G Protein Involvement in Receptor-Effector Coupling*

Les proteines regulatrices fixatrices de guanine assurent le transfert de l'information des recepteurs hormonaux a l'adenylate cyclase. Recapitulatif de l'etat des connaissances concernant la structure et le mecanisme d'action de ces proteines G


Overview
The discovery and subsequent characterization of guanine nucleotide-binding regulatory proteins (G proteins) confirmed the hypothesis of Rodbell and his colleagues (1) that a regulatory element is interposed between hormone receptors that control adenylyl cyclase activity and the enzyme itself. Investigation of the role of G proteins in the regulation of adenylyl cyclase (2), coupled with parallel study of GTP-mediated regulation of a cyclic GMP-specific phosphodiesterase in retinal rod outer segments (3,4), has yielded a wealth of information on the mechanisms by which cell-surface receptors communicate with their respective cellular effectors. Thus, it is now believed that the general features of the signal transduction pathway worked out for the regulation of adenylyl cyclase and cGMP phosphodiesterase, that information flows from extracellularly oriented receptor to G protein to intracellular effector, are shared by a variety of transmembrane signaling systems that are triggered by hormones, autacoids, and neurotransmitters (2, 5).
Metabolic events controlled by G proteins can be considered in two categories, those in which the direct involvement of a G protein has been conclusively demonstrated and those in which control by a G protein is likely but unproven. The two systems known to be controlled directly by G proteins are those noted above. Modulation of adenylyl cyclase activity in response to various stimulatory agonists (e.g. epinephrine, gonadotropins, and ACTH) or inhibitory agents (e.g. a2adrenergic and muscarinic agonists) is mediated by distinct G proteins, designated G, and Gi (for stimulatory and inhibitory), respectively. The concentration of cyclic GMP in retinal rod outer segments, a crucial determinant of visual excitation, is modulated through the ability of a G protein (transducin or G,) to activate a cyclic GMP-specific phosphodiesterase in response to photolyzed rhodopsin.
Probably the most widely studied system in which involvement of a G protein has been implicated, but not rigorously demonstrated, is the regulation of phosphoinositide turnover by a phosphatidylinositol-specific phospholipase C (6). The most compelling evidence presented to date involves the capacity of GTP and its nonhydrolyzable analogs to enhance the agonist-mediated accumulation of the products of this enzyme in permeabilized cells and/or membranes derived therefrom. The ability to block this response in some, but not all, cell types by treatment of the cells or membranes with pertussis toxin (which disrupts certain G protein-linked systems; see below) suggests that there may be cell-specific G proteins involved. However, there has as yet been no demon- stration of a direct effect of a G protein on phospholipase C.
Another system in which the question of regulation by a G protein is currently generating a great deal of excitement (and controversy) is that involving muscarinic cholinergic receptors and atrial potassium channels. Experiments similar to those discussed above, as well as reports of a direct effect of G proteins on the channel in excised patches (7, 8), provide very compelling evidence that this effector is under the control of a G protein. The controversy centers upon which of the G protein's subunits is responsible for the observed effects (see below). Other cellular events in which control by G proteins has been implicated include the regulation of neuronal Ca2+ channels (9), exocytotic secretory events (lo), olfactory transduction (11), protein translocation (12), and phospholipase Az activity (13).

Structure of G Proteins
The G protein family contains a relatively large number of very closely related members. Table I summarizes the properties of the relevant polypeptides (including those deduced from cDNA cloning) and their functional characteristics, both proven and inferred.
The several G proteins that have been characterized are heterotrimeric, with subunits designated a , ( 3 , and y (in order of decreasing mass). Differences in the a subunit presently serve to distinguish the various G protein oligomers. The a subunits contain a single, high-affinity binding site for guanine nucleotides and possess the GTPase activity that is crucial for the action of these proteins. This subunit also contains the site(s) for NAD-dependent ADP-ribosylation catalyzed by bacterial toxins. The a subunit of G. (Gee) can be ADP-ribosylated by cholera toxin; polypeptides currently designated as Gims and the very similar Go, can be ADPribosylated by pertussis toxin (also known as islet-activating protein); G,, can be modified by both toxins. ADP-ribosylation of these proteins results in a characteristic alteration of their function (activation of G, in the case of modification by cholera toxin and an impaired ability to interact with receptors in the case of ADP-ribosylation by pertussis toxin (14, 15)). The a subunits of Gi and Go, but apparently not those of G. and G,, are modified post-translationally by the addition of myristic acid (16). Although the function of this N-myristoylation is not known, it may play a role in the interaction of these relatively hydrophilic subunits with the plasma membrane. Masters et al. (17) have incorporated secondary structural predictions for an "average" a subunit into a model based on the crystal structure of the guanine nucleotidebinding protein elongation factor Tu. This view of G protein structure has provided a basis for speculation on the location of functional domains within the a subunits.
In addition to the heterogeneity that serves to divide G proteins into major classes (GB, Gi, Go, and G,), there are multiple forms of the subunit polypeptides within each class. The existence of two forms of G., was established when the protein was purified; the molecular basis for this heterogeneity was later shown to be due to alternative splicing of a single precursor mRNA (18,19). Recent evidence indicates that at least four species of G,, can be produced by this mechanism (20). Functional distinctions between the different forms of G, have not yet been reported, although hints of the involve-  (2,4). PAR, Pe These properties are assumed for those proteins that are known only as cDNA-deduced sequences.

2577
The relative quantities of the three Gi,s in "purified preparations of Gi used in functional studies is unclear. 'These proteins are candidates for involvement in the regulation of pertussis toxin-sensitive cellular events 'p, placental; not to be confused with the putative G protein that controls phosphoinositide hydrolysis.
The situation with respect to Gi is even more complicated. Gi, as originally defined, is the G protein responsible for receptor-mediated inhibition of adenylyl cyclase. It contains an a subunit of - [40][41] kDa that is a substrate for ADPribosylation by pertussis toxin. However, since hormonal inhibition of adenylyl cyclase may be mediated at least in part by the release of Py subunits, many G proteins may serve this role (see below). T o date, three highly homologous cDNAs that encode putative Gicr subunits have been isolated (2,22). The relationship between these deduced proteins and the Gis that have been purified to date is being studied, primarily by the use of specific antisera. A related protein, Go, has also been purified, and cDNAs that encode its a subunit have been cloned (23). This protein also possesses many of the properties associated with a Gi, although the sequence homology is not as great as that shared by the other three members of the group. In addition to the participation of Gi(s) in the hormonal inhibition of adenylyl cyclase, there is evidence that these proteins are involved in several other signaling systems (Table  I).
Heterogeneity is also apparent in G,,,, the G protein involved in visual transduction. Two highly homologous cDNAs that encode different forms of this polypeptide have been isolated. Antibodies raised aginst synthetic peptide sequences specific to each protein show that one, G,,,,, is localized exclusively in retinal rod outer segments, while the other, GLcrZ, is present only in cones (24). Both forms of G,, are presumed to be activators of cGMP phosphodiesterase in these cells.
The /3 and y subunits of G proteins are closely associated with each other and have been separated only after denaturation. The Py subunit complexes of G,, Gi, and Go can be interchanged in functional assays, and this complex may exchange between differing a subunits in vivo (see below). The complex may also participate in anchoring a to the plasma membrane (25).
Despite the fact that By subunit complexes can be exchanged among different a subunits, there are multiple forms of these polypeptides. Although the ,6 subunit of G, can be visualized on sodium dodecyl sulfate-polyacrylamide gels as a single band of 36 kDa, the P subunit of the other G proteins is a doublet of proteins with apparent masses of 36 kDa (PI) and 35 kDa (P2). cDNAs that encode two distinct forms of the P subunit have recently been isolated (26-28); the deduced amino acid sequences are 90% identical. Immunological evidence and studies involving expression of these cDNAs have revealed that one encodes PI and the other encodes Pz (29).
The evidence for heterogeneity of the y subunit is based on analysis of the proteins, since only G,, has been cloned to date. Comparison of peptide maps of G,, with those of human erythrocyte y indicates that these two proteins are distinct. Thus, the Py subunit complex of G, differs from that associated with at least some other G proteins, and indeed, the first report of functional differences between these entities has appeared recently (30). Additionally, electrophoretic and immunological evidence suggests that there is more than one non-G, y subunit (31,32); thus, functional differences may exist between the Py subunits of other G proteins as well. The requirement for both cy and by subunits for receptor-mediated activation of G proteins and the heterogeneity of each subunit suggest that each class of receptors might recognize a specific cy& structure. Other guanine nucleotide-binding proteins that do not fit the "definition" of a G protein developed above may also be important in signal transduction. These include ADP-ribosylation factor, a protein required for the covalent modification of G, by cholera toxin (33); G,, a protein of unknown function purified from placenta by Evans et al. (34); and the products of the ras oncogenes (35). These small (Mr = -21,000) proteins contain a high affinity guanine nucleotidebinding site, and a t least some of them are GTPases. However, they have not been reported to be substrates for ADP-ribosylatjon by bacterial toxins, and they appear to have little or no affinity for by. The most intensively studied of these proteins are the ras gene products. Mutations that inhibit GTPase activity are presumed to activate the protein and are associated with cellular transformation. There is preliminary evidence that these proteins can couple receptors to effectors, particularly growth factor receptors to a phosphatidylinositolspecific phospholipase C (36).

Mechanism
Purification of the components involved in the regulation of adenylyl cyclase and cyclic GMP phosphodiesterase, as well as the ability to reconstitute these entities in lipid bilayers of defined composition, has allowed detailed studies of the molecular mechanisms by which these enzymes are regulated. The essential protein-ligand and protein-protein interactions (both proven and presumed) are diagrammed in Fig. 1. The following discussion highlights the crucial points of this regulatory cycle. Those acquainted with the field will note that they have been spared a discussion of the effects of Mg2' on the various interactions; the influences of this important metal have been detailed elsewhere (2,37).
In the basal state, G proteins exist in their oligomeric form with GDP tightly bound to the cy subunit. The interaction of the G protein with an appropriately liganded receptor (H . R) stimulates the dissociation of GDP, presumably as a result of a conformational change that results in an "opening" of the guanine nucleotide-binding site. The H .R.G complex is ap- GTP; there are several reports of the copurification (at least for initial steps) of an H.R.G complex (presumed to be nucleotide-free) (38, 39). However, in the presence of the relatively high cellular concentrations of GTP, the "open" guanine nucleotide site is rapidly filled. The binding of GTP to H . R . G has two important consequences. The first is that the affinity of H for R (and R for G) is decreased. This negative heterotropic interaction between guanine nucleotide and hormone is a consequence of the utilization of the binding energy of R . H to drive G to the open form, thus necessarily weakening the observed binding of R for H. It is for this reason that antagonist-receptor complexes, in which the binding energy cannot be similarly utilized, have unchanged binding affinity in the presence of guanine nucleotide (40,41). Dissociation of H from R and, therefore, of R from G . GTP allows R to recycle and thus function catalytically in the activation of G (42). For the adenylyl cyclase system, Levitzki (43) has suggested that G, and the cyclase are always coupled, such that H.R interacts with the complex; a brief critique of this model is available (2).
The second major consequence of binding of GTP to G is the "activation" of the G protein, such that it interacts fruitfully with the appropriate effector. There is considerable evidence that such activation greatly reduces the affinity of the GTP-liganded cy subunit for the by complex and that the resulting dissociation of subunits is an important component of their mechanism. Much of this evidence comes from study of G protein activation by nonhydrolyzable analogs of GTP in detergent solution, although subunit dissociation has been decisively demonstrated with GTP in a reconstituted system containing rhodopsin and G, (44). The subunit dissociation model for the activation of G proteins has been challenged, primarily on the basis of kinetic analyses of hormonal activation of adenylyl cyclase in membranes (45; for a response, see Ref. 2). Direct demonstrations of subunit dissociation in intact membranes are required.
Numerous lines of evidence suggest that the G,, GTP complex is the actual regulator of the target effector in wellstudied G protein-linked systems (although recent reports suggest that the Py complex might serve this role in some systems; see below). The activation of G proteins by nonhydrolyzable analogs of GTP allows chromatographic resolution of the cy subunit-nucleotide complex from Py. In both the cyclic GMP phosphodiesterase and adenylyl cyclase systems, the capacity of oligomeric G, and G, to activate these effectors can be fully accounted for by the activity of the respective cynucleotide complex (2,4).
In contrast to the results obtained with G, and G,, activation of Gi by nonhydrolyzable analogs of GTP and separation of the a-nucleotide and By complexes revealed that the bulk of the capacity to inhibit adenylyl cyclase was in the by subunit and that the inhibition was largely dependent on the presence of G, (47). At least certain forms of the G, I GTPyS' complex are capable of inhibiting adenylyl cyclase, but the activity has never been impressive (48). These observations led to the subunit exchange model for the action of Gi. This hypothesis states that hormonal activation of Gi leads to dissociation of its subunits; the resultant increase in the concentration of free Py associated with the membrane drives interaction with G,, and, thus, indirect inhibition of adenylyl cyclase. The Py subunits of G,, Gi, and the abundant (in brain) Go are equipotent in their ability to deactivate G,,,. Thus, the combination of the subunit dissociation mechanism and a Py subunit complex that is shared among different a subunits provides a The abbreviation used is: GTPyS, guanosine 5'-O-(thiotriphosphate). mechanism for coordinate regulation of transmembrane signaling pathways, as discussed elsewhere (2). The basic premise of this hypothesis is that the receptor-mediated activation of a sufficient pool of G protein will increase the concentration of (37 to the point where it will inhibit some pathways, while the release of a-GTP can initiate others. As mentioned earlier, there are suggestions that the Pr subunit complex can regulate some effectors directly. Evidence comes primarily from study of atrial potassium channels and their activation by muscarinic cholinergic receptors. Logothetis et al. (8) perfused the intracellular surface of excised membrane patches from chick atrial cells with purified brain G protein subunits and found activation of K' channels by Pr subunits; unactivated a subunits reversed this effect.
These results have been challenged by Brown and Birnbaumer and their colleagues (49), who performed similar experiments on excised patches from guinea pig atria and found that the activated a subunit from a Gi-like G protein isolated from erythrocytes (which they now call Gk) was capable of activating the K' channel; the Pr subunits tested (from both brain and erythrocytes) were inert. The resolution of this controversy may require further definition and purification of the crucial components. The Pr complex has also been implicated in the activation of retinal rod phospholipase A, (13). Birnbaumer (50) has argued against the involvement of the Pr complex in any specifically regulated cellular event (e.g. inhibition of adenylyl cyclase or activation of K+ channels). He proposed that the role of (37 is that of a general attenuator of basal G protein activity, which would result in an enhanced signal to noise ratio of G protein-linked systems upon their specific activation. Although the suggestion is plausible, there is little evidence to support it. Deactivation of the G protein, which results in termination of the signal, appears to be primarily due to (or associated with) the GTPase activity of a.GTP. This is presumed to result in reassociation of a -G D P with By, as noted above. Since the relative capabilities of G,,. GDP and G,..GTP to interact with a given effector have not yet been determined, the contribution of subunit reassociation to the deactivation process is still unclear. The rate constant for hydrolysis of action. The alliance of all of these disciplines will facilitate investigation of G protein-linked signaling processes and of possible pathological disturbances therein.