Nγ-Aryl glutamine analogues as probes of the ASCT2 neutral amino acid transporter binding site

Article abstract of DOI:10.1016/j.bmc.2004.11.028

Analogues of l-glutamine were designed and synthesized to test a hydrogen-bond hypothesis between ligand and neutral amino acid transporter ASCT2. The key design feature contains a substituted phenyl ring on the amide nitrogen that contains electron withdrawing and electron donating groups that alter the pK_a of the amide NH. Through this study a preliminary binding site map has been developed, and a potent commercially available competitive inhibitor of the ASCT2 transporter has been identified.

Full text of DOI:10.1016/j.bmc.2004.11.028

Bioorganic & Medicinal Chemistry 13 (2005) 1111–1118
Nc-Aryl glutamine analogues as probes of the ASCT2
neutral amino acid transporter binding site
C. Sean Esslinger,* Kimberly A. Cybulski and Joseph F. Rhoderick
NIH COBRE Center for Structural and Functional Neuroscience, Department of Biomedical and Pharmaceutical Sciences,
The University of Montana, Missoula, MT 59812, USA
Received 29May 2004; revised 15 November 2004; accepted 15 November 2004
Available online 8 December 2004
Abstract—Analogues of L-glutamine were designed and synthesized to test a hydrogen-bond hypothesis between ligand and neutral
amino acid transporter ASCT2. The key design feature contains a substituted phenyl ring on the amide nitrogen that contains elec-
tron withdrawing and electron donating groups that alter the pKa of the amide NH. Through this study a preliminary binding site
map has been developed, and a potent commercially available competitive inhibitor of the ASCT2 transporter has been identified.
ꢀ 2004 Elsevier Ltd. All rights reserved.
1. Introduction
across the lipophilic cellular membrane. A number of
neutral amino acid transport systems have been identi-
fied, the main transport systems being the sodium
dependent systems A, ASC, N, and the sodium-indepen-
dent system L.13–15 As these transport systems are
responsible for the translocation of the neutral amino
acids (16 of the 20 common mammalian amino acids),
one may suspect there is considerable overlap of sub-
strate specificity. One common thread of these transport
systems is that they all transport L-glutamine.
L-Glutamine is the most abundant free amino acid in
mammalian blood plasma and cerebral spinal fluid.1
As such an abundant bio-molecule, L-glutamine is in-
volved in a variety of metabolic processes.2 L-Glutamine
is a major component of muscle,3,4 and in times of sur-
gical illness has been classified as a ꢀconditionally essen-
tialꢁ amino acid because of the increased metabolic
requirements.5 As a precursor to TCA cycle intermedi-
ates, glutamine is an important source of energy.6,7 Glu-
tamine acts as a vehicle for both carbon and nitrogen
shuttling8 and serves as an excretion route for toxic
ammonia.9 This systemic glutamine shuttling has been
referred to as the glutamine cycle,3 and is also partly
responsible for the pH buffering of the blood.10 Gluta-
mine is also accepted as being a major metabolic precur-
sor for the excitatory neurotransmitter L-glutamate11
and the inhibitory neurotransmitter GABA.12 In order
for glutamine to participate in all of these roles, this
small polar molecule must move throughout the body.
It is this glutamine movement in which we are interested,
and particularly transporter protein facilitated
movement.
Our interest focused on system ASC glutamine transport
(ASC for alanine, serine, cysteine). System ASC trans-
port has been identified as an electroneutral obligate ex-
changer,13 and depending on transmembrane substrate
ratios, it has the ability to operate in both influx and ef-
flux directions.16 This transporter functions through a
heteroexchange process, with glutamine efflux likely
coupled to the influx of the system ASC substrates ala-
nine, serine or cysteine.11,17 It has been proposed that
the transporter ASCT2 (system ASC isoform 2) is
responsible for part of the glutamine efflux from the
astrocytes involved in the glutamine/glutamate cycling
of glutamate neurotransmitter.11,15,17 The Km reported
for this transport process is in the range of 24–300 lM
for glutamine,18,19 has a lower throughput (Vmax ꢀ 8–
15 nmol/min/mg)18,19 than system A or N activity, and
has been proposed to contribute less of the overall glu-
tamine glial efflux than system N.7 However, it was re-
cently reported that a neuroblastoma cell line exhibits
glutamine transport primarily due to system ASC.20
Neutral amino acid transporter proteins are responsible
for facilitating movement of these highly polar solutes
Keywords: Pharmacophore; Modeling; Inhibition; H-Bonding.
*
Corresponding author. Tel.: +1 406 243 5228; fax: +1 406 243
4713; e-mail: christopher.esslinger@umontana.edu
0968-0896/$ - see front matter ꢀ 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmc.2004.11.028

1112
C. S. Esslinger et al. / Bioorg. Med. Chem. 13 (2005) 1111–1118
Similar glutamine transport activity reported in the liter-
ature occurs in the C6 rat glioma cell line and was fur-
ther characterized as activity due to ASCT2.21,22 These
observations could be the result of a general phenome-
non of cancerous cells up-regulating ASCT2 mediated
glutamine transport,23 and exploitation of this pheno-
menon poses an attractive target for cancer research.24,25
Despite the increased interest in the ASCT2 transporter,
little is known about the protein binding site and trans-
port requirements other than the endogenous L-amino
acid substrate profile (system ASC has a wide range of
substrate neutral amino acids, from small to medium
side chains). The goal of this study is to obtain informa-
tion about the ligand-protein binding interactions and
gain insight into the transport mechanism of the ASCT2
transporter.
ASCT2 binding site. D/L-Homocysteine (–SH
pKa = 10.52) was found to be a more potent inhibitor
of glutamine uptake. at pH 6.0 than L-glutamine
(amide–NH pKa = 16.52). These observations suggest
that the distal group of the ligand may participate in a
hydrogen bond donation that enhances the binding
affinity of the ligand dramatically, and the more acidic
the proton (without being fully ionized), the stronger
the hydrogen bond. This hypothesized hydrogen bond-
ing interaction may be a way to not only enhance the
potency of the synthetic ligands, but to impart selectivity
for the ASCT2 transporter over the other glutamine
transporters.
Therefore, a key design feature of the synthetic ana-
logues included a component to alter the pKa of the glu-
tamine amide NH to test the H-bonding hypothesis. It
was anticipated, in support of this hypothesis, that low-
ering the pKa of the amide NH would increase binding
affinity to the ASCT2 transporter. In order to affect
the acidity of the amide NH of L-glutamine, electron
withdrawing and electron donating groups could be
positioned such that the electronic effects could influence
the stability of the resulting negative charge forming on
the nitrogen. If substituents are placed directly on the
nitrogen, effects other than electronic (i.e., steric, H-
bonding, lipophilic, hydrophilic) may be the cause of
activity changes. We decided to use the N-phenyl substi-
tution as the template for altering the pKa of the NH. By
using this substitution, the conjugated ring allows more
direct substituent effects on the nitrogen electronic den-
sity while not influencing the steric environment in the
immediate NH vicinity.
Our study of the neutral amino acid transporter ASCT2
used the C6 rat glioma cell line as the model system. Our
pharmacological analysis found the C6 cell line func-
tionally exhibits glutamine uptake primarily by ASCT2
mediated transport (approximately 75% of total gluta-
mine uptake, results not shown). The remaining 25%
of glutamine uptake activity was consistent with system
N and to a minor extent system L (these results were in
the range of those reported independently).21 Uptake
experiments were therefore performed in the presence
of BCH (2-aminobicyclo-(2,2,1)heptane-2-carboxylic
acid, a system L selective inhibitor) and at pH 6.0 (to
eliminate the activity of system N) therefore isolating
ASCT2 glutamine transport activity.
In agreement with reported results,22 we observed that
glutamate was relatively inactive at inhibiting the uptake
of radiolabeled L-glutamine at pH 7.4, but was a potent
inhibitor at pH 6.0. This activity is hypothesized to
come from the protonation of glutamate rather than
from protonation of the protein, since the activity of
other substrates remained similar (Fig. 1). The proton-
ation of glutamate at the distal carboxylate forms a neu-
tral amino acid and resembles more closely the structure
of glutamine. The effective concentration of the proton-
ated form of 1 mM glutamate at pH 6.0 is 30 lM (based
on the calculated pKa of 4.57 for the distal acid group
using ACD 6.0 software), implying that the protonated
form of glutamate has a very high affinity for the
Another important design feature of the synthetic ana-
logues is the three-dimensional positioning of the func-
tional groups. Since little is known about the 3-D
requirements of binding to the ASCT2 transporter bind-
ing site, an acyclic template was decided upon. This
would impart little conformational bias between the dif-
ferent N-phenyl substituted analogues and allow more
direct analysis of the electronic effects of the substitu-
ents. Since glutamine has a slightly greater affinity for
the ASCT2 transporter than asparagine,20 the glutamine
length template was used. Therefore, the synthetic ana-
logue design to test the H-bond hypothesis was estab-
lished as the substituted c-N-phenyl L-glutamine
derivative illustrated in Figure 2.
Figure 1. pH dependence of amino acid substrates on the ability to
inhibit C6 glioma cell uptake of 3H-L-glutamine. While L-glutamate is
somewhat inactive at pH 7.4, it becomes a potent inhibitor when the
pH of the buffer is lowered to pH 6.0.
Figure 2. L-Glutamine analogue design. The substituents are electron
donating and electron withdrawing which influence the acidity of the
amide NH. This will test the proposed H-bond hypothesis at the
ASCT2 transporter.

C. S. Esslinger et al. / Bioorg. Med. Chem. 13 (2005) 1111–1118
1113
2. Results and discussion
The synthetic scheme to obtain the N-phenyl analogues
used L-glutamic acid as the chiral starting material,
following a reported sequence shown in Scheme 1.26
Nitrogen protection of glutamic acid with benzyl chloro-
formate resulted in the diacid carbamate 1. In order to
differentiate the two carboxylic acid groups, the selective
conversion to the azalactone 2 was performed in high
yield. The resulting monoacid 2 was treated with thionyl
chloride in refluxing methylene chloride to yield the acid
chloride that was immediately treated with the desired
aniline derivative in the presence of pyridine. This affor-
ded the N-phenyl amide derivative 3. Saponification of
the lactone moiety was accomplished using lithium
hydroxide, with subsequent deprotection of the carba-
mate using catalytic hydrogenation conditions, affording
the desired c-N-phenyl L-glutamine derivatives 4 in
modest yield (combined deprotection yields: 5 68%, 6
65%, 7 52%, 8 63%, 9 57%, 10 45%, 11 75%, 12 71%,
13 61%).
Figure 3. Synthetic glutamine analogues used to test the H-bond
hypothesis of neutral amino acid transporter ASCT2.
The N-aryl glutamine analogues shown in Figure 3 were
tested for the ability to inhibit the uptake of tritiated
glutamine in the C6 cell line at both pH 7.4 and 6.0 in
the presence of the system L inhibitor BCH. The results
shown in Table 1, noting the trend of increased ability to
inhibit the uptake of tritiated glutamine with decreased
pKa (N–H) of the ligands 5–8, support the H-bond
hypothesis for the ASCT2 binding site; that the distal
group of the ligand can participate in a hydrogen bond
donation that will enhance the binding affinity of the li-
gand, and as the pKa of the amide proton decreases, the
hydrogen bond interaction increases.
on the y-axis, suggesting a competitive mechanism of
inhibition and supporting the assumption that the effects
of glutamine uptake inhibition result from ligand–pro-
tein bonding interactions in the transporter binding site.
Presumably, the compounds act as reversible inhibitors
of glutamine uptake by acting as alternative substrates.
The actual transport of the analogues, however, has yet
to be shown.
The glutamine uptake inhibition data in Table 1 shows a
trend with respect to pKa, but also show anomalous re-
sults. The analogues 11, 12, and 13 containing the elec-
tron donating groups on the phenyl ring were expected
to show low glutamine uptake inhibition activity similar
to that of the unsubstituted analogue 5, based on similar
pKa (NH) values. On the contrary, these compounds
were quite potent inhibitors of glutamine uptake at
pH 6.0, suggesting lipophilic bonding interactions be-
tween ligand and protein occur within the binding site
of the transporter. On the other side of electronic
influence, the fluorophenyl analogues 9 and 10 were ex-
pected to show higher ability to inhibit the uptake of tri-
tiated glutamine than the chlorophenyl analogues 6
and 7, based on lower calculated pKa (NH) values.
To support the assumption that this effect in activity is
due to an increased affinity for the transporter binding
site, a kinetic analysis of the p-nitrophenyl anilide 8
was performed to determine the mechanism of inhibi-
tion. The concentration dependent inhibition curves
were used to fit the classical Michaelis–Menten equation
m = Vmax[S]/(Kmapp + [S]) with Kmapp = Km (1 + [I]/Ki)
by nonlinear regression (Fig. 4), where m is uptake velo-
city, [S] is radiolabeled glutamine concentration, and [I]
is inhibitor concentration. On replotting the data to ob-
tain a Lineweaver–Burk plot, the different lines intersect
Scheme 1. Synthesis of L-c-glutamyl anilide derivatives 4.

1114
C. S. Esslinger et al. / Bioorg. Med. Chem. 13 (2005) 1111–1118
Table 1. C6 L-glutamine uptake inhibition by synthetic analogues at
pH 6.0 and 7.4
Synthetic analogues
Percent of controla SD
Compound
pKa
(N–H)b
[Inhibitor]
(mM)
pH = 7.4
pH = 6.0
5
6
7
8
15.00
14.12
13.40
13.79
1
0.5
61 12
68
65 13
76
1
9
1
0.5
68 15
62
1
11
3
2
7
1
0.5
53
58
9
3
4
8
4
4
9
1
12 13
13
55 14
0
0
0.5
0.25
7
15
Figure 5. CoMFA with PLS STD*DEV COEFF shown. Steric
addition in the green area should increase activity supporting a
lipophilic pocket hypothesis, while steric addition in the yellow area
should decrease activity, supporting a benefit in NH accessibility. The
blue and red areas should enhance activity on increasing positive and
negative charges, respectively.
15 21
61
65 12
9
13.98
12.11
15.00
15.01
14.41
1
0.5
61 8
77 11
6
10
11
12
13
1
0.5
59
69
2
2
14
42
3
4
1
0.5
51
74
5
0
23
3
1
7
analogues is the substitution at the amide nitrogen, the
acyclic glutamine backbone was minimized and used
as the template for each analogue and re-minimized,
resulting in the same basic conformation for each ana-
logue. This helps to focus the analysis on the substituent
effects. Using the default parameters in the SYBYL 6.91
advanced computation module, the calculated steric,
electronic, and CoMFA properties along with the calcu-
lated pKa were compared to the estimated pKi values.
The resulting CoMFA model is shown in Figure 5.
The volumes of space surrounding the molecule (the p-
nitrophenyl derivative 8 shown) depict areas where
molecular properties are predicted to increase affinity
or decrease affinity for the protein. The green area
around the aromatic ring is where adding steric bulk
should increase ligand affinity (identifying a potential
lipophilic pocket). The yellow area above the amide
N–H is where adding steric bulk should decrease affin-
ity, supporting the hypothesis that the protein may oc-
cupy this region and act as the hydrogen bond
acceptor. The red region and the blue region support
ionic interactions from the zwitterionic form of the ana-
logues (which were initially used for the calculations).
1
0.5
55 16
6
0
40
6
4
78
1
0.5
54
69
3
0
5
8
21
4
a 100% of control = 0.5–1.8 nmol/mg/min, incubation time = 5 min,
n P 3 for all values, [3H-L-glutamine] = 100 lM.
b pKa calculations performed using ACD 6.0 software.
Kinetic Analysis of Glutamine Uptake Inhibition
for Analogue 8 at pH 6.0 in C6 Cells
0 uM
Vmax =855 9 pmol/min/mg
1000
750
500
250
10 uM
30 uM
100 uM
µ
Km =86 4
M
Ki =55 µM
6
Lineweaver-Burk Plot
0.0100
0.0075
0.0050
0.0025
0.0000
0 uM
10 uM
30 uM
100 uM
0.00
0.01
0.02
0.03
0.04
0.05
1/[S]
0
Although the CoMFA model supports the H-bond
hypothesis and identifies a possible lipophilic pocket,
the model does not explain the low glutamine uptake
inhibition activity of the fluorophenyl analogues 9 and
10 at pH 6.0. We believe that the increased anionic
character of the fluorine atoms may have a destabilizing
effect on bonding when positioned in this potential lipo-
philic pocket. To support this, log D (pH 6.0) values
(partition coefficients between octanol and water of the
ionic form at the corresponding pH) were calculated
(using ACD 6.0 software) and compared (Table 2). The
comparison does suggest a better solvation in water (a
more negative value) of the fluoro analogues than the
chloro analogues. A component making up the log D
property, the H-bond acceptor ability, was also com-
pared. Using the MOLCAD module of SYBYL 6.91, the
H-bond acceptor surface was calculated and the surface
0
250
500
1000
1250
750
µ
[L-Glutamine] ( M)
Figure 4. Inhibition curves for p-nitrophenyl glutamyl anilide 8 fit to
Michaelis–Menten kinetics. Inset shows Lineweaver–Burk plot of
same data to better illustrate competitive inhibition as lines converge at
1/[S] = 0.
Surprisingly, these compounds were somewhat inactive
at pH 6.0.
To help visualize these substituent effects, a Compara-
tive Molecular Field Analysis (CoMFA) was performed.
Using the Cheng–Prusoff relation, %inhib. = [I]/([I] + Ki
(1 + [S]/Km)), the Ki was estimated and used as the
dependent variable. Since the only difference between

C. S. Esslinger et al. / Bioorg. Med. Chem. 13 (2005) 1111–1118
Table 2. Log D values calculated for the synthetic analogues at pH 6.0
1115
Analogue
Log Da (pH 6.0)
Analogue
Log D (pH 6.0)
Analogue
Log D (pH 6.0)
H
N
H
N
H
N
O
O
O
À1.79
À1.22
À1.33
À1.78
CH3
NO
2
H3N
CO2
H3N
CO2
11
H N
3
CO
2
5
8
H
N
H
H
N
O
O
O
O
N
À0.82
À1.00
À1.40
À1.20
Cl
Cl
F
OCH3
H3N
CO2
H3N
CO2
9
H3N
CO2
12
6
Cl
F
CH3
H
N
H
N
H
N
O
F
O
À0.87
F
CH3
H3N
CO2
7
H3N
CO2
10
H3N
CO2
13
On comparing the fluoro analogues 9 and 10 with the chloro analogues 6 and 7, the fluoro analogues would appear to have a better solvation in water
by the more negative value. On comparing the fluoro analogues 9 and 10 with the alkyl/alkoxy analogues 11, 12, and 13, however, the fluoro
analogues would appear to not be as well solvated in water.
a Log D values calculated using ADC 6.0 software.
area of the molecule having H-bond acceptor property
density of 0.02 or higher displayed (Fig. 6). The resulting
H-bond acceptor surface area is significantly larger for
the trifluorophenyl analogue 10, than the other ana-
logues analyzed. The p-fluorophenyl derivative 9 shows
only a slight increase in H-bond acceptor surface area
over the p-chlorophenyl compound 6. Although the dif-
ference in H-bond acceptor surface area does not fully
account for the dramatic decrease in glutamine uptake
inhibition activity of compound 9, these combined re-
sults suggest a stronger solvation of the fluorophenyl
compounds in water. The lowered glutamine uptake
inhibition activity seen by the fluoro compounds could
be the result of the desolvation energy required to strip
the waters off, while in binding to the protein there is no
complimentary interaction to regain that solvation en-
ergy. This implies that within this region of space in
the binding site, there are no hydrogen bond donors,
supporting a lipophilic pocket.
3. Conclusion
Based on initial characterization, the ASCT2 H-bond
hypothesis was proposed. Through the use of glutamine
analogue probes, this hypothesis was supported and an
additional bonding hypothesis was generated: a lipo-
philic pocket is in the binding site. These hypotheses
are summarized in the graphic in Figure 7, illustrating
Figure 6. H-Bond acceptor surfaces for fluoro analogues 9 and 10 (A
and B, respectively) and the potent inhibitors p-nitrophenyl glutamyl
anilide 8 and 2,3-dichlorophenyl glutamyl anilide 7(C and D, respec-
tively). The surface is the H-bond acceptor density with a density value
of 0.02 and greater for the property. The areas calculated (SYBYL 6.91
2
2
2
˚
MOLCAD module) are: A A = 106.93, B A = 172.05, C A = 104.87,
˚
˚
2
and D A = 155.19, suggesting the trifluoro analogue 10,A has the
˚
most H-bond accepting surface of the four compounds, while the
monofluoro compound 9,B has less H-bond accepting surface than
the dichloro analogue 7,D and similar to 8,C. The lack of activity can
be explained by solvation effects for 10, but not readily for 9 through
this analysis.
Figure 7. Preliminary neutral amino acid transporter ASCT2 binding
site map based on the results of this study, illustrating the hydrogen-
bond hypothesis and the lipophilic pocket hypothesis. Other features
are also incorporated, such as an ꢀextendedꢁ conformation and the L-
configuration.

1116
C. S. Esslinger et al. / Bioorg. Med. Chem. 13 (2005) 1111–1118
L-3H-glutamine solution purchased from Perkin–Elmer
was spotted lightly on Whatman No 4 filter paper cut
into a strip 1 in. wide. A freshly prepared 1 mM gluta-
mine solution in water was spotted adjacent to the label
as a standard. The paper bottom is immersed in the mo-
bile phase containing 70% t-butanol, 15% formic acid,
15% water, and 5 mg/mL ninhydrin (Rf values for L-glu-
tamine, L-glutamate, and L-pyroglutamate in this sol-
vent system are 0.22, 0.31, and 0.32, respectively). The
addition of ninhydrin allows visualization of the gluta-
mine standard on drying the chromatogram in air. After
development, the chromatogram is dried in air and cut
into ten or more numbered horizontal sections while
noting where glutamine falls (identified by the standard)
The individual sections are put into vials containing
1 mL water and allowed to soak for 30 min. The compo-
nents are then quantified by the addition of 5 mL scintil-
lation fluid and the radiation counted by scintillation
counter. In most cases the radiolabel contained 10–
15% contaminant, which was used without purification.
Purification was performed by using tandem anion ex-
change/cation exchange cartridges (Whatman) and pass-
ing the volume of radiolabeled solution through using a
syringe. After washing the resin with two volumes of
deionized water, two volumes of 2 N ammonium
hydroxide solution was passed only through the cation
exchange cartridge. This effluent was lyophilized and
taken up in one volume of deionized water.
a preliminary binding site map for the ASCT2 neutral
amino acid transporter. Also through these experiments,
a commercially available potent inhibitor of the ASCT2
transporter has been identified, the p-nitrophenyl ana-
logue 8. Although the fluorophenyl derivatives 9 and
10 fit within the parameters to suggest high potency glu-
tamine uptake inhibition, these compounds are rela-
tively inactive. This inactivity is explained by increased
polarity of ring substituents resulting in a greater differ-
ence in solvation energy between bound and unbound
states, supporting a lipophilic pocket in the binding site.
A recent set of experiments suggest that within the brain
the SN1 transporter (the system N isoform prominent in
neuroglia) is the sole route for glial glutamine efflux in-
volved in the recycling of glutamate neurotransmitter
through the glutamine/glutamate cycle in the CNS.24
This leaves the ASCT2 transporter function to be deter-
mined. One hypothesis is that the ASCT2 transporter
functions as a facilitator of cellular needs, exchanging
the abundant glutamine for deficient amino acids. The
broad substrate profile supports this idea, and the up
regulation in cancerous cells requiring an abundant vari-
ety of amino acids for the uncontrolled growth also lend
to this hypothesis. Mentioned in a previous report,7 the
pH dependent transport of glutamate by the ASC trans-
port system offers an interesting possibility of aiding the
clearance of synaptic glutamate. Provided the trans-
porter is in the near vicinity of the vesicular glutamate
release (accompanied by an acidic environment) a rele-
vant contribution of glutamate uptake may be possible.
As this project advances, new compounds may aid in
identifying physiological roles of these transporters.
We are currently looking at the 3-D preference of the
functional groups, exploring the lipophilic pocket
hypothesis, and investigating in more detail the hydro-
gen bonding hypothesis further in order to develop more
potent and selective compounds to meet this end.
4.3. Cell culture
C6 rat glioma cells were grown in Dulbeccoꢁs Modified
Eagle Medium (Mediatech) (pH 7.4). The culture media
was supplemented with 10% fetal calf serum and a mix-
ture of penicillin (100 U/mL), streptomycin (100 mg/
mL) and fungizone (0.1 mg/mL). All cell lines were cul-
tured in 150 cm2 flasks (Corning) and maintained at
37 ꢁC in a humidified atmosphere of 5% CO2. Cells were
lifted from the flasks by the addition of trypsin (0.25%
final concentration) for 3–5 min at 37 ꢁC for passage
every 7 days. At this time cell culture stocks for cryo-
preservation were prepared by maintaining 1 · 106–
2 · 106 cells in the appropriate culture media supple-
mented with 20% FCS, 10% DMSO and storing at
À196 ꢁC under liquid N2. Cells formed a confluent
monolayer with an estimated density of 2.5 · 106 cells
and a protein concentration of 100–350 mg/well. Cells
were given the appropriate fresh medium 2 h prior to
an assay.
4. Experimental
4.1. General
L-[3,43H(N)]-Glutamine was purchased from Perkin–
Elmer and L-c-nitrophenyl glutamyl anilide was pur-
chased from ICN. D/L-Homocysteine, L-glutamate, ani-
lines, and Cbz-chloride were purchased from ACROS
Chemicals. NMR spectra were obtained on a Varian
400 MHz instrument using D2O with acetonitrile in
D2O (d 1.93 for proton, d 1.3 for carbon) as the external
standard, or DMSO with acetonitrile in DMSO as the
external standard. Optical rotations were taken on a
Perkin–Elmer polarimeter. HRMS were obtained
through the Mass Spectrometry and Proteomics Core
Facility, The University of Montana.
4.4. Cell culture glutamine transport assay
The C6 cell culture assays are carried out in 12-well
plates.27 Growth media is changed two hours prior to
carrying out the uptake assay. Cellular assays are more
consistent if the passage number does not exceed 60.
Each set of conditions is carried out in triplicate, allow-
ing for control uptake, two inhibitor experiments, and
three samples for protein assay per plate. The wells at
31 ꢁC are washed three times with 1 mL assay buffer at
pH 6.0 or 7.4 (containing: 137 mM NaCl, 5.1 mM
KCl, 0.77 mM KH2PO4, 0.71 mM MgSO4Æ7H2O,
1.1 mM CaCl2, 10 mM D-glucose, 10 mM HEPES,)
4.2. L-3H-glutamine analysis
The radiolabeled glutamine was analyzed for contami-
nation by glutamate/pyroglutamate using paper chro-
matography by the method of Levintow and Meister30
incorporating modifications by Khadem et al.31 The

C. S. Esslinger et al. / Bioorg. Med. Chem. 13 (2005) 1111–1118
1117
and allowed to preincubate for 5 min in 1 mL buffer.
The preincubation buffer is removed and H-glutamine
(100 lM, 500k) is added in same buffer (50 lL H-Gln
acidified to pH 2 with 6 N HCl, saturated with sodium
chloride, and extracted with ethyl acetate three times.
The combined ethyl acetate extracts were dried
(MgSO4), filtered, and concentrated to yield 70–98%
pure L-Z-glutamate.
3
3
in 20 mL buffer) with 5 mM BCH and with or without
inhibitor and allowed to incubate for 5 min. The radio-
label is removed after 5 min, washed with 1 mL ice cold
buffer three times, and lysed with 1 mL 0.4 N NaOH.
After 24 h, 0.4 mL of the NaOH sample is put in a scin-
tillation vial, 4 mL scintillation fluid is added, the sam-
ple is mixed and counted in a scintillation counter. All
values are corrected for background by subtracting
radioactivity accumulated in the presence of 10 mM glu-
tamine for pharmacology experiments or at 0 ꢁC using
identical conditions for kinetic experiments, and re-
ported as means +/À standard deviation of n assays.
Protein concentrations for plates were sampled from
three wells using the Pierce BCA (bicinchoninic acid)
protein assay (Pierce, Rockford, IL).
4-(2-Carboxy-ethyl)-5-oxo-oxazolidine-3-carboxylic acid
benzyl ester 2: To the 1 in toluene was added 1.5 equiv
paraformaldehyde and 1% toluenesulfonic acid (wt/
wt). The mixture was then refluxed under Dean–Stark
conditions for approximately 3 h. The mixture was di-
luted with ethyl acetate and washed with brine, dried
(MgSO4), filtered, and concentrated to yield the Z-
azalactone, which was used without further purification.
5-Oxo-4-(2-phenylcarbamoyl-ethyl)-oxazolidine-3-carb-
oxylic acid benzyl ester 3: To the anhydrous Z-azalac-
tone 2 in methylene chloride (1 M solution) was added
thionyl chloride and refluxed for 30 min at which time
the reaction mixture was cooled to room temperature
followed by the drop wise addition of 1 equiv aniline
derivative and 1 equiv pyridine in methylene chloride.
The mixture was stirred for an additional 15 min at
room temperature, diluted with methylene chloride,
and washed consecutively with 20% sodium phosphate
pH 3.0 and saturated sodium bicarbonate solution.
The organic solution was then dried with magnesium
sulfate, filtered, and concentrated to yield the Z-azalac-
tone amide 3, which was used in the nest step without
further purification. The amide 3 was dissolved in a min-
imum of THF, chilled to 5 ꢁC, and 1 equiv of a 1 M
solution of LiOH added. The mixture was stirred for
30 min at which time the THF was removed by rotory
evaporation. The remaining aqueous mixture was di-
luted with water, washed twice with ethyl acetate, then
acidified to pH 2 with 6 N HCl, saturated with NaCl(s),
and extracted with ethyl acetate three times. The com-
bined organic extracts were dried (MgSO4), filtered,
and concentrated to yield the pure N-Cbz carboxylic
acid which was carried on to the next step without fur-
ther purification.
4.5. Calculations
Physical properties (pKa and log D values) were calcu-
lated using ACD 6.0 software (Advanced Chemistry
Development Inc., Toronto, Ontario, Canada M5H
3V9, http://www.acdlabs.com). Nonlinear regression
was performed using GRAPHPAD PRISM version 4.00 for
Windows (GRAPHPAD Software, San Diego California
USA, http://www.graphpad.com). Kinetic data was fit-
ted to Michaelis–Menten kinetics with data points equal
to means SD or SEM of n experiments having nonspe-
cific binding subtracted, n P 3. Nonspecific binding
measured as radiolabel retained in the presence of
10 mM unlabeled glutamine for pharmacology or at
0 ꢁC for kinetics. All data were tested for significance
using Studentꢁs t-test, and only results with P < 0.05
were considered to be statistically significant.
4.6. Molecular modeling
SYBYL 6.91 was used for molecular modeling calcula-
tions (Tripos Inc., http://www.tripos.com) on a Slicon
Graphics O2 machine. Minimizations of the zwitterionic
forms of the amino acid analogues used the following
parameters: Tripos force field, Pullman charge calcula-
tion, dielectric constant = 80, minimization convergence
criteria of 0.001 kcal/mol, Powell minimization method.
CoMFA surfaces were calculated using SYBYL default
parameters with pKi (Ki estimated from the relation
%inhib. = [I]/([I] + Ki (1 + [S]/Km)) where [I] is the inhib-
itor concentration, [S] is the glutamine concentration,
and %inhib. = 100 À % of control obtained in Table 1)
as the dependent variable and pKa, electronic, steric,
and the SYBYL comfa measurement as the independent
variables.
Nc-Phenyl-glutamyl anilides 4: The N-Cbz carboxylic
acid was dissolved in methanol and the solution de-
gassed by evacuation and flushing with argon. Palla-
dium on carbon (10% Pd) was added to the
methanolic solution (10% wt/wt), the mixture purged
with hydrogen several times, and stirred under an atmo-
sphere of hydrogen until no starting material remained
by TLC. The reaction was filtered and the catalyst
washed with water and additional methanol. The meth-
anol was removed in vacuo and the water removed by
rotory evaporation without heating or by lyophilization
to yield pure glutamine derivative 4.
4.7. Chemistry
The phenyl analogue and all mono-substituted deriva-
tives (5, 6, 8, 9, 11, and 12)25,28,29 match reported
spectra.
4.7.1. General procedure. 2-Benzyloxycarbonylamino-5-
oxo-hexanedioic acid 1: To 1 equiv L-glutamate in 40%
triethylamine in methanol making a 1 M solution was
added benzyl chloroformate neat and stirred for 3 h.
The mixture was concentrated, diluted with water, and
washed with diethyl ether. The aqueous layer was then
1
L-c-N-2,3-Dichlorophenyl glutamyl anilide 7: H NMR
(DMSO): d 7.28–7.36 (m, 3H), 4.01 (ddd, J = 4.4, 4.5,
8.6 Hz, 1H), 2.48 (m, 2H), 2.07 (m, 1H), 1.83 (m, 1H).
13C NMR (DMSO): d 173.8, 170.9, 137.1, 128.5,

1118
C. S. Esslinger et al. / Bioorg. Med. Chem. 13 (2005) 1111–1118
6. Chaudhry, F. A.; Reimer, R. J.; Krizaj, D.; Barber, D.;
Storm-Mathisen, J.; Copenhagen, D. R.; Edwards, R. H.
Cell 1999, 99, 769–780.
7. Varoqui, H.; Zhu, H.; Yao, D.; Ming, H.; Erickson, J. D.
J. Biol. Chem. 2000, 275, 4049–4054.
8. Bode, B. P.; Souba, W. W. J. Parenter. Enteral Nutr. 1999,
23, S33.
9. Haussinger, D. Metabolism 1989, 38, 14–17.
10. Bourke, E.; Haussinger, D. Contrib. Nephrol. 1992, 100,
58–88.
128.0, 127.9, 126.7, 125.1, 53.4, 32.3 26.5. [a]D +25.0 (c
1.03); HRMS Calcd for C11H13N2O3Cl2 = 291.0303.
Found 291.0308.
L-c-N-2,3,4-Trifluorophenyl glutamyl anilide 10: 1H
NMR (DMSO): d 8.56 (br d, 2H), 7.55 (ddd, J = 61.5,
9.1, 5.2 Hz, 1H), 7.19 (ddd, J = 86, 8.74, 4.85 Hz, 1
H), 3.89(br, 1H), 2.60–2.35 (2dd, J = 8.4, 9.0 Hz, 2H),
2.11 (dd, J = 5.8,5.8 Hz, 1H), 2.01 (dd, J = 5.8, 5.8 Hz,
1H). 13C NMR (DMSO): d 173.7, 170.9, 152.1
(J = 260 Hz), 141.3 (J = 260 Hz), 138.14 (J = 260 Hz),
123.0, 119.0, 111.7, 53.3, 32.1, 26.4. [a]D +25.4 (c 1.32);
HRMS Calcd for C11H12N2O3F3 = 277.0800. Found
277.0821.
11. Deitmer, J. W.; Broer, A.; Broer, S. J. Neurochem. 2003,
87, 127–135.
12. Rae, C.; Hare, N.; Bubb, W. A.; McEwan, S. R.; Broer,
A.; McQuillan, J. A.; Balcar, V. J.; Conigrave, A. D.;
Broer, S. J. Neurochem. 2003, 85, 503–514.
13. Broer, S.; Brookes, N. J. Neurochem. 2001, 77, 705–719.
14. Chaudhry, F. A.; Schmitz, D.; Reimer, R. J.; Larsson, P.;
Gray, A. T.; Nicoll, R.; Kavanaugh, M.; Edwards, R. H.
J. Neurosci. 2002, 22, 62–72.
L-c-N-2,4-Dimethylphenyl glutamyl anilide 13: 1H
NMR (D2O): d 6.96 (br s, 1H), 6.89 (s, 2H), 3.91 (tapp
,
J = 6.7 Hz, 1H), 2.50 (m, 2H), 2.11 (m, 5H), 1.95 (s,
3H). 13C NMR (D2O): d 173.9, 172.0, 138.1, 131.4,
127.3, 126.8, 52.6, 31.3, 25.9, 20.2, 16.9. [a]D +24.1 (c
0.80); HRMS Calcd for C13H19N2O3 = 251.1409. Found
251.1396.
15. Bode, B. P. J. Nutr. 2001, 131, 2475S–2485S.
16. Zerangue, N.; Kavanaugh, M. P. J. Biol. Chem. 1996, 271,
27991–27994.
17. Broer, A.; Brookes, N.; Ganapathy, V.; Dimmer, K. S.;
Wagner, C. A.; Lang, F.; Broer, S. J. Neurochem. 1999, 73,
2184–2194.
18. Utsunomiya-Tate, N.; Endou, H.; Kanai, Y. J. Biol.
Chem. 1996, 271, 14883–14890.
19. Dolinska, M. D. A.; Hilgier, W.; Zielinska, M.; Zablocka,
B.; Buzanska, L.; Albrecht, J. J. Neurosci. Res. 2001, 66,
959–966.
20. Wasa, M.; Wang, H. S.; Okada, A. Am. J. Physiol. Cell.
Physiol. 2002, 282, C1246–C1253.
21. Dolinska, M.; Dybel, A.; Albrecht, J. Neurochem. Int.
2000, 37, 139–146.
22. Dolinska, M.; Dybel, A.; Zablocka, B.; Albrecht, J.
Neurochem. Int. 2003, 1353.
23. Li, R. Y. M.; Frolov, A.; Wheeler, T. M.; Scardino, P.;
Ohori, M.; Ayala, G. Anticancer Res. 2003, 23, 3413–
3418.
24. Boulland, J. L.; Osen, K. K.; Levy, L. M.; Danbolt, N. C.;
Edwards, R. H.; Storm-Mathisen, J.; Chaudhry, F. A.
Eur. J. Neurosci. 2002, 15, 1615–1631.
Acknowledgements
We would like to give a special thanks to Whitney Wid-
halm for all her efforts in helping the project along, Erin
M. OꢁBrien for her help with cell culture, Professor
Richard Bridges, Director of the Center for Structural
and Functional Neuroscience, for his input and sugges-
tions, Professor John Gerdes, Director of the Molecular
Computational Core Facility, and the Department of
Chemistry, The University of Montana and NSF for
the use of the NMR instrument. This research was made
possible by NIH Grant number P20 RR15583 from the
COBRE Program of the National Center for Research
Resources.
25. Parker, C. M.-B. E.; Meehan, G.; Piva. T.; Rigano, D.;
Favot, P.; West, M.; McCabe, M. G. P.; Miller, D. J.
World Patent 9,003,399, 1990.
References and notes
26. Lee, B. H.; Miller, M. J. Tetrahedron. Lett. 1984, 25, 927–
930.
27. Bridges, R.; Stanley, M. S.; Anderson, M. W.; Cotman, C.
W.; Chamberlin, A. R. J. Med. Chem. 1991, 34, 717–
725.
28. Menard, A. C. R.; Lherbet, C.; Rivard, C.; Roupioz, Y.;
Keillor, J. W. Biochemistry 2001, 42, 12678–12685.
29. Huang, X.; Luo, X.; Roupioz, Y.; Keillor, J. W. J. Org.
Chem. 1997, 62, 8821.
30. Levintow, L.; Meister, A. J. Biol. Chem. 1954, 209, 265.
31. El Khadem, H. S.; El-Shafei, Z. M.; Abdel Rahman, M.
M. A. Anal. Chem. 1963, 35, 1766.
1. Johansen, L.; Roberg, B.; Kvamme, E. Neurochem. Res.
1987, 12, 135–140.
2. J. Nutr. 2001, 131, Suppl. 9, 2447S–2590S.
3. Rennie, M. J.; Hundal, H. S.; Babij, P.; MacLennan, P.;
Taylor, P. M.; Watt, P. W.; Jepson, M. M.; Millward, D.
J. Lancet 1986, 2, 1008–1012.
4. Boelens, P. G.; Nijveldt, R. J.; Houdijk, A. P. J.; Meijer,
S.; Van Leeuwen, P. A. M. J. Nutr. 2001, 131, 2569S–
2577S.
5. Labow, B. I.; Souba, W. W. World J. Surg. 2001, 24,
1503–1513.