A family of highly selective allosteric modulators of the metabotropic glutamate receptor subtype 5

Article abstract of DOI:10.1124/mol.64.3.731

We have identified a family of highly selective allosteric modulators of the group I metabotropic glutamate receptor subtype 5 (mGluR5). This family of closely related analogs exerts a spectrum of effects, ranging from positive to negative allosteric modu

Full text of DOI:10.1124/mol.64.3.731

0026-895X/03/6403-731–740$7.00
MOLECULAR PHARMACOLOGY
Copyright © 2003 The American Society for Pharmacology and Experimental Therapeutics
Mol Pharmacol 64:731–740, 2003
Vol. 64, No. 3
2516/1089829
Printed in U.S.A.
A Family of Highly Selective Allosteric Modulators of the
Metabotropic Glutamate Receptor Subtype 5
JULIE A. O’BRIEN, WEI LEMAIRE, TSING-BAU CHEN, RAYMOND S. L. CHANG, MARLENE A. JACOBSON,
´
SOOKHEE N. HA, CRAIG W. LINDSLEY, HERVE J. SCHAFFHAUSER, CYRILLE SUR, DOUGLAS J. PETTIBONE,
P. JEFFREY CONN, and DAVID L. WILLIAMS, JR.
Neuroscience-WP (J.A.O., W.L., T.-B.C., R.S.L.C., M.A.J., C.S., D.J.P., P.J.C., D.L.W.) and Medicinal Chemistry (C.W.L.), Merck Research
Laboratories, West Point, Pennsylvania; Molecular Systems, Merck Research Laboratories, Rahway, New Jersey (S.N.H.); and Neurobiology,
Merck Research Laboratories, San Diego, California (H.J.S.)
Received March 11, 2003; accepted June 9, 2003
This article is available online at http://molpharm.aspetjournals.org
ABSTRACT
We have identified a family of highly selective allosteric modu- in fluorometric Ca2ϩ assays, whereas the analog 3,3Ј-dichloro-
lators of the group I metabotropic glutamate receptor subtype benzaldazine (DCB) does not exert any apparent modulatory
5 (mGluR5). This family of closely related analogs exerts a effect on mGluR5 activity. However, DCB seems to act as an
spectrum of effects, ranging from positive to negative allosteric allosteric ligand with neutral cooperativity, preventing the pos-
modulation, and includes compounds that do not themselves itive allosteric modulation of mGluRs by DFB as well as the
modulate mGluR5 agonist activity but rather prevent other fam- negative modulatory effect of DMeOB. None of these analogs
ily members from exerting their modulatory effects. 3,3Ј-Diflu- affects binding of [3H]quisqualate to the orthosteric (glutamate)
orobenzaldazine (DFB) has no agonist activity, but it acts as a site, but they do inhibit [3H]3-methoxy-5-(2-pyridinylethy-
selective positive allosteric modulator of human and rat nyl)pyridine binding to the site for 2-methyl-6-(phenylethynyl)-
mGluR5. DFB potentiates threshold responses to glutamate, pyridine, a previously identified negative allosteric modulator.
quisqualate, and 3,5-dihydroxyphenylglycine in fluorometric With the use of these compounds, we provide evidence that
Ca2ϩ assays 3- to 6-fold, with EC50 values in the 2 to 5 ␮M allosteric sites on GPCRs can respond to closely related li-
range, and at 10 to 100 ␮M, it shifts mGluR5 agonist concen- gands with a range of pharmacological activities from positive
tration-response curves approximately 2-fold to the left. The to negative modulation as well as to neutral competition of this
analog 3,3Ј-dimethoxybenzaldazine (DMeOB) acts as a nega- modulation.
tive modulator of mGluR5 agonist activity, with an IC50 of 3 ␮M
Metabotropic glutamate receptors (mGluRs) are G protein- ily located presynaptically and regulate the release of neuro-
coupled receptors that bind glutamate to exert a modulatory transmitters, including glutamate. Group II and III mGluRs
influence on neuronal excitability and synaptic transmission are coupled to G␣i and its associated effectors, such as ade-
in the central nervous system. The eight known members of nylate cyclase. These latter two groups are distinguished
the mGluR subfamily have been divided into three groups on from each other by their pharmacology; selective agonists
the basis of their sequence identity, pharmacology, and pre- and antagonists have been identified for each group (Conn
ferred signal transduction mechanism. Group I mGluRs and Pin, 1997). All mGluR subtypes possess a large (ϳ560
(mGluRs 1 and 5) are primarily localized postsynaptically amino acids) extracellular amino-terminal domain that con-
where they modulate ion channel activity and neuronal ex- tains the glutamate agonist binding site. Thus, in the
citability. The group I mGluRs are coupled to G␣q and its mGluRs, interaction of the agonist with the transmembrane
associated effectors, such as phospholipase C. Groups II domains is believed to be indirect (O’Hara et al., 1993; Conn
(mGluRs 2 and 3) and III (mGluRs 4, 6, 7, and 8) are primar- and Pin, 1997; Bockaert and Pin, 1999).
ABBREVIATIONS: mGluR, metabotropic glutamate receptor; DFB, 3,3Ј-difluorobenzaldazine; DCB, 3,3Ј-dichlorobenzaldazine; DMeOB, 3,3Ј-
dimethoxybenzaldazine; methoxy-PEPy, 3-methoxy-5-(2-pyridinylethynyl)pyridine; DHPG, 3,5-dihydroxy-phenylglycine; MPEP, 2-methyl-6-(phe-
nylethynyl)-pyridine; GPCR, G protein-coupled receptor; NPS 2390, 2-quinoxaline-carboxamide-N-adamantan-1-yl; SIB-1757, 6-methyl-2-(phe-
nylazo)-3-pyridinol; SIB-1893, (E)-2-methyl-6-(2-phenylethenyl)pyridine; CPCCOEt, 7-(hydroxyimino)cyclopropa[b]chromen-1a-carboxylate ethyl
ester; PHCCC, N-phenyl-7-(hydroxyimino)cyclopropa[b]chromen-1a-carboxamide; BAY 36-7620, (3aS,6aS)-6a-naphtalen-2-ylmethyl-5-methyli-
den-hexahydro-cyclopenta[c]furan-1-on; EM-TBPC, 1-ethyl-2-methyl-6-oxo-4-(1,2,4,5-tetrahydro-benzo[d]azepin-3-yl)-1,6-dihydro-pyrimidine-
5-carbonitrile; R214127, 1-(3,4-dihydro-2H-pyrano[2,3-b]quinolin-7-yl)-2-phenyl-1-ethanone; LCMS, liquid chromatography mass spectrometry;
TFA, trifluoroacetic acid; ELSD, evaporative light-scattering detection; HRMS, high-resolution mass spectrometry; CHO, Chinese hamster ovary;
FLIPR, fluorometric imaging plate reader; DMSO, dimethyl sulfoxide; KRB, Krebs’ bicarbonate buffer; PI, phosphatidyl inositol.
731

732
O’Brien et al.
It has been suggested that the group I receptor mGluR5 other family members from exerting their modulatory effects.
may play a role in a number of central nervous system These findings expand our understanding of allosteric sites
disease states, including pain (Salt and Binns, 2000; Bhave on GPCRs and suggest that small molecules can interact
et al., 2001), anxiety (Spooren et al., 2000; Tatarczynska et with allosteric sites with a range of intrinsic activities in a
al., 2001), addiction to cocaine (Chiamulera et al., 2001), and manner analogous to what is seen at traditional neurotrans-
schizophrenia (Chavez-Noriega et al., 2002). We are inter- mitter binding sites.
ested in mGluR5 function and initiated an effort to identify
and develop novel pharmacological tools to study the role of
mGluR5 in normal and pathological states. Most of the cur-
Materials and Methods
Compounds
rently available agonists and antagonists used to study
mGluR function have been developed as structural analogs of
glutamate, quisqualate, or phenylglycine (Schoepp et al.,
1999) and act through binding at or near the orthosteric
agonist site in the amino-terminal domain. Although it has
been possible to develop group-selective agonists and antag-
onists using this approach, it has been considerably more
challenging to obtain potent, agonist-site compounds that are
subtype-selective. A novel alternative approach has been to
identify compounds that act through allosteric mechanisms,
modulating the receptor by binding to sites different from the
highly conserved orthosteric binding site. Receptor modula-
tion by binding to an allosteric site may afford very high
subtype selectivity, assuming there would be less pressure to
conserve such a site within a subfamily of receptors. NPS
2390 (van Wagenen et al., 1998) and SIB-1757 and SIB-1893
(Varney et al., 1999) were the earliest compounds reported as
noncompetitive antagonists of mGluR1 and mGluR5, respec-
tively. The potent, noncompetitive group I antagonists MPEP
(mGluR5-selective) (Gasparini et al., 1999) and CPCCOEt
(mGluR1-selective) (Litschig et al., 1999) were the first re-
ported examples of negative allosteric modulators of group I
mGluRs. These compounds inhibit the agonist activation of
group I mGluRs in functional assays but do not affect agonist
affinity. They have been shown to exert their effects by bind-
ing to the transmembrane region of the receptors (Pagano et
al., 2000). Additional examples of negative allosteric modu-
lators of group I mGluRs include PHCCC (Annoura et al.,
1996), a structural analog of CPCCOEt, and BAY 36-7620
(mGluR1-selective) (Carroll et al., 2001); methoxymethyl-3-
[(2-methyl-1,3-thiazol-4-yl)ethynyl]pyridine, [3H]3-methoxy-
5-(2-pyridinylethynyl)pyridine (methoxy-PEPy), and 3-[(2-
methyl-1,3-thiazol-4-yl)ethynyl]-pyridine (mGluR5-selective)
(Cosford et al., 2003a,b); EM-TBPC (rat mGluR1-selective
(Malherbe et al., 2003); and R214127 (mGluR1-selective)
(Lavreysen et al., 2003). Recently, Knoflach et al. (2001)
reported examples of potent, positive allosteric modulators of
mGluR1a. These compounds also were shown to bind to the
transmembrane regions of the receptors.
A generalized synthetic scheme for the compounds reported in this
article is shown in Fig. 1.
Representative Synthesis (3,3-Difluorobenzaldazine). Into a
5-ml Smithsynthesizer reaction vial (Personal Chemistry, Foxboro,
MA) with a stir bar was placed 3-fluorobenzaldehyde (530 ␮l, 5
mmol, 2.0 equivalents) in 2.5 ml of acetonitrile. Then hydrazine (80
␮l, 2.5 mmol, 1.0 equivalents) was added, and the vial was capped.
The vial was then heated in the Smithsynthesizer reaction cavity at
150°C for 5 min. After 5 min, the reaction vial was rapidly cooled to
40°C. Upon removal from the reactor cavity, a bright yellow precip-
itate was collected by filtration and washed with water. After drying
overnight in a vacuum oven at 50°C, 378 mg (62%) of the title
compound was obtained as a bright yellow powder. Analytical LCMS
indicated a single peak (3.82 min, CH3CN/H2O/0.1% TFA, 4-min
gradient) that was Ͼ98% pure by UV (214 nm) and 100% pure by
ELSD analysis. 1H NMR (300 MHz, CDCl3) findings were the fol-
lowing: ␦ 8.61 (s, 2H), 7.59 (m, 4H), 7.43 (m, 2H), and 7.16 (m, 2H);
HRMS calculated for C14H10F2N2 (MϩH), was 245.0885; found
245.0867 (MϩH).
Analytical Data for Other Substituted Benzaldazines. For
2,2-difluorobenzaldazine, we obtained 408 mg (67%) of the title com-
pound as a bright yellow powder. Analytical LCMS indicated a single
peak (3.92 min, CH3CN/H2O/0.1% TFA, 4-min gradient) that was
Ͼ98% pure by UV (214 nm) and 100% pure by ELSD analysis. 1H
NMR (300 MHz, CDCl3) findings were the following: ␦ 8.93 (s, 2H),
8.13 (td, J ϭ 1.8, 7.35 Hz, 2H), 7.44 (m, 2H), 7.22 (m, 2H), and 7.13
(m, 2H); HRMS calculated for C14H10F2N2 (MϩH) was 245.0885, and
we found 245.0865 (MϩH). For 3,3-dimethoxybenzaldazine, we ob-
tained 536 mg (80%) of the title compound as a bright yellow floccu-
lent solid. Analytical LCMS indicated a single peak (3.65 min,
CH3CN/H2O/0.1% TFA, 4-min gradient) that was Ͼ98% pure by UV
(214 nm) and 100% pure by ELSD analysis. 1H NMR (300 MHz,
We have identified a family of selective allosteric modula-
tors of mGluR5. Unlike previously identified allosteric regu-
lators of GPCRs, this family of closely related analogs exerts
a spectrum of effects, ranging from positive to negative allo-
steric modulation, and includes compounds that do not them-
selves modulate mGluR5 agonist activity but rather prevent
Fig. 2. Quisqualate, glutamate, and DHPG activate Ca2ϩ transients in
human mGluR5-expressing CHO cells. Human mGluR5 CHO cells were
plated in clear-bottomed 384-well plates in glutamate/glutamine-free me-
dium, loaded the next day with the calcium-sensitive fluorescent dye
Fluo-4, and placed in FLIPR384. Agonists were added after 10 s of baseline
determination. Data were normalized to a glutamate (10 ␮M) control
maximum. Concentration-response curves were generated from the mean
data of five experiments. Error bars represent S.E.M. Similar results
were obtained for rat mGluR5 CHO cells. EC50 values for these cells are
shown here and are summarized in Table 3.
Fig. 1. Generalized synthesis scheme for benzaldazine analogs.

A Family of Allosteric mGluR5 Modulators
733
CDCl3) findings were the following: ␦ 8.63 (s, 2H), 7.44 (d, J ϭ 2.7 Hz, concentration of agonist (i.e., glutamate) for potentiation measure-
2H), 7.37 (s, 2H), 7.35 (m, 2H), 7.1 (m, 2H), and 3.88 (s, 6H); HRMS ments or inhibition measurements, respectively.
calculated for C16H16N2O2 (MϩH) was 269.1285, and we found
Reversibility of modulators was determined by adding a range of
269.1259 (MϩH). For 3,3-dichlorobenzaldazine, we obtained 346 mg concentrations of modulator to half of a plate of Fluo-4–loaded
(50%) of the title compound as a bright yellow solid. Analytical LCMS mGluR5 CHO cells (with vehicle added to the other half), allowing it
indicated a single peak (4.1 min, CH3CN/H2O/0.1% TFA, 4-min gra- to incubate for 5 min, washing with 100 ␮l of assay buffer 10 times,
dient) that was Ͼ98% pure by UV (214 nm) and 100% pure by ELSD allowing the plate to incubate for 15 min at 37°C, and washing
analysis. 1H NMR (300 MHz, CDCl3) findings were the following: ␦ another five times. Then the same range of concentrations of modu-
8.59 (s, 2H), 7.87 (m, 2H), 7.68 (dt, J ϭ 1.5, 7.2Hz, 2H), and 7.42 (m, lator was added to the vehicle-treated half of the plate, and vehicle
4H); HRMS calculated for C14H10Cl2N2 (MϩH) was 277.0294, and was added to the half originally treated with modulator and allowed
we found 277.0267 (MϩH).
to incubate for 5 min; then agonist was added in the FLIPR. The
response of the cells to agonist on the two halves of the plate were
compared to determine the effect of washing on the modulatory
effect. A rightward shift of the concentration-dependent modulatory
effect after washing indicated reversibility of the modulator. In the
case of compounds with neutral cooperativity, the assay was done as
above, except positive or negative modulators were added after all
wash steps to both halves of the plate in addition to the vehicle on the
treated half and the compound with neutral cooperativity on the
untreated half of the plate. The plate was allowed to incubate for 5
min, and agonist was added in the FLIPR. A rightward shift in the
effect of the neutrally cooperative compound in blocking the effect of
either the positive or negative allosteric modulator indicated revers-
ibility of the neutrally cooperative compound.
Cells
Cloned human mGluR5a receptors were transfected into CHO
cells (human mGluR5 CHO cells) using a cytomegalovirus-based
promoter to drive expression. Cells were grown in Dulbecco’s modi-
fied Eagle’s medium containing 10% dialyzed fetal bovine serum, 2
mM L-glutamine, 100 units/ml penicillin/streptomycin, nonessential
amino acids, 25 mM HEPES, 55 ␮M ␤-mercaptoethanol (all listed
materials were from Invitrogen, Carlsbad, CA), and 10 ␮g/ml puro-
mycin (BD Biosciences Clontech, Palo Alto, CA). Cloned rat
mGluR5a receptors were transfected, and the resulting cells (rat
mGluR5 CHO cells) were grown in the same manner. Group III
mGluRs were expressed in cell lines in the presence of chimeric or
promiscuous G proteins to make them compatible with assays mea-
suring increases in intracellular Ca2ϩ (mGluR4, mGluR7, and
mGluR8 were coexpressed with G␣qi5, G␣15, and G␣15, respectively).
Phosphoinositide Hydrolysis
Cells. CHO mGluR5 cells were loaded in their culture flasks
overnight with myo-[3H]inositol, suspended, and pooled. The radio-
tracer was removed, and glutamate-pyruvate transaminase, pyri-
Fluorometric Imaging Plate Reader
CHO cells expressing mGluR5 receptors (mGluR5 CHO cells) were doxal phosphate, and pyruvate were added to the buffer to reduce
plated in clear-bottomed, poly-D-lysine–coated 384-well plates (BD endogenous extracellular glutamate, and LiCl was added to inhibit
Biosciences, Franklin Lakes, NJ) in glutamate/glutamine-free me- the last step in the pathway of inositol trisphosphate breakdown.
dium using a Multidrop 384 cell dispenser (Thermo Electron Corpo- The cells were distributed to polypropylene culture tubes, and test
ration, Waltham, MA). The plated cells were grown overnight at compounds (serially diluted in 100% DMSO, then diluted in assay
37°C in the presence of 6% CO2. The next day, the cells were washed buffer to a 20ϫ stock at 1% DMSO, then added to assay for a final
with 3 ϫ 100 ␮l of assay buffer (Hanks’ balanced salt solution; DMSO concentration of 0.05%) or vehicle were added to each tube,
Invitrogen) containing 20 mM HEPES (Invitrogen), 2.5 mM probe- followed by quisqualate, an mGluR5 agonist that is not affected by
necid (Sigma Chemical Co., St. Louis, MO), and 0.1% bovine serum glutamate-pyruvate transaminase. The reaction was allowed to pro-
albumin (Sigma) using an EMBLA cell washer (Skatron, Lier, Nor- ceed for 45 min and was stopped by the addition of perchloric acid to
way). The cells were incubated with 1 ␮M Fluo-4AM (Molecular each tube. The perchloric acid was neutralized by the addition of
Probes, Eugene, OR) for 1 h at 37°C and 6% CO2. The extracellular KOH, and the contents of each tube were diluted and applied to
dye was removed by washing as described above. Ca2ϩ flux was Dowex 1 ϫ 8 columns (Dow Chemical Company, Midland, MI). The
measured using FLIPR384 fluorometric imaging plate reader (Molec- columns were washed with water and then with 60 mM ammonium
ular Devices Corp., Sunnyvale, CA). Compounds were serially di- formate and 5 mM sodium tetraborate. The columns were eluted
luted in 100% DMSO and then diluted in assay buffer to a 3ϫ stock with 1 M ammonium formate and 0.1 M formic acid. Scintillation
at 2% DMSO. This stock was then applied to the cells for a final
DMSO concentration of 0.67%. For potency determination, the cells
were preincubated with various concentrations of compound for 5
cocktail was added to the eluates, and the amount of radiolabel for
each sample was determine by scintillation counting.
Rat Hippocampal Slices. Rat pups (postnatal day 9) were de-
min and then stimulated for 3 min with either an EC20 or EC50 capitated, and their brains were removed and placed in ice-cold,
TABLE 1
Structures and activities of benzaldazine analogs
R
Allosteric Activity
Potency
␮M
3,3Ј-Difluorobenzaldazine
2,2Ј-Difluorobenzaldazine
4,4Ј-Difluorobenzaldazine
3,3Ј-Dimethoxybenzaldazine
4,4Ј-Dimethoxybenzaldazine
3,3Ј-Dichlorobenzaldazine
3,3Ј-F
Positive
Positive
Negative
Negative
Negative
Neutral
Neutral
Neutral
2.6 Ϯ 0.4
14 Ϯ 12
2,2Ј-F
4,4Ј-F
Ͼ100
3,3Ј-OCH3
4,4Ј-OCH3
3,3Ј-Cl
3.0 Ϯ 0.4
Ͼ100
7.6 Ϯ 1.9 vs. DFB
17 Ϯ 16 vs. DMeOB
Ͼ50
3,3Ј-Dihydroxybenzaldazine
3,3Ј-OH

734
O’Brien et al.
oxygenated Krebs’ bicarbonate buffer (KRB). The hippocampus was human mGluR5 CHO cells to a low concentration of gluta-
dissected on ice and sliced with use of a McIlwain tissue chopper
(Campden Instruments Ltd., Loughborough, UK). Slices were placed
in fresh ice-cold oxygenated KRB and dispersed, and then they were
placed in a 37°C water bath with gentle shaking. Note that decapi-
tation, dissection, and slicing were carried out to this point sepa-
rately for each animal to minimize the time between hippocampal
dissection and slice preparation. After the slices from the last animal
had incubated for 30 min, the slices were pooled in a conical-bot-
tomed centrifuge tube and allowed to settle by gravity. The super-
natant was aspirated off, and warmed oxygenated KRB was added to
mate (300 nM) without eliciting a response by themselves.
DFB was identified as active in this assay (Table 1).
DFB caused a concentration-dependent potentiation of the
response of human mGluR5 CHO cells to 300 nM glutamate
in this assay. The maximal potentiation at this concentration
of glutamate was approximately 3.1-fold, with an EC50 value
for potentiation of 2.6 Ϯ 0.4 ␮M (n ϭ 6, Fig. 3). As shown in
Table 2, DFB caused similar potentiation of the responses of
both human mGluR5 and rat mGluR5 CHO cells to gluta-
wash the slices. This was repeated two to three times. Finally, the mate, quisqualate, and DHPG with comparable potencies.
excess KRB was removed from the gravity-packed slices. Slices were
distributed to capped tubes containing 1 ␮Ci myo-[3H]inositol and
test compounds (diluted as a 50ϫ stock in 100% DMSO and then
added to assay for a final DMSO concentration of 2%), concluding
each addition by oxygenating and then capping the tubes. Slices in
the tubes were incubated at 37°C for 2 h. Then LiCl (to a final
concentration of 10 mM) and KRB or test compound (i.e., DFB) was
added to each tube, each of which was gassed and capped. The tubes
were incubated at 37°C for 15 min, and then agonist was added to the
tubes, which were again gassed and capped. After 20 min, chloro-
DFB also potentiated the glutamate response of human em-
bryonic kidney 293 cells expressing human mGluR5 (data not
shown). DFB alone (100 ␮M) caused no response on either
human or rat mGluR5 CHO (or human embryonic kidney
293) cells in this assay, indicating that the compound did not
act as an agonist.
Reversibility of DFB was determined by comparing the
FLIPR response of mGluR5 CHO cells to 300 nM glutamate
in the presence of a range of concentrations of DFB, with and
form/methanol (1:2) was added to each tube, and the tubes were without a washout step. In this series of experiments, the
vortexed for 1 min and then stood at room temperature for 15 min.
HCl and chloroform were added to each tube and mixed by vortexing.
Tubes were centrifuged to separate the phases, and an aliquot of the
aqueous (upper) phase was applied to a Dowex 1 ϫ 8 formate column.
The column was washed with 5 mM inositol and then 5 mM sodium
borate and 60 mM ammonium formate. The sample was eluted with
200 mM ammonium formate and 100 mM formic acid into scintilla-
tion vials. Scintillation cocktail (ReadySafe; Beckman Coulter, Ful-
lerton, CA) was added to each vial, and radioactivity was determined
by scintillation counting.
concentration-response curve of DFB was shifted Ͼ15-fold to
the right after washing (without washout, DFB EC50 ϭ 5.8 Ϯ
2.0 ␮M (n ϭ 3); with washout, DFB EC50 Ͼ 100 ␮M), indi-
cating that DFB was reversible.
In other functional studies, DFB did not potentiate the
activity of mGluR1, -2, -3, -4, -7, or -8. Although some antag-
onism (Ͼ30%) of mGluR4 and -8 occurred at 100 ␮M DFB,
none was observed at 30 ␮M. Thus, in FLIPR assays, the
effects of DFB (100 ␮M) as a percentage of the control re-
sponse were the following: mGluR1b, 80 Ϯ 10% at 50 ␮M
glutamate (ϳEC20 concentration); human mGluR4, 57 Ϯ 17%
and 71 Ϯ 12% at 2 ␮M and 30 ␮M glutamate, respectively
(ϳEC10 and ϳEC70 concentrations, respectively); human
mGluR7, 75 Ϯ 8% at 1 mM glutamate; and rat mGluR8, 25 Ϯ
Radioligand Binding Assays
The orthosteric agonist binding site radioligand [3H]quisqualate
and the allosteric inhibitor MPEP analog radioligand [3H]methoxy-
PEPy (Cosford et al., 2003) were used to evaluate the interaction of
DFB, DMeOB, and DCB with mGluR5. Membranes were prepared
from CHO mGluR5 cells or rat cerebral cortex. Aliquots of these
membranes were added to tubes containing test compound (serially
diluted in 100% DMSO, then diluted in assay buffer to a 4ϫ stock at
1% DMSO; stock was then added to assay for a final DMSO concen-
tration of 0.25%) or vehicle and either [3H]quisqualate (25 nM final
concentration in 20 mM HEPES, 2 mM CaCl2 and MgCl2, pH 7.2) or
[3H]methoxy-PEPy (1 or 2 nM final concentration in 50 mM Tris/
0.9% NaCl, pH 7.4). The tubes were incubated for 60 min at room
temperature with shaking, and the membrane-bound ligand was
separated from the free ligand by filtration onto glass-fiber filters
presoaked with 20 mM HEPES, 2 mM CaCl2 and MgCl2, pH 7.2, for
the [3H]quisqualate or with 0.2% polyethyleneimine for the [3H]me-
thoxy-PEPy. Filters were punched out into vials, Aquassure
(PerkinElmer Life Sciences, Boston, MA) was added, and membrane-
bound radioactivity was determined by scintillation counting of the
filters. Nondisplaceable binding was estimated using 1 mM gluta-
mate for the [3H]quisqualate assay and 1 ␮M MPEP for the [3H]me-
thoxy-PEPy assay.
Fig. 3. DFB potentiates mGluR5 activation by glutamate. mGluR5 CHO
cells were plated in clear-bottomed 384-well plates in glutamate/glu-
tamine-free medium, loaded the next day with the calcium-sensitive
fluorescent dye Fluo-4, and placed in FLIPR384. A range of concentrations
of DFB were added to cells (human mGluR5 CHO cells) after 10 s of
baseline determination. Five minutes later, a fixed concentration (ϳEC10
concentration) of agonist (glutamate) was added, and the Ca2ϩ response
was measured with the use of FLIPR384. Concentration-response curves
were generated from the mean data of six experiments. Error bars rep-
resent S.E.M. Results for glutamate, DHPG, and quisqualate on both
human and rat mGluR5 CHO cells are summarized in Table 2. Fold
potentiation was calculated from the maxima and minima determined by
Results
FLIPR Assays. In FLIPR384 assays, mGluR5 CHO cells
exhibited concentration-dependent increases in Fluo-4 fluo-
rescence in response to quisqualate, glutamate, and 3,5-di-
hydroxy-phenylglycine (DHPG) (Fig. 2). These compounds
seemed to act as full agonists, with potencies consistent with
EC50 values reported previously. Compounds were screened
in this assay for their ability to increase the response of nonlinear curve-fitting of the mean data.

A Family of Allosteric mGluR5 Modulators
735
1% and 59 Ϯ 15% at 400 nM and 10 ␮M glutamate, respec- 0.14 ␮M, n ϭ 3) (Fig. 6). This quisqualate-stimulated PI
tively (ϳEC10 and ϳEC70 concentrations, respectively). DFB hydrolysis was inhibited approximately 80% by MPEP (10
was also tested for any modulatory effect on the glutamate- ␮M), an mGluR5-selective negative allosteric modulator,
stimulated increases in guanosine 5Ј-O-(3-[35S]thio)triphos- suggesting that quisqualate was activating mGluR5 (data
phate binding to membranes prepared from human mGluR2 not shown). DFB (100 ␮M) shifted the quisqualate concen-
and human mGluR3-expressing cells. The effects of DFB as a tration-response curve to the left in this assay (EC50 ϭ 0.42 Ϯ
percentage of control response were the following: mGluR2, 0.10 ␮M, n ϭ 3) (Fig. 6), which is consistent with the data
100% and 81 Ϯ 7% at 1 and 100 ␮M glutamate, respectively obtained in the PI hydrolysis assays with the cloned recep-
(ϳEC5 and ϳEC90 concentrations, respectively); and tors expressed in CHO cells.
mGluR3, 100% at both 100 and 500 nM glutamate, respec-
Structure-Activity Relationships. In exploration of the
tively (ϳEC5 and ϳEC90 concentrations, respectively). DFB structure-activity relationships of DFB, we found that minor
(100 ␮M) also had no effect on responses to agonists of en- alterations in the structure afforded a variety of pharmaco-
dogenous G␣q-coupled receptors (e.g., 200 nM ATP, 200 nM logical profiles (Table 1). Changing the position of the fluoro
UTP, and 5 nM thrombin at purinergic or thrombin recep- groups from 3,3Ј to 2,2Ј decreased the potency of the poten-
tors) in the parental cell line (data not shown).
tiation of the FLIPR Ca2ϩ response caused by the compound
Increasing concentrations of DFB caused a parallel, left- by approximately 5-fold (EC50 ϭ 14 Ϯ 12 ␮M) without affect-
ward shift of mGluR5 CHO cell glutamate concentration- ing the maximal potentiation at this concentration of gluta-
response curves with no increase in maximal response (Fig. mate (Fig. 7). Replacement of the 3,3Ј fluoro groups on DFB
4). The maximum change in EC50 was 2.2-fold. Similar shifts with 3,3Ј methoxy groups (DMeOB) changed the pharmacol-
in agonist concentration-response curves were observed for ogy from potentiation of the glutamate response to negative
quisqualate and DHPG in both human and rat mGluR5 CHO
cells, again with no increase in the maximal response to the
agonists (Table 3).
Phosphatidyl Inositol Hydrolysis. To assess the effect
of DFB in other functional assays, phosphatidyl inositol hy-
drolysis assays were performed in mGluR5 CHO cells and in
rat brain slices. In the presence of LiCl, human mGluR5
CHO cells exhibited a concentration-dependent, quisqualate-
stimulated accumulation of inositol phosphates with an EC50
of 49.1 Ϯ 5.1 nM (n ϭ 3) and an Emax approximately 4.5-fold
over basal activity. DFB (10 ␮M) had no effect by itself on
basal activity in this assay, but it shifted the concentration-
response curve to quisqualate to the left (EC50 ϭ 27.1 Ϯ 2.6
nM, n ϭ 3, with no increase to maximal response) (Fig. 5),
which is consistent with the FLIPR fluorescence data. Re-
sults obtained for DHPG were similar to those for the agonist
(DHPG, EC50 ϭ 10.3 Ϯ 3.1 ␮M, n ϭ 3; DHPG ϩ 10 ␮M DFB,
EC50 ϭ 4.4 Ϯ 1.8 ␮M, n ϭ 3). In addition, similar results were
obtained with rat mGluR5 CHO cells with these two agonists
(quisqualate, EC50 ϭ 86.1 Ϯ 14.8 nM, n ϭ 3; quisqualate ϩ
Fig. 4. DFB potentiation of response to glutamate is manifested as
increased mGluR5 agonist sensitivity. mGluR5 CHO cells were plated in
clear-bottomed 384-well plates in glutamate/glutamine-free medium,
loaded the next day with the calcium-sensitive fluorescent dye Fluo-4,
and placed in FLIPR384. Several fixed concentrations of DFB were added
to cells (human mGluR5 CHO cells) after 10 s of baseline determination.
Five minutes later, a range of concentrations of agonist (glutamate) was
added. The leftward shift of the glutamate concentration-response curves
seems to approach a maximum at the highest concentrations of DFB.
Concentration-response curves were generated from the mean data of five
experiments. Error bars represent S.E.M. Results for glutamate, DHPG,
and quisqualate on both human and rat mGluR5 CHO cells are summa-
rized in Table 3 and were calculated from the EC50 values determined by
nonlinear curve-fitting of the mean data.
10 ␮M DFB, EC50 ϭ 28.3 Ϯ 8.8 nM, n ϭ 3; DHPG, EC50
ϭ
7.9 Ϯ 1.8 ␮M, n ϭ 4; DHPG ϩ 10 ␮M DFB, EC50 ϭ 3.0 Ϯ 0.4
␮M, n ϭ 3).
In hippocampal slices from young (postnatal day 9) rats,
quisqualate also stimulated phosphatidyl inositol (PI) hydro-
lysis in a concentration-dependent manner (EC50 ϭ 0.92 Ϯ
TABLE 3
TABLE 2
Effect of DFB on FLIPR agonist concentration-response curves: human
and rat mGluR5
EC50 change at 100 ␮M DFB was calculated from the EC50 values determined by
nonlinear curve fitting of the mean data (see Fig. 4) Data are expressed as mean Ϯ
S.E.M. (n)
DFB potentiation of human and rat mGluR5 activation by glutamate,
DHPG, and quisqualate
Maximum potentiation is the ratio of the maximum to the minimum response
determined by nonlinear curve fitting of mean data (see Fig. 3). Data are expressed
as mean Ϯ S.E.M. (n)
EC50 Change
EC50 for
Potentiation
Maximum
Potentiation
Agonist
Control Agonist EC50
at 100 ␮M
͓Agonist͔
DFB
␮M
-fold
-fold
Human mGluR5a CHO cells
300 nM Glutamate
1 ␮M DHPG
Human mGluR5a CHO cells
Glutamate
2.6 Ϯ 0.4 (6)
2.0 Ϯ 0.5 (6)
2.3 Ϯ 1.2 (6)
3.1
3.4
3.7
642 Ϯ 30 nM (5)
2.36 Ϯ 0.17 ␮M (5)
21.4 Ϯ 0.7 nM (5)
2.2
2.3
2.2
DHPG
8 nM Quisqualate
Rat mGluR5 CHO cells
300 nM Glutamate
1 ␮M DHPG
Quisqualate
Rat mGluR5 CHO cells
Glutamate
DHPG
Quisqualate
5.3 Ϯ 1.4 (5)
3.6 Ϯ 1.4 (5)
3.6 Ϯ 1.1 (4)
4.1
6.2
3.7
659 Ϯ 48 nM (5)
2.07 Ϯ 0.12 ␮M (5)
22.0 Ϯ 0.9 nM (5)
1.9
2.0
1.7
8 nM Quisqualate

736
O’Brien et al.
modulation with an IC50 of 3.0 Ϯ 0.4 ␮M) (Fig. 8). These
Substitution of the 3,3Ј-fluoro groups of DFB with 3,3Ј-
mGluR5 CHO cells exhibited no constitutive activity in chloro groups resulted in a compound (DCB) that acted to
FLIPR assays (mGluR5 receptor density was 240 fmol/mg), prevent the allosteric modulatory effects of other compounds
and it was therefore not possible to distinguish between in this structural class. For example, although DCB had no
antagonist-like or inverse agonist activity for DMeOB. The effect on the response of mGluR5 to glutamate alone, it
inhibitory effects of this compound are noncompetitive-like, reduced the ability of DFB to potentiate the FLIPR Ca2ϩ
because increasing concentrations of compound not only shift response of mGluR5 to 300 nM glutamate (Fig. 10A). In Fig.
the concentration-response curves of glutamate and quis- 10A, the upper and lower curves correspond to the responses
qualate to the right, but they also reduce the maximal re- of 600 and 300 nM glutamate, respectively, in the presence of
sponse to agonist similar to the behavior reported for MPEP increasing concentrations of DCB. At concentrations as high
(Fig. 9). Reversibility of DMeOB was determined by compar- as 100 ␮M, DCB has very little effect on the FLIPR Ca2ϩ
ing the FLIPR response of mGluR5 CHO cells to 600 nM
glutamate in the presence of a range of concentrations of
DMeOB, with and without a washout step. The concentra-
tion-response curve of DMeOB under these conditions was
shifted Ͼ15-fold to the right after washing, indicating that
DMeOB was reversible (data not shown).
Fig. 7. Positional isomers of DFB exhibit different potencies of potenti-
ation. Human mGluR5 CHO cells were plated in clear-bottomed 384-well
plates in glutamate/glutamine-free medium, loaded the next day with the
calcium-sensitive fluorescent dye Fluo-4, and placed in FLIPR384. A range
of concentrations of 2,2Ј-DFB and 3,3Ј-DFB were added to the cells after
10 s of baseline determination. Five minutes later, a fixed concentration
(ϳEC10 concentration) of glutamate was added, and the Ca2ϩ response
was measured with the use of FLIPR384. Concentration-response curves
Fig. 5. DFB potentiates quisqualate-stimulated PI hydrolysis in human
mGluR5 CHO cells. Phosphatidyl inositol hydrolysis was measured in
human mGluR5 CHO cells as outlined under Materials and Methods.
DFB (10 ␮M) had no effect by itself on basal activity. A representative
experiment is shown. Error bars represent S.D. Similar results were
obtained with rat mGluR5 CHO cells with these two agonists.
were generated from the mean data of five experiments. Error bars
represent S.E.M.
Fig. 8. DMeOB causes the inhibition of glutamate response. Human
mGluR5 CHO cells were plated in clear-bottomed 384-well plates in
glutamate/glutamine-free medium, loaded the next day with the calcium-
sensitive fluorescent dye Fluo-4, and placed in FLIPR384. A range of
concentrations of DMeOB was added to the cells after 10 s of baseline
determination. Five minutes later, a fixed concentration (ϳEC10 concen-
tration) of glutamate was added, and the Ca2ϩ response was measured
with the use of FLIPR384. Concentration-response curves were generated
from the mean data of three experiments. Error bars represent S.E.M.
Fig. 6. DFB potentiates quisqualate-stimulated PI hydrolysis in rat
postnatal day 9 (p9) hippocampal slices. Phosphatidyl inositol hydrolysis
was measured in rat postnatal day 9 hippocampal slices as described
under Materials and Methods. A representative experiment is shown.
Error bars represent S.D.

A Family of Allosteric mGluR5 Modulators
737
response of mGluR5 to glutamate alone. In the presence of 6 was weak (Ͼ50–100 ␮M), and they were not characterized
␮M DFB, the response of mGluR5 to 300 nM glutamate is further (Table 1).
potentiated in this system to approximate the response to
Neither DFB, DMeOB, nor DCB (each at 100 ␮M) had any
600 nM glutamate alone. The presence of DCB in this system effect on the binding of [3H]quisqualate (25 nM, Kd ϭ 110
attenuates the potentiation conferred by DFB in a concen- nM)) to membranes prepared from CHO mGluR5 cells (Fig.
tration-dependent manner (IC50 ϭ 7.6 Ϯ 1.9 ␮M).
12), suggesting that this series of analogs did not bind at the
In a similar manner, DCB relieved the inhibitory effect of orthosteric agonist binding site. Nondisplaceable [3H]quis-
8 ␮M DMeOB on the FLIPR Ca2ϩ response of mGluR5 to 600 qualate binding was estimated in the presence of 1 mM
nM glutamate (IC50 ϭ 17 Ϯ 16 ␮M, Fig. 10B) and blocked
approximately 50% of the inhibition of this system by 30 nM
MPEP (IC50 ϭ 3.0 Ϯ 0.9 ␮M, data not shown). Increasing
fixed concentrations of DCB shifted the MPEP inhibition
curves to the right in a parallel manner (Fig. 11).
Reversibility of DCB was determined by comparing the
FLIPR response of mGluR5 CHO cells to 300 nM glutamate
in the presence of 6 ␮M DFB or to 600 nM glutamate in the
presence of 8 ␮M DMeOB in the presence of a range of
concentrations of DCB, with and without a washout step. The
DFB or DMeOB were added after the washout step. The
concentration-response curve of DCB was shifted 3- to 4-fold
to the right after washing, indicating that DCB was revers-
ible (data not shown).
As was the case with DFB, none of the analogs reported
here interacted with human mGluR1b, -2, -3, or -7 or with
endogenous purinergic or thrombin receptors present in the
parental cell lines (data not shown). However, DMeOB was
found to act as an inhibitor of mGluR4 (IC50 of approximately
35 ␮M) and mGluR8 (IC50 of approximately 50 ␮M) activa-
tion. Also similar to findings with DFB, all analogs had the
same pharmacological profile on rat mGluR5a (data not
shown).
Several other close structural analogs exhibited activity as
negative modulators (4,4Ј-difluorobenzaldazine and 4,4Ј-di-
methoxybenzaldazine) or in preventing allosteric modulation
by other analogous compounds (3,3Ј-dihydroxybenzaldazine)
in ways similar to those of DMeOB and DCB, respectively.
However, the potency of these compounds in the FLIPR assay
Fig. 10. DCB reduces allosteric modulation by DFB and DMeOB.
mGluR5 CHO cells were plated in clear-bottomed 384-well plates in
glutamate/glutamine-free medium, loaded the next day with the calcium-
sensitive fluorescent dye Fluo-4, and placed in FLIPR384. A, a fixed
concentration of DFB (6 ␮M) and a range of concentrations of DCB were
added to the cells after 10 s of baseline determination. Five minutes later,
a fixed concentration (300 nM) of glutamate was added, and the Ca2ϩ
response was measured by FLIPR384. Squares represent the response to
300 nM glutamate (f, glutamate alone; Ⅺ, glutamate in the presence of
increasing concentrations of DCB). Circles represent the response to 600
nM glutamate (F, glutamate alone; E, glutamate in the presence of
increasing concentrations of DCB). , the response of 300 nM glutamate
potentiated by the presence of 6 ␮M DFB. B, a fixed concentration of
DMeOB (8 ␮M) and a range of concentrations of DCB were added to the
cells after 10 s of baseline determination. Five minutes later, a fixed
concentration (600 nM) of glutamate was added, and the Ca2ϩ response
was measured by FLIPR384. The symbols represent the same responses
described in A, except that Πrepresents the response of 600 nM gluta-
mate inhibited by the presence of 8 ␮M DMeOB. Concentration-response
curves were generated from the mean data of four experiments. Error
bars represent S.E.M.
Fig. 9. DMeOB seems to be a noncompetitive antagonist. mGluR5 CHO
cells were plated in clear-bottomed 384-well plates in glutamate/glu-
tamine-free medium, loaded the next day with the calcium-sensitive
fluorescent dye Fluo-4, and placed in FLIPR384. Several fixed concentra-
tions of DMeOB were added to cells (human mGluR5 CHO cells) after 10 s
of baseline determination. Five minutes later, a range of concentrations
of agonist (glutamate) was added. The concentration-response curves are
shifted to the right with increasing concentrations of DMeOB (EC50 for
glutamate ϭ 0.55 ␮M for vehicle control, 0.64 ␮M with 1 ␮M DMeOB,
1.36 ␮M with 10 ␮M DMeOB, and 3.84 ␮M with 100 ␮M DMeOB), and
the maximum response to glutamate is reduced at the highest concen-
trations of DMeOB. Concentration-response curves were generated from
the mean data of three experiments. Error bars represent S.E.M.

738
O’Brien et al.
glutamate. All three benzaldazine analogs inhibited binding thoxy-PEPy binding fully. MPEP (1 ␮M) was used to esti-
of the MPEP analog [3H]methoxy-PEPy (2 nM) (Cosford et mate nondisplaceable binding (9% of total binding). Similar
al., 2003) to these membranes (Fig. 13). DFB partially inhib- results were obtained using rat cortical membranes in place
ited [3H]methoxy-PEPy binding [51% inhibition, IC50
ϭ
of the human mGluR5-expressing CHO cell membranes.
8.5 Ϯ 4.4 ␮M (n ϭ 3)], whereas DCB and DMeOB inhibited
the binding more fully (75% and 92% inhibition, IC50 ϭ 5.8 Ϯ
2.6 ␮M (n ϭ 3) and 13.8 Ϯ 0.6 ␮M (n ϭ 3), respectively).
Saturation-binding studies using [3H]methoxy-PEPy suggest
that the presence of these analogs did not affect Bmax but did
increase Kd (Fig. 13B) (control, Bmax ϭ 143 fmol/mg protein,
Kd ϭ 4.7 Ϯ 0.4 nM; in the presence of 15 ␮M DFB, Bmax ϭ 139
fmol/mg protein, Kd ϭ 9.9 Ϯ 1.1 nM; in the presence of 6 ␮M
DMeOB, Bmax ϭ 141 fmol/mg protein, Kd ϭ 11.5 Ϯ 2.0 nM).
Under these conditions, MPEP was found to inhibit [3H]me-
Discussion
Potent and selective pharmacological tools are important
in the characterization of GPCR function, and in the case of
mGluRs, potent orthosteric agonists and antagonists have
been discovered (Conn and Pin, 1997; Schoepp et al., 1999).
With the advent of high-throughput functional assays, it has
been possible to expand the search for pharmacological tools
to include compounds that act on receptors at allosteric sites
rather than at the historically targeted orthosteric sites. For
the mGluRs, MPEP and CPCCOEt were the first compounds
that were clearly shown to be negative allosteric modulators
(selective for mGluR5 and mGluR1, respectively). Recently,
positive allosteric modulators selective for mGluR1b also
have been identified (Knoflach et al., 2001). In a continuation
of this approach of using functional assays to discover allo-
steric modulators, DFB was identified as a novel positive
allosteric modulator (i.e., potentiator) selective for mGluR5.
DFB was found to potentiate the activation of human and
rat mGluR5 receptors by glutamate, quisqualate, and DHPG,
although it was not itself an agonist. The potentiation seems
to be specific: it was observed for mGluR5 activation but not
for the activation of mGluR1, -2, -3, -4, -7, or -8 or for endog-
enous purinergic or thrombin receptors. Potentiation of
mGluR5 by DFB was observed in several functional assays in
cloned cells as well as tissue slices, and it is therefore un-
likely to be an artifact of the expression system.
Fig. 11. DCB reduces allosteric modulation by MPEP. mGluR5 CHO cells
were plated in clear-bottomed 384-well plates in glutamate/glutamine-
free medium, loaded the next day with the calcium-sensitive fluorescent
dye Fluo-4, and placed in FLIPR384. Several fixed concentrations of DFB
Studies to determine the structure-activity relationships of
DFB indicated that slight structural changes had profound
and a range of concentrations of MPEP were added to the cells after 10 s effects on the pharmacological function of the resulting com-
of baseline determination. Five minutes later, 600 nM glutamate was
pounds without altering selectivity for receptor subtype or
added, and the Ca2ϩ response was measured by FLIPR384. Concentration-
response curves were generated from the mean data of three experi-
ments. Data are plotted as a percentage of the maximum response to 600
nM glutamate. Error bars represent S.E.M.
Fig. 13. Benzaldazine analogs seem to interact with the MPEP binding
site. The assay conditions were the following: membranes prepared from
Fig. 12. Benzaldazine analogs have no effect on [3H]quisqualate binding
to human mGluR5 CHO cell membranes. The assay conditions were the
following: membranes prepared from human mGluR5 CHO cells were
incubated with [3H]quisqualate (25 nM final concentration in 20 mM
HEPES, 2 mM CaCl2 and MgCl2, pH 7.2) in the presence of 100 ␮M
compound or vehicle for 60 min at room temperature. Samples were
filtered onto glass fiber filters. Nondisplaceable binding (NSB) was esti-
mated with the use of 1 mM glutamate. A representative experiment is
shown. Error bars represent S.D.
human mGluR5 CHO cells were incubated with the radiolabeled MPEP
analog [3H]3-methoxy-5-(2-pyridinylethynyl)pyridine (1 or 2 nM final
concentration in 50 mM Tris and 0.9% NaCl, pH 7.4), for 60 min at room
temperature in the presence of varying concentrations of benzaldazine
analogs. Samples were then filtered onto glass fiber filters. Nondisplace-
able binding (NSB) was estimated with 1 ␮M MPEP. A representative
competition experiment is shown. Error bars show S.D. Inset, a repre-
sentative saturation experiment is shown in the presence and absence of
15 ␮M DFB.

A Family of Allosteric mGluR5 Modulators
739
species. Replacement of the fluoro groups with methoxy family 3 GPCR function that considers the possibility of
groups to give DMeOB changed the activity from potentia- variable coupling of the equilibrium between the open and
tion by the original compound to antagonism by the resulting closed conformations of the agonist binding region and the
compound. On the other hand, substitution of chloro groups equilibrium between the active and inactive conformations of
for the original fluoro groups to give DCB resulted in a the effector region. In this model, allosteric modulators could
compound that is neither an agonist, an inhibitor, nor an affect the coupling between these equilibria and/or the active/
allosteric potentiator. This otherwise pharmacologically “si- inactive conformational equilibrium of the effector region.
lent” compound was found to interact with mGluR5 to pre- Our results indicate that DFB did not increase the basal
vent the pharmacological actions of DFB and DMeOB. Thus, activity of the receptor (it did not exhibit apparent agonist
the compounds in this structural series exhibit a range of activity), which is not consistent with its action applied
pharmacological activities including positive allosteric mod- through altering the equilibrium between active and inactive
ulation, negative allosteric modulation, and neutral modula- conformations of the effector region. Parmentier et al. (2002)
tion (i.e., preventing positive or negative modulation). We also indicated that the maximum effect and affinity of ago-
have not found a compound in this series that exhibits ago- nists would increase in the presence of a positive allosteric
nist-like effects.
modulator. As noted above, we did not observe any effect of
The very close structural homology between these com- DFB on [3H]quisqualate binding, nor did we observe an in-
pounds and the ability of DCB to block completely both crease in the maximal effect of glutamate, DHPG, or quis-
potentiation by DFB as well as antagonism by DMeOB is qualate in the presence of DFB. At present, we do not have an
consistent with these compounds binding to the same site on explanation for this discrepancy between our results with
mGluR5. Binding studies are consistent with the interaction DFB and the results predicted for a positive allosteric mod-
of these analogs with the MPEP binding site. They inhibited ulator by this model. The alternative mode of action proposed
the binding of the MPEP analog [3H]methoxy-PEPy with by this model, involving alteration of the tightness of cou-
IC50 values in a potency range similar to that observed for pling between the agonist binding and effector regions of the
their functional effects. Saturation binding studies with receptor, may be consistent with the alteration in agonist
[3H]methoxy-PEPy were consistent with competitive inhibi- intrinsic efficacy suggested by the Ehlert model.
tion, because Bmax was not changed but Kd was increased.
Although receptors by definition exert their actions
Although inhibition of [3H]methoxy-PEPy binding by some of through allosteric mechanisms, it is only recently that inter-
these compounds was not complete, this may have been est in allosteric modulation of GPCR action by small mole-
caused by poor solubility in the longer binding assays, be- cules has progressed to active searches for pharmacological
cause data in other assays indicated a full effect of these tools acting through these mechanisms. The series of closely
compounds at 100 ␮M. Solubility may also play a role in the related analogs described in this article was identified from
slight reduction in the maximal response to glutamate in the an active search for such compounds, and they are members
presence of 100 ␮M DCB (Fig. 11). Most of these data are of a new class of pharmacological tools that includes the first
consistent with MPEP and the benzaldazine analogs binding reported example of an mGluR5-selective positive allosteric
to the same site, although the lack of complete inhibition of modulator, as well as an mGluR5-selective negative alloste-
[3H]methoxy-PEPy binding is not consistent with this con- ric modulator, and a pharmacologically “silent” allosteric li-
clusion. Identification of the amino-acid interactions that are gand with neutral cooperativity. The wide range in pharma-
important for the various pharmacologies of these com- cological profiles induced by subtle structural changes was
pounds will help clarify this and is currently being pursued unexpected, and therefore this set of tools will probably prove
with the construction of chimeric and mutated mGluR5 re- useful in studying the function of mGluR5. We believe that
ceptors. The lack of species selectivity between human and these compounds represent a rare example of a series of
rat suggests that the binding site for these compounds is allosteric modulators of GPCRs that range in pharmacologi-
conserved between these species and further indicates that cal activity from positive to negative modulation as well as
they may be suitable for studying the function of mGluR5 in neutral prevention of this modulation.
both primates as well as rodents.
We have compared the data obtained for these allosteric
modulators with published models of allosterism. One of the
Acknowledgments
We thank Blake Rowe for testing these compounds for activity on
human mGluR2 and mGluR3.
simplest models (Ehlert, 1988) describes the action of posi-
tive allosteric modulators through an increase in the intrin-
sic efficacy of agonists and/or an increase in affinity of the
agonist for its binding site. The potency and solubility of
these compounds did not permit us to make full use of the
pharmacological null methods described in this article. How-
ever, the lack of effect of these compounds on [3H]quisqualate
binding suggests that they do not affect the affinity of ago-
References
Annoura H, Fukunaga A, Uesugi M, Tatsuoka T, and Horikawa Y (1996) A novel
class of antagonists for metabotropic glutamate receptors, 7-(hydroxyimino)cyclo-
prop[b]chromen-1a-caroxylates. Bioorg Med Chem Lett 6:763–766.
Bhave G, Karim F, Carlton SM, and Gereau RW (2001) Peripheral group I metabo-
tropic glutamate receptors modulate nociception in mice. Nat Neurosci 4:417–423.
Bockaert J and Pin J-P (1999) Molecular tinkering of G protein-coupled receptors: an
evolutionary success. EMBO (Eur Mol Biol Organ) J 18:1723–1729.
Carroll FY, Stolle A, Beart PM, Voerste A, Brabet I, Mauler F, Joly C, Antonicek H,
Bockaert J, Mu¨ller T, et al. (2001) BAY36-7620: a potent non-competitive mGlu1
receptor antagonist with inverse agonist activity. Mol Pharmacol 59:965–973.
Chavez-Noriega LE, Schaffhauser HJ-L, and Campbell UC (2002) Metabotropic
glutamate receptors: potential drug targets for the treatment of schizophrenia.
Curr Drug Target CNS Neurol Disord 1:261–281.
Chiamulera C, Epping-Jordan MP, Zocchi A, Marcon C, Cottiny C, Tacconi S, Corsi
M, Orzi F, and Conquet F (2001) Reinforcing and locomotor stimulant effects of
cocaine are absent in mGluR5 null mutant mice. Nat Neurosci 4:873–874.
nists.
Furthermore,
4-carboxy,3-hydroxyphenylglycine,
which has no apparent mGluR5 agonist effects on its own
and antagonizes mGluR5 responses to glutamate, is revealed
as a partial agonist in the presence of 100 ␮M DFB (data not
shown). These results suggest that the benzaldazine analogs
may act by affecting the intrinsic efficacy of agonists.
Recently, Parmentier et al. (2002) proposed a model of

740
O’Brien et al.
Conn PJ and Pin J-P (1997) Pharmacology and functions of metabotropic glutamate
receptors. Annu Rev Pharmacol Toxicol 37:205–237.
Cosford NDP, Roppe J, Tehrani L, Schweiger EJ, Seiders TJ, Chaudary A, Rao S, and
Varney MA (2003a) [3H]-methoxymethyl-MTEP and [3H]-methoxy-PEPy: potent
and selective radioligands for the metabotropic glutamate subtype 5 (mGlu5)
receptor. Bioorg Med Chem Lett 13:345–348.
Cosford NDP, Tehrani L, Roppe J, Schweiger E, Smith ND, Anderson J, Bristow L,
Brodkin J, Jiang X, McDonald I, et al. (2003b) 3-[(2-Methyl-1,3-thiazol-4-
yl)ethynyl]-pyridine: a potent and highly selective metabotropic glutamate sub-
type 5 receptor antagonist with anxiolytic activity. J Med Chem 46:204–206.
Ehlert FJ (1988) Estimation of the affinities of allosteric ligands using radioligand
binding and pharmacological null methods. Mol Pharmacol 33:187–194.
Gasparini F, Lingenho¨hl K, Stoehr N, Flor PJ, Heinrich M, Vranesic I, Biollaz M,
Allgeier H, Heckendorn R, Urwyler S, et al. (1999) 2-Methyl-6-(phenylethynyl)-
pyridine (MPEP), a potent, selective and systemically active mGlu5 receptor
antagonist. Neuropharmacology 38:1493–1503.
Knoflach F, Mutel V, Jolidon S, Kew JNC, Malherbe P, Vieira E, Wichmann J, and
Kemp JA (2001) Positive allosteric modulators of metabotropic glutamate 1 recep-
tor: characterization, mechanism of action and binding site. Proc Natl Acad Sci
USA 98:13402–13407.
Lavreysen H, Janssen C, Bischoff F, Langlois X, Leysen JE, and Lesage ASJ (2003)
[3H]R214127: a novel high-affinity radioligand for the mGlu1 receptor reveals a
common binding site shared by multiple allosteric antagonists. Mol Pharmacol
63:1082–1093.
Litschig S, Gasparini F, Ru¨egg D, Stoehr N, Flor PJ, Vranesic I, Prezeau L, Pin J-P,
Thomsen C, and Kuhn R (1999) CPCCOEt, a noncompetitive metabotropic gluta-
mate receptor 1 antagonist, inhibits receptor signaling without affecting gluta-
mate binding. Mol Pharmacol 55:453–461.
Malherbe P, Kratochwil N, Knoflach F, Zenner M-T, Kew JNC, Kratzeisen C, Maerki
HP, Adam G, and Mutel V (2003) Mutational analysis and molecular modeling of
the allosteric binding site of a novel, selective, noncompetitive antagonist of the
metabotropic glutamate 1 receptor. J Biol Chem 278:8340–8347.
O’Hara PJ, Sheppard PO, Thogerson H, Venezia D, Haldeman BA, McGrane V,
Houamed KM, Thomsen C, Gilbert TL, and Mulvihill ER (1993) The ligand-
binding domain in metabotropic glutamate receptors is related to bacterial
periplasmic binding proteins. Neuron 11:41–52.
Pagano A, Ru¨egg D, Litschig S, Stoehr N, Stierlin C, Heinrich M, Floersheim P,
Prezeau L, Carroll F, Pin J-P, et al. (2000) The non-competitive antagonists
2-methyl-6-(phenylethynyl) pyridine and 7-hydroxyiminocyclopropan[b]chromen-
1␣-carboxylic acid ethyl ester interact with overlapping binding pockets in the
transmembrane region of group I metabotropic glutamate receptors. J Biol Chem
275:33750–33758.
Parmentier M-L, Pre´zeau L, Bockaert J, and Pin J-P (2002) A model for the func-
tioning of family 3 GPCRs. Trends Pharmacol Sci 23:268–274.
Salt TE and Binns KE (2000) Contributions of mGlu1 and mGlu5 receptors to
interactions with N-methyl-D-aspartate receptor-mediated responses and nocicep-
tive sensory responses of rat thalamic neurons. Neuroscience100:375–380.
Schoepp DD, Jane DE, and Monn JA (1999) Pharmacological agents acting at
subtypes of metabotropic glutamate receptors. Neuropharmacology 38:1431–1476.
Spooren WPJM, Vassout A, Neijt HC, Kuhn R, Gasparini F, Roux S, Porsolt RD, and
Gentsch C (2000) Anxiolytic-like effects of the prototypical metabotropic glutamate
receptor 5 antagonist 2-methyl-6-(phenylethynyl)pyridine in rodents. J Pharmacol
Exp Ther 295:1267–1275.
Tatarczynska E, Klodzinska A, Chojnacka-Wojcik E, Palucha A, Gasparini F, Kuhn
R, and Pilc A (2001) Potential anxiolytic- and antidepressant-like effects of MPEP,
a potent, selective and systemically active mGlu5 receptor antagonist. Br J Phar-
macol 132:1423–1430.
van Wagenen B, Artman L, Hammerland L, Hung B, Johnson J, Krapcho K,
Levinthal C, Logan M, Moe S, et al. (1998) In vitro pharmacological characteriza-
tions of NPS2390: a highly potent and selective, non-competitive antagonist of
group-I metabotropic glutamate receptors. Abstr Soc Neurosci 24:57619.
Varney MA, Cosford NDP, Jachec C, Rao SP, Sacaan A, Lin F-F, Bleicher L, Santori
EM, Flor PJ, Allgeier H, et al. (1999) SIB-1757 and SIB-1893: selective, noncom-
petitive antagonists of metabotropic glutamate receptor type 5. J Pharmacol Exp
Ther 290:170–181.
Address correspondence to: Dr. David L. Williams, Jr., WP46-300, Merck Re-
search Laboratories, West Point, PA 19486. E-mail: david_williams1@merck.com