Synthesis, binding affinity at glutamic acid receptors, neuroprotective effects, and molecular modeling investigation of novel dihydroisoxazole amino acids

Article abstract of DOI:10.1021/jm0504499

The four stereoisomers of 5-(2-amino-2-carboxyethyl)-4,5-dihydroisoxazole- 3-carboxylic acid-(+)-4, (-)-4, (+)-5, and (-)-5 were prepared by stereoselective synthesis of two pairs of enantiomers, which were subsequently resolved by enzymatic procedures. These four stereoisomers and the four stereoisomers of the bicyclic analogue 5-amino-4,5,6,6a-tetrahydro-3aH- cyclopenta[d]isoxazole-3,5-dicarboxylic acid (+)-2, (-)-2, (+)-3, and (-)-3 were tested at ionotropic and metabotropic glutamate receptor subtypes. The most potent NMDA receptor antagonists [(+)-2, (-)-4, and (+)-5] showed a significant neuroprotective effect when tested in an oxygen glucose deprivation (OGD) cell culture test. The same compounds were preliminarily assayed using Xenopus oocytes expressing cloned rat NMDA receptors containing the NR1 subunit in combination with either NR2A, NR2B, NR2C, or NR2D subunit. In this assay, all three derivatives showed high antagonist potency with preference for the NR2A and NR2B subtypes, with derivative (-)-4 behaving as the most potent antagonist. The biological data are discussed on the basis of homology models reported in the literature for NMDA receptors and mGluRs.

Full text of DOI:10.1021/jm0504499

J. Med. Chem. 2005, 48, 6315-6325
6315
Synthesis, Binding Affinity at Glutamic Acid Receptors, Neuroprotective
Effects, and Molecular Modeling Investigation of Novel Dihydroisoxazole Amino
Acids†
‡
‡
‡
‡
‡
§,#
Paola Conti, Marco De Amici, Giovanni Grazioso, Gabriella Roda, Andrea Pinto, Kasper Bø Hansen,
§
§
§
#
⊥
Birgitte Nielsen, Ulf Madsen, Hans Br a¨ uner-Osborne, Jan Egebjerg, Valentina Vestri,
Domenico E. Pellegrini-Giampietro, Pauline Sibille, Francine C. Acher, and Carlo De Micheli *
⊥
X
X
‡,
Istituto di Chimica Farmaceutica e Tossicologica, Universit a` degli Studi di Milano, Viale Abruzzi 42, 20131 Milano, Italy;
Department of Medicinal Chemistry, The Danish University of Pharmaceutical Sciences, Universitetsparken 2, DK-2500
Copenhagen, Denmark; Department of Molecular Biology, H. Lundbeck A/S, Valby, Denmark; Dipartimento di Farmacologia
Preclinica e Clinica, Universit a` degli Studi di Firenze, viale G. Pieraccini, 6, 50139 Firenze, Italy; and Laboratoire de Chimie
et Biochimie Pharmacologiques et Toxicologiques, UMR8601-CNRS, Universit e´ Ren e´ Descartes-Paris V, 45 rue des
Saints-P e` res, 75270 Paris Cedex 06, France
Received May 11, 2005
The four stereoisomers of 5-(2-amino-2-carboxyethyl)-4,5-dihydroisoxazole-3-carboxylic acid-
(
+)-4, (-)-4, (+)-5, and (-)-5 were prepared by stereoselective synthesis of two pairs of
enantiomers, which were subsequently resolved by enzymatic procedures. These four stereo-
isomers and the four stereoisomers of the bicyclic analogue 5-amino-4,5,6,6a-tetrahydro-3aH-
cyclopenta[d]isoxazole-3,5-dicarboxylic acid (+)-2, (-)-2, (+)-3, and (-)-3 were tested at
ionotropic and metabotropic glutamate receptor subtypes. The most potent NMDA receptor
antagonists [(+)-2, (-)-4, and (+)-5] showed a significant neuroprotective effect when tested
in an oxygen glucose deprivation (OGD) cell culture test. The same compounds were
preliminarily assayed using Xenopus oocytes expressing cloned rat NMDA receptors containing
the NR1 subunit in combination with either NR2A, NR2B, NR2C, or NR2D subunit. In this
assay, all three derivatives showed high antagonist potency with preference for the NR2A and
NR2B subtypes, with derivative (-)-4 behaving as the most potent antagonist. The biological
data are discussed on the basis of homology models reported in the literature for NMDA
receptors and mGluRs.
Introduction
The iGluRs are associated with integral cation-specific
ion channels and include the N-methyl-D-aspartate
The acidic amino acid glutamate (Glu), the major
neurotransmitter of the fast excitatory synapses in the
central nervous system (CNS), plays a pivotal role in
(
NMDA), 2-amino-3-(5-methyl-3-hydroxyisoxazol-4-yl)-
3
propanoic acid (AMPA), and kainate (KA) types.
Whereas the endogenous ligand for AMPA and KA
receptors is Glu, activation of NMDA receptors require
physiological processes ranging from learning and
memory to control of movements and pain sensitivity.1,2
5
simultaneous binding of both Glu and glycine. Since
Glu induces toxicity (excitotoxicity) by an overflow into
the synaptic cleft which can occur following metabolic
disturbances, and similar excitotoxicity can be caused
by overstimulation of Glu receptors by endogenous or
exogenous substances.1,2 Glu activity is mediated by
ionotropic (iGluRs) and metabotropic Glu receptors
overstimulation of NMDA receptors produces acute
neurodegeneration, much research has been focused on
the development of selective antagonists as neuropro-
tective agents.
In the past decade, the molecular biology of the
NMDA receptor family has been defined, and now we
know that these receptors are tetramers composed of
two NR1 subunits in combination with two NR2 sub-
3
,4
(
mGluRs). The mGluRs exert modulatory effects on
neuronal excitability and synaptic transmission. Up to
now, eight mGluR subtypes and multiple splice variants
have been cloned from mammalian brain. They have
been subdivided into three groups on the basis of their
sequence homology, coupling mechanisms, and phar-
macology. Group I (mGluR1,5) is positively coupled to
phospholipase C (PLC) whereas group II (mGluR2,3)
and group III (mGluR4,6,7,8) are coupled via Gi/o
4
6
7,8
units or, less commonly, NR3 subunits. Since non-
selective NMDA receptor antagonists are characterized
by a number of adverse CNS effects, including hal-
lucinations, a centrally mediated increase in blood
pressure and, at high doses, catatonia and anesthe-
9
,10
sia,
current interest largely centers on the develop-
4
proteins to adenylyl cyclase inhibition.
ment of potent and NR2B subtype-selective antago-
nists,1
1,12
e.g. Ifenprodil 1 (Figure 1). In preclinical tests
†
Dedicated to the memory of Professor Giorgio Bianchi.
these compounds appear to lack the side-effects com-
monly associated with nonselective NMDA receptor
*
To whom correspondence should be addressed. Phone:
+
39 02 50317538. Fax: +39 02 50317574. E-mail: carlo.demicheli@
unimi.it.
13
antagonists.
‡
Istituto di Chimica Farmaceutica.
In a recent paper,14 we reported the synthesis and
the pharmacological characterization of the racemic
form of the two acidic bicyclic amino acids, 5-amino-
4,5,6,6a-tetrahydro-3aH-cyclopenta[d]isoxazole-3,5-di-
§
#
⊥
Department of Medicinal Chemistry.
Department of Molecular Biology.
Dipartimento di Farmacologia Preclinica e Clinica.
Laboratoire de Chimie et Biochimie Pharmacologiques et Toxi-
X
cologiques.
1
0.1021/jm0504499 CCC: $30.25 © 2005 American Chemical Society
Published on Web 09/13/2005

6
316 Journal of Medicinal Chemistry, 2005, Vol. 48, No. 20
Conti et al.
Figure 1. Structures of ifenprodil (1) and target compounds.
carboxylic acids (()-2 and (()-3 (Figure 1). Derivative
Table 1. Enzymatic Hydrolyses Carried out on Diester (()-6
a
ee of
degree of conversion
(%)
(
()-2 behaved as a quite potent NMDA receptor an-
tagonist supported by antagonist activity at mGluR1,5
group I) and agonist activity at mGluR2. This phar-
enzyme
(+)-(5R,2′S)-8
E valueb
R- chymotrypsin
CAL-A
99
30
)
300
)
(
c
n.r.
macological profile candidates (()-2 as a new lead for
neuroprotective agents. The four stereoisomers (+)-2,
papain
99
30
)
300
)
lipase PS
proleather
CAL-B
n.r.
g 99
n.r.
95.6
n.r.
n.r.
(
-)-2, (+)-3, and (-)-3 were subsequently prepared and
50
)
>1000
)
105
)
)
the absolute configuration assigned by vibrational cir-
1
5
acylase
50
)
cular dichroism analyses. The structure-activity re-
lationship of (()-2 was extended through the synthesis
of monocyclic analogues, 5-(2-amino-2-carboxyethyl)-4,
PPL
lipase AK
)
a
Enantiomeric excess. b E: enantiomeric ratio, calculated ac-
5
-dihydroisoxazole-3-carboxylic acids (()-4 and (()-5,
c
cording to the equations reported in ref 19. n.r.: no reaction.
and evaluation of their biological activity. Both race-
mates were endowed with a remarkable NMDA antago-
nist activity devoid of any activity at mGluRs.16
Table 2. Enzymatic Hydrolyses Carried out on Diester (()-7
a
ee of
degree of conversion
(%)
This paper reports the synthesis of the four stereo-
isomers (+)-4, (-)-4, (+)-5, and (-)-5 and the results of
pharmacological investigations of the eight enantiomers
enzyme
(-)-(5S,2′S)-9
E valueb
R-chymotrypsin
CAL-A
n.r.c
n.r.
g 99
n.r.
g 99
n.r.
g 99
n.r.
n.r.
)
)
)
)
(+)-2, (-)-2, (+)-3, (-)-3, (+)-4, (-)-4, (+)-5, and (-)-5.
papain
50
)
>1000
)
The pharmacology at iGluRs and mGluRs was studied
by in vitro binding to rat cortical membranes and by
measurement of functional activity in second messenger
assays at cloned receptors expressed in Chinese hamster
ovary cells, respectively. The neuroprotective effects of
the most promising derivatives were then evaluated in
an in vitro model of cerebral ischemia on murine cortical
cells exposed to oxygen and glucose deprivation (OGD).
Furthermore, compounds (+)-2, (-)-4, and (+)-5, being
NMDA receptor antagonists, were assayed using Xeno-
pus oocytes expressing cloned rat NMDA receptors
containing NR1 in combination with either NR2A,
NR2B, NR2C, or NR2D, to investigate subunit selectiv-
ity. The overall results of the pharmacological investiga-
tions have been accounted for by docking the compounds
lipase PS
proleather
CAL-B
50
)
>1000
)
acylase
50
)
>1000
)
PPL
lipase AK
)
)
a
Enantiomeric excess. b E: enantiomeric ratio, calculated ac-
c
cording to the equations reported in ref 19. n.r.: no reaction.
efficiently hydrolyze the 2′ methyl ester of enantiomer
(+)-6 to the corresponding monoacid (+)-8 (Scheme 1).
It is worth pointing out that the reaction spontaneously
subsided at 50% conversion and was characterized by
a high degree of enantio-discrimination (E > 1000).19
As shown in Table 2, the same set of reactions per-
formed on derivative (()-7 demonstrated that Papain,
Proleather, and Acylase were equally efficient to hy-
drolyze enantiomer (-)-7 to monoacid (-)-9 (Scheme 1).
Since Papain, among these three enzymes, is less
expensive, we used this enzyme in the preparation of
17
in the binding site of an NR2A antagonist model as
1
8
well as in an mGluR2 agonist model.
Chemistry
The key step in the synthesis of target compounds
(-)-5 and (+)-5.
(
+)-4/(-)-4 and (+)-5/(-)-5 is the enzymatic resolution
of the previously reported isoxazoline derivatives (()-6
Intermediates (-)-6, (+)-8, (+)-7, and (-)-9 were
1
6
and (()-7. As shown in Tables 1 and 2, derivatives
()-6 and (()-7 were hydrolyzed in the presence of a
converted into target amino acids (-)-4, (+)-4, (+)-5, and
(-)-5, respectively, through standard transformations
(Scheme 1). Chiral HPLC analyses carried out on final
amino acids showed that all the compounds were
provided with a value of enantiomeric excess higher
(
number of commercially available enzymes. Among the
tested biocatalysts, Proleather turned out to be the most
selective enzyme for substrate (()-6 (Table 1), able to

Novel Dihydroisoxazole Amino Acids
Journal of Medicinal Chemistry, 2005, Vol. 48, No. 20 6317
Scheme 1a
a
(
a) Proleather, 0.1 M phosphate buffer/acetone; (b) Papain, 0.1 M phosphate buffer/acetone; (c) 1 N NaOH/H2O-EtOH; (d) 30%
CF3CO2H/CH2Cl2; (e) NaHCO3/AcOEt.
Table 3. Rat Cortical Membrane Receptor Binding and in Vitro Electrophysiological Data from the Rat Cortical Wedge Preparationa
receptor binding
electrophysiology
[3
H]CGP 39653
Ki (µM)
[3H]AMPA
IC50 (µM)
[3H]KA
IC50 (µM)
(rat cortical wedge):
b
compd
antagonism, IC50 (µM)
(
(
(
(
(
(
(
(
(
(
()-2c
+)-2
-)-2c
()-3
()-4d
+)-4
-)-4
()-5d
+)-5
-)-5
0.37 [0.34; 0.40]
0.233 [0.230; 0.237]
47 [43; 51]
>100
>100
>100
>100
>100
>100
>100
>100
>100
>100
>100
>100
>100
>100
>100
>100
>100
>100
>100
>100
14 ( 2
>100
0.21 [0.18; 0.23]
1.9 [1.8; 2.0]
0.10 [0.09; 0.12]
0.96 [0.88; 1.1]
0.51 [0.49; 0.53]
2.5 [2.4;2.7]
14.2 [12.9;15.7]
30.5 [24.9;37.4]
a
Values are expressed as the antilog to the log mean of three individual experiments. Numbers in parentheses [min; max] indicate (
b
c
d
SEM according to a logarithmic distribution. Antagonism of NMDA (10 µM) induced responses. Data from ref 14. Data from ref 16.
than 99%. Since the relative stereochemistry of amino
Results and Discussion
acids (()-4 and (()-5 was previously assigned by X-ray
analysis, the absolute configuration of the stereogenic
centers of the pairs of amino acids (-)-4/(+)-4 and
The three pairs of acidic amino acids (+)-2/(-)-2,
16
(
+)-4/(-)-4 and (+)-5/(-)-5 were assayed in vitro by
means of receptor binding to rat cortical membranes
(
+)-5/(-)-5 was assigned by chemical correlation
with enantiopure dipolarophile (+)-10. To this end,
,3-dipolar cycloaddition of (S)-(+)-N-BOC allylglycine
3
3
using the radioligands [ H]CGP 39653, [ H]AMPA and
3
[
H]KA as representatives for NMDA, AMPA and KA
1
22-24
receptors, respectively
(Table 3). The enantiomers
2
0
methyl ester [(+)-10] to ethoxycarbonylformonitrile
(
+)-3 and (-)-3 were not tested in these binding assays
oxide, generated in situ by treatment of ethyl 2-chloro-
due to the lack of affinity of the racemate (()-3. The
pairs of enantiomers (+)-2/(-)-2 and (+)-3/(-)-3, whose
racemates displayed activity at mGluRs,14 were assayed
at rat mGluR1 (representative of group I), at mGluR2
2
1
2
-(hydroxyimino)acetate with a base (Scheme 1),
yielded a 2:3 mixture of the stereomeric cycloadducts
+)-(5R,2′S)-6 and (-)-(5S,2′S)-7.
(

6
318 Journal of Medicinal Chemistry, 2005, Vol. 48, No. 20
Conti et al.
Table 4. Activities at Cloned Rat mGlu Receptors Expressed
in CHO Cellsa
mGluR1
mGluR2
mGluR4
i
(µM) EC50 (µM) K (µM)
compd EC50 (µM)
K
i
(µM) EC50 (µM)
K
i
(
(
(
(
(
(
(
(
()-2b
+)-2
-)-2
94 ( 16
c
>1000
>1000
>1000
>1000
weak
>1000
>1000
>1000
>1000
>1000
>1000
>1000
>1000
>1000 >1000
95 ( 13 weak
27 ( 7 16 ( 3
33 ( 6 9.1 ( 0.8
>1000 >1000
>1000 >1000
>1000 >1000
>1000
b
d
()-3
e
+)-3
-)-3
>1000
>1000
>1000
>1000
>1000
>1000
>1000
>1000
>1000
f
()-4
f
()-5
a
Data are given as mean ( SEM of at least three independent
b
c
experiments. Data from ref 14. 1 mM concentration produces
a 36 ( 7% activation. d Max. ) 59(7%. Max. ) 42(4%. Data
e
f
from ref 16.
(
representative of group II) and at mGluR4 (representa-
tive of group III); all the receptors were expressed in
CHO cells (Table 4). Due to inactivity of the race-
mates, the enantiomers (+)-4, (-)-4, (+)-5, and (-)-5
were not tested on mGluRs.
2
5
1
6
As shown in Table 3, the NMDA receptor activity,
previously observed for racemates (()-2, (()-4, and
1
4,16
(
(
()-5,
can be attributed to the enantiomers (+)-2,
-)-4, and (+)-5, respectively. Similarly, the mGluR
activity of derivatives (()-2, and (()-3 (Table 4),
14
resides exclusively in the enantiomers (-)-2 and (+)-3,
respectively. As a consequence, the mixed activity at
mGluRs and NMDA receptors of racemate (()-2 is the
overlap of the NMDA antagonist activity of its enanti-
omer (+)-2 with the mGluR activity of enantiomer
Figure 2. Derivatives (+)-2, (-)-4, and (+)-5 were tested at
concentrations of 0.3 µM and 3 µM for inhibition of currents
evoked by 10 µM Glu on Xenopus oocytes expressing cloned
rat NMDA receptors containing NR1 in combination with
either NR2A, NR2B, NR2C, or NR2D. The responses are
normalized to the response from 10 µM Glu.
(
-)-2.
To evaluate the subtype selectivity of derivatives
+)-2, (-)-4, and (+)-5 on cloned rat NMDA receptors,
(
they were tested on Xenopus oocytes expressing NR1
in combination with either NR2A, NR2B, NR2C, or
NR2D subunits. The antagonists were tested at con-
centrations of 0.3 µM and 3 µM for inhibition of currents
evoked by 10 µM Glu (Figure 2). At concentrations of 3
µM, all three derivatives antagonized responses from
NR2A and NR2B subtypes, and they showed no antago-
nistic effects on NR2C and NR2D subtypes. This shows
that the three antagonists have a preference for NR2A
and NR2B subtypes, as the potency of Glu at the four
receptor subtypes is similar (Glu EC50-values of
2
6
26
26
NR1/NR2A, NR1/NR2B, NR1/NR2C, and NR1/
NR2D are 2.9 µM, 1.8 µM, 1.5 µM, and 0.5 µM,
respectively, data not shown). Derivative (-)-4 seems
to be more potent than (+)-2 and (+)-5 since only this
compound antagonizes NR2A at concentrations of
Figure 3. Neuroprotection by derivatives (()-2 and (()-3
against OGD-induced neurotoxicity in murine cortical cultures.
Data are expressed as percent of OGD-induced neuronal
damage. Each point represents the mean ( SEM of three
experiments. *, p < 0.05 and **, p < 0.01 versus OGD (ANOVA
0
.3 µM. The results of this investigation show that
(
+)-2, (-)-4, and (+)-5 share similar NMDA receptor
subtype selectivity preferring NR2A and NR2B over
+
Tukey’s w test).
NR2C and NR2D (and that the order of potency [(-)-4
>
order of affinity obtained from binding to rat cortical
membranes (Table 3).
(+)-2 ) (+)-5)] is similar though not identical to the
medium, which was approximately 75% of that observed
by exposing the cultures to 1 mM Glu (data not
27
shown).
The neuroprotective effects of selected derivatives
were evaluated in cultured murine cortical cells exposed
to OGD, an in vitro model of cerebral ischemia (Figures
None of the tested compounds produced any signifi-
cant change in the release of LDH when added alone
into the incubation medium (not shown). When added
to the incubation medium during OGD and the subse-
quent 24 h recovery period, compound (()-2, but not
compound (()-3, dose-dependently attenuated OGD-
3
and 4). Exposure to OGD for 60 min produced an
intermediate level of neuronal damage, measured by the
release of lactate dehydrogenase (LDH) into the bathing

Novel Dihydroisoxazole Amino Acids
Journal of Medicinal Chemistry, 2005, Vol. 48, No. 20 6319
-)-4, and (+)-5. Worth noting is that the neuroprotec-
(
tive effect can be evidenced solely for derivatives
provided with high affinity for the NMDA receptor
binding site. Since a neuroprotective effect mediated by
mGluRs has been reported, the inefficacy of derivatives
(
-)-2 and (+)-3 may be attributed to their low activity
at mGluR1,5 coupled to a marginal activity at mGluR2.
To get insights into the relationship between stereo-
chemistry and biological activity, we performed docking
studies on the NMDA antagonists (+)-2, (-)-4, and (+)-5
into the recently proposed homology model of the
NR2A.17 Computational studies essentially confirm the
results reported for the 2-amino-5-phosphonopentanoic
acid (AP5-)-like NMDA-selective antagonists. Since the
amino acidic moieties of (+)-2, (-)-4, and (+)-5 adopt
the same binding mode, they were docked in accordance
to the mutational analysis reported for the NR2A
subunit.2 The flexible docking calculations reveal that
the carboxylic group interacts through a salt bridge with
the side chain of R518 and, in addition, a hydrogen bond
involving T513 stabilizes the ligand docking into the
pocket (Figure 5). At the same time, the amino group
gives rise to supplementary links with the backbone of
S511 residue and the side chains of amino acids T513,
H485, and D731. In our previous study,17 we assumed
that a crucial salt bridge interaction into the binding
site should involve the ligand distal carboxylic group
and residues S689-T690. This region of the receptor
protein constitutes the amino terminal portion of an
R-helix, and, as has been shown,30 it generates a
positively charged macro dipole which attracts nega-
tively charged moieties of the ligands (Figure 5). As a
consequence, the distal acidic group of the isoxazoline
ring of amino acids (+)-2, (-)-4, and (+)-5 is suitably
oriented to interact with the backbone NH groups of the
protein residues S689-T690. On the other hand, the
rigidity of enantiomer (-)-2 prevents a productive
interaction between the ω-CO2H group and the amino
terminal portion of the above-described R-helix.
The rank order in the flexible docking of the antago-
nists under investigation is in good agreement with the
rank order of their experimental binding affinities.
Moreover, going from (R)-AP5 to 2-amino-4-oxo-5-
phosphono-pentanoic acid [MDL 100,453] (Figure 6),
both NMDA selective antagonists, a doubling of the
binding affinity was reported,2 which parallels the
results observed for derivatives (+)-2 and (-)-4. This
outcome may be rationalized by admitting an extra
hydrogen bond between the MDL 100,453 carbonyl
group and the side chain of the Y730 residue. Indeed,
docking studies on compound (-)-4 indicate the same
additional hydrogen bond. In this case, the electron
donor oxygen atom of the isoxazoline ring of (-)-4 is
capable to interact with the OH group of Y730, mimick-
ing the carbonyl group of MDL 100,453. Due to the
opposite absolute configuration at position 5, stereoiso-
mer (+)-5 has a weaker interaction with Y730, ex-
pressed by the lower affinity, compared to (-)-4, which
parallels the low score obtained in the docking calcula-
tions.
8,29
Figure 4. Neuroprotective effects of (A) enantiomers (+)-2
and (-)-2 and (B) enantiomers (+)-4, (-)-4, (+)-5, and (-)-5
against OGD-induced neurotoxicity in murine cortical cultures.
Data are expressed as percent of OGD-induced neuronal
damage. Each bar represents the mean ( SEM of three
experiments. *, p < 0.05 and **, p < 0.01 versus OGD (ANOVA
+
Tukey’s w test).
induced neuronal death (Figure 3). The maximal effect
was observed at 300-1000 µM. When (+)-2 and (-)-2
were tested, enantiomer (+)-2 displayed a greater
neuroprotective effect than (-)-2 (Figure 4A), which is
compatible with the fact that (+)-2 has high affinity for
NMDA receptors whereas (-)-2 has weak affinity for
NMDA receptors and weak antagonist activity at
mGluR1 receptors. Similarly, among (()-4 and (()-5
enantiomers, the compounds displaying the most potent
activity as NMDA receptor antagonists, (-)-4 and
(
+)-5, were those that provided the most pronounced
neuroprotective effect (Figure 4B).
An overview of the results reported in Tables 3 and 4
displays the remarkable enantiopharmacology of the
compounds under study. Indeed, the activity displayed
by (()-2 at both NMDA and mGluRs is a combination
of a high affinity of enantiomer (+)-2 for NMDA recep-
tors with a weak activity of its antipode (-)-2 at both
mGluRs and NMDA receptors. In the remaining cases,
the biological activity of the racemates resides solely in
one enantiomer, i.e., the mGluRs activity of (()-3, is due
to the (+)-3 enantiomer, whereas enantiomers (-)-4 and
(
+)-5 carry the NMDA antagonist activity of the corre-
Furthermore, the enantiomers (+)-2, (-)-2, (+)-3, and
(-)-3, which are agonists or partial agonists at mGluR2
(Table 4), were analyzed using molecular modeling
tools.18 Among the four pairs of tested enantiomers, only
sponding racemates (()-4 and (()-5. These data are
further confirmed in the neuroprotective test (Figures
3
and 4), where the eutomers turned out to be (+)-2,

6
320 Journal of Medicinal Chemistry, 2005, Vol. 48, No. 20
Conti et al.
Figure 5. Docking of the new NMDA selective antagonists into the model of the NR2A subunit. (A) (+)-2; (B) (-)-4; (C) (+)-5.
The ligands are displayed as ball-and-stick models whereas amino acids involved in the ligand binding are in capped sticks color
code.
enantiomers (+)-2 and (-)-3 occupy a region where
substituents are poorly tolerated as defined in the
mGluR2 pharmacophore model31 and illustrated with
(-)-3 and (1R,3S)-ACPD in Figure 7. Distances between
amino and carboxylic functions at position 5 (proximal
functions) and carboxylate at position 3 (distal function)
are larger for 2 and 3 than those tolerated at the
mGluR1 binding site; this may explain their antagonist
properties. Indeed, a similar mGluR selectivity was
noted for phenylglycines such as (S)-4-carboxyphenyl-
glycine [(S)-4CPG] (Figure 6), which exhibits similar
functional group distances, agonist activity at mGluR2,
Figure 6. Structure of selective NMDA receptor antagonists
top) and selective mGlu2R agonists (bottom).
(
18
and antagonist activity at mGluR1. However, although
functional groups of 2 and 3 are well positioned for
mGluR2 activation, variable potencies have been de-
tected. These differences may result from steric restric-
tions at the binding site. In Figure 7A, the heterocycle
of (-)-2 lines a forbidden region situated between the
two acidic features of the model31 and may be respon-
sible for the weak agonist activity.
Observations on pharmacophore models can be cor-
related with those on homology models docked with the
agonists. The (-)-2 and (+)-3 isomers were docked in
the mGluR2 homology model that was previously gener-
ated.18 Functional groups of agonists were superimposed
to those of Glu in the 3D-homology model, and the
resulting system was minimized. The amino terminal
domain (ATD) of mGluRs folds in two lobes linked by a
hinge (Figure 8A). The agonists bind to lobe-I and are
then trapped upon closing of the second lobe.32 The
two displayed agonist activity at one subtype of mGluRs
mGluR2). Conformational analysis by high-tempera-
(
3
1
ture molecular dynamics showed that (-)-2 and (+)-3
adopt one stable conformation while (+)-4/(-)-4 and
(-)-5/(+)-5 are more flexible and assume conformations
different from that of (-)-2 and (+)-3. Worth mention-
ing, enantiomers (-)-2 and (+)-3 adopt an extended
conformation that fits the agonist pharmacophore model
31
that was previously generated. Superimposition of the
enantiomers to the mGluR2 agonist pharmacophore
model reveals that (-)-2 and (+)-3 are expected to be
the active enantiomers, since their cyclopentane rings
are located in the same region as those of known
agonists as (1S,3R)-1-amino-1,3-cyclopentanedicarboxy-
lic acid [(1S,3R)-ACPD] and 2-aminobicyclo[3.1.0]hexane-
2,6-dicarboxylic acid [LY354740] (Figures 6 and 7). The

Novel Dihydroisoxazole Amino Acids
Journal of Medicinal Chemistry, 2005, Vol. 48, No. 20 6321
Figure 7. (A) Side view of mGlu2R pharmacophore model from ref.31 with (-)-2 (pink) and (+)-3 (purple) superimposed. Potent
agonists [LY354740 and (1S,3R)-ACPD] are colored in green, moderate activity agonist [(1R,3S)-ACPD)] in orange, and inactive
analogue [(-)-3] in red. van der Waals spheres of oxygen and nitrogen atoms of proximal and distal functions are displayed as
red and blue dotted spheres, respectively, marking the ionizable features of the pharmacophore model (R2, R3, and R1). (B) Top
view of part A.
Figure 8. (A) 3D-Homology model of the ATD of mGlu2R docked with (+)-3.18 Lobe-I is colored in blue, lobe-II in magenta, and
the hinge in orange. Color code for the ligand: carbons in green, oxygens in red, nitrogens in blue. (B) Expanded view of part A,
with the same color code. (C) Expanded view of the docking of (-)-2 in the mGlu2R homology 3D-model; color code as in part A.
(
D) Superimposition of (-)-2, (+)-3, and LY354740 (yellow) docked in the mGlu2R binding site. Steric interactions are indicated
with white arrows; color code as in part A.
closed conformation of the ATD induces receptor activa-
tion. In Figure 8A, the mGluR2 ATD docked with the
most potent agonist (+)-3 is displayed. An expanded
view of the binding site bound with (+)-3 and (-)-2 is
shown in Figure 8, parts B and C. For comparison,
docking of LY354740 is displayed in Figure 8D. In the
two models, the proximal functions of (+)-3 and (-)-2
are bound to the conserved residues T168, S145, CO of

6
322 Journal of Medicinal Chemistry, 2005, Vol. 48, No. 20
Conti et al.
A166 (lobe-I), and Y216, D295 (lobe-II) as described for
all mGluR agonists.18,33 The distal functions bind to the
same cluster of basic (R57, R61, K377) and hydrophilic
acylase were purchased from Fluka. Proleather (Subtilysin
Carlsberg) and lipase AK were obtained from Amano Phar-
maceuticals Co.
1H NMR and 13C NMR spectra were recorded with a Varian
(S143) residues from lobe-I, as for other potent mGluR2
3 2
Mercury 300 (300 MHz) spectrometer in CDCl or D O solution
agonists. No distal bindings to lobe-II are detected;
however, water molecules can bridge the ligand to the
backbone or side chains of lobe II.1 The side chain of
R271 is also making van der Waals contacts with the
bicyclic structures of (-)-2 and (+)-3 as for LY354740.
Compound (+)-3 is a partial agonist, and this property
may be due to the steric bulk of the heterocycle, which
can come into steric interaction with G296 of lobe-II,
upon closing of the bilobate ATD (Figure 8D). We have
previously shown that such an interaction may be the
reason for antagonist properties of some ligands.34 In
the case of (+)-3, optimal closing is not reached;
however, the steric hindrance is not as drastic as with
R-alkylated ligands. In regards to (-)-2, its weak agonist
activity may be also related to the steric bulk of its
heterocycle. Yet because of a different configuration at
the bicyclic junction, this cycle is facing lobe-I. Steric
hindrance with side chains and backbone of lobe-I
residues may result in poor binding to lobe-I (Figure
at 20 °C. Chemical shifts (δ) are expressed in ppm and coupling
constants (J) in hertz. HPLC analyses were performed with a
Jasco PU-980 pump equipped with a UV-vis detector Jasco
UV-975 and with a Sedex 750 light scattering detector. Rotary
power determinations were carried out with a Perkin-Elmer
8
241 polarimeter coupled with a Haake N3-B thermostat. TLC
analyses were performed on commercial silica gel 60 F254
aluminum sheets; spots were further evidenced by spraying
with a dilute alkaline potassium permanganate solution or
with ninhydrin. Melting points were determined on a model
B 540 B u¨ chi apparatus and are uncorrected. Microanalyses
(C, H, N) of new compounds agreed with the theoretical value
within (0.3%.
Ethyl (5R*,2′S*)-5-(2-tert-Butoxycarbonylamino-2-meth-
oxycarbonylethyl)-4,5-dihydroisoxazole-3-carboxylate
[
(()-6] and Ethyl (5R*,2′R*)-5-(2-tert-Butoxycarbonyl-
amino-2-methoxycarbonylethyl)-4,5-dihydroisoxazole-3-
carboxylate [(()-7]. Cycloadducts (()-6 and (()-7 were
prepared according to the procedure previously reported.
1
6
HPLC retention times of the pairs of enantiomers are the
following. (-)-6: 17.36 min; (+)-6 19.88 min (column: CHIRAL-
CEL OD-H, Diacel; eluent: petroleum ether/2-propanol 97:3;
flow rate: 1.0 mL/min; λ ) 254 nm). (-)-7: 30.23 min; (+)-7
8
D) and consequently weaker potency.
In conclusion, we have investigated the pharmacology
3
3.67 min (column: CHIRALCEL OD-H, Diacel; eluent: pe-
of four pairs of acidic amino acids (+)-2/(-)-2, (+)-3/
-)-3, (+)-4/(-)-4, and (+)-5/(-)-5 and showed that the
troleum ether/2-propanol 95:5; flow rate: 1.0 mL/min; λ )
(
254 nm).
pharmacological activity is strictly related to their
absolute stereochemistry. In fact, in the case of com-
pounds (()-3, (()-4, and (()-5, only one enantiomer
possess biological activity, and, even more interestingly,
in the case of (()-2, the two enantiomers display
selectivity for the NMDA receptors and for the mGlu
receptors, respectively.
The results were rationalized by means of molecular
modeling studies. By using a homology model of the
NR2A subunit of the NMDA receptors, we have found
that the rank order in the flexible docking of the
antagonists (+)-2, (-)-4, and (+)-5 is comparable to the
rank order of their experimental binding affinities. The
higher affinity displayed by (-)-4, compared to the other
two derivatives, can be explained by an extra hydrogen
bond between the electron donor oxygen atom of the
isoxazoline ring and the OH group of Y730.
General Procedure for the Analytical Enzymatic Hy-
drolyses of Diesters (()-6 and (()-7. To a suspension of
(()-6 [or (()-7] (5 mg, 0.015 mmol) in 0.1 M phosphate buffer
pH 7.0 (1 mL) and acetone (200 µL) was added the enzyme
5 mg), and the mixture was stirred at room temperature. The
outcome of the reaction was monitored by HPLC; every hour,
0 µL of the mixture were withdrawn, made acidic with 2 N
(
2
HCl (10 µL), and extracted with ethyl acetate (20 µL). The
organic phase was analyzed by chiral HPLC under the protocol
described above.
Preparative Enzymatic Hydrolysis of Substrate
()-6. To a suspension of (()-6 (750 mg, 2.18 mmol) in 0.1 M
(
phosphate buffer pH 7.0 (8 mL) and acetone (2 mL) was added
Proleather (1.0 g), and the mixture was gently stirred at room
temperature for 2 h. The progress of the reaction was moni-
tored by chiral HPLC (column: CHIRALCEL OD-H, Diacel;
eluent: petroleum ether/2-propanol 97:3; flow rate: 1.0 mL/
min; λ ) 254 nm). The reaction spontaneously stopped at 50%
conversion to yield monoacid (+)-8 and residual ester (-)-6 in
3
e.e. higher than 99%. Solid NaHCO was added, and then the
On the other hand, the analysis of the mGluR2
agonist pharmacophore model shows that (-)-2 and
reaction mixture was extracted with diethyl ether and the
pooled organic layers were evaporated to recover the residual
ester (-)-6 (345 mg). The aqueous phase was made acidic with
(+)-3, which are weak or partial agonists, respectively,
2
N HCl, and the monoacid (+)-8 (315 mg) was obtained after
locate their cyclopentane rings in the same region as
those of known agonists as (1S,3R)-ACPD and LY354740,
while the inactivity of their enantiomers (+)-2 and (-)-3
can be explained by the fact that they occupy a region
where substituents are poorly tolerated. Moreover, the
weak agonist activity of (-)-2, as well as the partial
agonism of (+)-3, seem to be due to the steric bulk of
their heterocyclic moieties. These results correlate well
with those obtained by docking the compounds in the
homology model of the mGluR2.
extraction with diethyl ether, drying over anhydrous Na
and evaporation of the solvent.
2 4
SO ,
Ethyl (5S,2′R)-5-(2-tert-Butoxycarbonylamino-2-meth-
oxycarbonylethyl)-4,5-dihydroisoxazole-3-carboxylate
20
(
D
-)-6: [R] : -71.2 (c 1.0, chloroform); colorless oil; Anal.
(C H N O ) C, H, N. Spectroscopic data agreed with those
1
5
24
2
7
1
6
reported for the corresponding racemate.
Ethyl (5R,2′S)-5-(2-tert-Butoxycarbonylamino-2-car-
boxyethyl)-4,5-dihydroisoxazole-3-carboxylate
(+)-8:
20
[R]
D
: +54.8 (c 1.0, methanol); colorless oil; R
f
0.62 (chloroform/
methanol 9:1 + 50 µl acetic acid). HPLC retention time:
2
4.77 min (column: CHIRALCEL OD-H, Diacel; eluent: pe-
Experimental Section
troleum ether/2-propanol 97:3; flow rate: 1.0 mL/min; λ )
1
Material and Methods. (S)-(-)-2-Amino-4-pentenoic acid
was purchased from Fluka. Lipase B from Candida antarctica
3
254 nm). H NMR (CDCl ): 1.36 (t, J ) 7.3, 3H), 1.43 (s, 9H),
2.20 (m, 2H), 2.94 (dd, J ) 7.3, 17.5, 1H), 3.38 (dd, J ) 11.3,
17.5, 1H), 4.32 (q, J ) 7.3, 3H), 4.34 (m, 1H), 4.98 (m, 1H),
(
CALB) was bought from Roche Diagnostics (CHIRAZYME
L-2). Chymotrypsin, papain from Carica papaya, porcine
pancreatic lipase (PPL), lipase from Pseudomonas cepacia
22 2 7
5.38 (bd, J ) 6.2, 1H); Anal. (C14H N O ) C, H, N.
Preparative Enzymatic Hydrolysis of Substrate (()-
7. To a suspension of (()-7 (1.0 g, 2.90 mmol) in 0.1 M
(
lipase PS), lipase A from Candida antarctica (CALA), and

Novel Dihydroisoxazole Amino Acids
Journal of Medicinal Chemistry, 2005, Vol. 48, No. 20 6323
phosphate buffer pH 7.0 (50 mL) and acetone (10 mL) was
added papain (500 mg), and the mixture was gently stirred at
room temperature for 3 days. The progress of the reaction was
monitored by HPLC (column: CHIRALCEL OD-H, Diacel;
eluent: petroleum ether/2-propanol 95:5; flow rate: 1.0 mL/
min; λ ) 254 nm). When the enzymatic hydrolysis reached 50%
HPLC retention time: 10.47 min (column: CHIROBIOTIC
TAG, Astec; eluent: [ethanol/methanol 3:2]/water 4:1 + 50 mM
ammonium acetate + 0.1% acetic acid; flow rate: 1 mL/min;
light scattering detection); ee g 99%; Anal. (C
7 10 2 5
H N O ) C, H,
N. Spectroscopic data agreed with those reported for the
1
6
corresponding racemate.
5R,2′R)-5-(2-Amino-2-carboxy-ethyl)-4,5-dihydro-isox-
azole-3-carboxylic Acid (+)-5. Diester (+)-7 (445 mg,
.29 mmol) was transformed into final amino acid (+)-5
overall yield: 94%) following the procedure described above
for compound (-)-4. (+)-5: white prisms; decomposes in the
3
conversion, solid NaHCO was added, reaction mixture was
(
extracted with diethyl ether, and the pooled organic layers
were evaporated to recover the residual ester (+)-7 (445 mg).
The aqueous phase was made acidic with 2 N HCl, and the
monoacid (-)-9 (400 mg) was obtained after extraction with
1
(
2 4
diethyl ether, drying over anhydrous Na SO , and evaporation
of the solvent.
20
range 155-158 °C. [R]
D
: +68.0 (c 0.1, water/methanol 1:1);
HPLC retention time: 19.00 min (column: CHIROBIOTIC
TAG, Astec; eluent: [ethanol/methanol 3:2]/water 4:1 + 50 mM
ammonium acetate + 0.1% acetic acid; flow rate: 1 mL/min;
Ethyl (5R,2′R)-5-(2-tert-Butoxycarbonylamino-2-meth-
oxycarbonylethyl)-4,5-dihydroisoxazole-3-carboxylate
2
0
(
+)-7: [R]
D
: +79.0 (c 1.0, chloroform); colorless oil; Anal.
7 10 2 5
light scattering detection); ee g 99%; Anal. (C H N O ) C, H,
(
C
15
H
24
N
2
O
7
) C, H, N. Spectroscopic data agreed with those
N. Spectroscopic data agreed with those reported for the
1
6
16
reported for the corresponding racemate.
corresponding racemate.
Ethyl (5S,2′S)-5-(2-tert-Butoxycarbonylamino-2-car-
Methyl (S)-(+)-2-tert-Butoxycarbonylamino-4-pente-
noate (+)-10. Dipolarophile (+)-10 was prepared from com-
mercially available (S)-(-)-2-amino-4-pentenoic acid following
20
boxyethyl)-4,5-dihydroisoxazole-3-carboxylate (-)-9: [R]
171 (c 0.1, methanol); colorless oil; R 0.57 (chloroform/
methanol 9:1 + 50 µL acetic acid). HPLC retention time:
7.20 min (column: CHIRALCEL OD-H, Diacel; eluent: pe-
D
:
-
f
2
0
a literature procedure.
3
(
5R,2′S)-5-(2-tert-Butoxycarbonylamino-2-methoxycar-
troleum ether/2-propanol 95:5; flow rate: 1.0 mL/min; λ )
bonylethyl)-4,5-dihydroisoxazole-3-carboxylic Acid Eth-
1
2
1
54 nm). H NMR (CDCl
3
): 1.22 (t, J ) 7.2, 3H), 1.43 (s, 9H),
yl Ester (+)-6 and (5S,2′S)-5-(2-tert-Butoxycarbonylamino-
.92 (m, 2H), 2.22 (m, 1H), 2.93 (m, 1H), 3.34 (m, 1H), 4.08 (q,
2
-methoxycarbonylethyl)-4,5-dihydroisoxazole-3-
J ) 7.2, 3H), 4.30 (m, 1H), 4.92 (m, 1H), 5.12 (bs, 1H); Anal.
) C, H, N.
5S,2′R)-5-(2-Amino-2-carboxyethyl)-4,5-dihydroisox-
azole-3-carboxylic Acid (-)-4. Step A. Diester (-)-6
345 mg, 1.0 mmol) was dissolved in ethanol (3 mL) and
carboxylic Acid Ethyl Ester (-)-7. To a solution of (+)-10
710 mg, 3.10 mmol) in ethyl acetate (10 mL) was added ethyl
14 22 2 7
(C H N O
(
(
2-chloro-2-(hydroxyimino)acetate (705 mg, 4.65 mmol) and
sodium bicarbonate (1 g). The mixture was vigorously stirred
for 3 days at room temperature; the progress of the reaction
was monitored by TLC (petroleum ether/ethyl acetate 7:3).
Water was added to the reaction mixture and the organic layer
was separated and dried over anhydrous Na SO . The crude
(
treated with 1 N NaOH (3 mL) at room-temperature overnight.
The disappearance of the starting material was monitored by
TLC (chloroform/methanol 7:3 + 100 µL acetic acid). The
aqueous layer was washed with dichloromethane, made acidic
with 2 N HCl, and extracted with ethyl acetate. The organic
2
4
material, obtained after evaporation of the solvent, was
chromatographed on silica gel (petroleum ether/ ethyl acetate
4:1) to give 350 mg of (+)-6 and 556 mg of (-)-7. Overall yield
phase was dried over anhydrous Na
2 4
SO , and after evaporation
2
0
of the solvent, crude (5S,2′R)-5-(2-tert-butoxycarbonylamino-
85%. (+)-6: [R] : +70.6 (c 1.0, chloroform). Anal. (C H N O )
D
15 24
2
7
2
-carboxy)-4,5-dihydroisoxazole-3-carboxylic acid was obtained
as a white powder (220 mg, 0.73 mmol, yield 73%).
C, H, N. Spectroscopic data agreed with those reported for the
1
6
20
D
corresponding racemate. (-)-7: [R] : -78.2 (c 1.0, chloro-
Step B. The crude material, obtained from the previous
transformation (220 mg, 0.73 mmol), was treated with a 30%
dichloromethane solution of trifluoroacetic acid (562 µL,
form). Anal. (C15
24 2 7
H N O ) C, H, N. Spectroscopic data agreed
1
6
with those reported for the corresponding racemate.
Receptor Binding at iGluRs. The membrane prepara-
7
.3 mmol) at 0 °C. The solution was stirred at room temper-
tions used in receptor-binding experiments were prepared
3
5
ature for 3 h until disappearance of the starting material
according to Ransom and Stec with slight modifications, as
2
6
23
24
22
(
TLC: n-butanol/water/acetic acid 4:2:1). The volatiles were
previously described. Affinity for AMPA, KA, and NMDA
3
removed under vacuum, and the residue was washed with
receptor sites was determined using 5 nM [ H]AMPA, 5 nM
3
3
methanol and diethyl ether to give (-)-4: white prisms;
[ H]KA in the absence of CaCl , and 2 nM [ H]CGP 39653
2
2
0
3
decomposes in the range 175-182 °C. [R]
D
: -178.0 (c 0.1,
([ H]-(()-(E)-2-amino-4-propyl-5-phosphono-3-pentenoate), re-
water/methanol 1:1). R
f
0.37 (n-butanol/water/acetic acid
spectively. The tritiated ligands were all purchased from New
England Nuclear (NEN). The amount of bound radioactivity
was determined using a Packard TOP-COUNT microplate
scintillation counter. Binding data were analyzed by nonlinear
regression using GraphPad Prism 3.03 (GraphPad Software,
San Diego, CA). Data were fitted to the following equation:
4
:2:1). HPLC retention time: 28.15 min (column: CHIROBI-
OTIC TAG, Astec; eluant: [ethanol/methanol 3:2]/water 4:1
+
50 mM ammonium acetate + 0.1% acetic acid; flow rate:
1
mL/min; light scattering detection); ee g 99%; Anal.
(
C
7
H
10
N
2
O
5
) C, H, N. Spectroscopic data agreed with those
1
6
n
n
n
reported for the corresponding racemate.
5R,2′S)-5-(2-Amino-2-carboxyethyl)-4,5-dihydroisox-
azole-3-carboxylic Acid (+)-4. Monoacid (+)-8 (315 mg,
.95 mmol) was transformed into final amino acid (+)-4
overall yield: 60%) following the procedure described above
for compound (-)-4. (+)-4: white prisms; decomposes in the
B ) Bmax - (B
× [inhibition] )/(IC + [inhibitor] ), where
max
50
(
B is the binding, expressed as a percentage of total specific
binding, and n is the Hill coefficient.
0
(
Electrophysiology of NMDA Receptors in Xenopus
Oocytes. For expression in Xenopus oocytes, rat NR1-1a,
NR2A, NR2B, NR2C, and NR2D were subcloned into pCI-neo
containing a T7 site upstream from the 5′ untranslated region
(UTR). Linearized plasmids were used to produce NR1-1a,
NR2A, NR2B, NR2C, and NR2D cRNAs with mMessage
mMachine kits (Ambion). Oocytes were surgically removed
from mature female Xenopus laevis anaesthetized in a 0.4%
MS-222 (3-aminobenzoic acid ethyl ester, Sigma) solution for
10-15 min. To remove the follicle layer, the oocytes were
subsequently digested with 0.5 mg/mL collagenase (type IA,
20
range 155-170 °C; [R]
D
: +176.8 (c 0.1, water/methanol 1:1);
HPLC retention time: 13.75 min (column: CHIROBIOTIC
TAG, Astec; eluent: [ethanol/methanol 3:2]/water 4:1 + 50 mM
ammonium acetate + 0.1% acetic acid; flow rate: 1 mL/min;
light scattering detection); ee g 99%; Anal. (C
7 10 2 5
H N O ) C, H,
N. Spectroscopic data agreed with those reported for the
1
6
corresponding racemate.
5S,2′S)-5-(2-Amino-2-carboxyethyl)-4,5-dihydroisox-
azole-3-carboxylic Acid (-)-5. Monoacid (-)-9 (400 mg,
.21 mmol) was transformed into final amino acid (-)-5
overall yield: 85%) following the procedure described above
for compound (-)-4. (-)-5: white prisms; decomposes in the
(
2
Sigma) in OR-2 buffer (in mM: 82.5 NaCl, 2.0 KCl, 1.0 MgCl ,
1
(
and 5.0 HEPES pH 7.6) at room temperature for 2-3 h.
Healthy-looking stage V-VI oocytes were selected for injection
the following day. Oocytes were coinjected with cRNA encoding
NR1-1a and either of the NR2 subunits at a 1:1 ratio,
2
0
range 155-158 °C. [R]
D
: -69.0 (c 0.1, water/methanol 1:1);

6
324 Journal of Medicinal Chemistry, 2005, Vol. 48, No. 20
Conti et al.
respectively, and maintained in Barth’s solution (in mM:
tions were chosen to improve the quality of the docking results.
van der Waals and hydrogen bonding radii were set to 4.0 and
3.0 Å, respectively; the genetic algorithm parameters were kept
at the default value. Finally, the obtained docking results were
visually analyzed by Sybyl 6.8, to evaluate both the quality
and the accordance of the resulting binding mode with the
experimental data.
8
0
1
2
3 2 4
8 NaCl, 1.0 KCl, 2.4 NaHCO , 0.41 CaCl , 0.82 MgSO ,
.3 Ca(NO , 15 HEPES pH 7.5, 100 IU/mL penicillin, and
3 2
)
00 µg/mL streptomycin) at 18 °C. Recordings were performed
-4 days after injection using whole-cell two-electrode voltage-
2
+
2+
clamp at -40 mV to -100 mV in Ca - and Mg -free Ringers
solution containing (in mM) 115 NaCl, 2.5 KCl, 1.9 BaCl
2
, and
10 HEPES (pH 7.6). During recording, 10 µM glycine was
Molecular Modeling of mGluR2 Receptor Agonists.
Pharmacophore Models. The conformational flexibility of each
molecule was investigated through a stimulated annealing
included in all agonist and/or antagonist applications. Single
concentration tests were performed in at least triplicates with
two independent experiments. Electrophysiological data were
acquired using Clampex 7.0 (Axon Instruments) and subse-
quently processed with GraphPad Prism 4.0 (GraphPad Soft-
ware, San Diego, CA).
3
1
protocol as described by N. Jullian using the Discover
calculation engine with the CFF91 force field (Insight II
version 2001, Accelrys San Diego). It appears that the two
isomers (-)-2 and (+)-3 adopt only one stable conformation.
We superimposed them to the model by using the three
functional groups as anchor points (distal and proximal
carboxylate carbon atoms, amino group nitrogen atom) (Insight
II, Accelrys San Diego).
Metabotropic Testing. The mGluR subtypes mGluR1a,
mGluR2 and mGluR4a were stably expressed in Chinese
hamster ovary (CHO) cell lines. Cells were maintained and
tested as previously described by measurement of intracellular
2
5
Dockings. Each stereoisomer was docked at the Glu binding
calcium (mGluR1) or cyclic AMP (mGluR2,4) levels.
1
8
site of mGlu2R ATD model described by Bertrand et al. The
ligands were first manually positioned by superimposing the
R-carboxylate, the amino group, and the γ-carboxylate on those
of Glu. The obtained protein-ligand complex was therefore
subjected to energy minimization (steepest-descent conver-
Oxygen-Glucose Deprivation in Cortical Cell Cul-
tures. Cultures of mixed cortical cells containing both neuro-
nal and glial elements were prepared, used at 14 days in vitro,
and exposed to OGD as previously described.2 Briefly, culture
medium was replaced by a glucose-free balanced salt solution
7
-
1
-1
gence, 10 kcal. mol ‚Å ; conjugate-gradient convergence,
saturated with 95% N
were then sealed into an airtight incubation chamber equipped
with inlet and outlet valves and 95% N /5% CO was blown
2 2
/5% CO and heated to 37 °C. Multiwells
-
1
-1
0
.1 kcal. mol .Å ) while fixing with a restraint between 100
and 1000, the following distances: oxygen of the R-carboxylate
of ligand-S123 oxygen at 2.65 Å; ligand’s nitrogen-A144
oxygen of the carbonyl at 2.8 Å; ligand’s nitrogen-T146 oxygen
at 2.95 Å. These calculations were performed as previously
described using the Discover 3.00 calculation engine with the
2
2
through the chamber for 10 min to ensure maximal removal
of oxygen. The chamber was then sealed and placed into the
incubator at 37 °C for 60 min. OGD was terminated by
removing the cultures from the chamber, replacing the expo-
sure solution with oxygenated medium, and returning the
multiwells to the incubator under normoxic conditions. The
extent of neuronal death was assessed 24 h later. In this
system, 60 min OGD induced a neuronal damage that was
approximately 75% of the maximal degree of neuronal injury
achieved by exposing the cultures for 24 h to 1 mM glutamate.
OGD-induced cell injury was quantitatively evaluated by
measuring the amount of LDH released from injured cells into
culture media 24 h following exposure to OGD, as previously
18
CFF force field (Insight II version 2001, Accelrys, San Diego).
Acknowledgment. Professor Shigetada Nakanishi
is gratefully acknowledged for the kind gift of the
mGluR-expressing CHO cell lines. This work was
financially supported by MIUR (COFIN 2003) Rome and
Universit a` degli Studi di Milano (FIRST). G.G. wishes
to thank Professor Leonardo Scapozza (Laboratoire de
Chimie Th e´ rapeutique, Section des Sciences Pharma-
ceutiques, Universit e´ de Gen e´ ve, Switzerland.) for his
kind support in the utilization of the GOLD modeling
software. H.B.O. was supported by the Danish Medical
Research Council and the Novo Nordisk Foundation. FA
wishes to thank Dr H.-O. Bertrand (Accelerys, Orsay
France) for his constant and valuable support as well
as the C.N.R.S, the French ministry of reseach, RETINA-
France, and the Fondation de France for funding.
27
described. The LDH level corresponding to complete neuronal
death (with no glial death) was determined for each experi-
ment by assaying sister cultures exposed to 1 mM glutamate
for 24 h. Background LDH release was determined in control
cultures not exposed to OGD and subtracted from all experi-
mental values. The resulting value correlated linearly with
the degree of cell loss estimated by observation of cultures
under phase-contrast microscopy or under bright-field optics
following 5 min incubation with 0.4% trypan blue, which stains
debris and nonviable cells.
Molecular Modeling of NMDA Receptor Antagonists.
Supporting Information Available: Elemental analyses.
This material is available free of charge via the Internet at
http://pubs.acs.org.
Ligand Geometry and Conformation. All ligands, built using
36
Sybyl6.8, were preliminarily minimized by conjugated gradi-
3
6
ent utilizing the Tripos force field. The amino, carboxyl, and
phosphonate groups were ionized to better simulate the
physiological conditions. The charges were assigned by
MOPAC/PM3 calculation implemented in SYBYL environ-
ment.
References
(
1) Wheal, H. V., Thomson, A. M., Eds. Excitatory Amino Acids and
Synaptic Transmissions; Academic Press: London, 1995.
(
2) Br a¨ uner-Osborne, H.; Egebjerg, J.; Nielsen, E. Ø.; Madsen, U.;
Krogsgaard-Larsen, P. Ligands for glutamate receptors: Design
and therapeutic prospects. J. Med. Chem. 2000, 43, 2609-2645.
3) Dingledine, R.; Borges, K.; Bowie, D.; Traynelis, S. F. The
glutamate receptor ion channels. Pharmacol. Rev. 1999, 51,
Dockings. The open form of the NMDA/NR2A homology
model described by Grazioso et al.17 was chosen for the flexible
(
docking calculations. We used the putative binding site of the
NR2A protein surrounding the amino acids that have been
7
-61.
2
8,29
described as essential for l-Glu and (R)-AP5 binding.
GOLD
(4) Pin, J.-P.; Acher, F. The metabotropic glutamate receptors:
structure, activation mechanism and pharmacology. Curr. Drug
Targets - CNS & Neurol. Dis. 2002, 1, 297-317.
3
7,38
2
.1,
used for the molecular docking calculations, employs
a genetic algorithm (GA) wherein the molecular features of
the protein-ligand complexes are encoded as a chromosome.
The protein required for the calculations was considered with
all hydrogen atoms, and the implemented algorithm succes-
sively optimized the orientation of the hydroxyl hydrogen atom
of residues Tyr, Thr, Ser, and Lys located into the binding
pocket. Lone pairs were automatically added with the default
geometry; the cavity was detected with an active site radius
of 12.0 Å from the side chain hydrogen atom of residue T513.
The Goldscore fitness function and the torsion angle distribu-
(
5) Cull-Candy, S.; Brickley, S.; Farrant, M. NMDA receptor
subunits: diversity, development and disease. Curr. Opin.
Neurobiol. 2001, 11, 327-335.
(6) Schorge, S.; Colquhoun, D. Studies of NMDA receptor function
and stoichiometry with truncated and tandem subunits.
J. Neurosci. 2003, 23, 1151-1158.
(
7) Das, S.; Sasaki, Y. F.; Rothe, T.; Premkumar, L. S.; Takasu, M.;
Crandall, J. E.; Dikkes, P.; Conner, D. A.; Rayudu, P. V.; Cheung,
W.; Chen, H. S. V.; Lipton, S. A.; Nakanishi, N. Increased NMDA
current and spine density in mice lacking the NMDA receptor
subunit NR3A. Nature (London) 1998, 393, 377-381.

Novel Dihydroisoxazole Amino Acids
Journal of Medicinal Chemistry, 2005, Vol. 48, No. 20 6325
3
(
8) Chatterton, J. E.; Awobuluyl, M.; Premkumar, L. S.; Takanishi,
H.; Talantova, M.; Shin, Y.; Cui, J.; Tu, S.; Sevarino, K. A.;
Nakanishi, N.; Tong, G.; Lipton, S. A.; Zhang, D. Excitatory
glycine receptors containing the NR3 family of NMDA receptor
subunits. Nature (London) 2002, 415, 793-798.
(24) Braitman, D. J.; Coyle, J. T. Inhibition of [ H]kainic acid receptor
binding by divalent cations correlates with ion affinity for the
calcium channel. Neuropharmacology 1987, 26, 1247-1251.
(25) Bjerrum, E. J.; Kristensen, A. S.; Pickering, D. S.; Greenwood,
J. R.; Nielsen, B.; Liljefors, T.; Schousboe, A.; Br a¨ uner-Osborne,
H.; Madsen, U. Design, synthesis, and pharmacology of a highly
subtype-selective GluR1/2 agonist, (RS)-2-amino-3-(4-chloro-3-
hydroxy-5-isoxazolyl)propionic acid (Cl-HIBO). J. Med. Chem.
(
9) Kemp, J. A.; Kew, J. N. C.; Gill, R. Handbook of Experimental
Pharmacology; Jonas, P., Monyer, H., Eds.; Springer: Berlin,
1999; Vol. 141, pp 495-527.
(
10) Lees, K. R.; Asplund, K.; Carolei, A.; Davis, S. M.; Diener, H.
C.; Kaste, M.; Orgogozo, J. M.; Whitehead, J. Glycine antagonist
2003, 46, 2246-2249.
(
26) Clausen, R. P.; Hansen, K. B.; Cali, P.; Nielsen, B.; Greenwood,
J. R.; Begtrup, M.; Egebjerg, J.; Brauner-Osborne, H. The
respective N-hydroxypyrazole analogues of the classical glutamate
receptor ligands ibotenic acid and (R,S)-2-amino-2-(3-hydroxy-
(
gavestinel) in neuroprotection (GAIN International) in patients
with acute stroke: a randomised controlled trial. GAIN Inter-
national Investigators. Lancet 2000, 355, 1949-1954.
(
11) Nikam, S. S.; Meltzer, L. T. NR2B Selective NMDA Receptor
Antagonists. Curr. Pharm. Des. 2002, 8, 845-855.
12) Gotti, B.; Duverger, D.; Bertin, J.; Carter, C.; Dupont, R.; Frost,
J.; Gaudilliere, B.; MacKenzie, E. T.; Rousseau, J.; Scatton, B.
Ifenprodil and SL 82.0715 as cerebral anti-ischemic agents. I.
Evidence for efficacy in models of focal cerebral ischemia.
J. Pharmacol. Exp. Ther. 1988, 247, 1211-1221.
13) Kemp, J. A.; McKerman, R. M. NMDA receptor pathways as drug
targets. Nature Neurosci. 2002, 5, 1039-1042.
14) Conti, P.; De Amici, M.; Joppolo di Ventimiglia, S.; Stensbøl, T.
B.; Madsen, U.; Br a¨ uner-Osborne, H.; Russo, E.; De Sarro, G.;
Bruno, G.; De Micheli, C. Synthesis and anticonvulsant activity
of novel bicyclic acidic amino acids. J. Med. Chem. 2003, 46,
5
-methyl-4-isoxazolyl)acetic acid. Eur. J. Pharmacol. 2004, 499,
5-44.
3
(
(
27) For details see: Pellegrini-Giampietro, D. E.; Cozzi, A.; Leonardi,
P.; Peruginelli, F.; Meli, E.; Pellicciari, R.; Moroni, F. 1-Amino-
1
,5-dicarboxylic acid and (S)-(+)-2-(3′-carboxybicyclo[1.1.1]pentyl)-
glycine, two mGlu1 receptor-preferring antagonists, reduce
neuronal death in in vitro and in vivo models of cerebral
ischemia. Eur. J. Neurosci. 1999, 11, 3637-3647.
(
(
28) Anson, L. C.; Chen, P. E.; Wyllie, D. J. A.; Colquhoun, D.;
Schoepfer, R. Identification of amino acid residues of the NR2A
subunit that control glutamate potency in recombinant NR1/
NR2A NMDA receptors. J. Neurosci. 1998, 18, 581-589.
29) Lummis, S. C. R.; Fletcher, E. J.; Green, T. Identification of a
new site in the S1 ligand binding region of the NMDA receptor
NR2A subunit involved in receptor activation by glutamate.
Neuropharmacology 2002, 42, 437-443.
(30) Branden, C.; Tooze, J. Introduction to protein structure, 2nd ed.;
Garland Ed. Publishing: New York, 1999; pp16-18.
(31) Jullian, N.; Brabet, I.; Pin, J.-P.; Acher, F. Agonist selectivity of
mGluR1 and mGluR2 metabotropic receptors: a different en-
vironment but similar recognition of an extended glutamate
conformation. J. Med. Chem. 1999, 42, 1546-1555.
(32) Kunishima, N.; Shimada, Y.; Tsuji, Y.; Sato, T.; Yamamoto, M.;
Kumasaka, T.; Nakanishi, S.; Jingami, H.; Morikawa, K. Struc-
tural basis of glutamate recognition by a dimeric metabotropic
glutamate receptor. Nature (London) 2000, 407, 971-977.
(33) Acher, F.; Bertrand, H.-O. Amino acid recognition by Venus
Flytrap domains is encoded in an 8 residue motif. Biopolymers
2005, 80, 357-366.
(
(
3102-3108.
(
15) Roda, G.; Conti, P.; De Amici, M.; He, J.; Polavarapu, P. L.; De
Micheli, C. Enantiopure stereoisomeric homologues of glutamic
acid: chemoenzymatic synthesis and assignment of their abso-
lute configuration. Tetrahedron: Asymmetry 2004, 15, 3079-
3090.
(
16) Conti, P.; De Amici, M.; Grazioso, G.; Roda, G.; Barberis Negra,
F.; Nielsen, B.; Stensbøl, T. B.; Madsen, U.; Br a¨ uner-Osborne,
H.; Frydenvang, K.; De Sarro, G.; Toma, L.; De Micheli, C.
Design, synthesis and pharmacological characterization of novel
potent NMDA receptor antagonists. J. Med. Chem. 2004, 47,
6740-6748.
(
17) Grazioso G.; Moretti, L.; Scapozza, L.; De Amici, M.; De Micheli,
C. Development of a threee-dimensional model for the N-Methyl-
d-Aspartate NR2A subunit. J. Med. Chem. 2005, 48, 5489-5494.
18) Bertrand, H.-O.; Bessis; A.-S.; Pin, J.-P.; Acher, F. Common and
selective molecular determinants involved in metabotropic
glutamate receptor agonist activity. J. Med. Chem. 2002, 45,
(
(
34) Bessis, A.-S.; Rondard, P.; Gaven, F.; Brabet, I.; Triballeau, N.;
Pr e´ zeau, L.; Acher, F.; Pin, J.-P. Closure of the Venus flytrap
module of mGlu8 receptor and the activation process: insights
from mutations converting antagonists into agonists. Proc. Natl.
Acad. Sci. U.S.A. 2002, 99, 11097-11102.
3171-3183.
(
19) Sih, C. J.; Wu, S.-H. In Topics in Stereochemistry; Eliel, E. L.;
Wilen, S. H.; Eds.; John Wiley & Sons: New York, 1989; Vol.
19, pp 63-125.
(
20) Collier, P. N.; Campbell, A. D.; Patel, I.; Taylor, R. J. K. The
direct synthesis of novel enantiomerically pure R-amino acids
in protected form via Suzuki cross-coupling. Tetrahedron Lett.
3
(
35) Ransom, R. W.; Stec, N. L. Cooperative modulation of [ H]MK-
8
01 binding to the N-methyl-d-aspartate receptor-ion channel
complex by l-glutamate, glycine and polyamines. J. Neurochem.
988, 51, 830-836.
36) Sybyl 6.8, Tripos Inc., 1699 South Hanley Rd., St. Louis, MO
2
000, 41, 7115-7119.
1
(
21) Kozikowski, A. P.; Adamczyk, M. Methods for the stereoselective
(
(
(
cis-cyanohydroxylation of olefins. J. Org. Chem. 1983, 48, 366-
6
3144.
372.
37) GOLD v. 2.1, Cambridge Crystallographic Data Centre: Cam-
bridge, U.K.
38) Verdonk, M. L.; Cole, J. C.; Hartshorn, M. J.; Murray, C. W.;
Taylor, R. D. Improved Protein-Ligand Docking Using GOLD.
Proteins 2003, 52, 609-623. Available: http://www.ccdc.cam.ac.uk
(
22) Sills, M. A.; Fagg, G.; Pozza, M.; Angst, C.; Brundish, D. E.; Hurt,
S. D.; Wilusz, E. J.; Williams, M. [3H]CGP 39653: a new
N-methyl-d-aspartate antagonist radioligand with low nano-
molar affinity in rat brain. Eur. J. Pharmacol. 1991, 192, 19-
24.
(
23) Honor e´ , T.; Nielsen, M. Complex structure of quisqualate-
sensitive glutamate receptors in rat cortex. Neurosci. Lett. 1985,
54, 27-32.
JM0504499