Isoxazole analogues bind the System xc- transporter: Structure-activity relationship and pharmacophore model

Article abstract of DOI:10.1016/j.bmc.2009.11.001

Analogues of amino methylisoxazole propionic acid (AMPA), were prepared from a common intermediate 12, including lipophilic analogues using lateral metalation and electrophilic quenching, and were evaluated at System x_c^-. Both the 5-naphthylethyl-(16) and 5-naphthylmethoxymethyl-(17) analogues adopt an E-conformation in the solid state, yet while the former has robust binding at System x_c^-, the latter is virtually devoid of activity. The most potent analogues were amino acid naphthyl-ACPA 7g, and hydrazone carboxylic acid, 11e Y = Y′ = 3,5-(CF₃)₂, which both inhibited glutamate uptake by the System x_c^- transporter with comparable potency to the endogenous substrate cystine, whereas in contrast the closed isoxazolo[3,4-d] pyridazinones 13 have significantly lower activity. A preliminary pharmacophore model has been constructed to provide insight into the analogue structure-activity relationships.

Full text of DOI:10.1016/j.bmc.2009.11.001

Bioorganic & Medicinal Chemistry 18 (2010) 202–213
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry
journal homepage: www.elsevier.com/locate/bmc
ꢀ
Isoxazole analogues bind the System xc transporter: Structure–activity
relationship and pharmacophore model
Sarjubhai A. Patel a, Trideep Rajale a,b, Erin O’Brien a, David J. Burkhart b, , Jared K. Nelson b,à
,
Brendan Twamley b, Alex Blumenfeld b, Monika I. Szabon-Watola a,b, John M. Gerdes a,
Richard J. Bridges a, Nicholas R. Natale a,b,
*
a NIH COBRE Center for Structural and Functional Neuroscience, The University of Montana, Missoula, MT 59812, USA
b Department of Chemistry University of Idaho, Moscow, ID 83844-2343, USA
a r t i c l e i n f o
a b s t r a c t
Article history:
Analogues of amino methylisoxazole propionic acid (AMPA), were prepared from a common interme-
diate 12, including lipophilic analogues using lateral metalation and electrophilic quenching, and were
evaluated at System xcꢀ. Both the 5-naphthylethyl-(16) and 5-naphthylmethoxymethyl-(17) analogues
adopt an E-conformation in the solid state, yet while the former has robust binding at System xcꢀ, the
latter is virtually devoid of activity. The most potent analogues were amino acid naphthyl-ACPA 7g,
and hydrazone carboxylic acid, 11e Y = Y0 = 3,5-(CF3)2, which both inhibited glutamate uptake by the
Received 9 August 2009
Revised 30 October 2009
Accepted 2 November 2009
Available online 10 November 2009
Keywords:
ꢀ
System xc transporter with comparable potency to the endogenous substrate cystine, whereas in con-
Isoxazole
System xc
trast the closed isoxazolo[3,4-d] pyridazinones 13 have significantly lower activity. A preliminary
pharmacophore model has been constructed to provide insight into the analogue structure–activity
relationships.
ꢀ
Transporter
Pharmacophore
Ó 2009 Elsevier Ltd. All rights reserved.
Structure–activity relationship
1. Introduction
gradient generated by the EAATs as the driving force for the im-
port of L-cystine (L-Cys2, 2). Thus, the transporter mediates the
L
-Glutamate (
L
-Glu, 1, Chart 1) is the primary excitatory neu-
uptake of a vital sulfur-containing amino acid needed for the
rotransmitter in the mammalian CNS. Through its activation of a
wide variety of ionotropic (iGluRs) and metabotropic (mGluRs)
synthesis of glutathione (GSH) and oxidative protection,2 while
simultaneously producing an efflux of L-Glu that has the poten-
excitatory amino acid (EAA) receptors,
L
-glutamate-mediated sig-
tial to contribute to either excitatory signaling or excitotoxic
pathology. The significance of these actions is reflected in the
naling contributes to fast synaptic neurotransmission, higher or-
der signal processing (e.g., synaptic plasticity, development,
learning and memory), and even to neuropathology.1 Concentra-
ꢀ
range of CNS processes, to which Sxc has been linked, including:
drug addiction,3,4 brain tumor growth,5 oxidative protection,6 vir-
tions of L-Glu in the CNS are regulated by a family of excitatory
al pathology,7 the operation of the blood brain barrier,8 neuro-
amino acid transporters (EAATs) that rapidly sequester and con-
centrate this dicarboxylic amino acid in glia and neurons, and
thereby limit its extracellular accumulation and access to EAA
transmitter release,9 and synaptic organization.10
ꢀ
Initial pharmacological studies on Sxc established
L-Cys2 and
L-Glu as substrates, verified it operated as an obligate exchanger,
receptors. In contrast to the EAAT-mediated uptake of
L
-Glu,
and defined key features of its specificity, for example: (i)
the System xc (Sxcꢀ) transporter has been implicated in the ex-
L
-aspartate is neither a substrate nor inhibitor, (ii)
L
-homocyst-
ꢀ
eate is an inhibitor, (i.e., a SO3 can replace a distal COOꢀ) and
ꢀ
port of
L
-Glu from CNS cells in a manner that allows it to activate
ꢀ
EAA receptors. Sxc is a member of the heteromeric amino acid
(iii) L-a-aminoadipate and L-a-aminopimelate are inhibitors (i.e.,
transporter family (HATs; a.k.a. glycoprotein-associated amino
longer chain lengths are tolerated).11 Using CNS-derived tumor
cell lines that express high levels of Sxcꢀ, we have begun to more
acid exchangers) that functions as an obligate exchanger. Under
ꢀ
physiological conditions Sxc employs the
L-Glu concentration
thoroughly investigate the SAR’s governing binding and translo-
cation.12,13 In addition to the SO3 moieties (S-sulfo-
L
L
-CySH,
ꢀ
-serine-O-SO3ꢀ) the binding site also accommodates SO2 groups
-homocysteine-sulfinate), but not the PO3 group of
ꢀ
* Corresponding author. Tel.: +1 406 243 4132; fax: +1 406 243 5228.
E-mail address: Nicholas.natale@umontana.edu (N.R. Natale).
ꢀ2
(
L
L
-serine-
-Glu is
ꢀ2
O-PO3
.
The ability to bind higher homologues of
L
Present address: GSK Biologicals, 553 Old Corvallis Rd., Hamilton MT 59840, USA.
Present address: PharmAgra Labs, 158 McClean Rd., Brevard, NC 28712, USA.
à
substantiated by the actions of S-carboxymethyl- and S-carboxy-
0968-0896/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmc.2009.11.001

S. A. Patel et al. / Bioorg. Med. Chem. 18 (2010) 202–213
203
+NH3
COO-
2. Results and discussion
2.1. Preparation of amino acid analogues of AMPA
O
COOH
COO-
CO2H
COO-
O
NH
O
S
N
S
The amino acid analogues of AMPA were prepared using syn-
thetic methodology previously described by our laboratory, with
the exception of MOM- and BOM-ACPA, 7b and 7e, respectively,
which were prepared from the known 5-hydroxymethylene ace-
tal,18 via a straightforward Williamson ether synthesis, and carried
forward to the amino acids using our previously described
sequence.19
+NH3
+NH3
2 L-Cys2
+NH3
+NH3
4 QA
COO-
COO-
1 CPG
3 CPG
O
R
O
H3C
N
O
S
N
O
N
N
CO2R'
COO-
HN
OH
N
OH
2.2. Chemistry: steric and electronic influence on the
preparation of isoxazole–hydrazones verses isoxazolo[3,4-
d]pyridazinones
+NH3
+NH3
COO-
CO2H
6 AMPA
5 Sulfasalazine
7 ACPA
and analogs
ꢀ
The synthesis of isoxazole–hydrazones proceeds to the open
form only if electron withdrawing groups are present on the ring.20
In our hands, no conditions could be discovered to inhibit the cycli-
zation of aryl containing electron donating groups, and these pro-
ceed to the isoxazolo[3,4-d]pyridazinones 13 ([3,4-d]).21 Variable
temperature NMR studies indicate that the E- to Z-conversion be-
gins at elevated temperature (ca. 60 °C) and the ring closure
appears to be irreversible. Therefore, to deliberately synthesize
the closed [3,4-d] analogues 13 the reactions were conducted at re-
flux.22–32 In those cases where the open forms were isolated, the
best yields were obtained by conducting the reaction at room tem-
perature, and the subsequent hydrolysis at the lowest temperature
practical to effect conversion to the products 11, 16 and 17
(Scheme 1).
The lipophilic analogues listed are only representative, and
prepared using lateral metalation and electrophilic quenching as
previously described by our group.19 After deprotection and hydra-
zone formation E- and Z-isomers of the esters are almost always
observed. We have isolated and crystallized both E-14a (Fig. 4A)
and Z-14b (Fig. 4B), and examined their structures by X-ray crys-
tallography. Complete Crystallographic Information Files (CIF) are
contained in the Supplementary data. The Z-form 14b exhibits an
Chart 1. Ligand structures which bind the System xc transporter.
ethyl-
L
L-cysteine. However, there is
a limit to this trend, as
-homocystine and
L
-djenkolate exhibit reduced activities.12
Several conformationally constrained analogues of Glu also inhi-
bit Sxcꢀ, including quisqualate (QA, 4), 4-S-carboxy-phenylglycine
(4-S-CPG), ibotenate (IBO), (RS)-4-Br-homoibotenate, and (RS)-5-
Br-willardiine.12,14,15 Interestingly, several of these latter ana-
logues are much better known for activities at other iGluRs and
mGluRs, where the conformationally restricted positioning of
their functional groups have presumably increased their specific-
ity of action, as well as added to their value in delineating the
respective binding site pharmacophores for the various receptors.
Among these analogues, the actions of natural products QA and
ꢀ
IBO as Sxc inhibitors highlighted the potential use of isoxazoles
as a scaffold for the development of additional blockers for this
transporter. One of the best known of the isoxazoles is amino-
methyl isoxazole propionic acid (6, AMPA),16 which was the
defining agonist for the GluR1-4 receptor subtypes (a.k.a. AMPA
receptors).1 While AMPA itself exhibits little or no activity at
Sxcꢀ, we have begun to explore the potential inhibitory activity
of other AMPA analogues that have emerged from the pioneering
studies of Krogsgaard-Larsen and co-workers.17 In this work a
series of amino-3-carboxy-5-methylisoxazole propionic acid
ꢀ
(ACPA) analogues is evaluated for inhibitory activity at Sxc
.
Interestingly, we find that the introduction of lipophilic groups
to the ACPA base structure, as exemplified by 7, provides a point
of divergence that distinguishes the binding sites of GluR2 and
Sxcꢀ. This, in turn led us to examine non-amino acid bioisosteres
of ACPA, and we have found that several hydrazone acids (11 and
ꢀ
16, Chart 2) bind to the Sxc with affinities comparable to those
of the endogenous substrates. In contrast, the isoxazolo[3,4-d]
analogues 13 exhibit little or no binding to this transporter.
These novel isoxazole-based analogues are used in combination
with SAR data from other structurally diverse inhibitors (e.g.,
4-S-CPG, sulfasalazine (5)) to construct a pharmacophore model
ꢀ
Scheme 1. Analogs of AMPA: open 11, 16, 17 and closed isoxazolo[3,4-d]pyridaz-
inones 13.
of the Sxc substrate binding site.
Chart 2. Non-amino acid analogs of AMPA/ACPA.

204
S. A. Patel et al. / Bioorg. Med. Chem. 18 (2010) 202–213
almost orthogonal relationship between the isoxazole mean plane
and C@N hydrazone, and the NH is virtually perfectly poised for
cyclization to the [3,4-d] 13. On hydrolysis, the purified Z-forms
convert on standing in solution to the [3,4-d]. Of paramount
importance in comparing the biology of open to closed forms is
the caveat that it was necessary to make the solutions up freshly
before study, and to verify after the assay that the open formed
was indeed still open, which in some cases (i.e., 11a to 13a) occurs
in a matter of a few hours. The E-forms of the hydrazone acids with
electron withdrawing groups appear to be more conformationally
stable.
the apparent affinity of the ligands. To discern if these lipophilic
groups are interacting with distinct transporter domains a hybrid
analogue (16a) with lipophilic moieties at both C4 and C5 of the
isoxazole was prepared. Interestingly, 16a exhibited the same level
of inhibition observed with 11d, suggesting that the two different
aryl groups could be simultaneously accommodated by Sxcꢀ. There
does appear, however, to be some specificity with regard to these
lipophilic domains, as the presence of an additional oxygen in
the linkage significantly diminishes activity (i.e., compare 17 with
16a). More specifically, we previously demonstrated that the corre-
sponding esters of these two acids (14a and 15, respectively)35
adopt similar E-geometries about the hydrazone moiety, and
NMR chemical shifts indicate that this geometry is maintained
within the corresponding carboxylic acids 16 and 17. Worthy of
note is that the open hydrazones, such as 8, 9b, 9c, 10b and 11d
showed no detectable GluR2 binding in Gouaux’s S1S2 construct
filter assay36 (Gouaux, personal communication).
ꢀ
2.3. Inhibition of uptake of glutamate at the System xc
transporter
ꢀ
Inhibitory activity at Sxc was assessed by quantifying the abil-
ity of the analogues to attenuate the uptake of L
-3H]-Glu into hu-
man SNB-19 glioblastoma cells under Cl-dependent (Na-free)
conditions. These cells, as well as other glial tumor-derived lines,
Those compounds exhibiting significant levels of inhibition in
the initial screening assays were characterized in greater detail
by quantifying the concentration-dependence with which the ana-
ꢀ
exhibit much higher levels of Sxc activity than do primary astro-
cytes and are thus well suited for pharmacological assays.12,33
Compounds were initially screened at a single concentration (e.g.,
logues blocked Sxc and calculating corresponding Ki values. These
ꢀ
values, which are summarized in Table 3, were determined for
most of the blockers using the Cheng–Prusoff equation (i.e.,
Ki = (1 + [S]/Km)/[IC50] and IC50 values generated from standard
inhibition curves (see Fig. 1).37,38 To confirm that the inhibition
could be attributed to a competitive mechanism, a few of the ana-
logues, including the most potent inhibitor (S-2-naphthyl-ethyl-
ACPA, 7g), were also examined using a Michaelis–Menten analysis
in which both the concentration of the inhibitor and substrate
were systematically varied. A replot of the resulting Km,apparent val-
ues versus [I] was then used to determine a Ki value (Fig. 2). Con-
sistent with competitive inhibition, these plots revealed an
250 lM or 500 l
M) with Sxcꢀ activity reported as a % of control up-
take: 573 16 pmol/min/mg protein (mean SEM, n = 22). In this
respect, lower values denote greater levels of inhibition. As sum-
marized in Table 1, the first series of isoxazoles examined, which
included the well-known AMPA receptor ligands AMPA (6) and
ACPA (7), exhibited no discernable activity at Sxcꢀ. A similar lack
of activity was also observed with phosphonate derivative of ACPA
(8), as well as those isoxazoles to which small aliphatic groups had
been added to C(5), (7a–7c). In contrast, however, the placement of
aromatic functional groups, such as phenethyl (7d), benzyloxym-
ethyl (7e), and 1- or 2-naphthylethyl (7f and 7g, respectively), at
C(5) yielded amino acids with the capacity to markedly inhibit
the transporter. The increased inhibitory activity of the compounds
suggest that the presence of the larger aromatic moieties enhanced
binding to the transporter either by influencing the relative config-
urations of the functional groups participating in substrate binding
or through direct interaction with lipophilic domains adjacent to
the substrate site, or both. These structure–activity relationships
are also noteworthy in light of the observation that 7d, 7f and 7g
exhibit essentially no ability to displace radiolabeled AMPA in
brain slice experiments.34
increase in the apparent Km without a significant change in Vmax
.
A comparison of the Ki values demonstrate that inhibitors 7g,
11b, and 11e each bind to the transporter with an affinity compa-
rable to that of the endogenous substrate L-Cys2, about 50–60 lM.
Although all of the inhibitors share an isoxazole structure, the
diversity among these ligands also reveals a number of novel prop-
erties of the substrate binding site. Thus, 11b and 11e stand apart
ꢀ
from almost all of the recognized inhibitors of Sxc in that the
hydrazone linkage essentially removed the commonly found free
a-amino and a-carboxylate groups that typify the amino acids.
The aryl groups on these inhibitors, as well as on 7g, also suggest
the presence of lipophilic domains readily adjacent to the substrate
binding site that facilitate the binding of the compounds compared
to ACPA itself. Further, the fact that hybrid analogue 16a includes
aryl additions at both positions and retains inhibitory activity, sug-
gests that these lipophilic groups are interacting with distinct re-
gions of the protein.
ꢀ
This initial group of isoxazoles screened as Sxc inhibitors (Ta-
ble 1) also included modifications to the free amino group of ACPA,
including: semicarbazide hydrazone (9a and 9b), benzene–suf-
onylhydrazone (10a and 10b), phenylsemicarbazide hydrazone
(9c), and arylhydrazone (11e) bioisosteres. While for the most part
inactive, the latter two analogues (9c and 11e) exhibited a moder-
ate amount of inhibitory activity and prompted the synthesis and
characterization of the hydrazone derivatives listed in Table 2.
These compounds also included the corresponding pyridazones
that formed following intramolecular ring closure. In general these
bicyclic [3,4-d] analogues (13a–j) exhibited little or no activity and
in the instances where inhibition was observed, in every case the
structurally open parent hydrazones (11a–f) were markedly more
potent. For example: 11a produced 44% inhibition compared to
26% for 13a; 11b produced 78% inhibition compared to 15% for
13b; and 11e produced 86% inhibition compared to 18% for 13j.
While the results indicate that electron withdrawing groups on
2.4. Substrate activity
While the competition assays discussed above address the spec-
ificity of ligand binding, the results provide little insight into
whether the analogues are acting as alternative substrates or as
non-substrate inhibitors (i.e., compounds that bind but are not
translocated) to competitively block uptake. To examine this we
have developed an exchange assay in which the substrate-induced
efflux of
to NADPH as the extracellular
L
-Glu is quantified by measuring the conversion of NADP+
-Glu leaves the cells and is rapidly
L
ꢀ
the aryl moiety appear to correlate with higher levels of Sxc inhi-
metabolized by glutamate dehydrogenase (GDH) included in the
assay mixture.12,13 Thus, the addition of an Sxc substrate induces
ꢀ
bition, this may be the case only because electron donating groups
favor select diazene tautomers (Scheme 2) and, thus, encourage
cyclization to the less active closed [3,4-d] (13). As was observed
with modifications at C(5) (Table 1), the presence of lipophilic moi-
eties linked to the amino group also have the potential to increase
an efflux of
Representative traces (calibrated with a standard curve) for the
Sxcꢀ-mediated exchange of external
-Cys2 with internal -Glu in
SNB19 cells under Na-free conditions is shown in the Figure 3. In
L-Glu that can be followed fluorometrically in real time.
L
L

S. A. Patel et al. / Bioorg. Med. Chem. 18 (2010) 202–213
205
Table 1
Percent of control uptake of L-[3H]-Glu in the presence of ACPA derivatives
Table 1 (continued)
Compound
Sxcꢀ-mediated 3H-
L-Glu uptake
Compound
Sxcꢀ-mediated 3H-
L-Glu uptake
H3C
H2N
HOOC
O
N
95 2%a
101 6 (n = 3)
COOH
ACPA (racemic), 7
R = CH3
H3C
H2N
HOOC
O
N
98 12 (n = 3)
COOH
107 5 (n = 3)
(S)-ACPA 7,
R = CH
3
92 .5 (n = 3)
98 4 (n = 3)
115 12 (n = 3)
70 5 (n = 3)
70 6 (n = 4)
47 2 (n = 4)
37 4 (n = 4)
103 4 (n = 3)
96 5 (n = 3)
91 8 (n = 3)
101 15 (n = 3)
73 3 (n = 3)
O
N
O
N
COOH
76 5 (n = 3)
N
H
N
H
Phenyl-Semicarbazide Hydrazone
Acid, 9c
SNB-19 cells were assayed for
L
-[3H]-Glu (100
l
M) uptake under Cl-dependent (Na-
free) conditions in the presence of ACPA derivatives (250
l
M). Values are reported
as mean SEM (n P 3) of control activity (accumulation in the absence of inhibi-
tors, i.e. 100%). Rates of uptake in these control assays was 573 16 pmol/min/mg
protein (mean SEM, n = 22).
a
AMPA was previously assessed for inhibitory activity at 500
lM in LRM55
cells12 and is included for comparative purposes.

206
S. A. Patel et al. / Bioorg. Med. Chem. 18 (2010) 202–213
Table 2
Table 2 (continued)
Percent of control uptake of L-[3H]-Glu in the presence of isoxazole–hydrazone
derivatives
Compound
Sxcꢀ-mediated 3H-
uptake
L-Glu
Compound
Sxcꢀ-mediated 3H-
uptake
L-Glu
N
O
O
H3C
CO2H
N
N
N
74
85
96
94
4
1
2
3
O
N
O
H3C
56
2
H3C
N
13a, X=Y=Z= H
H3C
H
N
11a, X=Y=Z= H
O
H3C
CO2H
N
N
NO2
NO2
N
O
NO2
H3C
N
22 0.3
H3C
13b, X=Y= H, Z= NO2
N
H
H3C
N
O
O
11b, X=Y= H, Z= NO2
H3C
N
CO2H
N
N
O
H3C
N
13h
H3C
N
97
2
H3C
H
O2N
N
O
O
11c, X= NO Y=Z= H
2,
H3C
N
CH3
N
CO2H
N
H3C
O
NO2
13c X=Y= H, Z= CH3
N
H3C
N
45
86
1
2
H3C
O
F
H
N
O
O2N
N
F
11d, X= Z= NO2, Y= H
H3C
95
92
3
4
N
H3C
N
O
N
O
13g
13d
N
F
H3C
O
O
N
H3C
H3C
N
OCH3
CF3
N
13f
H3C
O
O
N
O
O
N
N
N
CF3
H3C
99
14
2
4
N
H3C
H3C
82
6
N
13e
H3C
CF3
Cl
CO2H
N
CF3
13j
13i
O
O
N
N
O
H3C
N
CF3
H3C
H
N
H3C
89
13
8
1
N
11e
H3C
Cl
N
CO2H
O
H
H2N
S
H2N
H
S
N
HN
HOOC
COOH
NO2
H3C
45
2
L-Cystine
O2N
SNB-19 cells were assayed for
L
-[3H]-Glu (100
l
M) uptake under Cl-dependent (Na-
16a, 2-Naphthyl ethyl
free) conditions in the presence of isoxazole–hydrazone derivatives (500
l
M).
Values are reported as mean SEM (n P 3) of control activity (accumulation in the
absence of inhibitors, i.e. 100%). Rates of uptake in these control assays was
573 16 pmol/min/mg protein (mean SEM, n = 22).
N
CO2H
O
O
N
HN
NO2
H3C
95
6
O2N
contrast to
did not induce an efflux of
L
-Cys2, the addition of S-2-naphthyl-ethyl-ACPA (7g)
-Glu over background levels, indicating
L
17, 2-Naphthyl methoxy-
Methyl
that it is not a substrate of the transporter. Consistent with the
action of a competitive inhibitor, however, the inclusion of 7g

S. A. Patel et al. / Bioorg. Med. Chem. 18 (2010) 202–213
207
O
O
O
R
O
R
R
R
N
N
N
N
H3C
HN
CO2R'
H3C
CO2R'
H3C
CO2R'
NH
O
H3C
N
N
N
N
X
N
N
Y
X
Y'
X
Y'
X
Y'
Y
Y
Z
Y
Z
Y'
Z
Z
13
19
18
11
Scheme 2. E- to Z-conversion of 11, 16, and 17 likely occurs via tautomerization to a diazene 18, and the Z-geometry 19 can close rapidly to the isoxazolo[3,4-d] pyridazinone
13.
Table 3
Ki values for the competitive inhibition of Sxcꢀ-mediated uptake of L-Glu
Compound
Ki (lM)
N
CO2H
O
O
7f
S-1-Napthyl-
117 15
ethyl-ACPA
CO2H
CO2H
H2N
N
7g
S-2-Napthyl-
52
5
ethyl-ACPA
CO2H
H2N
CO2H
N
O
NO2
11b
11e
60
64
2
6
N
Figure 1. IC50 determination for selected hydrazone inhibitors. SNB-19 cells were
assayed for
-[3H]-glutamate uptake under Cl-dependent (Na-free) conditions in the
presence of a range of inhibitor concentrations (10–1000 M). Data are plotted as
H3C
N
L
H3C
H
l
CO2H
CF3
mean SEM of control activity (accumulation in the absence of inhibitors, i.e.,
100%) and were pooled from at least three different assays. IC50 values were
determined by non-linear curve fitting (KaleidaGraph) and are reported in the inset.
N
O
N
H3C
N
H
CF3
NO2
H3C
CO2H
N
O
11d
16a
147 26
N
N
H3C
H3C
H
O2N
N
CO2H
O
N
HN
113
7
NO2
H3C
O2N
H
H2N
S
H2N
H
59 10
L
-Cystine
S
HOOC
COOH
SNB-19 cells were assayed for
L
-[3H]-glutamate uptake under Cl-dependent (Na-
free) conditions in the presence of range of inhibitor concentrations (10–
a
1000 M). Ki values were determined in one of to ways: (i) directly from a replot of
l
Km,app values determined by non-linear curve fitting of saturation curves (see Fig. 2)
or from IC50 values, determined in inhibition curves (see Fig. 1), using the Cheng–
Prusoff equation.37,38 Values are reported as mean SEM, n P 3.
did attenuate the ability of
Glu. When each of the identified isoxazole inhibitors listed in Table
3 was tested in this manner, none produced an efflux of -Glu that
was significantly different from background. These finding suggest
that while the substrate binding site can accommodate the vari-
ously modified isoxazoles, the compounds carrying these lipophilic
additions cannot be translocated across the membrane.
L-Cys2 to exchange with intracellular L-
Figure 2. Representative Michaelis–Menten analysis and Ki determination for S-2-
naphthyl-ethyl-ACPA. SNB-19 cells were assayed for
50, 100, 250, 500 M) under Cl-dependent (Na-free) conditions in the presence of a
range of inhibitor concentrations (20, 40, 200 M). Data are plotted as pmol/mg/ mg
L
-[3H]-glutamate uptake (20,
L
l
l
protein and have been corrected for non-specific uptake and leakage. Km,app values
were determined by non-linear curve fitting of the saturation curves (Kaleida-
Graph). Ki values were determined from a replot of Km,app versus [I] as shown in the
inset.

208
S. A. Patel et al. / Bioorg. Med. Chem. 18 (2010) 202–213
ꢀ
2.5. Computational modeling of the Sxc binding domain:
preliminary pharmacophore model
As a suitable crystal structure is lacking for the development of
ꢀ
an Sxc protein homology model, our approach has involved the
construction of a ligand-based, superposition, three-dimensional
(3D) pharmacophore model. Such models account for structural
and functional group similarities in addition to select low energy
ligand conformations that have been positioned in 3D space and
ꢀ
can be correlated to the SAR data generated from the Sxc trans-
port assays. A similar approach has been successfully employed
for delineating potent and selective ligands for the EAATs.39,40
The ligand training set (Fig. 5, Panel A) included the structurally di-
verse endogenous substrates L-Glu and L-Cys2 and the potent inhib-
itors quisqualate (QA) and (S)-4-carboxyphenyl-glycine (4-S-CPG).
The latter two provided elements of structural rigidity to the cen-
tral pharmacophore core region.12 The model employs both Sxc
ꢀ
substrates and non-substrate inhibitors, thereby, allowing it to be
consistent with the fundamental hypothesis that an obligate ex-
change transporter operating with an alternate access mechanism
will employ a single substrate binding site that may be defined
with discrete overlapping binding domains for the various ligands.
Conformational space searching of the training set ligands affor-
ded global energy minima for the substrates and inhibitors that
Figure 3. Representative fluorescent traces of Sxcꢀ-mediated exchange of extra-
cellular substrates for intracellular L-glutamate in SNB-19 cells. Representative
fluorometric assays following the substrate-mediated efflux of glutamate from SNB-
19 cells under Na+-free conditions as described in Methods. The traces were
calibrated using a standard curve in which aliquots of
the assay mixture in the absence of the SNB-19 cells. The addition of
(500 M) produced a significant efflux of -glutamate over baseline (absence of any
addition), while S-2-naphthyl-ACPA (500 M) did not. Consistent with a non-
L-glutamate were added to
L
-cystine
l
L
were then aligned in 3D space to maximize overlap of the
-COOH, -NH2, and distal COOH groups or related isosteric moi-
eties. Alignment of the training set -amino acid head groups af-
a-C,
l
substrate inhibitor, S-2-naphthyl-ACPA did, however, attenuate L-cystine-induced
efflux when both were added simultaneously.
a
a
L-a
fords the initial composite superposition pharmacophore model
shown in Figure 5, Panel B. A key model aspect is the pseudo
chair-like conformations of the substrate L-Cys2 (Panels B–D),
which are thought to correlate to the central core ring structures
of QA and 4-S-CPG. With the amino acid head groups aligned, the
model suggests the likelihood that there may be two ways in
which the binding site can accommodate the distal COOH group
(or isosteric moieties) as represented in Panels B–D. Although the
distal COOH of
upon rotation about the C6–7 bond (Fig. 5, Panels C and D), we
have utilized the -Cys2 conformation per Panel C in the model,
L-Cys2 could reach either of these positions, based
L
since it is consistent with a conformation in which intramolecular
interactions between the 4-position sulfur atom and a proton of
the C7 terminal amine group afford an intramolecular hydrogen
bonding interaction (depicted in Panel A, ligand 2) that is sup-
ported by NMR spectroscopy.41,42 The distances between the vari-
ous COOH binding domains (e.g., Panels C or D as C–C distances
between the respective COOH carbon atoms) are distinct from
those of
the inactivity of
-aminoadipate as a substrate.12 It is also notable that the confor-
L
-aspartate and
L-a-aminoadipate, further rationalizing
L-aspartate as an inhibitor and the activity of
L-
a
mational rigidity of the central rings of 4-S-CPG and QA may pro-
ꢀ
vide more optimal Sxc binding as reflected by lower Ki values
(e.g., ꢁ5
lM). The lack of flexibility due to the rings, however,
may also account for decreased substrate activity.12 Thus, 4-S-
CPG is essentially a non-substrate inhibitor and QA exchanges with
L
-Glu at a rate of only 35% that of
Three compounds of interest were incorporated into the initial
pharmacophore model, 7g and 11e (both of which exhibited Ki val-
ues comparable to -Cys2), as well as 16a, to identify previously
L-Cys2.
L
unrecognized regions of the transporter protein that are able to
accommodate lipophilic groups and steric bulk. Identifying and
optimizing such interactions has been key to the development of
high-affinity ligands for the EAATs.43 The challenge in trying to
understand the relative position these lipophilic substituents with
respect to the Figure 5 Panels A–D model lies with deciding how to
appropriately pair the two COOH groups of these newer com-
pounds with the
lead inhibitors, only 7g possesses an L-a
a
- and distal COOH model moieties. Of the three
-amino acid head group.
Figure 4. Panel A, X-ray structure of E-14a. Panel B. X-ray structure of Z-14b.

S. A. Patel et al. / Bioorg. Med. Chem. 18 (2010) 202–213
209
O
O
O
O
-
-
-
-
H
H
O
H
+
O
O
H
O
+
+
+
NH3
O
NH3
3
NH3
NH
H
O
S
S
7
N
N H
+
O
6
O
N
H
H
-
O
H
HO
O
O
O
Inhibitors
Substrates
L-Cys2
(2, chair form)
Quisqualate
(4, QA)
L-Glu
(1)
4-S-CPG
(3)
A
C
B
D
E
F
G
Head group
L-Amino acid
A
Lone pair
electrons or
Sulfur electrons
or aromatic ring
π-Bonds
hydrogen bonding
groups
E
B
Core
Ring
Aromatic F
groups
#1
C
Hydrogen
bonding group
D
Distal
carboxyl or
heteroatom
group
G
Aromatic
groups
#2
H
I
ꢀ
Figure 5. A ligand-based, superposition, three-dimensional (3D) pharmacophore model of the Sxc binding site. Panel A: ligand training set. Panels B–D: L-Glu, 1, yellow; L-
Cys2, 2, aqua; 4-S-CPG, 3, purple; quisqualate, 4, green. New lead compounds: Panel E, 7g, violet; Panel F, 7g, violet and hybrid 16a, tan; Panel G, hybrid 16a, tan and 11e, light
green. Panel H: sulfasalazine, 5, lipophilic group orientations, #1, purple and #2, tan. Panel I, summation pharmacophore cartoon model.
When the
a
-C,
a
-COOH and
a
-NH2 of 7g are aligned with the cor-
acknowledged that regions occupied by the respective lipophilic
moieties could be reversed resulting in altered alignments of the
COOH groups and related isosteric moieties.
responding groups of the initial superposition model, the naphthyl
ring moiety occupies the position as shown in Panel E of Figure 5.
When the naphthyl moiety position of 7g is used as a common 3D
descriptor for the same naphthyl moiety type of 16a, then the 3-
COOH group extending from the isoxazole ring of 16a is correlated
Similar alignment ambiguity arises when the more recently
identified inhibitor sulfasalazine 5 (5-[4-(2-pyridylsulfamoyl)pheny-
lazo]salicylic acid) is incorporated into the model. This compound
has attracted considerable attention because it provides a potential
to the distal COOH group of L-Glu within the model, resulting in a
ꢀ
second lipophilic model domain defined by the region occupied by
the 2,4-dinitrophenyl moiety, as shown in Figure 5, Panel F. In a
similar way, alignment of the 3,5-bis-trifluoromethyl-phenyl
group of 11e with the 2,4-dinitrophenyl group of 16a aligns the
3-COOH group extending from the isoxazole ring of 16a to the dis-
therapeutic link between Sxc inhibition and the treatment of CNS
glial tumors.5,44 Thus, the pyridylsulfamoylphenyl portion of sulfa-
salazine can be correlated to either of the lipophilic regions defined
by 7g, 11e, or 16a depending upon whether the COOH group of the
salicylic moiety of 5 is aligned with the
-Glu (Fig. 5, Panel H). Since either model alignments are possible
with sulfasalazine, then it is thought that the lipophilic domains
a- or distal COOH group of
tal COOH moiety of
philic group alignments are consistent to each other, it is
L
-Glu (Fig. 5, Panel G). Although the above lipo-
L

210
S. A. Patel et al. / Bioorg. Med. Chem. 18 (2010) 202–213
ꢀ
extending beyond the core Sxc substrate binding area imparts en-
4. Experimental section
ꢀ
hanced Sxc binding. Thus, while the parent analogues (e.g., ACPA
(7), S-ethyl-ACPA (7a)) exhibit little or no activity, the inclusion of
Commercial reagents are routinely examined for purity by NMR
and TLC, and recrystallized or distilled as appropriate. All reactions
were monitored by TLC. NMR was performed on a Varian Unity
Plus spectrometer at (400 MHz for 1H, 101 MHz for 13C) in deute-
rochloroform unless otherwise noted. Chemical shifts (d) are re-
ported using CHCl3 (7.26 ppm for 1H), CDCl3 (77 ppm for 13C) as
references. High resolution mass spectra (HRMS) were obtained
using a Micromass electrospray ionization (ESI)/time-of-flight
mass spectrometry (LC–TOF). Mass spectrometer samples were
introduced using a Waters model 2690 separations module HPLC
fitted with a C-18 reversed phase column (2.1 mm id, 5 cm). Ele-
mental analyses for C, H, and N were performed by Midwest Micro-
lab, Indianapolis, IN. Melting points were uncorrected.
lipophilic groups with sufficient
p-electron character and steric
bulk at the C5 of the isoxazole ring or via a hydrazone linkage to
the free amino group of ACPA, or both, markedly enhances the abil-
ꢀ
ity of these ligands to bind to the Sxc site.
The ligand alignment computational modeling and SAR in-
sights have allowed us to generate an initial perspective of an
ꢀ
Sxc pharmacophore model. A composite cartoon representation
of the model is shown in Figure 5, Panel I. The model central
core ring area is consistent with the notion that an alternate ac-
cess transporter can be defined with discrete overlapping binding
domains for the various ligands. The model regions that surround
the central core ring area include: (a) the
group; (b) an electron lone pair region or a hydrogen bonding
group, equivalent to the distal COOH group of -Glu or a hetero-
atom of the isoxazole ring of the analogues; (c) a hydrogen bond-
ing group, equivalent to the distal -amino acid moiety of -Cys2
L-amino acid head
L
4.1. Chemistry: amino acid analogues of AMPA
Amino acids, racemic ACPA (( )-7),45
(
)-8 (S)-ACPA, ((ꢀ)-7),
L
L
7d, 7g, 7f,19 7a and (S,S)-7c18 were prepared using methods previ-
ously described by our laboratories, and full experimental details
have been published. (S)-MOM-ACPA (7b) and BOM-ACPA (7e)46
were prepared from 12 via the known 5-hydroxyl-methyl acetal18
using the amino acid synthesis sequence previously described.19
or a second heteroatom of an isoxazole ring within an analogue,
(d) a distal carboxyl or heteroatom area, equivalent to the distal
COOH group of
L-Cys2 or 4-S-CPG; (e) lone pair electron or
p-
electron density, associated with the sulfur atoms of
L-Cys2 or
the aromatic ring of 4-S-CPG; (f) an aromatic ring group region
#1 which is represented by analogue 7g; and (g) an aromatic
ring group region #2 which found in analogue 16a. Importantly,
not all of the pharmacophore regions A–G need to be represented
4.1.1. 4-[(2S)-2-Amino-2-carboxyethyl]-5-[methoxymethyl] isox-
azole-3-carboxylic acid (7b, (S)-MOM-ACPA)
ꢀ
in each ligand to produce significant Sxc binding. Additionally,
Overall yield 12% for four steps, 7b was obtained as 60 mg of a
yellowish solid, mp = 166–168 °C (dec.). 1H NMR (D2O) d 3.25 (dd,
1H), 3.38 (dd, 1H), 3.41 (s, 3H), 4.29 (t, 1H, J = 6.8 Hz), 4.65 (s,
2H).13C NMR d 23.8, 53.6, 59.4, 63.8, 113.0, 157.3, 164.0, 169.8,
171.9 FAB+ (m-glycerol) m/z 245 [M+1]+ (38.8% rel. intensity-glyc-
erol matrix gave base peak at m/z 74.79). Anal. Calcd C9H13N2O6
[M+1]+ mass: 245.077361. Accurate mass: FAB+ 245.0769. Optical
ꢀ
higher affinity Sxc ligands may possess different select combina-
tions of the model A–G structural facets.
As previously mentioned, the increased inhibitory activity of
EAAT ligands possessing lipophilic groups has been similarly
used to infer the presence lipophilic domains adjacent to trans-
porter substrate binding sites. Interestingly, for both the EAAT
rotation: ½a 2D0
ꢀ5.3 (c 0.15, H2O). Anal. Calcd for C9H13N2O6Cl: C,
ꢂ
ꢀ
and Sxc transporters, the interactions between these lipophilic
domains and the ligands is also associated with a decreased abil-
ity of the appended analogues to act as alternative substrates.
Taken together, our initial SAR-based pharmacophore model sup-
38.51; H, 4.66; N, 9.98. Found: C, 38.51; H, 4.52; N, 9.73.
4.1.2. 4-[(2R,S)-2-Amino-2-carboxyethyl]-5-[benzyloxymethyl]
isoxazole-3-carboxylic acid (7e, BOM-ACPA)
ꢀ
ports a single site hypothesis for Sxc in which functional groups
on identified substrates and inhibitors most likely interact with
different subsets of available binding domain residues or the
same residues in the binding site altered with different relative
conformations. This is perhaps best illustrated by inhibitors such
as 11e and sulfasalazine, which may be binding to Sxc in a man-
ner that does involve the L-a-amino acid head group area, and
Overall yield 7% for four steps, 7e was obtained as 40 mg of a
yellowish solid, mp = 166–168 °C (dec.). 1H NMR (DMSO-d6) d
2.97 (dd, 1H), 3.16 (dd, 1H), 3.92 (t, 1H, J = 5.75 Hz), 4.55 (s, 2H),
4.65 (q, 2H, J = 13.05 Hz), 7.37 (m, 5H).13C NMR (126 MHz) d
23.5, 30.5, 33.02, 52.0, 58.5, 60.8, 63.4, 71.8, 110.8, 127.5, 128.1,
161.3, 170.1 FAB+ (m-glycerol/DMSP) m/z 321 [M+1]+, 277, 219,
91.6 Anal. Calcd for C15H16N2O6 (1.5 H2O): C, 51.87; H, 5.51; N,
8.06. Found: C, 51.48; H, 4.81; N, 8.80. HRMS calcd C15H17N2O6
[M+1]+: 321.1087, found: FAB+ 321.1073.
ꢀ
that also is defined by the endogenous
L-Glu and L-Cys2 sub-
strates for the transporter.
4.2. Chemistry bioisosteres of AMPA
3. Conclusions
Common intermediate 12 was prepared as described by dal
Piaz.21 Isoxazolo[3,4-d]pyridazinone 13a is known, also from that
group’s extensive work in the area.22–32 Compounds 13b–d and
13f–g, and 13j are reported in Chemical Abstracts, and are com-
mercially available from Ryan Scientific, Inc., but are not refer-
enced to Patent or Publication literature, therefore, the CAS
registry number and full characterization data is provided for each
in the Supplementary data. Open hydrazones 11a–e were prepared
in analogous fashion as previously described by our laboratories.20
In summary, analogues of AMPA have been prepared that bind
the Sxc transporter, with the most efficacious analogues amino
ꢀ
acid 7g and bioisostere 11e having comparable activity to the
endogenous substrate cystine. Both classes of compounds appear
to exhibit competitive binding. The isoxazolo[3,4-d] pyridazinones
ꢀ
13, in contrast exhibit only very weak binding at the Sxc antipor-
ter, although examination of their binding at other stages of the
glutamate–glutamine cycle could be worthwhile. Experiments to
address these and related model-derived questions are planned
to significantly advance the discovery paradigm. Future targets of
ꢀ
second generation refined Sxc inhibitor analogues will further
4.3. 5-Methyl-4-[1-(phenyl-hydrazono)-ethyl]-isoxazole-3-
carboxylic acid (11a)
elucidate which combinations of regions of the pharmacophore
ꢀ
model provide the greatest Sxc binding potency. Of particular
interest will be the further characterization of the lipophilic ligand
domains discovered in the present study.
Light brown solid; yield = 60%; mp = 138 °C; 1H NMR (CD3OD) d
2.18 (s, 3H), 2.58 (s, 3H), 6.78 (t, J = 8 Hz, 1H), 7.12 (d, J = 8.0 Hz,

S. A. Patel et al. / Bioorg. Med. Chem. 18 (2010) 202–213
211
2H), 7.19 (t, J = 8 Hz, 2H); 13C NMR (CD3OD) d 12.3, 16.6, 114.2 (3C),
120.8, 130.1 (3C), 134.9, 147.3, 163.5, 170.0; MS (ES+) 260 (M+1),
HRMS calcd for C13H14N3O3 [M+H]+ 260.1035, found 260.1021.
4.10. 3,4-Dimethyl-6-(3,5-dichlorophenyl)isoxazolo[3,4-d]pyri-
dazin-7(6H)-one (13i)
White solid; yield = 93% mp = 230–232 °C; 1H NMR (400 MHz,
CDCl3) d 2.56 (s, 3H), 2.89 (s, 3H), 7.34 (s, 1H), 7.59 (s, 2H); 13C
NMR (100 MHz, CDCl3) d 13.5, 19.8, 112.5, 124.5 (2C), 128.1 (2C),
135.1, 141.2, 142.4, 152.4, 153.0, 170.7; MS (ES+) 310 (M+), 312
(M+2); HRMS calcd for C13H10N3O2Cl2 [M+H] 310.0150, found
310.0190.
4.4. 5-Methyl-4-{1-[(4-nitro-phenyl)-hydrazono]-ethyl}-isoxazole-
3-carboxylic acid (11b)
Yellow solid; yield = 72%; mp = 240 °C; 1H NMR (acetone-d6) d
2.36 (s, 3H), 2.61 (s, 3H), 7.93 (d, J = 8.0 Hz, 2H), 8.44 (d,
J = 8.0 Hz, 2H); 13C NMR (acetone-d6) d 11.7, 16.1, 111.8, 112.2
(2C), 125.9 (2C), 138.8, 140.1, 151.2, 155.0, 161.4, 169.1; MS
(ES+) 305 (M+1), HRMS calcd for C13H13N4O5 [M+H]+ 305.0886,
found 305.0901.
4.11. 4-{1-[(2,4-Dinitro-phenyl)-hydrazono]-ethyl}-5-(2-naph-
thalen-2yl-ethyl)-isoxazole-3-carboxylic acid ethyl ester (14)
To a solution of ketone19 (0.56 g, 1.66 mmol) in THF (10 ml),
2,4-DNP reagent (10.9 ml, 1.66 mmol) was added drop wise at
0 °C. The mixture was stirred at room temperature for 6 h. Precip-
itate formed was filtered and purified by column chromatography
(30% ether in hexanes). The major product was obtained as orange
colored solid (14a), yield = 81%, mp = 160 °C; 1H NMR d 1.43 (t,
J = 7.1 Hz, 3H), 1.93 (s, 3H), 3.26 (t, J = 6.8 Hz, 2H), 3.40 (t,
J = 6.2 Hz, 2H), 4.44 (q, J = 7.1 Hz, 2H), 7.17 (dd, J = 8.4 & 1.7 Hz,
1H), 7.44 (m, 2H), 7.52 (m, 1H), 7.65 (m, 2H), 7.72 (d, J = 8.4 Hz,
1H), 7.76 (m, 1H), 8.14 (dd, J = 9.5 and 2.6 Hz, 1H), 9.14 (d,
J = 2.5 Hz, 1H), 10.98 (br s, 1H, NH). 13C NMR d 14.1, 17.2, 27.9,
34.0, 62.6, 116.2, 116.7, 123.3, 125.9, 126.4, 126.6, 126.8, 127.2,
127.6, 128.3, 129.9, 130.1, 132.2, 133.4, 136.8, 138.6, 144.5,
144.9, 153.9, 159.9, 172.2; MS (FAB): m/z 518(33), 517 (100), 501
(12), 248 (11), 154 (34), 141 (61), 136 (23).
4.5. 5-Methyl-4-{1-[(2-nitro-phenyl)-hydrazono]-ethyl}-isoxazole-
3-carboxylic acid (11c)
Orange solid; yield = 97%; mp = 168–169 °C; 1H NMR (acetone-
d6) d 2.36 (s, 3H), 2.65 (s, 3H), 6.96 (m, 1H), 7.64 (m, 1H), 7.89 (d,
J = 8.0 Hz, 1H), 8.18 (d, J = 8.0 Hz, 1H), 10.86 (br s, 1H); 13C NMR
(acetone-d6) d 11.6, 16.0, 116.0, 118.8 (2C), 125.9 (2C), 136.6
(2C), 141.2, 154.9, 161.3, 169.6; MS (ES+) 305 (M+1), HRMS calcd
for C13H13N4O5 [M+H]+ 305.0886, found 305.0903.
4.6. 4-{1-[(2,4-Dinitro-phenyl)-hydrazono]-ethyl}-5-methyl-
isoxazole-3-carboxylic acid (11d)
Anal. Calcd for C26H23N5O7: C, 60.35; H, 4.48; N, 13.53. Found: C,
60.51; H, 4.62; N, 13.41. Minor product (14b), yellow solid, 1H NMR
d 1.30 (t, J = 8.0 Hz, 3H), 2.08 (s, 3H), 3.21 (m, 4H), 4.32 (q,
J = 8.0 Hz, 2H), 7.16 (dd, J = 8.0 and 4.0 Hz, 1H), 7.30 (m, 2H), 7.44
(m, 1H), 7.58 (m, 3H), 7.76 (d, J = 8.0 Hz, 1H), 8.20 (dd, J = 8.0 &
4.0 Hz, 1H), 8.84 (d, J = 4.0 Hz, 1H), 10.37 (br s, 1H, NH).
Red solid; yield = 66%; mp = 189–191 °C; 1H NMR (CD3OD) d
2.39 (s, 3H), 2.64 (s, 3H), 7.96 (d, J = 8.0 Hz, 1H), 8.38 (d,
J = 8.0 Hz, 1H), 9.07 (s, 1H); 13C NMR (CD3OD) d 12.4, 17.1, 70.6,
117.2, 117.6 (2C), 124.1 (2C), 131.2 (2C), 146.2, 146.9, 171.5; MS
(ES+) 350 (M+1), HRMS calcd for C13H12N5O7 [M+H]+ 350.0737,
found 350.0767. Anal. Calcd for C13H12N5O7 (1.5 H2O): C, 41.49;
H, 3.75; N, 18.61. Found: C, 41.46; H, 3.47; N, 18.01.
4.12. 4-{1-[(2,4-Dinitro-phenyl)-hydrazono]-ethyl}-5-(2-
naphthalen-2-yl-ethyl)-isoxazole-3-carboxylic acid (16a)
To a solution of 14a (0.65 g) in THF, 2 ml of 3 M NaOH was
added drop wise at 0 °C. After 0.5 h stirring the disappearance of
ester was observed. The reaction mixture was acidified with 1 M
HCl to pH 2 and extracted with EtOAc. The organic extracts were
then washed with saturated NaHCO3 followed by brine and finally
dried over sodium sulfate. The organic layer was filtered and evap-
orated. The residue obtained was purified by column chromatogra-
phy with 5% MeOH in CHCl3 as eluent. The pure product was
obtained as yellow solid, yield = 67%, mp = 169–171 °C; 1H NMR
(acetone-d6): d 2.10 (s, 3H), 3.27 (t, J = 8.0 Hz, 2H), 3.50 (t,
J = 8.0 Hz, 2H), 7.34 (dd, J = 8.0 and 4.0 Hz, 1H), 7.41 (m, 2H), 7.66
(m, 1H), 7.71 (m, 1H), 7.79 (m, 2H), 7.88 (d, J = 8.0 Hz, 1H), 8.29
(dd, J = 8.0 and 4.0 Hz, 1H), 9.00 (m, 1H), 10.97 (br s, 1H, NH). 13C
NMR (acetone-d6) d 16.7, 30.4, 42.0, 94.6, 115.6 (2C), 123.0 (2C),
125.5 (2C), 126.2, 126.7 (2C), 127.5, 127.6, 127.8, 128.2 (2C),
130.5 (2C), 133.9, 134.0, 139.0, 172.7; MS (FAB): m/z 492 (9.0),
491 (33), 490 (100), 489 (32), 460 (27), 446 (51), 445 (25). Anal.
Calcd for (C24H19N5O7)2.H2O: C, 57.83; H, 4.04; N, 14.05. Found:
C, 57.74; H, 3.58; N, 13.80.
4.7. 4-{1-[(3,5-Bis-trifluoromethyl-phenyl)-hydrazono]-ethyl}-
5-methyl-isoxazole-3-carboxylic acid (11e)
Off white solid; yield = 95%; mp = 137–139 °C; 1H NMR (ace-
tone-d6) d 2.28 (s, 3H), 2.59 (s, 3H), 7.33 (s, 1H), 7.71 (s, 2H), 9.41
(br s, 1H); 13C NMR (acetone-d6) d 11.5, 16.4, 111.5, 112.6 (2C),
116.6, 131.9 (2C), 138.7 (2C), 147.5, 156.3, 162.3, 168.5, 185.5;
MS (ES+) 396 (M+1), HRMS calcd for C15H12N3O3F6 [M+H]+
396.0783, found 396.0801.
4.8. 3,4-Dimethyl-6-(4-(trifluoromethyl)phenyl)isoxazolo[3,4-
d]pyridazin-7(6H)-one (13e)
Brown crystalline solid; yield = 28% mp = 161–163 °C; 1H NMR d
2.56 (s, 3H), 2.90 (s, 3H), 7.72 (d, J = 8 Hz, 2H), 7.78 (d, J = 8 Hz, 2H);
13C NMR d 13.5, 19.8, 37.0, 112.6, 126.2, 129.7, 129.9, 141.1, 143.8,
152.5, 153.2, 170.6; MS (ES+) 310 (M+1), HRMS calcd for
C14H11N3O2F3 [M+H]+ 310.0803, found 310.0771.
4.9. 3,4-Dimethyl-6-(3-nitrophenyl)isoxazolo[3,4-d]pyridazin-
7(6H)-one (13h)
4.13. 4-{1-[(2,4-Dinitro-phenyl)-hydrazono]-ethyl}-5-(naphtha-
len-2-ylmethoxymethyl)-isoxazole-3-carboxylic acid (17)
White solid; yield = 96% mp = 187–188 °C; 1H NMR d 2.59 (s,
3H), 2.91 (s, 3H), 7.63 (m, 1H), 8.07 (m, 1H), 8.21 (m, 1H), 8.54
(s, 1H); 13C NMR d 13.5, 19.8, 112.6, 121.1, 122.5, 129.6, 131.7,
141.6, 141.8, 152.4, 153.2, 170.8; MS (ES+) 287 (M+1), HRMS calcd
for C13H11N4O4 [M+H]+ 287.0780, found 287.0785.
Yield: 60% The major isomer 1H NMR (DMSO) d 2.34 (s, 3H),
4.74 (s, 2H), 4.89 (s, 2H), 7.65 (d, 1H, J = 9.5 Hz), 7.80 (m, 4H),
7.91 (m, 3H), 8.15 (dd, 1H, J = 2.4, 9.5 Hz), 8.79 (d, 1H, J = 2.6 Hz),
10.87 (br s, 1H, NH). 13C NMR (500 MHz) d 15.3, 72.3, 73.4,
102.0, 120.9, 125.6, 125.7, 125.8, 125.9, 127.3, 127.5, 127.6,

212
S. A. Patel et al. / Bioorg. Med. Chem. 18 (2010) 202–213
127.7, 128.2, 129.6, 132.3, 132.6, 134.7, 135.6, 144.6, 146.2, 149.8,
magnetic stirrer platform. Confluent monolayers of SNB-19 glioma
cells grown on coverslips were paired, placed back to back and
rinsed in Na+-free HBHS buffer for 10 min. The coverslips were
transferred and incubated in a stirred quartz cuvette containing
2 ml HBHS at 30 °C. Assays were initiated with the addition of
NADP+ (1 mM) at 0 s followed by GDH (33 U) at 60 s. Subsequent
152.7, 190.8. Accurate mass Calcd for C24H20N5O8 [M+1]+
506.131187. Found: FAB+ 506.131188.
:
The minor isomer 1H NMR (DMSO) d 2.29 (s, 3H), 4.51 (s, 2H),
4.78 (s, 2H), 7.25 (d, 1H, J = 9.5 Hz), 7.42 (m, 4H), 7.54 (m, 3H),
8.40 (dd, 1H, J = 2.4, 9.5 Hz), 8.89 (d, 1H, J = 2.6 Hz), 10.55 (br s,
1H, NH). FAB+ m/z 506 [M+1]+, 490, 349, 306, 288, 272.
additions of L-cystine or other potential substrates (50–500 lM)
were made as detailed in the figure legends. Each experiment
was terminated with the addition of Triton X-100 (0.5%) to deter-
mine the total amount of L-Glu in the cells. The amount of L-Glu ef-
4.14. Cell culture
flux was determined via a standard curve generated in an identical
experiment but omitting the cells. Following the enzyme, seven
SNB-19 glioma cells, purchased from American Type Culture
Collection (Manassas, VA), were grown in Ham’s F-10 medium
(pH 7.4) containing 1 mM pyruvate and 16 mM NaHCO3 and sup-
plemented with 10% fetal calf serum. The cells were cultured in
150 cm2 flasks (Corning) and maintained at 37 °C in a humidified
sequential additions of 4 nmol
ear curve. Rates of -Glu efflux are reported as mean S.E.M and
are corrected for non-specific efflux (e.g., leakage) by subtracting
the amount of -Glu efflux in the absence of -cystine.
L-Glu were made yielding a non-lin-
L
L
L
atmosphere of 5% CO2. For L-glutamate uptake experiments, cells
were seeded in 12 well culture plates (Costar) at a density of
5 ꢃ 104 cells/well and for fluorometric assays cells were seeded
on 9.5 ꢃ 22 mm coverslips at a density of 1 ꢃ 105 cells/coverslip
and maintained for 5–8 days until 80–90% confluent. Cells formed
a confluent monolayer with an estimated density of 1–2.5 ꢃ
106 cells and a protein concentration of 100–350 mg/well or cover-
slip as determined by the bicinchoninic acid (BCA) method (Pierce).
Cells were given fresh medium 2 h prior to an assay.
4.17. Computational modeling
Computational modeling followed methods we have employed
previously for the production of a gas phase, steric strain, 3D super-
position, ligand-based pharmacophore model.39,40 Conformational
searching and alignment studies were carried out at the University
of Montana, Molecular Computational Core Facility. In essence,
training set ligands, including substrates L-Glu and L-Cys2 and inhib-
itors quisqualate (QA) and (S)-4-carboxyphenyl-glycine (4-S-CPG)
were built with SYBYL 7.0 software suite (Tripos, Inc.; St. Louis, MO)
and subjected to energy minimization (10,000 iterations) by Powell
minimization standard method. Initial Optimization and Termina-
tion parameters were set to None and Energy Change options,
respectively. Default parameters and values within the minimiza-
tion dialogue (minimize details) were used. Respective ligand con-
formers were stored as MOL2 files within a molecular spreadsheet
and sorted based on energy profile. The low energy conformation
was identified per ligand. Similar conformational searches were per-
formed for 7g, 11e, 16a and other analogues. Each low energy ana-
logue per ligand was imported with SYBYL and positioned in 3D
space utilizing the following common atoms (centroids) for align-
4.15. Glutamate uptake assay
Uptake of L-glutamate into cultured cells was quantified using a
modification of the procedure of Martin and Shane as previously
described.12,13,47 Individual wells, after removal of culture media,
were rinsed three times and pre-incubated in 1 ml Na+-free HEPES
buffered (pH 7.4) Hank’s balanced salt solution (HBHS) at 30 °C for
5 min. The Na+-free buffer contained: 137.5 mM choline Cl,
5.36 mM KCl, 0.77 mM KH2PO4, 0.71 mM MgSO4.7H2O, 1.1 mM
CaCl2, 10 mM
by aspiration of the preincubation buffer and the addition of a
500
l aliquot of Na+-free transport buffer containing -[3H]-gluta-
mate (4–16 mCi/ml) mixed with -glutamate (10–500 M, final
concentration). In those assays that evaluated inhibitor activity,
the 500 l aliquot of transport buffer contained both the
-[3H]-
D-glucose, and 10 mM HEPES. Uptake was initiated
l
L
L
l
ment, (a) amino acid head group:
a-carbon, carbon of the COOH
group and the amino moiety ( -NH2), and (b) the distal COOH group
a
l
L
carbon. When necessary, isosteric groups were directly correlated to
the above alignment moieties. The various composite alignments
were inspected visually in 3D and checked for desired atom super-
position. Composite MOL2 files were exported to MacPyMol soft-
ware (Version 0.99; DeLano Scientific, LLC; South San Francisco,
CA) to provide the Figure 5 molecular renderings.
glutamate and potential inhibitors to ensure simultaneous addi-
tion. Following a 5 min incubation at 30 °C, the assays were termi-
nated by three sequential 1 ml washes with ice cold buffer and
then the cells were dissolved in 1 ml of 0.4 M NaOH for 24 h. An
aliquot (200
ll) was then transferred into a 5 ml glass scintillation
vial and neutralized with the addition of 5
l
l glacial acetic acid
followed by 3.5 ml Liquiscint@ scintillation fluid (National Diag-
nostics) to each sample. Incorporation of radioactivity was quanti-
fied by liquid scintillation counting (LSC, Beckman LS 6500). Values
are reported as mean S.E.M and are corrected for non-specific
uptake (e.g., leakage and binding) by subtracting the amount of
Acknowledgments
The Bruker (Siemens) SMART APEX diffraction facility was
established at the University of Idaho with the assistance of the
NSF-EPSCoR program and the M. J. Murdock Charitable Trust, Van-
couver, WA, USA. The authors thank Drs. C. Sean Esslinger and
Mariusz Gajewski for helpful discussions of analogue design. The
molecular modeling studies were carried out in the U.M. Molecular
Core Computational Facility. This work was supported in part by
NIH NINDS NS038444 (N.N.,T.R.), NINDS NS30570 (R.J.B.), NCRR
P20RR015583 (R.J.B., N.N., J.G., S.P.), and the Malcolm and Carol
Renfrew Scholarship (M.I.S., T.R.).
L
-[3H]-glutamate accumulation at 4 °C. Inhibitory and Linewe-
aver–Burk plots and associated kinetic parameters for transport
inhibitors were estimated using a non-linear curve fitting analysis
(KaleidaGraph 3.6.5). Ki values were estimated on the basis of a
replot of Km,app values or using IC50 values and the Cheng–Prusoff
equation.37
4.16. Fluorometric assay
Fluorometric determination of L-glutamate efflux was quanti-
Supplementary data
fied using a modification of the procedures described by Nicholls
et al.48 and Vesce et al.49 NADPH fluorescence was quantified using
a Hitachi F-2000 fluorescence spectrophotometer fitted with a
thermostatted cuvette holder and a 1 cm2 electronically driven
Supplementary data associated with this article can be found, in
the online version, at doi:10.1016/j.bmc.2009.11.001.

S. A. Patel et al. / Bioorg. Med. Chem. 18 (2010) 202–213
24. Dal Piaz, V.; Ciciani, G.; Chimichi, S. Heterocycles 1985, 23, 365.
213
References and notes
25. Dal Piaz, V.; Ciciani, G.; Turco, G.; Giovannoni, M. P.; Miceli, M.; Pirisino, R.;
Perretti, M. J. J. Pharm. Sci. 1991, 80, 1417.
26. Dal Piaz, V.; Giovannoni, M. P.; Castellana, M. C.; Palacios, J. M.; Beleta, J.;
Domenech, T.; Segarra, V. J. Med. Chem. 1997, 40, 1417.
27. Dal Piaz, V.; Rascon, A.; Dubra, M. E.; Giovannoni, M. P.; Vergelli, C.; Castellana,
M. C. Il Farmaco 2002, 57, 89.
28. Dal Piaz, V.; Vergelli, C.; Giovannoni, M. P.; Scheideler, M. A.; Petrone, G.;
Zaratin, P. Il Farmaco 2003, 58, 1063.
29. Giovannoni, M. P.; Cesari, N.; Vergelli, C.; Graziano, A.; Biancalani, C.; Biagini,
P.; Ghelardini, C.; Vivoli, E.; Dal Piaz, V. J. Med. Chem. 2007, 50, 3945.
30. Montesano, F.; Barlocco, D.; Dal Piaz, V.; Poggesi, E.; Fanelli, F.; De Benedetti, P.
G. Bioorg. Med. Chem. 1998, 6, 925.
31. Renzi, G.; Pinazauti, S. Il Farmaco Ed. Sci. 1969, 885.
32. Vergelli, C.; Giovannoni, M. P.; Pieretti, S.; Di Giannuario, A.; Dal Piaz, V.;
Biagini, P.; Biancalani, C.; Graziano, A.; Cesari, N. Bioorg. Med. Chem. 2007, 15,
5563.
33. Ye, Z. C.; Sontheimer, H. Cancer Res. 1999, 59, 4383.
34. Magnusson, K. R. Mech. Ageing Dev. 1998, 104, 227.
35. Natale, N. R.; Szabon-Watola, M. I.; Twamley, B.; Bridges, R. J.; Patel, S.; Rajale,
T. Acta Crystallogr., Sect. E 2009, 65, 0144.
36. Chen, G. Q.; Gouaux, E. Proc. Nat. Acad. Sci. U.S.A. 1997, 94, 13431.
37. Cheng, Y. C.; Prusoff, W. H. Biochem. Pharmacol. 1973, 22, 3099.
38. Dowd, L. A.; Coyle, A. J.; Rothstein, J. D.; Pritchett, D. B.; Robinson, M. B. Mol.
Pharmacol. 1996, 49, 465.
1. Balazs, R.; Bridges, R. J.; Cotman, C. W. Excitatory Amino Acid Transmission in
Health and Disease; Oxford University Press: New York, 2006.
2. Kranich, O.; Dringen, R.; Sandberg, M.; Hamprecht, B. Glia 1998, 22, 11.
3. Baker, D. A.; McFarland, K.; Lake, R. W.; Shen, H.; Tang, X. C.; Toda, S.; Kalivas, P.
W. Nat. Neurosci. 2003, 6, 743.
4. Knackstedt, L. A.; Larowe, S.; Mardikian, P.; Upadhyaya, H.; Hedden, S.; Markou,
A.; Kalivas, P. Biol. Psychiat. 2008. Epub.
5. Chung, W. J.; Lyons, S. A.; Nelson, G. M.; Hamza, H.; Gladson, C. L.; Gillespie, G.
Y.; Sontheimer, H. J. Neurosci. 2005, 25, 7101.
6. Shih, A.; Erb, H.; Sun, X.; Toda, S.; Kalivas, P.; Murphy, T. J. Neurosci. 2006, 41,
10514.
7. Espey, M. G.; Kustova, Y.; Sei, Y.; Basile, A. S. J. Neurochem. 1998, 71, 2079.
8. Hosoya, K.; Tomi, M.; Ohtsuki, S.; Takanaga, H.; Saeki, S.; Kanai, Y.; Endou, H.;
Naito, M.; Tsuruo, T.; Terasaki, T. J. Pharmacol. Exp.Therapeut. 2002, 302, 225.
9. Baker, D. A.; Xi, Z. X.;Hui, S.; Swanson, C. J.;Kalivas,P. W. J. Neurosci. 2002, 22, 9134.
10. Augustin, H.; Grosjean, Y.; Chen, K.; Sheng, Q.; Featherstone, D. J. Neurosci.
2007, 27, 111.
11. Bannai, S. J. Biol. Chem. 1986, 261, 2256.
12. Patel, S. A.; Warren, B. A.; Rhoderick, J. F.; Bridges, R. J. Neuropharmacology
2004, 46, 273.
13. Warren, B. A.; Patel, S. A.; Nunn, P. B.; Bridges, R. J. Toxicol. Appl. Pharmacol.
2004, 200, 83.
14. Cho, Y.; Bannai, S. J. Neurochem. 1990, 55, 2091.
15. Ye, Z.; Rothstein, J. D.; Sontheimer, H. J. Neurosci. 1999, 19, 10767.
16. Madsen, U.; Slok, F. A.; Stensbol, T. B.; Brauner-Osborne, H.; Lutshoft, H. H.;
Poulsen, M. V.; Eriksen, L.; Krogsgaard-Larsen, P. Eur. J. Med. Chem. 2000, 35, 447.
17. Brauner-Osborne, H.; Egebjerg, J.; Nielsen, E. O.; Madsen, U.; Krogsgaard-
Larsen, P. J. Med. Chem. 2000, 43, 2609.
18. Nelson, J. K.; Twamley, B.; Villalobos, T. J.; Natale, N. Tetrahedron Lett. 2008, 49,
5957.
19. Burkhart, D. J.; McKenzie, A. R.; Nelson, J. K.; Myers, K. I.; Zhao, X.; Magnusson,
K. R.; Natale, N. R. Org. Lett. 2004, 6, 1285.
20. Burkhart, D. J.; Vij, A.; Natale, N. R. J. Chem. Crystallogr. 1999, 29, 749.
21. Renzi, G.; Dal Piaz, V. Gazz. Chim. Ital. 1965, 95, 1478.
22. Costantino, L.; Rastelli, G.; Gamberini, M. C.; Giovannoni, M. P.; Dal Piaz, V.;
Vianello, P.; Barlocco, D. J. Med. Chem. 1999, 42, 1894.
23. Dal Piaz, V.; Castellana, M. C.; Vergelli, C.; Giovannoni, M. P.; Gavalda, A.;
Segarra, V.; Beleta, J.; Ryder, H.; Palacios, J. M. J. Enzyme Inhib. Med. Chem. 2002,
17, 227.
39. Esslinger, C. S.; Agarwal, S.; Gerdes, J. M.; Wilson, P. A.; Davies, E. S.; Awes, A.
N.; O’Brien, E.; Mavencamp, T.; Koch, H. P.; Poulsen, D. J.; Chamberlin, A. R.;
Kavanaugh, M. P.; Bridges, R. J. Neuropharmacology 2005, 49, 850.
40. Mavencamp, T. L.; Rhoderick, J. F.; Bridges, R. J.; Esslinger, C. S. Bioorg. Med.
Chem. 2008, 16, 7740.
41. Hunston, R.; Gerothanassis, I.; Lauterwein, J. Org. Mag. Res. 1982, 18, 120.
42. Edmonds, D. T.; Summers, C. P. J. Mag. Res. 1969, 12, 134.
43. Bridges, R.; Patel, S. In Topics in Medicinal Chemistry; Napier, S., Ed.; Springer:
NY, 2009.
44. Sontheimer, H. J. Neurochem. 2008, 105, 287.
45. Burkhart, D. J.; Twamley, B.; Natale, N. R. Tetrahedron Lett. 2001, 42, 8415.
46. Szabon-Watola, M. I., Ph.D. Thesis, University of Idaho, 2006.
47. Martin, D. L.; Shain, W. J. Biol. Chem. 1979, 254, 7076.
48. Nicholls, D. G.; Sihra, T. S.; Sanches-Prieto, J. J. Neurochem. 1987, 1987, 50.
49. Vesce, S.; Bezzi, P.; Rossi, D.; Meldolesi, J.; Voltera, A. FEBS Lett. 1997, 411,
107.