THE JOURNAL OF BIOLOGICAL CHEMISTRY
© 2005 by The American Society for Biochemistry and Molecular Biology, Inc.
Vol. 280, No. 33, Issue of August 19, pp. 29788 –29795, 2005
Printed in U.S.A.
The Conformation, Location, and Dynamic Properties of the
Endocannabinoid Ligand Anandamide in a Membrane Bilayer*
Received for publication, March 16, 2005, and in revised form, June 9, 2005
Published, JBC Papers in Press, June 17, 2005, DOI 10.1074/jbc.M502925200
Xiaoyu Tian‡, Jianxin Guo‡, Fenmei Yao‡, De-Ping Yang§, and Alexandros Makriyannis‡¶
From the ‡Center for Drug Discovery, Northeastern University, Boston, Massachusetts 02115 and the §Department of
Physics, College of the Holy Cross, Worcester, Massachusetts 01610
The endogenous cannabinoid ligand anandamide is
biosynthesized from membrane phospholipid precursors and is believed to reach its sites of action on the
CB1 and CB2 receptors through fast lateral diffusion
within the cell membrane. To gain a better insight on
the stereochemical features of its association with the
cell membrane and its interaction with the cannabinoid
receptors, we have studied its conformation, location,
and dynamic properties in a dipalmitoylphosphatidylcholine multilamellar model membrane bilayer system.
By exploiting the bilayer lattice as an internal threedimensional reference grid, the conformation and location of anandamide were determined by measuring selected inter- and intramolecular distances between
strategically introduced isotopic labels using the rotational echo double resonance (REDOR) NMR method. A
molecular model was proposed to represent the structural features of our anandamide/lipid system and was
subsequently used in calculating the multispin dephasing curves. Our results demonstrate that anandamide
adopts an extended conformation within the membrane
with its headgroup at the level of the phospholipid polar
group and its terminal methyl group near the bilayer
center. Parallel static 2H NMR experiments further confirmed these findings and provided evidence that anandamide experiences dynamic properties similar to those
of the membrane phospholipids and produces no perturbation to the bilayer. Our results are congruent with a
hypothesis that anandamide approaches its binding site
by laterally diffusing within one membrane leaflet in an
extended conformation and interacts with a hydrophobic groove formed by helices 3 and 6 of CB1, where its
terminal carbon is positioned close to a key cysteine
residue in helix 6 leading to receptor activation.
The membrane lipid bilayer is a ubiquitous molecular assembly into which are embedded a variety of proteins, natural
hormones, and neurotransmitters. Accumulated evidence indicates that many fatty acid-derived lipophilic neurotransmitters
are synthesized, stored, and degraded and also exert their
functions within the lipid membrane (1, 2). Therefore, it has
been suggested that the conformation, location, and orientation
of the ligand in the membrane are critical in determining its
* This research was supported by Grants DA3801 and DA7251 and
Training Grant T32-DA7312 from the National Institute on Drug
Abuse. The costs of publication of this article were defrayed in part by
the payment of page charges. This article must therefore be hereby
marked “advertisement” in accordance with 18 U.S.C. Section 1734
solely to indicate this fact.
¶ To whom correspondence should be addressed: Center for Drug
Discovery, Rm. 116, Mugar Hall, Northeastern University, Boston, MA
02115. Tel.: 617-373-4200; Fax: 617-373-7493; E-mail: a.makriyannis@
neu.edu.
ability to reach and interact productively with its site of action
(3–5). Exploring the conformational and dynamic properties of
these ligands in the membrane can lead to a better understanding of the molecular features involved in their interactions with
the target proteins (6).
N-Arachidonoylethanolamine (anandamide), initially isolated from mammalian brain, has been identified as an endogenous ligand for the two known G protein-coupled cannabinoid
receptors (CB1 and CB2) (7). This endocannabinoid exerts its
activity by modulating several physiological functions such as
pain, cognition, and memory. Site-directed mutagenesis evidence has shown that the anandamide binding sites are embedded in the trans-membrane helices of the receptor (8, 9).
This correlates well with studies of other G protein-coupled
receptors, such as rhodopsin and the ␤-adrenergic receptor,
which provide evidence that their respective ligands interact
within the trans-membrane domains of the receptor (10). Moreover, enzymes that are involved in anandamide biosynthesis
and degradation, including a D-type phosphodiesterase (11) and
fatty acid amide hydrolase (7), are also membrane-bound, and
their respective substrates and products originate from membrane phospholipids (12). Because all of the above mentioned
endocannabinoid-related proteins are membrane-bound, the
conformation, location, and dynamic behavior of anandamide
within the cell membrane would be of particular importance for
a better understanding of the nature of receptor or enzyme
activation/deactivation and may also enhance our ability to
design novel therapeutic medications acting through the endocannabinoid system. Because of the high lipophilicity (clogP ⫽
6.3) of anandamide, this amphipathic ligand is expected to
reside nearly exclusively within the membrane bilayer.
Earlier computational work (13, 14) on the conformational
properties of anandamide has shown that its arachidonoyl component is capable of assuming a variety of conformations in
solution, which can be generally characterized as hairpin (Ushaped), J-shaped, and extended (Fig. 1). Because the structure
of anandamide is highly flexible, its conformation could be very
sensitive to its immediate environment, and the molecule may
exhibit different conformational preferences depending on its
surrounding media. However, there were no experimental
studies on the conformational properties of anandamide in a
bilayer membrane environment. From earlier work with (⫺)⌬9-tetrahydrocannabinol (⌬9-THC)1 we had demonstrated that
the orientation of a lipophilic ligand within an anisotropic
membrane environment does not always conform with the expected predictions based on optimal packing within the phospholipid amphipathic environment (15). Our experiments
showed that this amphipathic cannabinoid does not align as
1
The abbreviations used are: ⌬9-THC, (⫺)-⌬9-tetrahydrocannabinol;
DPPC, dipalmitoylphosphatidylcholine; REDOR, rotational echo double
resonance.
29788
This paper is available on line at http://www.jbc.org
Anandamide Conformation in Lipid Membrane Bilayers
29789
FIG. 1. Left, three possible anandamide
conformations suggested by computer
modeling (4). Right, DPPC structure derived from dimyristoyl phosphatidylcholine crystallographic data (28).
expected with the long axis of its tricyclic component parallel
to the phospholipid chains. Instead, the molecule assumes an
awkward orientation in which the long axis of its tricyclic
core is perpendicular to the bilayer chain (Fig. 2). We were
thus motivated to design a series of experiments to determine
experimentally which of the three computationally identified
conformations of anandamide is the dominant one in an
anisotropic membrane system and to study its location and
dynamic properties within the membrane. Traditionally, this
information is obtained with the help of high resolution NMR
where the ligand conformation is studied in SDS micellar or
other membrane-mimicking environments (16 –18). However,
such approaches suffer from several limitations, one of which
is the highly curved nature of micelles that may not serve as
an ideal membrane model (19). In the case of long chain lipid
messengers, such spectra are further complicated by the fact
that their respective 1H NMR resonances are unresolved
because of severe overlap. In this study, we have employed a
novel approach to examine the location and conformation of
anandamide in a model membrane system consisting of dipalmitoylphosphatidylcholine (DPPC) multilamellar bilayers
using the rotational echo double resonance (REDOR) NMR
technique. We have also used static solid-state 2H NMR to
obtain information on the dynamic properties of anandamide
in the bilayer.
Our experimental strategy was to employ the phospholipid
multilamellar bilayer system not only as a model membrane
environment but also as an internal reference grid for the
incorporated anandamide molecule because this supramolecular lipid assembly is a highly organized lattice (20). The
conformation of anandamide can then be obtained by determining the geometric relationships between anandamide and
the lipid molecules. This was accomplished by identifying key
atom pairs within the anandamide/DPPC bilayer assembly
and determining the respective intra- or intermolecular distances using rotational echo double resonance (REDOR), a
powerful solid-state NMR method capable of accurately
measuring distances between two different nuclei, such as
13
C– 31 P, 13 C– 15 N, and 13 C– 2 H (21, 22). This technique
uniquely allows the measurement of mid-range internuclear
distances and has been applied successfully to study peptide
backbone conformations and ligand-protein binding geometries (23, 24). A distance range of 8.0 Å may be reliably
measured for a pair of high ␥ nuclei, such as 13C and 31P.
Although the measurable distance is shorter for the low ␥ 2H
nucleus, this range is expanded in the experiments described
here because of the dipolar contributions of multiple 2H
FIG. 2. Structures of ⌬9-THC and anandamide. The orientation of
⌬9-THC in a lipid membrane bilayer is indicated by the dashed line,
which represents the long axis of the phospholipid bilayer chains. In
this orientation, the molecule positions its phenolic hydroxyl group
toward the polar side of the membrane and its long axis is perpendicular to the bilayer chains.
atoms with each of the observed 13C nuclei.
In this study, we have introduced 13C, 15N, and 2H isotopic
labeling in strategic positions within anandamide and the
surrounding DPPC molecules and determined the preferred
conformation of anandamide within the bilayer using selected internuclear distances between these labels through a
series of REDOR experiments (Fig. 3). The exact location and
conformation of anandamide in the lipid bilayer were obtained by simulating the multispin REDOR dephasing data
based on our proposed molecular model of the anandamide/
DPPC system. Parallel static solid-state 2H NMR experiments with the anandamide/DPPC system in the liquid crystalline phase (L␣) were used to confirm further our findings
and obtain information on the dynamic properties of this
endocannabinoid within the cell membrane from which it
originates and is believed to reside. This approach may also
be of general interest for studying the conformational properties of other small molecules, peptides, and integral membrane proteins within a membrane system.
29790
Anandamide Conformation in Lipid Membrane Bilayers
FIG. 4. Typical REDOR pulse program used with an XY-8 phase
alternation.
FIG. 3. Structure of DPPC and anandamide. Positions of specific
isotopic labeling in both molecules are shown. DPPC is 2H labeled
independently in each of the 2⬘, 7⬘, or 16⬘ positions. The arrows indicate
the atom pairs for which the dipolar couplings were measured in the
REDOR experiments. The solid line represents an observable REDOR
effect, and the dotted line reflects an internuclear distance beyond the
detection limit.
EXPERIMENTAL PROCEDURES
Materials
2
DPPC was H labeled independently at the 2⬘, 7⬘, and 16⬘ positions of
both sn-1 and sn-2 acyl chains (Fig. 3) to obtain 1,2-[2⬘-2H2]DPPC,
1,2-[7⬘-2H2]DPPC, 1,2-[16⬘-2H3]DPPC (25), whereas [1⬘-13C]anandamide, [20-13C,15N]anandamide, and [20-2H3]anandamide (Fig. 3) were
synthesized using methods described elsewhere (26). Deuterium-depleted water was purchased from Aldrich.
NMR Sample Preparation
Eleven samples of fully hydrated DPPC multilamellar bilayers with
and without anandamide (10 mol %) were prepared. Six of these were
used for two groups of REDOR experiments (see Fig. 3). The three
samples in the first group were [1⬘-13C]anandamide/[31P]DPPC, [1⬘13
C]anandamide/1,2-[2⬘-2H2]DPPC, and [1⬘-13C]anandamide/1,2-[7⬘2
H2]DPPC, where the anandamide 1⬘-13C label was used as the observe
nucleus and the DPPC labels 31P, 2⬘-2H2, or 7⬘-2H2 as the dephased
nucleus. The three samples in the second group were [20-13C]anandamide/1,2-[16⬘-2H3]DPPC, [20-13C]anandamide/1,2-[7⬘-2H2]DPPC, and
[20-13C, 15N]anandamide/unlabeled DPPC, where the anandamide 2013
C label was used as the observe nucleus and DPPC 16⬘-2H3, 7⬘-2H2, or
anandamide 15N as the dephased nucleus. The remaining five samples:
[20-2H3]anandamide/unlabeled DPPC, 1,2-[2⬘-2H2]DPPC, 1,2-[16⬘2
H3]DPPC, unlabeled anandamide/1,2-[2⬘-2H2]DPPC, and unlabeled
anandamide/1,2-[16⬘-2H3]DPPC, were used in the static solid-state 2H
NMR experiments. To prevent lipid oxidation, each sample preparation
was carried out in a nitrogen atmosphere using a glove box. Mixtures of
DPPC (45 mg) and anandamide were first dissolved in chloroform and
the solvent removed by passing a stream of nitrogen over the solution.
The residue was then placed under vacuum for 10 h, and 2H-depleted
water was added to the dried sample to produce a 50% (w/w) lipid/water
preparation. Each semisolid membrane preparation was then introduced into a 4-mm glass tube, sealed under vacuum, gently vortexed,
equilibrated at 50 °C for 20 min, and subsequently stored at ⫺20 °C.
Solid-state NMR Experiments
REDOR Experiments—All NMR experiments were carried out on a
Chemagnetics CMX300 solid-state spectrometer (Fort Collins, CO).
Each sealed sample was first allowed to equilibrate for 20 min at 50 °C
and packed into a 5-mm zirconia rotor. This was then inserted into a
Chemagnetics triple resonance probe and cooled rapidly to ⫺40 °C. The
rotor was driven by nitrogen gas, whereas the gas flow and temperature
were controlled by a Chemagnetics air and temperature controller
equipped with a dry ice/acetone bath. The magic angle spinning speed
was set to 4,000 Hz and controlled by the MAS controller with a
fluctuation of ⫾2 Hz. The observe nucleus was always 13C resonating at
75.43 MHz, whereas the dephased nucleus was 2H, 31P, or 15N at 46.05,
121.44, or 30.39 MHz, respectively. A contact pulse (1.6-ms duration
and 50-kHz power) was applied in the proton channel (299.99 MHz) for
cross-polarization, and its power was increased to 82 kHz for proton
decoupling (see Fig. 4). The observed 13C resonances in the experiments
are well resolved and separated from other chemical shift signals in the
13
C spectrum of the sample. To invert the deuterium nuclei for sufficient dephasing, a composite XY-4 ␲-pulse scheme was employed on the
2
H channel, whereas an XY-8 composite pulse scheme was used on the
13
C channel to compensate for possible pulse imperfections (27). A
sample of [1-13C,2-2H3]sodium acetate (Cambridge Isotope Laboratories, MA) was used as a standard for all 13C observe, 2H dephased
REDOR experiments.
Static Solid-state 2H NMR—All of the deuterium spectra were obtained using a Chemagnetics CMX300 spectrometer operating at 46.05
MHz with a wideline probe at 42 °C. The quadrupole echo pulse sequence, [(␲/2)x-␶-(␲/2)y], was employed with 2.5 ␮s for the 2H ␲/2 pulse,
35 ␮s for ␶, and 200 ms for recycle delay. A total of 5,000 echoes were
accumulated for each spectrum. Before recording a spectrum, each
sample was held at 50 °C in the probe for 15 min to ensure
complete equilibration.
RESULTS AND DISCUSSION
Solid-state REDOR NMR experiments were used to elucidate
the location and conformation of anandamide within the DPPC
multilamellar membrane bilayer by detecting heteronuclear
dipolar couplings between anandamide and the phospholipid
molecules, as well as between the 20-13C- and 15N-labeled sites
of anandamide. Each of our REDOR experiments consists of
two parallel parts: the 13C signal is first observed without the
dephasing pulse train and is then repeated with the dephasing
pulse train applied. The intensity difference ⌬S between the
two spectra can be directly related to the dipolar coupling of
each spin pair, from which the internuclear distance can be
deduced (28). The first group of experiments (1⬘-13C-observe)
was aimed at determining the location of the anandamide
headgroup, and we observed positive effects (⌬S ⱖ 0) in both
the 1⬘-13C-observe, 31P-dephased (1⬘-13C/31P) and the 1⬘-13Cobserve, 2⬘-2H2-dephased (1⬘-13C/2⬘-2H2) experiments. However, no effects (⌬S ⫽ 0) were observed in the 1⬘-13C/7⬘-2H2
REDOR experiment. The second group (20-13C-observe) was
designed to identify the location of the anandamide terminal
methyl within the DPPC bilayer. Only in the 20-13C/16⬘-2H3
experiment did we observe positive effects, whereas the 20-13C/
7⬘-2H2 and 20-13C/15N experiments showed no effects, evidence
that these internuclear distances are beyond the REDOR detection limits.
Location of Anandamide within the DPPC Lipid Bilayer
Anandamide Headgroup (1⬘-13C-Observe REDOR Experiments)—The location of the anandamide headgroup within the
lipid bilayer was identified from three REDOR experiments.
Fig. 5, left panel, shows four pairs of 1⬘-13C/31P REDOR spectra
Anandamide Conformation in Lipid Membrane Bilayers
29791
FIG. 5. Anandamide 13C REDOR spectral pairs dephased by
P (left panel) and 2H (right panel) of 1,2-[2ⴕ-2H2]DPPC. The
numbers of rotor cycles for each pair are as indicated, and, for each pair,
the full echo spectrum is shown on the left. All spectra are obtained with
128 scans at spinning speed of 4 kHz and repetition time of 5 s. The
peak at 46.0 ppm is the result of the 1⬘-13C label on the anandamide
headgroup.
FIG. 6. Anandamide 13C REDOR spectral pairs dephased by 2H
labels of 1,2-[16ⴕ-2H3]DPPC (left panel) and 1,2-[7ⴕ-2H2]DPPC
(right panel). The numbers of rotor cycles for each pair are as indicated, and, for each pair, the full echo spectrum is shown on the left. All
spectra are obtained with 128 scans at spinning speed of 4 kHz and
repetition time of 5 s. The 13C resonance at 16.0 ppm is the result of the
20-13C label at the anandamide terminal methyl group.
in which the 13C resonance was detected at 46.0 ppm. The
experiment revealed an intensity difference (⌬S) in each pair of
resonances which became progressively pronounced as the rotor cycle increased from 16 to 64. The 13C signal was almost
completely eliminated because of the 31P dephasing at rotor
cycle 64, indicating a strong coupling between the 1⬘-13C of
anandamide and 31P of the DPPC headgroup. In the second
experiment, pairs of 1⬘-13C/2⬘-2H2 REDOR spectra were collected, four of which are shown in Fig. 5, right panel. Here
again, the signal intensity difference ⌬S for each pair increased
noticeably as the rotor cycle progressed from 16 to 64, indicating that the 13C label of the anandamide headgroup is located
in the proximity of the 2⬘-2H labels of DPPC. Conversely, no
differences in the signal intensity were observed in the 1⬘-13C/
7⬘-2H2 experiments (data not shown), indicating that the dipolar coupling between 1⬘-13C of anandamide and the deuterons
at the 7⬘ position of DPPC is beyond the REDOR detection
limit. Collectively, the results from the above three sets of
experiments provide evidence that the headgroup of anandamide is located at the water/lipid interface. Such a location also
allows for possible intermolecular hydrogen bonding between
the anandamide headgroup hydroxyl and the phosphate
of DPPC.
Anandamide Terminal Methyl Group (20-13C-Observe REDOR Experiments)—To determine the location of the anandamide terminal methyl group within the lipid bilayer, we used
[20-13C]anandamide in 20-13C/16⬘-2H3, 20-13C/7⬘-2H2, and 2013
C/15N REDOR experiments. Pairs of 20-13C/16⬘-2H3 REDOR
spectra exhibited large ⌬S values (Fig. 6, left panel), providing
evidence for strong coupling and close proximity between the
20-13C label on anandamide and the terminal methyl deuterons of the DPPC acyl chains. Conversely, the 20-13C/7⬘-2H2
REDOR spectral pairs showed no discernible intensity differences (Fig. 6, right panel), indicating that the terminal methyl
group of anandamide is beyond the measurable range from the
middle of the DPPC acyl chains. The 20-13C/15N REDOR experiments using doubly labeled anandamide with 15N in its
headgroup and 13C in the terminal methyl showed essentially
no difference in their respective intensities (data not shown),
thus providing evidence that the anandamide terminal methyl
group is also not in close proximity with its own headgroup.
Conformation of Anandamide within the
DPPC Lipid Bilayer
31
Anandamide Adopts the Extended Conformation in Lipid
Bilayer—The above six REDOR experiments provided evidence
for the location of both the headgroup and terminal methyl of
anandamide with respect to the DPPC lipid bilayer. The results
also allow for a clear choice among the three possible anandamide conformations. In the U- or J-shaped conformation, the
anandamide 20-13C label would be located near the DPPC
2⬘-2H2 or 7⬘-2H2 label, respectively, and should have led to
strong REDOR effects in the corresponding experiments. However, if anandamide adopts the extended conformation in lipid
bilayers, we would expect the anandamide 20-13C label to be
near the DPPC 16⬘-methyl group. Because strong REDOR effects were observed only in the 20-13C/16⬘-2H3 experiment, we
therefore concluded that anandamide must exist, at least predominantly, in an extended conformation.
Structural Model for Anandamide within the DPPC Bilayer—In our experiment, each measured dipolar coupling may
result from interactions of the 13C-observe with nuclei from
multiple sites as one anandamide molecule is surrounded by a
number of DPPC molecules. To obtain a more quantitative
interpretation of our REDOR data, we have constructed a molecular model representing the geometric relationship between
anandamide and DPPC, using the Accelrys Insight II/Discover
modeling software. The DPPC bilayer assembly in the subgel
Lc⬘ phase was constructed based on earlier x-ray crystallographic data (20, 29). When an anandamide molecule adopts
the extended conformation, it structurally resembles a lipid
acyl chain. Based on this model, one anandamide molecule
occupies a site similar to that of one of the DPPC acyl chains as
shown in Fig. 7, laterally shifting the DPPC acyl chains from
sites AB to BC. In this representation, the tail of the anandamide molecule is surrounded by six DPPC acyl chains arranged in a hexagonal structure, whereas the anandamide
headgroup is surrounded by four phosphate headgroups arranged in a rhomboidal configuration. Fig. 8 depicts side and
top views of an anandamide molecule surrounded by four
DPPC molecules within one bilayer leaflet in which the intermolecular 1⬘-13C/31P, 1⬘-13C/2⬘-2H2, and 20-13C/16⬘-2H3 geometric relationships are represented.
29792
Anandamide Conformation in Lipid Membrane Bilayers
FIG. 7. Two-dimensional arrangement of the DPPC acyl chain lattice (●) and headgroup supralattice (f). In the left panel, the two
acyl chains of a DPPC molecule occupy sites A and B. In the right panel, an anandamide molecule occupies site A, laterally shifting the DPPC
molecule within the lattice to sites B and C. Anandamide is surrounded by six acyl chains (hexagon) and four headgroups (rhombus) of the four
neighboring DPPC molecules.
FIG. 8. Three-dimensional model of the anandamide/DPPC lattice. Left, the side view shows four DPPC molecules surrounding one
anandamide molecule, and the top view shows the projections of six acyl chains forming a hexagon and four headgroups in a rhomboidal
arrangement. Right, spatial relationships between 1⬘-13C of anandamide and the 31P of DPPC, between the 1⬘-13C of anandamide and the 2⬘-2H2
of the DPPC sn-2 chain, and between the 20-13C of anandamide and the 16-2H3 of DPPC.
Numerical Analysis of the Multispin REDOR Dephasing—
For a multispin system, the REDOR intensity difference ⌬S/S0
can be calculated by (30)
⌬S
1
⫽1⫺
S0
8␲ 2
冕 冕 冕 写 冋
2␲
␲
2␲
cos
␺⫽0
␪⫽0
␾⫽0
i
册
Nc␻Di
4 冑2关n̂⬘共i兲 䡠 ŷ兴关n̂⬘共i兲 䡠 ẑ兴
␯R
d␾sin␪d␪d␺
(Eq. 1)
where ␯R is the rotor frequency, Nc is the number of rotor
cycles,
␻ Di ⫽
␥ 1␥ 2h
4 ␲ 2r i3
(Eq. 2)
is the dipolar coupling constant of the ith 2H or 31P to 13C,
and n̂⬘(i) is the Euler rotation of the unit vector from the ith
2
H or 31P to 13C. To simulate the experimental REDOR
dephasing curve (⌬S/S0 versus Nc), we developed a computer
integration program using the Gaussian quadrature numerical algorithm under a Microsoft FORTRAN PowerStation.
Based on our model, the 1⬘-13C/31P rhomboidal arrangement
requires two distinct 13C–31P distances r1 and r2. In our
numerical simulation, a series of REDOR dephasing curves
are generated by systematically varying these two distances.
Fig. 9a shows three of the simulated curves where the best fit
is for values of r1 ⫽ 5.7 Å and r2 ⫽ 8.4 Å, each with an
uncertainty of about ⫾ 0.2 Å. According to our model, the 2H
labels contributing to the 13C–2H dephasing are equidistant
for the observed 1⬘- or 20-13C of anandamide. For this reason,
only one r value was required in our calculations for the
1⬘-13C/2⬘-2H2 and 20-13C/16⬘-2H3 experiments. The dephasing
curves for the anandamide 1⬘-13C and DPPC sn-2-[2⬘-2H2]
interactions within the triangular arrangement are shown in
Fig. 9b, with an optimal distance of 6.0 Å ⫾ 0.5 Å. For the
20-13C/16⬘-2H3 interactions within the hexagonal arrangement, the simulated REDOR curves are plotted in Fig. 9c.
The best fit for the experimental data corresponds to a distance of r ⫽ 6.5 Å, with an upper limit of 7.0 Å and a lower
limit of 6.0 Å.
Anandamide Conformation in Lipid Membrane Bilayers
29793
FIG. 9. Theoretical fittings (solid curves) to the REDOR experimental data (solid triangles) from the 1ⴕ-13C/31P (a), 1ⴕ-13C/2ⴕ-2H2 (b),
and 20-13C/16ⴕ-2H3 (c) experiments. Each curve was calculated based on the indicated distances and the specific geometric arrangement.
The obtained distances, in conjunction with the structural
coordinates of the DPPC supramolecular assembly, unequivocally establish that anandamide adopts an extended conformation with an estimated total “length” of 22 Å in the membrane
bilayer.
Dynamic Properties of Anandamide in the DPPC Liquid
Crystalline Bilayer
The above REDOR experiments were carried out at a low
temperature (⫺40 °C) by necessity to minimize molecular motions within the bilayer system and allow for the measurement
of dipolar interactions between specific atoms within the anandamide/DPPC assembly. To obtain parallel structural information on the anandamide/DPPC system under more “physiological” conditions, we carried out a series of static solid-state 2H
NMR experiments using DPPC 2H-labeled at the 2⬘ and 16⬘
positions as well as anandamide carrying 2H labels at its terminal methyl group.
The spectra in Fig. 10 were obtained at 42 °C, a temperature
at which the fully hydrated DPPC bilayer exists in the liquid
crystalline L␣ phase. The 2H spectrum from 1,2[2⬘-2H2]DPPC
confirmed that our membrane preparation exists in the liquid
crystalline L␣ phase and showed a set of three superimposed
Pake patterns with quadrupolar splittings of ⌬␯Q ⫽ 12.0, 17.8,
and 26.7 kHz, caused by 2-[2⬘S-2H], 2-[2⬘R-2H], and 1-[2⬘-2H2],
respectively, based on our earlier assignments (25). The 2H
spectrum caused by 1,2-[16⬘-2H3]DPPC from the terminal
methyl groups has a ⌬␯Q value of 3.1 kHz. This narrower
splitting reflects a considerably freer movement of this segment of the acyl chain. Addition of 10% (molar) anandamide to
each of the two DPPC preparations leads only to very small
increases in the ⌬␯Q values, indicating that this endocannabinoid ligand has only a minor effect on the dynamic properties of
the phospholipid bilayer.
The above data suggest an anandamide/DPPC model in
which the lipophilic ligand is well integrated into the bilayer
membrane with its long axis parallel to the phospholipid acyl
chains. Further support for this model is provided by the 2H
experiment from the terminal methyl group of anandamide in
which the quadrupolar splitting (⌬␯Q ⫽ 3.0 kHz) is very similar
to the respective spectrum from the 16⬘ methyl groups of
DPPC, suggesting that the terminal methyl groups from anandamide and DPPC exhibit similar motional properties and
arguably occupy a similar space within the bilayer.
29794
Anandamide Conformation in Lipid Membrane Bilayers
FIG. 10. 2H NMR spectra of a fully hydrated DPPC multilamellar membrane preparation of 1,2-[2⬘-2H2]DPPC (A) and 1,2-[16⬘-2H3]DPPC (B).
Spectra at the bottom are from the membrane preparation with 10 mol % anandamide incorporated. Spectrum C is from 10 mol % [202
H3]anandamide/DPPC preparation. All spectra are recorded at 42 °C.
FIG. 11. The endocannabinoid anandamide assumes an extended conformation in the bilayer with its polar
group at the same level as the polar
phospholipid head groups. It diffuses
laterally to the binding site of the CB1 receptor and interacts with a hydrophobic
groove formed by helices 3, 5, and 6. The
yellow star indicates the position of Cys-47
on helix 6 identified as a site of interaction
with the terminal carbon of anandamide
using covalent ligand experiments.
It is interesting to compare our findings with work related
to the structural properties of unsaturated membrane lipids because anandamide is derived from polyunsaturated
phospholipids. Recent studies involving unsaturated diaryl
phospholipids such as 1-palmitoyl-2-docosahexaenoyl-sn-glycero-3-phosphocholine (PDPC) and 1-stearoyl-2-docosahexaenoyl-sn-glycero-3-phosphocholine (SDPC) show that in the gel
L␤ phase, both the saturated and the polyunsaturated chains
adopt the all-trans and extended (angle iron or helical) conformations, respectively (31, 32). A similar result was obtained
from studies on PDPC and 1-stearoyl-2-arachidonoyl-sn-glycero-3-phosphocholine (SAPC) molecules in the liquid crystalline L␣ phase using either computational or 2H NMR approaches (33, 34). Our results now indicate that anandamide,
which represents a single chain polyunsaturated lipid and is
devoid of a phospholipid headgroup, also assumes such a preferred extended conformation when incorporated in a phospholipid bilayer.
How Anandamide Reaches Its Active Site
at the CB1 Receptor
The findings reported here serve to expand our current understanding on the structural features of the interaction of
anandamide with cell membrane and allow us to speculate on
the manner with which this endogenous ligand interacts with
its respective receptors. Based on substantial experimental
evidence, it has been proposed that lipophilic ligands interact
with their respective target proteins by first partitioning into
the membrane bilayer where they assume a preferred location
and orientation (5). We can thus postulate that the lipophilic
anandamide resides predominantly within the membrane bilayer either following its release within a neural synapse or
after its enzymatic synthesis within the cell membrane by
membrane-bound enzymes (7). While in the bilayer, the endocannabinoid ligand engages in fast lateral diffusion within the
bilayer leaflet before undergoing a productive interaction with
the receptor. Our results supporting an extended conformation
for anandamide with its headgroup at the level of the bilayer
phosphate and a length of ⬃22.0 Å suggest that the CB1
cannabinoid receptor may have a ligand entry port of equivalent size. Our results also demonstrate that such an alignment
does not lead to any unfavorable membrane perturbations.
Such an interaction allows anandamide to access helices 3 and
6, which are believed to be involved in CB1 receptor activation
(Fig. 11). This hypothesis is congruent with data2 suggesting
that the terminal five-carbon chain of the anandamide arachi2
R. Picone and A. Makriyannis, unpublished results.
Anandamide Conformation in Lipid Membrane Bilayers
donoyl moiety interacts with the CB1 receptor through a hydrophobic groove situated at the level of the bilayer center and
involved with the participation of hydrophobic residues V6.43
and I6.46 (35). We therefore propose that such an extended
conformation of anandamide in the lipid bilayer may facilitate
a productive interaction between this endocannabinoid and its
CB1 receptor. The same hydrophobic groove has been shown to
be involved in the interaction of the five carbon side chain of the
plant derived cannabinergic ligand, ⌬9-THC, and some of its
more potent longer chain classical cannabinoid analogs. Work
from our laboratory has confirmed that a cysteine residue in
helix 6 (Cys-47) reacts covalently with an electrophilic isothiocyanate group introduced at the terminal carbon of the cannabinoid ligand side chain. It is believed that this cysteine residue
is a conserved key residue involved in the activation of many G
protein-coupled receptors. Our hypothesis is strengthened further by recent data suggesting a similar covalent interaction
between the cysteine 47 and an anandamide derivative carrying an isothiocyanate group at its terminal C-20 carbon.3
CONCLUSIONS
Our solid-state REDOR experiments provide evidence for
anandamide existing in an extended conformation within
DPPC bilayers in the subgel phase. Parallel experiments using
static solid-state 2H NMR support a similar conformation for
anandamide in the L␣ liquid crystalline bilayer system. In the
phospholipid bilayer environment, the arachidonoyl chain of
anandamide assumes an orientation parallel to the lipid acyl
chains with its terminal methyl group near the bilayer center.
The anandamide headgroup is located near the bilayer lipid/
water interface and may be engaged in hydrogen bonding interactions with the lipid phosphate groups. Furthermore, our
2
H NMR experiments provide evidence that in the liquid crystalline phase, the dynamic properties of anandamide are generally similar to those of the bilayer phospholipids and indicate
that while in the above location and conformation, the endocannabinoid does not lead to unfavorable perturbation of the
membrane bilayer. The results are congruent with a hypothesis
that anandamide approaches its binding site by laterally diffusing within one membrane leaflet in an extended conformation and activates the cannabinoid receptors by interacting
with a hydrophobic groove formed by helices 3 and 6 while its
3
A. Makriyannis and I. Chen, unpublished results.
29795
terminal carbon is very closely positioned to a key cysteine
residue in helix 6.
REFERENCES
1. Bracey, M. H., Hanson, M. A., Masuda, K. R., Stevens, R. C., and Cravatt, B. F.
(2002) Science 298, 1793–1796
2. Piomelli, D. (2003) Nat. Rev. Neurosci. 4, 873– 884
3. Herbette, L. G., Chester, D. W., and Rhodes, D. G. (1986) Biophys. J. 49, 91–93
4. Makriyannis, A. (1995) in Cannabinoid Receptors (Pertwee, R. G., ed) pp.
87–115, Academic Press, London
5. Xie, X. Q., Melvin, L. S., and Makriyannis, A. (1996) J. Biol. Chem. 271,
10640 –10647
6. Inooka, H., Ohtaki, T., Kitahara, O., Ikegami, T., Endo, S., Kitada, C., Ogi, K.,
Onda, H., Fujino, M., and Shirakawa, M. (2001) Nat. Struct. Biol. 8,
161–165
7. Devane, W. A., Hanaus, L., Breuer, A., Pertwee, R., Stevenson, L., Griffing, F.,
Gibson, D., Mandelbaum, A., Etinger, A., and Mechoulam, R. (1992) Science
258, 1946 –1949
8. Hucho, F. (1993) Neurotransmitter Receptors 24, 3–14
9. Song, Z. H., and Bonner, T. I. (1996) Mol. Pharmacol. 49, 891– 896
10. Henderson, R., Baldwin, J. M., Ceska, T. A., Zemlin, F., Beckmann, E., and
Downing, K. H. (1990) J. Mol. Biol. 213, 899 –929
11. Cadas, H. G., Gaillet, S., Beltramo, M., Venance, L., and Piomelli, D. (1996)
J. Neurosci. 16, 3934 –3942
12. Childers, S. R., and Breivogel, C. S. (1998) Drug Alcohol Dependence 51,
173–187
13. Tong, W., Collantes, E. R., and Welsh, W. J. (1998) J. Med. Chem. 41,
4207– 4215
14. Barnett-Norris, J., Guarnieri, F., Hurst, D. P., and Reggio, P. H. (1998) J. Med.
Chem. 41, 4861– 4872
15. Makriyannis, A., Banijamali, A., Jarrell, H. C., and Yang, D. P. (1989) Biochim. Biophys. Acta 986, 141–145
16. Mcdonnell, P. A., and Opella, S. J. (1993) J. Magn. Reson. B. 102, 120 –125
17. Neidigh, J. W., and Andersen, N. H. (2002) Biopolymer 65, 354 –361
18. Palian, M. M., and Boguslavsky, V. I. (2003) J. Am. Chem. Soc. 125, 5823–5831
19. Guo, J., Pavlopoulos, S., Tian, X., Lu, D., Nikas, S. P., Yang, D.-P., and
Makriyannis, A. (2003) J. Med. Chem. 46, 4838 – 4846
20. Katsaras, J., Raghunathan, V. A., Dufourc, E. J., and Dufourcq, J. (1995)
Biochemistry 34, 4684 – 4688
21. Gullion, T., and Schaefer, J. (1989) J. Magn. Reson. 81, 196 –200
22. Schaefer, J. (1998) Ernst Schering Res. Found. Workshop 26, 25–52
23. Mueller, D. D., Schmidt, A., Pappan, K. L., McKay, R. A., and Schaefer, J.
(1995) Biochemistry 34, 5597–5603
24. Hirsh, D. J., Hammer, J., Maloy, W. L., Blazyk, J., and Schaefer, J. (1996)
Biochemistry 35, 12733–12741
25. Yang, D. P., Mavromoustakos, T., Beshah, K., and Makriyannis, A. (1992)
Biochim. Biophys. Acta 1103, 25–36
26. Yao, F.-M., Palmer, S. L., Khanolkar, A. D., Tian, X., Guo, J., and Makriyannis,
A. (2003) J. Labelled Compounds Radiopharmaceut. 46, 115–129
27. Sack, I., Goldbourt, A., Vega, S., and Buntkowsky, G. (1999) J. Magn. Reson.
138, 54 – 65
28. Pan, Y., Gullion, T., and Schaefer, J. (1990) J. Magn. Reson. 90, 330 –340
29. Hauser, H., Pascher, I., Pearson, R. H., and Sundell, S. (1981) Biochim.
Biophys. Acta 650, 21–51
30. Goetz, J. M., and Schaefer, J. (1997) J. Magn. Reson. 127, 147–154
31. Barry, J. A., Trouard, T. P., Salmon, A., and Brown, M. F. (1991) Biochemistry
30, 8386 – 8394
32. Binder, H., and Gawrisch, K. (2001) J. Phys. Chem. B 105, 12378 –12390
33. Applegate, K. R., and Glomset, J. A. (1986) J. Lipid Res. 27, 658 – 680
34. Applegate, K. R., and Glomset, J. A. (1991) J. Lipid Res. 32, 1635–1644
35. Barnett-Norris, J., Hurst, D. P., Buehner, K., Ballesteros, J. A., Guarnieri, F.,
and Reggio, P. H. (2002) Int. J. Quantum Chem. 88, 76 – 86