Stereochemical selectivity of methanandamides for the CB1 and CB2 cannabinoid receptors and their metabolic stability

Article abstract of DOI:10.1016/S0968-0896(01)00088-8

Several chiral, analogues of the endogenous cannabinoid receptor ligand, arachidonylethanolamide (anandamide), methylated at the 2,1′ and 2′ positions using asymmetric synthesis were evaluated in order to study (A) stereoselectivity of binding to CB1 and CB2 cannabinid receptors; and (b) metabolic stability with regard to anandamide amidase. Enantiomerically pure 2-methyl arachidonic acids were synthesized through diastereoselective methylation of the respective chiral 2-oxazolidinone enolate derivaties and CB1 and CB2 receptor affinities of the resulting chiral anandamides were evaluated using a standard receptor binding assay. Introduction of a single 2-methyl group increased affinity for CB1, led to limited enantioselectivity and only modestly improved metabolic stability. However, a high degree of enantio- and diastereoselectivity was observed for the 2,1′-dimethyl analogues. (R)-N-(1-methyl-2-hydroxyethyl)-2-(R)-methyl-arachidonamide (4) exhibited the highest CB1 receptor affinity in this series with a K_i of 7.42 nM, an at least 10-fold improvement on anandamide (K_i = 78.2 nM). The introduction of two methyl groups at the 2-position of anandamide led to no change in affinity for CB1 but somewhat enhanced metabolic stability. Conversely, chiral headgroup methylation in the 2-gem-dimethyl series led to chiral analogues possessing a wide range of CB1 affinities. Of these the (S)-2,2,2′-trimethyl analogue (12) had the highest affinity for CB1 almost equal to that of anandamide. In agreement with our previous anandamide structure-activity relationship work, the analogues in this study showed high selectivity for the CB1 receptor over CB2. The results are evaluated in terms of stereochemical factors affecting the ligand's affinity for CB1 using receptor-essential volume mapping as an aid. Based on the results, a partial CB1 receptor site model is proposed, that bears two hydrophobic pockets capable of accommodating 1′- and 2-methyl groups. Copyright

Full text of DOI:10.1016/S0968-0896(01)00088-8

Bioorganic & Medicinal Chemistry 9 (2001) 1673–1684
Stereochemical Selectivity of Methanandamides for the CB1 and
CB2 Cannabinoid Receptors and Their Metabolic Stability
Andreas Goutopoulos,a Pusheng Fan,a Atmaram D. Khanolkar,a Xian-Qun Xie,a
Sonyuan Lina and Alexandros Makriyannisa,b,
*
aDepartment of Pharmaceutical Sciences, and Center for Drug Discovery, University of Connecticut, Storrs, CT 06269, USA
bDepartment of Molecular and Cell Biology, University of Connecticut, Storrs, CT 06269, USA
Received 26 May 2000; accepted 13 February 2001
Abstract—Several chiral, analogues of the endogenous cannabinoid receptor ligand, arachidonylethanolamide (anandamide),
methylated at the 2,10 and 20 positions using asymmetric synthesis were evaluated in order to study (a) stereoselectivity of binding
to CB1 and CB2 cannabinoid receptors; and (b) metabolic stability with regard to anandamide amidase. Enantiomerically pure
2-methyl arachidonic acids were synthesized through diastereoselective methylation of the respective chiral 2-oxazolidinone enolate
derivatives and CB1 and CB2 receptor affinities of the resulting chiral anandamides were evaluated using a standard receptor
binding assay. Introduction of a single 2-methyl group increased affinity for CB1, led to limited enantioselectivity and only
modestly improved metabolic stability. However, a high degree of enantio- and diastereoselectivity was observed for the 2,10-
dimethyl analogues. (R)-N-(1-methyl-2-hydroxyethyl)-2-(R)-methyl-arachidonamide (4) exhibited the highest CB1 receptor affinity
in this series with a Ki of 7.42 nM, an at least 10-fold improvement on anandamide (Ki=78.2 nM). The introduction of two methyl
groups at the 2-position of anandamide led to no change in affinity for CB1 but somewhat enhanced metabolic stability. Con-
versely, chiral headgroup methylation in the 2-gem-dimethyl series led to chiral analogues possessing a wide range of CB1 affinities.
Of these the (S)-2,2,20-trimethyl analogue (12) had the highest affinity for CB1 almost equal to that of anandamide. In agreement
with our previous anandamide structure–activity relationship work, the analogues in this study showed high selectivity for the CB1
receptor over CB2. The results are evaluated in terms of stereochemical factors affecting the ligand’s affinity for CB1 using receptor-
essential volume mapping as an aid. Based on the results, a partial CB1 receptor site model is proposed, that bears two hydrophobic
pockets capable of accommodating 10- and 2-methyl groups # 2001 Elsevier Science Ltd. All rights reserved.
Introduction
Anandamide’s pharmacological profile was generally
found9 to parallel that of Á9-THC. However, the endo-
genous ligand is somewhat less potent, with a rapid
onset and shorter duration of action,9,10 presumably
due to metabolic instability which is attributed to the
presence of anandamide amidase, an enzyme possessing
amidohydrolase activity.12ꢀ15 This enzyme was shown
to be colocalized in the same brain regions as the CB1
cannabinoid receptor.14 The hydrolytic breakdown of
anandamide can be prevented in vitro by phenylmetha-
nesulfonyl fluoride (PMSF), a general serine protease
inhibitor.12,13 For this purpose, PMSF is included in the
binding assay in structure–activity relationship (SAR)
studies of anandamide analogues where amidase-cata-
lyzed anandamide hydrolysis might be a complicating
factor.16 In addition to anandamide, two other unsatu-
rated fatty acid ethanolamides, dihomo-g-linoleno-
lethanolamide and docosatetraenoylethanolamide, were
isolated from porcine brain and shown to possess
In 1992, arachidonylethanolamide (anandamide) was
isolated from porcine brain and considered the first,
putative endogenous cannabinoid ligand.1 Since then,
experimental evidence has supported the postulate that
this ligand plays an important role in cannabinoid biol-
ogy. Anandamide was, thus, shown to have inhibitory
effects in N-type calcium-currents through a pertussis
toxin-sensitive pathway in N18 neuroblastoma cells,2
forskolin-stimulated adenylate cyclase activity3,4 and
electrically evoked twitch response of the mouse vas
deferens.5,6 In vivo, it was shown7ꢀ11 to produce the
four characteristic effects of cannabimimetics, namely,
analgesia, hypothermia, hypoactivity and catalepsy.
*Corresponding author at first affiliation. Tel.: +1-860-486-2133; fax:
+1-860-486-3089; e-mail: makriyan@uconnvm.uconn.edu
0968-0896/01/$ - see front matter # 2001 Elsevier Science Ltd. All rights reserved.
PII: S0968-0896(01)00088-8

1674
A. Goutopoulos et al. / Bioorg. Med. Chem. 9 (2001) 1673–1684
cannabimimetic properties.17 Recently,
a
system
neurotransmitters, endogenous ligands and prototypical
drug molecules has served as an excellent vehicle for
studying stereochemical recognition in ligand–receptor
interactions, as well as a means of increasing hydrolytic
stability. Subsequently, the stereochemical implications
accompanying methyl group substitution in ananda-
mide are examined using molecular modeling as an aid.
responsible for the reuptake of anandamide was par-
tially characterized18 and the ligand’s structural
requirements for binding and transport were identi-
fied.19 Although the physiological role of anandamide is
not yet fully understood, recent evidence suggests that
this cannabinoid receptor ligand, at least in the dorsal
stratium, acts as an autacoid neuromodulator counter-
ing D2-activation.20,21
Results
Several SAR studies22ꢀ34 of anandamides have already
identified some of the stereoelectronic requirements for
interactions with the CB1 receptor. Most of the pub-
lished work has addressed the SAR of the head group
and more limited work has focused on structural mod-
ifications of the hydrophobic tail. It was thus shown
that, in close analogy with the classical cannabinoids,
replacement of the terminal five carbon atom chain of
anandamide by a dimethylheptyl chain, resulted in ana-
logues possessing higher affinity for CB1.22,23 Substitu-
tion of the arachidonyl chain with other fatty acid
chains having fewer than 20 carbon atoms or three
double bonds or with trans or non o-6 configuration,
leads to dramatic loss of affinity.4,24,29,30,34 In another
study, anandamide was oxidized by three different
lipoxygenases resulting in monohydroxy products at
different positions.32 Only the product of 11-lipoxy-
genase showed receptor affinity comparable to that of
anandamide. The CB1 SAR of the anandamide head
group can be summarized as follows: (a) phenyl or other
cyclic groups are not tolerated in that position;27 (b) the
terminal hydroxyl group can be replaced with
fluoro,24,26,30,33 chloro,30 vinyl,30 or a methyl25 group
with an increase in affinity while the O-methyl analogue
had diminished affinity;24,25 (c) a three carbon chain is
the optimal head group length for the N-hydroxyalkyl
or the N-alkyl derivatives;4,25 and (d) generally, tertiary
amides lack affinity for CB1 with the only known
exception of N,N-di-hydroxyethyl analogue.27
Chemistry
Anandamide analogues 1–7 (Table 1) were prepared by
reaction of the appropriate amino alcohols with either
(R)-, (S)-, or racemic 2-methyl arachidonic acid chlor-
ides which in turn were obtained from the correspond-
ing acids by treatment with oxalyl chloride in the
presence of dimethylformamide.28,30 Optically pure 2-
methyl-arachidonic acids were obtained by asymmetric
methylation of arachidonic acid which was derivatized
with 4-benzyl-2-oxazolidinone, a chiral auxiliary, first
introduced by Evans36ꢀ40 (Scheme 1, method A). Initi-
ally, arachidonic acid was coupled with lithiated benzyl-
oxazolidinones through its acid chloride. Subsequently,
the resulting N-arachidonyl-4-benzyl-2-oxazolidinones
were transformed into their chiral enolates with sodium
bis(trimethylsilyl)amide at ꢀ78 ꢁC in THF, which in
turn were methylated with a high degree of diasteros-
electivity (d.e. was determined to be between 90 and
94% by GC–MS analysis).
There is strong precedence37,40 that this reaction pro-
ceeds through the Z-sodium-enolate with the benzyl
group blocking one face of the nearly planar ring sys-
tem. Therefore, alkylation occurs predominately anti to
the benzyl group.
Purification by silica gel column chromatography led to
optically pure N-arachidonyl-4-benzyl-2-oxazolidi-
nones. As with earlier reports36ꢀ40 involving similar
asymmetric synthesis, the absolute configuration at the
2-position was assigned by analogy, based on the
expected direction of the methylation (anti to the benzyl
group). Unlike the earlier work by Ryan et al.,33 in
which the 2-methyl arachidonic acids were obtained by
chiral derivatization-resolution, the chiroselective
method described here allows us to assign the absolute
configuration of the resulting chiral methyl ananda-
mides, by analogy. Finally, the oxazolidinone chiral
auxiliary was removed hydrolytically (LiOH/H2O/THF)
at 0 ꢁC. Absence of racemization in this reaction was
demonstrated by recoupling the resulting methylated
arachidonic acid with optically pure 4-benzyl-2-oxazoli-
dinone, where a single reaction product was obtained.
Our initial efforts to develop metabolically stable ana-
ndamide analogues showed that this can be accom-
plished by sterically hindering the scissile amide bond
through the introduction of a-methyl groups. Of the
ensuing analogues, (R)-(+)-arachidonyl-10-hydroxy-20-
propylamide, or (R)-methanandamide (AM356) was the
most successful exhibiting a 4-fold higher affinity for
CB1 than anandamide, remarkable hydrolytic stabi-
lity28,35 and high in vivo potency.10,28
As an extension of our work that led to the discovery of
(R)-methanandamide we undertook a more compre-
hensive attempt to study the enantio- and diastereo-
selectivity of a family of 2-methylated anandamides
with regard to their affinities for the cannabinoid recep-
tors and their enzymatic stability. In view of the fact
that 2-methylation of anandamide is well tolerated in
terms of its CB1 affinity,26,33,34 we felt that we had good
chance of being successful. The results have allowed us
to propose a partial pharmacophoric model for the CB1
receptor with respect to anandamide binding. The work
was also motivated by earlier observations according to
which introduction of methyl substitution in
The effectiveness of the above method was validated by
resynthesizing diastereomers 4–7 via an alternative
route (Scheme 1, method B) involving the coupling of
racemic 2-methyl-arachidonic acid with enantiomeri-
cally pure 2-amino-1-propanol. The resulting diaster-
eomeric mixture was easily resolvable by silica gel
column chromatography. In this resolution, the higher

A. Goutopoulos et al. / Bioorg. Med. Chem. 9 (2001) 1673–1684
1675
Table 1. Affinities (Ki) of anandamide analogues for the cannabinoid receptors CB1 and CB2a
Affinity for the Cannabinoid receptors (Ki in nM)
CB1
CB2
Compound
with PMSF
without PMSF
1
2
54.1ꢂ5.2
457ꢂ32
4905ꢂ621
4259ꢂ524
35.3ꢂ4.3
45.8ꢂ6.2
3941ꢂ311
1865ꢂ255
3
4
3062ꢂ548
1952ꢂ212
32.5ꢂ5.1*53ꢂ15**
7.42ꢂ0.86
137ꢂ20**
40.4ꢂ3.9
5
6
7
8
185ꢂ12
389ꢂ72
233ꢂ69
211ꢂ29
1132ꢂ122
211ꢂ36
4876ꢂ381
2876ꢂ381
4695ꢂ646
>10ꢃ103
72.2ꢂ6.3
150ꢂ14
25.5ꢂ2.8* 47ꢂ3**
41ꢂ3**
257ꢂ28
9
220ꢂ21
>10ꢃ103
>10ꢃ103
>10ꢃ103
421ꢂ38
31.1ꢂ1.0*
876ꢂ92
10
11
12
13
1661ꢂ183
1794ꢂ194
273ꢂ50
191.4ꢂ24.5*
1288ꢂ133
153.9ꢂ30.0*
107ꢂ18
46.6ꢂ2.2*
20.6ꢂ1.6
28.3ꢂ3
868ꢂ132
(continued on next page)

1676
A. Goutopoulos et al. / Bioorg. Med. Chem. 9 (2001) 1673–1684
Table 1 (continued)
Affinity for the Cannabinoid receptors (Ki in nM)
CB1
CB2
Compound
with PMSF
without PMSF
14
15
173ꢂ46
268ꢂ101
8216ꢂ904
1926ꢂ4.5
78.2ꢂ1.6
1180
aCB1 affinities were determined using rat brain membranes and 0.8 nM[ 3H] CP-55,940 as the tritiated ligand essentially as previously described. By
running the assay both with and without PMSF the metabolic stability of the analogues is assessed. Mouse spleen was used as source of CB2
receptor. Data was analyzed using non-linear regression analysis. Ki values were obtained from a minimum of two independent experiments run in
duplicate. The 95% confidence limits are in parentheses. Ki values previously reported in the literature are also included: * indicates data from ref 34
and ** from ref 26. (R)-methanandamide (13), (S)-methanandamide (14) and anandamide (15) have been studied in previous published papers27,28
and are included in this table for the purpose of comparison.
Scheme 1. Synthesis of chiral methylated anandamide analogues.

A. Goutopoulos et al. / Bioorg. Med. Chem. 9 (2001) 1673–1684
1677
Rf value diastereomer was the R,R analogue (4) and the
more polar one was the S,R (5) as revealed by compar-
ison with the corresponding products obtained through
method A. Diastereomers 6 and 7 were also similarly
synthesized by coupling the above acid with (S)-(+)-2-
amino-1-propanol. All four diastereomers obtained
through this method were chromatographically, spec-
troscopically and optically identical with those obtained
through Method A. In Method B, the head group of
(R)-(and (S)-methanandamide serves in itself as a chiral
auxiliary in the resolution of a-methylated arachidonic
acids. Unfortunately, this method is very limited, sui-
table only for the synthesis of the 2,10 dimethyl analo-
gues. In the b-methylated headgroup series, obtained by
reaction of (ꢂ)-2-methyl-arachidonic acid with 1-
amino-2-propanol (data not shown), the stereochemical
communication between chiral methyl groups is lost and
the resulting diastereomeric mixture is not easily resol-
vable.
random search method while constraining the C5–C20
part of the molecule (torsion force: 100 kcal radꢀ2).
Within the searched low energy conformations, the
conformer shown in Figure 2 was selected as a preferred
low energy conformation for 4. Similarly, the corre-
sponding conformers for the other compounds were
chosen as low energy conformers. All other compounds
were built on the basis of 4 with corresponding
modifications. All conformers were energy minimized
again until the energy differences between successive
iterations were less than 0.001 kcal molꢀ1
A
ꢀ1. C4 and
the amide oxygen and nitrogen atoms were used as three
superposition points for the RMS alignment of the
analogues. Based on these superpositions, receptor
volume maps46 were generated using the Sybyl multiple-
contour tool. The methods used are described in more
detail elsewhere.47
Discussion
The 2,2-dimethyl analogues (8–12) were synthesized, as
previously,26,34 by coupling of the appropriate amino-
alcohols with 2,2-dimethyl arachidonic acid, which in
turn was synthesized by repeating the first step of
method 2, followed by hydrolysis.
Affinities for CB1 and CB2
In an earlier study involving the synthesis of chiral 10
and 20 monomethyl head group substituted ananda-
mides, we demonstrated that the (R)-(+)-arachidonyl-
10-hydroxy-20-propylamide [(R)-methanandamide] exhi-
bits the highest CB1 affinity and remarkable enzymatic
stability.28 Furthermore, other reports have shown that
certain enantiomeric or diastereomeric mixtures of 2-
methylated anandamides possess higher CB1 affinities
than anandamide.26,34 Two enantiomerically pure fluoro
analogues of 2-methylarachidonylethylamides33 were
prepared and found to exhibit high affinity but no
enantioselectivity in their binding with CB1. We now
report the synthesis and evaluation of a series of non-
fluorine containing anandamide analogues in which
chiral methyl groups are introduced systematically at
the head group as well as at the 2-position of the ara-
chidonyl moiety. No effort was made to introduce alkyl
groups larger than methyl at any of these positions since
such substitutions have been shown to significantly
lower the CB1 affinity.26,34
CB1 and CB2 receptor binding studies41,42
Prior to assaying for CB1 receptor affinity, the synap-
tosomal membranes were treated with 50 mM
PMSF,27,30 a general serine protease inhibitor, to pro-
tect the analogues from the hydrolytic activity of ana-
ndamide amidase. CB1 receptor binding assays were
also conducted in the absence of PMSF to assess the
metabolic stability of the analogues by comparing the
Ki’s obtained in the presence and absence of PMSF.27,30
For CB2 receptor affinity, mouse spleen membranes
were used without PMSF, since anandamide amidase is
not present in this preparation in significant amounts.43
The Ki values of the novel analogues for the CB1 and
CB2 are presented in Table 1, where (R)-methananda-
mide (13), (S)-methanandamide (14) and anandamide
(15) are included for the purpose of comparison.
Our results show that a chiral methyl group at the 2-
position of arachidonic acid leads to anandamide ana-
logues (1–3) with moderately improved affinities for
CB1 when compared to anandamide (15) but without
significant stereoselectivity. Lack of enantioselectivity at
that position was also demonstrated for the fluorinated
analogues reported by Ryan et al.33 The (S)-isomer (2)
exhibits slightly higher affinity, more than twice that of
the parent anandamide. However, introduction of a
second chiral methyl group at the ethanolamido head
group is accompanied with a major degree of enantio-
and diastereoselectivity with regard to CB1 binding.
Thus, the four analogues (4–7) encompassing two
enantiomeric pairs exhibit an approximately 50-fold
range of affinities for CB1. Of these, the 2(R),10(R)-
dimethyl analogue (4) had the highest affinity (Ki: 7.42
nM), 52-fold higher than its enantiomer 2(S),10(S)-
dimethyl anandamide (6) and an 11-fold improvement
over the parent anandamide (15). The other pair of
Computer molecular modeling
Computational studies on select analogues (1, 2, 4–7,
13–15) were carried out using the TRIPOS Sybyl 6.5
molecular modeling package with the Tripos empirical
forcefield,44 on an SGI Octane R10,000 workstation.
Among these analogues, 1, 2, 4, 13, and 15 were of high
affinity for CB1 (‘active’) whereas 5–7, and 14 were of
low affinity (‘inactive’). The most active analogue, 4,
was used as the starting point. The initial structure was
built using standard bond lengths and angles from the
Sybyl molecular modeling package. Atomic charges
were calculated for all molecules by the semi-empirical
method (MOPAC/AM1),45 and potential energies were
minimized by using the Tripos force field in two steps,
first using the steepest descent method for 100 iterations
and then the conjugate gradient method until the max-
imum derivative was less than 0.001 kcal molꢀ1 Aꢀ1
The conformation of 4 was determined by the Sybyl
.

1678
A. Goutopoulos et al. / Bioorg. Med. Chem. 9 (2001) 1673–1684
enantiomers, 2(S),10(R)-and 2(R),10(S)-dimethyl (5 and
7) had very similar affinities for CB1 with Ki values in
between those of the other enantiomeric pair.
head group produces striking effects on CB1 affinities
and leads to chiral analogues covering a 175-fold range
of Ki values. The causes for the above-mentioned steric
effects can be 2-fold. First, the presence of methyl
groups can lead to variations in the conformational
properties of the anandamide head groups which can, in
turn, play a role in the ligand’s interactions with its site
of action. Thus, if the preferred conformation of the
ligand resembles the pharmacophoric conformation of
anandamide at CB1, its ability to engage in a productive
interaction with receptor is enhanced. Conversely, a
conformation differing from the pharmacophoric one
may lead to reduced affinity. The second reason for the
broad spectrum of affinities of methylated anandamides
for CB1 may be related to whether individual methyl
groups can interact favorably with the receptor site by
‘fitting’ into a putative ‘hydrophobic pocket’, thus
resulting in enhanced affinity. Alternatively, methyl
substitution may be responsible for unfavorable inter-
actions with the binding site due to steric constraints
thus reducing the ligand’s affinity for the receptor.
Therefore, the introduction of two or three methyl sub-
stituents in anandamide may lead to a combination of
favorable and unfavorable steric interactions at the CB1
site, depending on the positions of individual methyl
groups and the absolute configurations of the methyl
substituted carbons.
As discussed in more detail in the following section, the
low level of enantioselectivity observed with the 2-
methyl analogues suggests that the steric requirements
for that position are not as strict as for the 10 position.
Although it is well known26,27,34 that alkyl groups larger
than methyl, at either of these positions, are detrimental
to CB1 affinity, it was shown26,34 that a 2-gem-dimethyl
group can be well tolerated. To further substantiate and
validate our pharmacophoric model, we synthesized the
previously reported 2,2-dimethyl analogue26,34 (8) as
well as its trimethyl congeners34 (9–12) in which a third
methyl group is introduced at the ethanolamido head
group. The CB1 binding data reported here are in gen-
eral agreement with those from earlier reports26,34
although there are some differences in the individual Ki
values. The results show that geminal dimethyl sub-
stitution alpha to the carbonyl group maintains essen-
tially the same affinity for CB1, in support of our
pharmacophoric model. However, introduction of a
third methyl group at the ethanolamine moiety leads to
reduction in affinity for CB1. Of the four trimethyl
compounds, the 2,2,20-(S)-trimethyl analogue (12) had
the highest affinity for CB1, slightly less than that of
anandamide, followed by the 2,2,10(R)-trimethyl (9)
with a Ki almost 3 times that of anandamide. The other
two analogues (10 and 11) exhibited considerably lower
affinities as Sheskin et al.34 have also reported. The
results from this series reveal that the SAR due to
methyl substitution at different positions of anandamide
is not always additive. However, some general trends
can be observed. Thus, in the trimethyl anandamide
series, 10-methyl substitution at the head group leads to
a considerable reduction in affinity when compared to
the 10-monomethyl and the 2,10-dimethyl series where
the presence of the 10-(R)-methyl group leads to sub-
stantial improvement in affinity for CB1. Similarly, 20-
methyl substitution in the trimethyl series also leads to
reduced affinity for CB1 unlike the monomethyl series
where the 20-(R)-methyl analogue has higher affinity
than anandamide. However, the effect of 20-methyl sub-
stitution in the 2,2,10- and 2,2,20-trimethyl analogues is
less striking than that due to 10-methyl substitution.
Examination of the binding data for the methylated
anandamides described here reveals that the variation in
affinities for CB1 cannot be solely attributed to the
conformational properties of the ligands. The wide dif-
ferences in affinities between the enantiomeric pairs of
ligands (e.g., 4 and 6) which are expected to have iden-
tical rotamer distributions, are congruent with such a
conclusion. Ligand conformational analyses which were
simplified by restricting our computation to the etha-
nolamido head group and the first five carbon atoms of
the arachidonyl tail, further support this interpretation.
In this study, several conformational minima were
identified for each compound, all within 1 Kcal/mol of
potential energy difference between them. Of these, we
chose the fully extended conformers for superposition
(Fig. 1). Based on the known anandamide SAR, we
All the analogues in this series exhibit a high degree of
selectivity for CB1 compared to CB2 with selectivity
ratios ranging from 4- to at least 260-fold. This finding
is in agreement with previous SAR studies27,31 in which
anandamide and its analogues were found to have low
affinities for CB2. The above results may argue that
anandamide is not the preferred endogenous ligand for
CB2. However, the characterization of a CB2 selective
endogenous ligand remains elusive, notwithstanding a
report attributing this role to palmitylethanolamide.48
Figure 1. CB1 receptor essential volume map as calculated by sub-
tracting the sum of molecular volumes of the aligned ‘inactive’ analo-
gues (5, 6, 7, and 14), from the sum of molecular volumes of the
aligned ‘active’ ones (1, 2, 4, 13, and 15). By extrapolation, the yellow
contours correspond to areas occupied by structural features of the
receptor.
Molecular modeling of CB1 binding stereochemical
requirements
It is clear from the above results that the stereoselective
introduction of methyl groups at or near the anandamide

A. Goutopoulos et al. / Bioorg. Med. Chem. 9 (2001) 1673–1684
1679
could assume that all the analogues in this study interact
with the CB1 receptor following the same binding motif,
which at least in the head group region, is dictated by the
anchoring of three principal pharmacophoric groups: the
terminal hydroxyl, the amide proton and the carbonyl
oxygen. Based on this model, the main difference
between the receptor-bound analogues is the direction
in which the methyl groups are projecting out of the
anandamide backbone. Therefore, the differences in
affinities can be attributed, at least in part, to the rela-
tive topologies of the methyl groups of the anandamide
analogues while interacting with the CB1 receptor site.
(receptor-excluded volume map). The yellow colored
clouds represent the volumes occupied by the methyl
groups of the most active analogue, 4. These methyl
groups are presumably interacting with putative small
hydrophobic pockets in the CB1 receptor site whose
outer circumscription are partially outlined by the yel-
low contours of Figure 1. These pockets can be accessed
simultaneously by analogue 4, which bears both methyl
groups in the appropriate relative configurations leading
to an optimal interaction with the receptor site. In con-
trast, the prefered conformers for analogues 5–7 place
their methyl group(s) beyond these pharmacophoric
pockets resulting in negative steric interactions with
CB1 and subsequently lower affinities. Based on this
pictorial model, the pocket that accommodates methyl
group(s) at the 2 position is larger than that of the 10
position. This postulate explains the low CB1 enantio-
selectivity for the 2-monomethyl analogues (1–2). The 2-
dimethyl congeners (8–12) were synthesized in order to
test and validate our model. Indeed, in support of our
hypothesis, we found the 2-gem-dimethyl group to be
well tolerated (8). However, the simultaneous presence
of a third methyl group in the ethanolamido moiety
(analogues 9–12) leads to unfavorable steric interactions
and lower CB1 affinities.
To gain more insight on the interactions of methylated
chiral anandamides with CB1, we employed molecular
modeling methods which allowed us to explore the
regio- and stereochemical requirements of methyl group
substituents for binding with CB1. The receptor-exclu-
ded and receptor-essential volume concepts46,47 were
used to develop a visual model of the receptor site
region that accommodates the head group moiety of
anandamide.
According to this approach, it is assumed that a family
of active ligands adopt similar shapes and occupy simi-
lar volumes at the receptor site, while inactive ligands
occupy different spaces which may negatively affect
their binding to the receptor. Our newly synthesized,
chiral 2-monomethyl analogues along with R-, S-
methanandamide and anandamide were used for the
construction of the model. More specifically, analogues
1, 2, 4, 13, and 15 of high affinity for CB1 (‘active’) and
analogues 5, 6, 7, and 14 of low affinity (‘inactive’) were
superimposed using the C5 and the amide nitrogen and
oxygen as superimposition points. The sum of the van
der Waals volumes of the ‘inactive’ analogues deviating
from the sum of volumes of the ‘active’ ones is illu-
strated by the yellow contours in Figure 1. According to
the ‘receptor essential volume’ approach, these yellow
contours can be extrapolated as areas occupied by
structural features of the receptor site. Occupancy of
these areas by ligand molecules is hindered by negative
steric interactions leading to lower binding affinities.
The sum of the van der Waals volumes of the ‘active’
analogues representing the free space within the recep-
tor region that accommodates the anandamide head
group is depicted in Figure 2 with the green cloud
Metabolic stability
In earlier publications,28,35 we had reported on the
remarkable stability of (R)-methanandamide (13) with
regard to enzymatic activity by anandamide amidase.
The introduction of a methyl group at the 10-position of
the head group of anandamide is governed by con-
siderable stereoselectivity since the S-enantiomer is
more susceptible to amidase activity than its enantio-
mer. Unlike previous reports,26,33 our present results
indicate that this metabolic stability imparted by the
presence of a methyl group alpha to the amide nitrogen
is only partially duplicated by the introduction of a
methyl group alpha to the amido carbonyl. Thus, the
respective two a-methyl enantiomers 1 and 2 are less
metabolically stable than (R)-methanandamide with the
R-isomer exhibiting the greater stability. These findings
were confirmed (unpublished results) by examining
these analogues as substrates of the anandamide ami-
dase on an HPLC assay developed in our laboratory.49
The combined presence of the above two chiral 10,2
methyl groups does not increase to any significant
extent the metabolic stability of anandamide although it
leads to some interesting observations. For example, the
2(R)- and 10(R)-methyl analogues are the most metabo-
lically stable chiral isomers within each enantiomeric
pair. This would lead to the prediction that the dimethyl
anandamide analogues incorporating the above two
chiral centers [i.e., 2R, 1R (4)] must also possess the
highest degree of metabolic stability. However, contrary
to our expectation, the results of this study find analo-
gues 2S, 10R (5) and 2R, 1S (7) to be the most enzyma-
tically stable dimethyl anandamides. This indicates that
the substrate-enzyme interactions cannot be always
predicted based on ‘additive SARs’ but involve more
complex contributions to the binding and/or catalysis at
the enzyme’s active site.
Figure 2. CB1 receptor excluded volume map as calculated by adding
the molecular volumes of the aligned ‘active’ analogues (1, 2, 4, 13,
and 15). The yellow colored clouds represent the volumes occupied by
the methyl groups of the most active analogue, 4.

1680
A. Goutopoulos et al. / Bioorg. Med. Chem. 9 (2001) 1673–1684
Finally, it appears that the incorporation of gem-dime-
thyl substituents at the 2-position of the arachidonic
acid moiety leads only to partial metabolic stability for
the respective anandamides. Our results clearly indicate
that metabolic stability with regard to anandamide
amidase is optimally accomplished by a stereospecific R-
methyl substitution at the carbon a to the amide nitro-
gen and is superior to substitution at the a-position to
the carbonyl group. This may appear to be a somewhat
surprising steric effect in view of the fact that the 10-
methyl group is two bonds removed from the carbonyl
site of catalytic action when compared to the 2-methyl
group which is only one bond away from that site. A
possible explanation for this observation is that a-ami-
domethyl protons are separated by five bonds from the
carbonyl oxygen and are then subject to the Newman’s
‘rule of six’50 which describes steric crowding during the
hydrolysis of carbonyl esters, while the a-arachidonyl
methyl hydrogens are only four bonds away from the
carbonyl oxygen, leading to less effective steric crowding
at the hydrolysable carbonyl group. According to the
‘rule of six’, the greater the number of atoms at the six
position, the greater the steric effect. This rule explains,
for example, why the ratio of the rates of hydrolysis for
ethyl acetate and ethyl isobutyrate is 8.5 whereas the
ratio of the rates of hydrolysis for methyl benzoates and
isopropyl benzoate is 20.50
chemical shifts are reported in ppm. Specific rotations
were determined using a Perkin-Elmer 241 polarimeter
within either a 1.00 or 0.01 dm cell. Satisfactory ele-
mental analyses was obtained for all the analogues syn-
thesized and are within ꢂ0.4% except for compounds 1
(0.42%), 5 (0.66%), and 8 (0.43%). The optically pure
amino alcohols were obtained from Aldrich Chemical
Company (Milwaukee, WI, USA). Arachidonic acid
and methyl ester were purchased from SIGMA Chemi-
cal Company and Nu-Chek-Prep. Inc.
(R)-N-arachidonyl-4-benzyl-2-oxazolidinone (16). To a
stirring solution of 500 mg (1.64 mmol) of arachidonic
acid and 274.5 mL (1.97 mmol) of triethylamine in 9 mL
of anhydrous THF at ꢀ78 ꢁC, 285.3 mL (1.97 mmol) of
pivaloyl chloride was added. A white suspension was
formed and stirring was continued for 10 min at ꢀ78 ꢁC
and 30 min at 0 ꢁC. The mixture was recooled to ꢀ78 ꢁC
and a ꢀ78 ꢁC solution of lithiated (R)-(+)-4-benzyl-
oxazolidinone was cannulated. The metallated oxazoli-
dinone was prepared by addition of 1.23 mL (1.97
mmol, 1.6 Min THF) of n-BuLi in a solution of 348.5
mg (1.97 mmol) of (R)-(+)-4-benzyl-oxazolidinone in 7
mL of THF at ꢀ78 ꢁC. The reaction mixture was stirred
for 3 h while raising slowly the temperature to 0 ꢁC and
then for 30 min at 0 ꢁC. The reaction was then quenched
by the addition of 10 mL of saturated aqueous solution
of ammonium chloride and extracted from ether (3ꢃ30
mL). The combined organic phases were washed with
water, dried over anhydrous magnesium sulfate and
stripped of volatiles. Purification with silica gel column
chromatography (eluent CH2Cl2/petroleum ether) yiel-
ded 692.1 mg (88%) of the title compound as a colorless
oil: Rf 0.51 (CH2Cl2); 1H NMR (200 MHz, CDCl3)
d?7.38–7.18 (m, 5H, aromatic-Hs), 5.49–5.29 (m, 8H,
vinylic-Hs), 4.72–4.60 (m, 1H, C40-H), 4.24–4.13 (m,
2H, C50-Hs), 3.30 (dd, 1H, J=3.2 Hz, 13.3 Hz, CHH-
Ph), 3.05–2.92 (m, 2H, C2-H), 2.89–2.69 (m, 7H, double
allylic Hs and CHH-Ph), 2.20–2.13 (m, 2H, C4-Hs),
2.10–2.00 (m, 2H, C16-Hs), 1.80 (quintet, 2H, J=7.3
Hz, C3-Hs) 1.42–1.23 (m, 6H, C17,18,19-Hs), 0.88 (t,
3H, J=6.8 Hz, C20-Hs); [a]28 ꢀ37.7ꢁ (neat). Anal. calcd
for C30H41O3N: C, 77.71; H, 8.91; N, 3.02. Found: C,
77.36; H, 9.19; N, 2.91.
Conclusions
Asymmetric methylation at the 2-position of ananda-
mide leads to analogues with only moderately improved
affinity for the CB1 receptor regardless of absolute
configuration and provides only modest protection
against anandamide amidase. However, when a second
methyl group is present at the ethanolamido head group
a high degree of enantio- and diastereoselectivity is
observed. The diastereomer with the highest affinity (4)
incorporates two methyl groups with an optimal abso-
lute configuration (R,R) suggesting the presence of two
respective small hydrophobic pockets at the receptor
site. A gem-dimethyl group at the 2-position is well tol-
erated in one of these pockets (the larger one) but
results in negative steric interactions when a third
methyl is introduced at the head group. The above
results support a CB1 receptor site model which incor-
porates two pockets capable of accommodating methyl
groups in the 10- and 2-positions of the molecule with
the 2-methyl pocket being significantly larger. Regard-
ing stability towards anandamide amidase imparted by
methyl group substitution, our present results confirm
that (R)-10-methylation leads to the most stable analo-
gues.
(R)-N-((R)-2-Methyl-arachidonyl)-benzyl-2-oxazolidi-
none (17). To a ꢀ78 ꢁC solution of 1.150 mL (1.15
mmol, 1.1 equiv, 1.0 Min THF) of sodium hexam-
ethyldisilamide, a 0 ꢁC solution of 500 mg (1.05 mmol,
1.0 equiv) of (R)-N-arachidonyl-4-benzyl-2-oxazolidi-
none (16) in 5 mL of anhydrous THF was added drop-
wise, via a cannula. The solution was stirred at ꢀ78 ꢁC
for 30 min and then 326 mL (5.24 mmol, 5 equiv) of
iodomethane was added and the reaction mixture was
stirred for 4 h at ꢀ78 ꢁC. The reaction was quenched by
the addition of 8 mL of saturated aqueous solution of
ammonium chloride. The product was extracted from
diethyl ether (3ꢃ30 mL). The combined ether extracts
were washed with water and dried over anhyd MgSO4.
Purification with silica gel column chromatography
(eluent CH2Cl2/petroleum ether) yielded 370.6 mg
(74%) of the title compound as a colorless oil: Rf 0.72
Experimental
Chemistry
1H NMR spectra were recorded either on a Bruker
DMX 500 MHz or a Bruker WP-200SY 200 MHz spec-
trometer using TMS as the internal standard. All
1
(CH2Cl2); H NMR (200 MHz, CDCl3) d 7.38–7.20 (m,

A. Goutopoulos et al. / Bioorg. Med. Chem. 9 (2001) 1673–1684
1681
5H, aromatic-Hs), 5.44–5.35 (m, 8H, vinylic Hs), 4.70–
4.63 (m, 1H, C40-H), 4.19–4.14 (m, 2H, C50-Hs), 3.72
(sext., 1H, J=6.9 Hz, C2-Hs), 3.26 (dd, 1H, J=3.2 Hz,
13.3 Hz, CHH-Ph), 2.90–2.71 (m, 7H, double allylic-Hs
and CHH-Ph), 2.16–2.00 (m, 4H, C4,16-Hs), 2.00–1.77
(m, 1H, CHHCH(CH3)), 1.50–140 (m, 1H,
CHHCH(CH3)), 1.40–1.23 (m, 9H, C17,18,19-Hs,
CH2CH(CH3)), 0.88 (t, 3H, J=6.8 Hz, C20-Hs); [a]28
ꢀ54.3ꢁ (neat). Anal. calcd for C31H43O3N: C, 77.94; H,
9.07; N, 2.93. Found: C, 78.28; H, 9.01; N, 2.82.
coupling (R)-2-methyl-arachidonic acid (18) with (S)-
(ꢀ)-2-amino-1-propanol. (yield: 79%) colorless oil: Rf
0.25 (40% CH2Cl2 in ethyl acetate); 1H NM
(500 MHz, CDCl3) d?5.57 (s, br, 1H, NH), 5.40–5.30 (m,
8H, vinylic-Hs), 4.10–4.04 (m, 1H, C10-H), 3.67–3.64
(m, 1H, CHHOH), 3.54–3.48 (m, 1H, CHHOH), 2.84–
2.79 (m, 6H, double vinylic Hs), 2.23–2.18 (m, 1H, C2-
H), 2.09–2.03 (m, 4H, C4,16-Hs), 1.76–1.70 (m, 1H,
CHHCH(CH3)), 1.48–1.42 (m, 1H, CHHCH(CH3)),
1.34–1.36 (m, 6H, C17, 18, 19-Hs), 1.18–1.14 (m, 6H,
CH(CH3), NH–CH(CH3)), 0.88, (t, 3H, J=6.8 Hz, C20-
Hs); [a]28 +16.2ꢁ (0.2 g/mL CH2Cl2). Anal. calcd for
C24H41O2N: C, 76.75; H, 11.00; N, 3.73. Found: C,
76.50; H, 10.92; N, 3.59.
R
(R)-2-Methyl-arachidonic acid (18). To a stirred solution
of 300 mg (0.63 mmol ) of (R)-N-((R)-2-methyl-arachi-
donyl)-benzyl-2-oxazolidinone (17) in 5 mL of THF at
0 ꢁC, 3 mL of 1.0 Maqueous solution of lithium
hydroxide was added. The reaction mixture was stirred
for 6 h at 0 ꢁC and then it was neutralized with satu-
rated aqueous solution of ammonium chloride. The
product was extracted from diethyl ether (3ꢃ30 mL),
washed with water and dried over anhyd MgSO4. Pur-
ification with silica gel column chromatography (eluent:
CH2Cl2/ethyl acetate) afforded 177.5 mg (89%) of the
title compound as a colorless oil:Rf 0.76 (10% ethyl
acetate in CH2Cl2); 1H NMR (500 MHz, CDCl3)
d?5.40–5.31 (m, 8H, vinylic Hs), 2.84–2.81 (m, 6H,
double allylic Hs), 2.49 (sext., 1H, J=6.9 Hz, C2-H),
2.14–2.00 (m, 4H, C4,16-Hs), 1.80–1.70 (m, 1H,
CHHCH(CH3), 1.56–1.45 (m, 1H, CHHCH(CH3)),
1.38–1.29 (m, 6H, C17,18,19-H), 1.20 (d, 3H, J=7.2 Hz,
CHHCH(CH3)), 0.88 (t, 3H, J=6.8 Hz, C20-H); [a]28
ꢀ20.2ꢁ (0.3 g/mL CH2Cl2). Anal. calcd for C21H34O2:
C, 79.19; H, 10.76. Found: C, 79.62; H, 10.90.
(R)-N-(2-Hydroxyethyl)-2-methyl-arachidonamide
(1).
Compound 1 was synthesized as 4 by coupling (R)-2-
methyl-arachidonic acid (18) with ethanolamine; (yield:
85%) colorless oil: Rf 0.47 (60% ethyl acetate in
CH2Cl2); 1H NMR (500 MHz, CDCl3) d 5.94 (s, br, 1H,
NH), 5.43–5.31 (m, 8H, vinylic-Hs), 3.73 (t, 2H, J=4.8
Hz, C10-Hs), 3.43 (quart., 2H, J=5.1 Hz, C20-Hs), 2.83–
2.80 (m, 6H, double allylic-Hs), 2.24 (sext., 1H, J=6.9
Hz, C2-Hs), 2.10–2.03 (m, 4H, C4,16-Hs), 1.77–1.72 (m,
1H,
CHHCH(CH3)),
1.50–1.45
(m,
1H,
CHHCH(CH3)), 1.37–1.26 (m, 6H, C17, 18, 19-Hs),
1.16 (d, 3H, J=6.9 Hz, CH(CH3)), 0.89 (t, 3H, J=6.0
Hz, C20-Hs); [a]28 ꢀ9.4ꢁ (0.1 g/mL CH2Cl2). Anal.
calcd for C23H39O2N: C, 76.40; H, 10.87; N, 3.87.
Found: C, 76.82; H, 10.98; N, 3.66.
(S)-N-Arachidonyl-4-benzyl-2-oxazolidinone. (S)-N-Ara-
chidonyl-4-benzyl-2-oxazolidinone was synthesized as
16 by coupling arachidonic acid (S)-(ꢀ)-4-benzyl-oxa-
zolidinone; (yield: 82%) colorless oil: Rf 0.51 (CH2Cl2);
1H NMR (200 MHz, CDCl3) d 7.38–7.18 (m, 5H, aro-
matic-Hs), 5.49–5.29 (m, 8H, vinylic-Hs), 4.72–4.60 (m,
1H, C40-H), 4.24–4.13 (m, 2H, C50-Hs), 3.30 (dd, 1H,
J=3.2 Hz, 13.3 Hz, CHH-Ph), 3.05–2.92 (m, 2H, C2-
Hs), 2.89–2.69 (m, 7H, double allylic-Hs and CHH-Ph),
2.20–2.13 (m, 2H, C4-Hs), 2.10–2.00 (m, 2H, C16-Hs),
1.80 (quintet, 2H, J=7.3 Hz, C3-Hs) 1.42–1.23 (m, 6H,
C17,18,19-Hs), 0.88 (t, 3H, J=6.8 Hz, C20-Hs); [a]26
+37.9ꢁ (neat).
(R)-N-(1-Methyl-2-hydroxyethyl)-2-(R)-methyl-arachido-
namide (4). To a stirred solution of 50 mg (0.16 mmol)
of (R)-2-methyl-arachidonic acid (18) and 12.4 mL (0.16
mmol) of DMF in 1 mL of dichloromethane at 0 ꢁC,
100 mL (0.20 mmol) of a 2.0 Msolution of oxalyl chlo-
ride in dichloromethane was added dropwise. The reac-
tion mixture was stirred at 0 ꢁC for 20 min. Then, 62.2
mL (0.80 mmol) of (R)-(ꢀ)-2-amino-1-propanol was
added and stirring was continued for 20 min at ambient
temperature. The reaction mixture was diluted with 3
mL of dichloromethane and washed with brine. The
organic phase was separated, dried over anhydrous
magnesium sulfate and stripped of volatiles in vacuo.
Purification by silica gel column chromatography (elu-
ent: CH2Cl2/ethyl acetate) yield 47.4 mg (79%) of 4 as a
(S)-N-((S)-2-Methyl-arachidonyl)-benzyl-2-oxazolidinone.
(S)-N-((S)-2-Methyl-arachidonyl)-benzyl-2-oxazolidinone
was synthesized as 17 by methylating (S)-N-arachido-
nyl-4-benzyl-2-oxazolidinone. Similar diastereoselect-
ivity was achieved as in preparation of 17; (yield: 71%);
colorless oil:Rf 0.72 (CH2Cl2); 1H NMR (200 MHz,
CDCl3) d 7.38–7.20 (m, 5H, aromatic-Hs), 5.44–5.35
(m, 8H, vinylic-Hs), 4.70–4.63 (m, 1H, C40-H), 4.19–
4.14 (m, 2H, C50-Hs), 3.72 (sext, 1H, J=6.9 Hz, C2-Hs),
3.26 (dd, 1H, J=3.2 Hz, 13.3 Hz, CHH-Ph), 2.90–2.71
(m, 7H, double allylic-Hs and CHH-Ph), 2.16–2.00 (m,
4H, C4,16-Hs), 2.00–1.77 (m, 1H, CHHCH(CH3)),
1.50–140 (m, 1H, CHHCH(CH3)), 1.40–1.23 (m, 9H,
C17, 18, 19-Hs, CHHCH(CH3)), 0.88 (t, 3H, J=6.8 Hz,
C20-Hs); [a]26 +54.5ꢁ (neat).
1
colorless oil: Rf 0.35 (40% CH2Cl2 in ethyl acetate); H
NMR (500 MHz, CDCl3) d 5.57 (s, br, 1H, NH), 5.40–
5.30 (m, 8H, vinylic Hs), 4.10–4.04 (m, 1H, C1-H),
3.67–3.64 (m, 1H, CHHOH), 3.54–3.48 (m, 1H,
CHHOH), 2.84–2.79 (m, 6H, double vinylic-Hs), 2.23–
2.18 (m, 1H, C2-H), 2.09–2.03 (m, 4H, C4, 16-Hs),
1.76–1.70 (m, 1H, CHHCH(CH3)), 1.48–1.42 (m, 1H,
CHHCH(CH3)), 1.34–1.36 (m, 6H, C17, 18, 19-Hs),
1.18–1.14 (m, 6H, CH(CH3), NH–CH(CH3)), 0.88, (t,
3H, J=6.8 Hz, C20-Hs); [a]28 ꢀ4.1ꢁ (0.2 g/mL CH2Cl2).
Anal. calcd for C24H41O2N: C, 76.75; H, 11.00; N, 3.73.
Found: C, 77.06; H, 11.24; N, 3.61.
(S)-N-(1-Methyl-2-hydroxyethyl)-2-(R)-methyl-arachido-
namide (7). Compound 7 was synthesized as 4 by
(S)-2-Methyl-arachidonic acid. (S)-2-Methyl-arachidonic
acid was synthesized as 18 by hydrolysis of (S)-N-((R)-

1682
A. Goutopoulos et al. / Bioorg. Med. Chem. 9 (2001) 1673–1684
2-methyl-arachidonyl)-benzyl-2-oxazolidinone; (yield:
86%); colorless oil:Rf 0.76 (10% ethyl acetate in
CH2Cl2); H NMR (500 MHz, CDCl3) d 5.40–5.31 (m,
8H, vinylic-Hs), 2.84–2.81 (m, 6H, double allylic-Hs),
2.49 (sext., 1H, J=6.9 Hz, C2-H), 2.14–2.00 (m, 4H,
C4,16-Hs), 1.80–1.70 (m, 1H, CHHCH(CH3), 1.56–1.45
(m, 1H, CHHCH(CH3)), 1.38–1.29 (m, 6H, C17,18,19-
H), 1.20 (d, 3H, J=7.2 Hz, CHHCH(CH3)), 0.88 (t, 3H,
J=6.8 Hz, C20-Hs); [a]28 +20.1ꢁ (0.3 g/mL CH2Cl2).
mL (0.75 mmol, 1.2 equiv) of a 2.0 Msolution of lithium
diisopropylamide (LDA) in heptane/THF/benzene was
added dropwise. The dark orange-red mixture was stir-
red for 45 min at ꢀ78 ꢁC and then 196 mL of MeI (3.2
mmol, 5 equiv) was added and the reaction mixture was
stirred for 2 h at ꢀ78 ꢁC. The temperature was raised
overnight to ambient while stirring was continued.
Another 5 equiv of MeI was added and 20 min later the
reaction was quenched with the addition of 3 mL aqu-
eous saturated solution of ammonium chloride. The
product was extracted with ether (3ꢃ20 mL), washed
with water and dried over anhyd MgSO4. Purification
with silica gel column chromatography afforded 133.1
mg (64%) of the title compound as a colorless liquid.Rf
0.58 (50% CH2Cl2/petroleum ether); 1H NM
(500 MHz, CDCl3) d 5.42–5.27 (m, 8H, vinylic-Hs), 3.67
(s, 3H, COOCH3), 2.85–2.80 (m, 6H, double vinylic
Hs), 2.47 (sext, 1H, J=7.0 Hz CHCH3), 2.08–2.03 (m,
4H, C4, 16-Hs), 1.79–1.72 (m, 1H, CHHCH(CH3)),
1.49–1.42 (m, 1H, CHHCH(CH3)), 1.37–1.15 (m, 6H,
C17, 18, 19-Hs), 1.16 (d, 3H, J=7.0 Hz, CH(CH3), 0.88
(t, 3H,J=6.7 Hz, C20-Hs).
1
(S)-N-(1-Methyl-2-hydroxyethyl)-2-(S)-methyl-arachido-
namide (6). Compound 6 was synthesized as 4 by cou-
pling (S)-2-methyl-arachidonic acid with (S)-(+)-2-
amino-1-propanol; (yield: 77%); colorless oil:Rf 0.35
(40% CH2Cl2 in ethyl acetate); 1H NMR (500 MHz,
CDCl3) d 5.57 (s, br, 1H, NH), 5.40–5.30 (m,
8H,vinylic-Hs), 4.10–4.04 (m, 1H, C10-H), 3.67–3.64 (m,
1H, CHHOH), 3.54–3.48 (m, 1H, CHHOH), 2.84–2.79
(m, 6H, double vinylic-Hs) 2.23–2.18 (m, 1H, C2-H),
2.09–2.03 (m, 4H, C4,16-Hs), 1.76–1.70 (m, 1H,
CHHCH(CH3)), 1.48–1.42 (m, 1H, CHHCH(CH3)),
1.34–1.36 (m, 6H, C17,18,19-Hs), 1.18–1.14 (m, 6H,
CH(CH3), NH–CH(CH3)), 0.88, (t, 3H,J=6.8 Hz, C20-
Hs); [a]28 +4.2ꢁ (0.2 g/mL CH2Cl2). Anal. calcd for
C24H41O2N: C, 76.75; H, 11.00; N, 3.73. Found: C,
76.56; H, 11.22; N, 3.63.
R
(ꢂ)-2-Methyl-arachidonic acid. (ꢂ)-2-Methyl-arachi-
donic acid methyl ester (100 mg, 0.30 mmol) was dis-
solved in 3 mL of a 15% solution of KOH in methanol/
water and stirred for 24 h at 50–60 ꢁC. The mixture was
then neutralized with 2 N HCl, most of methanol was
evaporated and extracted from ether (3ꢃ20 mL). Pur-
ification with silica gel column chromatography affor-
ded 65.8 mg (69%) of the title compound as a colorless
liquid: Rf 0.76 (10% ethyl acetate in CH2Cl2); 1H NM R
(500 MHz, CDCl3) d 5.40–5.31 (m, 8H, vinylic-Hs),
2.84–2.81 (m, 6H, double allylic-Hs), 2.49 (sext., 1H,
J=6.9 Hz, C2-H), 2.14–2.00 (m, 4H, C4, 16-Hs), 1.80–
1.70 (m, 1H, CHHCH(CH3), 1.56–1.45 (m, 1H,
CHHCH(CH3)), 1.38–1.29 (m, 6H, C17, 18, 19-Hs),
1.20 (d, 3H, J=7.2 Hz, CHHCH(CH3)), 0.88 (t, 3H,
J=6.8 Hz, C20-Hs).
(R)-N-(1-Methyl-2-hydroxyethyl)-2-(S)-methyl-arachido-
namide (5). Compound 5 was synthesized as 4 by cou-
pling (S)-2-methyl-arachidonic acid with (R)-(ꢀ)-2-
amino-1-propanol; (yield: 84%) colorless liquid:Rf 0.25
(40% CH2Cl2 in ethyl acetate); 1H NMR (500 MHz,
CDCl3) d 5.57 (s, br, 1H, NH), 5.40–5.30 (m, 8H,
vinylic-Hs), 4.10–4.04 (m, 1H, C10-H), 3.67–3.64 (m,
1H, CHHOH), 3.54–3.48 (m, 1H, CHHOH), 2.84–2.79
(m, 6H, double vinylic-Hs) 2.23–2.18 (m, 1H, C2-H),
2.09–2.03 (m, 4H, C4,16-Hs), 1.76–1.70 (m, 1H,
CHHCH(CH3)), 1.48–1.42 (m, 1H, CHHCH(CH3)),
1.34–1.36 (m, 6H, C17, 18, 19-Hs), 1.18–1.14 (m, 6H,
CH(CH3), NH-CH(CH3)), 0.88, (t, 3H, J=6.8 Hz, C20-
Hs); [a]28 ꢀ16.3ꢁ (0.2 g/mL CH2Cl2). Anal. calcd for
C24H41O2N: C, 76.75; H, 11.00; N, 3.73. Found: C,
77.41; H, 11.38; N, 3.34.
(ꢂ)-N-(2-Hydroxyethyl)-2-methyl-arachidonamide (3).
Compound 3 was synthesized as 4 by coupling (ꢂ)-2-
methyl-arachidonic acid with ethanolamine; (yield:
86%) colorless oil:Rf 0.47 (40% CH2Cl2 in ethyl ace-
1
(S)-N-(2-Hydroxyethyl)-2-methyl-arachidonamide
(2).
tate); H NMR (500 MHz, CDCl3) d 5.94 (s, br., 1H,
Compound 2 was synthesized as 4 by coupling (S)-2-
methyl-arachidonic acid with ethanolamine; (yield:
89%) colorless oil:Rf 0.47 (40% CH2Cl2 in ethyl ace-
NH), 5.43–5.31 (m, 8H, vinylic Hs), 3.73 (t, 2H, J=4.8
Hz, C1-H), 3.43 (quart., 2H, J=5.1 Hz, C2-Hs), 2.83–
2.80 (m, 6H, double allylic Hs), 2.24 (sext., 1H, J=6.9,
C2-Hs), 2.10–2.03 (m, 4H, C4,16-Hs), 1.77–1.72 (m, 1H,
CHHCH(CH3)), 1.50–1.45 (m, 1H, CHHCH(CH3)),
1.37–1.26 (m, 6H, C17, 18, 19-Hs), 1.16 (d, 3H, J=6.9
Hz, CH(CH3)), 0.89 (t, 3H, J=6.0 Hz, C20-Hs). Anal.
calcd for C24H41O2N: C, 76.75; H, 11.00; N, 3.73.
Found: C, 77.12; H, 11.18; N, 3.64.
1
tate); H NMR (500 MHz, CDCl3) d 5.94 (s, br, 1H,
NH), 5.43–5.31 (m, 8H, vinylic-Hs), 3.73 (t, 2H, J=4.8
Hz, C10-H), 3.43 (quart., 2H, J=5.1 Hz, C20-H), 2.83–
2.80 (m, 6H, double allylic Hs), 2.24 (sext., 1H, J=6.9
Hz, C2-Hs), 2.10–2.03 (m, 4H, C4,16-Hs), 1.77–1.72 (m,
1H,
CHHCH(CH3)),
1.50–1.45
(m,
1H,
CHHCH(CH3)), 1.37–1.26 (m, 6H, C17, 18, 19-Hs),
1.16 (d, 3H, J=6.9 Hz, CH(CH3)), 0.89 (t 3H, J=6.0
Hz, C20-Hs); [a]28 +9.6ꢁ (0.1 g/mL CH2Cl2). Anal.
calcd for C23H39O2N: C, 76.40; H, 10.87; N, 3.87.
Found: C, 76.68; H, 10.79; N, 3.70.
Preparation of 4 and 5 by method B. Compounds 4 and
5 were also synthesized by coupling, as described above,
(ꢂ)-2-methyl-arachidonic acid with (R)-(ꢀ)-2-amino-1-
propanol: To a stirred solution of 100 mg (0.32 mmol)
of (ꢂ)-2-methyl-arachidonic acid and 24.8 mL (0.32
mmol) of DMF in 2 mL of methylene chloride at 0 ꢁC,
200 mL (0.40 mmol) of 2.0 Msolution of oxalyl chloride
in methylene chloride was added dropwise. The reaction
(ꢂ)-2-Methyl-arachidonic acid methyl ester. To a stirred
solution of 200 mg (0.63 mmol, 1.0 equiv) arachidonic
acid methyl ester in 4 mL of anhyd THF, at ꢀ78 ꢁC, 375

A. Goutopoulos et al. / Bioorg. Med. Chem. 9 (2001) 1673–1684
1683
mixture was stirred at 0 ꢁC for 20 min. Then 48.8 mL
(0.80 mmol) of ethanolamine was added and stirring
was continued for 20 min at ambient temperature. The
reaction was stopped and the product was extracted
from diethyl ether (3ꢃ30 mL), dried over anhydrous
MgSO4 and stripped of volatiles. TLC indicated two
products with Rf 0.35 and 0.25 (40% CH2Cl2 in petro-
leum ether), corresponding to the two diastereomers 4
and 5. This diastereomeric mixture was separable by
silica gel column chromatography (eluent: CH2Cl2/ethyl
acetate) to afford 46.8 mg (39%) of 4 and 40.8 mg
(34%) of 5. The isolated compounds were found iden-
tical to the corresponding products from method A.
values. Data from at least two independent experiments
performed in duplicate were used to calculate IC50
values, which were converted to Ki values using the
assumptions of Cheng and Prusoff.52
For CB2 receptor binding studies, membranes were
prepared from frozen mouse spleen essentially accord-
ing to the procedure of Dodd et al.51 as described else-
where.27,30 Silanized centrifuge tubes were used
throughout to minimize receptor loss due to adsorption.
The CB2 binding assay was conducted in the same
manner as for CB1.
Preparation of 6 and 7by method B. Following the
above procedure, analogues 6 and 7 were also synthe-
sized by coupling (ꢂ)-2-methyl-arachidonic acid with
(S)-(+)-2-amino-1-propanol to afford, after resolution
by column chromatography 43.2 mg (36%) of 6 and
39.6 mg (33%) of 7. The isolated compounds were
found identical to the corresponding products from
method A.
Acknowledgements
This work was supported by grants DA-3801, DA-152,
DA-9158, and DA-7215 from the National Institute on
Drug Abuse. We thank Dr. L. S. Melvin from Pfizer,
Inc. for providing us with CP-55,490 and Packard
Instruments, Inc. for providing the Filtermate 196 and
Top-Count. A.G. thanks Boehringer Ingelheim Phar-
maceuticals Inc. for a Fellowship.
Preparation of compounds 8–12. Compounds 8–12 were
prepared as in ref 33.
References and Notes
Radioligand binding assay. For CB1, rat forebrain
membranes were prepared according to the procedure
of Dodd et al.51 The binding of the novel anandamide
analogues to the cannabinoid receptor was assessed as
previously described41,42 except that the membranes
were treated with PMSF. Membranes, previously frozen
at ꢀ80 ꢁC, were thawed on ice. To the stirred suspension
were added three volumes of 25 mMTris–HCl Buffer, 5
mMMgCl 2, and 1 mMEDTA, pH 7.4 (TME) contain-
ing 150 mMPMSF (made fresh in isopropanol as a 100
mMstock). The suspension was incubated at 4 ꢁC and
after 15 min a second addition of PMSF stock brought
the concentration to 300 mMof PMSF then incubated
for another 15 min. At the end of the second 15 min
incubation, the membranes were pelleted and washed
three times with TME to remove unreacted PMSF. The
treated membranes were subsequently used in the bind-
ing assay described below. Approximately 30 mg of
PMSF-treated membranes were incubated in silanized
96-well microtiter plate with TME containing 0.1%
essentially fatty acid free bovine serum albumin (BSA),
1. Devane, W. A.; Hanus, L.; Breuer, A.; Pertwee, R. G.;
Stevenson, L. A.; Griffin, G.; Gibson, D.; Mandelbaum, A.;
Etinger, A.; Mechoulam, R. Science 1992, 258, 1946.
2. Mackie, K.; Devane, W. A.; Hille, B. Mol. Pharmacol.
1993, 44, 498.
3. Vogel, Z.; Barg, J.; Levy, R.; Saya, D.; Heldman, E.;
Mechoulam, R. J. Neurochem. 1993, 61, 352.
4. Felder, C. C.; Briley, E. M.; Axelrod, J.; Simpson, J. T.;
Mackie, K.; Devane, W. A. Proc. Natl. Acad. Sci. U.S.A.
1993, 90, 7656.
5. Pertwee, R. G.; Stevenson, L. A.; Elrick, D. B.; Mechou-
lam, R.; Corbett, A. D. Br. J. Pharmacol. 1992, 105, 980.
6. Pertwee, R. G.; Fernando, S. R.; Griffin, G.; Abadji, V.;
Makriyannis, A. Eur. J. Pharmacol. 1995, 272, 73.
7. Fride, E.; Mechoulam, R. Eur. J. Pharmacol. 1993, 231, 313.
8. Crawley, J. N.; Corwin, R. L.; Robinson, J.; Felder, C. C.;
Devane, W. A.; Axelrod, J. Pharmacol. Biochem. Behav. 1993,
46, 967.
9. Smith, P. B.; Compton, D. R.; Welch, S. P.; Razdan, R. K.;
Mechoulam, R.; Martin, B. R. J. Pharmacol. Exp. Ther. 1994,
270, 219.
10. Jarbe, T. U. C.; Sheppard, R.; Lamb, R. J.; Makriyannis,
A.; Lin, S.; Goutopoulos, A. Behav. Pharmacol. 1998, 9, 169.
11. Romero, J.; Garcia-Palomero, E.; Lin, S.; Ramos, J. A.;
Makriyannis, A.; Fernadez-Ruiz, J. J. Life Sci. 1996, 58, 1249.
12. Deutsch, D. G.; Chin, S. A. Biochem. Pharmacol. 1993,
46, 791.
3
0.8 nM[ H]CP-55,940, and various concentrations of
anandamide analogues in a final volume of 200 mL.
Assays were incubated at 30 ꢁC for 1 h. The samples
were filtered using Packard Filtermate 196 and What-
man GF/C Filterplates and washed with wash buffer
(TME) containing 0.5% BSA. Radioactivity was detec-
ted using MicroScint 20 scintillation cocktail added
directly to the dried filterplates, and the filterplates were
counted using a Packard Instruments Top-Count.
Nonspecific binding was assessed using 100 nMCP-
55,940. Data collected from three independent experi-
ments performed with duplicate determinations were
normalized between 100 and 0% specific binding for
[3H]CP-55,940, determined using buffer and 100 nM
CP-55,940. The normalized data were analyzed using a
4 parameter nonlinear logistic equation to yield IC50
13. Childers, S. R.; Sexton, T.; Roy, M. B. Biochem. Pharma-
col. 1994, 47, 711.
14. Desarnaud, F.; Cadas, H.; Piomelli, D. J. Biol. Chem.
1995, 270, 6030.
15. Maurelli, S.; Bisogno, T.; De Petrocellis, L.; Di Luccia,
A.; Marino, G.; Di Marzo, V. FEBS Lett. 1995, 377, 82.
16. Deutsch D. G., Makriyannis A. NIDA Res. Monograph
1997, 173, 65.
17. Hanus, L.; Gopher, A.; Almog, S.; Mechoulam, R. J.
Med. Chem. 1993, 36, 3032.
18. Beltramo, M.; Stella, N.; Calignano, A.; Lin, S. Y.; Mak-
riyannis, A.; Piomelli, D. Science 1997, 277, 1094.

1684
A. Goutopoulos et al. / Bioorg. Med. Chem. 9 (2001) 1673–1684
19. Piomelli, D.; Beltramo, M.; Glasnapp, S.; Lin, S. Y.;
Goutopoulos, A.; Xie, X.-Q.; Makriyannis, A. Proc. Natl.
Acad. Sci. U.S.A. 1999, 96, 5802.
20. Giufrida, A.; Parsons, L. H.; Kerr, T. M.; Rodriguez de
Fonseca, F.; Navarro, M.; Piomelli, D. Nature Neurosci. 1999,
2, 358.
21. Self, D. W. Nature Neurosci. 1999, 2, 313.
22. Ryan, W. J.; Banner, W. K.; Wiley, J. L.; Martin, B. R.;
Rajdan, R. K. J. Med. Chem. 1997, 40, 3617.
34. Sheskin, T.; Hanus, L.; Slager, J.; Vogel, Z.; Mechoulam,
R. J. Med. Chem. 1997, 40, 659.
35. Lang, W.; Qin, C.; Lin, C.; Khanolkar, A. D.; Gouto-
poulos, A.; Fan, P.; Abouzid, K.; Meng, Z.; Biegel, D.; Mak-
riyannis, A. J. Med. Chem. 1999, 42, 896.
36. Evans, D. A.; Weber, A. E. J. Am. Chem. Soc. 1986, 108,
6757.
37. Evans, D. A.; Ennis, M. D.; Mathre, D. J. E. J. Am.
Chem. Soc. 1982, 104, 1737.
23. Seltzman, H. H.; Fleming, D. N.; Thomas, B. F.; Gilliam,
A. F.; McCallion, D. S.; Pertwee, R. G.; Compton, D. R.;
Martin, B. R. J. Med. Chem. 1997, 40, 3625.
38. Evans, D. A.; Bartoli, J. Tetrahedron Lett. 1982, 23, 807.
39. Evans, D. A.; Rieger, D. L.; Jones, T. K.; Kaldor, S. W. J.
Org. Chem. 1990, 55, 6260.
24. Adams, I. B.; Ryan, W.; Singer, M.; Thomas, B. F.;
Compton, D. R.; Razdan, R. K.; Martin, B. R. J. Pharmacol.
Exp. Ther. 1995, 273, 1172.
25. Pinto, J. C.; Potie, F.; Rice, K. C.; Boring, D.; Johnson,
M. R.; Evans, D. M.; Wilken, G. H.; Cantrell, C. H.; Howlett,
A. C. Mol. Pharmacol. 1994, 46, 516.
26. Adams, I. B.; Ryan, W.; Singer, M.; Razdan, R. K.;
Compton, D. R.; Martin, B. R. Life Sci. 1995, 56, 2041.
27. Khanolkar, A. D.; Abadji, V.; Lin, S. Y.; Hill, W. A. G.;
Taha, G.; Abouzid, K.; Meng, Z.; Fan, P.; Makriyannis, A. J.
Med. Chem. 1996, 39, 4515.
28. Abadji, V.; Lin, S. Y.; Taha, G.; Griffin, G.; Stevenson,
L. A.; Pertwee, R. G.; Makriyannis, A. J. Med. Chem. 1994,
37, 1889.
29. Wise, M. L.; Soderstrom, K.; Murray, T. F.; Gerwick,
W. H. Experientia 1996, 52, 88.
30. Lin, S. Y.; Khanolkar, A. D.; Fan, P.; Goutopoulos, A.;
Qin, C.; Papahatjis, D.; Makriyannis, A. J. Med. Chem. 1998,
41, 5353.
31. Hillard, C. J.; Manna, S.; Greenberg, M. J.; Dicamelli, R.;
Ross, R. A.; Stevenson, L. A.; Murphy, V.; Pertwee, R. G.;
Cambell, W. B. J. Pharmacol. Exp. Ther. 1999, 289, 1427.
32. Hampson, A. J.; Hill, W. A. G.; Zan Phillips, M.; Mak-
riyannis, A.; Leung, E.; Eglen, R. M.; Bornheim, L. M. Bio-
chim. Biophys. Acta 1995, 1259, 173.
40. Ager, D. J.; Prakash, I.; Schaad, D. R. Chem. Rev. 1996,
96, 835.
41. Devane, W. A.; Dysarz, F. A.; Johnson, R. M.; Melvin,
L. S.; Howlett, A. C. Mol. Pharmacol. 1988, 34, 605.
42. Charalambous, A.; Guo, Y.; Houston, D. B.; Howlett,
A. C.; Compton, D. R.; Martin, B. R.; Makriyannis, A. J.
Med. Chem. 1992, 35, 3076.
43. Hill, W. A. G., Morse, K. L., Colasanto, J., Khanolkar,
A., Makriyannis, A. Personal communication.
44. SYBYL (version 6. 5) molecular modeling software packa-
ges from TRIPOS Associates Inc., St. Louis, MO 63144, USA.
45. Dewar, M. A.; Zoebisch, E. G.; Healy, E. F.; Stewart, J. P.
J. Am. Chem. Soc. 1985, 107, 3902.
46. Sufrin, J. R.; Dunn, D.; Marshall, G. R. Mol. Pharmacol.
1981, 19, 307.
47. Xie, X.-Q.; Eissenstat, M.; Makriyannis, A. Life Sci. 1995,
56, 1963.
48. Facci, L.; Dal-Toso, R.; Romanelo, S.; Buriani, A.; Ska-
per, S. D.; Leon, A. Proc. Natl. Acad. Sci. U.S.A. 1995, 92,
3376.
49. Lang, W.; Qin, C.; Hill, W. A.; Lin, S.; Khanolkar, A.;
Makriyannis, A. Anal. Biochem. 1998, 261, 8.
50. Newman, M. S. J. Am. Chem. Soc. 1950, 72, 4783.
51. Dodd, P. R.; Hardy, J. A.; Oakley, A. E.; Edwardson,
J. A.; Perry, E. K.; Delaunoy, J.-P. Brain Res. 1981, 226, 107.
52. Cheng, Y. C.; Prusoff, W. H. Biochem. Pharmacol. 1973,
22, 3099.
33. Ryan, W. J.; Banner, K. W.; Crocker, P. J.; Martin, B. R.;
Razdan, R. K. Bioorg. Med. Chem. Lett. 1997, 7, 2269.