A Novel Site on the G a -protein That Recognizes Heptahelical Receptors*

Specific domains of the G-protein a subunit have been shown to control coupling to heptahelical receptors. The extreme N and C termini and a region between a 4 and a 5 helices of the G-protein a subunit are known to determine selective interaction with the receptors. The metabotropic glutamate receptor 2 activated both mouse G a 15 and its human homologue G a 16 , whereas metabotropic glutamate receptor 8 activated G a 15 only. The extreme C-terminal 20 amino acid residues are identical between the G a 15 and G a 16 and are therefore un- likely to be involved in coupling selectivity. Our data reveal two regions on G a 16 that inhibit its coupling to metabotropic glutamate receptor 8. On a three-dimen-sional model, both regions are found in a close proximity to the extreme C terminus of G a 16 . One module com- prises a 4 helix, a 4 2b 6 loop (L9 Loop), b 6 sheet, and a 5 helix. The other, not described previously, is located within the loop that links the N-terminal a helix to the b 1 G-protein a subunits, is the most variable region between of G a 15 and G a 16 proteins. Our results revealed that, as observed with family 1 GPCRs, a 4 helix, L9 loop, b 6 sheet, and a 5 helix together play an important role in controlling their coupling of G a -protein to family 3 receptors. This strengthens the hypoth-esis that these two distant heptahelical receptor types are recognized by similar molecular determinants on the G-protein a subunits. Our results also indicate that differences in amino acid residues in the region comprised by a 4 and a 5 helices are not sufficient to explain the differential coupling sensitivity of G a 15 and G a 16 to mGlu8 receptors. Indeed, we identified a second region in G a 16 that, once replaced by the equivalent G a 15 sequence, lead to a chimera showing distinctive and measurable coupling to mGlu8 receptors. This novel region is characterized by a 2-amino acid residue exchange between G a 15 and G a 16 , where Glu 39 and Glu 41 in G a 15 are replaced by Asp and Gly in G a 16 , respectively. These two residues are located in the loop linking the N-terminal a helix and the first b sheet of the Ras-like domain. Our three-dimensional model indicates that side chains of these residues are likely to be facing the surface of the protein in close proximity to the C terminus. As such they are likely to constitute a major deter-minant in defining the differential coupling observed for G a 15 and G a 16 to mGlu8 receptor. This region has never been pre viously reported as being involved in the control of receptor G-protein coupling, and it is therefore a novel possible contact site between the G-proteins and family 3 GPCRs. An alterna-tive that could not be ruled out in the present study is that these residues may affect flexibility of the N-terminal helix. our data

The specific signaling pathway of a given G-protein-coupled receptor (GPCR) 1 depends on the subset of G-proteins it can activate. Transduction of signals by activated GPCR requires the interaction of the receptors with heterotrimeric G-proteins. Recent progress in defining the structure of G-proteins and site-directed mutagenesis studies of receptors and G-proteins bring increasingly more specific and precise descriptions of the contact sites between these proteins (1,2). From the G-protein ␣, ␤, and ␥ subunits, probably both G␣ and ␤␥ dimers contact the receptors (3). The G␣ subunit is likely to play a decisive role in discriminating between different receptor subtypes (4 -6) and also between different functional states of receptors (7,8).
Several regions along the sequence of the G␣-protein are involved in the selectivity of its activation by GPCR (1,9,10). The best characterized is the extreme C terminus, where residues at position Ϫ3 and Ϫ4 (the residue Ϫ1 being the last one) are decisive for coupling of G␣ proteins with specific receptors (1,(11)(12)(13)(14)(15). Residue Ϫ4 in G␣ i , G␣ o , and G␣ t is the cysteine residue that is ADP-ribosylated by pertussis toxin, a covalent modification that prevents the interaction of the G-protein with the receptors. In the G q family, the residue Ϫ4 is a tyrosine residue that has to be phosphorylated for an efficient coupling to the PLC-activating receptors (16). Conformational changes in the C-terminal structures upon coupling to the GPCR probably cause activation of the G␣Ϫprotein. Another region determining coupling selectivity is the extreme N terminus. In the case of G q/11 -proteins, this N terminus has been shown to restrict coupling of these G-proteins to a subset of GPCRs, namely to those known to activate PLC (17). Finally, the region between ␣4 and ␣5 helices that includes the L9 loop and ␤6 sheet (18) is also involved in coupling selectivity probably by directly interacting with the receptors of the rhodopsin-like family (family 1 GPCRs) (10, 19 -22). The overall surface charge of a region that comprises the structures close to the C terminus is probably defining either selectivity or coupling properties in general (23).
Among the various G-proteins identified so far, the G␣ 15 subunit has unique properties. This G-protein that is found exclusively in the murine hematopoietic cell lineage shares the closest sequence similarity with the G␣ q protein and activates PLC␤s (24,25). In cells normally expressing this G-protein as well as in heterologous expression systems it was found that it can couple to many GPCRs, including those that naturally do not stimulate PLC. This was observed with the members of family 1 GPCRs (26 -28) and the mGlu receptor family (family 3 GPCRs) (29,30). G␣ 15 is therefore a G-protein that is not able to discriminate between a variety of receptors (26). Functional characterization of the human homologue G␣ 16 revealed that this protein also couples to many GPCRs (27) but is not as promiscuous as G␣ 15 . It is of interest to understand what determines the lack of selectivity of G␣ 15 and which sequences restrict promiscuity of the G␣ 16 (29,30). G␣ 15 and G␣ 16 proteins share 85% sequence identity. Interestingly, they share identical C termini, indicating this sequence element is not responsible for their differential coupling. Such a situation appears ideal to identify other regions in these G-proteins involved in their specific interaction with GPCRs. In the present study, the coupling of both mGlu2 and mGlu8 receptors to a series of G␣ 15 /G␣ 16 chimeric G-proteins was analyzed. This study revealed that amino acids within ␣4 helix, ␤6 strand, L9 loop, and ␣5 helix constitute one of the determinants that allows G-protein ␣ subunit to discriminate between members of family 3 receptors, a situation corresponding to data reported for family 1 GPCRs. In addition we identified a new region within the G-protein ␣ subunit involved in the selective recognition of group II versus group III mGlu receptors. This new region is located in the loop that connects ␣N helix with ␤1 sheet.

EXPERIMENTAL PROCEDURES
Materials-Chemicals including glutamate were obtained from Sigma (L'Isle d'Abeau, France) unless otherwise indicated. Serum, culture media, and other solutions used for cell culture were from Life Technologies, Inc. (Cergy Pontoise, France). The plasmids expressing mGlu receptors were as described previously (14) or modified (see below). The hemagglutinin epitope-tagged G␣ q was kindly provided by Dr. Bruce Conklin (The Gladstone Institute, San Francisco, CA). The plasmids pCIS-G␣ 15 and pCIS-G␣ 16 were kindly provided by Dr. M. Simon (Caltech, Los Angeles, CA).
Determination of Inositol Phosphate (IP) Accumulation-The procedure used for the determination of IP accumulation in transfected cells was adapted from previously published methods (33,34 . Cells were then washed again twice with the same buffer, and LiCl was added to a final concentration of 10 mM. The agonist was applied 5 min later and left for 30 min. The reaction was stopped by replacing the incubation medium with 0.5 ml of perchloric acid (5%), on ice. Supernatants were recovered, and the IPs were purified on Dowex columns. Total radioactivity remaining in the membrane fraction was counted after treatment with 10% Triton X-100, 0.1 N NaOH for 30 min and used as standard. Results are expressed as the amount of IP produced over the radioactivity present in the membranes. The doseresponse curves were fitted according to the equation y ϭ ((y max Ϫ y min )/1 ϩ (x/EC 50 ) n ) ϩ y min using the kaleidagraph program (Abelbeck software).
Antibodies Characterization-The anti-G␣ 15/16 antibodies were raised in rabbits against a synthetic peptide corresponding to the last 10 amino acid residues that are common for both the G␣ 15 and G␣ 16 . The antiserum was tested in several dilutions (data not shown). The 1:2000 dilution gave optimal results. ECL chemiluminescence system was used to stain the secondary antibodies (Amersham Pharmacia Biotech, Paris, France).
The chimeric G␣ qX5 and G␣ qX6 were constructed using a unique XbaI site of the cDNA encoding the G␣ q -HA-tagged protein. The XbaI sites in the C-terminal part of G␣ 15 and G␣ 16 were introduced by silent mutagenesis into the consensus sequence Met-Asp-Leu. This resulted in the expression of chimeric proteins that were detectable by both the HA antibodies and our anti-G␣ 15/16 C-terminal antibody.
Immunoblotting Analysis-Cells were transfected and treated as described above. Cells cultured in one well were harvested for immunodetection, and the rest of the cells were used for IP assay. Protein samples (10 g per lane) were separated using SDS-polyacrylamide gel electrophoresis and blotted onto nitrocellulose membrane. Prior to the immunoblotting, total protein on the membranes was visualized with Ponceau S to confirm that the same amounts of protein were loaded on the gel. The G-proteins were detected using primary monoclonal antibodies against hemagglutinin-epitope (generous gift of Dr. B. Mouillac, Montpellier, France) or the newly described polyclonal anti-G␣ 15/16 C-terminal antibodies. The His 6 -tagged receptors were detected using specific antibodies recognizing the MRGS-His epitope (Qiagen, Paris, France).
Construction of Plasmids Expressing Higher Levels of G␣ 16 -We have noticed considerable lower expression levels of the human G␣ 16 FIG. 1. Characterization of the anti-G␣ 15/16 antibodies. Protein extracts of cells transfected with a hemagglutinin epitope-tagged (HA) G␣ q subunit and chimeras between the G␣ q and G␣ 15 and G␣ 16 were resolved by SDS-polyacrylamide gel electrophoresis. The indicated antibodies were used for immunoblotting. Upper panel, immunoblot using the polyclonal anti-G␣ 15/16 antibodies. In the lower part, a portion of the same blot developed with the monoclonal anti-HA antibody. The bands corresponding to the G-proteins migrated with expected velocity. On the right, a schematic representation of the chimeric proteins between G␣ q -HA, G␣ 15 , and G␣ 16  protein than those of G␣ 15 in transfected HEK 293 cells. To get higher expression levels of G␣ 16 , we removed most of the 5Ј-untranslated region and introduced a Kozak sequence corresponding to that of G␣ 15 by polymerase chain reaction using the unique EcoRI-PstI sites. These modifications were sufficient to raise the G␣ 16 protein expression obtained with this new G␣ 16 PLUS construct to levels comparable with those reached by G␣ 15 . This construct was used through out the study and is referred as to G␣ 16 in this text.
Construction of Chimeric G-proteins-Silent mutagenesis approach using Pfu-based QuickChange TM technique (Stratagene) was used to introduce a BamHI, a PstI, and a BamHI restriction site at the positions indicated in Fig. 2 as sites 1, 2, and 4, respectively. Together with the existing XbaI site (position 3 in Figs. [2][3][4] and the HindIII site in the 3Ј-untranslated region, these sites were used to construct the chimeric proteins described in Figs. 3 and 4 using conventional subcloning techniques. The same approach was used for making the point mutations. Construction of Epitope-tagged Receptors-The mGlu2 and mGlu8 receptors were tagged with MRGS-His 6 epitope at the C termini as in our previous studies (14). Briefly, the pRK5 plasmid containing the sequence encoding the epitope MRGS-His 6 flanked by a unique NheI site at its 5Ј end, and an in-frame TAA stop codon at its 3Ј end was used. The receptors coding sequences were then introduced by polymerase chain reaction using NheI restriction site so that the stop codon was replaced by alanine followed by serine and the MRGS-His 6 epitope (where MRGS indicates: methionine, arginine, glycine, serine).
Molecular Modeling-The three-dimensional model of the G␣ 15 subunit was constructed by homology using the coordinates of different G-protein ␣ subunits in their GDP form obtained by x-ray crystallography. These include transducin in its GDP plus Mg 2ϩ form, the ␣t/␣i chimeric protein in its heterotrimeric form with ␤t␥t (35), the ␣i1 in its GDP plus Mg 2ϩ form (36), and in its heterotrimeric complex with ␤1␥2 (37) (Protein Data Bank accession numbers 1TAG, 1GOT, 1GDD, and 1GP2, respectively). Some structural elements were deleted due to their inappropriate folding in the expected heterotrimeric form bound to the receptor. These deletions include the N-terminal ␣ helix (the 24 Nterminal residues of the resolved structure) and the extreme C terminus (4 residues) of the ␣i1 in its GDP plus Mg 2ϩ form. Some constraints were also imposed during the modeling process of G␣ 15 : an ␣ helical secondary structure was imposed to residues 7-39, 305-325, 330 -334, and 349 -370. The sequence alignment and three-dimensional models were generated using the program Modeler (38) in the Insight-II environment (Molecular Simulation Inc., San Diego, CA) on a Silicon Graphic R10000 O 2 work station. A statistical evaluation of the threedimensional model was performed using the Verify 3D algorithm (39) and the Verify 3D Structure Evaluation Server. The model giving the best one-dimensional/three-dimensional scores was selected and subjected to energy minimization using the program Discover 2.9.7 (Molecular Simulation Inc.) and the CVFF force field. The extreme C terminus of the G␣ 15 subunit was modeled manually according to the resolved structure of the 10 C-terminal residues of transducin bound with rhodopsin (40). The ␤1␥2 subunits were added after superposition of the G␣ 15 model with the G␣ i1 subunit in its heterotrimeric complex. 15/16 Antibodies-The mGlu2 and mGlu8 receptors couple to the G i type of G-proteins and do not activate PLC pathway when expressed alone in HEK 293 cells. Because G␣ 15 and G␣ 16 activate PLCs, their efficient coupling to these mGlu receptors can easily be assayed by measuring the capability of glutamate to activate the PLC pathway in cells expressing both the receptor and one of these G-proteins. As mentioned in the introduction, we reported previously that glutamate activated PLC in HEK 293 cells coexpressing mGlu2 receptors and either G␣ 15 or G␣ 16 . In contrast, glutamate activated IP formation in cells expressing mGlu8 receptors and G␣ 15 but not in those coexpressing this receptor with G␣ 16 . To conclude that G␣ 16 can be activated by mGlu2 but not mGlu8 receptors, it was necessary to verify that both G-proteins were expressed at similar levels.

Characterization of Anti-G␣
We generated a new polyclonal antibody, called anti-G␣ 15/16 AB that recognizes equally well both G␣ 15 and G␣ 16 (Fig. 1). The antiserum was raised in rabbits immunized with a peptide corresponding to the extreme C-terminal region (10 amino acid residues) common to both proteins (Figs. 1 and 2). On Western blots, anti-G␣ 15/16 AB recognized a single major band that migrated at a velocity appropriate for these G-proteins (Fig. 1). This band was detected only in lanes where membrane proteins from HEK 293 cells expressing proteins possessing C termini of G␣ 15 or G␣ 16 (1-4). The empty bar at the C terminus indicates the sequence used to rise the polyclonal antibodies. In b the position of the secondary structures and the two coupling restrictive regions (hatched bars) responsible for the differential coupling of G␣ 15 and G␣ 16 to mGlu8 receptors are shown. ditional major band was detected, and no bands were detected in lanes where extracts from cells expressing other G␣-protein subunits were loaded (data not shown). To verify that anti-G␣ 15/16 AB was recognizing both G-proteins at the same extent, we expressed chimeric proteins corresponding to the HAtagged G␣ q protein with its last 51 C-terminal residues replaced by their corresponding 63 residues of either G␣ 15 or G␣ 16 (chimeras G␣ qX5 and G␣ qX6 , see Fig. 1). Those were expressed in HEK 293 cells at the same levels, as shown with the HA specific monoclonal antibody. When blots were reprobed with the new antibody anti-G␣ 15/16 AB, the bands corresponding to chimeras bearing G␣ 15 and G␣ 16 C termini were stained with similar intensity (Fig. 1).
Optimization of G␣ 16 Expression-Employing the anti-G␣ 15/16 AB antibody on immunoblots, we noticed that the level of expression of G␣ 16 was lower than that of G␣ 15 in cells transfected with the original vectors (data not shown). We therefore replaced the 5Ј-untranslated region of G␣ 16 cDNA, including the Kozak sequence with corresponding sequence from of G␣ 15 coding cDNA (see "Experimental Procedures"). This modification was found to raise the expression of G␣ 16 in transiently transfected HEK 293 cells to levels comparable with those of G␣ 15 (Figs. 1 and 3b).
In Contrast to mGlu2 Receptors, mGlu8 Receptors Do Not Efficiently Activate G␣ 16 -Using the expression plasmids described above, we examined the coupling of mGlu2 and mGlu8 receptors to both G␣ 15 and the modified G␣ 16 . As shown in Fig.  3a, glutamate stimulated IP formation in cells coexpressing mGlu2 receptors with either G␣ 15 or G␣ 16 , an effect that is dose-dependent (Fig. 6). Using the new G␣ 16 vector, the maximal glutamate effect was found to be slightly lower than that obtained with G␣ 15 in cells expressing mGlu2 receptors, whereas it was 50% smaller when the original plasmid was used (29,30). This is in agreement with the increased protein expression levels of G␣ 16 that resulted from the modification of the coding plasmid. In cells expressing mGlu8 receptors, a large increase in basal and glutamate-induced IP formation was detected when coexpressed with G␣ 15 , but no change was observed with G␣ 16 . All the glutamate effects were dose-dependent and the calculated EC 50 values (for mGlu2 receptor-G␣ 15 pair the EC 50 for glutamate was 8.2 Ϯ 0.8 M; mGlu2 receptor with G␣ 16 results in EC 50 8.0 Ϯ 0.3 M glutamate, and mGlu8 receptor coexpressed with G␣ 15 determined EC 50 for glutamate was 9.3 Ϯ 1.6 M) were close to those determined with these receptors using other assays (41). It was important to establish that the expression levels of both the receptors and the G-proteins are similar at their different combinations. At the receptor level this was confirmed by using His 6 -tagged receptors (data not shown). These data further confirm the discriminative coupling of G␣ 16 to mGlu2 and mGlu8 receptors.
Two Regions on G␣ 15 and G␣ 16 Are Recognized by mGlu Receptors-To identify the putative regions on the G-protein FIG. 3. a, chimeric proteins between G␣ 15 and G␣ 16 at the ␣4-␣5 region were constructed by swapping corresponding portions using silent restriction sites or by introducing point mutations. b, immunoblot analysis of G␣ 15 and G␣ 16 proteins and the chimeras and the point mutations expressed in HEK293 cells, stained with the anti-G␣ 16 antibodies. c, differential coupling of the G␣ 15 , G␣ 16 , and their reciprocal chimeras to mGlu2 and mGlu8 receptors. Basal (open bars) and 1 mM glutamate-induced (dark bars) IP formations were determined in HEK 293 cells coexpressing the mGlu2 (a) or mGlu8 (b) receptors with the G-protein ␣ subunits and their chimeras. The new constructs of the G␣ 16 proteins with augmented expression levels were used. Data represent the radioactivity in the IP fraction divided by the total radioactivity in the membranes. that cause the differential coupling of G␣ 15 and G␣ 16 to mGlu8 receptor, we constructed a series of G␣ 15  Chimera G␣ 15/16 C1, which corresponds to G␣ 15 with the Cterminal 44 residues from G␣ 16 (Fig. 3a), showed a smaller response to glutamate as compared with wild-type G␣ 15 (Fig.  3c). This indicates that the C-terminal portion of G␣ 16 contains a site that decreases G-protein coupling efficacy to mGlu8 receptors (Fig. 3c). Progressive exchange of G␣ 15 C terminus with corresponding sequences of G␣ 16 , as in chimera G␣ 15/16 C2, resulted in a more pronounced decrease in coupling efficiency to mGluR8 (Fig. 3c). The converse chimera G␣ 15/16 C3, which corresponds to G␣ 16 with the C-terminal 44 residues of G␣ 15 , was activated by mGlu8 receptors, although the glutamate response was smaller than that obtained with G␣ 15 (Fig. 3b). The module in G␣ 16 responsible for discriminating between the two receptors is most likely a region that includes ␣4 helix, L9 loop, the ␤6 sheet, and the ␣5 (residues 331-354). Residues that are different between G␣ 15 and G␣ 16 is these regions are: 10 in the ␣4 helix and N-terminal part of L9 loop, 3 in the C-terminal part of L9, 3 in ␤6, and 3 in ␣5 (Fig. 2). Attempts to further characterize the critical residues in this C-terminal region failed. For example, the comparison of the IP formation obtained with mGlu8 receptors expressed with the C1 and C2 G␣ 15/16 chimera suggests that the region between position 3 and 4 (Fig. 3a) decreases coupling efficacy of G␣ 16 . However, swapping this region of G␣ 16 for that of G␣ 15 in the chimera C4 did not improve the coupling compared with chimera C3, but rather diminished it. Also in chimera G␣ 15/16 C5, the exchange of the 3 residues of ␣5 in G␣ 15 by those of G␣ 16 lead to a decrease of functional responses obtained with both mGlu2 and mGlu8 receptors. The most likely explanation is that the region including ␣4, L9 loop, ␤6, and ␣5 discriminates the GPCRs as a net of many forces, a situation where a given charge of a side chain of an amino acid residue is less important than the outcome of additions of neighboring forces. One can imagine this situation also as a mosaic, where the whole picture is the decisive one. This is in agreement with results obtained previously by other researchers (23). Since the result of swapping this part of the G-proteins had only partial effect on the mGlu8 receptor-G␣-protein coupling, there had to be another discriminatory region. Swapping the N-terminal 32 residues of G␣ 16 by those of G␣ 15 resulted in chimera G␣ 15/16 N1, which showed a feeble coupling to mGlu8 receptors (Fig. 4). When the first 101 residues of G␣ 16 were exchanged by the corresponding region in G␣ 15 , chimera G␣ 15/16 N2, a significant increase in the capability of the chimeric G-protein to couple to mGlu8 receptors was observed. Interestingly, G␣ 15/16 N2 did not perform any better than the G␣ 15/16 C4 chimera (compare Figs. 3 and 4),suggesting that the region between position 101 and 313 did not play an important role in controlling coupling selectivity to mGlu2 and mGlu8 receptors. Taken together, these results indicate that two distinct regions in G␣ 16 , encompassing residues 32-101 and 331-354, respectively, are responsible for the observed decrease in coupling efficacy to mGlu8 receptors (see Fig. 2). In agreement with this hypothesis, chimera G␣ 15/16 N1C4 that is characterized by residues 331-354 of G␣ 15 and residues 32-101 of G␣ 16 was activated by mGlu8 receptors but not to the extent showed by wild type G␣ 15 . Chimera G␣ 15/16 N2C4, which is constituted by the backbone of G␣ 16 with both critical regions corresponding to residues 32-101 and 331-354 replaced by the respective G␣ 15 sequences, allowed a more efficient coupling of mGlu8 receptors to PLC. Considering the critical role played by residues 32-101 in conferring G-protein coupling specificity, we decided to make point mutations within this region in both G␣ 15 and G␣ 16 .
The ␣N-␤1 Loop of the G␣-protein Selectively Modifies Cou-pling to mGlu Receptors-Within region 32-101, several amino acid residues differ between G␣ 15 and G␣ 16 (Fig. 2). Two of these nonconserved residues are located in the loop that links the N-terminal ␣ helix and the first ␤ strand of the Ras-like domain (the ␣N-␤1 loop), the other residues are located in the ␣A helix of the helix-rich domain. Among this set of residues, those located in the ␣N-␤1 loop are the most likely to affect G-protein coupling selectivity, since they are located in close proximity to the known G-protein-mGlu receptors interacting site, the extreme C terminus, and are exposed on the surface of the protein. Interestingly, the ␤1 sheet is highly conserved within the G␣-protein family, whereas the contiguous ␣N-␤1 loop is extremely variable. As mentioned above, two residues in the ␣N-␤1 loop are different between G␣ 15 and G␣ 16 : glutamic acid 39 and 41 in G␣ 15 are replaced by an aspartic acid and a glycine in G␣ 16 , respectively (Fig. 2). Our results indicate that both residues at positions 39 and 41 are involved in determining coupling selectivity to mGlu receptors. To reach maximal mGlu8 receptor coupling activity, both positions in G␣ 16 have to be replaced by the corresponding residues of G␣ 15 (chimera G␣ 16 E-E; Figs. 5 and 6); this is in agreement with what has been observed with chimera N2 (Fig. 5). Moreover, the reverse mutant G␣ 15 D-G showed a significant decrease of coupling performance with mGlu8 receptor as compared with wild-type G␣ 15 . All mutants exhibited comparable good coupling to mGlu2 receptor, even though with slightly lower efficacy than G␣ 15 . Accordingly, basal IP formation induced by mGlu2 receptor was lower with these mutants than with G␣ 15 (Figs. 5 and 6). It is important to mention again that the expression levels of the G-proteins and their mutants and the receptor levels were similar in all experiments, even when cDNAs coding for different G-protein constructs were used.

Molecular Modeling of G␣ 15 Reveals Clusters of Residues Involved in Receptor Recognition in a Close
Proximity to the C Terminus-A three-dimensional model of G␣ 15 subunit was constructed to determine the exact spatial location of the two critical regions, which allow coupling of G␣ 15 to mGlu8 receptors. This model was generated using the coordinates of the ␣ subunits transducin and G␣ i in their GDP-Mg 2ϩ form (18,42), since these are expected to correspond to the conformation of the G-protein that interacts with the receptor. Within the region between helices ␣4 and ␣5 (residues 313-354), these residues (except Val 352 ) are on the surface of the protein and are found in a relative close proximity to the C terminus (Fig. 7). Within the N-terminal region (residues 32-101), the two residues that we identified as being important for the efficient coupling of G␣ 15 to mGlu8 receptors (Glu 39 and Glu 41 ) are in close proximity to the C terminus (Fig. 7). The majority of the other residues of G␣ 15 that are not conserved in G␣ 16 are in our model far away from the C terminus of the ␣ subunit (the only exception are Lys 200 and Lys 203 in G␣ 15 replaced by Gln and Asn in G␣ 16 , respectively) and as such are unlikely to affect coupling selectivity of the heterotrimeric G-protein. DISCUSSION In this study we showed that two distinct regions along the sequences of G␣ 15 and G␣ 16 control their coupling to heptahelical receptors. One such region is between residues 311 and 354 that includes the C-terminal part of the ␣4 helix, L9 loop, the ␤6 strand, and the ␣5 helix. The second region comprises residues 39 and 41, which lie within the loop that links the N-terminal ␣ helix with the first ␤ strand of the Ras-like domain of the G-protein ␣ subunit. Both regions are in close proximity to the extreme C terminus in the spatial arrangement of the proteins. These data further strengthen the role of the area located in close proximity to the C terminus of G␣Ϫprotein in controlling coupling to heptahelical receptors and extent the putative surface on G␣Ϫprotein that modulates the interaction with this class of receptors.
To better characterize the regions in G-proteins involved in GPCR recognition, we took advantage of the differential coupling of G␣ 15 and G␣ 16 to mGlu2 and mGlu8 receptors. The In red are the 2 residues that are located in the loop linking the N-terminal ␣ helix and the first ␤ sheet of the Ras-like domain residues that are not conserved in G␣ 16 and that are compatible with the coupling of G␣ 15 , but not of G␣ 16 to mGlu8 receptor. The orange residues are trhose that are different between G␣ 15 and G␣ 16 at the ␣4 helix, L9 loop, ␤6 sheet, and ␣5 helix region. The yellow residues are those that are different between G␣ 15 and G␣ 16 but had no effect on the differential coupling. On the left is the view of the G-protein trimer from the receptor-oriented side. On the right is the same model rotated 90°(side view). The ␤ subunit is in light blue, and the ␥ subunit is in green. mGlu receptors and G␣ 15/16 are unlikely to naturally interact. The mouse G␣ 15 and human G␣ 16 are known to be exclusively expressed in the hematopoietic cell lineage, whereas mGlu receptors are mostly expressed in neurons and/or in astrocytes in the central nervous system. However, several consideration prompted us to perform this study. The interaction between GPCR and the heterotrimeric G-protein is known to involve several regions on the G␣-protein, the best characterized being the extreme C terminus. Since G␣ 15 and G␣ 16 share identical C termini, this study allowed us to identify additional regions with coupling-restrictive abilities. All G␣ subunits share high sequence similarity and therefore have highly conserved structures; thus, identification of additional regions involved in coupling selectivity of G␣ 15/16 would help us to better map residues in all the ␣ subunits that control the coupling selectivity and/or efficacy to their partner receptors. The mGlu receptors are members of a subfamily of GPCRs (family 3) that is distinct from the rhodopsin-like family 1 GPCRs. Thus, identification of regions on a G-protein ␣ subunit involved in specific interaction with these receptors should allow a comparison of the coupling mechanisms between family 1 and family 3 receptors. G␣ 15/16 are known to couple to a wide variety of GPCRs, making these G-proteins a very useful tool for the establishment of functional tests for receptors for which no information on the G-protein coupling selectivity is available (28). Identification of the specific epitopes on G␣ 15/16 that influence their promiscuous coupling could constitute an helpful information for the construction of modified G-proteins with even lower selectivity.
We were expecting that the residues responsible for the differential coupling of G␣ 15 and G␣ 16 toward mGlu8 receptors would be located in the L9 (␣4-␤6) loop. This region was described as a key site in G-protein ␣ subunits discriminating between members of family 1 GPCRs (9, 10, 19 -21). This loop, which is longer than the corresponding sequences of other G-protein ␣ subunits, is the most variable region between of G␣ 15 and G␣ 16 proteins. Our results revealed that, as observed with family 1 GPCRs, ␣4 helix, L9 loop, ␤6 sheet, and ␣5 helix together play an important role in controlling their coupling of G␣-protein to family 3 receptors. This strengthens the hypothesis that these two distant heptahelical receptor types are recognized by similar molecular determinants on the G-protein ␣ subunits. Our results also indicate that differences in amino acid residues in the region comprised by ␣4 and ␣5 helices are not sufficient to explain the differential coupling sensitivity of G␣ 15 and G␣ 16 to mGlu8 receptors. Indeed, we identified a second region in G␣ 16 that, once replaced by the equivalent G␣ 15 sequence, lead to a chimera showing distinctive and measurable coupling to mGlu8 receptors. This novel region is characterized by a 2-amino acid residue exchange between G␣ 15 and G␣ 16 , where Glu 39 and Glu 41 in G␣ 15 are replaced by Asp and Gly in G␣ 16 , respectively. These two residues are located in the loop linking the N-terminal ␣ helix and the first ␤ sheet of the Ras-like domain. Our three-dimensional model indicates that side chains of these residues are likely to be facing the surface of the protein in close proximity to the C terminus. As such they are likely to constitute a major determinant in defining the differential coupling observed for G␣ 15 and G␣ 16 to mGlu8 receptor. This region has never been previously reported as being involved in the control of receptor G-protein coupling, and it is therefore a novel possible contact site between the G-proteins and family 3 GPCRs. An alternative that could not be ruled out in the present study is that these residues may affect flexibility of the N-terminal helix. Since the N terminus of G␣-protein contacts the ␤ subunit, this might influence ␤-␥ coupling and so coupling to receptors (43).
Taken together, our data indicate that at least two sets of residues in G␣ 15 and G␣ 16 contribute to coupling restriction of these G-proteins to the mGlu8 receptors. The residues comprised in these two regions are located on both sides of the extreme C-terminal domain known to interact with the intracellular loops of the heptahelical receptors. This cooperativity between different portions of the G-protein ␣ subunit, that leads to differential coupling to heptahelical receptors, is reminiscent of the cooperativity between different intracellular portions of the receptors in controlling their coupling to Gproteins. This strengthens the concept that coupling efficacy and/or selectivity between heptahelical receptors and the Gproteins results from a multiplicity of contact zones. It remains to be clarified which portion of a G-protein is recognized by a given intracellular region on the receptors.