Activation of the Inhibitory GTP-binding Protein of Adenylate Cyclase, Gi, by @-Adrenergic Receptors in Reconstituted Phospholipid Vesicles*

beta-Adrenergic receptors and the inhibitory GTP-binding protein, Gi of the adenylate cyclase system were reconstituted into phospholipid vesicles by the method described previously for reconstituting receptors and the stimulatory GTP-binding protein, Gs (Brandt, D. R., Asano, T., Pedersen, S. E., and Ross, E. M. (1983) Biochemistry 22, 4357-4362). In the receptor-Gi vesicles, beta-adrenergic agonists stimulated both the high-affinity binding of guanosine 5'-O-(3-thiotriphosphate) (GTP gamma S) to Gi and GTPase activity to an extent similar to that observed in vesicles containing beta-adrenergic receptors and Gs. Stimulation required receptors and displayed appropriate beta-adrenergic specificity. The prior treatment of receptor-Gi vesicles with islet-activating protein (pertussis toxin) plus NAD markedly inhibited both the isoproterenol-stimulated binding of GTP gamma S and the isoproterenol-stimulated GTPase activity. No contamination of Gi by Gs was apparent. These data suggest that receptors that typically stimulate adenylate cyclase activity may also activate the inhibitory system, perhaps as one mechanism of desensitization.

5 Present address, Department of Physiological Chemistry, Faculty of Pharmaceutical Sciences, Hokkaido University, Sapporo, Hokkaido, Japan.
7i Established Investigator of the American Heart Heart Association.
GTPyS,' activation of G. is reversed only slowly by the dissociation of nucleotide.
G, is composed of three subunits. The cy subunit (M, = 45,000 or 52,000) binds GTP and is capable of stimulating adenylate cyclase. The /3 subunit (Mr = 35,000) regulates the binding of nucleotide to the cy subunit and dissociates from 01 when G, is activated (2,3). A y subunit (Mr -8000) sometimes copurifies with G,. It is of unknown function and appears to remain associated with the ( 3 subunit when G, is activated (4). Hormonal inhibition of adenylate cyclase is a strikingly similar process. Inhibitory receptors promote the binding of G T P to an inhibitory regulatory protein, Gi, thereby activating it (1). Gi is structurally homologous to G,, having a homologous GTP-binding cy subunit (Mr = 41,000), a very similar or identical / 3 subunit (5,6), and a presumably identical y subunit (4). The activation of Gi also causes the cy and By subunits to dissociate. This has suggested that Gi-mediated inhibition primarily reflects an increased concentration of free by that promotes the reassociation of fir with the 01 subunit of G, to cause its deactivation (7,8). It is also likely that the 01 subunit of Gi causes inhibition of adenylate cyclase, although not necessarily directly (8)(9)(10).
Given the homology of stimulatory and inhibitory G proteins, it was of interest to determine whether G, or Gi could interact with a receptor from the opposite pathway. We have recently reconstituted purified G, and partially purified padrenergic receptors into unilamellar phospholipid vesicles in order to study their interactions directly (11, Reconstitution efficiently restores catecholamine-stimulated GTPase activity (1 1) and rapid catecholamine stimulation of GTPyS binding to G, and G. activation, allowing relatively detailed studies of these More recently, we have combined pure (>95%) receptors from turkey erythrocytes and pure G, from rabbit liver into phosphatidy1serine:phosphatidyleth-ano1amine:cholesterol vesicles (12). Receptor-G, coupling in this system indicates that the single receptor polypeptide (Mr = 43,000) can interact directly with G,. We have now used these procedures to probe the coupling of p-adrenergic receptors, which generally activate adenylate cyclase, to G,. We report here the efficient interaction of these proteins in reconstituted lipid vesicles.

MATERIALS AND METHODS
Most of the procedures used in this study have been described in detail previously ( l l ) . 2 3 3 G, and G. were purified from rabbit liver according to Bokoch et ai. (13) and Sternweis et al. (14), respectively. Purified Gi was free of the 01 subunit of G. according to silver-stained sodium dodecyl sulfate-polyacrylamide gels. No G. regulatory activity was found in the G, preparations after exhaustive MF-promoted activation in detergent solution by 50 FM GTPrS when assayed by reconstitution into G.-deficient cyc-membranes (14). @-Adrenergic receptors were purified about 3000-fold from turkey erythrocyte plasma membranes by affinity chromatography on alprenolol-agarose  Activation of Gi by &Adrenergic Receptors that catalyzes the ADP-ribosylation of Gi) was a gift of Dr. Michio Ui (Hokkaido University). [35S]GTPyS was purchased from New England Nuclear, and unlabeled GTPyS was purified as described (11). &Adrenergic receptors and either G. or G; were reconsituted into vesicles composed of dimyristoylphosphatidylcholine and turkey erythrocyte polar lipids as described previously (ll)? The recovery of Gi was 45 k 10% of that added to the original mixture (n = five preparations). The recovery of receptor was 27 k 2%. The molar ratio of Gi to receptor was 2-6 in the vesicle preparations used here.
The assays for [Y3]GTPyS binding (11); GTPase ( l l ) , and ['TI iodocyanopindolol binding (11) have been described, as has the assay for the activation of G. by GTPyS (16). Details are also provided in the figure legends. Other ancillary assays and procedures are described elsewhere (11, 16)?.3 To ADP-ribosylate Gi, reconstituted vesicles were incubated for 10 min at 30 "C with 25 pg/ml of IAP and 1 mM NAD in 20 mM Na Hepes (pH 8.0), 1 mM EDTA, 2 mM M e & , 100 mM NaC1, 1 mM dithiothreitol.

RESULTS
When either Gi or G, was reconstituted into vesicles with P-adrenergic receptors, the @-adrenergic agonist isoproterenol markedly stimulated the rate of quasi-irreversible binding of [35S]GTPyS to each G protein ( Fig. 1). In both cases, binding was rapid and the agonist-stimulated reaction displayed an apparent first-order rate constant of 1. to Gi was usually similar to that to G,, and no consistent difference between the rates of binding to the two G proteins has been noted. At short times, isoproterenol stimulated the binding of GTPyS to G. by about 6.5-fold and to Gi by about 3.6-fold in Fig. 1. This degree of stimulation was rather low. The average relative stimulation of Gi by isoproterenol was 5.6-fold (range = 3.6-12.4) in 10 experiments. It is our impression that Gi reconstitutes slightly less efficiently than does G, under the conditions used here. Thus, only about 35% of the total Gi in the vesicles could be stimulated by receptor, while 50% or more of reconstituted G, could usually be stimulated.
We do not know if this difference represents an important aspect of receptor-Gi coupling or is a consequence of the reconstitution protocol.
The data of Fig. 2 indicate that the stimulation of GTPyS binding to Gi displays ,&adrenergic specificity. Selectivity among catecholamine agonists was appropriate for prather than a-adrenergic effects, and appropriate stereoselectivity was observed. (-)-Propranolol was a potent antagonist, but a-adrenergic blockers were ineffective at inhibiting the stimulation of binding by isoproterenol.
Receptor-mediated stimulation of the binding of GTPyS to Gi was blocked efficiently if the vesicles were treated with NAD and IAP, a toxin from Bordetella pertussis that specifically ADP-ribosylates and inactivates Gi (17) (Fig. 3). The effect of IAP was greatest on the agonist-stimulated binding reaction, where inhibition was 82% of the isoproterenol-stimulated increment above basal. This is in contrast to results obtained using soluble Gi, where little or no effect of IAP treatment on GTPyS binding was observed. Inhibition of the basal GTPyS binding reaction in the vesicles was relatively slight. Treatment with IAP had no effect on the ability of the receptors to bind the &adrenergic ligand  Fig. 3 indicate that the @-adrenergic stimulation of GTPyS binding in the receptor-Gi vesicles is not to contaminating G, in the Gi or receptor preparations. As a further control, receptor-Gi vesicles were assayed for the adenylate cyclase stimulatory activity that is characteristic of G,. Receptor-Gi vesicles were incubated with isoproterenol plus 0.1 PM GTPyS at 30 "C for 5 min and then reconstituted into the G,-deficient plasma membranes of cyc-S49 lymphoma cells (16). In such experiments, the receptor-Gi vesicles contained no G, at a lower level of detection of 1.6 mol% of the Gi present, consistent with the results of dodecyl sulfate-polyacrylamide gel electrophoresis of the Gi used to prepare the vesicles.
Receptor-G, vesicles have been shown to display p-adrenergic catecholamine-stimulated GTPase activity ( l l ) , and the receptor Gi vesicles display this activity as well (Fig. 4). The molar turnover number of Gi-catalyzed G T P hydrolysis was 0.33 min-' in this experiment, somewhat lower than that observed in receptor-G, vesicles. In two other experiments, turnover numbers were calculated to be 0.27 and 0.29 min-'. The agonist-stimulated GTPase activity was inhibited 90% by treating the vesicles with IAP plus NAD. The basal GTPase activity was inhibited only slightly by IAP, as was observed in the GTPyS binding experiment shown in Fig. 3. Thus, a t a descriptive level, IAP appears to uncouple receptor from Gi in the vesicles, as is the case in native membranes (see Ref. 18).

DISCUSSION
The data presented indicate that P-adrenergic receptors, which generally stimulate adenylate cyclase via the GTPbinding protein, G,, can also stimulate the activation of the inhibitory GTP-binding protein, Gi. Stimulation was demonstrated as an increased rate of GTPyS binding and an increased GTPase activity caused by p-adrenergic agonists. The blockade of these effects by treatment with IAP and NAD, the purity of the Gi preparation, and the absence of any G, activity in the receptor-Gi vesicles indicate that the GTPase activity and GTPyS binding are due to Gi. The pharmacological specificity of stimulation indicates that observed effects are mediated via the p-adrenergic receptor.
It is tempting to speculate on the possible physiological significance of P-adrenergic activation of Gi. However, it is possible that the unexpected interaction of &adrenergic receptors and Gi might reflect the partial denaturation of one or both proteins, the nonphysiological phospholipid environment, the phylogenetic distance between turkeys and rabbits, or some other oddity of our reconstituted system that minimizes the differences between two highly homologous but not totally identical proteins.
Alternatively, the activation of Gi by receptors that are normally considered to stimulate adenylate cyclase might represent an important regulatory mechanism. Murayama and Ui (19) have presented data suggesting that P-adrenergic stimulation of rat adipocyte membranes can lead to the inhibition of adenylate cyclase and that this inhibition is blocked by treatment with IAP. Similar data have been presented recently for platelets by Jakobs et al. (20). Both reports are consistent with the &adrenergic stimulation of Gi. A different corroborative experiment has been performed by Abramson and M~l i n o f f .~ These authors found that P-adrenergic receptors in cyc-S49 lymphoma cell membranes bind the agonist [3H]hydroxybenzylisoproterenol with the high affinity that is usually ascribed to the receptor-G, complex (see Ref. 1 for review). This high-affinity binding was inhibited by the addition of GTP. However, cyc-cells lack functional G. and have normal levels of Gi (Refs. 9, 10, 21, and 22; see Ref. 1 for review), and Gi may promote the high-affinity agonist binding observed by these authors. The finding that cyc-cells can become desensitized either to P-adrenergic agonists or to prostaglandin El (23) may also reflect an interaction of Gi with stimulatory receptors.
Thus, the interaction of P-adrenergic receptors with Gi may be of importance in native membranes, perhaps acting as a mechanism to attenuate hormonal stimulation of the adenylate cyclase system. Such attenuation might be rapid and act as a buffer on stimulation, or it might be delayed and form the basis for a novel form of heterologous desensitization. Gi might also act to bind receptors in a nonproductive complex, S. Abramson and P. Molinoff, personal communication.

9354
Activation of Gi by @-Adrenergic Receptors thereby causing receptor-specific, or homologous, desensitization. These questions can be addressed quantitatively in vesicles containing receptors and either Gi or G,, or both G proteins. IAP and cholera toxin will be useful in the analysis of these effects in native membranes.