The glutamate receptor GluR5 agonist (S)-2-Amino-3-(3-hydroxy-7,8-dihydro- 6H-cyclohepta[d]isoxazol-4-yl)propionic acid and the 8-methyl analogue: Synthesis, molecular pharmacology, and biostructural characterization

Article abstract of DOI:10.1021/jm900565c

The design, synthesis, and pharmacological characterization of a highly potent and selective glutamate GluR5 agonist is reported. (S)-2-Amino-3-((RS)-3- hydroxy-8-methyl-7,8-dihydro-6H-cyclohepta-[d]isoxazol-4-yl)propionic acid (5) is the 8-methyl analogu

Full text of DOI:10.1021/jm900565c

J. Med. Chem. 2009, 52, 4911–4922 4911
DOI: 10.1021/jm900565c
The Glutamate Receptor GluR5 Agonist (S)-2-Amino-3-(3-hydroxy-7,8-dihydro-6H-
cyclohepta[d]isoxazol-4-yl)propionic Acid and the 8-Methyl Analogue: Synthesis,
Molecular Pharmacology, and Biostructural Characterization†
Rasmus P. Clausen,*,‡ Peter Naur,‡ Anders S. Kristensen,‡ Jeremy R. Greenwood,‡ Mette Strange,‡ Hans Brauner-Osborne,‡
Anders A. Jensen,‡ Anne Sophie T. Nielsen,‡ Ulla Geneser,‡ Lone M. Ringgaard,‡ Birgitte Nielsen,‡ Darryl S. Pickering,§
Lotte Brehm,‡ Michael Gajhede,‡ Povl Krogsgaard-Larsen,‡ and Jette S. Kastrup‡
§
‡Department of Medicinal Chemistry and Department of Pharmacology and Pharmacotherapy, Faculty of Pharmaceutical Sciences, University
of Copenhagen, 2 Universitetsparken, DK-2100 Copenhagen, Denmark
Received May 3, 2009
The design, synthesis, and pharmacological characterization of a highly potent and selective glutamate
GluR5 agonist is reported. (S)-2-Amino-3-((RS)-3-hydroxy-8-methyl-7,8-dihydro-6H-cyclohepta-
[d]isoxazol-4-yl)propionic acid (5) is the 8-methyl analogue of (S)-2-amino-3-(3-hydroxy-7,8-dihy-
dro-6H-cyclohepta[d]isoxazol-4-yl)propionic acid ((S)-4-AHCP, 4). Compound 5 displays an impro-
ved selectivity profile compared to 4. A versatile stereoselective synthetic route for this class of
compounds is presented along with the characterization of the binding affinity of 5 to ionotropic
glutamate receptors (iGluRs). Functional characterization of 5 at cloned iGluRs using a calcium
imaging assay and voltage-clamp recordings show a different activation of GluR5 compared to (S)-
glutamic acid (Glu), kainic acid (KA, 1), and (S)-2-amino-3-(3-hydroxy-5-tert-butyl-4-isoxazolyl)pro-
pionic acid ((S)-ATPA, 3) as previously demonstrated for 4. An X-ray crystallographic analysis of 4 and
computational analyses of 4 and 5 bound to the GluR5 agonist binding domain (ABD) are presented,
including a watermap analysis, which suggests that water molecules in the agonist binding site are
important selectivity determinants.
Introduction
and NR3A-B, and KA receptors of GluR5-7 and KA1-2
subunits.1,6-8 In recent years, X-ray crystallographic studies
of a soluble construct (GluR2-S1S2) of the agonist binding
domain (ABD) of GluR2 containing various AMPA receptor
ligands have provided structural information about ligand
recognition as well as activation and desensitization mechan-
isms.9-11 Recently, ABDs of NR1, NR2A, NR3A-B, GluR3-
GluR6, and theδ2 receptor have appeared.12-21 TheABDhas
a clamshell-like structure that closes upon ligand binding.
This motion probably leads to opening of the ion channel. It
has been observed that the degree of domain closure induced
by ligands in GluR2 correlates with the relative agonist
efficacy of ligands measured by the magnitude of the induced
ion current.22,23 These constructs crystallize as dimers, and it
is now generally believed that the iGluR complexes are dimers
of dimer assemblies containing a total of four ligand binding
sites.
Glutamic acid (Glua) is the primary central excitatory
neurotransmitter in the brain and plays a key role in normal
functions of the central nervous system (CNS) but also in a
number of CNS disorders.1-4 Upon synaptic release, Glu
activates a highly heterogeneous group of receptors compris-
ing both the G-protein coupled metabotropic Glu receptors
(mGluRs) as well as the ionotropic Glu receptors (iGluRs)
that are ligand-gated ion channels.1,5 On the basis of struct-
ural, functional, and pharmacological characteristics, iGluRs
are categorized into three classes according to selective activa-
tion by the agonists N-methyl-D-aspartic acid (NMDA), (S)-
2-amino-3-(3-hydroxy-5-methylisoxazol-4-yl)propionic acid
((S)-AMPA, 2), and kainic acid (KA, 1) (Figure 1). The
iGluRs are homo- or heteromeric assemblies of subunits
forming ion channels, AMPA receptors being made up of
the GluR1-4 subunits, NMDA receptors of NR1, NR2A-D,
We recently reported that (S)-2-amino-3-(3-hydroxy-7,8-
dihydro-6H-cyclohepta[d]isoxazol-4-yl)propionic acid ((S)-
4-AHCP, 4) is a potent and selective agonist at homomeric
GluR5 receptors.24 This finding explained the paradoxically
high potency in the activation of cat spinal interneurons but
relatively low affinity to known Glu binding sites previously
observed.25 The X-ray crystal structure of 4 in the GluR2-
S1S2 construct was recently reported and in this structure
the ABD is partially closed, which is in agreement with
previous observations that 4 is a partial agonist at GluR2.26
However, at GluR5, 4 shows superagonistic characteristics by
inducing a larger current than the full agonists KA and Glu, in
†PDB ID: 2WKY.
*To whom correspondence should be addressed. Phone: þ45 35 33 65
66. Fax: þ45 35 33 60 40. E-mail: rac@farma.ku.dk.
a Nonstandard abbreviations: 4-AHCP, 2-amino-3-(3-hydroxy-7,8-
dihydro-6H-cyclohepta[d]isoxazol-4-yl)propionic acid; iGluRs, ionotropic
glutamate receptors; KA, kainic acid; ATPA, 2-amino-3-(3-hydroxy-5-
tert-butylisoxazol-4-yl)propionic acid; ABD, agonist binding domain;
Glu, (S)-glutamic acid; CNS, central nervous system; mGluRs, meta-
botropic glutamate receptors; NMDA, N-methyl-D-aspartic acid; AMPA,
2-amino-3-(3-hydroxy-5-methylisoxazol-4-yl)propionic acid; ConA, Con-
canavalin A; GluR2-S1S2, soluble construct of the agonist binding
domain of GluR2; GluR5-S1S2, soluble construct of the agonist binding
domain of GluR5; DFT, density functional theory.
r
2009 American Chemical Society
Published on Web 07/09/2009
pubs.acs.org/jmc

4912 Journal of Medicinal Chemistry, 2009, Vol. 52, No. 15
Clausen et al.
Scheme 1a
Figure 1. Structures of Glu and selective iGluR agonists 1-5.
voltage-clamp measurements using Xenopus laevis oocytes
and from measurements of intracellular calcium concentra-
tions in HEK293 cells,24 when these systems are expressing
GluR5(Q) and treated with Concanavalin A (ConA) to block
desensitization. Thus, 4 is an important tool for studying the
functional properties of GluR5 and other KA receptor sub-
types because it displays a different pharmacology compared
to other GluR5 ligands.
Twosyntheticroutes to4have beenreportedof whichoneis
stereoselective, but the yields are rather low. Thus, a more
efficient route to 4 and analogues is desirable. Furthermore,
4 also has significant AMPA receptor activity. We here
present a preparative synthetic route to 4, and the analogue
(S)-2-amino-3-((RS)-3-hydroxy-8-methyl-7,8-dihydro-6H-cyclo-
hepta[d]isoxazol-4-yl)proionic acid (5) with improved phar-
macological properties, together with an X-ray crystal struc-
ture of 4 in the soluble construct of the GluR5 ABD. The
interactions of 4 and 5 with the AMPA and KA receptor
subtypes GluR2 and GluR5, respectively, are compared using
computational binding site analysis methods including the
recently reported watermap analysis.
Results
Chemistry. The new synthetic route to 4 (Scheme 1) com-
menced like previously reported syntheses from O-alkyl
ketone 9a,24,25 which was prepared from ethyl 2-oxohepta-
necarboxylate 6a in three steps via 7a and 8a. Similarly, the
methylated analogue 9b could be synthesized from ethyl
(RS)-3-methyl-2-oxoheptanecarboxylate 6b27 via 7b and
8b. Treatment of ketones 9a and 9b with hydrazine gave
the hydrazones 10a and 10b in 97% and 89% yield, respec-
tively. Subsequently, treatment with iodine yielded vinyl
iodides 11a and 11b, in 83% and 53% yield, respectively,
and these iodides were substrates in Negishi couplings. Thus,
the (S)-serine derived iodide 12 was converted to the zinc
iodide 13,28 which was used directly in Pd-catalyzed cross-
coupling reactions with vinyl iodides 11a and 11b, furnishing
14a and 14b in 38% and 55% yield, respectively. Deprotec-
tion of 14a upon treatment with 6 M HCl in AcOH for 4 h at
120 ꢀC yielded 4 in 89% yield with 87% ee. Similarly,
compound 5 was obtained in 86% yield with 86% ee from
14b. Recrystallizations improved the enantiomeric excess
to>95%. Attempts to separate the two diasteromers of 5
were not successful by either crystallization or HPLC. In an
attempt to synthesize the chain homologue of 4, the aspartic
acid derived iodide 1529 was converted to the zinc iodide 16
that was employed in a cross-coupling reaction with vinyl
a Reagents: (a) NH2OH, H2O; (b) iPr-Br, DMF, K2CO3; (c) Na2-
Cr2O7 2H2O, H2SO4, AcOH; (d) NH2NH2, EtOH; (e) I2, THF Et3N;
3
(f) Zn, DMF; (g) Pd(OAc2), dicyclohexyl-m-biphenyl phosphine; (h) 12 M
aq. HCl, AcOH.
iodide 11a, providing the protected homologue 17 in 57%
yield. However, deprotection attempts under various acidic
conditions yielded complex mixtures from which we were
unable to isolate the desired final product.
We also investigated the possibility of making six-mem-
bered ring homologues of 4 and 5 using the same synthetic
strategy (Scheme 2). Thus, ketones 21a and 21b were ob-
tained in three steps from 18a and 18b via isoxazolols 19a and
19b and the O-alkylated 20a and 20b, respectively. These
ketones were converted into hydrazones 22a and 22b in 99%
and 89% yield, respectively, that upon treatment with iodine
yielded vinyl iodides 23a and 23b in 63% and 53% yield,

Article
Journal of Medicinal Chemistry, 2009, Vol. 52, No. 15 4913
respectively. The vinyl iodides 23a and 23b were employed
in a Pd-catalyzed cross-coupling with zinc iodide 13 provid-
ing the protected ring homologues 24a and 24b in 52%
and 55% yield, respectively. Again, deprotection of these
protected amino acids under acidic conditions yielded
complex mixtures and we did not obtain the desired final
products.
in [3H]AMPA, [3H]KA, and [3H]CGP 39653 competition
radioligand displacement assays, respectively, using cell mem-
branes prepared from rat cortical brain tissue (Table 1). The
affinity of 5 for AMPA binding sites is more than an order of
magnitude lower than that of 4 and 5-fold weaker than that
of (S)-2-amino-3-(3-hydroxy-5-tert-butyl-4-isoxazolyl)pro-
pionic acid ((S)-ATPA, 3). Also, the affinity of 5 for KA
binding sites is reduced 5-fold compared to 4, although it is
still 3-fold more potent than 3. The affinities of compounds 4
and 5 were further evaluated at recombinant GluR1-7
expressed as homomeric receptors in Sf9 cells (Table 2). In
this assay, compounds 4 and 5 are virtually equipotent at
GluR5, but at all other subtypes, compound 5 displays
approximately an order of magnitude lower affinity than
compound 4.
Scheme 2a
Compound 5 was further characterized functionally using
HEK293 cell lines stably expressing homomeric GluR1-6
by measuring the increase of intracellular Ca2þ upon iGluR
activation using Fluo-4/AM as a fluorescent Ca2þ indicator
(Table 3).30 In this assay, compound 5 exhibit reduced
agonist potencies at GluR1-4 AMPA receptors compared
to 4, whereas the potency at iGluR5 is maintained. No
response at GluR6 is observed for 4 or 5 at concentrations
up to 1 mM. Thus, compound 5 more selectively activates
GluR5 than 4. As for 1 and 4, compound 5 is a partial agonist
at AMPA receptors in this assay, whereas 3 is a full agonist.
As observed previously for 4, the response induced by 5 at
ConA treated GluR5 is greater than those elicited by Glu, 1,
and 3. To further establish the functional characteristics of
5 seen in the Fluo-4/AM assay, we performed functional
characterization using voltage-clamp measurements at oocytes
expressing GluR1 and GluR5 (Table 4, Figure 2). Again, 5 is
more selective as compared to previously reported data for 3
and 4.24,31
a Reagents: (a) (1) NH2OH, H2O; (b) iPr-Br, DMF, K2CO3; (c)
Na2Cr2O7 2H2O, H2SO4, AcOH; (d) NH2NH2, EtOH; (e) I2, THF
Et3N; (f) 11, Pd(OAc2), dicyclohexyl-m-biphenyl phosphine.
3
X-ray Crystal Structure. The soluble GluR5-S1S2 con-
struct comprising the agonist binding domain of the recep-
tor16 was cocrystallized with 4 in order to acquire detailed
knowledge on the binding mode of 4 in GluR5 (Figure 3A).
The protein crystallized in space group P41212, with two
molecules forming a dimer in the asymmetric unit and
Table 1. Binding Data on Ionotropic Glutamate Receptors in Rat
Cortical Membranesa
IC50 (μM)
IC50 (μM)
Ki (μM)
compd
[3H]AMPA
[3H]KA
[3H]CGP 39653
˚
reflections extended to 2.2 A resolution (Table 5). Investiga-
2b
3c
4e
5
0.039
1.8
0.28
>100
23
1.6
>100
>100d
>100
>100
tion of the ligand-binding cavity reveals that the ligand is
bound with the R-amino acid part of the molecule anchored
primarily by charge-charge interactions to Arg523 and
Glu738 in a manner very similar to that observed in other
structures of the ABD of iGluRs and with additional hydro-
gen bonds to backbone as well as side chain atoms of the
protein (Table 6 and Figure 3B). In the distal part of the
ligand, the exocyclic oxygen and the nitrogen atom of the
isoxazole ring act as a mimic of the γ-carboxylate moiety of
9.7 [5.02 ( 0.05]
8.3 [5.09 ( 0.06]
a The numbers in brackets [min, max] indicate mean(SEM accord-
ing to a logarithmic distribution. b From ref 52. c From ref 53. d [3H] CPP.
e From ref 24.
Pharmacology. The affinity of compound 5 for native
AMPA, KA, and NMDA receptors was initially determined
Table 2. Receptor Binding Affinity at Cloned Subtypes Expressed in Sf9 Cells
compounds
3
4
5
b
receptor
GluR1o
Ki (nM) pKi ( SEM]a
nH
Ki (nM) pKi ( SEM]
nH
Ki (nM) pKi ( SEM]
nH
2780 [3.526 ( 0.050]
1660 [3.213 ( 0.051]
2040 [3.309 ( 0.014]
2140 [3.322 ( 0.060]
2.13 [0.325 ( 0.041]
>1000000
1.10 ( 0.11
0.81 ( 0.05
0.75 ( 0.03
0.89 ( 0.03
0.96 ( 0.03
232 [2.364 ( 0.030]
175 [2.243 ( 0.020]
173 [2.231 ( 0.058]
202 [2.295 ( 0.065]
2.57 [0.399 ( 0.057]
>100000
1.03 ( 0.01
0.95 ( 0.02
0.97 ( 0.05
0.90 ( 0.05
0.93 ( 0.01
1550 [3.177 ( 0.077]
6310 [3.782 ( 0.091]
1110 [3.024 ( 0.081]
2200 [3.338 ( 0.031]
5.08 [0.696 ( 0.066]
>100000
0.93 ( 0.04
1.04 ( 0.10
0.98 ( 0.04
1.04 ( 0.02
0.98 ( 0.02
GluR2(R)o
GluR3o
GluR4o
GluR5(Q)
GluR6(VCR)A
GluR7A
2530 [3.399 ( 0.045]
0.87 ( 0.08
199 [2.247 ( 0.101]
0.92 ( 0.05
1380 [3.113 ( 0.076]
1.04 ( 0.04
a Mean pKi(SEM (nM) from at least three experiments conducted in triplicate at 16 concentrations of the compound. b Hill coefficient, mean(SEM.

4914 Journal of Medicinal Chemistry, 2009, Vol. 52, No. 15
Clausen et al.
Table 3. Functional Characterization of Glu, 1, 3, 4, and 5 at Stable GluR HEK293 Cell Lines in the Fluo-4/AM Assaya
compounds
receptor
GluR1i
Glub
1b
3b
4b
32
5
EC50 (μM)
max (%)
51
100
190
100
52
23
18
380
15
40
45
47
55
37
97
730
101
70 [4.2 ( 0.2]
29 ( 3
160 [3.8 ( 0.4]
11 ( 3
48 [4.2 ( 0.04]
46
R
28
92
24
13
54
GluR2Qi
GluR3i
EC50 (μM)
Rmax (%)
>1000
EC50 (μM)
630
100
150
102
R
max (%)
100
20
GluR4i
EC50 (μM)
Rmax (%)
6.7
79
0.58
35 [3.8 ( 0.4]
68 ( 6
0.86 [6.1 ( 0.2]
147 ( 8
>1000
100
140
100
73
GluR5Q
GluR6Q
EC50 (μM)
1.8
R
max (%)
114
>1000
147
>1000
EC50 (μM)
Rmax (%)
1.8
98
100
a EC50 values and maximal response values (Rmax as % of Rmax of Glu) are given. For compound 5, pEC50 ( SEM values in brackets are given. b From
Strange et al.30
Table 4. Potencies from Functional Characterization of 3, 4, and 5
Using Two-Electrode Voltage Clamp Recordings on Xenopus laevis
Oocytes Expressing Cloned GluR Subtypes
compounds
3
4
5
receptor
EC50 (μM) EC50 (μM)
EC50 (μM)
nHill
GluR5(Q)
GluR1
NR1/NR2A
0.6631
6231
0.1324
4.524
ND
0.24 [6.6 ( 0.8] 1.29 ( 0.12
70 [4.2 ( 0.3] 0.91 ( 0.13
ND
>1000
ND
Glu when compared to the structure of GluR5-S1S2 com-
plexed with Glu (Figure 3C).16 Introduction of the seven-
membered ring has two consequences: (1) Four water mole-
cules (W3, W5, W6, and W7, Figure 3C) participating in a
hydrogen-bond network between the protein and the endo-
genous ligand Glu are displaced by the ring, which thereby
almost entirely fills this subcavity of the protein. (2) The ring
acts as a spacer that keeps the ABD in a partially open
conformation (9-11ꢀ more open as compared to GluR5 in
complex with Glu, PDB code 1YCJ16).
In the structure of GluR2-S1S2 in complex with 4, the
ligand was found in two conformations caused by a ring
flip in the seven-membered ring (Figure 3D).26 In the GluR5-
S1S2 structure, the majority of the electron density can be
explained by a single conformation of 4.
Figure 2. (A) Representative recording of currents evoked by 5
and Glu obtained from a Xenopus laevis oocyte expressing ConA-
treated homomeric GluR5 using a two-electrode voltage clamp.
(B) Concentration-response curves of 5 at homomeric GluR5 (9),
homomeric GluR1i (b), NR1/NR2A (O). Responses are normal-
ized to the response evoked by a saturating concentration of Glu
(100 μM). Each data point represent averaged responses from at
least five different oocytes. Error bars are SEM recordings from
NR1/NR2A expressing oocytes were performed in the presence of
100 μM glycine.
Molecular Modeling. Docking and Design. When 4 was
first docked to GluR2 and a homology model of GluR524
and compared with the GluR5-selective ligand (S)-2-amino-
3-(5-tert-butyl-3-hydroxyisothiazol-4-yl)propionic acid (3a)
(PDB: 2AIX), docking scores and structural analysis indi-
cated that GluR5 probably accommodates a methyl substi-
tuent at position 8 better than GluR2. In particular, Ser721
and Ser741 in GluR5 (Thr686 and Met708 in GluR2) were
hypothesized to provide extra space for or clash with the
substituent, respectively. Solving the experimental structures
of 4 in complex with GluR2-S1S2 (PDB: 1WVJ)26 and
GluR5-S1S2 (PDB: YYYY; present study) supports this
idea, and accordingly, analogue 5 was synthesized, expecting
that introduction of a methyl group at position 8 would
increase potency at GluR5 and decrease potency at GluR2,
thus improving GluR5 versus GluR2 selectivity.
the plane of the ring system (that is, on the same side of or
opposite to the R-amino acid side chain). Density functional
theory (DFT) and ab initio calculations in water indicate
that for 4, two of these conformations are more favorable: those
with the position 6 carbon atom opposite to the side chain.
The same two conformations were observed in the structure
of GluR2-S1S2:4 (PDB: 1WVJ).26 The conformation observed
in GluR5-S1S2 is 1.0 (DFT) or 0.4 kcal/mol (ab initio) higher
in energy than the global minimum. Examination in GluR5
of the electron density around the seven-membered ring, with
re-refinement using alternative conformations, suggests that the
bound structure is mostly in this higher-energy conformation,
Conformational Preferences of 4 and 5. In principle, the
seven-membered rings of 4 and of the diastereomers (2S,8R)-
5 and (2S,8S)-5 can assume four puckering modes, depending
on whether the carbon atom in position 7 is in or out of the
isoxazole plane and whether position 6 is located above or below

Article
Journal of Medicinal Chemistry, 2009, Vol. 52, No. 15 4915
Figure 3. Structure of 4 bound to the agonist binding domain of GluR5. (A) A 2-fold symmetric dimer of GluR5-S1S2 is present in the
asymmetric unit of the crystal (molA is shown in green and molB in orange). 4 is displayed in blue. (B) Close-up view of the ligand binding site
˚
(molA), including potential hydrogen bonds within 3.2 A (dashed lines). Bonds of protein and ligand are shown in green and blue, respectively.
Oxygen atoms are colored red, nitrogen atoms blue, and sulfur atoms yellow. Water molecules are shown as red spheres. (C) Superposition (on
all residues) of the structures of 4 and Glu (molA, PDB: 1YCJ) in complex with GluR5-S1S2, respectively. The protein residues have been left
out for clarity. 4 and water molecules are colored as in (B). In the GluR5-S1S2:Glu complex, the ligand and water molecules are colored gray.
Potential hydrogen bonds from ligands to water molecules are shown. (D) Visualization of the ligand binding cavity of GluR2 and GluR5 with
4 bound. The cavity was computed using the program PASS with water molecules treated as part of the protein. Side chains of amino acids
differing between GluR2 and GluR5 are shown.

4916 Journal of Medicinal Chemistry, 2009, Vol. 52, No. 15
Table 5. Data Collection and Refinement Statistics for GluR5-S1S2:4
Clausen et al.
Watermap Analysis. A watermap analysis was performed
in order to establish whether selectivity can be understood in
terms of the pattern and energies of water molecules in the
binding pocket. Watermap is a binding site analysis tool
based on inhomogeneous solvation theory, capable of de-
scribing the entropic and enthalpic contributions to the free
energies of individual receptor hydration sites, by clustering
the results from an explicit solvent molecular dynamics
simulation.32,33 Using the proteins only from GluR2-
S1S2:4 (PDB: 1WVJ)26 and from GluR5-S1S2:4 (molB), a
pair of MD simulations were run and analyzed, leading to
the watermaps shown in Figure 5. The agreement between
the watermap sites and the experimental water and ligand
heteroatom positions was generally very tight. Although
there were minor differences in the networks of water
molecules displaced by ligand in each receptor, 4 displaces
roughly 10 water molecules in both cases and stabilizes W1
and W2, seen in the majority of iGluR agonist complexes.
However, in GluR2 at the degree of domain closure induced
by 4, the ligand sequesters two high-energy water molecules:
W3 (5.5 kcal/mol) and W8 (7.0 kcal/mol). In GluR5-S1S2:4,
W3 is absent both in the watermap and in the X-ray
structure, while W8, which is experimentally observed in
molB, is much lower in watermap energy (0.2 kcal/mol);
two changes that are expected to confer selectivity. Further-
more, in GluR5, a new water molecule near the ligand, W9, is
favorably stabilized. Ser741 and Ser721 (Met708 and Thr686
in GluR2) contribute to the stability of W8 and W9 in
GluR5. Finally, in the space between Val685 and the ligand
in GluR5 (but not between Leu650 and the ligand in GluR2),
watermap predicts a mildly unfavorable displaceable hydra-
tion site Wx (1.2 kcal/mol), although only at partial occu-
pancy (64%) and not experimentally observed. This suggests
an opportunity for increased selectivity if Wx is displaced, as
would be the case by the methyl group of (2S,8R)-5. In
summary, the watermaps indicate that the selectivities of 4
and 5 can be understood in terms of the free energies of the
binding site water molecules that differ between GluR2 and
GluR5 near the ligand.
space group
˚
no. per a.u.a
crystal mosaicity (deg)
P41212
unit cell (A)
a=b=71.3, c=234.6
2
0.33
˚
resolution (A)
30-2.20
386202
31679
total observations
unique observations
I/σ(I)b
12.6 (2.6)
99.4 (99.3)
9.2 (43.9)
21.1 (25.6)
24.8 (31.6)
4042/34/2/295
27.1/26.4/30.9/31.6
completeness (%)b
Rsym (%)b,c
R
work (%)d
Rfree (%)e
no. protein/ligand/ion/water atoms
average B-values protein/
2
ligand/ion/water (A )
˚
˚
rmsd bond lengths (A)
0.007
1.3
97.2
rmsd bond angles (deg)
residues in allowed regions (%)f
a Number of protein molecules per asymmetric unit (a.u). b Values in
˚
parentheses are statistics for the highest resolution bin (2.34-2.20 A).
P
P
c Rsym(I)= hkl|Ihkl-ÆIhklæ|/ hkl Ihkl, where Ihkl is the measured inten-
P
sity of the reflections with indices hkl. d Rwork
=
hkl( Fo,hkl|-|Fc,hkl )/
|Fo,hkl|, where |Fo,hkl| and |Fc,hkl| are the observed and calculated struc-
ture factor amplitudes for reflection hkl. e Five percent of the reflections
in the data set were set aside for calculation of the Rfree value. f The
Ramachandran plot was calculated according to Kleywegt et al.54
˚
Table 6. Potential Hydrogen Bonds within 3.2 A between GluR5-S1S2
and 4a
molA
molB
4 R carboxylate oxygen 1:
Thr518 N
2.8
2.6
2.8
2.8
Arg523 Nη1
4 R carboxylate oxygen 2:
Arg523 Nη2
Ser689 N
2.8
2.9
2.8
2.8
4 R-ammonium nitrogen:
Pro516
2.8
3.0
2.9
2.6
2.9
2.9
Thr518 Oγ1
Glu738 Oε1
Binding Mode of 5 and Stereoselectivity. To model the
binding mode of 5 and determine which diastereomer is likely
to be more active, four local minimum-energy conforma-
tions of (2S,8R)-5 and (2S,8S)-5 with the charge-bearing
heteroatoms constrained to approximate the experimental
positions were least-squares superimposed with the experi-
mental coordinates of 4. The complexes were minimized,
treating the ligand with DFT and the receptor with molec-
ular mechanics (QM-MM). Side chains in contact with the
ligand were allowed to move freely during minimization. In
particular, the conformation of Met737 was found to adjust
to maximize favorable contact with the ligand. Although it
was not possible to determine with certainty the precise
preferred ring-puckering mode of 5 when bound, Figure 4
depicts a low-energy complex for each diastereomer. In
general, (2S,8R)-5 produced lower energy complexes than
(2S,8S)-5, indicating (2S,8R)-5 to be a more active eutomer.
Figure 4 also shows the hydrophobic and hydrophilic sur-
faces (shown as site maps) in the vicinity of the 8-position.
These observations illustrate that above the isoxazole plane
of the ligands is a more hydrophilic region with limited space.
The 8-methyl group of (2S,8S)-5 must bend away to avoid it,
forcing part of the seven-membered ring downward below
the preferred zone indicated by the binding mode of 4. By
contrast, the 8-methyl group of (2S,8R)-5 can be neatly
4 distal hydroxy group:
W1
2.9
2.9
4 distal ring nitrogen:
Thr690 Oγ1
W2
2.6
3.0
2.6
3.1
4 distal ring oxygen:
Glu738 N
a The chemical structure of 4 is shown in Figure 1.
3.2
3.1
but some contribution from the alternative low-energy ring
puckering seen in GluR2 cannot be ruled out. The reason for
the slightly higher energy ring conformation with the carbon
atom in position 7 out of the isoxazole plane in GluR5 is
avoidance of close approach to the γ-hydroxyl group of
Ser741, whereas the γ-methylene of Met708 in GluR2 is further
away. For 5, the conformational distribution is further affected
by the preference of the 8-methyl substituent to rotate out of the
isoxazole plane, disfavoring one of the conformations of each
diastereomer and in particular the ring conformation of
(2S,8R)-5 corresponding to 4 in GluR5 is higher in energy.
Two low energy conformations of (2S,8R)-5 and (2S,8S)-5 are
presented in Figure 4.

Article
Journal of Medicinal Chemistry, 2009, Vol. 52, No. 15 4917
Figure 4. QM-MM minimized complexes of (2S,8R)-5 (black) and (2S,8S)-5 (gray) in GluR5, superimposed with 4 (brown). Hydrophobic
and hydrophilic site map isopotentials near chiral position 8 are indicated in tan and cyan, respectively.
acccomodated by a hydrophobic volume below the isoxazole
plane, in contact with Val685, Ser721 and Met737. The
smaller volume of the corresponding hydrophobic space in
GluR2 due to Thr686 and Leu650, along with the aforemen-
tioned weak hydration site Wx in GluR5, suggest that it is the
(2S,8R)-5 diastereomer that confers the observed increased
GluR5 selectivity of 5 compared with 4. Thus, these studies
indicate that selectivity and potency of this class of com-
pounds may be improved further by synthesizing (2S,8R)-5
analogues stereoselectively. However, initial attempts to
develop a synthetic route for this were not successful and
the diastereomers could not be separated.
investigated the possibility of using the synthetic strategy to
obtain an alkylated analogue of 4 as well as ring and chain
homologues. However, the deprotection conditions produ-
ced the desired compounds as the main product only for the
8-methylated compound 5, whereas the other analogues
yielded complex reaction mixtures in which the desired pro-
duct could not be identified. A new synthetic route to ketone
7a employing a different O-protecting group would be re-
quired to further improve this synthesis. Control of the
stereochemistry at the 8-position of 5 would also be an
objective of such a route.
A pharmacological and biostructural characterization of 4
and 5 was performed as well. The pharmacological studies
conclude that 5 is more selective than 4 in two different
functional assays and this selectivity profile was also seen in
competition binding experiments. The selectivity of 5 as a
GluR5 agonist is comparable to that of the standard GluR5
ligand 3. Both compounds 4 and 5 induced an increased
response at homomeric GluR5 receptors compared to the full
agonists Glu, 1, and 3 when the receptor is treated with ConA
to block desensitizitation (Figure 2 and Table 3). These results
suggest that these two compounds behave like superagonists
of GluR5, inducing a larger current than glutamate at satur-
ating concentrations when receptor densitization is inhibited
by ConA. This behavior has to our best knowledge not
previously been observed with any other iGluR agonist and
may explain previously reported high activity of 4 observed at
cat spinal interneurons.25 The superagonist behavior cannot
be readily understood at present and requires more detailed
kinetic studies of GluR5 activation, and 4 and 5 will be
important tools in such studies. Although these compounds
behave as superagonists, the crystal structure of 4 in the
GluR5 ABD displays only partial domain closure. In X-ray
crystal structures of GluR2, this is observed with compounds
that are partial agonists and also in the structure of the partial
agonist domoic acid in GluR5-S1S2.34 Thus, the current
structure presents a challenge to the previously observed
relationship between the degree of domain closure and the
response induced by orthosteric ligands,35 which has led to the
Discussion
We have presented a new stereoselective synthetic route to
the specific GluR5 agonist 4 as well as the X-ray crystal
structure of 4 bound to the GluR5 ABD. Furthermore, we
have presented structure-based design, synthesis, and phar-
macological characterization of a new analogue 5, which
turned out to be one of the most potent and selective GluR5
agonists yet described. Compound 4 is interesting because it
induces an increased response at ConA-treated homomeric
GluR5 receptors compared to the full agonists Glu, 3, and 1.24
This is observed when GluR5 is expressed either transiently in
Xenopus laevis oocytes or stably in HEK293 cells.24,30 We
decided to investigate the possibility of improving the synth-
esis of 4 and the pharmacological profile by making a new
designed analogue 5. Comparison of the X-ray structure of 4
in the GluR2 agonist binding domain construct with 4 docked
in a homology model of the GluR5 agonist binding domain
suggested that the selectivity could be improved by introdu-
cing increased steric bulk at the 8-position.
We focused on improving the synthetic route from 7a,
which in the first reported procedure25 provided racemic 4
in 3% yield and in the later stereoselective route24 led to (S)-4
with a yield of 0.4%. Inspired by the use of a Negishi type
cross-coupling reaction for introducing the alanine side chain
by Jackson and co-workers,28 4 could be synthesized from 7a
in 28% yield with an enantiomeric excess of 89%. We

4918 Journal of Medicinal Chemistry, 2009, Vol. 52, No. 15
Clausen et al.
Figure 5. Watermaps of GluR2 and GluR5 from molecular dynamics, using the protein only from GluR2-S1S2J:4 (PDB: 1WVJ) and from
GluR5-S1S2:4 (molB). Experimental ligand (gray) and water positions included for comparison (ball-and-stick) together with H-bond network
(fine dashed lines). Transparent spheres indicate clustered water positions, color-coded by entropic (green) and enthalpic (cyan) contribution to
free energy (darker is higher ΔG). Key water molecules nearby but not displaced by ligand are labeled W1-W3, W7-8, and Wx, together with
selected ΔGs in kcal/mol. Ligand-stabilized water molecules W1 and W2 are observed in both GluR2 and GluR5. High energy water W3 in
GluR2 is not predicted or observed in GluR5. W8 is greatly stabilized in GluR5 vs GluR2. Low energy water W9 in GluR5 is not predicted or
observed in GluR2. Low occupancy hydration site Wx (64%) is predicted but too diffuse for observation in GluR5 and absent in GluR2.
hypothesis that agonist efficacy is positively correlated to a
high degree of domain closure in kainate receptors. However,
a possible explanation for the observed discrepancy could also
be that the use of ConA in our functional assays does not

Article
Journal of Medicinal Chemistry, 2009, Vol. 52, No. 15 4919
completely inhibit GluR5 desensitization and that 4 and 5
desensitizes GluR5 to a lesser extent than glutamate.
GluR5-S1S2 in complex with 4 shows many similar binding
features to the GluR2 structure but also some key differences
in the binding pocket in particular with respect to the pattern
of waters. We therefore performed a watermap analysis of the
GluR5 agonist binding site in comparison with GluR2 to
understand the role of individual water molecule binding sites,
both observed and virtual, in the selectivity observed. This
analysis disclosed that the increased activity of 4 and 5 at
GluR5 compared to GluR2 is clearly linked to changes in the
energy of particular water molecules in the binding pocket
that are stabilized, sequestered, or displaced by the ligand,
affirming the utility of this kind of analysis. The high potency
and selectivity most likely resides with the (2S,8R) diastereo-
mer of 5 according to binding site mapping and QM-MM
calculations.
was added to a heated (85 ꢀC) solution of 12 M HCl (2.5 mL).
The resulting reaction mixture was refluxed for 1.5 h, reduced in
vacuo, and added H2O (50 mL). Extraction with AcOEt (3ꢀ50 mL),
drying, filtration, evaporation, and CC (toluene-AcOEt (4:1)
and 1% AcOH) gave 7b (1.43 g, 82%), which was sufficiently
pure to take further. A small sample was recrystallized (toluene-
petroleum ether): mp 125.5 -126.5 ꢀC. 1H NMR δ 1.28 (3H, d,
J=7.2 Hz), 1.88-1.46 (5H, m), 1.96 (1H, m), 2.33 (1H, ddd, J=
3.5, 8.0, and 15.9 Hz), 2.44 (1H, ddd, J=3.5, 7.5, and 15.9 Hz),
2.94 (1H, m), 11.56 (1H, br s). 13C NMR δ 17.3, 20.7, 27.64,
27.66, 34.0, 34.3, 105.8, 169.7, 173.7. Anal. (C9H13NO2) C, H, N.
(RS)-3-Isopropoxy-8-methyl-5,6,7,8-tetrahydro-4H-cyclohepta-
[d]isoxazole (8b). Potassium carbonate (840 mg, 6.1 mmol) was
added to a solution of 7b (920 mg, 5.50 mmol) in dry DMF
(15 mL), and the reaction mixture was left stirring for 30 min at
60 ꢀC. Isopropyl bromide (775 μL, 8.26 mmol) was added, and
the resulting reaction mixture was stirred overnight at 60 ꢀC.
The reaction mixture was cooled, H2O (20 mL) was added, and
the mixture was extracted with Et2O (3ꢀ50 mL). The combined
organic phases were dried, filtered, evaporated, and CC (toluene-
AcOEt (9:1)) gave 8b (890 mg, 77%) sufficiently pure for the
next reaction. A sample was distilled in a Kugelrohr apparatus:
bp 225 ꢀC/12 mmHg. 1H NMR δ 1.29 (3H, d, J=6.9 Hz), 1.37
(6H, d, J=6.1 Hz), 1.42-1.74 (4H, m), 1.74-1.85 (1H, m), 1.94
(1H, m), 2.31 (2H, m), 2.94 (1H, m), 4.87 (1H, heptet, J=6.1 Hz).
13C NMR δ 17.46, 20.75, 22.05, 22.07, 27.56, 27.78, 33.92, 34.30,
72.56, 105.38, 169.46, 173.27. Anal. (C12H19NO2) C, H; N:
calcd, 6.69; found, 6.09.
In conclusion, compound 5 is one of the most potent and
selective GluR5 agonists yet known and could be an impor-
tant tool for further studying the functional characteristics of
GluR5 in particular with respect to the increased response as
well as the general physiological role of GluR5 receptors.
Experimental Section
Chemistry. General Procedures. Analytical thin-layer chro-
matography (TLC) was performed on silica gel F254 plates
(Merck). All compounds were detected using UV light and/or by
spraying with diluted solutions of KMnO4 or H2SO4/Ce(SO4)2/
ammonium molybdate. Ketones, 3-isoxazolols, and amino
groups were developed using standard spraying solutions of
2,4-dinitrophenylhydrazine, FeCl3, and ninhydrin, respectively.
Column chromatography (CC) was performed on silica gel
Matrex LC 60A (0.070-0.200 mm). Unless otherwise stated,
the petroleum ether used for CC and recrystallizations had
distillation ranges 80-100 ꢀC and 40-60 ꢀC, respectively. Orga-
nic phases were dried using MgSO4. Melting points were deter-
(RS)-3-Isopropoxy-8-methyl-5,6,7,8-tetrahydro-4H-cyclohepta-
[d]isoxazol-4-one (9b). To a solution of 8b (550 mg, 2.63 mmol)
in AcOH (99.9%, 3 mL) and concentrated H2SO4 (400 mL) kept
at 0-10 ꢀC was added a solution of Na2Cr2O7 2H2O (818 mg,
3
2.74 mmol) in AcOH (5 mL). After stirring for 4 h at rt, the
reaction mixture was neutralized with 12 M NaOH, H2O was
added (50 mL) and the mixture was extracted with Et2O(3ꢀ 50 mL).
The combined organic phases were dried and evaporated,
followed by CC (toluene-AcOEt (9:1)). First starting material
was eluted (130 mg, 24%), then 9b (180 mg, 31%) as a yellow oil.
1H NMR δ 1.42, 1.43, 1.44 (3 ꢀ 3H, 3d, J=7, 6, and 6 Hz), 1.58-
2.60 (3H, m), 2.15 (1H, m), 2.70 (2H, m), 3.20 (1H, m), 4.94 (1H,
heptet, J=6.2 Hz). 13C NMR δ 18.55, 20.28, 21.71, 32.76, 34.73,
44.77, 73.95, 108.22, 168.69, 179.79, 194.12.
mined in capillary tubes and are uncorrected. 1H (300 MHz), 13
C
(75 MHz), and associated proton test (APT) NMR spectra were
recorded on a Varian Gemini-2000 BB spectrometer. Unless
otherwise stated, CDCl3 was used as solvent. Chemical shifts are
given in ppm (δ) using TMS [1H NMR (δ 0.0)] or the CDCl3
peak [13C NMR (δ 77.00)] as internal standards. For NMR
(RS)-4-Hydrazono-3-isopropoxy-8-methyl-5,6,7,8-tetrahydro-
4H-cyclohepta[d]isoxazole (10b). A solution of the ketone 9b
(175 mg, 0.78 mmol) in EtOH (10 mL) was treated with Et3N
(190 μL, 3.91 mmol) and hydrazine hydrate (1.1 mL, 7.89 mmol).
The reaction mixture was heated under reflux for 3 h and left
stirring overnight at rt. The solvent was evaporated and the
residue dissolved in CH2Cl2 (20 mL). The solution was washed
with H2O (20 mL) and the aqueous phase extracted with CH2Cl2
(3ꢀ20 mL). The combined organic phases were dried and eva-
porated to give a (1:1) mixture of stereoisomers of 10b (110 mg,
59%) as a solid. 1H NMR δ 1.31 and 1.32 (3H, 2d, J=6.9 and
7.5 Hz), 1.38-1.46 (6H, m), 1.56-1.72 (1H, m), 1.73-2.09
(3H, m), 2.50 (2H, m), 3.04 (1H, m), 4.93 and 5.01 (1H, 2 ꢀ
heptet, J=6.2 Hz). A sample was recrystallized (AcOEt-petro-
leum ether), giving a single stereoisomer of 10b: mp 118.0-
118.5 ꢀC. 1H NMR δ 1.31 (3H, d, J=6.9 Hz), 1.41 (3H, d, J=
6 Hz), 1.44 (3H, d, J=6 Hz), 1.58-1.73 (2H, m), 1.87-2.10 (2H,
m), 2.50 (2H, m), 3.04 (1H, m), 4.93 (1H, heptet, J=6.2 Hz), 5.33
(2H, s br). 13C NMR δ 19.45, 21.31, 22.00, 22.05, 29.40, 32.16,
34.37, 73.62, 105.26, 142.61, 168.53, 172.73. Anal. (C12H19N3O2)
C, H, N.
spectra recorded in D2O, 1,4-dioxane [1H NMR (δ 3.70) and 13
C
NMR (δ 69.48)] was used as an internal standard. Elemental
analyses were performed at Analytical Research Department,
H. Lundbeck A/S, Denmark, or by J. Theiner, Microanalytical
Laboratory, Department of Physical Chemistry, University of
Vienna, Austria, and are within (0.4% of the theoretical values
unless otherwise stated. Stereochemical purity, expressed as
enantiomeric excess (ee), was determined using a Chirobiotic
T column (4.6 mmꢀ150 mm, ASTEC) equipped with a Chiro-
biotic T guard column (4.6 mmꢀ50 mm, ASTEC). The column
was eluted at 0.5 mL/min with an aqueous solution of ammo-
nium acetate buffer (15 mM; adjusted to pH 4.0 using AcOH)
and EtOH (40/60; v/v %). A TSP system consisting of a P2000
pump, an AS3000 autoinjector, and an SM5000 PDA detector
was used. The % ee values were determined from peak areas at
210 nm.
(RS)-8-Methyl-5,6,7,8-tetrahydro-4H-cyclohepta[d]isoxazol-
3-ol (7b). A solution of ethyl 3-methyl-2-oxoheptanecarboxylate
(6b)27 (2.07 g, 10.4 mmol) in MeOH (2 mL) was carefully added
to NaOH (437 mg, 10.9 mmol) dissolved in MeOH (7 mL) and
H2O (0.5 mL) at -70 ꢀC. To this solution was added a mixture of
(RS)-4-Iodo-3-isopropoxy-8-methyl-7,8-tetrahydro-6H-cyclo-
hepta[d]isoxazole (11b). A solution of I2 (3.06 g, 12.1 mmol) in
dry THF (20 mL) was dropwise added to a solution of hydra-
zone 10b (1.15 g, 4.85 mmol) in dry THF (30 mL) and Et3N
(6.7 mL, 48.1 mmol) while keeping the temperature below 10 ꢀC.
The reaction mixture was then stirred for 30 min at rt followed
by addition of H2O (10 mL). The aqueous phase was extracted
NH2OH HCl (1.45 g, 21 mmol) and NaOH (0.87 g, 22 mmol) in
3
MeOH (8 mL) and H2O (1 mL) filtered and precooled to -70 ꢀC.
The reaction mixture was allowed to warm to 0 ꢀC over a period
of 2 h. Upon addition of acetone (1 mL) the reaction mixture

4920 Journal of Medicinal Chemistry, 2009, Vol. 52, No. 15
Clausen et al.
with Et2O (3 ꢀ 10 mL), and the combined organic phases were
dried, evaporated, and subjected to CC (toluene-AcOEt (4:1)),
yielding 11b (1.26 g, 78%) as a dark oil. 1H NMR δ 1.32 (3H, d,
J=6.9 Hz), 1.436 and 1.441 (2 ꢀ 3H, 2 ꢀ d, J=6 and 6 Hz), 1.68
(1H, m), 2.05 (1H, m), 2.15 (2H, m), 3.13 (1H, hextet, J=7.5 Hz),
4.94 (1H, heptet, J=6.0 Hz), 6.78 (1H, t, J=6.8 Hz). 13C NMR δ
18.62, 21.96, 29.24, 33.36, 34.09, 73.82, 78.36, 105.28, 144.17,
166.96, 172.73.
membranes prepared and used for binding as previously de-
tailed.41,42 The affinities of compound 5 at GluR1, GluR2(R),
GluR3, and GluR4 (all as flop isoforms) were determined from
competition experiments with 2-5 nM [3H]AMPA, at GluR5-
(Q)1b with 1-2 nM [3H]SYM 2081, at GluR6(V,C,R)a using
5 nM [3H] KA, and at GluR7a using 2-4 nM [3H]SYM 2081.
Italic letters in parentheses indicate the RNA-edited isoforms of
the subunits used.
Methyl (2S, 8RS)-2-[(tert-Butoxycarbonyl)amino]-3-(3-iso-
propoxy-8-methyl-7,8-dihydro-6H cyclohepta[d]isoxazol-4-yl)propi-
onate (14b). 1,2-Dibromoethane (117 μL, 1.36 mmol) was added
to a stirring suspension of Zn dust (1.80 g, 28 mmol) in dry
DMF (2.5 mL). The mixture was repeatedly heated to reflux
and allowed to cool to rt (ꢀ3). Trimethylsilyl chloride (35 μL,
0.28 mmol) was added to the mixture, which was stirred for
further 45 min. The iodide (R)-1228 (1.29 g, 3.92 mmol) was
dissolved in dry DMF (2.5 mL) under N2 and added to the Zn
dispersion to generate 13. The mixture was stirred at 50 ꢀC until
no starting material could be observed by TLC (ca. 45 min). The
Zn dispersion was allowed to settle, and the supernatant
was transferred by syringe to a mixture of the vinyl iodide 11b
(1.20 g, 3.60 mmol), Pd(AcO)2 (41 mg, 0.18 mmol), and dicy-
clohexyl-m-biphenyl phosphine (200 mg, 0.57 mmol) in dry
THF (2.5 mL). The remainder of the Zn dispersion was washed
with dry DMF (2 ꢀ 500 μL) and transferred to the resulting
reaction mixture, which was stirred overnight at rt. The mixture
was filtered through celite and added H2O (15 mL). Extraction
with AcOEt (3ꢀ50 mL), drying, filtration, evaporation, and CC
(toluene-AcOEt (6:1)) gave 14b (1.13 g, 77%) as an oil. 1H
NMR δ 1.30 (1.5H, d, J=7.5 Hz), 1.33 (1.5H, d, J=7.5 Hz),
1.45-1.38 (15H, m), 1.52-1.68 (1H, m), 1.85-2.00 (1H, m),
2.10-2.36 (2H, m), 2.45 (0.5H, dd, J=9 and 13 Hz), 2.70 (0.5H,
dd, J=9 and 13 Hz), 2.92 (0.5H, dd, J=6 and 13 Hz), 3.10-3.19
(1.5H, m), 3.68 and 3.71 (3H, s), 4.26-4.35 (1H, m), 4.93-5.04
(2H, m), 5.73 (1H, t, J=5.9 Hz). 13C NMR δ 19.1, 19.4, 21.8,
25.2, 25.6, 28.2, 31.7, 31.9, 34.1, 34.6, 37.7, 37.8, 51.76, 51.83,
53.3, 53.5, 73.4, 79.4, 103.1, 103.3, 125.0, 125.2, 132.1, 154.7,
154.8, 167.6, 172.9, 173.0, 174.3, 174.4.
(2S, 8RS)-2-Amino-3-(3-hydroxy-8-methyl-7,8-dihydro-6H-cyclo-
hepta[d]isoxazol-4-yl)propionic Acid Hydrochloride (5). Com-
pound 14b(1.09 g, 2.67 mmol) was dissolved in a mixture of 12 M
HCl (6 mL), H2O (3 mL) and AcOH (3 mL), and the mixture was
heated at 120 ꢀC for ca. 3 h. After evaporation and re-evaporation
from H2O (2ꢀ10 mL), the brownish residue was dissolved in H2O
(20 mL) and washed with CH2Cl2 (2 ꢀ 20 mL). The combined
CH2Cl2 phases were extracted with H2O. The combined H2O
phases was evaporated, giving 5 as a brownish-violet solid residue
(660 mg, 86%, 89% ee), which was>95% pure according to NMR,
TLC, and HPLC. Recrystallization (2-propanol-H2O) gave violet
crystals (140 mg, 45% ee) and the mother liquor furnished
after addition of ether 516 mg (95% ee) that upon recrystallization
(2-propanol-H2O-Et2O) gave 5 (212 mg, 28%, ee 98.7%): mp
>200 ꢀC. 1H NMR (D2O) δ 1.22 (1.5H, d, J=7 Hz), 1.25 (1.5H, d,
J=7 Hz), 1.60 (1H, m), 1.88 (1H, m), 2.27 (2H, m), 2.78 (0.5H, dd,
J=8.0 and 14.3 Hz), 2.98-3.22 (2.5H, m), 4.07 (1H, m), 5.87 (1H,
q, J=6.1 Hz). 13C NMR (D2O) δ 20.86, 20.97, 28.28, 28.43, 32.74,
32.87, 37.06, 37.22, 38.53, 38.65, 55.53, 55.73, 105.79, 105.97,
124.54, 124.65, 138.00, 138.17, 170.90, 174.40, 179.46, 179.56. Anal.
In Vitro cRNA Transcription. The cDNA encoding rat
GluR1i, rat NR1, rat NR2A, and rat GluR5-1a(Q) subunits
inserted into the vectors pGEM-HE (GluR1 and GluR5) and
pCIneo (NR1-1a and NR2A; kind gift from Dr.
Kasper Hansen, Emory University) were used for preparation
of capped cRNA transcripts. Capped mRNA was synthesized
from 1 μg of linearized plasmid DNAs by in vitro transcription
and purified using the mMESSAGE mMACHINE T7 mRNA-
capping kit (Ambion, Austin, TX) according to the protocol
supplied by the manufacturer. Purified cRNA was resuspended
in H2O to a concentration of 100 ng/μL and stored at -80 ꢀC
until use.
Electrophysiology. Preparation of Oocytes. Mature female
Xenopus laevis (NASCO, CA) were anesthetized by submersion
into a solution of 0.1% ethyl 3-aminobenzoate dissolved in
double-distilled water with 1 g/L CaCO3 and ovaries were
surgically removed. The ovarian tissue was dissected and treated
with 1.5 mg/mL collagenase in Ca2þ-free Barth’s medium (in
mM: 88 NaCl, 1 KCl, 15 Ca(NO3)2, 0.82 MgSO4, 2.4 NaHCO3,
5 HEPES, pH 7.4) for 45-90 min at room temperature. Col-
lagenase-treated oocytes were subsequently maintained in
Barth’s medium supplemented with 0.41 mM CaCl2. On the
second day, oocytes were injected with 25 nL of diluted
cRNA (∼0.03 ng/nL of GluR1i or ∼0.001 ng/nL of a 1:1
mixture of NR1-1a and NR2A) and incubated in Barth’s
medium with gentamicin (50 μg/mL; G1397, Sigma, St. Louis,
MO) at 17 ꢀC. Oocytes were used for recordings from 3 to 6 days
postinjection.
Two-Electrode Voltage Clamp Electrophysiology and Deter-
mination of Agonist EC50. Recording of membrane currents in
oocytes was performed by inserting two glass microelectrodes
(resistance: 0.5-2.5 MΩ) filled with 3 M KCl into an oocyte
emerged in frog Ringer’s solution (in mM: 115 NaCl, 2 KCl, 1.8
BaCl2, 5 HEPES, pH 7.6). Electrodes were voltage clamped with
the use of a two-electrode voltage clamp amplifier (TEV-200A,
Dagan Corporation, Minneapolis, MN). Recordings of mem-
brane currents were made while the oocytes were continuously
perfused with frog Ringer’s solution. Drugs were dissolved in
frog Ringer’s solution and added to oocytes by full bath
application. Agonist concentration-response relationships
were constructed by measuring the maximal current induced
by bath application of a saturating concentration of agonist to
the oocyte (100 μM glutamate for GluR1; 100 μM glutamate
and 100 μM glycine for NR1/NR2A), followed by application of
compounds in increasing concentrations. The current responses
(I) induced by application of a specific concentration of agonist
was normalized to the response induced by saturating concen-
tration of Glu (IGLU). The mean of normalized responses from
at least five oocytes were plotted as a function of agonist
concentration. The concentration-response relationship was
determined by curve-fitting of the four parameter logistic equa-
tion: I/I0 = 1/(1 þ (EC50/[agonist])n), where I is the agonist-
evoked current, I0 is the agonist-evoked current in the absence of
antagonist, [agonist] is the concentration of agonist, and n is the
Hill slope.
(C12H16N2O4 HCl) C, H, Cl, N.
3
In Vitro Pharmacology. Receptor Binding Assays. Affinities
for native AMPA, KA, and NMDA receptors in rat cortical
synaptosomes were determined using 5 nM [3H]AMPA (55.5 Ci/
mmol),36 5 nM [3H]KA (58.0 Ci/mmol),37 and 2 nM [3H]CGP
39653 (Kd=6 nM, 50.0 Ci/mmol),38 respectively, with minor modi-
fications as previously described.39 Rat brain membrane pre-
parations used in these receptor binding experiments were
prepared according to a method previously described.40
The Fluo-4 Intracellular Calcium Imaging Assay. Stable
GluR1i, GluR2(Q)i, GluR3i, GluR4i, GluR5(Q), and GluR6(Q)
HEK293 cell lines were constructed and the functional proper-
ties of Glu and compounds 1 and 3-5 were characterized in the
Fluo-4/AM assay as previously described.30 The cells were
split into poly-D-lysine-coated black 96-well plates with clear
bottoms (BD Biosciences, Bedford, MA). After 16-24 h, the
Recombinant Receptor Binding Assays. Sf9 cells were cultured
and infected with recombinant baculovirus of rat AMPA re-
ceptors (GluR1o-4o) or rat KA receptors (GluR5-7) and

Article
Journal of Medicinal Chemistry, 2009, Vol. 52, No. 15 4921
medium was aspirated and the cells were incubated in 50 μL
loading buffer (Hanks Buffered Saline Solution (HBSS) con-
taining 20 mM HEPES, 1 mM CaCl2, 1 mM MgCl2 and 2.5 mM
probenecid, pH 7.4) supplemented with 6 μM Fluo-4/AM
(Molecular Probes, Eugene, OR) at 37 ꢀC for 1 h. The loading
buffer was aspirated, the cells were washed with 100 μL
loading buffer, and then 100 μL assay buffer ((HBSS) contain-
ing 20 mM HEPES, 10 mM CaCl2, 1 mM MgCl2, and 2.5 mM
probenecid, pH 7.4, supplemented with 100 μM CTZ in the
case of GluR1-4 and with 1 mg/mL ConA in the case of
GluR5-6) was added to the wells. Then the 96-well plate was
assayed in a NOVOstar microplate reader (BMG Labtechnolo-
gies, Offenburg, Germany), measuring emission (in fluorescence
units (FU)) at 520 nm caused by excitation at 485 nm before and
up to 60 s after addition of 33 μL agonist solution (the agonists
were dissolved in assay buffer). The experiments were performed
in triplicate at least three times for each compound at each
iGluR.
Acknowledgment. The work was supported by grants
from The Lundbeck Foundation, The Dansync Center for
Synchrotron Radiation, The Drug Research Academy, and
The Danish Medical Research Council. We thank people at
ID14-1 and V. Soroka for data collection.
Supporting Information Available: Experimental procedures
for synthesis of compounds 4, 10a, 11a, 14a, 17, 19b, 20a, 20b,
21a, 21b, 22a, 22b, 23a, 23b, 24a, and 24b. This material is
available free of charge via the Internet at http://pubs.acs.org.
References
(1) Brauner-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.
(2) Dingledine, R.; Borges, K.; Bowie, D.; Traynelis, S. F. The
Glutamate Receptor Ion Channels. Pharmacol. Rev. 1999, 51, 7–62.
(3) Javitt, D. C. Glutamate as a therapeutic target in psychiatric
disorders. Mol. Psychiatry 2004, 9, 984–997.
(4) Parsons, C. G.; Danysz, W.; Quack, G. Glutamate in CNS
disorders as a target for drug development: an update. Drug News
Perspect. 1998, 11, 523–569.
(5) James, N. C. K.; John, A. K. Ionotropic and metabotropic
glutamate receptor structure and pharmacology. Psychopharma-
cology (Berlin) 2005, 4–29.
(6) Erreger, K.; Chen, P. E.; Wyllie, D. J.; Traynelis, S. F. Glutamate
receptor gating. Crit Rev. Neurobiol. 2004, 16, 187–224.
(7) Mayer, M. L. Glutamate receptor ion channels. Curr. Opin.
Neurobiol. 2005, 15, 282–288.
(8) Mansour, M.; Nagarajan, N.; Nehring, R. B.; Clements, J. D.;
Rosenmund, C. Heteromeric AMPA receptors assemble with a
preferred subunit stoichiometry and spatial arrangement. Neuron
2001, 32, 841–853.
(9) Armstrong, N.; Sun, Y.; Chen, G. Q.; Gouaux, E. Structure of a
glutamate-receptor ligand-binding core in complex with kainate.
Nature 1998, 395, 913–917.
(10) Armstrong, N.; Gouaux, E. Mechanisms for activation and antag-
onism of an AMPA-sensitive glutamate receptor: crystal structures
of the GluR2 ligand binding core. Neuron 2000, 28, 165–181.
(11) Hogner, A.; Kastrup, J.; Jin, R.; Liljefors, T.; Mayer, M.; Egebjerg,
J.; Larsen, I.; Gouaux, E. Structural basis for AMPA receptor
activation and ligand selectivity: crystal structures of five agonist
complexes with the GluR2 ligand-binding core. J. Mol. Biol. 2002,
322, 93–109.
X-ray Structure Determination. Expression and crystalliza-
tion of the GluR5-S1S2 construct comprising the ABD of
GluR5 was achieved as previously described.16 A protein solu-
tion of 4 mg/mL in 10 mM HEPES pH 7.0, 10 mM NaCl, 1 mM
EDTA, and 10 mM 4 was used for crystallization with the
hanging drop vapor diffusion method by mixing 1 μL of protein
solution with 1 μL reservoir solution consisting of 18%
PEG4000, 0.3 M Li2SO4, and 0.1 M Tris-HCl pH 7.5. Crystals
grew within two days at 6 ꢀC and were flash-frozen in liquid
nitrogen after transfer through a cryoprotectant of 20% glycerol
in reservoir solution. Data were collected at beamline ID14-1,
˚
ESRF, Grenoble, France, to 2.2 A resolution and were pro-
cessed using the HKL package.43 The structure was solved by
molecular replacement with the structure of GluR5-S1S2 in
complex with domoic acid as template.34 The program ARP/
wARP44 was used to automatically build 94% of the residues.
Following manual rebuilding in program COOT,45 the structure
was refined with the program CNS46 and the ligand could
hereafter unambiguously be built into the electron density.
The quality of the structure was validated using PROCHECK.47
Domain openings were determined using DynDom.48 Figure 3
was prepared in PyMOL.49
Computational Details. For the ligands 4 and 5 in solution, ab
initio (LMP2/6-311þG**) and density functional theory (M06-
2X/6-311þG**) optimizations including Poisson-Boltzmann
self-consistent reaction field treatment of continuum solvation
(PB-SCRF) were performed in Jaguar 7.5.50 All atom models
from both chains of GluR5-S1S2J:4 as well as GluR2-S1S2J:4
(PDB: 1WVJ) were prepared according to the standard protocol
for bond assignment, hydrogen addition, H-bonding network
optimization, and restrained minimization in Maestro 8.5.49
QM-MM minimizations were performed in Qsite 1.7, which
uses Jaguar 7.5 and Impact 5.0.49 The QM region consisted
of the each ligand (4, (2S,8R)-5, and (2S,8S)-5) at M06-2X/
6-311G** and the MM region was handled by the OPLS2005
forcefield, with constraints on all backbone atoms and
(12) Furukawa, H.; Gouaux, E. Mechanisms of activation, inhibition
and specificity: crystal structures of the NMDA receptor NR1
ligand-binding core. EMBO J. 2003, 22, 2873–2885.
(13) Furukawa, H.; Singh, S. K.; Mancusso, R.; Gouaux, E. Subunit
arrangement and function in NMDA receptors. Nature 2005, 438,
185–192.
(14) Mayer, M. L. Crystal structures of the GluR5 and GluR6 ligand
binding cores: molecular mechanisms underlying kainate receptor
selectivity. Neuron 2005, 45, 539–552.
(15) Nanao, M. H.; Green, T.; Stern-Bach, Y.; Heinemann, S. F.; Choe,
S. Structure of the kainate receptor subunit GluR6 agonist-binding
domain complexed with domoic acid. Proc. Natl. Acad. Sci. U.S.A.
2005, 102, 1708–1713.
(16) Naur, P.; Vestergaard, B.; Skov, L. K.; Egebjerg, J.; Gajhede, M.;
Kastrup, J. S. Crystal structure of the kainate receptor GluR5
ligand-binding core in complex with (S)-glutamate. FEBS Lett.
2005, 579, 1154–1160.
(17) Naur, P.; Hansen, K. B.; Kristensen, A. S.; Dravid, S. M.; Pickering,
D. S.; Olsen, L.; Vestergaard, B.; Egebjerg, J.; Gajhede, M.; Traynelis,
S. F.; Kastrup, J. S. Ionotropic glutamate-like receptor delta 2 binds
D-serine and glycine. Proc. Natl. Acad. Sci. U.S.A. 2007, 104, 14116–
14121.
(18) Yao, Y.; Harrison, C. B.; Freddolino, P. L.; Schulten, K.; Mayer,
M. L. Molecular mechanism of ligand recognition by NR3 subtype
glutamate receptors. EMBO J. 2008, 27, 2158–2170.
(19) Yao, Y.; Mayer, M. L. Characterization of a soluble ligand binding
domain of the NMDA receptor regulatory subunit NR3A. J.
Neurosci. 2006, 26, 4559–4566.
(20) Ahmed, A. H.; Wang, Q.; Sondermann, H.; Oswald, R. E. Struc-
ture of the S1S2 glutamate binding domain of GLUR3. Proteins
2008, 75, 628–637.
˚
side chains further than 6 A from the ligand. Four input
conformations each for 4, (2S,8R)-5, and (2S,8S)-5 were derived
from conformational searches in Macromodel 9.649 in which the
five fully or partially charged heteroatoms were constrained to
the experimental positions and then least-squares superimposed
with the native ligand in molA and molB. Hydrophobic and
hydrophilic isopotentials were calculated on molA with the
ligand, waters, and other heteroatoms removed, using Sitemap
2.2.49 The enthalpies, entropies, and free energies of binding site
water molecules with the ligand removed were calculated in
˚
Watermap 1.0 within a 5 A radius of the experimental posi-
tions of 4 in GluR2 (PDB: 1WVJ) and GluR5 (q.v., mol B) based
on a pair of 9 ns molecular dynamics simulations in Desmond
2.0,49,51 all with default settings, according to the method of
Abel et al.32,33 Figures depicting watermaps and site maps were
produced in Maestro 8.5.49
(21) Kasper, C.; Frydenvang, K.; Naur, P.; Gajhede, M.; Pickering,
D. S.; Kastrup, J. S. Molecular mechanism of agonist recognition
by the ligand-binding core of the ionotropic glutamate receptor 4.
FEBS Lett. 2008, 582, 4089–4094.

4922 Journal of Medicinal Chemistry, 2009, Vol. 52, No. 15
Clausen et al.
(22) Jin, R.; Banke, T. G.; Mayer, M. L.; Traynelis, S. F.; Gouaux, E.
Structural basis for partial agonist action at ionotropic glutamate
receptors. Nat. Neurosci. 2003, 6, 803–810.
(23) Frandsen, A.; Pickering, D. S.; Vestergaard, B.; Kasper, C.;
Nielsen, B. B.; Greenwood, J. R.; Campiani, G.; Fattorusso, C.;
Gajhede, M.; Schousboe, A.; Kastrup, J. S. Tyr702 is an important
determinant of agonist binding and domain closure of the ligand-
binding core of GluR2. Mol. Pharmacol. 2005, 67, 703–713.
(24) Brehm, L.; Greenwood, J. R.; Hansen, K. B.; Nielsen, B.; Egebjerg,
(37) Braitman, D. J.; Coyle, J. T. Inhibition of [3H]kainic acid receptor
binding by divalent cations correlates with ion affinity for the
calcium channel. Neuropharmacology 1987, 26, 1247–1251.
(38) Sills, M. A.; Fagg, D.; 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 nanomolar affnity in
rat brain. Eur. J. Pharmacol. 1991, 192, 19–24.
(39) Hermit, M. B.; Greenwood, J. R.; Nielsen, B.; Bunch, L.; Jorgensen,
e
C. G.; Vestergaard, H. T.; Stensbol, T. B.; Sanchez, C.; Krogsgaard-
e
J.; Stensbol, T. B.; Brauner-Osborne, H.; Sløk, F. A.; Kronborg,
Larsen, P.; Madsen, U.; Brauner-Osborne, H. Ibotenic acid and
thioibotenic acid: a remarkable difference in activity at group III
metabotropic glutamate receptors. Eur. J. Pharmacol. 2004, 486, 241–
250.
e
T. T. A.; Krogsgaard-Larsen, P. (S)-2-Amino-3-(3-hydroxy-7,8-
dihydro-6H-cyclohepta[d]isoxazol-4-yl)propionic acid, a potent
and selective agonist at the GluR5 subtype of ionotropic glutamate
receptors. Synthesis, modeling, and molecular pharmacology. J.
Med. Chem. 2003, 46, 1350–1358.
(40) Ransom, R. W.; Stec, N. L. Cooperative modulation of
[3H]MK801 binding to the N-methyl-D-aspartate receptor ion
channel complex by L-glutamate, glycine and polyamines. J. Neu-
rochem. 1988, 51, 830–836.
(41) Vogensen, S. B.; Clausen, R. P.; Greenwood, J. R.; Johansen,
T. N.; Pickering, D. S.; Nielsen, B.; Ebert, B.; Krogsgaard-Larsen,
P. Convergent synthesis and pharmacology of substituted tetra-
zolyl-2-amino-3-(3-hydroxy-5-methyl-4-isoxazolyl)propionic acid
analogues. J. Med. Chem. 2005, 48, 3438–3442.
(42) Sagot, E.; Pickering, D. S.; Pu, X.; Umberti, M.; Stensbol, T. B.;
Nielsen, B.; Chapelet, M.; Bolte, J.; Gefflaut, T.; Bunch, L. Chemo-
enzymatic synthesis of a series of 2,4-syn-functionalized (S)-gluta-
mate analogues: new insight into the structure-activity relation of
ionotropic glutamate receptor subtypes 5, 6, and 7. J. Med. Chem.
2008, 51, 4093–4103.
(25) Krogsgaard-Larsen, P.; Nielsen, E. O.; Curtis, D. R. Ibotenic acid
e
analogs;synthesis and biological and in vitro activity of confor-
mationally restricted agonists at central excitatory amino acid
receptors. J. Med. Chem. 1984, 27, 585–591.
(26) Nielsen, B. B.; Pickering, D. S.; Greenwood, J. R.; Brehm, L.;
Gajhede, M.; Schousboe, A.; Kastrup, J. S. Exploring the GluR2
ligand-binding core in complex with the bicyclical AMPA analogue
(S)-4-AHCP. FEBS J. 2005, 272, 1639–1648.
(27) Borowitz, I. J.; Bandurco, V.; Beller, H.; Williams, G. J.; Suciu, N.;
Gross, L.; Rigby, R. D. G.; Kurland, D. Medium ring compounds
0.7. Synthesis of 2-methyl-7-ketoundecanolide, 8-ketoundecano-
lide, and 2,4,6-trimethyl-7-ketodecanolide. J. Org. Chem. 1972, 37,
581–588.
(28) Jackson, R. F. W.; Moore, R. J.; Dexter, C. S.; Elliot, J.; Mowbray,
C. E. Concise synthesis of enantiomerically pure phenylalanine,
homophenylalanine, and bishomophenylalanine derivatives using
organozinc chemistry: NMR studies of amino acid-derived orga-
nozinc reagents. J. Org. Chem. 1998, 63, 7875–7884.
(29) Siebum, A. H. G.; Woo, W. S.; Raap, J.; Lugtenburg, J. Access to
any site-directed isotopomer of methionine, selenomethionine,
cysteine, and selenocysteine;use of simple, efficient modular
synthetic reaction schemes for isotope incorporation. Eur. J. Org.
Chem. 2004, 2905–2913.
(30) Strange, M.; Brauner-Osborne, H.; Jensen, A. A. Functional
characterisation of homomeric ionotropic glutamate receptors
GluR1-GluR6 in a fluorescence-based high throughput screening
assay. Comb. Chem. High Throughput Screening 2006, 9, 147–158.
(31) Nielsen, M. M.; Liljefors, T.; Krogsgaard-Larsen, P.; Egebjerg, J.
The Selective Activation of the Glutamate Receptor GluR5
by ATPA Is Controlled by Serine 741. Mol. Pharmacol. 2003, 63,
19–25.
(43) Otwinowski, Z.; Minor, W. Processing of X-ray diffraction data
collected in oscillation mode. Macromol. Crystallogr., Part A 1997,
276, 307–326.
(44) Perrakis, A.; Morris, R.; Lamzin, V. S. Automated protein model
building combined with iterative structure refinement. Nat. Struct.
Biol. 1999, 6, 458–463.
(45) Emsley, P.; Cowtan, K. Coot: model-building tools for molecular
graphics. Acta Crystallogr., Sect D: Biol. Crystallogr. 2004, 60,
2126–2132.
(46) Brunger, A. T.; Adams, P. D.; Clore, G. M.; Delano, W. L.; Gros,
P.; Grosse-Kunstleve, R. W.; Jiang, J. S.; Kuszewski, J.; Nilges, M.;
Pannu, N. S.; Read, R. J.; Rice, L. M.; Simonson, T.; Warren, G. L.
Crystallography & NMR system: a new software suite for macro-
molecular structure determination. Acta Crystallogr., Sect D: Biol.
Crystallogr. 1998, 54, 905–921.
(47) Laskowski, R. A.; Macarthur, M. W.; Moss, D. S.; Thornton, J. M.
Procheck;A Program to Check the Stereochemical Quality of
Protein Structures. J. Appl. Crystallogr. 1993, 26, 283–291.
(48) Hayward, S.; Lee, R. A. Improvements in the analysis of domain
motions in proteins from conformational change: DynDom ver-
sion 1.50. J. Mol. Graph. Model. 2002, 21, 181–183.
(32) Abel, R.; Young, T.; Farid, R.; Berne, B. J.; Friesner, R. A. Role of
the active-site solvent in the thermodynamics of factor Xa ligand
binding. J. Am. Chem. Soc. 2008, 130, 2817–2831.
(33) Young, T.; Abel, R.; Kim, B.; Berne, B. J.; Friesner, R. A. Motifs
for molecular recognition exploiting hydrophobic enclosure in
protein-ligand binding. Proc. Natl. Acad. Sci. U.S.A. 2007, 104,
808–813.
(34) Hald, H.; Naur, P.; Pickering, D. S.; Sprogoe, D.; Madsen, U.;
Timmermann, D. B.; Ahring, P. K.; Liljefors, T.; Schousboe, A.;
Egebjerg, J.; Gajhede, M.; Kastrup, J. S. Partial agonism and
antagonism of the ionotropic glutamate receptor iGluR5;Struc-
tures of the ligand-binding core in complex with domoic acid and
2-amino-3-[5-tert-butyl-3-(phosphonomethoxy)-4-isoxazolyl]propio-
nic acid. J. Biol. Chem. 2007, 282, 25726–25736.
(35) Frandsen, A.; Pickering, D. S.; Vestergaard, B.; Kasper, C.;
Nielsen, B. B.; Greenwood, J. R.; Campiani, G.; Fattorusso, C.;
Gajhede, M.; Schousboe, A.; Kastrup, J. S. Tyr702 is an important
determinant of agonist binding and domain closure of the ligand-
binding core of GluR2. Mol. Pharmacol. 2005, 67, 703–713.
(36) Honore, T.; Nielsen, M. Complex structure of quisqualate-sensi-
tive glutamate receptors in rat cortex. Neurosci. Lett. 1985, 54,
27–32.
(49) Delano, W. L. The PyMOL Molecular Graphics System; Delano
Scientific: San Carlos, CA, 2002.
(50) Schrodinger Suite 2008; Schrodinger LLC: 101 SW Main Street, Suite
1300, Portland, OR 97204, 2008.
(51) Desmond 2.0; D. E. Shaw Research LLC: 120 West 45th Street, 39th
Floor, New York, NY 10036, 2008.
(52) Vogensen, S. B.; Greenwood, J. R.; Varming, A. R.; Brehm, L.;
Pickering, D. S.; Nielsen, B.; Liljefors, T.; Clausen, R. P.; Johansen,
T. N.; Krogsgaard-Larsen, P. A stereochemical anomaly: the
cyclised (R)-AMPA analogue (R)-3-hydroxy-4,5,6,7-tetrahydro-
isoxazolo[5,4-c]pyridine-5-carboxylic acid [(R)-5-HPCA] resem-
bles (S)-AMPA at glutamate receptors. Org. Biomol. Chem. 2004,
2, 206–213.
(53) Stensbol, T. B.; Borre, L.; Johansen, T. N.; Egebjerg, J.; Madsen,
e
U.; Ebert, B.; Krogsgaard-Larsen, P. Resolution, absolute stereo-
chemistry and molecular pharmacology of the enantiomers of
ATPA. Eur. J. Pharmacol. 1999, 380, 153–162.
(54) Kleywegt, G. J.; Jones, T. A. Phi/psi-chology: Ramachandran
revisited. Structure 1996, 4, 1395–1400.