Selectivity and Evolutionary Divergence of Metabotropic Glutamate Receptors for Endogenous Ligands and G Proteins Coupled to Phospholipase C or TRP Channels*

Background: Endogenous ligands and G-proteins regulate responses of metabotropic glutamate receptors. Results: l-serine-O-phosphate activates Group III while antagonizing Group I & II. mGluR1 couples to Gαi/o as well as Gαq/11. Conclusion: l-Serine-O-phosphate differentially interacts with mGluR. mGluR1 has a bifurcating G-protein coupling. Significance: Defining the interactions of mGluR with endogenous ligands and G-proteins is important for understanding their roles in human neuronal function and for design of therapeutics. To define the upstream and downstream signaling specificities of metabotropic glutamate receptors (mGluR), we have examined the ability of representative mGluR of group I, II, and III to be activated by endogenous amino acids and catalyze activation of G proteins coupled to phospholipase C (PLC), or activation of Gi/o proteins coupled to the ion channel TRPC4β. Fluorescence-based assays have allowed us to observe interactions not previously reported or clearly identified. We have found that the specificity for endogenous amino acids is remarkably stringent. Even at millimolar levels, structurally similar compounds do not elicit significant activation. As reported previously, the clear exception is l-serine-O-phosphate (l-SOP), which strongly activates group III mGluR, especially mGluR4,-6,-8 but not group I or II mGluR. Whereas l-SOP cannot activate mGluR1 or mGluR2, it acts as a weak antagonist for mGluR1 and a potent antagonist for mGluR2, suggesting that co-recognition of l-glutamate and l-SOP arose early in evolution, and was followed later by divergence of group I and group II mGluR versus group III in l-SOP responses. mGluR7 has low affinity and efficacy for activation by both l-glutamate and l-SOP. Molecular docking studies suggested that residue 74 corresponding to lysine in mGluR4 and asparagine in mGluR7 might play a key role, and, indeed, mutagenesis experiments demonstrated that mutating this residue to lysine in mGluR7 enhances the potency of l-SOP. Experiments with pertussis toxin and dominant-negative Gαi/o proteins revealed that mGluR1 couples strongly to TRPC4β through Gαi/o, in addition to coupling to PLC through Gαq/11.

To define the upstream and downstream signaling specificities of metabotropic glutamate receptors (mGluR), we have examined the ability of representative mGluR of group I, II, and III to be activated by endogenous amino acids and catalyze activation of G proteins coupled to phospholipase C (PLC), or activation of G i/o proteins coupled to the ion channel TRPC4␤. Fluorescence-based assays have allowed us to observe interactions not previously reported or clearly identified. We have found that the specificity for endogenous amino acids is remarkably stringent. Even at millimolar levels, structurally similar compounds do not elicit significant activation. As reported previously, the clear exception is L-serine-O-phosphate (L-SOP), which strongly activates group III mGluR, especially mGluR4,-6,-8 but not group I or II mGluR. Whereas L-SOP cannot activate mGluR1 or mGluR2, it acts as a weak antagonist for mGluR1 and a potent antagonist for mGluR2, suggesting that co-recognition of L-glutamate and L-SOP arose early in evolution, and was followed later by divergence of group I and group II mGluR versus group III in L-SOP responses. mGluR7 has low affinity and efficacy for activation by both L-glutamate and L-SOP. Molecular docking studies suggested that residue 74 corresponding to lysine in mGluR4 and asparagine in mGluR7 might play a key role, and, indeed, mutagenesis experiments demonstrated that mutating this residue to lysine in mGluR7 enhances the potency of L-SOP. Experiments with pertussis toxin and dominant-negative G␣ i/o proteins revealed that mGluR1 couples strongly to TRPC4␤ through G␣ i/o , in addition to coupling to PLC through G␣ q/11 . Responses to glutamate, the main excitatory neurotransmitter in the central nervous system, are mediated primarily either by ionotropic receptors, which lead to fast membrane depolarization driving action potentials, or by metabotropic receptors of the class C G protein-coupled receptor (GPCR) 2 family (1). These metabotropic, or mGluR, receptors lead to slower but longer-lasting modulatory effects driven either by activation of phospholipase C (PLC, for group I mGluR, mGluR1, and mGluR5) and release of Ca 2ϩ from intracellular stores or by activation of G i/o class G proteins (for group II and III) coupled strongly to diverse ion channels, weakly to PLC, and under special circumstances (strong activation of adenylyl cyclase), to suppression of cAMP synthesis. mGluR proteins are of considerable interest for both basic and translational neuroscience because of their critical roles in modulatory neuronal signaling, emotions, learning and memory, and neurodegeneration. They are considered prime targets for psychotropic or neuroprotective drugs, so the structural and mechanistic bases for their specificities are of considerable importance for drug discovery and design.
Specific recognition of L-glutamate poses a structural challenge, because there are abundant amino acids in the central nervous system, which are close structural homologues. These include D-glutamate, L-glutamine, L-aspartate, L-and D-serine, D-serine-O-phosphate, and L-serine-O-phosphate (L-SOP). In general, other amino acid receptors, including those which recognize L-glutamate, demonstrate relatively low specificity. Bacterial amino acid receptors, from whose venus flytrap domain the ligand recognition domain of mGluR is derived (2), and taste receptor T1R1ϩT1R3 (3,4) for L-glutamate, have considerable cross-recognition of similar amino acids (4) as do other GPCRs, such as calcium-sensing receptors (3) and ionotropic glutamate receptors (6 -8). Previously, it was reported that there is stringent specificity of L-glutamate for all subtypes of mGluR (9). Here we tested not only the amino acids studied in that report, but also L-SOP, and structural analogues such as * This work was supported by National Institutes of Health Grants R01-D-serine-O-phosphate or D-serine to verify specificity of these endogenous ligands for mGluR. L-SOP is of particular interest, as concentrations of L-SOP in brain extracts have been reported to be comparable to or exceed those needed to activate Group III mGluR, and, in addition, the enzymes needed for production and degradation of L-SOP are found in both neurons and glial cells (10).
The studies reported here aim to address the following questions: how stringent is the recognition specificity of each of the mGluR groups for endogenous amino acid ligands, what are their specificities for G proteins, and can we use molecular docking approaches to identify structural determinants of specificity? Although there have been a number of previous studies using a variety of assays to address these questions for individual mGluR, the studies in this work represent a uniquely comprehensive survey of representatives of all three sub-groups using parallel assays and expression systems. In addition, our novel assays have allowed us to ask these questions with higher sensitivity than has previously been possible, permitting the observation of previously undetected overlaps in specificity.

MATERIALS AND METHODS
Compounds and cDNA Clones-L-glutamate, L-serine-Ophosphate, L-glutamine, D-serine, L-aspartate, GABA, L-arginine, taurine, and ␤-alanine were purchased from Sigma Aldrich Inc. L-serine was purchased from Spectrum Chemicals Corp. and D-SOP was purchased from Santa Cruz Biotechnology. LY367385 was purchased from Tocris Bioscience. Pertussis toxin was purchased from Calbiochem Inc. The rat mGluR1a cDNA was tagged at N-terminal with triple myc epitope (a gift from Dr. Anna Francesconi, Albert Einstein College of Medicine). N-terminal Myc-tagged rat mGluR2 and rat mGluR4a constructs were gifts from Hans Bräuner-Osborne (University of Copenhagen, Denmark). Rat mGluR7 cDNA construct is a gift from Dr. Shigetada Nakanishi (Osaka Bioscience Institute, Japan). cDNA encoding mGluR7 was subcloned into backbone of mGluR4 where cDNA of mGluR4 is excised out and replaced by cDNA of mGluR7. mGluR6 was PCRamplified from mouse retina cDNA (11) using primers 5Ј-CCAGAGGTTGGCTCAGTCCAGGAGC-3Ј and 5Ј-CAACC-TTGCTACTTGGCGTCCTCTG-3Ј and cloned. The open reading frame was then subcloned into pCDNA3.1 using PCR to add the kozak sequence GCCACC immediately preceding the start codon. cDNA for G ␣15 and G ␣16 were kindly provided by Dr. Melvin Simon (Caltech) and chimeric G proteins G␣ qi and G␣ qo were gifts from Dr. Frank Conklin (UCSF). mGluR7 mutant N74K and dominant negative mutants of G␣ i1 (G202T), G␣ i2 (G203T), G␣ i3 (G202T), and G␣ oA (G203T) were generated using the QuickChange mutagenesis kit (Stratagene, La Jolla, CA).
Cell Culture and Transfection-HEK-293 cells were maintained with complete Dulbecco's modified Eagle's medium (DMEM), which is composed of 10% fetal bovine serum (FBS), 2 mM L-glutamine, 100 units/ml penicillin G, 100 g/ml streptomycin at 37°C in the presence of CO 2 . HEK/TRPC4 cells which stably express TRPC4␤ were maintained in complete DMEM with 0.5 g/liter of G418. Cells were plated in 96 well, black-walled, clear-bottomed poly-d-lysine coated plate at a density of 75,000 cells per well and transfected with Lipofectamine 2000 as recommended by manufacturer. mGluR-expressing plasmids were transfected with or without plasmids directing expression of promiscuous G proteins.
Calcium Mobilization Assay-After 36 h following transfection, cells were washed with Krebs/Ringer/Hepes (KRH) buffer (120 mM NaCl, 4.7 mM KCl, 1.1 mM CaCl 2 , 10 mM HEPES, 1.2 mM KH 2 PO 4 , 1.2 mM MgSO 4 , pH 7.4) supplemented with 1.8 g/liter glucose, and 1 mM probenecid to inhibit dye efflux. Cells were incubated with 2.5 M Fluo-4 AM (Invitrogen) for 1 h at room temperature. Afterwards cells were washed twice, and then incubated with 80 l of buffer at room temperature for 30 min. Drug plates were prepared with different concentrations of different drugs to be tested at 3 times the desired final concentration, and 40 l of drug solution was injected after 20 s of recording (to determine baseline signal). Ca 2ϩ -enhanced fluorescence (excitation 485 nm and emission 525 nm) was detected using a FlexStation 3 microplate reader (Molecular Devices). As controls for background signals, cells transfected with vector lacking an expression construct were challenged with drugs in parallel.
Membrane Potential Assay-HEK-293/TRPC4␤ cells (a gift from Dr. Michael X. Zhu, UT Health Science Center at Houston, TX) were maintained with complete Dulbecco's modified Eagle's medium (DMEM) which is composed of 10% fetal bovine serum (FBS), 2 mM L-glutamine, 100 units/ml penicillin G, 100 g/ml streptomycin, 0.5 g/liter G418 at 37°C in the presence of CO 2 . After 24 -48 h of transfection, media was removed, and cells were incubated with membrane potential dye and quencher combination (Molecular Devices) diluted in KRH buffer for 30 min at room temperature. For testing PTX effects, cells were pre-incubated with the media or PTX (100 ng/ml) for 10 -13 h, and then the media with or without PTX was washed away and replaced with KRH buffer prior to the assays. Assays were performed at 32°C. 40 l of 3ϫ drug solution was injected after 20 s of baseline recording, and signal was recorded for a total of 150 s. Data were analyzed with GraphPad Prizm software.
Data Analysis for Dose Response Curves-For all dose-response curves, different concentrations of ligand were added to different wells on the same plate, and the time course of responses recorded for each. We have analyzed the data in the following ways: First, we have corrected for the artifactual change in fluorescence observed upon dilution when ligand is added by subtracting the time-varying background signal (determined with mock ligand addition or mock transfection) from each time course. Then we have averaged the three points of maximum value up to 150 s (when the carbachol control was added) to minimize effects of fluctuations in signal, to obtain the maximum value used to determine each point in the doseresponse curves. In fitting the dose response curves, the maximum value was allowed to float, as were the K 1/2 value and the base (minimum) value. The Hill coefficient was fixed at a value of n ϭ 1, i.e. the fitting function was a simple hyperbolic saturation curve. Although there is no reason a priori to rule out some cooperativity with dimeric receptors, there was no higher value of n that consistently yielded better fits than n ϭ 1. For each fit, the standard error of the fit for the Ϫlog 10 (EC 50 ) value was calculated for the three replicates, and then propagation of errors was used to determine the standard error of the mean for the independent experiments.
Molecular Docking-Molecular docking of L-SOP into the ligand-binding region of mGluR7 (PDB ID 2E4Z) was performed using RosettaLigand (12,13) and the Automatic Roset-taLigand Setup (ARLS) scripts. A library of twenty-five low energy L-SOP conformations was generated using OpenEye's Omega (14) software and each was placed within an area of 8.0 Å centered on the experimentally determined binding pocket. Initially rigid-body translation was performed until the center of the ligand is free from steric clashes with the receptor and then rotated into 1000 orientations to determine optimum compatibility with nearby atoms at that location. The best (or one of the best) orientation was then perturbed translationally and rotationally in a Monte Carlo minimization scheme, side chains, and ligand were repacked, and all degrees of freedom were minimized. Final minimization involved backbone torsion angle minimization and side-chain rotamer sampling. In all, over 40,000 trajectories for the ligand-receptor pair were performed.
Evolutionary Trace Analysis-A combination of sequence and pattern analysis was used to determine the functionally important residues that 1) differentiate G ␣q/11 and G␣ i/o activation by group I mGluR, 2) characterize specificity toward L-SOP activation for group III mGluR and 3) define the unique properties of mGluR7 such as weak activation of G␣ i/o or lack of TRPC4␤ coupling. Evolutionary trace analysis was performed with rvET (15). rvET algorithm measures the importance of a residue based on its variation pattern in the context of the phylogenetic tree. To identify the likely functional residues leading to specificity, we focused on the top ranked ET residues (top 45%) that also followed a set amino acid patterns within subfamilies.
To construct a multiple sequence alignment, protein sequences were gathered from a BLAST search of the UniRef100 database and aligned with PROMALS3D (16). To identify functional homologues, the proteins sequences were filtered based on sequence identity (Ͼ30%). Evolutionary trace analysis was performed on 240 mGluR homologues. We focused the analysis on the transmembrane region of the protein as defined by current mGluR1 structure PDBID 4OR2.

Stringent Specificity of Endogenous
Ligand for mGluR-Ten amino acids that are known to be present at substantial concentrations in cerebrospinal fluid (17) and D-SOP were applied to HEK 293 cells transfected with representative members for all subtypes of mGluR. mGluR1 naturally couples to G␣ q so it was transfected alone but the other mGluR, which are G␣ i/o coupled receptors, were co-transfected with a promiscuous G protein, either G␣ 15 , or G␣ qo . When 1 mM L-glutamate was applied to mGluR1-transfected HEK cells, a robust calcium signal was produced (Fig. 1A). However, none of other amino acids at 1 mM concentration produced detectable signal in the cells transfected with mGluR1, indicating L-glutamate is the only endogenous ligand which activates mGluR1. Similar results were observed for mGluR2 transfected HEK cells indicating that L-glutamate is highly specific to mGluR1 and mGluR2 over other endogenous ligands (Fig. 1B). However, for mGluR4 transfected cells, not only L-glutamate but also L-SOP generated enhanced intracellular calcium responses (Fig. 1C). Similar results were observed for mGluR6-and mGluR7-transfected cell as well (Fig. 1, D and E). L-SOP Behaves as an Antagonist for mGluR1 and mGluR2-Next, 11 amino acids were tested again for their potential behavior as antagonists for mGluR1 and mGluR2. A mixture of 1 mM of each amino acid and 3 M L-glutamate (EC 50 for FIGURE 1. Agonism of endogenous amino acids for representative mGluR of group I, II, and III. Ca 2ϩ responses elicited in response to 1 mM amino acids were measured by fluorescence intensity changes (Relative Fluorescence Unit, RFU) in a microplate reader. HEK-293 cells were transiently transfected with an mGluR-expressing plasmid with or without promiscuous G ␣qo protein, loaded with dye, and treated with each ligand at t ϭ 20s. A, mGluR1 (group I), B, mGluR2 (group II), C, mGluR4 (group III), D, mGluR6 (group III), E, mGluR7 (group III). The drug response for mock-transfected cells was subtracted from each drug response. Each column is the mean Ϯ S.E. (*, p Ͻ 0.05) of triplicates from three independent experiments. p values were calculated from independent two-tailed t tests. mGluR1) was applied to mGluR1 transfected HEK cells. As shown in Fig. 2A, L-SOP inhibited the L-glutamate mediated mGluR1 response (p Ͻ 0.005). When different concentrations of L-SOP were applied to mGluR1-transfected cells, there was increased inhibition of the L-glutamate response at higher concentrations (Fig. 2B). The estimated K i derived from this graph is about 1 mM. In addition, we also tested the antagonistic effect of L-SOP on mGluR2. Interestingly, L-SOP displayed a substantially more potent inhibition of mGluR2 activation (Fig. 2B); the estimated K i value was 1 M, three orders-of-magnitude more potent than for mGluR1. Thus L-SOP binding at the L-glutamate site is conserved across all three mGluR groups, but it behaves as a potent antagonist toward mGluR2 and weak antagonist toward mGluR1, while acting as a potent agonist for group III mGluR.
A Key Residue in L-SOP Recognition-A puzzling feature of group III mGluR is that all members except mGluR7 show similar potencies or activation by L-glutamate or L-SOP, yet mGluR7, with 68.6% sequence identity and 82% sequence similarity to mGluR4 responds to L-SOP with two orders of magnitude lower potency (Fig. 3C). To gain insights into mechanisms behind this evolutionary divergence, we took advantage of the availability of a crystal structure of the mGluR7 ligand binding domain (18) (PDB ID: 2E4Z) by probing potential binding determinants for L-SOP using molecular docking. One of the highest scoring positions (ranked by the RosettaLigand binding energy function) suggested that the phosphate ester on L-SOP would likely be positioned near asparagine 74 in the lowest energy docked complexes, but would have the closest potential hydrogen bond acceptor of L-SOP 5.0 Å away from the potential hydrogen bond donor in the amide nitrogen of the asparagine 74 side chain (Fig. 3A). Computational substitution of the residue found at that position in mGluR4, lysine, suggested a much more favorable distance (2.5 Å) as well as a full positive charge, which would provide favorable electrostatic interactions with the phosphate (Fig. 3B). Accordingly, we substituted lysine for asparagine at position 74 in mGluR7 and tested its responses to L-SOP. The mutation shifted the potency by a factor of 5 to an EC 50 of 3.7 M, as compared with an EC 50 of 19 M for wild type mGluR7 (Fig. 3C). This result confirms that residue 74 plays a critical role in determining the affinity of group III for L-SOP. Interestingly, other group III mGluR have Q, N, or K at that position, while it is variably Y in group I or R in group II, which bind but do not respond to L-SOP (Fig. 3D). The extended length and the positive charge on the arginine side chain in mGluR2 may explain the much greater potency with which it is inhibited by L-SOP as compared with mGluR1. To investigate the role in group 1 and group 2 receptors of the residues corresponding to 74 in mGluR4 and mGluR7, lysine substitutions in mGluR1 (mGluR1-Y74K) and mGluR2 (mGluR2-R74K) were generated respectively and tested against for both L-SOP and L-glutamate. mGluR1-Y74K abolished responses to L-glutamate (Fig. 3F) and L-SOP, although it was expressed well at the cell surface (data not shown). On the other hand, mGluR2-R74K showed lowered potency for both L-glutamate and L-SOP by ϳ45 Ϯ 13-fold and 13 Ϯ 2-fold, respectively (Fig. 3E). Thus, although the potency of L-SOP decreased, selectivity for L-SOP over glutamate increased more than 3-fold. These data indicate that position 74 plays a critical role in determining the affinity of group I for L-glutamate and of group II for L-glutamate and L-SOP as well as its significant role in group III for L-SOP specificity.
Selectivity of mGluR for Phospholipase C Activating G Proteins-Four different phospholipase C (PLC)-activating G protein ␣ subunits were expressed with mGluR to test the ability of group II and group III mGluR to activate the PLC pathway and induce Ca 2ϩ release monitored with fluorescent dyes. Naturally occurring G ␣15 and G␣ 16 are known to couple to many GPCR (19,20), whereas G␣ qo and G␣ qi are synthetic constructs with C-terminal tails from G␣ o or G␣ i2 , respectively, substituted for the C-terminal sequence of G␣ q (21,22). As shown in Fig. 4A, calcium release was observed in cells that expressed mGluR1 only or cells that co-expressed mGluR1 and other promiscuous G proteins. Interestingly, there was enhanced calcium signal from the cells which expressed promiscuous G proteins and mGluR1, compared with the signal from cells which expressed mGluR1 alone (Fig. 4A), indicating that mGluR1 can couple to PLC through promiscuous G proteins, including those containing C-terminal sequences from G␣ o or G␣ i2 . This result fits well with the observation (see below) that mGluR1 couples well to G␣ i/o proteins. In contrast, calcium responses were not observed in cells expressing mGluR2, mGluR4, mGluR6, or mGluR7 receptors without supplemental G proteins (Fig. 4, B-E), indicating that these receptors do not couple to the endogenous G␣ q/11 proteins in HEK cells. On the other hand, cells co-expressing group II and III receptor and the PLCcoupled G␣ subunits all displayed robust release of Ca 2ϩ upon   together with A, mGluR1, B, mGluR2, C, mGluR4, D, mGluR6, or E, mGluR7. Fluorescence from dye-loaded cells was measured for 20 s prior to addition at t ϭ 20 s of 100 M L-glutamate or 100 M L-SOP. Each bar represents the mean peak fluorescence Ϯ S.E. (*, p Ͻ 0.0001) of triplicates from three independent experimentations. The p value was calculated from independent two tailed t test.
L-glutamate addition. These results confirm that these group II and III mGluR do not couple detectably to endogenous G␣ q/11 , in contrast to mGluR1, which elicits robust Ca 2ϩ release in response to L-glutamate without co-expression of exogenous G␣.
As shown in Fig. 4A, in addition to its ability to activate endogenous G␣ q/11 , mGluR1 differs from group II and group III mGluR in that it does not appear to couple well to G␣ 15 . However, it does display robust coupling to G␣ 16 , and also to the synthetic constructs G␣ qo and G␣ qi , a result that reveals that the C-terminal tails of G␣ o and G␣ i , while conferring coupling of these proteins to group II and group III do not greatly interfere with coupling to mGluR1, perhaps as a result of differences in G protein recognition mechanisms used by group I versus group II and group III mGluR.
Under the simplifying assumption that mGluR1 coupling efficiency is given by the difference between the response for mock-transfected cells and the response for mGluR1-transfected cells for endogenous G␣ q/11 , and by the difference between the response for cells with endogenous proteins only and those for cells transfected with specific G␣ constructs, the relative G␣ coupling efficiencies were G␣ q ϳ G␣ qo Ͼ G␣ 16 Ͼ G␣ qi Ͼ G␣ 15 ϳ 0 for mGluR1. For group II mGluR2, the relative G␣ coupling efficiencies were G␣ qo ϳ G␣ 16 Ͼ G␣ qi Ͼ G␣ 15 (Fig. 4B). For group III mGluR4, they were G␣ qo Ͼ G␣ qi ϳ G␣ 15 Ͼ G␣ 16 . For group III mGluR6, they were G␣ qo Ͼ G␣ 15 Ͼ G␣ qi ϳ G␣ 16 , and for mGluR7, they were G␣ qo Ͼ G␣ 15 Ͼ G␣ qi Ͼ G␣ 16 , with G␣ 16 activation being barely detectable (Fig. 4, C-E). Because in the long run we are interested in using evolutionary approaches (23) to understanding mGluR, we chose to use G␣ 15 or G␣ qo for subsequent Ca 2ϩ release assays, as relatively robustly but promiscuously coupled PLCcoupled G␣ subunits.
L-Glutamate and/or L-SOP Induced Membrane Potential Change in HEK/TRPC4␤ Cells Transiently Expressing mGluR-Next, we examined if group II or group III mGluR can couple to the TRPC4␤ channel through activation of G␣ i/o . For this purpose we used an HEK-derived cell line stably expressing TRPC4␤ channels (HEK/TRPC4␤) (24,25). As shown in Fig.  5A, application of L-glutamate to HEK/TRPC4␤ cells transiently transfected with mGluR2 resulted in a concentrationdependent increase in membrane potential; no response was elicited in empty vector-transfected HEK/TRPC4␤ cells. Plotting the maximal response, corrected for background as described under "Materials and Methods," as a function of concentration generated a dose response curve, which fit well when a non-linear regression was applied (Fig. 5B). The Ϫlog 10 (EC 50 ) from mGluR2 transfected HEK/TRPC4␤ cells is 4.63 Ϯ 0.12. The potency observed in these experiments is within the previously reported Ϫlog 10 (EC 50 ) range of 4.69 -5.39 for mGluR2 (26), although less potent than reported for responses based on G protein-regulated inwardly rectifying potassium channels (GIRK) and a thallium flux assay, where the Ϫlog 10 (EC 50 ) was found to be 5.1 (27). Similar results were observed for HEK/ A, B, mGluR2, C, D, mGluR4 or E, F, mGluR6. The assay was performed by reading baseline fluorescence for 20 s followed by addition of increasing concentrations of L-glutamate. Fluorescence of mock-transfected cells was subtracted from each trace recorded in the mGluR-transfected cells. B, D, F, maximal responses between 20 s and 150 s after L-glutamate injection plotted as a function of L-glutamate concentration. Responses from cells were represented in the presence (Ⅺ) or absence (f) of 100 ng/ml pertussis toxin (PTX). The dose response curves are fits to a hyperbolic binding isotherm generated using a non-linear regression algorithm in the Prizm graphPad software package, as described in "Materials and Methods." TRPC4␤ cells transfected with mGluR4 or mGluR6 (Fig. 5, C and E). The Ϫlog 10 (EC 50 ) for mGluR4 was 4.07 Ϯ 0.10 and the Ϫlog 10 (EC 50 ) for mGluR6 was 4.13 Ϯ 0.06, a little lower than the previously reported (1) values 5.52-4.70 and 4.80 for mGluR4 and mGluR6, respectively (Fig. 5, D and F), suggesting that coupling to the TRPC4␤ channel might be less efficient than coupling to previously studied effector molecules. Note that because the L-glutamate concentrations did not extend into the high millimolar range to avoid nonspecific effects on signaling, the true Ϫlog 10 (EC 50 ) values may be even higher, indicating that signaling through TRPC4␤ may require more G protein activation than previously studied pathways. Surprisingly, when L-glutamate was applied to HEK/TRPC4␤ cells transfected with mGluR7, no significant response was observed.

FIGURE 5. Membrane potential changes induced by L-glutamate induced in TRPC4␤-expressing cells transiently transfected with constructs expressing
Another endogenous ligand, L-SOP, was also tested with HEK/TRPC4␤ cells which were transfected with mGluR4 or mGluR6. L-SOP addition to mGluR4-transfected HEK/TRPC4␤ cells resulted in a concentration-dependent fluorescence increase as shown in Fig. 6A. The Ϫlog 10 (EC 50 ) value calculated from the dose response curve was 5.01 Ϯ 0.06 (Fig. 6B), similar to the previously reported values of 5.4 -6.3 (10,28) and somewhat less potent than the 6.5 -log 10 (EC 50 ) determined in our calcium assays in HEK cells co-transfected with G␣ qo . Similarly, Ϫlog 10 (EC 50 ) for mGluR6 was 4.83 Ϯ 0.05 in this assay (Fig. 6, C and D), comparable to previously reported Ϫlog 10 (EC 50 ) values of 6.4 -5.5 (28). However, there was little response from mGluR7-transfected cells (Fig. 9A), even at L-SOP concentrations giving robust responses in cells co-expressing promiscuous PLC-coupled G proteins. To rule out the possibility of mislocalization of mGluR7 in HEK/TRPC4 cells, we performed experiments to determine whether in the TRPC4␤-expressing cell line mGluR7 can activate G␣ qo and trigger Ca 2ϩ release. Since it does so (Fig. 9B), it is clear that mGluR7 is localized to the cell surface membrane properly, as verified by ELISA measurement of cell surface expression (Fig.   9C). Since mGluR7 is well documented to activate G␣ i/o proteins (29 -31), and is clearly functional in our assays, we hypothesize that because mGlurR7 is a weaker activator than mGluR4 or mGluR6, even at saturating agonist concentrations, it may be that a necessary threshold level of G i/o activation needed for stimulation of TRPC4␤ is not reached. In order to gain insights into a possible structural basis for this lack of TRPC4 coupling to mGluR7, we performed evolutionary trace analysis on group III mGluRs and found several candidate residue located in G protein interacting sites, as discussed below under "Evolutionary Trace Analysis." L-Glutamate and L-SOP-induced TRPC4␤ Activation Is Mediated by G␣ i/o Proteins-To test if mGluR coupling to theTRPC4␤ channel is mediated by G␣ i/o , HEK/TRPC4␤ cells transfected with mGluR2 were treated overnight with 100 ng/ml pertussis toxin (PTX) to ADP-ribosylate G␣ i/o proteins in trimeric G protein complexes and inhibit G protein coupling to the receptor. This treatment significantly reduced the L-glutamate-mediated membrane potential response (Fig. 5B), indicating that the L-glutamate mediated response for mGluR2 is dependent on G␣ i/o . Similarly, there was PTX sensitive inhibition of mGluR4 or mGluR6 in transfected HEK/TRPC4␤ cells activated either by L-glutamate or L-SOP, indicating these mGluR couple to the TRPC4␤ channel via the G␣ i/o pathway (Figs. 5, D and F, and 6, B and D).
The Proteins of G␣ i/o Family Play a Role in mGluR1 Signaling-To extend our observation of the enhanced Ca 2ϩ responses in cells that co-expressed mGluR1 and promiscuous G proteins (Fig. 4A), we examined whether mGluR1, which is known to couple primarily to proteins of the G␣ q family, can couple to the TRCP4␤ channel via G␣ i/o . Interestingly, an increase in membrane potential was observed upon L-glutamate challenge of mGluR1-expressing cells. This result may be due at least partially to the known ability of mGluR1 to activate G␣ q/11 protein (32, 33), because the TRPC4 ␤ channel is known to couple to G q -coupled receptors as well as G␣ i/o -coupled  OCTOBER 24, 2014 • VOLUME 289 • NUMBER 43 receptors (34 -36). To determine whether G␣ i/o proteins might also be involved in mGluR1 signaling, PTX was incubated with mGluR1 expressing HEK/TRPC4␤ cells. Surprisingly the L-glutamate mediated mGluR1 response was partially diminished by PTX (Fig. 7A). After confirming that mGluR1 is involved in PTX-sensitive G␣ i/o signaling, we further investigated which subtypes are involved in mGluR1 signaling. We applied dominant negative mutant G␣ i/o proteins to compete with endogenous G proteins and prevent them from coupling to mGluR1 (Figs. 7, 8). When dominant negative mutant forms of G␣ i1 (G202T), G␣ i2 (G203T), G␣ i3 (G202T), or G␣ oA (G203T) were co-transfected with mGluR1, L-glutamate mediated responses were inhibited by all these mutant G␣ proteins. It is noteworthy that decreased response was observed in response to L-glutamate but not in response to carbachol (whose activation of TRPC4␤ is mediated by G␣ q/11 ) showing that this inhibition by dominant negative G proteins is specific for mGluR1, rather than a general block of signaling by all GPCR or of TRPC4␤. This conclusion is further supported by the failure of dominant negative G␣ oA to block G q/11 -mediated Ca 2ϩ release in response to activation of mGluR1 (Figs. 8, 9).

Endogenous Ligand and G Protein Specificity in mGluR
Evolutionary Trace Analysis-To gain insights into the structural and evolutionary basis for the overlapping but distinctive profiles of specificities, we applied computational approaches based on the Evolutionary Trace (ET; see "Materials and Methods"). To understand the functional divergence in G protein specificity between group I, which efficiently activate G␣ q/11 , and groups II and III, which do not, we examined the highly ranked ET residues specific to group I (invariant in mGluR1 and mGluR5 homologues) and identified two distinct groups of residues. One is found on the intracellular loop side of the transmembrane region (blue boxes, Fig. 10B), as might be expected because these loops would likely form the G protein binding site (blue spheres in Fig. 10C). We also noted the second intracellular loop to be enriched in such residues relative to the other intracellular loops, including a two residue insertion in group I relative to groups II and III. This loop has previously been noted to be key for G protein recognition (20). A second group of important residues specific to group I was found in the extracellular side of the transmembrane region and suggests an allosteric site (blue boxes and spheres in Fig. 11). Although it is not amenable to ET analysis due to sequence divergence among groups, the C terminus may also play a role at the cytoplasmic surface. Group I has a much longer terminal (over 300 hundred amino acids longer than groups II and III) and could also be a key difference needed for G q function.

Endogenous Ligand and G Protein Specificity in mGluR
For insights into contributions of the transmembrane domain to L-SOP activation specificity, we focused on the important positions that were specific to group III (invariant within mGluR4/6/7/8 homologues, but divergent in groups I and II). One of the key differences was in the extracellular loops where group III is longer and well conserved within group III. We marked the residues (green spheres in Fig. 10C) that did appear in mGluR1 structure (PDBID 4OR2). We noted that mGluR7 diverges from the rest of group III in the second extracellular loop, suggesting a possible contribution to the lower activation efficiency found in mGluR7 (red box in Fig. 10A). A second group of important residues specific to group III comprises four residues that make up a cluster on the cytoplasmic side of the transmembrane region (Fig. 11B).
To search for residues that might contribute to the inability of mGluR7 to activate G proteins efficiently enough to stimulate TRPC4␤ we focused on highly ranked residues specific to mGluR7 and found three highly ranked residues specific to mGluR7 in the intracellular loops (Fig. 10, B and  C), where they could impact G protein interactions. As noted above, mGluR7 also diverges from the rest of group III in the second extracellular loop (Fig. 10, A and C). In addition, we noted multiple residues throughout the transmembrane region specific to mGluR7 (Fig. 11), and these may influence the distributions of conformations induced by agonist binding.

DISCUSSION
The results of these studies represent a systematic characterization of specificity in upstream and downstream signaling of representative mGluR of groups I, II, and III. Using a highly sensitive assay we demonstrated the ability of all groups of mGluR to discriminate with high stringency against all endogenous amino acid ligands other than L-glutamate and L-SOP, including close structural analogues, even at millimolar concentrations (Fig. 1). This result is consistent with a previous study (9), which demonstrated stringent specificity of L-glutamate for all subtypes of mGluR. A first new finding is the stringent specificity for L-SOP of group III mGluR. Structural analogues of L-SOP such as D-SOP or L-serine did not elicit signals with group III. In addition, L-SOP, while not activating mGluR1 or mGluR2, was recognized by them as an antagonist with IC 50 values of ϳ1 mM or 1 M respectively (Fig. 2). This second new finding suggests that L-SOP may have a physiological role as a negative regulator of group II mGluR, as well as activation of group III. It has been reported that L-SOP inhibits the accumulation of 3 H-inositolmonophosphate elicited by ibotenic acid, a specific agonist for group I mGluR in rat hippocampal slices incubated in the presence of 7 mM Li ϩ (37). Also, L-SOP has been shown to displace [ 3 H]glutamate from specific binding sites on L-AP4 sensitive receptors, presumably group III mGluR, in the rat brain, and this result was supported by electrophysiological measurements (38). It is clear from the results presented here that L-SOP can compete with L-glutamate in the binding site of mGluR regardless of group. Specifically, L-SOP binds all groups of mGluR, but it activates group III as an agonist and inhibits group I or group II as an antagonist.
Interactions of L-SOP with group II and group III mGluR are likely to be physiologically important. Concentrations of L-SOP in rat brain extracts have been reported to be Ͼ 5.4 M (10). The concentration in healthy human brain has been reported to be ϳ300 M, and as high as 1 mM in patients with Alzheimer disease (39). Another study found L-SOP concentrations in the micromolar range in brains from pigs, mice, and chickens (40). Given that the total brain concentrations reported are sufficient for activation of group III mGluR or inhibition of group II, and that local concentrations near sites of synthesis and release are likely much higher, it seems very likely that L-SOP action on group II and group III mGluR is of physiological significance, either for normal cell signaling, or as a result of pathological release in brain injury or degeneration. However, an effect on group I mGluR is much less likely, suggesting a functional divergence of evolutionary importance.
This evolutionary divergence of L-SOP responsiveness suggests that it may have been a characteristic of an ancestral mGluR that was selectively lost by group II while being conserved by group III, or conversely, selectively acquired by group III. These findings lend support to the idea that L-SOP signaling is physiologically important, because the patterns of responses suggest that as the functions of mGluR diverged, retaining or eliminating L-SOP responsiveness conferred distinct selective advantages according to the function of each group. Among group III, mGluR7 responds to L-SOP less strongly than mGluR4 does. Using molecular docking, we have found that a key residue for conferring this distinction is asparagine 74 (Fig.  3). A role for this residue in ligand recognition is consistent with previous reports that the mGluR7 mutant N74K decreased the EC 50 of the group III-specific synthetic agonist L-AP4 by about 10-fold as compared with wild type mGluR7 (41), and that a K74Y mutation in mGluR4 enhanced its affinity for the mGluR1-specific synthetic agonist ibotenic acid (42). However, the evolutionary and functional significance of this residue can only be understood in the context of the endogenous ligands with which the receptors have been challenged over the course of evolution. Comparison of results from mGluR1, mGluR2, mGluR4, and mGluR6, reveals that lysine, glutamine, and arginine at this position are all compatible with potent activation or inhibition by L-SOP. Apart from mGluR7, all other group III mGluR, across vertebrate species, have lysine or glutamine at this position, suggesting that differentially selective L-SOP sensing is important for their functions, whereas the shift to lower L-SOP sensitivity is important for the function of mGluR7. The divergence in group III sensitivity to L-SOP appears to have occurred relatively late in evolution. Whereas all vertebrate mGluR7 have an asparagine at position 74, the proteins with the most similar sequence encoded in the Caenorhabditis elegans and Drosophila genomes have arginine at position 74.
Obviously position 74 accounts for only part of the functional differences between mGluR7 and other group III receptors, as EC 50 for L-SOP is still much higher that of mGluR4 or mGluR6, and it does not activate TRPC4␤ (Fig. 9A). Based on ET, other possible contributors are the second extracellular loop, which may be involved in coupling between the ligand binding domain and/or cysteine-rich domain and the transmembrane domain. This region may also contribute to the very different L-SOP responses of group II and group 1 versus that of group III. Based on the crystal structure of the transmembrane domain of mGluR1, this loop has been suggested as a potential coupling site between the extracellular domain and transmembrane domain (43). Divergence of sequence between mGluR7 and other group III receptors is also observed at the second and the third intracellular loops and C terminus (red boxes, Fig. 10B, red spheres in Fig. 10C). These residues are located in the cytosolic face of the transmembrane, that it, the G protein interacting site. These candidate residues may be contribute to unique mGluR7 properties such as lack of coupling to the TRPC4 channel and generally weaker downstream signaling as compared with other group III mGluR (44).
A third new set of surprising results was observed among the different mGluR with respect to G protein coupling using either a calcium mobilization assay based on phospholipase C activation or a membrane potential assay based on TRPC4␤ channel coupling. Group II and group III mGluR are conventionally described as sharing common G protein specificity, i.e. as being specific for G i and G o , and having no activity toward G q/11 , whereas group I mGluR are considered unique among mGluR in their specificity for G q/11 and G s and discrimination against G i/o (1,32,33). However, our results from calcium mobilization assays reveal that there is considerable overlap among the specificities of groups I, II, and III, in that all efficiently activate G␣ qo and, to a lesser extent, G␣ 15 and G␣ qi , whereas group III mGluR are poor at activation of G␣ 16 (Fig. 4). A broad specificity of G proteins for mGluR1 was observed using the membrane potential assay. Using treatment with pertussis toxin (PTX) or cotransfection with dominant negative G␣ i/o proteins, we found that mGluR1 can couple to G␣ i/o as well as to G␣ q/11 (Fig. 7A). Among the four types of dominant negative mutants of G␣ i/o proteins (i1, i2, i3, and o A ), the dominant negative mutants of G␣ i2 and G␣ o blocked the L-glutamate mediated mGluR1 activation of TRPC4␤ the most, and mutant G␣ i1 or G␣ i3 blocked the response to a lesser extent (Fig. 7, C-F). To test for the possibility that the dominant negative isoforms at high levels simply block activation of G q , we titrated cells with increasing amounts of dominant-negative-expressing plasmids (Fig. 7, C-F) to reveal that mGluR1 signaling to TRPC4␤ can be almost completely suppressed, whereas the same level of dominant negative subunits has little effect on signaling to TRPC4␤ by the M3 muscarinic acetylcholine receptor, a Class A GPCR known to work through activation of G␣ q/11 (45,46). In addition, the same levels of dominant-negative G␣ o that completely suppress TRPC4␤ activation, have no effect on Ca 2ϩ release induced by endogenous G q/11 in the same cell line (Fig. 8, A-B). These data indicate that the dominant negative G␣ o completely suppresses G i/o signaling by mGluR1, but has little or no effect on G␣ q/11 activation by either mGluR1 or the M3 muscarinic receptor. Recent evidence from behavioral and brain slice experiments in wild type and TRPC4-knock-out mice suggests a physiologically important role for mGluR1 activation of TRPC4; however, in this work it was assumed that mGluR1 regulates TRPC4 channels exclusively through G␣ q/11 stimulation of PLC (47). Both the inhibitory effects of PTX and dominant negative G␣ subunits on mGluR1 activation of TRPC4␤ FIGURE 10. Sequence alignment and ET analysis of the loops of the transmembrane domain of the mGluR family. Dark gray shading indicates residues with charged side chains, and light gray shading indicates hydrophobic residues. A, E1-E3, extracellular loops. B, I1-I3, intracellular loops, and C-term, initial, conserved portion of C-terminal domain. The remainder of the C-terminal domain is highly divergent. The coloring of the mGluR7 residues listed below the other sequences indicates their evolutionary importance as determined by ET percentile rank (5) (according to the displayed ET importance spectrum. The green boxes around subsets of residues indicate those that are conserved within Group III, but diverge in both Group I and Group II, and are therefore associated with L-SOP activation. The cyan boxes indicate residues that are conserved among Group II and Group III, but diverge in Group I, and are therefore associated with G q/11 activation. The red boxes indicate residues whose identities are unique to mGluR7, but common among mGluR7 orthologues from different species, and therefore associated with the unique functional features of mGluR7 within Group III, such as weak activation of G i/o . C, the structure of the transmembrane domain of mGluR1 (PDB:4OR2 (43)) with loop residues colored as for boxes. The box around the extracellular loop residues is drawn to call attention to residues missing in the structure because of the insertions in Group III as compared with Groups I and II, and because of lack of density in the x-ray data. and the positive responses in Ca 2ϩ -based fluorescence assays using promiscuous G proteins whose C termini are identical to those in G␣ o or G␣ i , support a significant role for G␣ o and G␣ i subunits coupling to mGluR1. Evolutionary trace analysis (Fig.  10, B and C) identifies a subset of residues that may be important for activation of G q/11 by mGluR1 and mGluR5. The importance of the intracellular loops and C terminus has been well documented in previous reports, with the second intracellular loop reported to be necessary but not sufficient for G protein specificity (44,48).
Taken together, the results demonstrate that fluorescence depolarization assays based on activation of the TRPC4␤ channel provide a sensitive continuous read-out of responses triggered by mGluR coupled to G␣ i/o . This approach can serve as a general method for development of high throughput assays for other receptors coupled to G␣ i/o using their physiological part-ners, rather than promiscuous phospholipase C-activating G proteins, whose responses may differ substantially from those of G␣ i/o .