Tetrapeptide inhibitors of the glutamate vesicular transporter (VGLUT)

Article abstract of DOI:10.1016/j.bmcl.2007.07.006

Quinoline-2,4-dicarboxylic acids (QDCs) bearing lipophilic substituents in the 6- or 7-position were shown to be inhibitors of the glutamate vesicular transporter (VGLUT). Using the arrangement of the QDC lipophilic substituents as a template, libraries of X₁X₂EF and X₁X₂EW tetrapeptides were synthesized and tested as VGLUT inhibitors. The peptides QIEW and WNEF were found to be the most potent. Further stereochemical deconvolution of these two peptides showed dQlIdElW to be the best inhibitor (K_i = 828 ± 252 μM). Modeling and overlay of the tetrapeptide inhibitors with the existing pharmacophore showed that H-bonding and lipophilic residues are important for VGLUT binding.

Full text of DOI:10.1016/j.bmcl.2007.07.006

Bioorganic & Medicinal Chemistry Letters 17 (2007) 5125–5128
Tetrapeptide inhibitors of the glutamate vesicular
transporter (VGLUT)
Sarjubhai A. Patel, Jon O. Nagy, Erin D. Bolstad, John M. Gerdes
and Charles M. Thompson*
Center for Structural and Functional Neuroscience, Department of Biomedical and Pharmaceutical Sciences,
The University of Montana, Missoula, MT 59812, USA
Received 5 February 2007; revised 1 July 2007; accepted 3 July 2007
Available online 26 July 2007
Abstract—Quinoline-2,4-dicarboxylic acids (QDCs) bearing lipophilic substituents in the 6- or 7-position were shown to be inhib-
itors of the glutamate vesicular transporter (VGLUT). Using the arrangement of the QDC lipophilic substituents as a template,
libraries of X1X2EF and X1X2EW tetrapeptides were synthesized and tested as VGLUT inhibitors. The peptides QIEW and WNEF
were found to be the most potent. Further stereochemical deconvolution of these two peptides showed DQLIDELW to be the best
inhibitor (Ki = 828 252 lM). Modeling and overlay of the tetrapeptide inhibitors with the existing pharmacophore showed that H-
bonding and lipophilic residues are important for VGLUT binding.
Ó 2007 Elsevier Ltd. All rights reserved.
L-Glutamate is stored in synaptic vesicles prior to its
H
depolarization-triggered, calcium-dependent release
from neuron terminals1–4 and is transported into the
vesicles in an ATP-dependent manner by the glutamate
vesicular transporter (VGLUT). Unlike the plasma
membrane neurotransmitter transporters, VGLUT is
stimulated by low, physiologically relevant concentra-
tioꢀns of Clꢀ ion,5 although the contribution of the
Cl -to DpH has been debated.2–7 VGLUT is specific
for glutamate but it has low affinity (Km = 1–3 mM),
which contrasts with the plasma membrane transporters
that are specific for glutamate but with high affinity
(Km = 5–50 lM).8–11 To differentiate between these
transporters, potent and selective inhibitors of VGLUT
are needed.
HO
O
N
O
O
N
H
H
O
HN
Br
H
N
H
Me
bromocryptine (22
μM)
HO3S
NH2
Me
OH
N
N
HO3S
N
SO3H
N
HO
Me
H2N
SO3H
trypan blue (50 nM)
Figure 1. Structures of VGLUT inhibitors.
The main VGLUT inhibitor structures have been re-
cently reviewed.1 In brief, aspartate5,12 and simple glu-
tamate analogs are not good inhibitors of VGLUT,
whereas some kynurenate analogs showed modest activ-
ity. The alkaloid bromocryptine (Ki = 20 lM) and cer-
tain azo dyes (e.g., Trypan blue) are among the most
potent VGLUT inhibitors (Fig. 1).13 We reported a sys-
tematic, structure-activity study of quinoline 2,4-dicar-
boxylic acids (QDCs; Fig. 2) as inhibitors.14,15 that
seeded the development of the first pharmacophore
model.1 for VGLUT and the use of QDCs as a key motif
for future inhibitor design and substituent variation.
Some of the more potent QDC-based inhibitors con-
tained lipophilic groups at position 6 or a hydroxyl at
position 8. Combining these favorable substituents into
the QDC template led to the observation that this
Keywords: Vesicular glutamate transporter (VGLUT); Inhibitor; Tet-
rapeptides; Glutamate; Uptake; Vesicle; Trypan blue.
*
Corresponding author. Tel./fax: +1 4062434643; e-mail: charles.
thompson@umontana.edu
0960-894X/$ - see front matter Ó 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2007.07.006

5126
S. A. Patel et al. / Bioorg. Med. Chem. Lett. 17 (2007) 5125–5128
Table 1. Inhibition of VGLUT by tetrapeptidesa
CO2H
O
-(O)C
HN
CO2H
4
X1
X2
X3
X4
3H-L-Glu uptake
(% of control)b
6
NH2
N
N
H
N
2 C(O)X
H
Library 1
AAc
Q
8
OH
O
R
AA
AA
W
E
W
W
Quinoline-2,4-dicarboxylic acid (QDC)
X = peptide
X
E
56
66
38
35
28
1
4
5
3
3
W - E - X peptide
Q
E
W
W
Q
I
E
Figure 2. Proposed QDC inhibitor structural relationship to peptides.
D-Q
L-Q
D-I
D-I
L-E
L-E
D-W
D-W
Library 2
AA
pattern overlays with a peptide that contains (HO2C)-
WEX(NH2) (Fig. 2). The very weak basicity of the
QDC nitrogen also suggested that a peptide amide
might be an appropriate isostere. This prompted an
investigation of small peptides that might be capable
of binding VGLUT. Peptide-based inhibitors are also
possible leads to uncover protein interactions.
AA
AA
AA
N
E
F
N
W
E
F
63 17
E
F
36
13
41
2
3
1
W
E
F
D-W
L-N
D-E
D-F
Other
Congo red (2 lM)
31
2
a Tetrapeptides tested as racemic mixtures at 2 mM.
Based on observations that QDCs containing an embed-
ded glutamate moiety and lipophilic substituents (phe-
nyl, styryl, etc.) confer greater inhibitory activity, a
peptide library was envisioned in which the C-terminus
amino acid was occupied by either tryptophan (W) or
phenylalanine (F) to represent the lipophilic substituent
and the adjacent position occupied by a glutamate (E)
residue. The N-terminus and second residue (X1 and
X2) were systematically varied to investigate how these
positions could enhance binding (Fig. 3).
b Control rate for 3H-L-glutamate uptake was 1847 130 (n = 17)
pmol/min/mg protein.
c AA = 19 different amino acids.
Cl
Cl
To further refine our binding requirements and increase
the overall library diversity, either D- or L-amino acids
were used. Further rationale for the incorporation of
D-glutamate into the libraries is based on the modest
activity of this enantiomer as an inhibitor of VGLUT.11
Overall, stereoisomeric tetrapeptides X1X2EW(F) were
prepared and evaluated as VGLUT inhibitors (Fig. 3,
Table 1).
Method 1 and
(D,L) Fmoc-F-OH or
(D,L) Fmoc-W(tBoc)-OH
F or W(tBoc)-NH2
Method 2 and
(D,L) Fmoc-E-(γ-OtBu)-OH
F or W(tBoc)-E(γ-OtBu)-NH2
Peptide synthesis.16 Tetrapeptides were synthesized
according to Scheme 1 and structures determined by
NMR and/or mass spectrometry.
1. Method 2 and
(D,L) Fmoc-X2-OH
2. Method 2 and
(D,L) Fmoc-X1-OH
F or W(tBoc)-E(γ-OtBu)-X2-X1
Inhibition of VGLUT by tetrapeptides.17 Screening of
the peptide libraries as VGLUT inhibitors was carried
1. 7:2:1, CH2Cl2/HOAc/CF3CH2OH
2. 95:2.5:2.5, CF3CO2H/H2O/(iPr)3SiH
3
out using H-glutamate as substrate and the ability of
3
test compounds to block the uptake of H-glutamate
H2N-X1-X2-E-(F or W)-CO2H
into synaptic vesicles isolated from rat forebrain. Tetra-
peptide sub-libraries of the type X1X2EW (Library 1)
or X1X2EF (Library 2), where X1 and X2 were varied
as amino acids (AA) in D- or L- form (except cysteine),
and where the identity of X1 was known, were tested as
inhibitors of VGLUT.18 The sub-libraries were screened
and the pools showing the most inhibition of uptake
were deconvoluted to identify X2 residues (Table 1).
Method 1. EtN(iPr)2/CH2Cl2/DMF
then 20% piperidine/DMF
Method 2. HOBt, DIPCDI, CH2Cl2/DMF
then 20% piperidine/DMF
Scheme 1. Synthesis of target tetrapeptides containing a glutamate at
position 3.
We next reasoned that the various stereoisomers com-
prising a deconvoluted sub-library would be approxi-
mately equal in concentration, allowing the assay to be
a reasonable guide to selection of the pools used for fur-
ther deconvolution. In library 1, the only tetrapeptide
sublibrary to inhibit glutamate uptake at VGLUT was
structures with X1 = Q (56% uptake). Systematic
screening of the QX2EW tetrapeptides indicated the
CO2H
O
W(F)
X1 X2
N
H
O
Figure 3. Tetrapeptide design based on the QDC-template.

S. A. Patel et al. / Bioorg. Med. Chem. Lett. 17 (2007) 5125–5128
5127
order of inhibitory potency for position X2 as I >> Q,
W > N. Therefore, QIEW was selected for deconvolu-
tion into its stereoisomers, six of which inhibited the up-
take of glutamate: LDLD (28%), DDLD (35%), LDDD
(60%), DDDD (63%), LDLL (69%), and LLDL (82%). The
QWEW panel was active but allowed 66% residual
transport. Based on the lesser activity and solubility lim-
its, this panel was not deconvoluted for further analysis.
Although a glutamic acid is fixed at position 3 in all li-
braries, charged residues at positions X1 and X2 resulted
in poor VGLUT inhibition. For X2, both libraries prefer
hydrophobic or amide-containing residues. Q, N, W,
and I consistently appeared in the panels that showed
inhibition of glutamate uptake. This finding both corre-
19
lates with the structure of bromocryptine (Fig. 1),
a
‘peptide containing’ alkaloid, and with the reported
pharmacophore. Bromocryptine is a potent inhibitor
that lacks an acidic side chain group like glutamate
but contains homologous lipophilic residues along a
peptide-like framework.
The results from Library 2 showed that aryl or amide
residues are preferred at position X1 (N-term) with less
uptake allowed following the order W (36%) < N
(63%). As found with Library 1, Library 2 showed activ-
ity with a wider range of possible residues occupying po-
sition 2. For X2 in Library 2 (WX2EF), the order was N
(13%) > Q, W > I, F > H. As with library 1, the WNEF
tetrapeptide was selected for further stereoisomer decon-
volution. Interestingly, the individual enantiomers were
not as active as the mixture and only one isomer,
DWLNDEDF, showed inhibition of glutamate uptake
(41%).
In an effort to gain insight into the plausible interactions
with VGLUT as a function of the QDC-based ligand de-
sign features (Fig. 2) and the inhibition (Table 1, Fig. 3)
an initial alignment of an extended conformation of the
stereoisomer LQDILEDW (C- to N-terminal) was made
against our established inhibitor-based VGLUT phar-
macophore model (Fig. 5).1
The inhibitor pharmacophore model, derived using the
automated three-dimensional (3D) ligand alignment
protocol GASP (Tripos Inc., St. Louis), is defined with
To better understand the inhibitory activity, a more
thorough analysis of DQLIDELW was conducted. Using
the Cheng–Prusoff relationship to estimate from IC50
values, the Ki was determined from non-linear regres-
sion analyses of sigmoidal inhibitory dose–response
curves. The inhibitory dose–response for DQLIDELW
was plotted (Fig. 4) and the Ki was determined to be
828 252 lM as means SEM (n = 3). In separate
experiments, it was determined that neither QIEW nor
WNEF blocks v-ATPase activity in rat synaptic vesicles
at the concentrations tested for VGLUT inhibition.
6-(40-biphenyl)-quinoline-2,4-dicarboxylate
(QDC,
gray), one half of the symmetric structure of Chicago
Sky Blue (yellow, CSB), and bromocryptine (orange,
BCP). The superposition model describes three distinct
regions in 3D space, including c-carboxyl groups,
H-bonding acceptors, and lipophilic pocket moieties
brought together with discrete inter-atomic distances
˚
between the regional points (A values, Fig. 5). Since
small peptides may assume a host of conformations,
a plausible extended conformation of LQDILEDW was
selected for
comparison.
a
VGLUT pharmacophore model
Analysis of the LQDILEDW conformation alignment re-
veals that the N-terminal arginine (Q) correlates to the
Figure 5. An extended conformation of the tetrapeptide LQDILEDW
(green) aligned in relation to a computationally derived VGLUT
superposition pharmacophore model.1 composed with QDC (gray),
one half of CSB (yellow), and BCP (orange), in which the ligands are
correlated together with defined common pharmacophore comparison
points (c-carboxyl group, H-bonding acceptor, and lipophilic pocket).
Inter-atomic distances noted.
Figure 4. Demonstration of the inhibitory dose–response dependency
of DQLIDELW on the uptake of 3H-L-glutamate (0.25 mM) into rat
brain synaptic vesicles. Representative plot yielded a Ki = 0.465 mM.
The control rate for L-glutamate uptake was 1912 nmol/min/mg
protein and Km = 2.1 mM.

5128
S. A. Patel et al. / Bioorg. Med. Chem. Lett. 17 (2007) 5125–5128
c-carboxyl model region, the glutamic acid (E) side
chain carboxyl moiety corresponds to the pharmaco-
phore H-bonding acceptor group, and the C-terminal
tryptophan (W) is consistent with the aromatic ring lipo-
philic pocket superposition area. Taken together, QIEW
may block glutamate transport at VGLUT because it
contains three ligand structural binding elements similar
in 3D space to the more potent inhibitor ligand pharma-
cophore groups. Although the lesser potency of QIEW
likely results from the conformational flexibility, its
identity as a VGLUT inhibitor is consistent with the ba-
sic QDC-template peptide design strategy (Fig. 2).
References and notes
1. Thompson, C. M.; Davis, E.; Carrigan, C. N.; Cox, H. D.;
Bridges, R. J.; Gerdes, J. M. Curr. Med. Chem. 2005, 12,
2041.
2. Fykse, E. M.; Fonnum, F. Neurochem. Res. 1996, 21,
1053.
3. Maycox, P. R.; Deckwerth, T.; Hell, J. W.; Jahn, R. J.
Biol. Chem. 1988, 263, 15423.
4. Tabb, J. S.; Kish, P. E.; Van Dyke, R.; Ueda, T. J. Biol.
Chem. 1992, 267, 15412.
5. Wolosker, H.; de Souza, D. O.; de Meis, L. J. Biol. Chem.
1996, 271, 11726.
6. Cidon, S.; Sihra, T. S. J. Biol. Chem. 1989, 264, 8281.
7. Hartinger, J.; Jahn, R. J. Biol. Chem. 1993, 268, 23122.
8. Bartlett, R. D.; Esslinger, C. S.; Thompson, C. M.;
Bridges, R. J. Neuropharmacology 1998, 37, 839.
9. Bridges, R. J.; Lovering, F. E.; Koch, H.; Cotman, C. W.;
Chamberlin, A. R. Neurosci. Lett. 1994, 174, 193.
10. Garlin, A. B.; Sinor, A. D.;Sinor, J. D.;Jee, S. H.;Grinspan,
J. B.; Robinson, M. B. J. Neurochem. 1995, 64, 2572.
11. Naito, S.; Ueda, T. J. Neurochem. 1985, 44, 99.
12. Bridges, R. J.; Kavanaugh, M. P.; Chamberlin, A. R.
Curr. Pharm. Des. 1999, 5, 363.
13. Keller, B. U.; Blaschke, M.; Rivosecchi, R.; Hollmann,
M.; Heinemann, S. F.; Konnerth, A. Proc. Natl. Acad. Sci.
U.S.A. 1993, 90, 605.
14. Carrigan, C. N.; Bartlett, R. D.; Esslinger, C. S.; Cybulski,
K. A.; Tongcharoensirikul, P.; Bridges, R. J.; Thompson,
C. M. J. Med. Chem. 2002, 45, 2260.
15. Carrigan, C. N.; Esslinger, C. S.; Bartlett, R. D.; Bridges, R.
J.;Thompson, C.M. Bioorg. Med.Chem.Lett. 1999,9, 2607.
16. Peptides were synthesized with DIPCDI/HOBt using
Fmoc methodology on a Cl–Trt–Cl (2-chlorotrityl chlo-
ride) resin and cleaved from the resin bead by treatment
with 4:1 CH2Cl2/HOAc for 3 h. Peptides were treated with
TFA to remove protecting groups.
17. Vesicular transport was quantified as described Kish, P. E.;
Ueda, T. Methods Enzymol. 1989, 174, 9, Assays were
initiated by the addition of 3H-L-glutamate inhibitors
(0.01–5 mM) to the synaptic vesicles (approx. 0.1 mg
protein). Rates of uptake were normalized to protein
content.
Moreover, examination of the Q-X2-E-W library
reveals an enhanced peptide inhibition efficacy when
X2 = isoleucine (I). Thus, ligands with terminal aro-
matic ring groups and an additional lipophilic side
chain residue (e.g., W and I of QIEW; W and F
of WNEF, not shown) might be related to the dis-
tinct dual lipophilic moieties (the terminal aromatic
rings and distal propyl groups) of the more potent
VGLUT inhibitor bromocryptine. Collectively, the
lipophilic peptide and bromocryptine moieties could
be important structural facets for enhanced VGLUT
inhibition. Full conformational analyses of QIEW
and WNEF and their superposition within the phar-
macophore model are currently underway to discern
the relative 3D arrangement of the lipophilic
groups.
The discovery of tetrapeptide inhibitors raises the possi-
bility that they may represent protein motifs that may
bind VGLUT. BLAST analysis (rodentia)20,21 matched
a Ca+-transporting ATPase (WNEF), neuroprotective
protein (QIEW), and vasohibin (QIEW). We are cur-
rently examining the possibility of protein binding to
VGLUT using these leads.
Acknowledgments
18. Highly lipophilic peptides displayed poor solubility and
were first solubilized in acetonitrile and then diluted into
the Hepes assay buffer.
19. Carlson, M. D.; Kish, P. E.; Ueda, T. J. Biol. Chem. 1989,
264, 7369.
20. Altschul, S. F.; Koonin, E. V. Trends Biochem. Sci. 1998,
23, 444.
21. Altschul, S. F.; Madden, T. L.; Schaffer, A. A.; Zhang, J.;
Zhang, Z.; Miller, W.; Lipman, D. J. Nucleic Acids Res.
1997, 25, 3389.
This research was made possible by grants from the
NIH NS38248 (C.M.T.) and P20 RR15583 from the
COBRE Program of the National Center for Research
Resources. We are grateful for support to the Molecular
Computational Core Facility (NIH NOT-RR-02-005)
and expert assistance of Rohn Wood and Dr. Wes
Smith.