The Non-competitive Antagonists 2-Methyl-6-(phenylethynyl)pyridine and 7-Hydroxyiminocyclopropan[b]chromen-1a-carboxylic Acid Ethyl Ester Interact with Overlapping Binding Pockets in the Transmembrane Region of Group I Metabotropic Glutamate Receptors*

We have investigated the mechanism of inhibition and site of action of the novel human metabotropic glutamate receptor 5 (hmGluR5) antagonist 2-methyl-6-(phenylethynyl)pyridine (MPEP), which is structurally unrelated to classical metabotropic glutamate receptor (mGluR) ligands. Schild analysis indicated that MPEP acts in a non-competitive manner. MPEP also inhibited to a large extent constitutive receptor activity in cells transiently overexpressing rat mGluR5, suggesting that MPEP acts as an inverse agonist. To investigate the molecular determinants that govern selective ligand binding, a mutagenesis study was performed using chimeras and single amino acid substitutions of hmGluR1 and hmGluR5. The mutants were tested for binding of the novel mGluR5 radioligand [3H]2-methyl-6-(3-methoxyphenyl)ethynyl pyridine (M-MPEP), a close analog of MPEP. Replacement of Ala-810 in transmembrane (TM) VII or Pro-655 and Ser-658 in TMIII with the homologous residues of hmGluR1 abolished radioligand binding. In contrast, the reciprocal hmGluR1 mutant bearing these three residues of hmGluR5 showed high affinity for [3H]M-MPEP. Radioligand binding to these mutants was also inhibited by 7-hydroxyiminocyclopropan[b]chromen-1a-carboxylic acid ethyl ester (CPCCOEt), a structurally unrelated non-competitive mGluR1 antagonist previously shown to interact with residues Thr-815 and Ala-818 in TMVII of hmGluR1. These results indicate that MPEP and CPCCOEt bind to overlapping binding pockets in the TM region of group I mGluRs but interact with different non-conserved residues.

Metabotropic glutamate receptors are G protein-coupled receptors that play important roles in regulating the activity of many synapses in the central nervous system. At present, eight mGluR 1 subtypes (mGluR1 through mGluR8) have been cloned and functionally expressed (1,2). Based on their amino acid sequence homologies, pharmacology, and functional profiles, these subtypes are classified further into three groups. Members of group I (mGluR1 and -5) stimulate the activity of phospholipase C and mobilize intracellular Ca 2ϩ . Members of group II (mGluR2 and -3) and group III (mGluR4, -6, -7, -8) inhibit adenylyl cyclase. Despite the differences in primary structures and functional roles, all mGluRs feature a large conserved N-terminal extracellular domain, which is involved in the recognition of agonists and competitive antagonists (3)(4)(5)(6)(7)(8).
Most ligands for mGluRs were derived from amino acids and act at the conserved glutamate binding site (9). Recently, novel subtype-selective group I mGluR antagonists emerged that are structurally unrelated to amino acids and to each other. The first non-amino acid-like antagonist described was CPCCOEt (Fig. 1), a selective mGluR1 antagonist (10,11). CPCCOEt inhibits receptor activity by a non-competitive mechanism which does not affect the binding affinity of glutamate (12,13). Molecular characterization of the site of inhibition in mGluR1 revealed that CPCCOEt interacts with two non-conserved residues at the top of transmembrane (TM) helix VII (13). The first described selective mGluR5 antagonists, SIB-1757 and SIB-1893 ( Fig. 1), are also unrelated to amino acids and were shown to act via a non-competitive mechanism (14).
To address the question whether these structurally unrelated mGluR1 and mGluR5 antagonists interact with different sites of the mGluR subtypes or share a common binding site in the 7TM domain, we have studied the binding site and mode of action of the selective mGluR5 antagonist MPEP (15). MPEP is a novel derivative of SIB-1893 with nanomolar potency (Fig. 1); it is an effective antihyperalgesic in animal models of chronic inflammatory pain (16), a neuroprotectant in excitotoxin-induced striatal lesions (17) and an anticonvulsant in several epilepsy models (18). We generated a number of chimeric receptors and point mutations in which segments or single residues of hmGluR5 were exchanged with the corresponding residues of hmGluR1 and vice versa. These mutants were tested for inhibition of glutamate-induced calcium signals by MPEP * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

Compounds
MPEP and CPCCOEt were synthesized as described previously (15,10). [ 3 H]M-MPEP was synthesized using 2-methyl-6-(3-hydroxy-phenylethynyl)pyridine as starting material by methylation with tritiated methyliodide. Further details of the synthesis and characteristics of the radioligand will be published elsewhere. 2 Glutamate was obtained from Tocris (Bristol, United Kingdom). Other chemicals were purchased from Sigma (Buchs, Switzerland).

Plasmids
CDNAs encoding wild-type hmGluR1b and hmGluR5a and the chimeric hmGluR1/5a and hmGluR5/1b receptors termed p254, p255, p317, p257, p305, p306, p310, and p322 were described previously (13). Additional chimeras were constructed in pCMV-T7-3 (19) using standard cloning techniques based on unique restriction sites in hmGluR1b and -5a, novel restriction sites introduced by site-directed mutagenesis or the polymerase chain reaction-based overlap extension technique. The authenticity of the chimeric cDNAs (Table I) was confirmed by sequencing of all amplified DNA fragments. Point mutations in TMIII and TMVII of hmGluR1b and hmGluR5a cDNAs were generated using the QuickChange site-directed mutagenesis kit (Stratagene, La Jolla, CA). The authenticity of each point mutation was confirmed by DNA sequencing.

Cell Culture and Transfections
Chinese hamster ovary and L cell lines stably expressing human mGluR1b and -5a were grown in Dulbecco's modified Eagle's medium supplemented with 10% dialyzed fetal calf serum as described previously (13). For radioligand binding experiments, wild-type and mutant cDNAs were transfected into COS1 cells using the DEAE-dextran method (23). For measurements of intracellular [Ca 2ϩ ] i , mutant cDNAs were transiently expressed in HEK293 cells by electroporation using a Gene Pulser apparatus (Bio-Rad). Briefly, 5 g of plasmid DNA were used to transfect 1.5 ϫ 10 6 cells in a total volume of 150 l of electroporation buffer (K 2 HPO 4 , 50 mM; CH 3 COOK, 20 mM; KOH, 20 mM, pH 7.4). After electroporation (250 V, 300 microfarads), cells were resuspended in Dulbecco's modified Eagle's medium (Life Technologies, Inc.) supplemented with 10% fetal calf serum and antibiotics. 5 ϫ 10 5 cells were plated on glass coverslips (9 ϫ 18 mm) coated with 100 mg/ml poly-D-lysine (Sigma).

IP and [Ca 2ϩ ] i Measurements
Cells were seeded in 24-well tissue culture plates and were labeled to equilibrium with myo-[ 3 H]inositol at 2 ϫ 10 Ϫ6 Ci/ml in Dulbecco's modified Eagle's medium for 20 h. Preincubation with LiCl, stimulation with agonist and/or antagonist, and extraction of total inositol phosphates (IP) were performed as described previously (13). For the experiments addressing constitutive receptor activity, the cells were incubated in the presence of a glutamate-degrading enzyme (1 unit/ml glutamate pyruvate transaminase plus 2 mM pyruvate), 1 h before and during the incubation period with LiCl, except when cells were stimulated by added glutamate. Concentration-response curves were obtained by fitting the four-parametric logistic equation to the data using Prism2.0 (GraphPad Software, San Diego, CA).

Ligand Binding
Membranes from transfected COS1 cells were collected 2 days after transfection. Cells were washed with phosphate-buffered saline and mechanically detached in ice-cold phosphate-buffered saline containing 10 mM EDTA. Cells were centrifuged at 4000 rpm for 20 min at 4°C and resuspended in binding buffer (30 mM NaHepes, 110 mM NaCl, 1.2 mM MgCl 2 , 5 mM KCl, 2.5 mM CaCl 2 2H 2 O, pH 8.0). Cells were then disrupted on ice with a Polytron homogenizer for 20 s, and membranes were collected by centrifugation at 18,000 rpm for 20 min at 4°C. The pellet was resuspended in binding buffer, homogenized with a Teflon homogenizer, and used immediately for binding. Ligand binding assays were performed using [ 3 H]M-MPEP and cold M-MPEP (10 Ϫ6 M) to determine nonspecific binding. 2 Briefly, samples consisted of 200 l of membrane suspension (50 g), 25 l of radioligand (2-30 nM), and 25 l of binding buffer. The reaction was terminated after a 30-min incubation at 25°C by dilution and rapid filtration through Whatman GF/B glass fiber filters. The filters were washed three times with cold binding buffer, and the bound radioactivity was counted using a ␤-counter in 5 ml of Ultima Gold NV Packard (Canberra Packard, Zurich, Switzerland). Specific [ 3 H]M-MPEP binding was defined as total binding minus nonspecific binding in the presence of 1 ϫ 10 Ϫ6 M cold M-MPEP. In one experiment all measurements were performed in duplicate. Saturation analysis and competition curves were analyzed using Prism2.0 (Graph-Pad Software).

Molecular Modeling
Construction of a hmGluR5 Transmembrane Domain Model-To suggest a plausible binding mode of MPEP to hmGluR5, a seventransmembrane model was built and optimized using the programs SYBYL (SYBYL 6.4 software; Tripos Inc., St. Louis, MO) and X-PLOR 2 I. Vranesic, personal communication.  (X-PLOR 3.1 software; Molecular Simulations, Inc., San Diego, CA) based on the ␣-carbon template of the transmembrane helices of the rhodopsin receptor family (24,25). In brief, seven individual polyalanine standard ␣ helices of lengths 27, 27, 35, 25, 30, 30, and 24 were built, and each helix was superimposed on the corresponding helix of the ␣-carbon template derived from the rhodopsin family of GPCRs (25). The root-mean-square distances of these superimposed C-␣ atoms of TM helices I-VII were 0.14, 0.79, 0.19, 0.13, 0.84, 1.35, and 0.12, respectively. All 198 alanines were mutated according to the putative assignment of transmembrane segments given in Table II by making the corresponding side chain changes of alanines. All prolines were fixed using the SYBYL Biopolymer FIX_PROLINE command, hydrogen atoms were added with the ADDH command, and Gasteiger partial charges were computed with the CHARGE GAST_HUC command. Owing to the presence of charged residues (arginines (4), lysines (11), aspartic acid (2), and glutamic acids (2)), the net charge was ϩ11e. Groups forming termini of helices where kept neutral as NH 2 and COOH.
This raw model was optimized with X-PLOR using the conjugate gradient method and the Tripos force field (X-PLOR/TAFF). Harmonic restraints with a force constant of 1.0 kcal/mol Å were applied to the initial coordinates of all 198 C-␣ and 188 C-␤ atoms. Minimization was carried out until the gradient grad(E) and the energy E were 0.4 kcal/mol A and Ϫ1814 kcal/mol Å, respectively.
The extracellular domain between TMVI and TMVII is, according to our sequence assignment to helical transmembrane segments, formed by a short sequence of about four residues, which may be part of the binding pocket. A plausible third extracellular loop between helix VI and VII was searched with SYBYL/BIOPOLYMER/LOOP assigning residues Asn-796, Tyr-797, and Met-802 as anchor regions and the four residues Lys-798 to Thr-801 as window region. A loop with a sequence homology of 26% and a root-mean-square fit of 0.22 of backbone atoms in the anchor region was selected for the automatic construction of the loop. After calculation of all partial charges the model was again minimized (X-PLOR/TAFF, grad(E) ϭ 0.4 kcal/mol Å, E ϭ Ϫ1870 kcal/mol) using the same constraints of the previous minimization.
Docking of MPEP-To this initial mGluR5 model we have manually docked MPEP with the pyridine N-atom accepting a H-bond from the hydroxyl group of Ser-658 of helix III and the 2-methyl in hydrophobic contact with the pyrrolidine ring of Pro-655 also in helix III. This model was optimized in 48 independent runs using an X-PLOR molecular dynamics protocol involving heating from 25 to 500 K in 2 ps and cooling again to 25 K in 2 ps, followed by 2000 steps of TAFF/X-PLOR minimization while keeping harmonic restraints to the previous reference coordinates of C-␣ and C-␤ atoms.
Docking of CPCCOEt to the Mutant hmGluR5-M802T,S805A-The initial receptor model of hmGluR5-M802T,S805A was obtained by changing in the initial wild-type hmGluR5 model the side chains of Met-802 and Ser-805 into respective threonine and alanine side chains, recalculation of partial charges, and X-PLOR/TAFF minimization. To this model CPCCOEt was manually docked such that its oxime group formed a contact with the hydroxyl group of Thr-802 and its benzene moiety was pointing down into the transmembrane region between TMIII and TMVII. Optimization of this complex was performed as described above for the complex of MPEP with wild-type hmGluR5.

MPEP Is a Non-competitive Antagonist and Inverse
Agonist-We have shown previously that MPEP potently and selectively reduces the quisqualate-induced IP production in a concentration-dependent manner in L cells stably expressing hmGluR5a with an IC 50 value of 39 nM (15). To analyze the mode of inhibition of MPEP, the concentration dependence of the stimulation of IP production in response to glutamate was compared in the absence and presence of 10, 30, 100, and 300 nM MPEP. With increasing concentrations of MPEP, a profound reduction in the amplitude of IP production was observed as compared with the stimulation evoked by glutamate alone (Fig. 2A). The reduction in the amplitude was not accompanied by a change in the EC 50 value or the Hill coefficient of glutamate, indicating that MPEP is a non-competitive antagonist (Fig. 2, B and C).
We have previously reported that rat mGluR5a displayed detectable constitutive activity upon transient overexpression in either LLCPK1 (21) or HEK293 cells (26). This constitutive activity could not be antagonized by competitive group I mGluR antagonists such as MCPG. Given its non-competitive mode of action, we wondered whether MPEP could inhibit the rat mGluR5a constitutive activity, and thus act as an inverse agonist. To keep the extracellular glutamate concentration in this study as low as possible, HEK293 cells were cotransfected with rat mGluR5a and the glutamate transporter EAAC1. Basal activity was measured in the presence of the glutamate-de-  grading enzyme glutamate pyruvate transaminase in the incubation medium. Under these conditions the basal IP levels were 3.6 Ϯ 0.5-fold (n ϭ 4) higher in cells coexpressing rat mGluR5a and EAAC1 than in cells expressing the transporter alone (Fig. 3A). A further increase in basal IP production was observed by coexpression of mGluR5a with the G protein G␣ q as described previously for rat mGluR1a (8). In the presence of G␣ q , the basal level of IP production in mGluR5a-expressing cells was 5.6 Ϯ 1.6-fold (n ϭ 4) higher than in cells expressing G␣ q alone (Fig. 3B). Although MCPG antagonized the action of glutamate, the basal IP production in the presence of G␣ q was not inhibited by MCPG (Fig. 3B). MPEP, however, when applied on cells expressing rat mGluR5a (Fig. 3A) or rat mGluR5a plus G␣ q (Fig. 3B), led to a significant but not complete inhibition of basal activity. The inverse agonism of MPEP was dose-dependent ( Fig. 3C) with an IC 50 of 13.6 Ϯ 4.9 nM and a Hill coefficient of 1.1 Ϯ 0.2 (n ϭ 4), in agreement with the potency of MPEP on quisqualate-stimulated IP production in stable cell lines expressing hmGluR5a (15). Inhibition by MPEP Is Mediated by the Transmembrane Domain of mGluR5-The non-competitive mode of inhibition of MPEP prompted us to speculate that the compound does not act at the glutamate-binding site located in the large N-terminal extracellular domain. In order to localize the structural determinants mediating this inhibition, we first used two chimeras of hmGluR1 and hmGluR5 (hmGluR1/5 and -5/1) with a fusion site after the first TM segment (Fig. 4). When HEK293 cells were transfected with wild type hmGluR1 and hmGluR5 as well as two chimeras hmGluR1/5 and -5/1 (p305, p254), all transfected cells responded with a transient increase in [Ca 2ϩ ] i upon application of 50 M glutamate (Fig. 4). Coapplication of glutamate and MPEP (1 M) caused a complete inhibition of [Ca 2ϩ ] i responses in cells transfected with wild type hmGluR5 and the chimera hmGluR1/5 (p305). In contrast, the glutamate-stimulated change in [Ca 2ϩ ] i was not affected by MPEP in cells expressing hmGluR1 and hmGluR5/1 (p254). This indicated that the inhibitory effect of MPEP is mediated by the C-terminal part of the receptor including the TM segments II-VII and/or intracellular and extracellular loops of hmGluR5. Reversibility of the inhibition by MPEP was demonstrated by reapplying glutamate after a 10-min washout period.
[ 3 H]M-MPEP Binding Requires Residues in TMIII and TM-VII of mGluR5-To identify the molecular determinants governing MPEP selectivity and inhibition, we constructed a series of chimeric hmGluR1/5 and -5/1 receptors fused at different position in the TM domains (Fig. 5). Each of these mutants was tested in a radioligand binding assay using the novel antagonist [ 3 H]M-MPEP (Fig. 1), a close analog of MPEP with a K D of 3.5 Ϯ 0.7 nM on wild-type hmGluR5a. [ 3 H]M-MPEP was displaced by MPEP with an IC 50 of 16 nM (data not shown). Saturation binding studies with membranes prepared from cells transfected with the hmGluR1/5 chimera p305 yielded a K D value of 4.0 Ϯ 0.5 nM (Fig. 5). Binding affinities comparable to wild type hmGluR5 (K D Ͻ 5 nM) were also obtained for all mutants containing the third and seventh TM segments of hmGluR5 (p257, p305, p310, p316, p323, p324). In contrast, no significant binding was observed in wild type hmGluR1 or chimeric receptors lacking TMIII and/or TMVII of hmGluR5 (p254, p255, p317, p306, p312, p361, p322). All chimeras were shown to be functional as revealed by a transient increase in [Ca 2ϩ ] i upon stimulation with 50 M glutamate. Consistent with the ligand binding data, chimeras containing TMIII and TMVII of hmGluR5 were also functionally inhibited by MPEP (1 M), whereas mutants lacking one or both of these TM segments of hmGluR5 were insensitive to application of MPEP (data not shown).
Sequence alignments of the hmGluR subtypes revealed a high degree of conservation of TMIII and TMVII. In fact, only FIG. 3. MPEP inhibits basal constitutive activity of rat mGluR5a. A, IP production was measured in HEK293 cells (control (Ctrl)) and cells transiently expressing rat mGluR5a. The constitutive activity is evident when comparing columns labeled Ctrl Basal and mGluR5a Basal. MPEP was applied at 1 M and glutamate (Glu) at 3 M. B, same as in A, except that the G protein ␣ q subunit was overexpressed alone (Ctrl-Basal) or together with mGluR5a in order to increase the constitutive activity. MPEP was applied at 1 M, MCPG at 3 mM, and glutamate (Glu) at 3 M. C, concentration-response curve of the inhibitory effect of MPEP on the basal IP production measured in cells expressing both mGluR5a and the G␣ q protein. In all experiments, the glutamate transporter EAAC1 was coexpressed to deplete the extracellular medium of glutamate and the glutamate degrading enzyme glutamate-pyruvate transaminase (plus 2 mM pyruvate) was added to avoid a putative activation of the mGluR5a by residual extracellular glutamate produced by the cells. Results are expressed as the percentage of IP production over the total radioactivity remaining in the membrane fraction of the cells. The data presented in A and B are the means Ϯ S.E. of data from four different experiments performed in triplicate, and data shown in C are a representative example of three separate experiments performed in triplicate. six residues in TMIII and six residues in TMVII differ between hmGluR1 and hmGluR5 (Fig. 6). To precisely identify the residues governing the specific binding of [ 3 H]M-MPEP, we constructed point mutants of hmGluR5, in which the non-conserved amino acids in TMIII and TMVII were substituted by the homologous amino acids of the closely related subtype hmGluR1. The analysis of the TMVII mutants R5-M802T, -S805A, -V819T, -V822M, and -L826I revealed no significant decrease in binding affinity of [ 3 H]M-MPEP when compared with wild-type hmGluR5 (Table III) To exclude the possibility that the mutation of Pro-655, Ser-658, and Ala-810 disrupt the structure of the binding site, these residues were introduced singly or in combination at the corresponding position of hmGluR1. Single substitutions such as R1-V823A or double substitutions such as R1-S668P,C671S revealed no significant affinity for [ 3 H]M-MPEP as observed with the wild type hmGluR1 receptor (Table III) To demonstrate that the mutants with reduced binding affinity or loss of binding affinity were properly expressed and targeted to the membrane, these mutants were tested in the calcium assay. All mutants tested showed transient elevations of [Ca 2ϩ ] i upon application of 50 M glutamate. Furthermore, the mutants that showed a complete loss of binding affinity also failed to show inhibition by MPEP (data not shown).
The mGluR1 Antagonist CPCCOEt and the mGluR5 Antagonist MPEP Bind to Overlapping Binding Pockets of Group I mGluRs-We have shown earlier that the selective mGluR1 antagonist CPCCOEt specifically interacts with the residues Thr-815 and Ala-818 in TMVII of hmGluR1 (13). Here we report that the selective mGluR5 antagonist MPEP requires the non-conserved amino acid residues Pro-655 and Ser-658 in TMIII and Ala-810 in TMVII. We wondered whether these antagonists bind to different or mutually exclusive binding sites in the transmembrane region of group I mGluRs. To address this question, CPCCOEt was tested for inhibition of radioligand binding in the hmGluR5 mutant R5-M802T,S805A (p330) and the hmGluR1 mutant R1-V664I,S668P, C671S,V823A (p381), which showed high affinity binding for [ 3 H]M-MPEP (Table III) as well as functional inhibition by CPCCOEt (for p330, see Ref. 13; for p381, data not shown).
In parallel, we performed molecular modeling of the TM region of wild-type hmGluR5 and the mutant R5-M802T,S805A and manually docked MPEP and CPCCOEt to the key residues identified by site-directed mutagenesis (Fig. 8). After a series of 48 minimizations using a molecular dynamic protocol (for details, see "Materials and Methods"), the lowest energy structures were taken as a model for MPEP binding to hmGluR5 and CPCCOEt binding to the mutant R5-M802T,S805A, respectively. The MPEP/hmGluR5 model supports the experimental findings by predicting close interaction of MPEP with the side chains of the amino acid residues Val-806, Ser-809, and Ala-810 in TMVII and Pro-655, Ser-658, and Tyr-659 in TMIII. Further favorable contacts are suggested with Lys-638 in TMII. The CPCCOEt/hmGluR5-mutant model predicts favorable contacts to the side chains of Thr-802, Ala-805, and Ser-809 in TMVII and Lys-798 in the third extracellular loop. Further contacts are suggested with the side chains of Pro-655, Ser-658, and Tyr-659 in TMIII. A comparison of both models suggests that the pyridine ring of MPEP occupies precisely the same space between TMVII and TMIII than the benzene ring of CPCCOEt.

DISCUSSION
The major findings of this study concern the involvement of the hmGluR5 7TM domain in the high affinity binding of MPEP, which mediates non-competitive inhibition. This suggests that MPEP inhibits receptor activity without changing the affinity of glutamate to its binding site, and thus MPEP might act at a receptor site different from the glutamate binding domain. In agreement with this hypothesis is the inability of MPEP to displace [ 3 H]quisqualate binding to rat mGluR5a 3 and the lack of effect of mGluR5 agonists on [ 3 H]M-MPEP binding. 2 In addition to its non-competitive antagonist action, MPEP also decreased basal IP production in the absence of an agonist 3 V. Mutel, personal communication. in cells transiently overexpressing rat mGluR5a indicating an inverse agonist activity. Indeed, increased basal receptor activity was still detected in the presence of the glutamate transporter EAAC1 and the glutamate degrading enzyme glutamate pyruvate transaminase, conditions expected to drastically decrease the extracellular glutamate concentration to levels below that required for activation of mGluR5. As previously reported for rat mGluR1a (8), co-expression of rat mGluR5a with the G protein G␣ q further increased basal IP production. The competitive antagonist MCPG did not block any basal IP formation at a concentration of 3 mM but almost fully inhibited the action of glutamate. This shows that the basal activity does not result from the activation of the receptor by an endogenous agonist, and therefore likely results from the constitutive activity of the receptor. In contrast, when MPEP was applied alone on cells expressing mGluR5a or mGluR5a plus G␣ q , a significant but not complete inhibition of the basal activity was detected. This inverse agonism of MPEP, so far not described for other antagonists of family 3 (mGluR-like) GPCRs, was dose-dependent and in agreement with the potency of MPEP in inhibiting the effect of glutamate.
Studies of a large number of antagonists of family 1 (rhodopsin-like) GPCRs led to the hypothesis that these receptors oscillate between (at least) two conformational states, an inactive and an active one. Agonists stabilize the receptor in the active state. Antagonists, divided in two categories, are either neutral antagonists, which have the same affinity to both the inactive and active conformational state, or inverse agonists, which stabilize the inactive state and therefore inhibit the constitutive activity. In contrast to family 1 GPCRs, mGluRs and other family 3 GPCRs consist of two distinct domains, the large extracellular agonist binding domain and the 7TM region, which directly interacts with the G proteins. As none of the known competitive mGluR1 antagonists interacting at the glutamate binding site have been shown to inhibit constitutive activity of mGluRs (27), we speculate that the constitutive activity of mGluRs might result from an equilibrium between two conformational states of the 7TM region rather than from an equilibrium between an active and inactive state of the agonist binding domain. According to this hypothesis, one is expecting that non-competitive antagonists acting on the 7TM region of mGluRs are more likely to act as inverse agonist than competitive antagonists acting at the extracellular glutamate binding site.
A detailed molecular investigation using chimeric receptors and point mutants of hmGluR5 and hmGluR1 revealed that MPEP binds to and interacts with Ala-810 in TMVII and Ile-651, Pro-655, and Ser-658 in TMIII, respectively. Replacement It has not yet been determined whether the structurally unrelated non-competitive antagonists MPEP and CPCCOEt interact with different sites of mGluR subtypes or share a common binding site in the TM domain. We showed in a previous study that two residues unique to hmGluR1, Thr-815 and Ala-818 on the extracellular surface of TMVII, were responsible for the selective action of the non-competitive mGluR1 antagonist CPCCOEt (13). The present study shows that binding of the mGluR5 antagonist MPEP requires an interaction with Ala-810 further down in the transmembrane helix VII as well as additional interactions with Pro-655 and Ser-658 in TMIII. Using hmGluR1 and -5 mutants (R1-V664I,S668P, C671S,V823A and R5-M802T,S805A), which show functional inhibition by MPEP and CPCCOEt, we unequivocally demonstrated complete inhibition of [ 3 H]M-MPEP binding by CPC-COEt in a concentration dependent manner. This is further supported by docking studies of MPEP and CPCCOEt to 7TM domain models. These models suggest that the pyridine ring of MPEP precisely occupies the same space between TMVII and TMIII as the benzene ring of CPCCOEt, providing a molecular explanation for the displacement of MPEP by CPCCOEt. However, other parts of these antagonists do not overlap and suggest interactions with different TM helices. Thus, although MPEP and CPCCOEt are structurally unrelated, they recognize overlapping binding pockets in the 7TM region of group I mGluRs that are sufficiently diverse to allow subtype-specific interaction with different classes of compounds.
These findings may have important implications for the design of novel mGluR antagonists. Up to now, most compounds acting at the eight subtypes of mGluRs are phenylglycines or rigidified amino acid analogs such as LY354740 (9). These compounds possess a wide spectrum of agonist, partial agonist, and antagonist activities and interact at the conserved glutamate binding site located in the large N-terminal extracellular domain. Although some of these compounds have reached nanomolar potencies and are able to discriminate among groups of mGluRs, as yet there are no known agonists or competitive antagonists that are able to sufficiently discriminate individual mGluR subtypes most likely due to structural constraints of the glutamate binding site. In contrast, the overlapping binding sites for the non-competitive antagonists CPC-COEt and MPEP in the 7TM region seem to be less conserved, and thus to tolerate binding of structurally diverse compounds. Therefore, it can be speculated that medicinal chemistry efforts toward non-competitive antagonists acting in the 7TM domain are more likely to generate subtype-selective compounds than efforts on competitive antagonists acting at the extracellular glutamate binding site.