Synthesis and biology of the conformationally restricted ACPD analogue, 2-aminobicyclo[2.1.1]hexane-2,5-dicarboxylic Acid-I, a potent mGluR agonist

Article abstract of DOI:10.1021/jm970719q

To better characterize the roles of metabotropic glutamate receptors (mGluRs) in physiological and pathophysiological processes, there is an important need to learn more about the structural features relevant to the design of novel, high-affinity ligands that are family and subtype specific. To date, many of the biological studies that have been conducted in the area of mGluR research have made use of the agonist (1S,3R)-ACPD. This compound has been shown to act as an agonist at both the group I and group II receptors while showing little selectivity among the four subtypes belonging to these two groups. Moreover, (1S,3S)-ACPD, the cis isomer, shows negligible activity at group I receptors and is a good agonist of mGluR2. Since ACPD is itself somewhat flexible, with four distinctive conformations being identified from molecular modeling studies for the trans isomer and five conformations for the cis isomer, we believed that it would be of interest to examine the activity of an ACPD analogue that has been constrained through the introduction of a single carbon atom bridge. Accordingly, we have prepared an aminobicyclo[2.1.1]hexanedicarboxylic acid (ABHxD-I) analogue of ACPD. The synthesis of this compound was accomplished by use of an intramolecular [2 + 2] photocycloaddition reaction, in which four distinct isomers were isolated. Of these four compounds, only a single isomer, ABHxD- I (6a), was found to be a potent agonist of the mGluRs. This compound, which expresses the fully extended glutamate conformation, was found to be more potent than ACPD at all six of the eight mGluR subtypes that were investigated and to be comparable to or more potent than the endogenous ligand, glutamate, for these receptors. Interestingly, despite its fixed conformation, ABHxD-I, like glutamate, shows little subtype selectivity. Through modeling studies of ABHxD-I (6a), ABHD-VI, LY354740, (1S,3R)-ACPD, (1S,3S)-ACPD, and L-glutamate, we conclude that the aa conformation of L- glutamate is the active conformation for both group I and group II mGluRs. Moreover, the modeling-based comparisons of these ligands suggest that the selectivity exhibited by LY354740 between the group I and group II mGluRs is not a consequence of different conformations of L-glutamate being required for recognition at these mGluRs but rather is related to certain structural elements within certain regions having a very different impact on the group I and group II mGluR activity. The enhanced potency of ABHxD-I relative to trans-ACPD commends it as a useful starting point in the design of subtype selective mGluR ligands.

Full text of DOI:10.1021/jm970719q

J . Med. Chem. 1998, 41, 1641-1650
1641
Syn th esis a n d Biology of th e Con for m a tion a lly Restr icted ACP D An a logu e,
2-Am in obicyclo[2.1.1]h exa n e-2,5-d ica r boxylic Acid -I, a P oten t m Glu R Agon ist
Alan P. Kozikowski,*,† Darryl Steensma,† Gian Luca Araldi,† Werner Tu¨ckmantel,† Shaomeng Wang,†
Sergey Pshenichkin,‡ Elena Surina,‡ and J arda T. Wroblewski‡
Drug Discovery Program, Institute of Cognitive and Computational Sciences, Georgetown University Medical Center,
3970 Reservoir Road, NW, Washington, D.C. 20007-2197, and Department of Pharmacology,
Georgetown University Medical Center, 3900 Reservoir Road, NW, Washington, D.C. 20007
Received October 21, 1997
To better characterize the roles of metabotropic glutamate receptors (mGluRs) in physiological
and pathophysiological processes, there is an important need to learn more about the structural
features relevant to the design of novel, high-affinity ligands that are family and subtype
specific. To date, many of the biological studies that have been conducted in the area of mGluR
research have made use of the agonist (1S,3R)-ACPD. This compound has been shown to act
as an agonist at both the group I and group II receptors while showing little selectivity among
the four subtypes belonging to these two groups. Moreover, (1S,3S)-ACPD, the cis isomer,
shows negligible activity at group I receptors and is a good agonist of mGluR2. Since ACPD
is itself somewhat flexible, with four distinctive conformations being identified from molecular
modeling studies for the trans isomer and five conformations for the cis isomer, we believed
that it would be of interest to examine the activity of an ACPD analogue that has been
constrained through the introduction of a single carbon atom bridge. Accordingly, we have
prepared an aminobicyclo[2.1.1]hexanedicarboxylic acid (ABHxD-I) analogue of ACPD. The
synthesis of this compound was accomplished by use of an intramolecular [2 + 2] photocy-
cloaddition reaction, in which four distinct isomers were isolated. Of these four compounds,
only a single isomer, ABHxD-I (6a ), was found to be a potent agonist of the mGluRs. This
compound, which expresses the fully extended glutamate conformation, was found to be more
potent than ACPD at all six of the eight mGluR subtypes that were investigated and to be
comparable to or more potent than the endogenous ligand, glutamate, for these receptors.
Interestingly, despite its fixed conformation, ABHxD-I, like glutamate, shows little subtype
selectivity. Through modeling studies of ABHxD-I (6a ), ABHD-VI, LY354740, (1S,3R)-ACPD,
(1S,3S)-ACPD, and L-glutamate, we conclude that the aa conformation of L-glutamate is the
active conformation for both group I and group II mGluRs. Moreover, the modeling-based
comparisons of these ligands suggest that the selectivity exhibited by LY354740 between the
group I and group II mGluRs is not a consequence of different conformations of L-glutamate
being required for recognition at these mGluRs but rather is related to certain structural
elements within certain regions having a very different impact on the group I and group II
mGluR activity. The enhanced potency of ABHxD-I relative to trans-ACPD commends it as a
useful starting point in the design of subtype selective mGluR ligands.
In tr od u ction
selective agonist (1S,3R)-1-aminocyclopentane-1,3-di-
carboxylic acid [(1S,3R)-ACPD] generally through
measurements involving phosphoinositide hydrolysis or
Ca2+ mobilization. To date the use of expression cloning
techniques has led to the identification of eight mGluR
subtypes which have been placed into three major
categories based on their molecular structure, signal
transduction mechanisms, and pharmacological proper-
ties.3 Group I mGluRs (mGluR1 and 5) are coupled to
phosphoinositide (PI) hydrolysis, whereas group II
(mGluR2 and 3) and group III (mGluR4, 6, 7, and 8)
are negatively linked to adenylyl cyclase activity. The
group I receptors are more sensitive to quisqualic acid
than they are to ACPD, the group II receptors are more
sensitive to ACPD than quisqualic acid, and the group
III receptors are most sensitive to 2-amino-4-phospho-
nobutyric acid (L-AP4). While the mGluRs have been
shown to play a fundamental role in the modulation of
synaptic transmission, considerable attention has also
The amino acid glutamate (Glu) plays a pivotal role
in biological processes ranging from memory and learn-
ing to neuronal degeneration. This major excitatory
amino acid (EAA) acts through disparate Glu receptors,
which can be categorized into two distinct types, the
ionotropic glutamate receptors and the metabotropic
glutamate receptors. The ionotropic Glu receptors, or
iGluRs, are associated with integral cation-specific ion
channels and include the NMDA [N-methyl-D-aspartic
acid], AMPA [2-amino-3-(3-hydroxy-5-methylisoxazol-
4-yl)propanoic acid], and KA (kainic acid) subtypes.1 On
the other hand, the metabotropic Glu receptors (mGluR)
are coupled to cellular effectors through GTP-binding
proteins.2 The mGluRs have been distinguished phar-
macologically from the iGluRs by the use of the mGluR-
† Drug Discovery Program.
‡ Department of Pharmacology.
S0022-2623(97)00719-X CCC: $15.00 © 1998 American Chemical Society
Published on Web 04/11/1998

1642 J ournal of Medicinal Chemistry, 1998, Vol. 41, No. 10
Kozikowski et al.
been given to their role in neurodegenerative diseases
including Alzheimer’s disease, cerebral ischemia, Par-
kinsonism, spinal cord injury, and epilepsy.4
cal [2 + 2] cycloaddition reaction of an appropriately
functionalized dehydroglutamate. One compelling vir-
tue of this approach was that we could employ L-serine
as a chiral educt, thus allowing us to access the required
amino diacids in optically pure form. Accordingly, by
employing the procedure first developed by Seebach,14
the oxazolidine 1 was generated in good yield and in
excellent optical purity (Scheme 1). Next, treatment of
1 with HCl in methanol afforded R-allylserine methyl
ester which was protected with benzyl chloroformate to
give compound 2. The alcohol was oxidized under
Swern conditions, and the resulting aldehyde was
treated with methyl (triphenylphosphoranylidene)ac-
etate to produce the R,â-unsaturated ester 3, the
required precursor for the [2 + 2] photocycloaddition
chemistry.15 Diene 3 was then irradiated in the pres-
ence of acetophenone in benzene as solvent using either
a quartz or Pyrex apparatus to afford a mixture of all
four possible bicyclic products 4a -d .
Compounds 4a -c were isolated as a chromatographi-
cally inseparable mixture while 4d was isolated as a
single compound. The mixture of compounds 4a -c was
deprotected with Pd/C under an atmosphere of H2 to
afford compounds 5a -c as three chromatographically
separable isomers. Compound 4d was also deprotected
using H2 and Pd/C. The diesters 5a -d were individu-
ally treated with 6 N HCl at reflux for 1 h to afford the
final products 6a -d .
To better characterize the roles of GluRs in physi-
ological and pathophysiological processes, there is an
important need to learn more about the structural
features relevant to the design of novel, high-affinity
ligands that are family and subtype specific. To date,
many of the biological studies that have been conducted
in the area of mGluR research have made use of the
agonist (1S,3R)-ACPD. This compound has been shown
to act as an agonist at both the group I and group II
receptors while showing little selectivity among the four
subtypes belonging to these two groups. Moreover,
(1S,3S)-ACPD, the cis isomer, shows negligible activity
at group I receptors and is a good agonist of mGluR2.5
Since ACPD is itself somewhat flexible, with four
distinctive conformations being identified from molec-
ular modeling studies for the trans isomer, and five
conformations for the cis isomer,6 we believed that it
would be of interest to examine the activity of ACPD
analogues that have been constrained through the
introduction of a single carbon atom bridge. Specifical-
ly, we envisioned adding the methylene group so as to
bridge carbon atoms 2 and 4 of the cyclopentane ring.
In pursuit of this idea, it was of interest to find a
literature report describing the synthesis of several
norbornane analogues of ACPD.7 However, none of
these compounds were found to be particularly promis-
ing because of their low potency, with the best com-
pound acting as an antagonist at mGluR1a with a KB
value of 300 µM. Also, this series of molecules was
prepared only in racemic form. As will be seen below,
the biological activity found for our one carbon lower
homologues stands in stark contrast to the activity
reported for the norbornanes.
Ster eoch em ica l Assign m en ts
The assignment of stereochemistry to each of the four
diastereoisomeric amino acids 6a -d was based on high-
field NMR experiments of the methyl esters 5a -d .
From the 1D 1H NMR spectrum, the diagnostic W-
coupling constant J 1-4 for the bicyclo[2.1.1]hexanes
5a -d was determined to be 6.9 Hz, a number which is
typical for the coupling of bridgehead protons. This
value is also in agreement with that reported in the
literature for 2-azabicyclo[2.1.1]hexane.16 A remarkably
large W-coupling of about 8 Hz was observed between
H5 and H6R for both compounds 5a and 5b. Also
characteristic for these two derivatives is the W-coupling
between H3â and H6â of 2.4 and 2.7 Hz. 2D-NOESY
experiments on compound 5a show correlations between
H5 and H3â, H5 and NH2, and H6R and H3R (Scheme 2).
These results suggest that the carboxylate groups in 5a
have a cis relationship. The 2D NOESY experiments
on 5b revealed correlations between H6R and H3R and
between H6R and NH2. These results define a trans
relationship for the two carboxylate groups (Scheme 2).
Within the field of new compound development for the
mGluRs, notable recent achievements include the iden-
tification of 2-aminobicyclo[3.1.0]hexane-2,6-dicarboxylic
acid (LY354740) as a potent group II agonist (EC50
)
5.1 nM at mGluR2),8 and reports of the enhanced
potency of (2S,1′S,2′S,3′R)-2-(2′-carboxy-3′-phenylcyclo-
propyl)glycine (PCCG-4)9 and (2S,1′S,2′S)-2-(2′-carboxy-
cyclopropyl)-3-(9-xanthenyl)alanine (LY341495)10 as
mGluR2 antagonists. (2R,4R)-4-Aminopyrrolidine-2,4-
dicarboxylic acid (2R,4R-APDC), a structural analogue
of (1S,3R)-ACPD, is selective for human mGluR2 and
mGluR3 and is inactive at other mGluR subtypes.11 The
1-benzyl derivative of (2R,4R)-APDC, on the other hand,
shows some selectivity for mGluR6.6 3,5-Dihydroxy-
phenylglycine is a selective agonist for group I mGluRs
with no activity at the group II and group III mGluRs.12
Last, a recent article provides an account of the mGluR
activity of ACPD analogues containing an additional
carboxyl group.13
1
The 1D H NMR spectra for compounds 5c and 5d
show no coupling between H5 and H6R and exhibit small
coupling constants (2.1 Hz) between H3â and H6â. The
depicted stereochemistry of 5c follows from the observa-
tion that this compound undergoes a facile intramo-
lecular cyclization reaction to afford the corresponding
five-membered ring lactam.
The above spectral and chemical information thus
provides strong support for our stereochemical assign-
ments.
P h a r m a cologica l Resu lts
Ch em ica l Syn th esis of th e ABHxDs
As an obvious synthetic approach to the ABHxDs, we
considered carrying out an intramolecular photochemi-
The activity of ABHxD isomers was tested in cell lines
expressing the individual mGluRs. Group I receptors

Conformationally Restricted ACPD Analogue
J ournal of Medicinal Chemistry, 1998, Vol. 41, No. 10 1643
Sch em e 1. Synthetic Route to the ABHxD Analogues of ACPD
Sch em e 2. Structural Elucidation of Compounds 5a -d
mGluRs), and Figure 3 (group III mGluRs). These
studies are summarized in Table 1, which shows the
EC50 values at all six mGluR subtypes. Among the
tested compounds, ABHxD-I is the most potent in
activating all six mGluR subtypes, with potencies in
most cases comparable to or higher than that of the
endogenous agonist glutamate. In contrast, ABHxD-
IV shows little activity at any of the tested mGluRs. In
all cases in which agonist activity was not observed at
1 mM concentrations, compounds were also tested for
possible antagonist activity. However, none of the
ABHxD isomers showed antagonist action at any of the
six mGluRs (data not shown).
To further establish their selectivity, the ABHxD
isomers were tested for possible activity at ionotropic
glutamate receptors by measuring 45Ca2+ influx in
primary cultures of cerebellar neurons.19 As shown in
Table 2, in conditions where agonists of all ionotropic
glutamate receptors are active, none of the ABHxD
isomers stimulated 45Ca2+ influx. Tests of possible
antagonistic action showed that several ABHxD isomers
used at 1 mM concentrations partially inhibited NMDA
responses. The strongest inhibition was produced by
ABHxD-IV, which was inactive at mGluRs. The re-
sponses induced by kainate and AMPA were not sub-
stantially inhibited by any of the ABHxD isomers.
ABHxD-I, the most potent mGluR agonist, produced a
slight inhibition of NMDA and kainate responses;
however, this effect appeared at a 1 mM concentration,
which is 3 orders of magnitude higher than the potency
of ABHxD-I to activate most of the mGluRs.
(mGluR1a and mGluR5a) are coupled to phospholipase
C, and their activity was determined by accumulation
of inositol phosphates. Group II receptors included the
cAMP-coupled mGluR2, as well as the chimeric receptor
mGluR3/1a which, as shown previously,17 combines the
pharmacological properties of mGluR3 with the activa-
tion of phospholipase C. Group III receptors were
represented by mGluR4a and mGluR6, both coupled to
the inhibition of cAMP formation.18 The agonist dose-
response curves for all four ABHxD isomers are shown
in Figure 1 (group I mGluRs), Figure 2 (group II
Molecu la r Mod elin g Stu d ies
To gain further insights into the structural/confor-
mational requirements for ligand potency and selectivity
among the mGluR groups, we have performed confor-

1644 J ournal of Medicinal Chemistry, 1998, Vol. 41, No. 10
Kozikowski et al.
F igu r e 1. Agonist dose-response curves of the different
ABHxD isomers at group I mGluRs. IP formation was mea-
sured in cells expressing mGluR1a (A) and mGluR5a (B)
receptors after a 30 min incubation with the indicated con-
centration of respective agonists. Data are expressed as the
percent of maximal response obtained with 1 mM glutamate
after subtracting the basal IP formation. Values are means (
SEM of four to six independent experiments performed in
duplicate.
F igu r e 2. Agonist dose-response curves of the different
ABHxD isomers at group II mGluRs. In cells expressing
mGluR2 (A), agonist activity was measured as the decrease
of forskolin-induced cAMP accumulation during a 10 min
incubation with the indicated concentrations of agonists. In
cells expressing the chimeric mGluR3/1a (B), the IP formation
was measured after a 30 min incubation with agonists. Data
are expressed as the percent of maximal response obtained
with 1 mM glutamate after subtracting the basal values.
Values are means ( SEM of four to six independent experi-
ments performed in duplicate.
mational analyses on all four ABHxD isomers using
high-temperature molecular dynamics simulations fol-
lowed by energy minimization of each conformation
generated by the simulations. These studies confirmed
the expectation that the ABHxD isomers are rather
inflexible. The superposition of the energy-minimized
conformations of each of these isomers using all heavy
atoms, excluding the four terminal oxygens, gave a root-
mean-square value of less than 0.1 Å, indicating that
each isomer essentially adopts only one low-energy
conformation.
The low-energy conformation of each compound was
then compared to that of L-glutamate, the endogenous
ligand, and LY354740 (structure shown at the bottom
of Table 3), a potent and selective group II mGluR
agonist. L-glutamate adopts nine low-energy, distinct
conformations. The aa conformation of L-glutamate is
believed to be the biologically active conformation in
binding to mGluRs.6,9 We have shown previously that
LY354740 can adopt two slightly different, low-energy
conformations, both of which have excellent overlap
(RMS ) 0.17 Å) with L-glutamate’s aa conformation.6
LY354740 is a potent group II mGluR agonist with an
EC50 value of 5.1 nM at human mGluR2 and 24.3 nM
at human mGluR3, while it has no activity at cells
expressing human mGluR4 or mGluR7 (group III) and
fails to activate or inhibit human mGluR1a or mGluR5a
when tested at concentrations up to 1 µM.8 Consistent
with the findings of others,8,9,21 we have therefore
concluded that the aa conformation of L-glutamate is
the active conformation for the group II mGluRs.
Among the four ABHxD isomers, ABHxD-II (6b),
ABHxD-III (6c), and ABHxD-IV (6d ) have very poor
overlap with the aa conformation of L-glutamate (Table
3 and Figure 4), and their RMS values range from 0.58
Å (6b) to 0.89 Å (6d ). These results are consistent with
the poor activity of 6b-d at the group II mGluRs (Table
1). On the other hand, ABHxD-I (6a ) has an excellent
overlap with the aa conformation of L-glutamate, with
an RMS value of 0.18 Å (Table 3 and Figure 4). This
result is again consistent with the activity of 6a at group
II mGluRs (Table 1).
ABHxD-I (6a ) also shows fairly potent activity at the
group I mGluRs (Table 1). Although both ABHxD-I and
LY354740 are rigid structures and exhibit excellent
overlap with the aa conformation of L-glutamate (RMS
value 0.17 vs 0.18 Å), it is intriguing that LY354740 is
a very selective group II mGluR agonist, while ABHxD-I
(6a ) exhibits little selectivity among group I, group II,
and group III mGluRs. Accordingly, it is clear that
structural elements other than conformation must play
an important role in determining the selectivity of these
compounds for the mGluR subtypes.

Conformationally Restricted ACPD Analogue
J ournal of Medicinal Chemistry, 1998, Vol. 41, No. 10 1645
regions these two compounds occupy similar three-
dimensional space. Because LY354740 is a potent and
selective group II mGluR ligand and ABHxD-I (6a ) is a
fairly potent but nonselective mGluR ligand, occupancy
of region A appears to be beneficial to the activity at
group II mGluRs and/or to be detrimental to activity at
group I and group III mGluRs. The crucial structural
difference between ABHD-VI and ABHxD-I (6a ) occurs
in region B. The very low potency of ABHD-VI at both
group I and group II mGluRs clearly suggests that
the occupancy of region B by a ligand may be detri-
mental to its activity at both group I and group II
mGluRs.
(1S,3R)-ACPD has been used as a mGluR ligand
although it has little selectivity between group I and
group II mGluRs. On the other hand, (1S,3S)-ACPD is
a selective ligand for group II mGluRs. However, both
ligands are not very potent. Conformational analysis
showed that (1S,3R)-ACPD can adopt four different
conformations and (1S,3S)-ACPD can adopt five differ-
ent conformations using an RMS equal to 0.1 Å as the
cutoff value.20 Superposition of these low-energy, dis-
tant conformations of (1S,3R)-ACPD on the aa confor-
mation of L-glutamate showed that none of these
conformations displays substantially good overlap, with
an average RMS value of 0.59 Å being observed. The
previously proposed mGluR active conformation for
(1S,3R)-ACPD has an RMS value of 0.62 Å when
superimposed on the aa conformation of L-glutamate
(Figure 6).7 The modeling results for (1S,3R)-ACPD are
consistent with the experimental results that (1S,3R)-
ACPD is not a very potent mGluR ligand.
F igu r e 3. Agonist dose-response curves of the different
ABHxD isomers at group III mGluRs. In cells expressing
mGluR4a (A) and mGluR6 (B), agonist activity was measured
as the decrease of forskolin-induced cAMP accumulation
during a 10 min incubation with the indicated concentrations
of agonists. Data are expressed as the percent of maximal
response obtained with 1 mM glutamate for mGluR4a and
with 1 mM AP4 for mGluR6 after subtracting the basal values.
Values are means ( SEM of four to six independent experi-
ments performed in duplicate.
For (1S,3S)-ACPD, one of the low-energy conforma-
tions has an excellent overlap with the aa conformation
of L-glutamate (Figure 6), with an RMS value of 0.20
Å, while the other four conformations overlap poorly
with the aa conformation of L-glutamate (the RMS
values range from 0.47 to 0.78 Å). Based solely upon
the excellent overlap of (1S,3S)-ACPD on the aa con-
formation of L-glutamate, one might predict (1S,3S)-
ACPD to be a potent mGluR ligand. Experimentally,
(1S,3S)-ACPD is a modestly active ligand at group II
(13 µM at mGluR22 which compares to an EC50 of ∼2
µM in our own system) and group III (50 µM at
mGluR4a)2 receptors as compared to L-glutamate, while
its activity at group I is >300 µM. In comparison with
ABHD-VI (Figure 6), one of the carbon atoms of (1S,3S)-
ACPD was found to occupy region B. In view of the fact
that (1S,3S)-ACPD, like ABHD-VI, is poorly active at
group I receptors, one may conclude that occupancy of
region B is detrimental to group I activity, but that
occupancy of this region does not preclude group II
activity. These results once again reveal that an
appropriate ligand conformation, while necessary for
potent mGluR activity, is not in and of itself sufficient
for achieving such activity.
To shed more light on the possible structural elements
that may be important to the selectivity and/or potency
of these compounds, we have compared LY354740,
ABHxD-I (6a ), and the aa conformation of L-glutamate.
In addition, we have included one of the aminobicyclo-
[2.2.1]heptanedicarboxylic acids (ABHD-VI), which can
be viewed as an analogue of cis-ACPD that has been
rigidified by the incorporation of a two-carbon bridge
(see bottom of Table 3). ABHD-VI was found to have
only weak activity at group I (antagonist at mGluR1a)
and group II (agonist at mGluR2) mGluRs when tested
at a concentration of 3 mM.7 The superposition of these
compounds on the aa conformation of L-glutamate is
shown in Figure 5. Like LY354740 and ABHxD-I (6a ),
ABHD-VI was also found to show excellent overlap with
the aa conformation of L-glutamate (RMS ) 0.17 Å,
Table 3 and Figure 5). This result clearly suggests that
good overlap with the aa conformation of L-glutamate
is a necessary but not a sufficient condition to achieve
potent activity at group I and group II mGluRs. As can
be observed in Figure 5, the comparison of these
compounds reveals two regions where significant struc-
tural differences occur among these compounds. The
crucial structural difference between LY354740 and
ABHxD-I (6a ) occurs in region A, while in all other
In summary, through modeling studies of ABHxD-I
(6a ), ABHD-VI, LY354740, (1S,3R)-ACPD, (1S,3S)-
ACPD, and L-glutamate, we conclude that the aa
conformation of L-glutamate is the active conformation
for both group I and group II mGluRs. The selectivity
exhibited by LY354740 between the group I and group
II mGluRs is not a consequence of different conforma-
tions of L-glutamate being required for recognition at

1646 J ournal of Medicinal Chemistry, 1998, Vol. 41, No. 10
Kozikowski et al.
Ta ble 1. EC50 Values Obtained for the ABHxDs at Six of the mGluR Subtypesa
EC50 values, µM
group II mGluRs
group I mGluRs
mGluR1a mGluR5a
4.9 ( 0.21
15 ( 2.2
>300b
1.6 ( 0.14
121 ( 10
232 ( 23
>1000
group III mGluRs
agonist
mGluR2
mGluR3/1a
mGluR4a
mGluR6
Glu
3.1 ( 0.23
23 ( 2.6
0.29 ( 0.07
2.0( 0.3
1.9 ( 0.31
40 ( 5.8
9.8 ( 0.81
4.9 ( 0.37
82 ( 6.2
(()-trans-ACPD
(1S,3S)-ACPD
AP4
ABHxD-I (6a )
ABHxD-II (6b)
ABHxD-III (6c)
ABHxD-IV (6d )
∼800
50b
0.33 ( 0.09
23 ( 7.1
>300b
13b
30 (mGluR3)b
0.28 ( 0.03
5.3 ( 0.93
0.72 ( 0.11
57 ( 6
91 ( 11
0.33 ( 0.06
54 ( 9
38 ( 10
2.2 ( 1.5
∼1000c
∼1000
∼800
∼1000c
>1000
>1000
>1000
>1000
>1000
>1000
>300c
a
EC50 values were calculated by fitting the normalized data (shown in Figures 1-3) to the logistic equation by nonlinear regression.
b
Values represent means ( SEM from four to six experiments performed in duplicate as described in the Experimental Section. Data
taken from ref 2. c These EC50 values were estimated assuming that the compounds act as full agonists.
Ta ble 2. Effects of ABHxD Compounds on Ionotropic Glutamate Receptors in Cerebellar Neuronsa
antagonist activityc
agonist activityb
% of basal activity
% of NMDA
effect
% of kainate
effect
% of AMPA
effect
additions
NMDA
kainate
AMPA
MK-801
NBQX
ABHxD-I
ABHxD-II
ABHxD-III
ABHxD-IV
100 µM
50 µM
30 µM
1 µM
10 µM
1 mM
1 mM
1 mM
1 mM
245 ( 24
243 ( 14
389 ( 23
62 ( 14
79 ( 5.9
102 ( 15
82 ( 7.2
93 ( 4.5
69 ( 11
100 ( 9.8
100 ( 5.9
100 ( 6.0
0 ( 0.0
3 ( 0.3
77 ( 4.9
94 ( 8.1
97 ( 3.4
91 ( 3.5
11 ( 0.6
95 ( 9.9
88 ( 3.2
100 ( 2.0
86 ( 2.8
70 ( 3.9
58 ( 2.5
96 ( 4.4
17 ( 0.9
a
Activity at ionotropic glutamate receptors was measured as 45Ca2+ influx in cerebellar granule neurons as described in the Experimental
b
Section. All values represent means ( SEM from six measurements performed on two separate preparations of neuronal cultures. Agonist
activity at all ionotropic glutamate receptors was measured in Mg2+-free medium supplemented with 1 µM glycine to allow for activity
at NMDA receptors, and with 10 µM cyclothiazide to inhibit desensitization of AMPA receptors, and is expressed as percent of basal
45Ca2+ influx. c Antagonist activity at NMDA receptors was measured in a Mg2+-free medium in the presence of 100 µM NMDA, 1 µM
glycine, and 10 µM NBQX to inhibit non-NMDA receptors. Antagonism of kainate action was tested in the presence of 1 mM Mg2+, 50 µM
kainate, and 1 µM MK-801 to inhibit NMDA receptors, while the effects on AMPA action were measured in the presence of 1 mM Mg2+
,
30 µM AMPA, 10 µM cyclothiazide, and 1 µM MK-801. All antagonist effects are expressed as percent of 45Ca2+ influx induced by agonist
alone.
Ta ble 3. Calculated RMS Deviations for Selected mGluR
Ligands in Comparison to the aa Conformation of L-Glutamatea
ligand
RMS (Å)
ligand
RMS (Å)
ABHxD-I (6a )
ABHxD-II (6b)
ABHxD-III (6c)
ABHxD-IV (6d )
0.18
0.58
0.77
0.89
LY354740
ABHD-VI
(1S,3R)-ACPD
(1S,3S)-ACPD
0.17
0.17
0.62
0.20
a
the group I and group II mGluRs, but rather is related
to certain structural elements within a certain region
(region A) having a very different impact on the group
I and group II mGluR activity. We have determined
that the occupancy of region A by a ligand may enhance
its activity at group II mGluRs, while possibly being
detrimental to its group I and group III mGluR activity.
Modeling comparisons of ABHD-VI also suggests that
the occupancy of region B is prohibitive to achieving
good activity at group I mGluRs. Last, on the basis of
the fairly good activity of ABHxD-I (6a ) at group III
mGluRs, we propose that the active conformation for
the group III mGluRs should be one that resembles the
aa conformation of L-glutamate. The current modeling
studies thus offer additional insights into the design of
potent, group specific mGluR ligands.
F igu r e 4. Overlay of the minimized conformations of the
ABHxD isomers 6a -d with the aa conformation of L-
glutamate.
Discu ssion
As is apparent from the data presented in Table 1,
the bicycle ABHxD-I (6a ) is more potent than ACPD as
an agonist at all six of the eight subtypes tested. The
increase in potency ranges from 6- to 34-fold. The other
three isomers show much poorer or little activity. It is
interesting to note that the rank order of potency of
trans-ACPD at the mGluRs is mGluR2 > 1a > 5a > 3
> 6 > 4a, while for the bicycle 6a the rank order is
mGluR2 > 5a > 1a > 3 > 6 > 4a. These potency orders

Conformationally Restricted ACPD Analogue
J ournal of Medicinal Chemistry, 1998, Vol. 41, No. 10 1647
son to the readily mobile ligand L-glutamate. Therefore,
to achieve any real differences in subtype selectivity
with ABHxD-I or related molecules will require the
addition of other functionality capable of either prevent-
ing or enhancing receptor interactions through a com-
bination of steric and electronic effects. In particular,
as already delineated in the modeling section, occupancy
of region A (Figure 5) by a ligand appears to enhance
group II activity, while possibly being detrimental to
group I and group III activity. Moreover, ligand oc-
cupancy of region B appears to be prohibitive to group
I activity.
In light of the higher potency of 6a relative to ACPD,
we would suggest the former to be the preferred agent
for use in future biological studies of mGluRs in cases
where subtype selectivity is not required. In fact,
studies of a ligand capable of interacting much like
glutamate with the individual mGluRs while being
devoid of ionotropic activity may prove intriguing.
ABHxD-I (6a ) is a possible candidate for future drug
design work, as it bears an additional carbon atom,
relative to ACPD, to which diverse substituents may be
appended in order to obtain compounds that are subtype
selective.
F igu r e 5. Overlay of ABHxD-I, LY354740, and ABHD-VI
with the aa conformation of L-glutamate. Region A highlights
the extra space occupied by LY354740, while region B demar-
cates the extra steric volume of ABHD-VI.
Exp er im en ta l Section
Gen er a l. Starting materials were obtained from Aldrich
Chemical Co. or from other commercial suppliers. Solvents
were purified as follows: diethyl ether was distilled from
phosphorus pentoxide; THF was freshly distilled under nitro-
gen from sodium benzophenone.
IR spectra were recorded on an ATI Mattson Genesis spec-
trometer. 1H and 13C NMR spectra were obtained with a
Varian Unity Inova instrument at 300 and 75.46 MHz, re-
spectively. 1H chemical shifts (δ) are reported in ppm down-
field from internal TMS. 13C chemical shifts are referenced
to CDCl3 (central peak, δ ) 77.0 ppm), benzene-d6 (central
peak, δ ) 128.0 ppm), or DMSO-d6 (central peak, δ ) 39.7
ppm). NMR assignments were made with the help of COSY,
DEPT, HETCOR, and NOESY experiments. The subscript
numbers used to identify the ABHxD protons are defined
according to the following numbering scheme:
F igu r e 6. Overlay comparisons of (1S,3R)-ACPD, (1S,3S)-
ACPD, ABHD-VI, and L-glutamate, which emphasizes partial
occupation of region B by (1S,3S)-ACPD.
are basically the same, except for the switch in activity
at 1 and 5. The value of 6a as a research tool is further
enhanced by its general lack of activity for members of
the ionotropic receptor family.
Previously it has been proposed by Pellicciari that the
fully extended conformation of glutamate represents the
bioactive conformation at the group II mGluRs.8,9,21 Due
to the rigidity of the bicycle 6a , it is clear that its
embedded glutamate moiety may adopt only the fully
extended (aa) conformation. This structural preference
is supported by the modeling studies, as detailed in the
previous section. The excellent potency of 6a for
mGluR2 therefore provides further experimental sup-
port for this proposed structural preference. In view of
the more or less parallel activity of trans-ACPD and 6a
at each of the six subtypes tested, it would also be
reasonable to conclude that the active conformation of
ACPD at each of these receptor subtypes must be that
which best mimics an extended conformation.
In light of the fact that ABHxD-I shows little real
selectivity for any of the mGluRs, with the exception of
mGluR4a at which it is the least potent, one could argue
that the extended glutamate conformation is readily
accommodated by the majority of the mGluR subtypes.22
In fact, it is rather intriguing that a rigid molecule like
this shows so little difference in its ability to interact
with and to activate the individual mGluRs in compari-
Melting points were determined in Pyrex capillaries with a
Thomas-Hoover Unimelt apparatus and are uncorrected.
Mass spectra were measured in the EI mode at an ionization
potential of 70 eV. TLC was performed on Merck silica gel
60F254 glass plates; column chromatography was performed
using Merck silica gel (60-200 mesh). The following ab-
breviations are used: DMSO ) dimethyl sulfoxide, ether )
diethyl ether; THF ) tetrahydrofuran; DCM ) dichlo-
romethane.
(2S)-Meth yl 2-[(Ben zyloxycar bon yl)am in o]-2-(h ydr oxy-
m eth yl)-4-p en ten oa te (2). Acetyl chloride (460 mL, 6.47
mol) was added via addition funnel to methanol (1.08 L) at 0
°C. A solution of 1 (18.3 g, 71.8 mmol) in methanol (10 mL)
was added and the reaction mixture refluxed for 3 h. Evapo-
ration of the volatiles in vacuo produced a white solid which
was dissolved in dioxane (150 mL) and H2O (50 mL). NaHCO3
(12.1 g, 143.7 mmol) and benzyl chloroformate (20.5 mL, 143.67
mmol) were added, and the reaction mixture was stirred
overnight. The reaction mixture was extracted with EtOAc
(3 × 200 mL), and the combined organic phases were washed

1648 J ournal of Medicinal Chemistry, 1998, Vol. 41, No. 10
Kozikowski et al.
with brine, dried over MgSO4, filtered, and concentrated.
Column chromatography (SiO2, 50% EtOAc/hexane) afforded
3H), 4.82 (br s, NH2); 13C NMR (methanol-d4) δ 36.9 (CH2),
41.3 (CH2), 42.9 (CH), 52.4 (CH3), 53.0 (CH3), 53.9 (CH), 55.5
(CH), 64.5 (C), 174.6 (CO), 176.9 (CO).
2 (12.3 g, 58%) as a colorless oil: [R]25 -0.4° (c 2.1, CHCl3);
D
1H NMR (CDCl3) δ 2.52 (dd, J ) 14.1 and 7.2 Hz, 1H), 2.78
(dd, J ) 13.8 and 7.8 Hz, 1H), 3.51 (br s, 1H), 3.75 (s, 3H),
3.88-3.76 (m, 1H), 4.18-4.07 (m, 1H), 5.15-5.01 (m, 4H),
5.68-5.52 (m, 1H), 5.79 (br s, 1H), 7.37-7.29 (m, 5H); 13C NMR
(CDCl3) δ 36.7, 52.7, 64.9, 65.1, 66.7, 119.9, 127.9, 128.0, 128.4,
(1S,2S,4S,5S)-Dim eth yl 2-a m in obicyclo[2.1.1]h exa n e-
2,5-d ica r boxyla te (5c): yield 64 mg (20%); Rf 0.80 (EtOAc/
MeOH, 8/1); 1H NMR (CDCl3) δ 1.07 (d, J ) 7.5 Hz, H6R), 1.46-
1.56 (m, H6â), 1.67 (br s, NH2), 1.97 (dd, J ) 2.1 and 11.4 Hz,
H
3â), 2.35 (d, J ) 11.4 Hz, H3R), 2.61 (m, H5), 2.64-2.74 (m,
131.0, 136.0, 155.2, 172.3; IR (film) 3407, 1724 cm-1
.
H4), 2.9-3.0 (dt, J ) 2.7 and 6.9 Hz, H1), 3.60 (s, 3H), 3.71 (s,
3H); 13C NMR (CDCl3) δ 36.4, 36.6, 39.7, 49.4, 50.9, 52.0, 52.6,
63.3, 172.4, 176.4.
(2S)-Meth yl 4-Allyl-4-[(ben zyloxycar bon yl)am in o]glu ta-
con a te (3). A solution of DMSO (0.57 mL, 7.8 mmol) in CH2-
Cl2 (8 mL) was added to a solution of oxalyl chloride (0.51 mL,
5.9 mmol) in CH2Cl2 (4 mL) over 15 min. Alcohol 2 (1.15 g,
3.95 mmol) in CH2Cl2 (20 mL) was added over 10 min. After
an additional 10 min of stirring, Et3N (2.73 mL, 19.6 mmol)
in CH2Cl2 (15 mL) was added over 10 min and the temperature
was maintained between -60 and -70 °C for 30 min. The
reaction mixture was quenched with a saturated solution of
NH4Cl (30 mL) and extracted with CH2Cl2 (5 × 40 mL). The
combined organic fractions were dried over MgSO4, filtered,
and concentrated. The residue was dissolved in CH2Cl2 (45
mL), methyl (triphenylphosphoranylidene)acetate (1.97 g, 5.88
mmol) was added, and the mixture was stirred at room
temperature for 4 h. The reaction mixture was then poured
into H2O and extracted with CH2Cl2 (5 × 40 mL). The
combined organic phases were dried over MgSO4, filtered, and
concentrated. Purification of the residue by column chroma-
tography (SiO2, 33% EtOAc/hexane) afforded compound 3 (1.36
g, quant) as a colorless oil: [R]25D -2.2° (c 1.6, CHCl3); 1H NMR
(CDCl3) δ 2.67 (dd, J ) 13.5, 7.5 Hz, 1H), 2.89 (dd, J ) 13.2,
6.6 Hz, 1H), 3.78 (s, 3H), 3.74 (s, 3H), 5.20-5.00 (m, 4H), 5.65-
5.50 (m, 1H), 5.73 (br s, 1H), 5.98 (d, J ) 15.9 Hz, 1H), 7.16
(d, J ) 15.9 Hz, 1H), 7.41-7.27 (m, 5H); 13C NMR (CDCl3) δ
40.3, 51.7, 53.3, 63.2, 66.9, 120.9, 121.5, 128.1, 128.2, 128.5,
130.4, 136.0, 145.3, 154.2, 166.2, 170.8; IR (film) 3352, 1724
(1R,2S,4R,5S)-Dim eth yl 2-a m in obicyclo[2.1.1]h exa n e-
2,5-d ica r boxyla te (5d ). Compound 4d (95 mg, 0.27 mmol)
and 10% Pd/C (15 mg) were stirred in MeOH (5 mL) under an
atmosphere of H2. After 3 h, the reaction mixture was filtered
and concentrated. Purification of the residue by silica gel
column chromatography (MeOH/EtOAc, 1/9) afforded com-
pound 5d (58 mg, 100%) as a colorless oil: 1H NMR (CDCl3) δ
1.46 (d, J ) 11.7 Hz, 1H), 1.5-1.6 (m, 1H), 1.62 (d, J ) 7.2
Hz, 1H), 1.85 (br s, 2H), 2.63-2.70 (m, 2H), 2.82 (dd, J ) 1.8
and 11.7 Hz, 1H), 2.87-2.94 (m, 1H), 3.56 (s, 3H), 3.69 (s, 3H).
Com p ou n d s 6a -d . Compound 5a -d were stirred with 6
N HCl at reflux for 1 h. Evaporation afforded compounds
6a -d .
(1S,2S,4S,5S)-2-Am in ob icyclo[2.1.1]h exa n e-2,5-d ica r -
boxylic a cid h yd r och lor id e (6a ): [R]25D +15° (c 0.5, MeOH);
1
mp >220 °C; H NMR (D2O) δ 1.87 (t, J ) 8.7 Hz, H6â), 2.05
(br d, J ) 12.6 Hz, H3â), 2.30-2.40 (m, H6â), 2.65 (d, J ) 12.6
Hz, H3R), 2.80-2.90 (m, H4), 2.92 (d, J ) 8.7 Hz, H5), 3.05-
3.20 (dd, J ) 2.1 and 6.9 Hz, H1); 13C NMR (D2O) δ 36.8 (CH2),
37.5 (CH2), 41.6 (CH), 51.1 (CH), 53.7 (CH), 64.0 (C), 173.0
(CO), 175.6 (CO). Anal. (C8H11NO4‚HCl‚0.5H2O) C, H, N.
(1R,2S,4R,5R)-2-Am in obicyclo[2.1.1]h exa n e-2,5-d ica r -
boxylic a cid h yd r och lor id e (6b): [R]25 -7.0° (c 0.5, H2O);
D
1
mp >220 °C; H NMR (D2O) δ 1.65 (t, J ) 9.0 Hz, H6R), 2.01
cm-1
.
(d, J ) 12.6 Hz, H3R), 2.46 (dt, J ) 2.4 and 9.6 Hz, H6â), 2.63
(br d, 12.9 Hz, H3â), 2.80-2.90 (m, H4), 3.08 (dd, J ) 2.4 and
6.9 Hz, H1), 3.15 (d, J ) 8.7 Hz, H5); 13C NMR (D2O) δ
36.6 (CH2), 38.0 (CH2), 41.5 (CH), 51.0 (CH), 53.7 (CH), 63.6
(C), 173.4 (CO), 176.0 (CO). Anal. (C8H11NO4‚HCl‚0.5H2O)
C, H, N.
Dim eth yl 2-[(Ben zyloxyca r bon yl)a m in o]bicyclo[2.1.1]-
h exa n e-2,5-d ica r boxyla te (4a -d ). Diene 3 (3.84 g, 11.5
mmol) and acetophenone (0.40 mL) in benzene (400 mL) were
first purged with argon and then irradiated with a Hanovia
450 W medium-pressure mercury lamp through a Pyrex filter.
After 4.5 d, the reaction mixture was evaporated and the
residue purified by column chromatography (SiO2, 33% EtOAc/
hexane) to afford a mixture of cis- and trans-3 (Rf 0.8, 50%
EtOAc/hexane, 1.16 g, 30%), 4a -c (Rf 0.65, 50% EtOAc/
hexane, 1.22 g, 32%), and 4d (Rf 0.53, 50% EtOAc/hexane, 197
(1S,2S,4S,5R)-2-Am in obicyclo[2.1.1]h exa n e-2,5-d ica r -
boxylic a cid h yd r och lor id e (6c): [R]25D -5.4° (c 0.25, H2O);
1
mp 198-200 °C; H NMR (D2O) δ 1.51 (d, J ) 8.1 Hz, H6R),
1.65-1.75 (m, H6â), 2.02 (dd, J ) 2.7 and 13.2 Hz, H3â), 2.70
(br d, J ) 13.2 Hz, H3R), 2.90-3.00 (m, H4), 3.06 (br s, H5),
3.10-3.20 (m, H1); 13C NMR (D2O) δ 34.3 (CH2), 36.3 (CH2),
43.2 (CH), 50.1 (CH), 52.0 (CH), 63.6 (C), 173.0 (CO), 177.0
(CO). Anal. (C8H11NO4‚HCl‚0.5H2O) C, H, N.
mg, 5%) as colorless oils: [R]25 +21° (c 0.7, CHCl3); 1H NMR
D
(CDCl3) δ 1.43 (d, J ) 7.2 Hz, H6R), 1.50-1.53 (m, H6â), 1.77
(br d, J ) 12.3 Hz, H3R), 2.67 (br s, H5), 2.72-2.80 (m, H4),
3.06 (dd, J ) 1.8 and 12.3 Hz, H3â), 3.20 (br s, H1), 3.60 (br s,
6H), 5.07 (s, CH2), 5.63 (br s, NH), 7.33-7.27 (m, 5H); 13C NMR
(CDCl3) δ 36.1, 37.1, 39.8, 49.6, 51.2, 51.3, 52.2, 63.3, 66.8,
128.0, 128.4, 136.1, 155.22, 170.5, 172.4.
(1R,2S,4R,5S)-2-Am in obicyclo[2.1.1]h exa n e-2,5-d ica r -
boxylic a cid h yd r och lor id e (6d ): [R]25 +6° (c 0.25, H2O);
D
1
mp (dec) 205 °C; H NMR (D2O) δ 1.45 (d, J ) 8.7 Hz, H6R),
1.80-1.90 (m, H6â), 1.95 (br d, J ) 13.2 Hz, H3R), 2.75-2.90
(m, H4 and H3â), 2.97 (br s, H5), 3.20-3.30 (m, H1). Anal.
(C8H11NO4‚HCl‚0.5H2O) C, H, N.
Com p ou n d s 5a -c. Compounds 4a -c (520 mg, 1.50 mmol)
and 10% Pd/C (80 mg) were stirred in MeOH (25 mL) under
an atmosphere of H2. After 3 h, the reaction mixture was
filtered and concentrated. Silica gel column chromatography
(MeOH/EtOAc, 1/9) of the crude mixture led to the separation
of the three products 5a -c.
Molecu la r Mod elin g. All molecular modeling studies were
conducted using the QUANTA program,23 and all molecular
mechanics and molecular dynamics simulations were per-
formed using the CHARMM program.24 To access the different
conformations that may be adopted by each of these ligands,
high-temperature molecular dynamics simulations were per-
formed as follows: (1) the temperature of the system was
raised to 1000 from 0 K within 1 ps; (2) the system was then
equilibrated for 30 ps at 1000 K; and (3) the system was finally
simulated for 50 ps. Trajectories were recorded every 0.1 ps
in the final simulation stage and subsequently minimized for
5000 steps, or until convergence, defined as an energy gradient
tolerance of 0.001 kcal mol-1 Å-1, using an adopted-basis
Newton-Raphson algorithm, as implemented in the CHARMM
program. The dielectric constant was set to 80 to simulate
an aqueous environment. The carboxyl groups in these ligands
was set to be deprotonated and the amine groups to be
protonated, as expected under physiological conditions. These
minimized structures were clustered using the Cluster Analy-
(1R,2S,4R,5R)-Dim eth yl 2-a m in obicyclo[2.1.1]h exa n e-
2,5-d ica r boxyla te (5a ): yield 65 mg (20%); Rf 0.6 (EtOAc/
MeOH, 8/1); 1H NMR (CDCl3) δ 1.29 (t, J ) 8.1 Hz, H6R), 1.54
(dd, J ) 2.4 and 11.7 Hz, H3â), 1.78 (br s, NH2), 2.20-2.30 (m,
H
6â), 2.47 (d, J ) 11.7 Hz, H3R), 2.64-2.74 (m, H4), 2.77 (dd, J
) 2.7 and 6.9 Hz, H1), 3.01 (d, J ) 7.8 Hz, H5), 3.73 (s, 3H),
3.74 (s, 3H); 13C NMR (CDCl3) δ 37.6 (CH2), 39.9 (CH2), 41.6
(CH), 51.6 (CH3), 51.7 (CH3), 52.1 (CH), 52.8 (CH), 63.3 (C),
173.7 (CO), 176.1 (CO).
(1S,2S,4S,5R)-Dim eth yl 2-a m in obicyclo[2.1.1]h exa n e-
2,5-d ica r boxyla te (5b): yield 156 mg (49%); Rf 0.49 (EtOAc/
1
MeOH, 8/1); H NMR (methanol-d4) δ 1.54 (dd, J ) 0.9 and
12.0 Hz, H3R), 1.72 (t, J ) 7.8 Hz, H6R), 2.26 (ddd, J ) 2.7, 5.4
and 8.1 Hz, H6â), 2.48-2.56 (m, H3â and H5), 2.62-2.68 (m,
H4), 2.72 (dd, J ) 2.4 and 6.9 Hz, H1), 3.68 (s, 3H), 3.76 (s,

Conformationally Restricted ACPD Analogue
J ournal of Medicinal Chemistry, 1998, Vol. 41, No. 10 1649
sis module in the QUANTA program using all non-hydrogen
atoms as the reference atoms, unless otherwise stated, to
obtain truly distinct low-energy conformations. In each con-
formational cluster, the lowest energy conformation was
selected as the representative conformation for the cluster. The
low-energy, distinct conformations for each ligand were com-
pared to the aa conformation of L-glutamate employing the
Molecular Similarity module in the QUANTA program using
the five carbon atoms and one nitrogen atom in L-glutamate
as the reference atoms. This comparison essentially provides
the overall goodness of overlap between the low energy
conformation of a mGluR ligand and the low energy conforma-
tion of L-glutamate. The root-mean-square (RMS) values were
obtained for each ligand and provide a quantitative measure
of the goodness of the overlap.
was tested in the presence of 1 mM Mg2+, 50 µM kainate, and
1 µM MK-801 to inhibit NMDA receptors, while the effects on
AMPA action were measured in the presence of 1 mM Mg2+
,
30 µM AMPA, 10 µM cyclothiazide, and 1 µM MK-801.
Incubations were terminated by three washes with ice-cold
buffer, and the cells were dissolved in 0.5 M NaOH for
determination of accumulated radioactivity and protein con-
tent.
Ack n ow led gm en t. We are indebted in part to
Pharmacia-Upjohn and the NIH (Grants NS28130 and
NS01720) for their support of this research. We thank
Dr. Francine Acher for a preprint of the paper cited in
ref 7.
Biologica l Assa ys. Cell Cu ltu r es. Lines of Chinese
hamster ovary (CHO) cells with stable expression of mGluR1a,
mGluR2, mGluR5a, mGluR6, and chimeric mGluR3/1a recep-
tors were prepared and cultured as described previously.17 The
baby hamster kidney (BHK) cell line expressing mGluR4a was
a generous gift from Zymogenetics Inc. (Seattle). All cell lines
were cultured in 96-well plates and passaged every 6-7 days.
Primary cultures of neurons were prepared from cerebella of
7-day-old rats and cultured as described previously18 in 96-
well plates. Cerebellar neurons were used for experiments
after 8-9 days in culture.
Refer en ces
(1) Meldrum, B.; Garthwaite, J . Excitatory amino acid neurotoxicity
and neurodegenerative disease. Trends Pharmacol. Sci. 1990,
11, 379-387.
(2) Conn, P. J .; Pin, J .-P. Pharmacology and functions of metabo-
tropic glutamate receptors. Annu. Rev. Pharmacol. Toxicol. 1997,
37, 205-237. In this review, the EC50 values (µM) for glutamate
are reported to be 9-13 at mGluR1a, 3-10 at mGluR5a, 4-20
at mGluR2, 4-5 at mGluR3, 3-20 at mGluR4a, and 16 at
mGluR6. The higher potencies found for glutamate in our system
(Table 1) probably reflect differences in the levels of receptor
expression.
Assa ys of In ositol P h osp h a tes (IP s) Accu m u la tion .
The activity of agonists at phospholipase C-coupled receptors
(mGluR1a, mGluR5a, and mGluR3/1a) was determined by
measurements of their ability to increase the hydrolysis of
membrane phosphoinositides. CHO cells expressing mGluR1,
mGluR5, or mGluR3/1a cultured in 96-well plates were
incubated overnight in glutamine-free culture medium supple-
mented with 0.75 µCi myo-[3H]inositol (Amersham) to label
the cell membrane phosphoinositides. Incubations with ago-
nists were carried out for 30 min at 37 °C in Locke’s buffer
(156 mM NaCl, 5.6 mM KCl, 3.6 mM NaHCO3, 1 mM MgCl2,
1.3 mM CaCl2, 5.6 mM glucose, and 20 mM Hepes, pH 7.4)
containing 20 mM LiCl to block IP degradation. The reaction
was terminated by aspiration, and IPs were extracted with
0.1 M HCl. The separation of [3H]IPs was performed by anion
exchange chromatography as described previously.18 In each
experiment, the stimulations induced by the various com-
pounds were expressed as percent of maximal response
induced by 1 mM glutamate.
(3) Nakanishi, S. Molecular diversity of glutamate receptors and
implications for brain function. Science 1992, 258, 597-603.
(4) Kno¨pfel, T.; Kuhn, R.; Allgeier, H. Metabotropic glutamate
receptors: novel targets for drug development. J . Med. Chem.
1995, 38, 1417-1426. Riedel, G. Function of metabotropic
glutamate receptors in learning and memory. Trends Neurosci.
1996, 19, 219-224.
(5) J oly, C.; Gomeza, J .; Brabet, I.; Curry, K.; Bockaert, J .; Pin, J .-
P. Molecular, functional and pharmacological characterization
of the metabotropic glutamate receptor type 5 splice variants:
comparison with mGluR1. J . Neurosci. 1995, 15, 3970-3981.
(6) Tu¨ckmantel, W.; Kozikowski, A. P.; Wang, S.; Pshenichkin, S.;
Wroblewski, J . T. Synthesis, molecular modeling, and biology
of the 1-benzyl derivative of APDCsan apparent mGluR6
selective ligand. Bioorg. Med. Chem. Lett. 1997, 7, 601-606. The
aa conformation of glutamate is depicted below in which C1 and
C4 are anti, and C2 and C5 are anti.
Assa y of cAMP F or m a tion . The activity of agonists at
receptors coupled negatively to adenylyl cyclase (mGluR2,
mGluR4a, mGluR6) was determined by measurements of their
ability to decrease the forskolin-induced elevation of cyclic
AMP formation as described previously.17 Cells cultured in
96-well culture plates were preincubated 10 min at 37 °C in
Locke’s medium containing 300 µM isobutylmethylxanthine
to inhibit the activity of phosphodiesterases which degrade
cAMP. Then 5 µM forskolin was added without or with mGluR
agonists, and the incubation was continued for 10 min. After
the incubation, the medium was rapidly aspirated and cAMP
was extracted with 0.1 M HCl and measured by radioimmu-
noassay using a magnetic Amerlex RIA kit. In each experi-
ment, the results were expressed as percent of the maximal
response induced by 1 mM glutamate for mGluR2 and
mGluR4, and by 1 mM AP4 for mGluR6.
Deter m in a tion of Ca lciu m In flu x. Activity of ionotropic
glutamate receptors was determined by measurements of
45Ca2+ influx in cerebellar neurons, as described previously.19
Cells cultured in 96-well plates were incubated for 15 min at
room temperature in Locke’s buffer containing 1 µCi/mL of
45CaCl2 and the indicated additions. Agonist activity at all
ionotropic glutamate receptors was measured in Mg2+-free
medium supplemented with 1 µM glycine to allow for activity
at NMDA receptors and with 10 µM cyclothiazide to inhibit
desensitization of AMPA receptors. Antagonist activity at
NMDA receptors was measured in a Mg2+-free medium in the
presence of 100 µM NMDA, 1 µM glycine, and 10 µM NBQX
to inhibit non-NMDA receptors. Antagonism of kainate action
(7) Tellier, F.; Acher, F.; Brabet, I.; Pin, J .-P.; Brockaert, J .; Azerad,
R. Synthesis of conformationally constrained stereospecific
analogues of glutamic acid as antagonists of metabotropic
receptors. Bioorg. Med. Chem. Lett. 1995, 5, 2627-2632, Tellier,
F.; Acher, F.; Brabet, I.; Pin, J .-P.; Azerad, R. Aminobicyclo[2.2.1]-
heptane dicarboxylic acid (ABHD), rigid analogs of ACPD and
glutamic acid: synthesis and pharmacological activity on me-
tabotropic receptors mGluR1 and mGluR2. Bioorg. Med. Chem.,
in press.
(8) Monn, J . A.; Valli, M. J .; Massey, S. M.; Wright, R. A.; Salhoff,
C. R.; J ohnson, B. G.; Howe, T.; Alt, C. A.; Rhodes, G. A.; Robey,
R. L.; Griffey, K. R.; Tizzano, J . P.; Kallman, M. J .; Helton, D.
R.; Schoepp D. D. Design, synthesis, and pharmacological
characterization of (+)-2-aminobicyclo[3.1.0]hexane-2,6-dicar-
boxylic acid (LY354740): a potent, selective, and orally active
group 2 metabotropic glutamate receptor agonist possessing
anticonvulsant and anxiolytic properties. J . Med. Chem. 1997,
40, 528-537. Schoepp, D. D.; J ohnson, B. G.; Wright, R. A.;
Salhoff, C. R.; Mayne, N. G.; Wu, S.; Cockerham,. S. L.; Burnett,
J . P.; Belegaje, R.; Bleakman, D.; Monn, J . A. LY354740 is a
potent and highly selective group II metabotropic glutamate
receptor agonist in cells expressing human glutamate receptors.
Neuropharmacology 1997, 36, 1-11.
(9) Pellicciari, R.; Marinozzi, M.; Natalini, B.; Costantino, G.;
Luneia, R.; Giorgi, G.; Moroni, F.; Thomsen, C. Synthesis and
pharmacological characterization of all sixteen stereoisomers
of 2-(2′-carboxy-3′-phenylcyclopropyl)glycine. Focus on
(2S,1′S,2′S,3′R)-2-(2′-carboxy-3′-phenylcyclopropyl)glycine, a novel
and selective group II metabotropic glutamate receptor antago-
nist. J . Med. Chem. 1996, 39, 2259-2269.
(10) Ornstein, P. L.; Bleisch, T. J .; Arnold, M. B.; Wright, R. A.;
J ohnson, B. G.; Tizzano, J . P.; Helton, D. R.; Kallman, M. J .;
Schoepp, D. D. Novel antagonists for metabotropic glutamate
receptors. Neuropharmacology 1995, 35, A22.

1650 J ournal of Medicinal Chemistry, 1998, Vol. 41, No. 10
Kozikowski et al.
(11) Monn, J . A.; Valli, M. J .; J ohnson, B. G.; Salhoff, C. R.; Wright,
R. A.; Howe, T.; Bond, A.; Lodge, D.; Spangle, L. A.; Paschal, J .
W.; Campbell, J . B.; Griffey, K.; Tizzano, J . P.; Schoepp, D. D.
Synthesis of the four isomers of 4-aminopyrrolidine-2,4-dicar-
boxylate (APDC): identification of a potent, highly selective and
systemically-active agonist for metabotropic glutamate receptors
negatively coupled to adenylate cyclase. J . Med. Chem. 1996,
39, 2990-3000.
(12) Schoepp, D. D.; Goldsworthy, J .; J ohnson, B. G.; Salhoff, C. R.;
Baker, S. R. 3,5-Dihydroxyphenylglycine is a highly selective
agonist for phosphoinositide-linked metabotropic glutamate re-
ceptors in the rat hippocampus. J . Neurochem. 1994, 63, 769-
772.
(13) Acher, F. C.; Tellier, F. J .; Azerad, R.; Brabet, I. N.; Fagni, L.;
Pin J .-P. R. Synthesis and pharmacological characterization of
aminocyclopentanetricarboxylic acids: new tools to discriminate
between metabotropic glutamate receptor subtypes. J . Med.
Chem. 1997, 40, 3119-3129.
(14) Seebach, D.; Aebi, J . D.; Gander-Coquoz, M.; Naef, R. Stereo-
selektive Alkylierung an C(R) von Serin, Glycerinsa¨ure, Threonin
und Weinsa¨ure u¨ber heterocyclische Enolate mit exocyclischer
Doppelbindung. Helv. Chim. Acta 1987, 70, 1194-1216.
(15) Eaton, P. E. Photochemical reactions of simple alicyclic enones.
Acc. Chem. Res. 1968, 1, 50-57.
(16) Stevens, C.; De Kimpe, N. A new entry into 2-azabicyclo[2.1.1]-
hexanes via 3-(chloromethyl)cyclobutanone. J . Org. Chem. 1996,
61, 2174-2178.
(19) Wroblewski, J . T.; Nicoletti, F.; Costa, E. Different coupling of
excitatory amino acid receptors with Ca2+ channels in primary
cultures of cerebellar granule cells. Neuropharmacology 1985,
24, 919-921.
(20) Larue, V.; Gharbi-Benarous, J .; Acher, F.; Valle, G.; Crisma, M.;
Toniolo, C.; Azerad, R.; Girault, J .-P. Conformational analysis
by NMR spectroscopy, molecular dynamics simulation in water
and X-ray crystallography of glutamic acid analogues: isomers
of 1-aminocyclopentane-1,3-dicarboxylic acid. J . Chem. Soc.,
Perkin Trans. 2 1995, 1111-1126.
(21) Costantino, G.; Natalini, B.; Pellicciari, R.; Lombardi, G.; Moroni,
F. Definition of a pharmacophore for metabotropic glutamate
receptors negatively linked to adenylyl cyclase. Bioorg. Med.
Chem. 1993, 1, 259-265; also, see: Shimamoto, K.; Ohfune, Y.
Syntheses and conformational analyses of glutamate analogs:
2-(2-carboxy-3-substituted-cyclopropyl)glycines as useful probes
for excitatory amino acid receptors. J . Med. Chem. 1996, 39,
407-423.
(22) J ohansen, P. A.; Chase, L. A.; Sinor, A. D.; Koerner, J . F.;
J ohnson, R. L.; Robinson, M. B. Type 4a metabotropic glutamate
receptor: identification of new potent agonists and differentia-
tion from the L-(+)-2-amino-4-phosphonobutanoic acid-sensitive
receptor in the lateral perforant pathway in rats. Mol. Phar-
macol. 1995, 48, 140-149.
(23) QUANTA, a molecular modeling system, is supplied by Molec-
ular Simulations, Inc., 200 Fifth Ave, Waltham, MA 01803-
5279.
(24) Brooks, B. R.; Bruccoleri, R. E.; Olafson, B. D.; States, D. J .;
Swaminathan, S.; Karplus, M. CHARMM: A program for
macromolecular energy, minimization, and dynamics calcula-
tions. J . Comput. Chem. 1983, 4, 187-217.
(17) Wroblewska, B.; Wroblewski, J . T.; Pshenichkin, S.; Surin, A.;
Sullivan S.; Neale, J . H. N-Acetylaspartylglutamate selectively
activates mGluR3 receptors in transfected cells. J . Neurochem.
1997, 69, 174-181.
(18) Raulli, R.; Danysz, W.; Wroblewski, J . T. Pretreatment of
cerebellar granule cells with concanavalin
A potentiates
quisqualate-stimulated phosphoinositide hydrolysis. J . Neuro-
chem. 1991, 56, 2116-2124.
J M970719Q