Transposition of Three Amino Acids Transforms the Human Metabotropic Glutamate Receptor (mGluR)-3-Positive Allosteric Modulation Site to mGluR2, and Additional Characterization of the mGluR2-Positive Allosteric Modulation Site

Glutamate is the primary excitatory neurotransmitter in the mammalian central nervous system, and its activity is mediated by two classes of receptors: ionotropic glutamate receptors, which are multimeric ion channels responsible for fast synaptic transmission, and metabotropic glutamate receptors (mGluRs), which are seven-transmembrane receptors that couple to G proteins to modulate slower synaptic transmission through intracellular second messengers. Because ionotropic glutamate receptors are expressed by nearly all types of neurons and mediate fast excitatory neurotransmission throughout the brain, direct pharmacological manipulation of this group of receptors has been difficult because inhibition could produce widespread disruption of brain function, which could lead to negative side effects. Conversely, mGluRs that activate intracellular signaling cascades offer an opportunity for developing drugs that regulate glutamate neurotransmission in a functionally selective manner.

mGluRs belong to the G protein-coupled receptor (GPCR) superfamily and have eight known subtypes (mGluR1–8). These receptors divide into three groups based upon sequence homology, intracellular messengers, and pharmacology. Group I receptors (mGluR1 and 5) activate phospholipase C via G_q proteins and initiate an inositol triphosphate/diacylglycerol second messenger cascade. Group II (mGluR2 and 3) and group III (mGluR4, 6, 7, and 8) receptors inhibit adenylate cyclase via G_i/o proteins (Schoepp, 1994; Conn and Pin, 1997; Schoepp, 2001).

mGluR2 and mGluR3 are located primarily presynaptically in the hippocampus, cortex, striatum, thalamus, and amygdala (Ohishi et al., 1993). Activation of these receptors has been shown to decrease synaptic transmission and glutamate release in the hippocampus (Macek et al., 1996). The distribution and prominent functions of group II mGluRs in neuronal excitability and synaptic transmission suggest that modulation of these GPCRs is a promising strategy for the treatment of neurological and neuropsychiatric disorders, such as anxiety, schizophrenia, and pain (Conn and Pin, 1997; Chavez-Noriega et al., 2002; Varney and Gereau, 2002; Lorrain et al., 2003). Furthermore, mGluR-mediated glutamate activity has been shown to influence calcium and potassium ion channels, modulate N-methyl-d-aspartate and α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptor currents, and GABA release (Lea and Faden, 2003). The ability of mGluRs to modulate neuroprotective and neurotoxic systems suggests that some mGluRs play a role in neuronal injury following trauma. Indeed, studies have shown group II and III mGluRs to have a neuroprotective role (for review, see Lea and Faden, 2003).

Several lines of evidence suggest that positive allosteric modulators of the mGluRs may offer advantages over classical GPCR agonists by overcoming receptor desensitization that often occurs after repeated dosing of agonists (Pin et al., 2001; May and Christopoulos, 2003). Furthermore, in contrast to full agonists, potentiators can be designed to be devoid of any agonist activity or act as weak partial agonists. However, when agonist is present, potentiators can significantly increase potency and/or efficacy of the agonist. Development of potentiators could also be useful in disease states in which the level of the endogenous ligand is attenuated. Patients administered a potentiator could achieve normal agonist responses. Due to the high level of conservation of the agonist binding site of mGluRs, developing subtype-specific agonists has proven to be difficult. To date, no agonist has been reported to discriminate between mGluR2 and mGluR3. However, as described in this article and previously, several mGluR2-specific positive allosteric modulators have been identified (Johnson et al., 2003; Schaffhauser et al., 2003; Hu et al., 2004; Pinkerton et al., 2004a,b, 2005; Bonnefous et al., 2005).

The present study had three main goals. First, we wanted to provide additional pharmacological characterizations of a series of novel and structurally distinct mGluR2 potentiators. We tested two indanones (Pinkerton et al., 2005), a biphenyl-indanone (Bonnefous et al., 2005) and a pyrimidine methyl anilines (Hu et al., 2004). Second, we wanted to determine whether this diverse set of mGluR2 allosteric modulators also binds to the known mGluR2 potentiator site defined by amino acids Ser⁶⁸⁸, Gly⁶⁸⁹, and Asn⁷³⁵ (Schaffhauser et al., 2003) or to determine whether another novel potentiator site exists. Third, we wanted to develop a high-throughput transfection system and dual scintillation proximity assay (SPA) using hmGluR2-wild type (WT) and the hmGluR2_Leu⁶⁸⁸-, Val⁶⁸⁹-, and Asp⁷³⁵-expressing cells. With these assays, we could easily identify hmGluR2-specific potentiators (i.e., positive only in the hmGluR2-WT assay) versus allosteric modulators of both hmGluR2 and hmGluR3 (i.e., active in both hmGluR2-WT and the hmGluR2_Leu⁶⁸⁸, Val⁶⁸⁹, and Asp⁷³⁵ assays).

Materials and Methods

Materials. Glutamate, GDP, guanosine 5′[γ-thio]-triphosphate (GTPγS), glutamate pyruvate transaminase (GPT), coenzyme pyridoxal phosphate, and pyruvate were obtained from Sigma-Aldrich (St. Louis, MO). [³⁵S]GTPγS (1250 Ci/mmol), Unifilter GF/B plates, and MicroScint 20 were purchased from PerkinElmer Life and Analytical Sciences (Boston, MA). (2S,2′R,3′R)-2-(2′,3′)-dicarboxycyclopropyl)glycine (DCG-IV) was purchased from Tocris Cookson Inc. (Ellisville, MO). Male Sprague-Dawley rats (250–300 g) were purchased from Harlan (Indianapolis, IN). The pcDNA3.0, pcDNA3.1-based mammalian expression vectors were purchased from Invitrogen (Carlsbad, CA). A QuikChange site-directed mutagenesis kit was obtained from Stratagene (La Jolla, CA), and an Effectene kit was purchased from QIAGEN (Valencia, CA). BioCoat Fibronectin 24-well tissue culture plates were obtained from BD Biosciences (Bedford, MA). Dowex-1-X8 (200–300 mesh in the formate form) was obtained from Bio-Rad (Hercules, CA). Myo-[2-³H]inositol (specific activity 18 Ci/mmol) and RNA binding YSi SPA beads were purchased from GE Healthcare (Piscataway, NJ). Protease inhibitor cocktail was obtained from Roche Diagnostics (Indianapolis, IN). Picoplate 96-well TopCount plates and TopSeal were purchased from PerkinElmer Life and Analytical Sciences. The compounds MRLSD-650, MRLSD-230, MRLSD-041, and MRLSD-973 (Fig. 1, A–D) have described previously (Hu et al., 2004; Bonnefous et al., 2005; Pinkerton et al., 2005), and MRLSD-064, MRLSD-726, MRLSD-438, and MRLSD-667 (Fig. 1, E–H) are novel (Cube et al., 2004; Pinkerton et al., 2006a,b; Govek et al., 2006). All MRLSD-compounds were synthesized at Merck Research Laboratories (San Diego, CA).

Membrane Preparations. Membranes were isolated from hmGluR2- and hmGluR3-expressing stable cell lines as described previously (Schaffhauser et al., 2003). In brief, confluent cells were washed into phosphate-buffered saline and harvested by centrifugation (200g). The resulting cell pellet was 1) homogenized in buffer A (20 mM HEPES and 10 mM EDTA, pH 7.4) and centrifuged at 40,000g (20 min at 4°C), 2) the resulting pellet was washed one time in buffer A and then one time in buffer B (20 mM HEPES and 0.1 mM EDTA, pH 7.4), and 3) resuspended in buffer B. Membranes were stored at –80°C; protein concentration was ∼1 mg/ml.

Whole-rat brain membranes were isolated as described previously (Schaffhauser et al., 2003). Briefly, rats were decapitated, and whole brains were homogenized in 6 volumes (w/v) of 10% sucrose at 4°C using a glass-Teflon homogenizer. The homogenate was centrifuged (1000g for 10 min), and the supernatant was removed and centrifuged again (40,000g for 20 min at 4°C). The pellet was resuspended in 25 ml of water and centrifuged (8000g for 10 min). The supernatant and the buffy coat layer were removed and centrifuged (40,000g for 20 min at 4°C). The pellet was resuspended in buffer C (5 mM HEPES-KOH, pH 7.4), freeze-thawed twice, and then centrifuged (40,000g for 20 min). The resulting pellet was resuspended in buffer C and stored at –80°C at a protein concentration of ∼3 mg/ml. Protein measurements were determined with the Bio-Rad DC kit.

[³⁵S]GTPγS Binding Assay. This assay was performed as described in Schaffhauser et al. (2003). Briefly, mGluR2, mGluR3, or rat brain membranes were thawed, further processed, and then resuspended in assay buffer (20 mM HEPES, pH 7.4, 100 mM NaCl, and 3 mM MgCl₂) to a final protein concentration of 0.5 mg/ml (hmGluR2- and hmGluR3-expressing membranes) or 0.1 mg/ml (rat brain). In a 96-well plate (Beckman Coulter, Fullerton, CA), test compounds were added (300 pM–10 μM) along with 5 μM GDP, membrane (10 μg/well rat brain and 50 μg for recombinant cells), and 0.05 nM [³⁵S]GTPγS (total volume 0.5 ml). Plates were incubated at 30°C for 1 h, and then the assay was terminated by rapid filtration (Unifilter GF/B plate with 96-well cell harvester; Brandel Inc., Gaithersburg, MD). Plates were then rinsed, dried, and MicroScint 20 was added to each well, followed by counting in a TopCount scintillation counter (PerkinElmer Life and Analytical Sciences).

Cloning of Wild-Type hmGluR2 and hmGluR3 cDNAs. Cloning of WT hmGluR2 cDNA (Flor et al., 1995; Schaffhauser et al., 2003) and WT hmGluR3 cDNA was performed as described previously (Varney et al., 1998; Schaffhauser et al., 2003) and then cloned into pcDNA 3.0-based mammalian expression vectors.

Construction of Chimeric and Point Mutations of hmGluR2/3 and hmGluR3/2 cDNAs. The mGluR2 and mGluR3 chimeric receptor cDNAs were generated by site-directed mutagenesis from WT cDNAs to introduce single or multiple mutations as described previously (Schaffhauser et al., 2003). All of the single and multiple point mutations (PMs) of hmGluR2/3 receptor cDNAs were synthesized using the QuikChange site-directed mutagenesis kit and cloned into the pcDNA3.1-based mammalian expression vectors.

Transient Transfection of Chimeric hmGluR2/3 cDNAs into HEK293 Cells. All generated cDNAs constructs, including WT, were cotransfected with Gα₁₆ cDNA (kindly provided by Aurora Pharmaceuticals, currently Pfizer, San Diego, CA) into HEK293 cells by a modified reverse transfection protocol using a lipid-based transfection reagent (Ziauddin and Sabatini, 2001; Schaffhauser et al., 2003). Briefly, hmGluR2 or hmGluR3, Gα₁₆, and enhanced green fluorescent protein cDNAs were mixed with DNA condensation buffer plus sucrose (0.3 M). Enhancer solution was added to the mixture, mixed, and then incubated at room temperature for 5 min. Effectene reagent was then added to the mixture, mixed, and then incubated for 10 min at room temperature. Finally, 0.25% glycogen was added and mixed, and then the resulting DNA-lipid-polymer transfection reagent was evenly added to the bottom of a fibronectin-coated tissue culture plate. The plates were then allowed to incubate overnight at 4°C. The resulting plates were dried under vacuum and seeded with HEK293 cells (0.7 × 10⁶ cells/well) and incubated in a 37°C, CO₂ incubator for 36 h.

Measurement of Phosphoinositide Hydrolysis. [³H]Inositol monophosphate ([³H]IP₁) assays were performed as described previously (Schaffhauser et al., 2003). Briefly, transfected HEK293 cells were labeled overnight with 1 μCi/well myo-[2-³H]inositol in glutamine-free Dulbecco's modified Eagle's medium. Subsequently, the cells were washed with HBS buffer containing 125 mM NaCl, 5 mM KCl, 0.62 mM MgSO₄, 1.8 mM CaCl₂, 6 mM glucose, and 20 mM HEPES, pH 7.4, for 45 min at 37°C. Following the washes, cells were incubated with HBS buffer containing 10 mM LiCl for 20 min. After incubation, 5 μM glutamate alone (EC₁₀), 1 mM glutamate (EC₁₀₀), or 5 μM glutamate (EC₁₀) in combination with varying concentrations of test compounds was added and incubated for 1 h at 37°C. The reactions were terminated, and accumulated [³H]IP₁ was extracted by adding 4°C chloroform/methanol/HCl (4 N). The mixtures were transferred to new tubes and then extracted with chloroform and H₂O. The aqueous phase was separated from the organic phase by settling for 15 min. The [³H]IP₁ fraction was isolated using Dowex-1-X8 and quantified by liquid scintillation counting. Data were normalized to the response seen with 1 mM glutamate. For hmGluR3 chimera transfections, residual glutamate was removed by preincubating and running the assay in HBS and HBS/LiCl buffers containing 3.0 mg of GPT, 3.0 mg of pyridoxal phosphate, and 15.0 mg of pyruvate per 50 ml. After 20-min incubation in the LiCl-containing buffer, 10 nM DCG-IV alone (EC₁₀), 3 μM DCG-IV (EC₁₀₀), or 10 nM DCG-IV (EC₁₀) in combination with varying concentrations of test compounds was added and incubated for 1 h at 37°C. The reactions were terminated, and the assay was completed as described above.

SPA for IP₁ in Transiently Transfected mGluR2 Cells. The assay was performed as reported previously (Brandish et al., 2003), with modifications. HEK293 transfected cells were labeled overnight with 2 μCi/well myo-[2-³H]inositol in glutamate-free Dulbecco's modified Eagle's medium in a 96-well plate. The following day (18 h after myo-[2-³H]inositol addition), the medium was removed, and the cells were washed two times for 40 min in HBS buffer containing 125 mM NaCl, 5 mM KCl, 0.62 mM MgSO₄, 1.8 mM CaCl₂, 6 mM glucose, and 20 mM HEPES, pH 7.4, for 45 min at 37°C. Following the washes, cells were incubated with HBS buffer containing 10 mM LiCl (20 min). After incubation, 5 μM glutamate alone (EC₁₀), 1 mM glutamate (EC₁₀₀), or 5 μM glutamate (EC₁₀) in combination with varying concentrations of test compounds was added, and cells were incubated for an additional hour at 37°C. For hmGluR3 chimeras, we used DCG-IV at 3 μM for the 100% response and 30 nM for ≈10% response. The wells were aspirated to terminate the incubation, and 50 μl of 10 mM formic acid was added to each well and allowed to incubate at room temperature for 20 min. The plate was placed on shaker table for 5 min. In a 96-well Picoplate, 80 μl/well SPA beads was added to 40 μl of assay lysate, which was mixed with the beads by pipetting. The plates were sealed with TopSeal and placed on a shaker table for 1 h. Plates were counted in a TopCount-NXT (PerkinElmer Life and Analytical Sciences) after 2-h delay.

Data and Statistical Analysis. All experiments were performed in triplicate and repeated in three separate assays. Data were normalized to the response obtained with 1 mM glutamate or 3 μM DCG-IV. The obtained curves were fitted to a four-parameter logistic equation giving EC₅₀ values, Hill coefficient, and maximal effect using Prism (GraphPad Software Inc., San Diego, CA).

The data were analyzed via one-way analysis of variance as appropriate followed by post hoc comparisons. P values < 0.05 for all tests and comparisons were deemed significant unless otherwise indicated. A logarithmic scale was used to satisfy assumptions of equal variance and normal distribution.

Results

Pharmacological Characterization of MRLSD-650 on Rat and Recombinant Human Group II mGluRs. Stimulation of [³⁵S]GTPγS binding in native and recombinant hmGluR2 membrane preparations was performed to provide a functional measure of Gα_i-coupled receptors (Lazareno and Birdsall, 1993). Dose-response analysis of MRLSD-650 (Fig. 1A) alone determined no stimulation of [³⁵S]GTPγS binding in membranes prepared from rat brain or from cells expressing hmGluR2 (Fig. 2, A and B). Conversely, when increasing concentrations of MRLSD-650 were combined with an EC₁₀ concentration of glutamate (1 μM), there was a significant increase in the magnitude of the glutamate-induced [³⁵S]GTPγS binding. A similar level of potentiation was seen in both the rat brain and hmGluR2 membrane preparations when normalized to the response of 1 mM glutamate, and the potentiation occurred in a concentration-dependent manner. Table 1 summarizes the EC₅₀ values and E_max values for MRLSD-650 in both rat brain and hmGluR2 membranes. MRLSD-650 increased the potency of glutamate ∼2-fold in hmGluR2 membranes from glutamate alone: from an EC₅₀ value of 11.61 μM (11.09–12.16 μM) to an EC₅₀ value of 5.15 μM (4.90–5.42 μM) for glutamate + MRLSD-650 (Fig. 2C).

We also investigated the specificity of MRLSD-650 in other mGluR-expressing cells and determined that MRLSD-650 has no detectable glutamate-induced agonist, antagonist, or potentiator activity in cell lines expressing the hmGluR1, hmGluR3, hmGluR4, hmGluR5a, or hmGluR7a receptor (data not shown). Hence, we determined that MRLSD-650 is a selective mGluR2-positive modulator that increases both the potency and efficacy of the agonist ligand glutamate.

Positive Allosteric Modulation by MRLSD-650 Is Mediated by Three Amino Acid Residues in the Transmembrane Domains of hmGluR2. To determine whether the location of the binding site of MRLSD-650 on hmGluR2 was the same three residues determined previously to bind the mGluR2 potentiator LY487379 (Schaffhauser et al., 2003), we investigated the effect of MRLSD-650 on the mGluR2/3 chimeric cDNAs reported previously. The various mutant receptors were prepared by exchanging either segments, single amino acid residues, or multiple amino acids residues between hmGluR3 and hmGluR2 (Fig. 3). Transient transfection of HEK293 cells with WT hmGluR2 in conjunction with Gα₁₆ resulted in glutamate-induced production of [³H]IP₁ in a concentration-dependent manner, with a geometric mean of EC₅₀ = 40 μM (range 30–53 μM). The addition of MRLSD-650 alone (10 μM) increased production of [³H]IP₁ 4-fold over the response observed with 5 μM glutamate (Fig. 3A, black bar). This “basal” increase in [³H]IP₁ production seen in the presence of no exogenously added glutamate most likely results from endogenous glutamate being released from the cultured cells due to cell rupture and death. We attempted to remove this endogenous glutamate with multiple media changes before each assay, but it is very difficult to remove from the media of living cells. However, when 10 μM MRLSD-650 was added in the presence of an EC₁₀ concentration of glutamate (i.e., 5 μM glutamate), we saw a significant increase in the magnitude of the glutamate-induced phosphoinositide hydrolysis (Fig. 3A, striped bar).

To localize the site of the potentiator interaction, we transfected HEK293 cells with three different chimeric constructs and determined that the exchange of the transmembrane regions TMI-III (Thr⁵⁵⁰ to Leu⁶⁵⁶) or a single amino acid residue in TMVII (Ser⁸¹⁷) of hmGluR2 with hmGluR3 sequences did not significantly effect the potentiation of MRLSD-650 on glutamate-induced [³H]IP₁ accumulation (Fig. 3, B and D, respectively). In contrast, the exchange of the mGluR2 amino acid residues present in TMIII-V (Leu⁶⁵⁶ to Arg⁷⁵⁰) with homologous hmGluR3 sequences resulted in the complete loss of the potentiator activity of MRLSD-650 (Fig. 3C), and also the loss of the basal activity. Previously, we had identified this same domain as the binding region of the mGluR2 potentiator LY487379. The magnitude of glutamate-stimulated [³H]IP₁ was similar to WT mGluR2, suggesting that the loss of effect of MRLSD-650 was mediated by alteration of the amino residues in the allosteric binding site (present between Leu⁶⁵⁶ and Arg⁷⁵⁰), and not by alteration of the glutamate binding site. Sequence alignments of hmGluR2 and hmGluR3 in this TMIII-V segment revealed multiple amino acid differences in residues contained within these transmembrane domains and extracellular loop (Fig. 4).

To identify the precise amino acid residues involved in the positive allosteric modulation of the hmGluR2 by MRLSD-650, we used various multiple or single amino acid residues hmGluR2/3 chimeric constructs described previously in Schaffhauser et al. (2003) (Table 2). We determined by Dunnett's post hoc analysis that all of the chimeric constructs responded to glutamate (P < 0.05) and only S688L/G689V (PM2) gave a glutamate stimulation greater than wild type (P < 0.05). No other significant differences were seen between the chimeric constructs and WT hmGluR2 in the limited agonist pharmacology profiling performed. TMIV substitutions of either the single residue A681F (PM1) or double residues V695S/A696V (PM5) had no significant effect on the activity of MRLSD-650 (Table 2). A significant reduction of the potentiation by MRLSD-650 of glutamate-induced [³H]IP₁ accumulation was observed with the hmGluR2 S688L/G689V (PM2) receptor mutant (Fig. 5A). Interestingly the single point mutations S688L (PM3) or G689V (PM4) had no significant effect individually, suggesting that these residues need to be present together to interact with MRLSD-650 and potentiate hmGluR2 in the presence of agonist (Fig. 5A). Further analysis of the amino acid residues that varied between mGluR2 and mGluR3 in the extracellular segments of TMIV-V TMV confirmed that the single mutation H723V (PM8), G730I (PM9), or A740I (PM10) or double mutations A710L/P711A (PM6) or V716T/T718I (PM7) had no significant reduction in the potentiator effect of MRLSD-650 on glutamate-induced [³H]IP₁ production. In contrast, we saw a significant reduction in the positive modulation by MRLSD-650 in the double mutant A733T/N735D (PM11) (Fig. 5B; Table 2). Interestingly, when the A733T and N735D mutations were assayed separately, only the single point mutation N735D (PM13) significantly reduced the effect of MRLSD-650, suggesting that Ala⁷³³ (PM12) is not involved in the potentiation of this compound (Fig. 5B; Table 2). The combined double mutants S688L/N735D (PM14), G689V/N735D (PM15) or the triple mutant S688L/G689V/N735D (PM16) all completely disrupted the activity of MRLSD-650 (Fig. 5C; Table 2). We determined by Western blot analysis that the lack of potentiation by each of these chimeric constructs was not due to a different level of protein expression or sensitivity to glutamate since all constructs expressed protein and responded to glutamate very similarly to WT mGluR2 (Schaffhauser et al., 2003).

Clearly, the S688L, G689V, and N735D (PM16) substitutions rendered MRLSD-650 inactive (Figs. 5C); interestingly, these are the same residues described previously for the interaction site of LY487379 with mGluR2 (Schaffhauser et al., 2003). However, this analysis does not definitively prove that these are the only three residues involved in the activity of these compounds. Therefore, to prove that these three residues were involved in the potentiation effect, we made one additional chimeric construct in which these amino acids were exchanged in hmGluR3 to the hmGluR2 residues (i.e., L688S/V689G/D735N). HEK293 cells were then transiently transfected with this chimeric construct or the wild-type hmGluR3 receptor, along with Gα₁₆. Glutamate induced the production of [³H]IP₁ in a very similar concentration-dependent manner in the wild-type receptor and the L688S, V689G, and D735N mutant. In WT hmGluR3, the EC₅₀ value for glutamate was 360 nM (227–571 nM), whereas the EC₅₀ value for hmGluR3 (L688S/V689G/D735N) was 378 nM (273–525 nM) (Fig. 6A). During these experiments, we found that hmGluR3 transiently transfected cells were very sensitive to glutamate, and any residual glutamate from culturing of the cells was enough to stimulate the cells treated with 10 μM MRLSD-650 to full potentiation. To circumvent this issue, we needed to add GPT, pyridoxal phosphate, and pyruvate to all wash and assay buffers to remove this residual glutamate. Under these assay conditions, we also used DCG-IV as the agonist (10 nM for EC₁₀ and 3 μM for EC₁₀₀). As anticipated, when cells transiently transfected with the WT hmGluR3 were exposed to MRLSD-650 + 10 nM DCG-IV, they failed to induce the production of [³H]IP₁ over 10 nM DCG-IV alone (Fig. 6B). In contrast, in cells transiently transfected with hmGluR3 L688S/V689G/D735N mutations, production of [³H]IP₁ was ≥100% of control (Fig. 6C), suggesting that the hmGluR3 L688S, V689G, and D735N mutations created an active mGluR2-like allosteric modulation binding site on the hmGluR3 receptor.

To further evaluate the hmGluR2 potentiator binding pocket, we developed a high-throughput screening (HTS) 96-well SPA. In this HTS assay, we can rapidly measure the production of IP₁ in HEK293 cells transiently transfected with either WT mGluR2 or S688L/G689V/N735D chimeric constructs. This assay allowed us to screen four times as many compounds, at more concentrations and in less time than with the traditional [³H]IP₁ assay using columns. From these studies, we determined that both of these constructs induced [³H]IP₁ production in response to glutamate in a very similar, concentration-dependent manner (Fig. 7A). WT hmGluR2 induced a glutamate EC₅₀ of 57.3 μM (50.5–65.1 μM), whereas the hmGluR2 S688L/G689V/N735D chimera construct induced a glutamate EC₅₀ value of 72.2 μM (65.9–79.1 μM). As seen in Fig. 7, B and C, MRLSD-650 and three additional selective hmGluR2 potentiators—MRLSD-230 (Fig. 1B), MRLSD-041 (Fig. 1C), and MRLSD-973 (Fig. 1D) (Hu et al., 2004; Bonnefous et al., 2005; Pinkerton et al., 2005)—were assayed in this HTS assay using transiently transfected WT hmGluR2 or hmGluR2 S688L/G689V/N735D triple mutant cells (Table 3). We were able to rapidly determine EC₅₀ values that were comparable with those determined in the [³H]IP₁ assay and also confirm that these four compounds were not active when assayed with cells expressing the triple mutant chimera. This assay was faster, required much less DNA for transfections, and lent itself to 96-well automation systems not available in our other 12-well IP₁ assay. Hence, this HTS assay allows for rapid screening of potentiators in a much more cost-effective and timely paradigm.

Transmembrane Domain of hmGluR3 Also Mediates Positive Allosteric Modulation of Select Compounds. Interestingly, we also identified other structurally distinct compounds—MRLSD-064 (Fig. 1E), MRLSD-726 (Fig. 1F), MRLSD-438 (Fig. 1G), and MRLSD-667 (Fig. 1H)—that had some hmGluR3 [³⁵S]GTPγS binding activity in cells transiently transfected with the hmGluR2 S688L/G689V/N735D chimera. These compounds were then assayed using the recombinant hmGluR3 membranes and found to be active with there as well (Fig. 8; Table 4). Hence, we have determined that these compounds are both hmGluR2 potentiators (46–197 μM) and weak hmGluR3-positive allosteric modulators (438–950 μM).

Discussion

To date, various studies have described group II mGluR receptor-selective agonists, including the heterobicyclic amino acids LY354740, LY379268, and MGS0028, which are believed to act at the glutamate binding site and activate both mGluR2 and mGluR3 (Monn et al., 1999; Pin et al., 1999; Kunishima et al., 2000; Nakazato et al., 2000). In the current study, we characterized a novel mGluR2-specific, positive allosteric modulator and identified that the site of action of this compound is separate from the agonist binding site on the receptor. The presence of another modulatory site on mGluR2 and mGluR3 proteins allows for the development of compounds that selectively potentiate mGluR2 but not mGluR3 or vice versa. MRLSD-650 is a novel, low molecular-weight (mol. wt. = 500.4) compound that acts as a selective mGluR2-positive allosteric modulator in both native (rat brain membranes) and recombinant (stably or transiently transfected) cells expressing human mGluR2.

Group II mGluR receptors have been reported previously to stimulate [³⁵S]GTPγS binding in recombinant membrane preparations (Kowal et al., 1998). In the current study, MRLSD-650 alone did not enhance [³⁵S]GTPγS binding in membranes prepared from rat brain or in membranes prepared from cells stably expressing human mGluR2. However, MRLSD-650 produced very robust, positive modulation of glutamate-induced responses in both recombinant human mGluR2 and native rat mGluR2 membranes. These results are similar to the results reported for the positive allosteric mGluR1 modulator Ro 67-7476 (Knoflach et al., 2001) and the GABA_B(1b/2A) receptor modulator CGP7930 (Urwyler et al., 2001). The potentiation observed was highly selective for mGluR2, since no modulatory activity was seen in cells stably expressing human mGluR1, mGluR3, mGluR4, mGluR5, or mGluR7.

The addition of 1 μM MRLSD-650 increased the potency of glutamate by 2-fold, and MRLSD-650 also increased the maximal response to glutamate. Previous radioligand binding studies (Schaffhauser et al., 2003) support the idea that this change in potency might reflect a change in the affinity of the receptor for the agonist. Mechanistically, MRLSD-650 might produce this effect by either 1) increasing the affinity of the mGluR2 for the agonist, 2) increasing coupling potency and or coupling efficiency by enhancing the ability of the activated receptor to couple more efficiently to its endogenous G protein(s), or 3) allowing more receptors to remain in the active configuration for longer periods or increasing the total number of receptors in the active configuration (Hall, 2000; Parmentier et al., 2002; Gilchrist, 2007; Kenakin, 2007; Leach et al., 2007; Schwartz and Holst, 2007). Unfortunately, to date our data have not allowed us to discriminate between these three potential mechanisms, although we did determine that the potentiator increases the affinity of the agonist for mGluR2 (Schaffhauser et al., 2003).

The site of action of MRLSD-650 was found to be to the same three amino acid residues that bind the novel mGluR2 potentiator LY487379 (Schaffhauser et al., 2003). Our detailed molecular investigation of these mGluR2 and mGluR3 chimeric receptors and point mutations also determined that our transient assay systems (PI and SPA) express high levels of the mGluR2. Interestingly, MRLSD-650 could evoke a small agonist response in these systems in the absence of exogenous glutamate. Knoflach et al. (2001) reported similar effects with an mGluR1 allosteric modulator. No intrinsic activity was observed for MRLSD-650 in assays using well washed native rat brain membrane preparations or hmGluR2 membranes from stably transfected cells, in which contaminating glutamate levels were very low, suggesting that the intact transfected cells may have intrinsic glutamate activity in this transient assay system. Nevertheless, addition of MRLSD-650 in combination with a low level of glutamate potentiated the response above that observed with glutamate alone, except when the TMIII-IV domain was exchanged with that of hmGluR3. This work confirms our previous findings that the TMIII-V domain is critical for hmGluR2 potentiation.

Previously, we reported that the combination of amino acids Ser⁶⁸⁸ or Gly⁶⁸⁹ in TMIV with Asn⁷³⁵ in TMV, or the combination of these two residues together with Asn⁷³⁵, were the critical amino residues in forming the binding pocket for LY487379 (Schaffhauser et al., 2003). This was confirmed by a recent study (Hemstapat et al., 2007). We have now confirmed that MRLSD-650 also binds to the same residues. Additionally, the importance of these residues was further determined when we substituted these same three amino acids, Ser⁶⁸⁸, Gly⁶⁸⁹, and Asn⁷³⁵, into the homologous region of hmGluR3. This hmGluR2-like binding site on mGluR3 was potentiated by the mGluR2-specific MRLSD-650, suggesting that we had created a functional hmGluR2-like potentiator binding site in the mGluR3. Following the development of the SPA assay, we identified three additional compounds, MRLSD-230, MRLSD-041, and MRLSD-973 (Fig. 1, B–D), that potentiate glutamate activity on hmGluR2, but not when the three critical residues were exchanged. In contrast to these mGluR2-specific potentiators, we also identified a series of compounds that potentiated the agonist response in the hmGluR2 S688L, G689V, and N735D chimera: MRLSD-064, MRLSD-726, MRLSD-438, and MRLSD-667 (Fig. 1, E–H). These were tested and found to be potentiators in hmGluR3 membranes as well. Thus, the hmGluR2 S688L, G689V, and N735D mutant may be used to discriminate between selective and nonselective modulators of hmGluR2 in a single assay. This avoids having to use the ultrasensitive hmGluR3 membranes, or hmGluR3 transiently transfected cells, which can be difficult to work with due to endogenous glutamate issues.

Despite finding MRLSD-064, MRLSD-726, MRLSD-438, and MRLSD-667 to be active in human mGluR2 and human mGluR3 recombinant membranes as well as in transiently transfected cells, we did not see biphasic curves with the rat brain membranes. A recent study (Gu et al., 2008) gives a very detailed report on the distribution of mGluR2 and mGluR3 in the rat forebrain. The two receptors seem to be fairly evenly distributed throughout the forebrain, suggesting that making large-scale membrane preparations from specific brain regions to favor mGluR2 or mGluR3 would be very difficult. However, it would seem from their study that sufficient levels of mGluR3 should have been present in the membranes prepared from rat whole brain to see an agonist or allosteric modulator response. Our recombinant mGluR3 membranes had an EC₅₀ value of just >100 nM (data not shown), and it is difficult to remove endogenous glutamate in the rat brain preparations to levels below 1 μM to use for mGluR2 testing. It is likely that residual glutamate was already near the mGluR3 E_max value before the addition of glutamate. Therefore, at the 1 μM concentration of glutamate we used in the assay, the native rat brain mGluR3 would likely have been fully activated by residual glutamate. Any additional mGluR3 activity caused by the compounds would be above the 100% glutamate response (i.e., from 100 to 110%). This activity was probably masked by the much more potent mGluR2 potentiator response, where the response goes from 10% of the mGluR2 glutamate stimulation to 90% or higher. Attempts to further remove the residual glutamate from the rat brain preparations resulted in increasing loss of activity.

Novel negative allosteric binding sites have been identified for the noncompetitive mGluR1 antagonist 1-ethyl-2-methyl-6-oxo-4-(1,2,4,5-tetrahydro-benzo[d]azepin-3-yl)-1,6-dihydro-pyrimidine-5-carbonitrile, which involves amino acids in multiple transmembrane domains (Val⁷⁵⁷ in TMV; Trp⁷⁹⁸, Phe⁸⁰¹, and Tyr⁸⁰⁵ in TMVI; and Thr⁸¹⁵ in TMVII) (Malherbe et al., 2003), and recently for mGluR2 (Hemstapat et al., 2007). Although the binding pocket for the mGluR2 negative allosteric modulator has not yet been determined, it does not seem that the negative allosteric binding sites overlap with the residues identified for positive modulation (Hemstapat et al., 2007). Knowledge obtained from the mapping of new allosteric modulators is helping to determine which regions of the mGluRs are critical for modulation of agonist activity.

The use of highly selective mGluR-positive allosteric modulators in the therapeutic treatment mGluR-related diseases might be an attractive approach since these modulatory compounds would only be efficacious in the presence of endogenous agonist (i.e., use-dependent). They might also elicit less tachyphylaxis and/or receptor desensitization than competitive agonists (Raddatz et al., 2007). Furthermore, it seems to be easier to find very selective ligands that bind outside of the agonist binding domains, which are highly conserved between all known mGluRs. MRLSD-650 and related compounds therefore constitute valuable tools to explore the role of mGluR2 in neuronal function, and the potential clinical utility of selective mGluR2 allosteric potentiators.