbered relative to this
residue. In this numbering system, for example, the most
highly conserved residue in TMH2 of the mouse CB1receptor
is D2.50(164). The residue that immediately precedes it is
A2.49(163). Figure 1 serves as a reference for this numbering
system in CB1.
Definition of the R otameric State of ?1. Different
nomenclatures have been used to define the rotameric state
of side chain torsion angles. The nomenclature employed here
for the ?1 (chi 1) torsion angle is that described by Shi and
co-workers.8When the heavy atom at the γ position is at a
position opposite to the backbone nitrogen when viewed from
the ?-carbon to the R-carbon, the ?1 is defined to be trans.
When theheavy atom at theγ position is at a position opposite
tothe backbone carbon when viewed from the ?-carbon tothe
R-carbon, the?1 is defined tobegauche+ (g+). When theheavy
atom at the γ position is at a position opposite to the
R-hydrogen when viewed from the ?-carbon to the R-carbon,
the ?1 is defined tobe gauche- (g-). Using this nomenclature
system, the side chain conformations were categorized intog-
(0° <?1 < 120 °), trans (120° < ?1 < 240°), or g+ (240° < ?1 <
360°).
R to R * T ransition in GPCR s. Because agonists are
thought to have higher affinity for the activated form of
GPCRs,16agonist ligands in the work reported here were
docked in a model of theactivated state(R*) of CB1(seebelow).
This R* CB1 model was created by modification of our
rhodopsin (Rho)-based model of the inactive (R) form of CB1
(see below) and guided by the biophysical literature on the R
to R* transition. It has now been well-established in the
Biophysical literature that the R toR*transition in GPCRs is
accompanied by significant changes in the transmembrane
helix bundle (see refs 15 and 36 for reviews). These studies
haveindicated that activation of Rhois accompanied by a rigid
domain motion of TMH6 relative to TMH3.10J avitch and co-
workers14documented that Cys 6.47 becomes available tothe
binding pocket only in a constitutively active ?-2 mutant. The
acquired accessibility of Cys 6.47 in the mutant was hypoth-
esized to result from a counterclockwise rotation (from an
extracellular view) and/or tilting of the sixth membrane
spanning segment associated with activation of the receptor.
Lin and Sakmar51reported that perturbations in the environ-
ment of W3.41 (along with W6.48) of Rho occur during the
conformational change concomitant with receptor activation.
W3.41(126) faces lipid in the inactive state of Rho.26Lin and
Sakmar51attributed the observed differential shift of the Lb
absorption of the indole side chain of Trp126(W3.41) in Rho
during activation to the decreased hydrophobicity of its
environment. This has been interpreted as originating from a
counterclockwise rotation of TMH3 (from an extracellular
view),15which would move this residue into the more polar
environment of the TMH3-TMH4 interface. A ∼20° rotation
of TMH3 would be required to rotate this residue in the Rho
crystal structure out of lipid and into the TMH3-TMH4
interface.
J ensen and co-workers13recently demonstrated through
fluorescence studies of the ?-2-adrenergic receptor that P6.50
in the highly conserved CWXP motif of TMH6 can act as a
flexible hinge that mediates the transition from R to R*. In
the R state, these investigators propose that TMH6 is kinked
at P6.50 such that its cytoplasmic end is nearly perpendicular
to the membrane and close to the cytoplasmic end of TMH3.
The transition tothe R* state is accomplished by the straight-
ening of TMH6 such that the cytoplasmic part of TMH6 moves
away from the receptor core and upward toward the lipid
bilayer.13Ballesteros and co-workers52recently proposed that
a salt bridge between R3.50 and E6.30 at the intracellular end
of the ?-2-adrenergic receptor stabilizes this receptor in its
inactive state.
In the present study, the literature on GPCR activation
discussed above was used togenerate an R* CB1TMH bundle
from a model of the inactive (R) CB1receptor based on the 2.8
Å crystal structure of rhodopsin.26The creation of these two
forms of CB1is described below.
Model of Inactive State (R ) F orm of CB1. A model of
theR form of CB1was created using the2.8 Å crystal structure
of bovineRho.26First, thesequenceof themouseCB1receptor53
(see Figure 1) was aligned with the sequence of bovine Rho
using the same highly conserved residues as alignment guides
that were used initially to generate our first model of CB1.54
TMH5 in CB1lacks the highly conserved proline in TMH5 of
Rho. Therefore, the sequence of CB1in the TMH5 region was
aligned with that of Rho as described previously using its
hydrophobicity profile.54The mouse CB1sequence53is 97.7%
identical to the human CB1 sequence55overall and 100%
identical within the transmembrane regions. The mouse
sequence is one residue longer (473 residues) than the human
sequence (472 residues) due toan additional residue in the N
terminus.
Initial helix ends for mouseCB1werechosen in analogy with
those of Rho:26TMH1, N1.28(113)-R1.61(146); TMH2, R2.37-
(151)-H2.68(182); TMH3, S3.21(186)-R3.56(221); TMH4,
T4.38(230)-C4.66(258); TMH5, H5.34(271)-K5.64(301); TMH6,
R6.28(337)-K6.62(371); TMH7, K7.32(377)-S7.57(402); in-
tracellular extension of TMH7, D7.59(404)-C7.71(416). With
the exception of TMH1, these helix ends were found to be
within one turn of the helix ends originally calculated by us
and reported in 1995.54Two changes dictated by the CB1
sequence were made in the helix ends. The shortness of the
E1 loop region in CB1 necessitated starting TMH3 at 3.23
[N3.23(188)-R3.56(221)]. The break in helicity caused by the
GWNC sequence motif on the extracellular end of TMH4
necessitated that TMH4 end at 4.62 instead of 4.66 (as is found
in Rho). Changes to the general Rho structure that were
5148J ournal of Medicinal Chemistry, 2003, Vol. 46, No. 24 McAllister et al.

Page 11

necessitated by sequence divergences included the absence of
helix-kinking proline residues in TMH1 and TMH5, the lack
of a GG motif in TMH2, as well as, the presence of extra
flexibility in TMH6.9
Because TMH6 figures prominently in the R to R* transi-
tion,13we have studied the conformations accessible toTMH6
in CB1and CB2using conformational memories (CM).9These
studies revealed that TMH6 in CB1 (but not CB2) has high
flexibility due to the small size of residue 6.49 (a glycine)
immediately preceding Pro 6.50. Two families of conformers
were identified by CM for TMH6 in CB1. Cluster 1 showed a
pronounced prolinekink (40 members out of 100, 71.2°average
kink angle). Cluster 2 contained helices with less pronounced
kinks (51 members out of 100, 30.1° average kink angle). A
conformer from the more kinked CM family of CB1 TMH6s
(cluster 1) was used in our model of the R state of CB1. This
conformer was selected (Prokink angle) 53.1°) sothat R3.50-
(215) and D6.30(339) could form a salt bridge at the intracel-
lular ends of TMHs 3 and 6 in the CB1 TMH bundle. An
analogous salt bridge has been shown to be an important
stabilizer of the inactive state of the ?2adrenergic receptor52
and to be present in Rho.26Because of the extreme flexibility
of TMH6 in CB1, we have proposed that an additional
TMH3-6 salt bridge, K3.28(193)-D6.58(367), stabilizes the
inactive state on the extracellular side of the TMH bundle.6
Model of Active (R *) F orm of CB1. On the basis of
experimental results for rhodopsin and the ?-2-adrenergic
receptor,10,13,14,51,52theR*(active) CB1bundlewas created from
theinactive(R) model of CB1by rotating TMH3 sothat residue
3.41 moves into the less hydrophobic environment of the
TMH3-4 interface.51This was accomplished by a 20° coun-
terclockwise (extracellular view) rotation of TMH3 from its
orientation in the inactive (R) bundle. In the R* bundle, a
TMH6 conformer from thesecond major conformational family
(less kinked: 21.8° kink angle) identified by CM9was substi-
tuted for the TMH6 conformer used in the inactive model of
CB1. This conformer was chosen sothat the salt bridge in the
inactive state between R3.50(214) and D6.30(338) would be
broken due tothe movement of the intracellular end of TMH6
away from that of TMH3 and out into lipid.52TMH6 was also
rotated (counterclockwise from extracellular view) sothat Cys
6.47 became accessible from inside the binding site crevice.14
Preparation of Helices. Each helix of the model was
capped as the acetamide at its N-terminus and as the N-
methyl amide at its C-terminus. Ionizable residues in the first
turn of either end of the helix were neutralized, as were any
lipid facing charged residues. Ionizable residues were consid-
ered charged if they appeared anywhere else in the helix.
L igand Conformations and Docking Positions. The
binding-site conformations and anchoring interactions inside
the receptor used for each ligand were based on our published
computational and experimental work and that of others as
detailed below.
SR 141716A. We have recently shown in a mutant cycle
study that the C3 substituent of SR141716A has a direct
interaction with K3.28,6soK3.28 was used as theanchor point
for docking. Since chimera studies have identified that the
TMH4-TMH5 region of CB1is involved in SR141716A binding,3
the ligand was docked in the TMH3-4-5 region of CB1. Our
SR141716A conformational analysis studies identified two
principal conformers (of theC3 substituent),6oneof which was
identified as thebioactiveconformer based on our mutant cycle
experimental results.6This conformer produces the highest
negative electrostatic potential in the C-3 substituent region
(0.92 kcal/mol above global minimum as indicated by ab initio
Hartree-Fock 6-31G* calculations). In this conformation, the
piperidine ring is in a chair conformation with the nitrogen
lone pair of electrons pointing in the same direction as the
carboxamide oxygen. The carboxamide of the C3 substituent
of SR141716A is in a trans geometry. A recent crystal structure
of SR141716A confirms this carboxamide trans geometry (C.
George, Laboratory for the Structure of Matter, Naval Re-
search Laboratory, personal communication). This bioactive
conformer of SR141716A was docked such that its carboxamide
oxygen formed a hydrogen bond with K3.28 and the ligand
occupied the TMH3-4-5 region of CB1without van der Waals
overlaps. The resultant ligand/CB1complex was energy mini-
mized as described below.
WIN55,212-2. CB1/CB2 chimera studies have suggested
that WIN55,212-2 binds in the TMH4-TMH5 region of CB1.3
In agreement with these results, we have shown that the CB1
affinity of WIN55,212-2 increases in a V5.46F mutant,
suggesting that TMH5 is part of the WIN55,212-2 binding
pocket.5WIN55212-2 was therefore docked in the TMH4-5
region of CB1. Our synthesis of rigid indene analogues that
mimic the s-cis and s-trans conformers of WIN55,212-2
suggested that the s-trans conformation of WIN55,212-2 is
its bioactive conformation.45WIN55,212-2, in its s-trans
conformation, was docked in the TM4-5 region of CB1using
interactive computer graphics with the naphthoyl group
oriented either extracellularly or intracellularly and each
ligand/CB1complex was energy minimized. These calculations
suggested that a naphthoyl intracellular orientation resulted
in the lower energy complex.
Anandamide. CB1human K3.28(192)A mutation binding
results for theendogenous agonist anandamidehavesuggested
that K3.28 is a primary interaction site.29SAR studies reported
by Pintoand co-workers have suggested that the carboxamide
group, rather than the hydroxyl group of anandamide, is
important for high CB1 affinity.30We therefore used an
interaction between the caboxamide oxygen and K3.28 as an
anchor point for docking anandamide at CB1. We have used
the CM method, a Monte Carlo/simulated annealing method,
to identify low free energy conformations of anandamide.25,27
In contrast to SR141716A and WIN55,212-2, which have
relatively few low-energy conformers, anandamide was found
to have many low-energy conformations. The CM method
identified twomajor conformational families for anandamide,
a U-shaped cluster (49/100 structures) and an extended-shape
cluster (29/100 structures).25These results combined with CM
results for several acyl chain congeners of anandamide sug-
gested that anandamideanalogues (from theU-shaped cluster)
that can form tightly curved structures should have the
highest affinity for CB1.25These results led us to select
structures from the anandamide U-shaped cluster25for dock-
ing. A representative structure from the U-shaped cluster was
selected such that when docked in our CB1 R* model, the
carboxamide oxygen of this conformer could form a hydrogen
bond with K3.28 but create no van der Waals overlaps with
other residues in the binding site crevice. The ligand/CB1
complex was then energy minimized as described below.
E nergy Minimization: L igand-R eceptor Complexes
and Unoccupied R eceptor States. Each ligand was docked
using interactive computer graphics. SR141716A was docked
in theinactivestate(R) model of CB1, becauseinverseagonists
have higher affinity for this state,6while WIN55,212-2 and
anandamide were docked in the model of the CB1 active R*
state, because agonists have higher affinity for this state. The
energy of the ligand/CB1 R or ligand/CB1 R* TMH bundle
complex was minimized using the AMBER*united atom force
field in Macromodel 6.5 (Schro ¨dinger Inc., Portland, OR). A
distancedependent dielectric, 8.0 Å extended nonbonded cutoff
(updated every 10 steps), 20.0 Å electrostatic cutoff, 4.0 Å
hydrogen bond cutoff, and explicit hydrogens on sp2carbons
were used. The first stage of the calculation consisted of 2000
steps of Polak-Ribier conjugate gradient (CG) minimization
in which a force constant of 225 kJ /mol was used on the helix
backbone atoms in order to hold the TMH backbones fixed,
whilethesidechains werepermitted torelax. Thesecond stage
of the calculation consisted of 100 steps of CG in which the
force constant on the helix backbone atoms was reduced to50
kJ /mol in order toallow the helix backbones toadjust. Stages
one and two were repeated with the number of CG steps in
stage two incremented from 100 to 500 steps until a gradient
of 0.001 kJ /(mol‚Å
protocol was followed for the unoccupied receptor R and R*
models.
2) was reached. This same minimization
Aromatic Microdomain at the Cannabinoid CB1ReceptorJ ournal of Medicinal Chemistry, 2003, Vol. 46, No. 245149

Page 12

Assessment of Aromatic Stacking Interactions. Resi-
dues and/or ligand regions were designated here as participat-
ing in an aromatic stacking interaction if subject rings had
centroid tocentroid distances (d) between 4.5 and 7.0 Å. These
interactions were further classified as “tilted-T” arrangements
if 30° e R e 90° and as parallel arrangements for R < 30°
(where R is the angle between normal vectors of interacting
rings). Parallel arrangements were considered favorable only
if the interacting rings were offset from each other.38All
measurements were made using Macromodel 6.5 (Schro ¨dinger
Inc.).
Mutation Studies. Materials. [3H]CP55,940, SR141716A,
and CP55,940 were obtained from the National Institutes on
Drug Abuse. WIN55,212-2 was purchased from RBI (Natick,
MA) and anandamide was purchased from Tocris. The mouse
CB1cDNA was cloned in our lab.53
Mutagenesis. Mutations were introduced with the Quik-
Change site-directed mutagenesis kit (Stratagene) as previ-
ously described.56This method allows mutagenesis to be
performed in any vector, hence we used mouse CB1subcloned
into pcDNA3 (Invitrogen). The DNAs were sequenced to
confirm mutation in the desired regions only. The mutations
were made with the following primer sets: (forward) F3.25A-
(190) CAA AGA TAG TCC CAA TGT GGC TCT GTT CAA ACT
GGG TGG, F3.36A(201) GTG GGG TTA CCG CCT CCG CCA
CAG CAT CTG TG, W4.64A(256) CTC TCC TGG GCG CGA
ACT GCA AGA AGC TGC, W5.43A(280) GAA ACC TAC CTG
ATG TTC GCG ATC GGA GTC ACC AGT G, W6.48A(357)
TGT TGA TCA TCT GCG CGG GCC CTC TGC TTG CGA TC.
The reverse primers were the reverse complement of the
forward primer sequences.
Cell Culture and T ransfection. Cell lines were created
by transfection of wild type or mutant CB1pcDNA3 into HEK
293 cells by the Lipofectamine reagent (Invitrogen) and
cultured as previously described.56Cell lines containing mod-
erate to high levels of receptor mRNA, assessed by Northern
analysis, were tested for receptor binding and signal trans-
duction properties. Cell lines with the most similar receptor
expression profile, as ascertained by Bmaxvalues, were chosen
for further analysis (Table 2).
Cannabinoid R eceptor R adioligand Binding Deter-
minations. The current assay has been previously de-
scribed.56,57Briefly, cells wereharvested in phosphate-buffered
saline containing 1 mM EDTA and centrifuged at 500g for 5
min. The cell pellet was homogenized and centrifuged three
times at 1600g (10 min). The combined supernatants were
centrifuged at 100 000g (60 min). The pellet (P2 membrane)
was resuspended in 3 mL of buffer B (50 mM Tris-HCl, 1 mM
EDTA, 3 mM MgCl2, pH 7.4) to yield a protein concentration
of approximately 1 mg/mL. Binding was initiated by the
addition of 25-75 µg of P2 membrane protein to silanized
tubes containing [3H]CP55,940 (102.9 Ci/mmol) and a suf-
ficient volume of buffer C [50 mM Tris-HCl, 1 mM EDTA, 3
mM MgCl2, and 5 mg/mL fatty acid freebovineserum albumin
(BSA), pH 7.4] to bring the total volume to 0.5 mL. The
addition of 1 µM of unlabeled CP55,940 was used to assess
nonspecific binding. Specific binding averaged >50% of total
binding in all cell lines used in the analysis. Following
incubation (30 °C for 1 h), binding was terminated by the
addition of 2 mL of ice cold buffer D (50 mM Tris-HCl, pH 7.4,
plus 1 mg/mL BSA) and rapid vacuum filtration through
Whatman GF/C filters (pretreated with 0.1% polyethylene-
iminefor at least 2 h). CP55,940 and all cannabinoid analogues
were prepared by suspension in assay buffer from a 1 mg/mL
ethanolic stock without evaporation of the ethanol (final
concentration of no more than 0.4%). When anandamide was
used as a displacing ligand, experiments were performed in
the presence of phenylmethylsulfonyl fluoride to inhibit the
break down of the ligand by endogenous esterases (50 µM).
Saturation experiments were conducted with six concentra-
tions of [3H]CP55,940 ranging from 250 pM to 10 nM.
Competition assays were conducted with 1 nM [3H]CP55,940
and six concentrations (0.001 nM to10 µM) displacing ligands.
Bmax and Kd values were calculated by unweighted least-
squares nonlinear regression of log concentration values versus
binding of pmol/mg protein. These data was fit to a one-site
binding model using GraphPad Prism (GraphPad). Displace-
ment IC50 values were determined by unweighted least-
squares nonlinear regression of log concentration-percent
displacement data and then converted to Kivalues using the
method of Cheng and Prusoff58and analyzed with GraphPad
Prism.
Immunocytochemistry. Cells wereplated ontocover slips
that had been pretreated for 1 h with poly D-lysine (0.02 mg/
mL; Sigma) and maintained in a humidified atmosphere of
5% CO2 in air at 37 °C. Cells were washed with 10 mM
HEPES-buffered saline (comprising in mM: NaCl 130, D-
glucose 25, HEPES 10, KCl 5.4, CaCl21.8, MgCl21) at room
temperature and incubated for 1 h with a polyclonal rabbit
anti-CB1 receptor antibody (Cayman Chemical) at a final
concentration of 8 µg/mL. Cells were washed three times with
HEPES-buffered saline and fixed with 4% paraformaldehyde
for 10 min at room temperature. CB1 receptor antibody was
labeled for 40 min at room temperature with Alexa Fluor 488
goat anti-rabbit secondary antibody (1:500; Molecular Probes)
followed by three washes in buffer. Cover slips were mounted
on slides in Vectashield (Vector Laboratories) and cell surface
labeling was visualized with a fluorescencemicroscope(Nikon).
Wild-type CB1 expressing cells or transient cells that were
transfected with W4.64A mutant CB1receptors were labeled.
Time after transfection did not alter the labeling as a similar
pattern of labeling was observed when stable or transient cells
transfected with wild-type mouse CB1receptor were labeled.
In control experiments, no labeling was observed with the
secondary antibody alone or when the primary antibody was
incubated with CB1 receptor blocking peptide at 100 µg/mL
(Cayman Chemical).
Statistical Analyses. The Kiand Kdvalues in the mutant
versus wild-typecell lines werecompared by analysis of logged
data (GraphPad Prism) using analysis of variance (ANOVA)
or the unpaired Student’s t-test, where suitable. Bonferroni-
Dunn post-hoc analyses were conducted when appropriate. P
values <0.05 defined statistical significance.
Acknowledgment.
Beverly Brookshire for her technical assistance in the
preparation of themanuscript. This work was supported
by National Institute on Drug Abuse grants DA05274
and DA09978 (to M.E.A.) and DA03934 and DA00489
(to P.H.R.).
The authors wish to thank
R eferences
(1) Onaivi, E. TheBiology of Marijuana: From Geneto Behaviour;
Harwood Academic Publishers: Reading, U.K., 2002.
(2) Shire, D.; Calandra, B.; Delpech, M.; Dumont, X.; Kaghad, M.;
Fur, G. L.; Caput, D.; Ferrara, P. Structural features of the
central cannabinoid CB1receptor involved in the binding of the
specific CB1antagonist SR 141716A. J . Biol. Chem. 1996, 271,
6941-6946.
(3) Shire, D.; Calandra, B.; Bouaboula, M.; Barth, F.; Rinaldi-
Carmona, M.; Casellas, P.; Ferrara, P. Cannabinoid receptor
interactions with the antagonists SR 141716A and SR 144528.
Life Sci. 1999, 65, 627-635.
(4) Chin, C. N.; Murphy, J . W.; Huffman, J . W.; Kendall, D. A. The
third transmembrane helix of the cannabinoid receptor plays a
role in the selectivity of aminoalkylindoles for CB2, peripheral
cannabinoid receptor. J . Pharmacol. Exp. Ther. 1999, 291, 837-
844.
(5) Song, Z. H.; Slowey, C. A.; Hurst, D. P.; Reggio, P. H. The
difference between the CB(1) and CB(2) cannabinoid receptors
at position 5.46 is crucial for the selectivity of WIN55212-2 for
CB(2). Mol. Pharmacol. 1999, 56, 834-840.
(6) Hurst, D. P.; Lynch, D. L.; Barnett-Norris, J .; Hyatt, S. M.;
Seltzman, H. H.; Zhong, M.; Song, Z. H.; Nie, J .; Lewis, D.;
Reggio, P. H. N-(Piperidin-1-yl)-5-(4-chlorophenyl)-1-(2,4-dichlo-
rophenyl)-4-methyl-1 H-pyrazole-3-carboxamide (SR141716A)
Interaction with LYS 3.28(192) Is Crucial for Its Inverse
Agonism at the Cannabinoid CB1 Receptor. Mol. Pharmacol.
2002, 62, 1274-1287.
5150 J ournal of Medicinal Chemistry, 2003, Vol. 46, No. 24McAllister et al.

Page 13

(7) McAllister, S. D.; Tao, Q.; Barnett-Norris, J .; Buehner, K.; Hurst,
D. P.; Guarnieri, F.; Reggio, P. H.; Nowell Harmon, K. W.;
Cabral, G. A.; Abood, M. E. A critical role for a tyrosine residue
in the cannabinoid receptors for ligand recognition. Biochem.
Pharmacol. 2002, 63, 2121-2136.
(8) Shi, L.; Liapakis, G.; Xu, R.; Guarnieri, F.; Ballesteros, J . A.;
J avitch, J . A. Beta2 adrenergic receptor activation. Modulation
of the proline kink in transmembrane 6 by a rotamer toggle
switch. J . Biol. Chem. 2002, 277, 40989-40996.
(9) Barnett-Norris, J .; Hurst, D. P.; Buehner, K.; Ballesteros, J . A.;
Guarnieri, F.; Reggio, P. H. Agonist Alkyl Tail Interaction with
Cannabinoid CB1Receptor V6.43/I6.46 Groove Induces a Helix
6 Active Conformation. Int. J . Quantum Chem. 2002.
(10) Farrens, D.; Altenbach, C.; Ynag, K.; Hubbell, W.; Khorana, H.
Requirement of rigid-body motion of transmembrane helices for
light activation of rhodopsin. Science 1996, 274, 768-770.
(11) Gether, U.; Lin, S.; Ghanouni, P.; Ballesteros, J .; Weinstein, H.;
Kobilka, B. Agonists induce conformational changes in trans-
membranedomains III and VI of thebeta2 adrenoceptor. EMBO
J . 1997, 16, 6737-6747.
(12) Ghanouni, P.; Steenhuis, J . J .; Farrens, D. L.; Kobilka, B. K.
Agonist-induced conformational changes in the G-protein-
coupling domain of the beta 2 adrenergic receptor. Proc. Natl.
Acad. Sci. U.S.A. 2001, 98, 5997-6002.
(13) J ensen, A. D.; Guarnieri, F.; Rasmussen, S. G.; Asmar, F.;
Ballesteros, J . A.; Gether, U. Agonist-induced conformational
changes at the cytoplasmic side of transmembrane segment 6
in the beta 2 adrenergic receptor mapped by site-selective
fluorescent labeling. J Biol. Chem. 2001, 276, 9279-9290.
(14) J avitch, J . A.; Fu, D.; Liapakis, G.; Chen, J . Constitutive
activation of the b2 adrenergic receptor alters the orientation
of its sixth membrane-spanning segment. J . Biol. Chem. 1997,
272, 18546-18549.
(15) Hulme, E. C.; Lu, Z. L.; Ward, S.; Allman, K.; Curtis, C. The
conformational switch in 7-transmembrane receptors: the mus-
carinic receptor paradigm. Eur. J . Pharmacol. 1999, 375, 247-
260.
(16) Leff, P. The two-state model of receptor activation. Trends
Pharmacol. Sci. 1995, 16, 89-97.
(17) Samama, P.; Pei, G.; Costa, T.; Cotecchia, S.; Lefkowitz, R. J .
Negative antagonists promote an inactive conformation of the
beta 2-adrenergic receptor. Mol. Pharmacol. 1994, 45, 390-394.
(18) Burley, S.; Petsko, G. Aromatic-aromatic interaction: A mech-
anism of protein structure stabilization. Science1985, 229, 23-
28.
(19) Bouaboula, M.; Perrachon, S.; Milligan, L.; Canat, X.; Rinaldi-
Carmona, M.; Portier, M.; Barth, F.; Calandra, B.; Pecceu, F.;
Lupker, J .; Maffrand, J .-P.; LeFur, G.; Casellas, P. A selective
inverse agonist for central cannabinoid receptor inhibits mito-
gen-activated protein kinase activation stimulated by insulin or
insulin-like growth factor 1. J . Biol. Chem. 1997, 272, 22330-
22339.
(20) Pan, X.; Ikeda, S.; Lewis, D. SR 141716A acts as an inverse
agonist to increase neuronal voltage-dependent Ca2+ currents
by reversal of tonic CB1 cannabinoid receptor activity. Mol
Pharmacol 1998, 54, 1064-1072.
(21) Meschler, J . P.; Kraichely, D. M.; Wilken, G. H.; Howlett, A. C.
Inverse agonist properties of N-(piperidin-1-yl)-5-(4-chlorophe-
nyl)-1-(2,4-dichlorophenyl)-4-methyl-1H-pyrazole-3-carboxam-
ide HCl (SR141716A) and 1-(2-chlorophenyl)-4-cyano-5-(4-
methoxyphenyl)-1H-pyrazole-3-carboxylic acid phenylamide(CP-
272871) for the CB(1) cannabinoid receptor. Biochem. Pharmacol.
2000, 60, 1315-1323.
(22) Thomas, B. F.; Gilliam, A. F.; Burch, D. F.; Roche, M. J .;
Seltzman, H. H. Comparative receptor binding analyses of
cannabinoid agonists and antagonists. J . Pharmacol. Exp. Ther.
1998, 285, 285-292.
(23) Katoch-Rouse, R.; Pavlova, O. A.; Caulder, T.; Hoffman, A. F.;
Mukhin, A. G.; Horti, A. G. Synthesis, structure-activity
relationship, and evaluation of SR141716 analogues: Develop-
ment of central cannabinoid receptor ligands with lower lipo-
philicity. J . Med. Chem. 2003, 46, 642-645.
(24) D’Ambra, T. E.; Estep, K. G.; Bell, M. R.; Eissenstat, M. A.; J osef,
K. A.; Ward, S. J .; Haycock, D. A.; Baizman, E. R.; Casiano, F.
M.; Beglin, N. C. Conformationally restrained analogues of
pravadoline: nanomolar potent, enantioselective, (aminoalkyl)-
indoleagonists of thecannabinoid receptor. J . Med. Chem. 1992,
35, 124-135.
(25) Barnett-Norris, J .; Hurst, D. P.; Lynch, D. L.; Guarnieri, F.;
Makriyannis, A.; Reggio, P. H. Conformational memories and
the endocannabinoid binding site at the cannabinoid CB1recep-
tor. J . Med. Chem. 2002, 45, 3649-3659.
(26) Palczewski, K.; Kumasaka, T.; Hori, T.; Behnke, C. A.; Mo-
toshima, H.; Fox, B. A.; Le Trong, I.; Teller, D. C.; Okada, T.;
Stenkamp, R. E.; Yamamoto, M.; Miyano, M. Crystal structure
of rhodopsin: A G protein-coupled receptor. Science 2000, 289,
739-745.
(27) Barnett-Norris, J .; Guarnieri, F.; Hurst, D. P.; Reggio, P. H.
Exploration of biologically relevant conformations of anand-
amide, 2-arachidonylglycerol, and their analogues using confor-
mational memories. J . Med. Chem. 1998, 41, 4861-4872.
(28) LaLonde, J . M.; Levenson, M. A.; Roe, J . J .; Bernlohr, D. A.;
Banaszak, L. J . Adipocyte lipid-binding protein complexed with
arachidonic acid. Titration calorimetry and X-ray crystallo-
graphic studies. J . Biol. Chem. 1994, 269, 25339-25347.
(29) Song, Z.-H.; Bonner, T. I. A lysine residue of the cannabinoid
receptor is critical for receptor recognition by several agonists
but not WIN55212-2. Mol. Pharmacol. 1996, 49, 891-896.
(30) Pinto, J . C.; Potie, F.; Rice, K. C.; Boring, D.; J ohnson, M. R.;
Evans, D. M.; Wilken, G. H.; Cantrell, C. H.; Howlett, A. C.
Cannabinoid receptor binding and agonist activity of amides and
esters of arachidonic acid. Mol. Pharmacol. 1994, 46, 516-522.
(31) Hillard, C. J .; Manna, S.; Greenberg, M. J .; DiCamelli, R.; Ross,
R. A.; Stevenson, L. A.; Murphy, V.; Pertwee, R. G.; Campbell,
W. B. Synthesis and characterization of potent and selective
agonists of the neuronal cannabinoid receptor (CB1). J . Phar-
macol. Exp. Ther. 1999, 289, 1427-1433.
(32) Bonechi, C.; Brizzi, A.; Brizzi, V.; Francoli, M.; Donati, A.; Rossi,
C. Conformational Analysis of N-arachidonylethanolamide
(Anandamide) Using Nuclear Magnetic Resonance and Theoreti-
cal Calculations. Magn. Reson. Chem. 2001, 39, 432-437.
(33) Barnett-Norris, J .; Hurst, D. P.; Reggio, P. H. The influence of
cannabinoid receptor second extracellular loop conformation on
the binding of CP55940. 2003 Symposium on theCannabinoids;
International Cannabinoid Research Society, Burlington, VT:
Cornwall, Ontario, Canada, 2003; p 78.
(34) Rhee, M.-H.; Nevo, I.; Bayewitch, M. L.; Zagoory, O.; Vogel, Z.
Functional role of tryptophan residues in the forurth transmem-
brane domain of the CB2cannabinoid receptor. J . Neurochem.
2000, 75, 2485-2491.
(35) Ward, W. H.; Timms, D.; Fersht, A. R. Protein engineering and
the study of structuresfunction relationships in receptors.
Trends. Pharmacol. Sci. 1990, 11, 280-284.
(36) Sansom, M. S.; Weinstein, H. Hinges, swivels and switches: the
role of prolines in signaling via transmembrane alpha-helices.
Trends. Pharmacol. Sci. 2000, 21, 445-451.
(37) Ballesteros, J . A.; Deupi, X.; Olivella, M.; Haaksma, E. E.; Pardo,
L. Serine and threonine residues bend alpha-helices in the chi-
(1) ) g(-) conformation. Biophys. J . 2000, 79, 2754-2760.
(38) Hunter, C.; Singh, J .; Thornton, J . π-π interactions: The
geometry and energetics of phenylalanine-phenylalanine inter-
actions in proteins. J . Mol. Biol. 1991, 218.
(39) Sinnokrot, M. O.; Valeev, E. F.; Sherrill, C. D. Estimates of the
ab initiolimit for pi-pi interactions: The benzene dimer. J . Am.
Chem. Soc. 2002, 124, 10887-10893.
(40) Lan, R.; Liu, Q.; Fan, P.; Lin, S.; Fernando, S. R.; McCallion,
D.; Pertwee, R.; Makriyannis, A. Structure-activity relation-
ships of pyrazole derivatives as cannabinoid receptor antago-
nists. J . Med. Chem. 1999, 42, 769-776.
(41) Wiley, J . L.; J efferson, R. G.; Grier, M. C.; Mahadevan, A.;
Razdan, R. K.; Martin, B. R. Novel pyrazole cannabinoids:
insights into CB(1) receptor recognition and activation. J .
Pharmacol. Exp. Ther. 2001, 296, 1013-1022.
(42) Shim, J . Y.; Welsh, W. J .; Cartier, E.; Edwards, J . L.; Howlett,
A. C. Molecular interaction of the antagonist N-(piperidin-1-yl)-
5-(4-chlorophenyl)-1-(2,4-dichlorophenyl)-4-methyl-1H-pyrazole-
3-carboxamidewith theCB1cannabinoid receptor. J . Med. Chem.
2002, 45, 1447-1459.
(43) Huffman, J .; Dai, D.; Martin, B.; Compton, D. Design, Synthesis
and Pharmacology of Cannabimimetic Indoles. Bioorg. Med.
Chem. Lett. 1994, 4, 563-566.
(44) Eissenstat, M. A.; Bell, M. R.; D’Ambra, T. E.; Alexander, E. J .;
Daum, S. J .; Ackerman, J . H.; Gruett, M. D.; Kumar, V.; Estep,
K. G.; Olefirowicz, E. M. Aminoalkylindoles: structure-activity
relationships of novel cannabinoid mimetics. J . Med. Chem.
1995, 38, 3094-3105.
(45) Reggio, P. H.; Basu-Dutt, S.; Barnett-Norris, J .; Castro, M. T.;
Hurst, D. P.; Seltzman, H. H.; Roche, M. J .; Gilliam, A. F.;
Thomas, B. F.; Stevenson, L. A.; Pertwee, R. G.; Abood, M. E.
The bioactive conformation of aminoalkylindoles at the cannab-
inoid CB1and CB2receptors: Insights gained from (E)- and (Z)-
naphthylidene indenes. J . Med. Chem. 1998, 41, 5177-5187.
(46) Huffman, J .; Mabon, R.; Wua, M.-J .; Lu, J .; Hart, R.; Hurst, D.;
Reggio, P.; Wiley, J .; Martin, B. 3-Indolyl-1-naphthylmethanes:
New Cannabimimetic Indoles Provide Evidence for Aromatic
Stacking Interactions with the CB1 Cannabinoid Receptor.
Bioorg. Med. Chem. 2003, 11, 539-549.
(47) Brandl, M.; Weiss, M. S.; J abs, A.; Suhnel, J .; Hilgenfeld, R.
C-H‚‚‚[pi]-Interactions in Proteins. J . Mol. Biol. 2001, 307, 357-
377.
(48) Bracey, M. H.; Hanson, M. A.; Masuda, K. R.; Stevens, R. C.;
Cravatt, B. F. Structural adaptations in a membrane enzyme
that terminates endocannabinoid signaling. Science 2002, 298,
1793-1796.
Aromatic Microdomain at the Cannabinoid CB1ReceptorJ ournal of Medicinal Chemistry, 2003, Vol. 46, No. 24 5151

Page 14

(49) Malkowski, M. G.; Ginell, S. L.; Smith, W. L.; Garavito, R. M.
The productive conformation of arachidonic acid bound to
prostaglandin synthase. Science 2000, 289, 1933-1937.
(50) Ballesteros, J . A.; Weinstein, H. Integrated methods for the
construction of three-dimensional models and computational
probing of structure function relations in G protein-coupled
receptors. Methods in Neuroscience; Academic Press: San Diego,
1995; pp 366-428.
(51) Lin, S.; Sakmar, T. Specific tryptophan UV-absorbance changes
are probes of the transition of rhodopsin to its active state.
Biochemistry 1996, 35, 11149-11159.
(52) Ballesteros, J .; J ensen, A.; Liapakis, G.; Rasmussen, S.; Shi, L.;
Gether, U.; J avitch, J . Activation of the ?2adrenegic receptor
involves disruption of an ionic link between thecytoplasmic ends
of transmembrane segments 3 and 6. J . Biol. Chem. 2001, 276,
29171-29177.
(53) Abood, M. E.; Ditto, K. E.; Noel, M. A.; Showalter, V. M.; Tao,
Q. Isolation and expression of a mouseCB1cannabinoid receptor
gene. Biochem. Pharmacol. 1997, 53, 207-214.
(54) Bramblett, R. D.; Panu, A. M.; Ballesteros, J . A.; Reggio, P. H.
Construction of a 3D model of the cannabinoid CB1 receptor:
Determination of helix ends and helix orientation. LifeSci. 1995,
56, 1971-1982.
(55) Gerard, C. M.; Mollereau, C.; Vassart, G.; Parmentier, M.
Molecular cloning of a human cannabinoid receptor which is also
expressed in testis. Biochem. J . 1991, 279, 129-134.
(56) Tao, Q.; McAllister, S. D.; Andreassi, J .; Nowell, K. W.; Cabral,
G. A.; Hurst, D. P.; Bachtel, K.; Ekman, M. C.; Reggio, P. H.;
Abood, M. E. Role of a conserved lysine residue in the peripheral
cannabinoid receptor (CB2): Evidence for subtype specificity.
Mol. Pharmacol. 1999, 55, 605-613.
(57) Tao, Q.; Abood, M. E. Mutation of thehighly conserved aspartate
residue in the CB1 and CB2 receptors disrupts G-protein
coupling. J . Pharmacol. Exp. Ther. 1998, 285, 651-658.
(58) Cheng, Y. C.; Prusoff, W. H. Relationships between theinhibition
constant (Ki) and the concentration of inhibitor that causes 50%
inhibition (IC50) on an enzymatic reaction. Biochem Pharmacol
1973, 22, 3099-3108.
J M0302647
5152 J ournal of Medicinal Chemistry, 2003, Vol. 46, No. 24McAllister et al.