Role of the Second and Third Intracellular Loops of Metabotropic Glutamate Receptors in Mediating Dual Signal Transduction Activation*

On the basis of sequence homology and structural similarities, metabotropic glutamate receptors (mGluRs), extracellular Ca2+-sensing receptor, γ-aminobutyric acid type B receptor, and pheromone receptors are enlisted in a distinct family within the larger G protein-coupled receptor superfamily. When expressed in heterologous systems, group I mGluRs can activate dual signal transduction pathways, phosphoinositides turnover and cAMP production. To investigate the structural basis of these coupling properties, we introduced single amino acid substitutions within the second and third intracellular loops (i2 and i3) of mGluR1α. Wild-type and mutant receptors were expressed in human embryonic kidney 293 cells and analyzed for their capacity to stimulate both signaling cascades. Each domain appeared to be critical for the coupling to phospholipase C and adenylyl cyclase. Within i2, Thr695, Lys697, and Ser702 were found to be selectively involved in the interaction with Gq class α subunit(s), whereas mutation of Pro698 and the deletion Cys694-Thr695 affected only Gscoupling. Furthermore, the mutation K690A profoundly altered mGluR1α signaling properties and imparted to the receptor the ability to couple to the inhibitory cAMP pathway. Within i3, we uncovered two residues, Arg775 and Phe781, that are crucial for coupling to both pathways, since their substitution leads to receptor inactivation.

Metabotropic glutamate receptors (mGluRs) 1 are coupled to heterotrimeric G proteins, and through this interaction they regulate the intracellular level of second messenger molecules such as inositol trisphosphate and cyclic AMP (cAMP). Furthermore, they have been shown to modulate the activity of voltage-sensitive Ca 2ϩ and K ϩ channels, G protein-regulated inward rectifier K ϩ channels, as well as GABA A , ␣-amino-3hydroxy-5-methylisoxazole-4-propionate, and N-methyl-D-aspartate receptors (1)(2). Thus far, eight mGluRs (mGluR1 through mGluR8; see Refs. 3-10) have been cloned. They have been categorized into three groups according to their sequence homology, agonist selectivity, and main signal transduction pathway activated in heterologous systems (1,(11)(12). Group I mGluRs (mGluR1 and -5) mobilize intracellular Ca 2ϩ by stimulating phosphoinositide (PI) turnover and promote cAMP accumulation; group II (mGluR2 and 3) and group III (mGluR4, -6 -8) receptors inhibit adenylyl cyclase.
mGluRs, together with the parathyroid Ca 2ϩ -sensing receptors (PCaR1 (13)) and the GABA B receptor (14), form a separate family within the seven transmembrane domain G protein-coupled receptor (GPCR) superfamily (15)(16)(17). mGluRs are characterized by a large amino-terminal extracellular domain that comprises the glutamate binding site (18 -19). The intracellular loops connecting the putative membrane-spanning helices are also distinctive. In mGluRs, they are relatively small compared with those of seven transmembrane domain receptors belonging to the rhodopsin/␤-adrenergic family. Most importantly, there is no significant sequence homology between members of the mGluR family and other cloned GPCRs.
The structure-function relationships supporting the coupling to G proteins have been well investigated for members of the rhodopsin/␤-adrenergic family. Although all the cytoplasmic domains of these receptors take part in G protein activation to some degree, i3 appears to harbor the structural elements that impart specificity to the interaction (20 -21).
However, much less is known about the structural determinants of the coupling of mGluRs to G proteins. Recently, the analysis of chimeric receptors derived from the G i -coupled mGluR3 bearing different portions of the cytoplasmic domains of the G q -coupled mGluR1 has shown that i2 of group I receptors is necessary, but not sufficient, for the specific activation of phospholipase C (PLC) and that both i3 and the cytoplasmic tail of the receptor appear to be also necessary for efficient coupling to this pathway (22)(23). The finding that a point mutation in i3 of the human PCaR1 is responsible for loss of receptor function (24) also suggests that this domain could be important for mGluR-G protein interaction.
In heterologous systems, group I mGluRs have been shown to stimulate cAMP accumulation (25)(26). However, the domains of the receptor and the signaling partners involved in the activation of this pathway have not been characterized.
In this work, we introduced single amino acid substitutions within i2 and i3 of mGluR1 to identify specific residues that contribute to the interaction with G proteins. To gain insights * This work was supported in part by National Institutes of Health Grant EY09534 and by the Samuel and May Rudin Foundation. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18  on the structural elements involved in selective interaction with either pathway, we tested the mutant receptors for their ability to induce PI turnover and cAMP accumulation. Here we show that both i2 and i3 play a role in determining selective coupling to G s and G q class ␣ subunits and that mutations of single residues differentially affect the activation of the two transduction pathways.

EXPERIMENTAL PROCEDURES
Site-directed Mutagenesis-Site-directed mutagenesis was carried out by a modification of the polymerase chain reaction (PCR) technique as described by Nelson and Long (27). The oligonucleotides required for the mutagenesis (Ransom Hill Bioscience, Ramona, CA; Table I) consisted of the following: (a) forward mutagenic primers containing single/ multiple base mismatches; (b) reverse hybrid primer serving as 3Јanchor, harboring at the 5Ј-end 20 nucleotides randomly selected and unique and not complementary to the mGluR1␣ sequence and, at the 3Ј-end, 21 complementary nucleotides; (c) forward 5Ј-anchor primer; (d) reverse complementary hybrid primer (c-primer) identical to the random 20-mer. All polymerase chain reaction reactions were carried out in a 50-l reaction volume containing 200 M deoxynucleotides and 2.5 units of Taq polymerase (Boehringer Mannheim). In the Step 1 polymerase chain reaction reaction, 50 pmol each of mutagenic and hybrid primer were used with the template pRc-mGluR1␣ (pRc-CMV vector; Invitrogen, San Diego, CA) to yield an amplified product comprising the mutation and the unique sequence.
Step 1 was 30 cycles of 1 min at 94°C denaturation, 3 min at 37°C annealing, 3 min at 74°C elongation. The amplified DNA was fractionated on agarose gel, purified, and used together with pRC-mGluR1␣ in the Step 2 reaction. Step 2 was a single cycle of 5 min at 94°C, 2 min at 37°C, and 10 min at 74°C. At Step 2 completion, 50 pmol each of the 5Ј-anchor and c-hybrid primer was added, and 30 additional cycles were completed (Step 3).
Step 3 products were digested with BglII and SphI for all mutants but i2* (Table I), which were digested with NcoI and SphI and cloned into pBS-mGluR1␣ (pBlueScriptSK(Ϫ) vector; Stratagene, San Diego, CA). The sequence of the amplified DNAs was determined by the Sanger method (Sequenase, U. S. Biochemical Corp./Amersham Life Science, Inc.). Mutant receptors were subcloned from the pBlueScript construct into pRc-CMV by digestion with SacII and ApaI (i2 and i2* mutants) or SacII and BspEI (i3 mutants).
Expression in HEK 293 Cells-HEK 293 cells were cultured in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum (FBS). Cells were transfected by electroporation with an Electro Cell Manipulator (BTX, San Diego, CA) using 4-mm gap cuvettes. The DNA used in transfection experiments was purified with Qiagen columns (Qiagen, Chatsworth, CA) according to the procedure recommended by the manufacturer. In a standard experiment, cells were harvested by trypsinization and washed 3 times with PBS; for each electroporation 5 or 10 million cells were used with either 10 or 20 g of DNA (according to the purpose of the experiment) in 0.3-ml final volume. Electroporations were conducted at 260 V, 950 microfarads, 72 ⍀, yielding 20 ms as average time constant. The transfection efficiency, about 30%, was determined by co-transfection (1:1 ratio) with the pRc-lacZ vector and subsequent chromogenic reaction. All the tissue culture reagents were purchased from Life Technologies Inc. HEK 293 were generously provided by Dr. M. Chao at Cornell University Medical College.
Inositol Phosphates Accumulation-Following electroporation, cells were left to recover in DMEM supplemented with 10% dialyzed FBS and 2 mM GlutaMAX TM (Life Technologies Inc.) for 3-5 h and metabolically labeled for approximately 16 h with 1 Ci ml Ϫ1 [ 3 H]myo-inositol (NEN Life Science Products) in inositol-free DMEM. Cells were washed with PBS and incubated for 2 h with 2 mM pyruvate and 2 units/ml glutamic-pyruvic transaminase (Sigma) (28). After washing with PBS, cells were incubated for 10 min in Hanks' balanced saline solution containing 10 mM LiCl; 1 mM Glu (Sigma) was then added for 20 min. Agonist stimulation was stopped by placing the cells on ice and adding 5% ice-cold trichloroacetic acid. Separation of total [ 3 H]inositol phosphates was carried out essentially as described (29) by chromatography on AG 1-X8 anion-exchange resin (Bio-Rad); total inositol phosphates were eluted with 700 mM ammonium formate, 100 mM formic acid.
In Vivo cAMP Accumulation-Three hours after electroporation, cells were metabolically labeled for approximately 16 h with 1 Ci ml Ϫ1 [ 3 H]adenine (NEN Life Science Products) in DMEM supplemented with 10% dialyzed FBS and 2 mM GlutaMAX TM . Cells were then washed with PBS, treated with glutamic-pyruvic transaminase for 2 h (see above), and preincubated for 30 min with Hanks' balanced saline solution containing 1 mM 3-isobutyl-1-methylxanthine (Sigma). 1 mM Glu in the presence of 3-isobutyl-1-methylxanthine was then added for 20 min. Cells were lysed with ice-cold 5% trichloroacetic acid, and adenine nucleotides were fractionated according to the method of Salomon (30).
Alternatively, after stimulation with Glu, unlabeled cells were lysed with 65% ethanol, 4 mM EDTA, and cAMP concentration was determined by the cyclic AMP 3 H assay system (Amersham Life Science Inc.) a i2* indicates mutants that have been subcloned with different sets of restriction enzymes (see "Experimental Procedures"). b Mismatches introduced in the sequence to direct amino acid substitutions are underlined.
according to the procedure recommended by the manufacturer. Immunofluorescence and Immunoblot Analysis-For immunofluorescence studies, electroporated cells were grown for up to 18 h in serumcontaining medium on gelatin-coated coverslips to facilitate adhesion. Cells were then washed with PBS containing 2 mM MgCl 2 , fixed for 2 min with ice-cold methanol, and permeabilized with 0.4% Triton X-100. The primary antibody M5-14 made in rabbit and directed to the carboxyl terminus of the rat mGluR1␣ (31) was diluted 1:400 and reacted with a fluorescein isothiocyanate-conjugated goat anti-rabbit secondary antibody (Jackson ImmunoResearch Laboratories, West Grove, PA). Coverslips were washed, mounted on glass slides with Vectashield (Vector Laboratories, Burlingame, CA), and analyzed using a Sarastro Phoibos 1000 confocal laser scanning system attached to a Nikon microscope.
For immunoblot analysis, transfected cells were washed and incubated on ice with pre-chilled PBS containing 5 mM EDTA for 20 min. The cells were then scraped off the dish with a rubber policeman, collected by centrifugation, and lysed in alkaline sample buffer (0.12 M Tris base, pH 9.5, 3.3% sodium dodecyl sulfate, 10% glycerol, 3.1% dithiothreitol, and 0.004% bromphenol blue). Total proteins were quantified with the Bio-Rad Protein Assay System (Bio-Rad) according to the procedure recommended by the manufacturer. Prior to 6% polyacrylamide gel electrophoresis separation, the samples were heated at 60°C for 3 min. Semi-dry protein transfer on polyvinylidene difluoride membrane (Micron Separations, Westboro, MA) was performed as indicated by the manufacturer. The blots were then blocked with 10% non-fat dry milk in Tris-buffered saline (TBS: 100 mM Tris-Cl, pH 7.5, 0.9% NaCl) at room temperature for 1 h, washed three times with TBS-0.05% Tween (TBS-T), and incubated overnight at 4°C with the polyclonal antibody M5-14 diluted 1:2500 in TBS-5% dry milk. After washing with TBS-T, the membranes were incubated with a horseradish peroxidaseconjugated goat anti-rabbit antibody (Jackson ImmunoResearch Laboratories) diluted 1:1000. Chemiluminescence detection was carried out with the ECL TM Detection System (Amersham Life Science) following the protocol indicated by the manufacturer, and the membranes were quickly exposed for autoradiography for 30 s, 5 min, and 1 h. Autoradiographies exposed for 5 min, all within the linear range of signal intensity, were used for the densitometric quantification. Following immunodetection, the blots were stained with india ink (32) to allow relative quantification of the transferred proteins. Images of autora-diographies and membranes were acquired with an Arcus II scanner (AGFA), and densitometric analysis was performed with the software package NIH-Image 1.60. 2

RESULTS
Receptors belonging to the rhodopsin/␤-adrenergic family are characterized by the presence of short amphipathic ␣-helices at the amino and carboxyl termini of their third intracellular loop. Amino acids within these microdomains play a key role in determining selective coupling to heterotrimeric G proteins (20 -21). Unlike these GPCRs, mGluR1 i3 is small, nonhomologous, and well conserved among all members of the family, despite their different coupling properties (Fig. 1a).
To evaluate the contribution of single amino acid residues in i3 of mGluR1␣ to G protein coupling, we mutagenized Arg 775 (R775W and R775K), corresponding to Arg 796 in PCaR1; Pro 778 (P778Q); Phe 781 (F781P and F781S); Asn 782 (N782I); and Glu 783 (E783Q). These residues are conserved among all metabotropic and calcium-sensing receptors and were substituted with amino acids designed to change the charge distribution of the loop (Fig. 1b).
The only residue in i3 that distinguishes receptors coupled to different signal transduction pathways is Ala 779 , which is conserved in group I mGluRs, whereas Glu is present in groups II and III receptors. Since this seems to suggest that the amino acid at this position could participate in coupling selectivity, we replaced Ala 779 with Glu (A779E).
Within i2 (the largest intracellular loop in mGluRs) only few amino acids are in common between groups I and II/III receptors (Fig. 1c). The regions of homology are concentrated at the amino and carboxyl termini of the domain, which are predicted to adopt ␣-amphipathic conformation. The intervening stretch 2 Developed at the National Institutes of Health and available by anonymous ftp from zippy.nimh.nih.gov. of residues is highly hydrophilic and therefore constitutes a potential protein interaction site. Previous studies conducted with chimeric receptors showed that the 16-amino acid region from Lys 691 to Gln 706 is required to impart the ability to activate PLC to G i -coupled mGluRs (22)(23). We mutagenized residues within this region that, according to sequence alignment, are conserved among all mGluRs (Lys 690 , Pro 698 , and Ser 702 ) and residues that are shared only by PLC-coupled receptors (Lys 692 , Thr 695 , Arg 696 , Lys 697 , and Ala 705 ) (Fig. 1c). Among them, the positively charged amino acids that are part of the central hydrophilic region of i2 were substituted with hydrophobic residues to perturb the charge distribution. Group I mGluRs also differ from group II and III in the size of i2 (26 versus 24 amino acids). To test whether such a difference is structurally relevant for G protein coupling, we generated two deletion mutants. In the Cys 694 -Thr 695 mutant, Cys 694 and Thr 695 , which are shared only by group I receptors and are located in the center of i2, were deleted. In the K692G/Cys 694 -Thr 695 mutant, the same deletion was accompanied by the substitution of Lys 692 to Gly; this substitution reduces the number of positively charged residues located in the amino terminus of i2, thus increasing the structural similarity of the mutant mGluR1 with group II and III receptors.
Receptor-mediated Stimulation of PI Turnover-To assess the ability of the mutant receptors to activate the PLC signaling cascade, we measured the accumulation of inositol phosphates (IPs) upon stimulation with Glu in HEK 293 cells. Wild-type receptor activation induces a 3-4-fold increase of IP production compared with basal level (not-stimulated cells); the specificity of this response is confirmed by the absence of effect of Glu in cells transfected with the reporter construct pRc-lacZ alone (Fig. 2a). Consistent with recent reports, mGluR1␣ expression generates a high basal level of inositol phosphates; this effect, distinctive to the long splice variant of the receptor, has been attributed by other investigators to intrinsic (agonist-independent) activity (28).
In HEK 293 cells, mGluR1␣-mediated stimulation of the PLC pathway is insensitive to pertussis toxin (PTx) (data not shown). In contrast, both in Xenopus oocytes and Chinese hamster ovary (CHO) cells, the receptor activates the same signaling pathway through PTx-sensitive and -insensitive G proteins, G o and G 11 , respectively (25,(33)(34). Since HEK 293 cells do not express G o (35), PLC stimulation is likely to depend solely on the interaction of the receptor with ␣ subunits of the G q family.
Upon analysis of receptors mutated in i3, we observed that the Glu-stimulated IP accumulation is suppressed in the R775W, F781S, and F781P mutants (Fig. 2a) and significantly reduced in the P778Q. The R775W substitution introduced in rat mGluR1 corresponds to the R796W mutation in the human PCaR1 linked to familial hypocalciuric hypercalcemia (24); such mutation causes loss of function of the human receptor. In an analogous fashion, the same substitution in mGluR1␣ leads to loss of the Glu-dependent PLC stimulation but does not affect the high basal activity of the receptor. Notably, substitution of Arg 775 with Lys (R775K) produces the opposite effect; this mutation generates a receptor that retains wild-type-like Glu-dependent activity but lacks intrinsic activity.
Substitution of Phe 781 with either a polar (F781S) or a hydrophobic residue (F781P) completely blocks receptor-medi- ated activation of the PLC pathway, both Glu-dependent and constitutive. In contrast, the mutations A779E, N782I, and E783Q do not alter significantly the Glu-dependent activity of the receptor. However, both the N782I and the E783Q mutants do not possess agonist-independent basal activity.
i2 mutant receptors carrying the substitutions T695A, R696G, P698R, S702T, and A705S do not differ from wild type in stimulating PI turnover (Fig. 2b), although a significantly higher level of basal activity is observed in the P698R mutant. On the other hand, the mutations K690A, T695E, K697M, and S702R cause a significant decrease in Glu-stimulated IP production. In addition, with the exception of K690A, these receptors show a significantly reduced intrinsic activity. None of the mutations introduced in i2 results in a nonfunctional receptor.
Receptor-mediated Stimulation of the cAMP Pathway-Wildtype and mutant receptors were tested for their ability to induce cAMP accumulation in HEK 293 cells. The endogenous adenylyl cyclase isoform in this cell line has not been identified yet; however, it is known to be G s and forskolin-activated but not sensitive to Ca 2ϩ , protein kinase C, or ␤␥ stimulation (36). Upon stimulation with Glu, mGluR1␣ induces a small but significant increase in cAMP (Fig. 3, a and b). The properties of the endogenous cyclase make it likely that such effect is generated through direct activation of G s , although with low efficiency. This was ascertained by co-transfecting the wild-type mGluR1␣ together with increasing amount of the cDNA coding for G s , which resulted in a proportional increase of Glu-induced mGluR1␣-mediated cAMP accumulation (Fig. 3a).
Next, we tested for cAMP accumulation those i2 mutants that are affected in their ability to stimulate PI turnover. The parallel analysis of two separate signaling cascades should help determine whether the mutations affect selectively the coupling to the G q class ␣ subunits or whether they cause a general decrease in the affinity of the activated receptor for G proteins. All the mutants tested, except K690A, have activities comparable with wild type (Fig. 3b). Remarkably, the K690A mutant shows a much increased basal level and Glu-dependent stimulation of cAMP; this is likely to result from an increased affinity for G s .
However, since the level of cAMP production is generally quite small, it is possible that some change in the efficiency of coupling could go undetected. To circumvent this possible problem we have increased the basal level of cyclase to amplify the receptor response. This was achieved by co-transfecting HEK 293 cells with type II adenylyl cyclase together with wild-type and mutant receptors. The type II adenylyl cyclase isoform was chosen because it is not Ca 2ϩ -regulated and therefore not sensitive to variations in intracellular Ca 2ϩ induced by receptormediated activation of the PLC pathway (37). In this experimental paradigm, stimulation of the wild-type receptor with Glu results in about 4-fold increase of cAMP accumulation (Fig.  4a). Consistent with our previous observation, the K690A mutant has greater activity than wild type (about 8-fold). All the other mutations in i2 do not affect receptor activity, with the exception of the Cys 694 -Thr 695 and P698R mutants which are less efficient than wild type (only 2-fold stimulation).
Among receptors mutated in i3 (Fig. 4b), substitution of Phe 781 with either Pro or Ser (F781P; F781S) causes receptor loss of function; in addition, the substitutions R775W and N782I also affect receptor activity, resulting in a decreased efficiency in the stimulation of the cAMP pathway. All other i3 mutants do not differ from the wild-type receptor.
Properties of the K690A Mutant-From the characterization of the K690A mutant activity profile, it became clear that Lys 690 is critical for coupling to both G q and G s . In particular, in the K690A mutant the interaction with G q is weakened, whereas that with G s is strengthened compared with wild type.
It has been suggested that mGluR1␣ could interact with PTx-sensitive G protein(s) coupled to inhibition of adenylyl cyclase in CHO cells (25). Since the K690A substitution apparently changes the affinity of the receptor for at least two different G proteins, we examined the possibility that the same mutation could enable the receptor to interact with members of the G i family. To this end, we co-transfected the ␤ 2 -adrenergic receptor (20) with either mGluR1␣, mGluR2, or K690A cDNAs in HEK 293 cells. ␤ 2 -Adrenergic receptor potently stimulates adenylyl cyclase via coupling to G s , allowing any inhibitory action of the mGluRs to be more easily detected. Expression of mGluR2, which is used as control for effective inhibition of adenylyl cyclase (6), induces in this experimental paradigm about 30% inhibition of cAMP accumulation (Fig. 5a). Although mGluR1␣ has no effect, the K690A mutant is able to inhibit cAMP accumulation with an efficiency comparable to mGluR2 (Fig. 5a). This effect is blocked by preincubating the transfected cells with PTx before stimulation with Glu (Fig. 5b), whereas no significant change is observed in the wild-type receptor (data not shown). This finding indicates a direct coupling of the K690A mutant to inhibitory G proteins.
Expression Level and Localization of mGluR1␣ and Mutant Receptors-To verify that the effects brought about by the amino acid substitutions on receptor effector-function are not due to changes in the synthesis and/or localization of the protein, we analyzed the expression level and distribution of wildtype and mutant receptors, transfected in HEK 293 cells, by Western blot and immunofluorescence. In immunoblots, the polyclonal antibody M5-14 detects selectively the receptor protein, thus yielding a single band of 145 kDa estimated molecular mass, as predicted for mGluR1␣ (Fig. 6a). The expression level of the receptor increases in linear fashion with the amount of DNA introduced into the cells. To detect variations in the syntheses of wild-type and mutant receptors, we transfected HEK 293 cells with an adequate amount of the corresponding DNAs (2.5 g) to obtain a level of protein expression within the linear range of increase (Fig. 6b). This analysis does not show any significant difference in the level of synthesis of the mutant receptors tested compared with the wild type.
The correct localization at the plasma membrane of mGluR1␣ and mutant receptors was validated by immunofluorescence combined with laser scanning confocal microscopy. Representative mutants are shown in Fig. 6c. None of the introduced substitutions or deletions appear to affect the sorting and overall structure of the receptor protein, as can be inferred by the correct targeting to the plasma membrane. mGluR1␣-like immunofluorescence is also detectable intracellularly in permeabilized preparations, probably in endosomes.

DISCUSSION
Physiological and pharmacological studies have shown that mGluRs activate multiple second messenger cascades, inhibit neurotransmitter release, and regulate K ϩ and Ca 2ϩ channel activity (1,38,39). These multi-signaling effects are thought to be mediated by multiple mGluRs. In addition, it is possible that some receptors have the capacity to interact with several different subsets of G protein ␣ subunits (3)(4)25). Furthermore, the associated ␤␥ subunits could also contribute to specific signaling by regulating channel opening (40), different isoforms of adenylyl cyclase (36,41), and PLC (42,43). Hence, the final output of mGluR activation in the central nervous system may also depend on the repertoire of signaling molecules available in the vicinity of the receptor. This could be particularly important in neurons, which can undergo activity-dependent changes in dendritic and axonal compartments.
In neurons, group I mGluRs have been shown to stimulate local intracellular calcium release via the PLC cascade and modulate other biochemical pathways and channel activities. This results from the stimulation of protein kinase C following the release of diacylglycerol. Although there is, thus far, no unequivocal demonstration for the direct involvement of group I mGluRs in the activation of the cAMP stimulatory pathway in vivo, evidence for the activation of the cAMP stimulatory pathway in cortical neurons following both group I and group II receptor activation has been presented (44 -45). This discrepancy with studies conducted on receptors expressed in heterologous systems could be due to the paucity of pharmacological tools capable of discriminating selectively group I receptors.
Here, we focused on the signaling properties of mGluR1␣ in a heterologous system, HEK 293 cells. In these cells we have demonstrated that the receptor activates at least two different signaling cascades, the PLC pathway (via interaction with G q ) and the stimulatory cAMP pathway (via interaction with G s ).
Recently several other GPCRs, such as ␣ 2 -adrenergic receptors, muscarinic receptors, and the parathyroid hormone (PTH)/ PTH-related peptide receptor, have been shown to possess dual or multiple effector functions. There is now accumulating evidence that this multi-signaling activity is due to coupling to different G protein ␣ subunits, sometimes leading to opposite effects, for example G s and G i as in the case of the ␣ 2 -adrenergic receptor (46). In several instances, the structural domains involved in differential coupling have been partially uncovered; however, it is not yet possible to draw general rules from these studies. For example, in the case of the PTH/PTH-related peptide receptor, some residues within the amino terminus of i3 appear to be independently involved either in PLC or adenylyl cyclase stimulation, whereas mutation of other residues affects both cascades (47). On the other hand, in the case of the ␣ 2 -adrenergic receptor, both i2 and the amino terminus of i3 are involved in stimulating cAMP production, and the deter- Shown is a typical assay performed in duplicate; similar results were obtained in another independent experiment. Error bars indicate the deviation from the mean. cAMP accumulation is expressed as the cAMP/ATPϩcAMP ratio. minants of coupling to G i are located in different regions of i3 (48,49).
Since the mGluRs bear no sequence homology with the other GPCR families, their study, together with that of other GPCRs, will unravel conserved structural features involved in coupling to G proteins. By mutating amino acids that are shared by all mGluRs, we have identified two residues in i3, Arg 775 and Phe 781 , that are required for productive coupling to both the PLC and stimulatory cAMP pathway. The finding that the R775W substitution can be "rescued" by replacing Arg 775 with another positively charged residue points to a requirement for a positive charge at this position. A more dramatic phenotype is obtained by substituting Phe 781 with either hydrophobic or polar residues; both mutations cause complete loss of function of the receptor. This finding suggests that Phe 781 is a key residue at the surface of interaction between receptor and G proteins. Together these observations indicate that i3 could play a pivotal role in providing a site for interactions with both G q and G s . On the other hand, two other mutants in i3 appear to operate independently in their ability to couple to G proteins FIG. 6. Expression level and localization of mGluR1␣, i2, and i3 mutant receptors in HEK 293 cells. a, Western blot analysis of total protein lysate (approximately 20 g per lane) from HEK 293 cells transfected with increasing amount of mGluR1␣ cDNA (the total DNA amount (20 g) used in the transfection was held constant with carrier). Lane 1, untransfected cells; lanes 2-6, 0.5, 1, 2.5, 5, and 10 g, respectively, of pRc-mGluR1␣. The inset shows the quantification of mGluR1␣ expression level; arbitrary units represent the ratio between the optical density of mGluR1␣-specific signal on autoradiographic film and that of the proteins transferred on the blot and visualized with india ink. b, relative expression level of wild-type, i2, and i3 mutant receptors as determined by Western blot analysis. 2.5 g of each mutant and wild-type DNAs were transfected in HEK 293 cells; total protein lysate (20 g) was used for immunoblot analysis. Quantification of the expression level was conducted as described above. Results represent mean Ϯ S.E. of at least two independent experiments. c, transfected cells were immunolabeled with an antibody directed against mGluR1␣ and then analyzed by confocal microscopy. Untransfected cells have a weak background fluorescence. as follows: Pro 778 seems to be important for a full Glu-dependent G q stimulation, and Asn 782 is required for a full activation of the stimulatory cAMP pathway.
Our study has uncovered several residues within i2 that are selectively involved in either the PLC pathway or the stimulatory cAMP pathway, suggesting the existence of two independent G protein-specific interaction domains. One example is the deletion of Cys 694 and Thr 695 . These residues are shared only by group I receptors and are located in the center of i2, between two predicted ␣-helical regions at the amino and carboxyl termini of this loop. They appear to be necessary to support coupling to G s but not to G q . Interestingly, the K690A substitution in i2 generates a receptor that is less effective than wild type in activating G q but more effective in activating G s . This suggests that Lys 690 could be located at a position critical for both G s and G q coupling. Searching for common motifs shared by mGluR1␣ i2 and G s -coupled GPCRs, we found an interesting homology between the amino terminus of i2 and the equivalent portion of the human ␣ 2A -adrenergic receptor (␣ 2 C10) i3, which is known to be required for coupling to G s (48). Sequence alignment shows that the K690A substitution increases the sequence similarity of mGluR1 with these receptors (Fig. 7).
Remarkably, the K690A mutation also confers to mGluR1␣ the ability to couple to G i . Thus, a single amino acid residue appears to act as a molecular switch for receptor-G protein coupling selectivity. A similar structural control of coupling specificity is mirrored in the ␣ subunit of G proteins; in fact, the four carboxyl-terminal amino acids of ␣ q , ␣ i2 , ␣ o , ␣ z , and ␣ s determine their coupling specificity (50), and only one of these four residues is different between ␣ o and ␣ i2 . An alternative explanation for the phenotype of the K690A mutant is that mGluR1␣ itself might be able to weakly stimulate G i and that the mutation enhances this interaction. This interpretation would be consistent with the observation by Aramori and Nakanishi (25) that the stimulation of cAMP accumulation induced by mGluR1␣ in CHO cells is potentiated by PTx.
Several mutations introduced in i3 and i2 of mGluR1␣ appeared to specifically affect either the constitutive or the agonistdependent activity of the receptor. As for several GPCRs, mGluR1␣ has recently been shown to be able to stimulate IP production in an agonist-independent fashion (28). This constitutive activity is thought to result from the ability of the receptor to adopt an active conformation in the absence of bound ligand. Our observations suggest that by altering the conformation of the G protein-coupling domains of the receptor, it is possibletospecificallyaffecteithertheconstitutiveortheagonistdependent activity. Similar mutants have been described for several other GPCRs and have led to the formulation of the "allosteric ternary complex" model (51), and the phenotype of the mutations described here can be accounted for by this model. The extension of this model to mGluRs is particularly significant because mGluRs differ from other GPCRs in their mode of ligand binding; indeed, the ligand binding site of mGluRs resides in the large extracellular domain (18 -19), whereas in most GPCRs it is located in the transmembrane region. Because inverse agonists are not yet available, the physiological relevance of the constitutive activity of mGluR1␣ in neurons cannot be assessed.
Taken together, these findings suggest that for mGluRs, as for other GPCRs, more than one domain is required to form the surface for the interaction with G protein ␣ subunits. This improved understanding of mGluR-G protein interaction will provide a valuable paradigm to analyze the coupling properties of GABA B receptors, the recently discovered pheromone receptors (52)(53), and possibly other related receptors yet to be discovered. FIG. 7. Sequence alignment of the first 11 amino acids of i3 of different ␣ 2 -adrenergic receptors (this domain is critical for coupling to G s ) with the first 11 residues of the i2 of mGluR1␣. Identical residues are boxed in black, whereas amino acids not identical but homologous are boxed in white. The arrow indicates Lys 690 ; the mutant receptor K690A shows increased affinity for G s .