Functional Coupling of a Human Retinal Metabotropic Glutamate Receptor (hmGluR6) to Bovine Rod Transducin and Rat Go in an in Vitro Reconstitution System*

The cDNA encoding hmGluR6, appended with a 15-amino acid antibody epitope (1D4), was transiently transfected in COS-7 cells. The receptor was purified from COS cell membranes using an antibody affinity column. The purified receptor was then reconstituted into lipid vesicles, and its ability to activate either transducin, the rod photoreceptor-specific GTP-binding protein, or the α subunit of Go was assayed in vitro using a guanosine 5′-3-O-(thio)triphosphate binding assay. Activation of both transducin and Go was observed. The rate of Goactivation was 18-fold greater than the rate of transducin activation. This indicates that the coupling of mGluR6 to Go is more efficient and suggests that Go may be involved in coupling to mGluR6 in ON-bipolar cells.

Glutamate, the major excitatory neurotransmitter in the central nervous system, activates both ionotropic receptors and metabotropic receptors (mGluR) 1 coupled to GTP-binding proteins (G-proteins). Recent molecular cloning studies have identified eight different subtypes of glutamate metabotropic receptors (1,2). These receptors possess seven putative membranespanning domains preceded by a large extracellular domain that probably functions as the ligand-binding domain (3,4). Despite having an overall design similar to other G-proteincoupled receptors, mGluRs do not have sequence similarity to other G-protein-coupled receptors, except the recently cloned ␥-aminobutyric acid ␤ receptors where the identity is only 18 -23% (5). The coupling of mGluRs to G-proteins and second messenger systems has been studied in heterologous expression systems using either Xenopus oocytes or stable cell lines. From these studies, it appears that a variety of different Gproteins are involved in the coupling of mGluRs to either phospholipase C or adenyl cyclase (1,2).
In the vertebrate retina, glutamate is the neurotransmitter released by photoreceptors in the dark. At the synapse between photoreceptors and bipolar cells, glutamate functions as both an excitatory and inhibitory neurotransmitter, exciting OFF-bipolar cells by opening ionotropic glutamate receptors and inhibiting ON-bipolar cells by activating a metabotropic receptor (6,7). L-2-Amino-4-phosphonobutyrate (L-AP4) mimics the action of L-glutamate at ON-bipolar cells and selectively hyperpolarizes these cells (8,9). Based on a series of physiological and pharmacological studies, it has been suggested that both L-glutamate and L-AP4 activate the G-protein-coupled mGluR expressed in ON-bipolar cells and activate a cGMP biochemical cascade similar to the light-activated phototransduction cascade in the rod and cone photoreceptors. It has further been suggested that upon activation, mGluR6 can stimulate a transducin-like G-protein that activates a cGMP phosphodiesterase (10,11). The resulting decrease in intracellular concentrations of cGMP leads to the closure of cGMP-gated, cation-selective ion channels and results in hyperpolarization of ON-bipolar cells (12)(13)(14). Light, which hyperpolarizes the photoreceptors, transiently depolarizes the ON-bipolar cells. Thus, this metabotropic receptor, mGluR6, is likely the only metabotropic receptor that directly mediates synaptic transmission.
Although there is considerable evidence to support the involvement of an mGluR (mGluR6) in mediating the dark, hyperpolarizing response of ON-bipolar cells, there is little evidence supporting the idea that the components of the mGluR6linked second messenger pathway are similar to the proteins involved in the cGMP cascade found in photoreceptors. Using genetically engineered mice, it has recently been demonstrated that mGluR6 mediates the synaptic effects of glutamate in both rod and cone ON-bipolar cells and is essential for normal visual processing (15,16). MGluR6 was originally cloned from a rat retinal cDNA library and expressed in Chinese hamster ovary cells (17). In this heterologous expression system the activation of mGluR6 inhibited forskolin-stimulated adenyl cyclase with no indication of involvement in a cGMP cascade (17). A similar observation has been made with the human homolog, hmGluR6 (18). An alternate approach to attempt to link mGluR6 to a cGMP biochemical cascade has been to probe bipolar cells with immunological probes. No cross-reactivity was observed in vertebrate bipolar cells with antibodies to transducin, cGMP phosphodiesterase, cGMP-gated channel, or arrestin, which are some of the proteins involved in mediating the cGMP cascade of rod photoreceptors, suggesting that the proteins involved in the second messenger system in bipolar cells are not identical with those expressed in rod photoreceptors (19,20). A recent physiological study demonstrated that the cGMP-gated channel expressed in cat bipolar cells has properties different from the cyclic nucleotide-gated channels found in either photoreceptors or olfactory receptors (14), suggesting that a different class of cGMP-gated channels may be expressed in ON-bipolar cells. Furthermore, Noga et al. found that a G o -specific antibody cross-reacted with rod bipolar cells, suggesting that mGluR6 may activate a G o -like G-protein (19). Additional experiments are needed to identify the second messenger system that links mGluR6 activation to hyperpolarization of retinal ON-bipolar cells.
To address the coupling of mGluR6 to a G-protein, we examined it in an in vitro reconstitution system. We cloned and modified the human mGluR6, expressed it in COS-7 cells, purified it, and assayed its function in an in vitro GTP binding assay. Our results demonstrate that mGluR6 can activate both transducin and G o in an agonist-dependent fashion. The rate of G o activation was 18-fold greater than the rate of transducin activation, suggesting that a G o type G-protein may be activated by mGluR6 in vivo.

EXPERIMENTAL PROCEDURES
Materials-All reagents, except where indicated, were purchased from Sigma. L-AP4 was purchased from Tocris Neuramin. [ 35 S]GTP␥S was from NEN, and nonradiolabeled GTP␥S (tetra lithium salt) was from Boehringer Mannheim. LipofectAMINE was from Life Technologies, Inc. G o was obtained from Calbiochem. Frozen bovine retinas were obtained from Schenk Packing Company (Stanwood, WA). The monoclonal antibody rhodopsin 1D4, which is specific for the C terminus of rhodopsin has been previously described (21). Peptide I (DEASTTVSK-TETSQVAPA) was purchased from the American Peptide Co., Inc.
Twelve partial hmGluR6 cDNA clones were obtained (METAB65-METAB76) from this screen. Sequence analysis of these cDNAs revealed that METAB75 was the largest partial mGluR6 cDNA (nt 794 -3105) isolated. To obtain the 5Ј portion of human mGluR6, a specifically primed human retinal gt10 cDNA library was constructed using an antisense mGluR6 oligonucleotide (nt 1083-1058) to prime human retinal poly(A) ϩ RNA essentially as described by Gubler and Hoffman (23) and Lapeyre and Amalric (24). Approximately 1.3 ϫ 10 6 recombinants from the gt10 specifically primed library were screened with a 582base pair SmaI fragment (nt 820 -1401) from METAB75 at the hybridization and washing stringencies described above for the retinal library screening. Twenty hybridizing plaques were identified in this screen and eight putative human mGluR6 clones (METAB77 to METAB85) were isolated and characterized. The full-length pCMV-2(ϪSA/SD)hm-GluR6 construct was prepared using the METAB85 (nt Ϫ84 to 1083) and METAB75 (nt 794 -3105) cDNAs. Initially, 1.1-and 2.3-kilobase EcoRI fragments isolated from METAB85 and METAB75, respectively, were subcloned into the pGEM7Z vector (Promega, Madison, WI). An FIG. 1. Analysis of the purified hmGluR6 by silver stain and immunoblot. A, silver stain analysis of proteins in a COS-7 cell membrane preparation after transfection with the modified vector pmt2 containing the modified hmGluR6 gene and subsequent purification of hmGluR6 using immunoaffinity chromatography. Lane 1 contains 5 l of solubilized membranes from a 600-l membrane preparation prepared from six 10-cm tissue culture plates with cells at a density of 5 ϫ 10 6 cells/plate. Lane 2 is 20 l of a 300-l lipid vesicle-receptor preparation that has receptor purified from six tissue culture plates as described above. These samples were analyzed on a 15% SDS/PAGE gel. B, immunoblot analysis of purified hmGluR6 and bovine rhodopsin. Lane 1 is 20 l of a purified receptor-vesicle preparation, and the control in lane 2 is 15 pmol of bovine rhodopsin. The proteins were initially separated on a 10% SDS/PAGE gel.

FIG. 2. L-AP4 activated hmGluR6 activates bovine transducin.
A, the time course for the binding of GTP␥S to transducin as catalyzed by L-AP4-activated hmGluR6 purified from transfected COS cells and reconstituted into lipid vesicles. The assay was performed as described under "Experimental Procedures." 6 mM L-AP4 (f) was used in this assay and compared with an identical reaction in the absence of agonist (q). The rate of 0.065 pmol of GTP␥S bound/min is the slope of the line. This rate is approximately 4-fold greater than the rate observed in the absence of agonist. The data are from one of six similar experiments (Table I). B, light-activated bovine rhodopsin activates bovine transducin. The time course for the binding of GTP␥S to transducin as catalyzed by light-activated bovine rhodopsin purified from transfected COS cells and reconstituted into lipid vesicles. The assay was performed as described under "Experimental Procedures." Three aliquots were assayed in the dark (q) at 30-s intervals. The remaining reaction was exposed to room light for 30 s, and five additional aliquots were removed and assayed (f). The light-activated rate of 0.73 pmol of GTP␥S bound/ min is the slope of the line. This rate is approximately 12-fold greater than the rate observed in the dark. The data are from one of three similar experiments.
Modification of the hmGluR6 Clone-The cDNA for the human mGluR6 was modified by the addition to the C terminus of 15 amino acids (STTVSKTETSQVAPA) corresponding to the epitope for the rhodopsin monoclonal 1D4. This epitope tag has been used with other transmembrane proteins (26,27). This modified hmGluR6 cDNA was subcloned into a modified version of the eukaryotic expression vector pMT-2 (28,29).
Expression and Purification of hmGluR6 -The hmGluR6 clone was expressed in COS-7 cells following transfection with LipofectAMINE. Cells were harvested 48 h posttransfection for purification. The transfected cells were then solubilized in 50 mM Tris buffer, pH 7.0, containing 140 mM NaCl, 1 mM dithiothreitol, 1% CHAPS, and 10 mg/ml asolectin for 40 min at 4°C. Expressed proteins were purified by immunoaffinity chromatography using the bovine rhodopsin antibody, 1D4, as described by Oprian et al. (30). The purified receptor in a CHAPS asolectin solution was applied to a Sephadex G-50 column to remove the detergent and form lipid vesicles containing the purified receptor (31). The fractions containing the vesicles were pooled and concentrated by centrifugation in a Centricon-30. These vesicles were then assayed for activity, and the amount of expressed protein was estimated by Western blotting using the antibody 1D4 to visualize the receptor.
Expression and Purification of Bovine Rhodopsin-Wild type bovine opsin was expressed, purified, and reconstituted into asolectin vesicles as described above for hmGluR6. Vesicles containing wild type bovine opsin were then incubated with the chromophore 11-cis-retinal (0.2 mM) in the dark for at least 1 h.
Purification of Transducin-Transducin was purified from bovine retina according to the procedure of Wessling-Resnick and Johnson (32) and then subjected to ion exchange chromatography on DE-52 as described by Baehr et al. (33).
Assay for Activation of Transducin and G o -Purified hmGluR6 in lipid vesicles was assayed for its ability to catalytically activate transducin or G o by following the binding of [ 35 S]GTP␥S as has been previously described for rhodopsin (34). Purified hmGluR6 in lipid vesicles (final concentration, 2.6 ϫ 10 Ϫ8 M) was incubated with the indicated amounts of L-AP4 in reaction buffer containing 1 mM Tris, 10 mM NaCl, 0.5 mM MgCl 2 , and 10 M EDTA, pH 7.0, at 30°C for 30 min. The reaction was initiated by the addition of transducin or G o and GTP␥S resulting a final concentration of 1 M. Aliquots were assayed at indicated times by filtering through nitrocellulose filters. Rhodopsin activation of transducin was performed as described previously (34).
SDS/PAGE-SDS/PAGE was performed according to Laemmli using both 10 and 15% polyacrylamide gels (35). Proteins were visualized using silver staining according to the procedure of Bloom et al. (36).
Immunological Analysis-Immunoblot analysis was performed according to the methods of Burnette (37). Rhodopsin and the modified hmGluR6 were detected using the monoclonal antibody 1D4. Antibody binding was visualized using a radiolabeled secondary antibody that was visualized using a PhosphorImager (Molecular Dynamics).

RESULTS
A full-length human cDNA clone for mGluR6 was constructed from two overlapping cDNA fragments isolated from human retinal cDNA libraries. The percentage of identity of the human and rat mature mGluR6 protein is 94.8%, and the divergence is largely due to the divergence of the signal peptide. The four putative Asn glycosylation sites and two putative PKC phosphorylation sites are conserved between the human and rat sequences. The deduced amino acid sequence of hm-GluR6 reported here is 99.8% identical to the hmGLuR6 sequence reported by Laurie et al. (18). The only differences occur at Arg 59 3 Pro and Val 557 3 Ala (the sequence of Laurie et al. mutated to our sequence). The underlined residue is identical to those reported for the rat sequence (17).
Expression and Purification of hmGluR6 -The cDNA clone for hmGluR6 was modified with the addition of a 15-amino acid epitope tag at the C-terminal tail, which corresponds to the epitope for the rhodopsin monoclonal antibody 1D4 (21). The strategy of adding this epitope has been utilized previously for human color visual pigments and the ␤ 2 -adrenergic receptor (26,27). In these instances the additional amino acids did not alter the properties of the membrane proteins but allowed for easy purification using immunoaffinity chromatography (26,27).
The hmGluR6 protein was expressed in COS-7 cells and purified by immunoaffinity chromatography using the 1D4 antibody. Using this procedure a major protein band with an apparent molecular mass of approximately 103,000 Da is visualized on an SDS/PAGE gel using silver staining (Fig. 1A, lanes   FIG. 3. L-AP4 activated hmGluR6 activates rat G o . The time course for the binding of GTP␥S to G o as catalyzed by L-AP4-activated hmGluR6 purified from transfected COS cells and reconstituted into lipid vesicles is shown. The assay was performed as described under "Experimental Procedures." 6 mM L-AP4 (f) was used in this assay and compared with identical reactions in the absence of agonist (q) and with 10 M MPPG (OE), the Class III-specific antagonist. The rate of 2.1 pmol of GTP␥S bound/min is the slope of the line and represents the rate of activation in the presence of agonist. This rate is approximately 5-fold greater than the rate observed in the absence of agonist and 2.6-fold greater than the rate in the presence of the inhibitor MPPG. The data are from one of three similar experiments.

TABLE I The G-protein activation by hmGluR6 and rhodopsin
The initial rate of G-protein activation is indicated as pmol of GTP␥S bound/min and was determined as described in the Fig. 2 legend. There is an intrinsic activity, agonist-insensitive activity that has not been subtracted from these rates. The rates represent the means Ϯ S.E. of the indicated number of similar experiments. 1 and 2). Immunoblot analysis of a similar gel reveals proteins with apparent molecular masses of 50,000, 103,000, and 206,00 Da that cross-react with the rhodopsin monoclonal antibody 1D4 (Fig. 1B, lane 1). The 50-kDa band is most likely a proteolytic fragment and represents only 5% of the total protein visualized in this blot. A similar cleavage fragment has been observed with the in vitro expression of mGluR5 (38). The two major bands (103,000 and 206,00 kDa) appear to be the monomer and dimer of the same protein. The monomer has an apparent molecular mass that is approximately 5 kDa greater than that predicted from the amino acid sequence of the gene and the additional epitope tag. This increase is probably due to posttranslational modifications such as glycosylation as has been shown for other mGluRs (39). Rhodopsin was expressed and purified using a similar procedure and a protein of approximately 40 kDa and a dimer of 80 kDa cross-reacted with the 1D4 antibody (Fig. 1B, lane 2). The reason for the apparent dimerization of these G-protein-coupled receptors observed in SDS/PAGE gels is not known, but it is observed with rhodopsin (40) and other mGluRs (25,(41)(42)(43).

Activation of Transducin and G O by hmGluR6 in the Presence of L-AP4 -
The functional coupling of G-protein-coupled receptors can be examined in functional reconstituted systems with the receptor and the G-protein using a GTP␥S binding assay. GTP␥S binding requires activation of the G-protein by an activated receptor. We used this method to examine the coupling of the purified mGluR6 protein with transducin purified from bovine retina as we have previously used to examine the coupling of rhodopsin with transducin. Using this preparation, we observed a time-dependent activation of transducin following the addition of 6 mM L-AP4 shown by a time-dependent increase in GTP␥S binding ( Fig. 2A). The average activation observed with 6 mM L-AP4 was 4-fold (n ϭ 6). In control experiments in the absence of mGluR6 protein, L-AP4 does not activate transducin. Furthermore, we observed no GTP␥S binding in the absence of transducin (data not shown). The EC 50 for L-AP4 activation of mGluR6 as measured by transducin activation was 8.4 mM.
To understand the level of activation of transducin by mGluR6 we observed in this reconstitution system, we examined the activation of transducin by rhodopsin purified in the same manner. We observed that transducin activation, determined by the rate of GTP␥S binding, by light-activated wild type bovine rhodopsin expressed in COS and reconstituted in lipid vesicles is 14-fold greater than the rate observed when transducin is activated by mGluR6 under similar conditions (Fig. 2B). These results indicate that coupling of hmGluR6 to transducin is weak when compared with the activation observed with rhodopsin. This suggests that another G-protein may be the preferred substrate for hmGluR6.
An immunological study by Noga et al. (19) has suggested that ON-bipolar cells do not contain transducin but do include a G o -type G-protein. To further investigate the activation of hmGluR6 in this in vitro reconstitution system, we examined the ability of hmGluR6 to activate rat G o . We observed that G o was activated by hmGluR6 in the presence of the agonist L-AP4 (Fig. 3) and that the rate and extent of binding of GTP␥S to G o in the presence of activated hmGluR6 is much greater than the rate observed with transducin (ϳ18-fold greater under comparable conditions; Table I). The EC 50 for L-AP4 activation of mGluR6 as measured by G o activation was 3.1 mM. Because the in vivo agonist for hmGluR6 is most likely glutamate, the ability of glutamate-activated mGluR6 to activate G o was also tested. A 3-fold increase in GTP␥S binding was observed in the presence of 10 mM glutamate. In contrast, the activation of transducin under identical conditions was only 1.4-fold (data not shown). To further investigate the specificity of the coupling of hmGluR6 to G o in this reconstitution system, we examined the ability of the Class III-specific mGluR antagonist MPPG (44) to inhibit the L-AP4-stimulated GTP␥S binding. Preincubation with 10 M MPPG in the binding assay inhibited the L-AP4-stimulated GTP␥S binding by 42% (n ϭ 3, Fig. 3 and Table II).
In summary, the hmGluR6 protein has been expressed in vitro, purified, and functionally reconstituted into an in vitro G-protein activation assay. We observed activation of two Gproteins, transducin and G o , by agonist activated hmGluR6; the activation of G o was 18-fold greater than activation of transducin.

DISCUSSION
In rodents, mGluR6 is nearly exclusively expressed in ONbipolar (15). Activation of this G-protein-coupled receptor by glutamate leads to the activation of a biochemical cascade that results in the hyperpolarization of ON-bipolar cells (8,9). Electrophysiological and pharmacological studies have implicated a cGMP biochemical cascade in this signaling pathway (10,11). The identity of the specific G-protein activated in this system as well as the effector enzyme have yet to be conclusively identified. In an attempt to address this problem, we have cloned, expressed, purified, and reconstituted the hmGlur6 protein in an in vitro system where the coupling of the receptor to different G-proteins can be explored.
We found that hmGluR6 protein expressed in vitro and reconstituted into lipid vesicles can functionally couple to two known G-proteins. When activated by the agonist L-AP4, hm-GluR6 can couple to both transducin and a G o . Activation was measured by the binding of GTP␥S to the G-protein. The rate of activation of G o is 18-fold greater than the rate of activation of transducin. Glutamate can also activate hmGluR6 in this reconstituted system and agonist (L-AP4) activation is inhibited by the Class III-specific antagonist MPPG. These results indicate that we are measuring functional coupling of hmGluR6 to both G o and transducin. This is the first report of the purification and functional coupling of mGluR6 to a G-protein, or any glutamate metabotropic receptor, in an in vitro reconstitution system.
The results reported here do not identify the G-protein that actually couples to the hmGluR6 in the ON-bipolar cells. They suggest that the specificity of hmGluR6 for G o is greater than it is for transducin. This is consistent with the finding that antibodies to G o but not transducin localize to ON-bipolar cells (19). The data presented here suggest that hmGluR6 can also activate transducin. Both G o and transducin are members of the same family of G-proteins (45). It is possible that the G-protein in ON-bipolar cells has characteristics of both G o and transducin. Alternatively, there may be multiple G-proteins activated by mGluR6 in vivo. The second messanger system  Fig. 3. All assays were performed in the presence of 6 mM L-AP4. There is an intrinsic activity, agonist-independent activity, that is 20 -30% of stimulated activity. This has not been subtracted from any of the rates. The percentage of activation was calculated in each experiment. Data represent the means Ϯ S.E. of three similar experiments.
[MPPG] Activity (n ϭ 3) M % 0 100 10 58 Ϯ 5.9 30 50 Ϯ 1.7 1000 37 Ϯ 2.1 mediating the glutamate-dependent hyperpolarization in ONbipolar cells remains to elucidated. Effectors enzymes activated by G o are now possible candidates for involvement in this second messanger system. Transducin is not known to be a promiscuous G-protein. It is not activated by many G-protein-coupled receptors and seems to retain some degree of specificity for visual opsins. It can be activated by the ␣ 2 -adrenergic receptor (46). The results reported here are important because they suggest that the hm-GluR6 can activate transducin, the rod-specific G-protein. Because hmGluR6 is a G-protein-coupled receptor that does not have substantial sequence homology to opsin, these results suggest a novel coupling of a non-opsin G-protein-coupled receptor to transducin.