The Bombesin Receptor Subtypes Have Distinct G Protein Specificities*

We used an in situ reconstitution assay to examine the receptor coupling to purified G protein α subunits by the bombesin receptor family, including gastrin-releasing peptide receptor (GRP-R), neuromedin B receptor (NMB-R), and bombesin receptor subtype 3 (BRS-3). Cells expressing GRP-R or NMB-R catalyzed the activation of squid retinal Gαq and mouse Gαq but not bovine retinal Gαt or bovine brain Gαi/o. The GRP-R- and NMB-R-catalyzed activations of Gαq were dependent upon and enhanced by different βγ dimers in the same rank order as follows: bovine brain βγ > β1γ2 ≫ β1γ1. Despite these qualitative similarities, GRP-R and NMB-R had distinct kinetic properties in receptor-G protein coupling. GRP-R had higher affinities for bovine brain βγ, β1γ1, and β1γ2 and squid retinal Gαq. In addition, GRP-R showed higher catalytic activity on squid Gαq. Like GRP-R and NMB-R, BRS-3 did not catalyze GTPγS binding to Gαi/o or Gαt. However, BRS-3 showed little, if any, coupling with squid Gαq but clearly activated mouse Gαq. GRP-R and NMB-R catalyzed GTPγS binding to both squid and mouse Gαq, with GRP-R activating squid Gαq more effectively, and NMB-R also showed slight preference for squid Gαq. These studies reveal that the structurally similar bombesin receptor subtypes, in particular BRS-3, possess distinct coupling preferences among members of the Gαq family.

Mammalian bombesin-like peptides, gastrin-releasing peptide (GRP) 1 and neuromedin B (NMB), are widely distributed in the nervous system and the gut. They regulate various physiological processes such as secretion, growth, muscle contraction, and neuromodulation through high affinity receptors (1,2). Three pharmacologically and structurally distinct bomb-esin receptor subtypes have been cloned and characterized in mammals as follows: the GRP-preferring receptor (GRP-R), the neuromedin B-preferring receptor (NMB-R), and bombesin receptor subtype 3 (BRS-3) which has a structure related to GRP-R and NMB-R but for which no high affinity, naturally occurring ligand has been identified as yet (2). Comparison of the predicted amino acid sequences (2) of the bombesin receptor subtypes shows all three to be structurally related members of the G protein-coupled receptor superfamily with pairwise sequence identity ranging from 48 to 54% (see Fig. 1). Upon agonist binding, G protein-coupled receptors activate specific heterotrimeric G proteins, which in turn regulate a variety of intracellular effectors such as adenylyl cyclase, phospholipase C, ion channels, and cGMP-phosphodiesterase (3).
Heterotrimeric G proteins are composed of three polypeptides as follows: an ␣ subunit and a ␤␥ dimer that acts as a functional monomer. Ligand-activated G protein-coupled receptors catalyze the exchange of GTP for GDP bound to the G␣ subunit, resulting in dissociation of the GTP-activated ␣ subunit from both its cognate G␤␥ dimer and the receptor. The GTP-activated ␣ subunit as well as dissociated G␤␥ dimer in turn regulate intracellular effectors. At least 20 different ␣ subunits, 5 ␤ subunits, and 12 ␥ subunits have been identified to date. The G␣ subunits have been divided into four groups based upon sequence homology and intracellular effector regulation (4,5). The G␣ q subfamily, which includes G␣ q , G␣ 11 , G␣ 14 , and G␣ 15/16 , stimulates phosphoinositide hydrolysis by activating phospholipase C-␤ (6 -10). In addition, G␤␥ subunits can also stimulate phospholipase C-␤s in concert with G␣ q (11,12).
Given that the seven transmembrane domain receptor superfamily consists of thousands of distinct receptors, and the family of heterotrimeric G proteins involved in receptor coupling is also very diverse, a central issue in receptor signaling is how these protein families contribute to the diversity of receptor/G protein-mediated responses while conserving the specificity of each response. One level of specificity is likely to be determined by the thermodynamics of protein-protein interactions between subunits of the heterotrimeric G protein and the receptor. An in situ reconstitution procedure has been used successfully to study receptor-G protein interactions for baculovirus-infected Sf9 cell membranes expressing the 5-HT 2c receptor (13), and for mouse fibroblast cell membranes expressing stably transfected GRP-R (14). This technique utilizes chaotrope-extracted membrane fractions in which endogenous GTP-binding proteins as well as other extrinsic membrane proteins are removed or inactivated by urea, while leaving uncoupled receptors fully functional when reconstituted with agonist and purified G protein subunits.
Since mammalian bombesin receptors stimulate phospho-* 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 U.S.C. Section 1734 solely to indicate this fact.
The inositide hydrolysis (15)(16)(17), it has been assumed that agonist stimulation of bombesin receptors leads to activation of a G␣ q , which in turn activates a phospholipase C-␤ isozyme. Antisense oligonucleotide injection of Xenopus oocytes (18) has identified G␣ q as a mediator of the NMB-R response. However, in Xenopus oocytes neither G␣ q nor G␣ 11 antisense injection had any effect on GRP-R signal transduction, although the in situ reconstitution of GRP-R with purified G protein subunits shows explicitly that GRP-R activates a G␣ q but not G␣ i/o or G␣ t (14). Such observations raise a possibility that the ambiguity in the antisense oligonucleotide experiments could be due to a difference in the relative affinity or activity of GRP-R for G␣ q as compared with that of NMB-R for G␣ q . The assessment of this possibility requires a quantitative comparison of GRP-R and NMB-R coupling with purified G protein subunits in vitro.
In this report we compare the receptor-G protein interactions within the structurally related bombesin receptor family using the in situ reconstitution assay. We have quantitatively examined the G protein activation by these related receptor structures using homogeneous preparations of defined G protein subunits. Our studies revealed that whereas GRP-R and NMB-R selectively coupled with squid and mouse G␣ q in an agonist-and G␤␥-dependent manner, their coupling properties were distinct. GRP-R had higher affinities for G␣ q and G␤␥ dimers, higher catalytic activity for nucleotide exchange on G␣ q , and a higher ratio of agonist stimulated to basal activity than NMB-R. On the other hand, the structurally related BRS-3 was different from GRP-R and NMB-R in selectivity for G␣ q . It strongly preferred mouse G␣ q over squid G␣ q in the in situ reconstitution assay, whereas GRP-R clearly preferred squid G␣ q , and NMB-R also showed a slight preference for squid G␣ q .
Formation of Recombinant BRS-3 Baculovirus-A cDNA fragment encoding the open reading frame of the human BRS-3 (hBRS-3) flanked by FLAG epitope tag at the 5Ј end was cloned into EcoRI site of a transfer vector pBacPAK8 (CLONTECH). The sequence at the 5Ј end is 5Ј-AATTCGGCTTGCGCGCATGGACTACAGGACGACGATGACAAG-GCTCAAAGGCAG-3Ј. The sequence at the 3Ј end is identical to that of the original hBRS-3 clone inserted into an EcoRI site (23). Insect cell culture, transfection, plaque purification, and virus amplification of BRS-3 were carried out according to the manufacturer's protocol (CLONTECH).
Urea Extraction of Receptor-containing Membranes-We modified our previously published urea extraction procedure (13,14). The P2 membrane pellet was resuspended in ice-cold solution A (10 mM Hepes, pH 7.4, 1 mM EGTA, 100 M 4-(2-aminoethyl)benzenesulfonyl fluoride HCl) containing 7 M urea. After incubation in 7 M urea for 30 min on ice, the membrane solution was diluted to less than 4 M urea with solution A and then sedimented at 142,000 ϫ g for 30 min at 4°C. Following a single urea extraction and centrifugation, the membrane pellet was washed once with solution A alone and recollected by sedimentation as before. The final pellet was resuspended in solution A with 12% (w/v) sucrose, and aliquots were frozen and stored at Ϫ80°C.
GDP/GTP␥S Exchange Assay-The receptor-catalyzed GDP/GTP␥S exchange on G␣ was determined essentially as described previously (32) with the addition of 2 M GDP to compete for uncatalyzed GTP␥S binding (14). Receptor-containing membranes were mixed with G protein subunits and with or without agonist on ice in a total volume of 30 l. An addition of 20 l of reaction solution was used to initiate the reactions. The reactions contained a final concentration of 50 mM MOPS, pH 7.5, 100 mM NaCl, 1 mM EDTA, 3 mM MgSO 4 , 1 mM dithiothreitol, 3 mg/ml bovine serum albumin, 2 M GDP, and rified G protein subunits. These receptor membranes include rod outer segment disc membranes of bovine retina (29), baculovirus-infected Sf9 cell membranes containing the 5-HT 2c receptor (13), and stably transfected fibroblast cell membranes containing GRP-R (14). We applied this in situ receptor reconstitution technique to the bombesin receptor family to compare the G protein coupling properties of bombesin receptor subtypes, which share 48 to 54% amino acid homology (Fig. 1).
We modified the previously published urea extraction procedure to obtain more consistent receptor recovery and G protein depletion (see "Experimental Procedures"). Table I summarizes the effects of the modified procedure on receptor-binding sites and GTP␥S binding activity of GRP-R-and NMB-R-containing membranes. Compared with the 6 M urea extraction procedure used previously, 7 M urea required only one instead of two or three extractions, removed more endogenous GTP␥S binding activity (94 -96% versus 92%), while consistently maintaining high recovery of ligand-binding sites. The GRP-R-binding site abundance was actually enriched more than 3-fold by 7 M urea extraction, since 100% of the antagonist binding activity was recovered, whereas 71% of the membrane protein was removed. Furthermore, for both GRP-R and NMB-R, agonist-stimulated GTP␥S binding in the absence of exogenous G proteins was also abolished more thoroughly by 7 M urea extraction than by 6 M urea extraction used previously (data not shown), suggesting 7 M urea treatment resulted in a more homogeneous population  urea extraction on ligand-binding sites and endogenous GTP␥S binding activity of the GRP-R and NMB-R membranes Membranes were prepared from Balb 3T3 mouse fibroblast cell lines stably expressing GRP-R or NMB-R (see "Experimental Procedures"). After urea extraction, the membrane protein concentrations decreased from 3.5 to 1 and from 6 to 1.6 g/l, for GRP-R and NMB-R, respectively. For the receptor ligand binding assays, 3.5 g of membrane protein from unextracted GRP-R, 1 g of membrane protein from 7 M urea-extracted GRP-R, and 6 g of membrane protein from unextracted NMB-R membranes were used. GRP-R and NMB-R concentrations were determined by Scatchard analysis of 125 I-697 and 125 I-ME binding as described under "Experimental Procedures." GTP␥S binding assays of 0.75 nM GRP-R or 1.0 nM NMB-R, unextracted or 7 M urea-extracted, proceeded for 10 min at 30°C, and bound GTP␥S was determined as described under "Experimental Procedures." All values presented are the means Ϯ S.D. of data obtained from three independent experiments. Table I also shows that 125 I-697 (universal bombesin receptor agonist) and 125 I-ME (GRP-R-specific antagonist) measured identical binding site abundance on membranes before urea extraction, indicating that we could use 125 I-697 to determine the receptor concentration for NMB-R and BRS-3, for which a radiolabeled high affinity antagonist is not available. Because of the decreased affinities of uncoupled receptors for agonists, 125 I-697-binding sites of 7 M urea-extracted GRP-R, NMB-R, and BRS-3 could not be determined accurately.
Since 7 M urea extraction removed more endogenous GTP␥S binding from GRP-R-containing membranes, we tested whether these uncoupled receptors could couple with purified squid retinal G␣ q and bovine brain G␤␥ as was shown previously for 6 M urea-treated GRP-R (14). Fig. 2 shows the results for reconstitution of membranes containing GRP-R and NMB-R either untreated or 7 M urea-extracted. To facilitate the comparison of the efficiency of reconstitution, we have tested ap-proximately equal catalytic activities of the two receptor types at K m concentrations of ␣ q and near-saturating ␤␥. Thus we are directly comparing both the success of reconstitution and the catalytic properties of the receptors. These experiments demonstrated that (i) very little agonist-stimulated exchange of GDP for GTP␥S on exogenously added G␣ q was detected for unextracted GRP-R or NMB-R (Fig. 2, A and C); (ii) both 7 M urea-extracted GRP-R and 7 M urea-extracted NMB-R coupled with squid G␣ q (Fig. 2, B and D); (iii) like GRP-R, NMB-Rcatalyzed activation of G␣ q was also dependent on both agonist and ␤␥ subunits (Fig. 2, B and D). Despite the qualitative similarities between GRP-R and NMB-R in G protein coupling, they were clearly different in the ratio of agonist-independent (basal) to agonist-stimulated activity (Fig. 2, B and D).
In order to obtain initial rate estimates for the receptor-catalyzed GTP␥S binding, we performed the progress analyses for GRP-R-and NMB-R-catalyzed reactions shown in Fig. 3. For both 7 M urea-extracted GRP-R and NMB-R and G protein alone, the binding of GTP␥S progressed at a very low rate; G proteinreconstituted receptors without agonist showed an increased rate of binding, whereas the addition of agonist increased the reaction rate to the highest values. Moreover, for all of these conditions the GTP␥S binding was approximately linear with time for the initial 10 min of the reaction. Therefore we have used 10 min as a fixed time point in the GDP/GTP␥S exchange assay to measure the initial velocity of the receptor-catalyzed activation in all subsequent experiments. The greatly accelerated initial rates of GTP␥S binding in the presence of agonist represent our measure of receptor-catalyzed G protein activation. The GRP-R, NMB-R, and BRS-3 were expressed at widely varying abundance in the Balb 3T3 fibroblasts. Therefore, we have used the modified GTP␥S binding procedures (14) including trace [ 35 S]GTP␥S and 2 M GDP to suppress residual nucleotide binding activity of the urea-extracted membranes rather than our initial procedures that utilize 1 M GTP␥S with no competing nucleotide (13). Our modified procedure also accommodates the comparison of the family of G␣ proteins that differ in spontaneous binding exchange rates. Because the chemical concentration of GTP␥S (4 -8 nM) limits the binding reactions, the plateau values obtained in these experiments are not stoichiometric binding of GTP␥S to the G␣ q . Rather, they represent consumption of the [ 35 S]GTP␥S trace in the binding reactions. That these receptors are indeed catalytic was demonstrated by additional experiments using 1 M GTP␥S without competing GDP in which 1 nM GRP-R or NMB-R activated the entire 100 nM G␣ q in about 40 min (data not shown).
Our previous study has shown selective coupling of GRP-R with G␣ q but not G␣ i/o or G␣ t (14). In order to know whether the other members of the bombesin receptor family share the same selectivity for G␣ q , we tested the ability of urea-extracted membranes to catalyze exchange of GDP for GTP␥S on squid retinal G␣ q , bovine retinal G␣ t , or bovine brain G␣ i/o in the presence of bovine brain G␤␥. Both GRP-R (Fig. 4A) and NMB-R (Fig. 4B) selectively catalyze the exchange reaction on squid G␣ q in an agonist-dependent manner. However, BRS-3 did not activate any of these G protein preparations using the universal bombesin receptor agonist 697 (Fig. 4C). In order to understand why BRS-3 failed to catalyze nucleotide exchange on all tested G proteins, we have attempted to exclude the possibility that peptide 697 is only a partial agonist of the BRS-3. We tested a mutated BRS-3 receptor in which four amino acid residues critical for ligand selectivity were replaced with their counterparts in NMB-R and GRP-R (R127Q, S205P, H294R, and S315A, see Fig. 1). This mutant, 4⌬BRS-3, displays 2 and 3 orders of magnitude increase in affinities for GRP (21) and NMB (33), respectively. Fibroblast cells expressing 4⌬BRS-3

FIG. 3. Kinetics of GRP-R-and NMB-R-catalyzed GTP␥S binding to G␣ q .
Urea-extracted membranes providing final concentrations of 0.5 nM GRP-R (A) or 0.85 nM NMB-R (B) were assayed alone (‚) or reconstituted with 840 nM bovine brain ␤␥ and 150 nM squid retinal G␣ q . Reconstituted membranes were assayed in the absence (E) or presence (q) of 1 M GRP or NMB. The binding of GTP␥S to the G protein subunits in the absence of membranes was also determined (ϫ). For all conditions, the reaction volumes were scaled up to 150 l; the binding reaction was conducted at 30°C, and 10-l aliquots were removed at the indicated times for the determination of GTP␥S binding as described under "Experimental Procedures." The lines drawn for G protein-reconstituted samples are the best-fit simple exponential curves using "Grafit." show NMB-stimulated inositol phosphate increases (33). As shown in Fig. 4D, 4⌬BRS-3 did not catalyze GTP␥S binding on any of the tested G␣ subunits in the presence of NMB. The enhanced GTP␥S binding on G␣ i/o in the presence of all of the bombesin receptors seems to reflect a nonspecific interaction independent of receptors, because the level of GTP␥S binding was proportional to total membrane protein concentration instead of receptor concentration (data not shown).
The lack of coupling of BRS-3 and 4⌬BRS-3 with the G protein subunits tested could be due to the low receptor abundance in Balb 3T3 fibroblast cells (0.26 and 0.33 pmol of receptor/mg of membrane protein for BRS-3 and 4⌬BRS-3, respectively, versus 3.7 and 2.2 pmol/mg for GRP-R and NMB-R) or the absence of essential G protein subunits. To achieve high receptor abundance, Sf9 cells were used to express recombinant BRS-3 encoded by a baculovirus. To test the possibility that BRS-3 can couple with a mammalian G␣ q rather than squid G␣ q , recombinant mouse G␣ q was purified from baculovirus-infected Sf9 cells and used in the reconstitution assays. GRP-R expressed in Sf9 cells was also compared with that expressed in fibroblast cells in order to establish that the receptors expressed in these different cells have the same coupling properties. As shown in Fig. 5, GRP-R expressed in mouse fibroblast cells and insect Sf9 cells behaved the same way. They activated both mouse G␣ q and squid G␣ q but with higher catalytic activity for the latter. Although BRS-3 expressed in fibroblast cells failed to show agonist-stimulated activity with either G␣ q (most likely due to the low receptor abundance), Sf9 cell-expressed BRS-3 clearly showed coupling with mouse G␣ q , but little if any coupling with squid G␣ q . Another member of the bombesin receptor family, NMB-R, falls in between GRP-R and BRS-3 in selectivity for mouse and squid G␣ q . NMB-R showed slightly more efficient coupling with squid G␣ q than with mouse G␣ q .
One of the unique advantages of this in situ reconstitution technique is that it allows a quantitative assessment of receptor-G protein coupling. To determine how well a receptor couples with a G protein as well as to compare coupling efficiency between different receptors, we have performed saturation analysis of the receptor-catalyzed GTP␥S exchange with the G protein subunits. Fig. 6A shows the saturation of the exchange reaction catalyzed by GRP-R and NMB-R with squid G␣ q . The initial velocities conformed to a single-site model with K m values of 58 nM for GRP-R and 112 nM for NMB-R. The catalytic activities of GRP-R and NMB-R were also different, with V max 2 values of 8.5 ϫ 10 Ϫ3 M GTP␥S bound M receptor Ϫ1 s Ϫ1 for GRP-R and 4.1 ϫ 10 Ϫ3 M GTP␥S bound M receptor Ϫ1 s Ϫ1 for NMB-R. Fig. 6B shows the saturation of the catalysis with bovine brain G␤␥. These data also fit well to a single-site model with a K 1/2 of 115 2 These values almost certainly underestimate the catalytic constants of GRP-R and NMB-R for G protein activation. Our GTP␥S binding reactions included 2 M GDP to suppress the receptor-independent binding to G␣. Since the reactions included carrier-free [ 35 S]GTP␥S at 4 -8 nM, the rates for GTP␥S binding in the absence of competing GDP would be much higher. nM for GRP-R and 238 nM for NMB-R. The differences between GRP-R and NMB-R in affinity and catalytic activity for G protein subunits were statistically significant, as summarized in Table II.
Given the diversity of G␤␥ dimers, the receptor-G protein coupling selectivity is unlikely to be restricted to the G␣ subunit alone. To address the question of whether the bombesin receptors also have selectivity for ␤␥ dimers, we tested the ability of GRP-R (Fig. 7A) and NMB-R (Fig. 7B) to activate G␣ q with different ␤␥ dimers, including bovine brain ␤␥, bovine retinal ␤␥ (␤ 1 ␥ 1 ), and ␤ 1 ␥ 2 . When tested at a concentration of 0.25 M, bovine brain ␤␥ showed the greatest enhancement of GRP-R-or NMB-R-catalyzed exchange reaction, ␤ 1 ␥ 2 the second highest, whereas ␤ 1 ␥ 1 hardly affected the binding of GTP␥S to G␣ q . At a concentration of 1 M, ␤ 1 ␥ 1 also enhanced the receptor-catalyzed GTP␥S binding but incompletely, whereas 0.74 M of ␤ 1 ␥ 2 produced the greatest enhancement.
To compare the affinities of GRP-R and NMB-R for G␤␥ dimers, we performed saturation analysis of the receptor-catalyzed GTP␥S binding with G␤ 1 ␥ 1 and G␤ 1 ␥ 2 . As shown in Fig.  8 and summarized in Table II, GRP-R consistently showed higher affinity for the G␤␥ dimers that we tested. For a given G␤␥ preparation, the ratio of K 1/2 of NMB-R and GRP-R ranged from 2.4-to 4.7-fold. Despite the quantitative differences between GRP-R and NMB-R, they showed same rank order of preference among the three ␤␥ preparations: bovine brain ␤␥ Ͼ ␤ 1 ␥ 2 Ͼ Ͼ ␤ 1 ␥ 1 .

DISCUSSION
In this study we adapted a published in situ receptor reconstitution method utilizing membranes from cells expressing recombinant GRP-R, NMB-R, BRS-3, or 4⌬BRS-3 which have been extracted with 7 M urea to remove endogenous GTPbinding proteins. The urea extraction procedure yielded a homogenous population of uncoupled receptors with 100% recovery of receptor ligand-binding sites. Such receptor preparations were functional when reconstituted with heterotrimeric G protein subunits as shown by the assay measuring the first biochemical event in G protein activation: receptor-catalyzed exchange of GTP for GDP on a G␣ subunit. The in situ receptor reconstitution technique has been used successfully in our earlier studies using membranes from baculovirus-infected Sf9 cells expressing 5-HT 2c receptor (13), membranes from stably transfected fibroblast cells expressing GRP-R (14), and now NMB-R and BRS-3. We believe that this method should be applicable to study virtually any receptor-G protein coupling. The major limitation we have observed is the receptor abundance in the membrane fraction.
We found all three bombesin receptors selectively coupled with a G␣ q but not G␣ i/o or G␣ t . However, three similar receptors were different in coupling selectivity toward members of the G␣ q family. Although GRP-R and NMB-R coupled to both squid and mouse G␣ q , GRP-R had a much stronger preference for squid G␣ q , and NMB-R showed only a slight preference for squid G␣ q . In contrast to GRP-R and NMB-R, the structurally related BRS-3 did not couple with squid G␣ q , whereas it clearly coupled with mouse G␣ q . Given that the differences between squid G␣ q and mouse G␣ q structures are not much greater than the ones among mouse G␣ q subtypes themselves (34 -36), it will be interesting to investigate coupling of the bombesin receptors to various G␣ q family subtypes within the same species in future studies.
The kinetic analysis of receptor-G protein interactions presented in this report also revealed a quantitative difference between GRP-R and NMB-R. The controlled, independent manipulation of receptor and G protein subunit concentrations required for this analysis is not possible using a whole cell system or prior reconstitution methods using purified, detergent-solubilized receptors and G proteins in phospholipid vesicles. GRP-R and NMB-R, although similar in their selectivity for G␣ q and rank-order preference for G␤␥ in the receptor-G  protein coupling, were different in the catalytic activity toward G␣ q and affinities for G proteins. GRP-R showed higher catalytic activity on squid G␣ q and higher affinities for both G␣ q and G␤␥ dimers than NMB-R. These results may partially explain an ambiguity noted in antisense oligonucleotide experiments in which individual G␣ subunits were depleted (18). In those experiments, Xenopus laevis oocytes expressing either GRP-R or NMB-R were microinjected with antisense phosphorothioate oligonucleotides complementary to specific regions of either Xenopus G␣ q or G␣ 11 to deplete selectively G␣ q or G␣ 11 protein. Following application of agonist, the activity of the calcium-activated chloride channel was measured under whole cell voltage clamp conditions. These experiments showed that treatment with the G␣ q antisense oligonucleotides could inhibit up to 74% of the response of the NMB-R but had no effect on the GRP-R response. G␣ 11 antisense, on the other hand, had little effect on either GRP-Ror NMB-R-mediated responses. The data reported here showed GRP-R coupled more effectively with squid G␣ q than with mouse G␣ q . Squid G␣ q is 74 -78% identical to mouse G␣ q , G␣ 11 , and G␣ 14 . As the sequence identity between mouse G␣ 14 and G␣ q or G␣ 11 is 80 or 81%, respectively, it is likely that GRP-R couples primarily with G␣ 14 instead of G␣ q or G␣ 11 . It is also possible that due to the higher affinity as well as higher catalytic activity for G␣ q , it would be easier to observe the influence of G␣ q depletion on NMB-R-regulated response than on GRP-R response. In those experiments, the antisense depletion taking place might simply fail to reduce G␣ q to a level that would impair GRP-R response.
GRP-R and NMB-R not only showed selective coupling with G␣ q but also showed a clear discrimination between different ␤␥ dimers. We provide two arguments that this result suggests the different ␤␥ dimers have different affinity and/or efficacy for bombesin receptors, rather than reflecting different affinities of G␣ q for ␤␥ dimers. First, instead of a uniform difference between G␤␥ dimers, GRP-R and NMB-R had different K 1/2 ratio for ␤ 1 ␥ 1 and ␤ 1 ␥ 2 (8-fold versus 4.6-fold). Second, rat 5-HT 2c receptor has also been shown to couple with squid retinal G␣ q in the in situ reconstitution assay (13). But unlike GRP-R or NMB-R, it has low affinity for both bovine brain ␤␥ (estimated K 1/2 is about 600 nM) and bovine retinal ␤␥, i.e. ␤ 1 ␥ 1 FIG. 7. G␤␥ coupling preference of the GRP-R and NMB-R. Urea-extracted membranes providing a final concentration of 1.25 nM GRP-R (A) or 2.5 nM NMB-R (B) were mixed with 100 nM squid G␣ q , 1 M agonist 697, and the indicated concentrations of ␤␥ dimers. BB␤␥ is the phenyl-Sepharose isolated ␤␥ fraction from bovine cortex; ␤ 1 ␥ 1 is the bovine retinal ␤␥, and ␤ 1 ␥ 2 is the recombinant dimer from baculovirus-infected Sf9 cells. G␤␥ dimer concentrations were determined by Amido Black staining. GTP␥S binding reactions proceeded for 10 min at 30°C, and binding was determined as described under "Experimental Procedures." The values presented are the means of triplicate determinations (bars, S.D.), and the results are representative of three independent experiments.
FIG. 8. G␤ 1 ␥ 1 and G␤ 1 ␥ 2 saturation of GRP-R-and NMB-Rcatalyzed GTP␥S binding. Varying concentrations as indicated of G␤ 1 ␥ 1 and G␤ 1 ␥ 2 were included in reactions containing 100 nM squid retinal G␣ q and 1.0 nM GRP-R with 1 M GRP (A) or 1.4 nM NMB-R with 1 M NMB (B). The GTP␥S binding reactions proceeded for 10 min at 30°C, and bound GTP␥S was determined as described under "Experimental Procedures." The values presented are from single determinations. The lines drawn are the best-fit curves for single site saturation. The results are representative of three to six independent experiments. (13). Other receptor-G protein coupling studies also supported the notion that receptors can have different affinities for ␤␥ dimers. Studies of bovine rhodopsin activation of ␣ t have found differences in apparent affinity among tissue-derived ␤␥ dimers of defined compositions or recombinant ␤␥ dimers, whereas ␣ t shows essentially no preference for ␤␥ dimers (31)(32)37). The fact that both GRP-R and NMB-R preferred bovine brain ␤␥ over ␤ 1 ␥ 2 and the diverse composition of bovine brain ␤␥ dimers (37,38) suggest there may be other ␤␥ dimer(s) having higher affinity or/and efficacy than ␤ 1 ␥ 2 in enhancing the catalytic activity of GRP-R and NMB-R on G␣ q .
In situ receptor reconstitution has been proven to be a useful methodology for detailed kinetic analysis of receptor-G protein coupling. It allows the identity and concentration of each coupling component to be defined and manipulated, while preserving the receptors in their native phospholipid environment. By using this method, we established a significant quantitative difference between GRP-R and NMB-R for interaction with the same squid G␣ q and G␤␥ proteins and, in addition, a qualitative difference between those two receptors and BRS-3 which did not interact with that same G␣ q . Combining the currently available high expression systems (e.g. baculovirus infection of insect Sf9 cells, transfection of mouse fibroblast cells) with the in situ receptor reconstitution technique, it should be feasible to study functional coupling between any recombinant receptor and G protein subunits, advancing our understanding of the molecular mechanisms governing the signal transduction pathway for G protein-coupled receptors.