The AT2 Receptor Selectively Associates with Giα2 and Giα3 in the Rat Fetus

The effects of angiotensin II are mediated by a family of seven transmembrane receptors. In the adult, the majority of the receptors are of the AT1 isoform, which is coupled to heterotrimeric G proteins (either Gqα or Giα). In contrast, the AT2 receptor is expressed at low levels in the adult but is the major form expressed in the fetal and neonatal animal. Previous results have failed to show G protein coupling of the AT2 receptor in the fetus. We now provide evidence that the AT2 receptor is G protein-coupled. An antibody that binds several Gα subunits immunoselected angiotensin II receptor-Gα complexes. In addition, Giα1-3 antibody, which recognizes Giα1, Giα2 and Giα3, also co-immunoselect the AT2 receptor. Anti-Giα2 and anti-Giα3 antibodies were both able to co-immunoselected AT2 receptor-Giα complexes, but consistent with the lack of Giα1 in the fetal extracts, anti-Giα1 antibodies did not nor did any other G protein-directed antisera. The finding that AT2 receptor couples to both Giα2 and Giα3 raises the possibility that selective interactions between AT2 receptor and different G proteins may result in specific cellular effects mediated by AT2 stimulation.

The effects of angiotensin II are mediated by a family of seven transmembrane receptors. In the adult, the majority of the receptors are of the AT 1 isoform, which is coupled to heterotrimeric G proteins (either G q␣ or G i␣ ). In contrast, the AT 2 receptor is expressed at low levels in the adult but is the major form expressed in the fetal and neonatal animal. Previous results have failed to show G protein coupling of the AT 2 receptor in the fetus. We now provide evidence that the AT 2 receptor is G protein-coupled. An antibody that binds several G ␣ subunits immunoselected angiotensin II receptor-G ␣ complexes. In addition, G i␣1-3 antibody, which recognizes G i␣1 , G i␣2 and G i␣3 , also co-immunoselect the AT 2 receptor. Anti-G i␣2 and anti-G i␣3 antibodies were both able to co-immunoselected AT 2 receptor-G i␣ complexes, but consistent with the lack of G i␣1 in the fetal extracts, anti-G i␣1 antibodies did not nor did any other G proteindirected antisera. The finding that AT 2 receptor couples to both G i␣2 and G i␣3 raises the possibility that selective interactions between AT 2 receptor and different G proteins may result in specific cellular effects mediated by AT 2 stimulation.
Two Ang II 1 receptor subtypes (AT 1 and AT 2 ) have been defined (1,2). Most known physiological actions of Ang II are mediated by the AT 1 receptor (1)(2)(3)(4). On the other hand, the AT 2 receptor remains an enigma. Cloning of AT 2 receptor by our laboratory and others (5)(6)(7)(8) has revealed that like the AT 1 receptor, the AT 2 receptor also possesses a seven transmembrane domain structure, which is similar to that of the heterotrimeric guanine nucleotide binding protein (G protein)-coupled receptors (9). However, the biochemical association of this receptor to this class of protein has not been demonstrated.
Heterotrimeric G proteins, comprised of ␣, ␤, and ␥ subunits, function as intermediates that couple cell surface receptors to intracellular effectors (10,11). Much of the specificity of receptor action is dictated by the ␣ subunit to which it is coupled. This subunit can be grouped into four major classes, each comprised of multiple members that are related by sequence homologies (␣s (␣s and ␣olf), ␣i (␣i1,␣i2 and ␣i3, ␣o, ␣t1, ␣t2, ␣gust and ␣z), ␣q (␣q, ␣11, ␣14, ␣15, and ␣16) and the ␣12 (␣12 and ␣13)) (12). Interestingly, many seven-transmembrane receptors can, in fact, couple to several G proteins, increasing the complexity of the systems. Once thought to be only negative regulators of ␣ subunit activity, the ␤␥ subunits have also been shown to be linked to the activation of various intracellular pathways (10 -12).
In radioligand binding and in autoradiographic studies, the AT 1 receptor is known to be sensitive to GTP analogs (1,2,13), suggesting the coupling to G proteins. Stimulation of Ang II receptor in AT 1 -rich tissues results in calcium mobilization through G q/11␣ and modulation of cAMP levels through G i␣ (14 -16). In contrast to the AT 1 receptor, in most tissues examined, binding to the AT 2 receptor is not susceptible to GTP analogs (1,2,13,(17)(18)(19), which has led to the suggestion that the AT 2 receptor does not interact with G proteins (13). However, indirect evidence has suggested that AT 2 receptor may also couple to G proteins (20,21). In certain regions of the rat brain, binding to the AT 2 receptor was sensitive to GTP analogs and to pertussis toxin treatment (22). Similarly, in PC12w cells, the AT 2 receptor-mediated inhibition of phosphotyrosine phosphatase was reversed by pertussis toxin treatment (23). More recently, Kang et al. (24) reported that in vivo in cultured rat neurons, AT 2 receptor-mediated stimulation of delayed rectifier K ϩ current (I k ) could be abolished by intracellular application of an anti-G i␣ antibody.
The above data demonstrating sensitivity (or insensitivity) of radioligand binding to GTP␥S and pertussis toxin are indirect evidence for G protein coupling (or uncoupling). However, these approaches suffer from several caveats. Although the work of Kang et al. (24) is more direct, the use of cultured cells, which require several weeks of culture in order to develop an AT 2 receptor-mediated action, is also problematic. Therefore, the purpose of the present work was to demonstrate a direct, biochemical association of AT 2 receptor and its coupled G proteins in the rat fetus. The rat fetus was chosen because the AT 2 receptor is expressed at extraordinarily high levels in many fetal tissues.
Radioligand Binding Assay-Membrane (10 g of protein) or immunoselected fractions were incubated with 2 nM 125 I-Ang II (2,176 Ci/mM) at room temperature for 60 min in 100 l of 20 mM Tris-HCl, pH 7.4, 10 mM MgCl 2 , 0.25% bovine serum albumin. Bound and free ligands were separated by rapid filtration through GF/B Whatman glass fiber filters. For solubilized membranes, the filters were presoaked in 0.5% polyethylenimine. The filters were washed three times, and the bound radioligand was quantitated. Each point was performed in duplicate. Non-specific binding was defined as radioactivity bound in the presence of 1 M unlabeled ligand. For characterization of AT 2 receptor subtypes, the selective AT 1 receptor antagonist DuP 753 (losartan) and the selective AT 2 ligands CGP42112A or PD123319 (each at 1 M) were employed.
Western Blots-Solubilized membrane samples in SDS-polyacrylamide gel (PAGE) buffer were incubated in boiling water bath for 5 min. Approximately 30 g of protein was subjected to 10% SDS-PAGE and transferred onto nitrocellulose. Immunoblotting was carried out according to ECL Western blotting protocol (Amersham Life Science Inc.).
Immunoselection of AT 2 Receptor-G Protein Complexes-Solubilized AT 2 receptor preparations (1 ml of a 1 mg/ml solution) diluted in solubilization buffer containing 0.3% CHAPS were incubated with anti-G protein antibodies or antisera overnight at 4°C with constant shaking (amounts of the antibody or antisera are indicated in the figure legends). Protein A-agarose (60 l of 50% w/v) was then added and incubated with rocking for an additional 4 h at 4°C. The samples were centrifuged at 14,000 rpm. The supernatant was removed, and the immunoselections were washed three times with 1 ml of ice-cold solubilization buffer containing 0.3% CHAPS. The AT 2 receptor in the immunoselection and in the supernatant was detected by radioligand binding assays as described above. In some experiments, 10 mM NaCl or 10 mM NaF plus 10 M AlF 3 were added to the solubilized membrane preparations at the time of the immunoselection.
[ 35 S]GTP␥S Binding Assay-[ 35 S]GTP␥S binding was measured with a modification of assay described by Johnson and Corbin (26). Immunoselections were mixed with incubation buffer (20 mM Tris-HCl, 25 mM MgCl 2 , 1 mM EDTA, 100 mM NaCl, and 1 mM dithiothreitol, pH 8) and 5 nM [ 35 S]GTP␥S (1, 332.0 Ci/mmol) in 100 l. The binding was initiated by the addition of the immunoselections to the incubation buffer and was carried out for 1 h at 30°C. The reactions were terminated by rapid filtration through GF/B Whatman glass fiber filters. The filters were washed three times with 5 ml of ice-cold buffer (20 mM Tris-HCl, pH 8.0, 25 mM MgCl 2 , 100 mM NaCl) and then quantitated. Nonspecific binding was determined in the presence of 1 mM GTP and subtracted from total bound radioactivity.
ADP-ribosylation of Solubilized and Immunoselected Proteins with Pertussis Toxin-Pertussis toxin was preactivated by incubation with 20 mM dithiothreitol, 0.125% SDS and 1 mg/ml BSA for 30 min at 30°C. Solubilized membrane (30 g) or immunoselections were incubated with pertussis toxin (10 g/ml) and 5 Ci of [ 32 P]NAD in the presence or in the absence of 100 M NAD for 1 h at 30°C, in a final volume of 100 l of reaction buffer (pH 7.4, 1 mM EDTA, 10 mM thymidine, 10 mM Hepes, 0.5 mM ATP, 0.1 mM GTP) (27,28). The proteins were loaded into 10% SDS-PAGE gels. Gels were stained, dried, and autoradiographed at Ϫ70°C for 5-24 h.
Materials-125 I-Ang II (2, 176 Ci/mmol) was purchased from Robert Speth at Washington State University. Dithiothreitol, GTP␥S, and CHAPS were from Sigma. Losartan was provided by Ronald Smith of DuPont-Merck. PD123319 was provided by Joan Keiser at Parke-Davis. CGP42112A was provided by Marc de Gasparo of Ciba-Geigy. ECL Western blotting detection reagents was from Amersham International Plc. Antisera sources are listed in Table I. (29 -33).
Calculations and Statistics-Data are presented as mean values Ϯ S.E. Student's t test for unpaired observation and one-way analysis of variance were used for statistical evaluation. p Ͻ 0.05 was considered significant. Specific binding was calculated by subtracting nonspecific binding measured in the presence of unlabeled ligand from total binding. Saturation analysis were performed using nonlinear regression curve-fitting. Dissociation constant K d and B max were calculated from Scatchard plot using a linear regression program (Statview).

Pharmacological Analysis of AT 2 Receptor in Rat Fetus
Membrane Preparations-Binding of 125 I-Ang II to membranes prepared from 18-day-old rat fetus was of high affinity, saturable, and linearly proportional to the protein concentrations tested (1-60 g, data not shown). Scatchard analysis revealed a single class of binding sites (Fig. 1A), yielding a K d of 1.89 Ϯ 0.11 nM and B max of 1.91 Ϯ 0.10 pmol/mg protein (n ϭ 3). Nonspecific binding was determined in the presence of 1 M CGP42112A and was less than 15% of total binding at the K d of binding. Binding of 125 I-Ang II to the membrane preparation was competed by AT 2 selective ligand CGP42112A (91 Ϯ 3.7% displacement at 1 M, n ϭ 3) and PD123319 (78 Ϯ 3.1% displacement at 1 M, n ϭ 3) but was not sensitive to the AT 1selective antagonist DuP 753 (7 Ϯ 2.1% displacement at 1 M, n ϭ 3). (Fig. 1B).
We next tested the ability of detergents to solubilize the fetal membranes and release AT 2 receptors, which retain binding capacity. As shown in Fig. 1C, optimal solubilization of AT 2 receptor was achieved with 0.5-1% CHAPS, which yielded 70 -60% of specific 125 I-Ang II binding and 50 -60% of total protein. Because solubilization of AT 2 receptor with 1% CHAPS resulted in less nonspecific binding but higher AT 2 receptor binding in immunoselections (data not shown), 1% CHAPS was used in subsequent experiments. 125 I-Ang II binding to solubilized membrane preparations was specific and saturable, with moderate decrease in the affinity and B max (Fig. 1A, K d ϭ 3.46 Ϯ 0.12 nM, B max ϭ 1.33 Ϯ 0.11 pmol/mg protein.). Competition studies with receptor specific ligands (losartan, PD123319, and CGP42112A) and nonsubtype selective ligands (Sar-1, Ile-8 Ang II, and Ang II) exhibited an inhibitory pattern similar to that seen in crude membrane preparation (Fig. 1B) and consistent with the pharmacology of the AT 2 receptor.
Presence of G Protein ␣ and ␤ Subunit Immunoreactivity in Solubilized Rat Fetus Membrane Preparation-Prior to an ex- amination of the coupling of the AT 2 receptor to specific G proteins, we first assessed the expression of the various G proteins in the solubilized 18-day-old rat fetus membranes using Western analysis. A majority of the anti-G protein antibodies tested yielded strong signals with appropriate molecular sizes (Fig. 2). Specific immunoreactivity was detected at 40 -41 kDa with G i␣2 and G i␣3 antibodies. Anti-G q/11␣ antibody detected proteins of molecular masses of 42-43 kDa. Anti-G z␣ antibody recognized a major band at 41 kDa. Anti-G ␤common recognized a band at approximately 35-36 kDa. A weak band at 39 kDa was detected by anti-G o␣ antibody. However, several different anti-G i␣1 antibodies as well as antibodies specific for G 12␣ and G 13␣ failed to detect protein, suggesting the low expression of these proteins in the whole rat fetus. Control immunoblots using IgG were negative (Fig. 2).

Immunoselection of [ 35 S]GTP␥S Binding Proteins with Anti-G Protein
Antibodies-In order to test the immunoselection procedure, the presence of GTP binding proteins in the immunoselections was examined by measuring the specific binding of [ 35 S]GTP␥S. Binding of [ 35 S]GTP␥S to both solubilized membrane preparation and to immunoselections was concentrationdependent and specific (data not shown). Under our experimental conditions, nonspecific binding defined in the presence of 1 mM of GTP was less than 10%. As shown in Fig. 3, polyclonal antibodies raised against several G proteins (G i␣1-3 , G z␣ , and G q/11␣ ) were tested and shown to immunoselect, in a concentration-dependent manner, GTP binding proteins. Similar numbers of GTP binding sites were immunoselected with each antibody. The percentage of total GTP binding sites in the solubilized membrane that was immunoselected with the antibodies specific for G i␣1-3 , G z␣ , and G q/11␣ were 1.8, 1.4, and 1.2%, respectively. This must be considered as an underestimate because solubilized membrane has been reported to contain GTP binding activities distinct from G ␣ proteins (34).  Immunoblotting of ␤ and various ␣ subunits of G proteins in solubilized rat fetus membrane preparation. Solubilized membranes (30 g protein) were size-separated by SDS-PAGE and blotted to nitrocellulose membrane. G ␣ and G ␤ proteins were immunodetected using anti-G ␤common (SC-378) and various specific anti-G ␣ antibodies, respectively (Asano, anti-G i␣1 ; Asano, anti-G i␣2 ; 371729, an-tiG i␣3; SC-387, anti-G o␣ SC-388, anti-G za; SC-392, anti-G q/11␣; SC-409, antiG 12␣ ; and SC-410, antiG 13␣ ). IgG (0.1 g/ml) was used as control. The antibody-G protein complex was detected with horseradish peroxidase-conjugated donkey anti-rabbit antibody (1:2,500) and ECL-reagent per instructions. All primary antibodies were used at a final dilution of 1:1,000 (0.1 g/ml). Molecular mass size markers (kDa) are indicated to the left. Shown is a representative example of two separate experiments.
A second method was used to test the immunoselection procedure. Because G i␣1-3 are known to be sensitive to pertussis toxin, the presence of pertussis toxin-sensitive G proteins was also investigated. Labeling of both solubilized membrane preparation and SC-262-immunoselected fractions but not IgG-immunoselected fractions with 32 P-NAD in the presence but not in the absence of pertussis toxin resulted in a radiolabeled band corresponding to molecular mass of approximately 40 -41 kDa (Fig. 4). This band was not observed when 100 M NAD was added, suggesting pertussis toxin-specific reaction of ADP-ribosylation (Fig. 4).
Immunoselection of AT 2 Receptor-G Protein Complexes with Anti-G Protein Antisera-We next tested the ability of antibodies raised against the G proteins to co-select angiotensin II binding sites. Initially, we used an anti-G ␣common antibody (SC-386), which interacts with G i␣1-3 , G o␣ , G z␣ , and G t␣ . The results demonstrate the selection of a low level of Ang II binding sites that were not observed with IgG (Fig. 5A). Next, antibodies more selective against specific G proteins were tested. Consistently, anti-G i␣1-3 (SC-626) concentration-dependently co-immunoselect the solubilized Ang II binding sites (Fig. 5B). Interestingly, this antibody selected a higher amount of Ang II binding sites compared with G ␣common antibody. To address specificity, SC-262P, the peptide to which SC-262 was raised completely blocked the ability of this polyclonal antibody to immunoselect the Ang II binding sites. SC-262P itself did not interfere with 125 I-Ang II binding (Fig. 5C). No angiotensin II binding sites were co-immunoselected with control IgG. As shown in Fig. 5D, binding of 125 I-Ang II to SC-262-immunoselected samples was of high affinity and yielded saturation binding curves comparable with those observed with solubilized membrane preparations. Scatchard analysis revealed that the K d of the immunoselected binding sites were comparable with that observed for the binding sites in the solubilized membrane fraction. In contrast, no specific Ang II binding sites were detected in immunoselections by anti-G ␤ , anti-G z␣ , anti-G o␣ , anti-G 13␣ , anti-G 12␣ , and anti-G q/11␣ antibodies (Fig. 6), even though anti-G q/11␣ and anti-G z␣ were almost as effective in concentration-dependently immunoselecting GTP binding proteins (Fig. 3).
These Ang II binding sites represented AT 2 receptors. In experiments where specific binding was defined independently with Sar-1, Ile-8 Ang II, or CGP42112, the compounds were shown to compete to similar levels (Fig. 7). Moreover, DuP 753 was not able to compete 125 I-Ang II binding to the immunoselected material. In addition, binding with 125 I-CGP42112 was performed in selective experiments and demonstrated results comparable with those obtained with the 125 I-Ang II (data not shown).
The above results suggest the close association between the AT 2 receptor and G i␣ , which may be an actual coupling of the receptor to this G protein. Alternatively, this association may only indicate the presence of these proteins within the same mixed micelles, which are formed during the solubilization step. The latter explanation is unlikely because antibodies specific for several other G proteins failed to immunoselect AT 2 binding sites; the proteins in the mixed micelles should be randomly distributed. However, to provide further and more direct data concerning this point, we examined the ability of aluminum fluoride, which is known to disrupt G protein-receptor complexes, to inhibit the co-immunoselection of the AT 2 binding sites by the anti-G i␣ antibody. The addition of NaF plus AlF to the immunoselection decreased the ability of the anti-G i␣ antibody to immunoselect AT 2 binding sites by 38% (specific binding obtained with anti-G i␣ antibody, 1200 Ϯ 66; anti-G i␣ antibody plus NaF plus AlF; 740 Ϯ 98) However, NaCl at the same concentration was without effect. Thus, taken together, the results are consistent with a coupling of the AT 2 receptor and G i␣ .
To determine which subtype of G i␣ was involved, several anti-G i␣1 -, G i␣2 -, and G i␣3 -specific antibodies were tested. An anti-G i␣2 antibody (J-883) co-immunoselected AT 2 receptor-G i␣ complexes but to an extent less than that selected with anti-   (Fig. 8). Similarly, an anti-G i␣3 antibody (NEI-803) also co-immunoselected AT 2 receptor, again to an extent less than that observed with anti-G i␣1-3 (Fig. 8). Several other anti-G i␣ subtype antibodies yield similar results (G i␣2 (1521), 670 Ϯ 93 cpm; G i␣2 (Asano), 675 Ϯ 83 cpm; G i␣3 (06 -270), 504 Ϯ 67 cpm; G i␣3 (371729), 431 Ϯ 84 cpm). We next tested the ability of anti-G i␣2 and anti-G i␣3 antibodies to immunoselect Ang II binding sites in sequential experiments. Previous immunoreaction of the membrane preparation with anti-G i␣3 (NEI-803) antibody did not affect the ability of anti-G i␣2 (J-883) antibody to immunoselect Ang II binding sites. Similarly, previous immunoreaction of the membrane preparation with anti-G i␣2 (J-883) antibody did not affect the ability of anti-G i␣3 (NEI-803) antibody to immunoselect Ang II binding sites. On the other hand, prior treatment with anti-G i␣1-3 (SC-262) antibody abolished the subsequent immunoselection with anti-G i␣1-3 antibody (Table II).

DISCUSSION
Based on the ability of highly specific antibodies raised against the G ␣ subunits to co-immunoselect AT 2 binding sites, we conclude that the AT 2 receptor is G protein-coupled, at least in the rat fetus. The specificity of this approach was provided by the fact that control IgG or nonimmune sera failed to select Ang II binding sites. Importantly, antibodies against G q/11␣ and G z␣ also failed to immunoselect Ang II binding sites, despite their ability to immunoselect comparable numbers of GTP␥S binding sites. The immunoselected Ang II binding sites were of high affinity with saturation curve similar to that observed with solubilized membranes.
Ang II acts at multiple sites within the cardiovascular system to assist in the regulation of cardiovascular homeostasis (1,2). These effects are mediated by high affinity Ang II binding sites, which, at least in the adult, have been shown to be of the AT 1 subtype (1, 2). This receptor is a seven transmembranespanning receptor (35) that is coupled primarily to G q␣ resulting in the activation of phospholipase C and the subsequent release of IP 3 and diacyl glycerol, leading to increased intracellular calcium and activation of protein kinase C (14 -16). Alternatively, this receptor is coupled to G i␣ , resulting in the inhibition of adenylyl cyclase activity (14,15). Activation of the AT 1 receptor induces a variety of actions depending on the tissue, exhibiting specific adrenal, renal, and vascular effects (1)(2)(3)(4).
In contrast to the AT 1 receptor, the AT 2 receptor remains a mystery. Initially described on the basis of selective binding to specific ligands, this receptor was found to be expressed at very low levels in the adult and only in selective tissues (1,2). On the other hand, binding studies revealed that this receptor was highly expressed in embryonic, fetal, and neonatal tissues, which lead to the speculation that this receptor may mediate some undefined action of Ang II in the processes associated with growth, development, or differentiation (18,36,37). Ang II has been shown, both in culture as well as in vivo, to induce growth and alter cellular phenotype (39,40). However, the ability of AT 1 -specific antagonists to block these effects of Ang II, at least in the adult, argued against this role for the AT 2 receptor (41,42). Recently, we have found that the AT 2 receptor mediates an antigrowth effect, counteracting the growth promoting effects mediated by the AT 1 receptor (43). These results were obtained in a variety of models and may depend on the AT 2 -mediated decrease in MAP kinase activity.
The discovery that the AT 2 receptor shared the structural characteristics of the serpentine receptor family strongly suggested that this receptor may couple to G proteins (5)(6)(7)(8). Indirect evidence, such as the ability of pertussis toxin to block certain actions ascribed to the AT 2 receptor, was consistent with this suggestion (20 -23). In the course of our study, we became aware of the work of Kang et al. (24), who demonstrated that the AT 2 receptor mediated an activation of a po- FIG. 8. The AT 2 receptor is coupled to both G i␣2 and G i␣3 . Solubilized rat fetal membranes (1 mg/ml) were incubated with antiG i␣1 , G i␣2, antiG i␣3 , antiG i␣1-3 (10 g), nonimmune serum (NI, 10 l), or IgG (10 g) overnight at 4°C. Antibodies were then collected with protein A-agarose. The immunoselections were incubated with 125 I-Ang II (2 nM) for 1 h at room temperature. Nonspecific binding was defined in the presence of 1 M CGP42112A. Data are presented as means Ϯ S.E., n ϭ 3. The antiG i␣2 (J-883) and antiG i␣3 (NEI-803) antibodies co-immunoselected AT 2 -G i␣ complexes but to an extent less than that selected with antiG i␣1-3 (SC-262). Several other anti-G i␣ subtype antibodies yield similar results (see text). *, p Ͻ 0.05 compared with nonimmune serum. †, p Ͻ 0.05 compared with IgG control. tassium current in cultured neonatal rat neurons. Intracellular delivery of an anti-G i␣ antibody recognizing all of G i␣1 , G i␣2 and G i␣3 by micropipette abolished this current, suggesting a role for G i␣ . Although an elegant and well controlled approach, the intracellular delivery of antibodies suffers from certain caveats, the major being a lack of this approach to demonstrate direct receptor-G protein interaction. Moreover, the ability of Ang II to induce this potassium current is not found in freshly isolated neurons but only in cells cultured for more than 12 days. Thus the possibility that this observation represented what could happen and not necessarily what does happen in vivo remained. Our direct evidence of an AT 2 receptor-G i␣1-3 interaction is therefore complimentary to their approach.
What are the intracellular signals induced by the AT 2 receptor by way of G i␣ (and associated ␤␥)? It has been shown that all three G i␣ subtypes (G i␣1 , G i␣2 , and G i␣3 ) are able to stimulate the G protein-gated atrial potassium channel and to inhibit adenylyl cyclase (44 -46). The activation of the neuronal potassium current is consistent with the idea that G i␣ is involved as is the evidence that pertussis toxin will block certain actions of the receptor. Our evidence for an antiproliferative action mediated by the AT 2 receptor is harder to explain. Other results from our laboratory demonstrate that the activation of AT 2 receptor leads to a decrease in growth and an inhibition of mitogen-activated protein kinase activity (43). The results in the current study demonstrate the coupling (or the physical association) of the receptor to G i␣ ; especially G i␣2 presents a conundrum. An increasing body of evidence suggests that G i␣2 stimulates growth via a stimulation of the mitogen-activated protein kinase pathway (47)(48)(49). It could be argued that the association of the AT 2 receptor and G i␣2 may not occur in vivo but is an artifact of the extraction procedure exists; however, we have recently obtained immunohistochemical evidence that at least in the fetal aorta, the AT 2 receptor and G i␣2 are expressed in the same area of the vessel wall. 2 Although compartmentalization of receptors and G proteins in different membrane microdomains exist in some cells (38), such compartmentalization has not been described in smooth muscle but is usually associated with epithelial cell types. Thus, we are left without a suitable explanation. One potential explanation is that in whole cells, the receptor is more strongly associated with the other ␣ subtypes, G i␣3 or perhaps even G i␣1 . Another is that the receptor-G i␣2 interaction involves a unique ␤␥ subunit that is actually mediating the anti-growth response, overriding the stimulating effect of G i␣2 . Other explanations may involve the inhibition of G i␣2 activity following stimulation of the AT 2 receptor or the association of the AT 2 receptor with a novel G i␣ , which cross reacts with the anti-G i␣ antibodies. Clearly, this area requires further study.
The direct evidence for the association of the AT 2 receptor and G i␣ brings up another issue, that of the lack of sensitivity to GTP analogs such as GTP␥S. It is known that G proteins cycle between an inactive (GDP bound form) and active (GTP bound) states (10 -12). The GDP bound-␣ subunit associates with ␤␥ subunits to form the heterotrimer, which is able to associate with ligand-free receptors. Receptors coupled to ␣␤␥ hetrotrimers usually bind ligand with high affinity. When the receptor is activated by agonist it undergoes a conformational change that is transmitted to the ␣ subunit, promoting the exchange of GDP for GTP. This causes a dissociation of the ␣ subunit both from the receptor and from ␤␥ and frees the ␣ subunit to interact with its appropriate effector. Receptors not associated with ␣␤␥ hetrotrimers usually exhibit a lower affinity for agonist. Once GTP is cleaved to GDP, the ␣ subunit ceases its effector stimulation, reassociates with ␤␥, and subsequently can reassociate with the receptor. Substitution of GTP with a nonhydrolyzable analog such as GTP␥S or Gpp(NH)P forces the ␣ subunit to remain in its active form. Subsequently, no reassociation with ␤␥ and receptor occurs, and the affinity of the receptor thus remains in a low affinity state. The only evidence for a GTP␥S shift in affinity state of the AT 2 receptor has been obtained in radioligand autoradiographic studies or certain brain nuclei (20,21); all other published studies using whole cells or membrane fractions have been negative (1,2,13,(17)(18)(19). This inability of GTP␥S to shift the AT 2 receptor to a low affinity state has often been cited as evidence that the receptor in not G protein-coupled. The inability of GTP␥S to induce a low affinity shift of AT 2 receptors in membranes from whole rat fetus may suggest that a majority of the receptors are not G protein-coupled. Consistent with this, the co-immunoselection of the receptor using anti-G protein antibodies was inefficient, yielding only a small fraction of the total AT 2 binding sites. Alternatively, the AT 2 receptor may couple to G proteins by a nonconventional mechanism or may couple to a G protein with a low rate of GDP-GTP exchange. Consistent with this latter possibility, in our study, GTP␥S did not influence binding to the immunoselected AT 2 receptor nor did it inhibit the immunoselection of the receptor by the G protein antibody, whereas AlF, which binds to the GDP bound G protein, did attenuate the ability of G protein antibody to co-immunoselect AT 2 binding sites.
In conclusion, co-immunoselection AT 2 receptor with anti-G protein antibodies proved useful in identification of direct association of AT 2 receptor and its coupled G proteins. The present study provide the first biochemical evidence that AT 2 receptor is G protein-coupled. The demonstration that AT 2 receptor is able to couple to both G i␣2 and G i␣3 may enhance our understanding of the signal transduction pathways underlying the physiological functions mediated by AT 2 receptor stimulation. 2 H. Hutchinson, J. Zhang, and R. E. Pratt, unpublished observations.

TABLE II Association of the AT 2 receptor with both G i␣2 and G i␣3
Round I experiments were performed by incubating solubilized rat fetus membrane with anti-G i␣2 anti-G i␣3 or anti-G i␣1-3 or their appropriate controls (nonimmune serum, control IgG) overnight at 4°C. Antibody-G protein complexes were then collected with 60 l (50% w/v) protein A-agarose. The immunoselections were incubated with 125 I-Ang II (2 nM) for 1 h at room temperature. Nonspecific binding was defined in the presence of 1 M CGP42112A. The same procedure was followed with round II, except that the supernatants of round I were used. The data are presented as means Ϯ S.E. (n ϭ 2). NI, nonimmune serum.