Coupling of the aZA-Adrenergic Receptor to Multiple G-proteins A SIMPLE APPROACH FOR ESTIMATING RECEYI'OR-G-PROTEIN COUPLING EFFICIENCY IN A TRANSIENT EXPRESSION SYSTEM*

It is now widely appreciated that G-protein-coupled cell-surface receptors can modulate distinct signal transduction pathways via coupling to different GTP-binding proteins. In the present study, we have used a transient co-expression approach to study the coupling of a single az-adrenergic receptor (auAR) population to three different G protein subtypes (Gi, G,, and G,) acting on two different cellular effectors in HEK 293 cells. In all cases, the affinity of the receptor for the au-adrenergic agonist, UK14304, is unchanged (KO = 670 m). However, there is a dramatic difference in the ECm of UK14304 in eliciting inhibition of endogenous adenylyl cyclase via endogenous Gi (0.09 m) uereue activation of phospholipase C via co-transfected G, (50 m) or stimulation of endogenous adenylyl cyclase via co-transfected G. (70 m) in HEK 293 cells. These findings are consistent with the interpretations that the a&R preferentially interacts with Gi rather than G, or G,. When the auAR was mutated at Asp79, a residue highly conserved among G-protein-coupled receptors, the mutant D79N auAR lost the ability to couple to G, and G. and, although it was able

It is now widely appreciated that G-protein-coupled cell-surface receptors can modulate distinct signal transduction pathways via coupling to different GTPbinding proteins. In the present study, we have used a transient co-expression approach to study the coupling of a single az-adrenergic receptor the interpretations that the a&R preferentially interacts with Gi rather than G, or G,. When the auAR was mutated at Asp79, a residue highly conserved among Gprotein-coupled receptors, the mutant D79N auAR lost the ability to couple to G, and G. and, although it was able to couple to inhibition of cyclase via pertussis toxin-sensitive pathways (Gi), it did so with a lower potency than observed for the wild-type auAR (EC50 = 7.2 MI). The most straightforward interpretation of these data is that the D79N mutation in the a " reduces the efficiency of coupling of the a&R to all G-proteins, thus eliminating signal transduction through those pathways less efficiently coupled to the auAR. Since the transient expression assays described permit manipulation of the structure of both the receptor or the G-protein, the present strategies could be exploited to delineate the complementary domains specifying the affinity andor efficacy of receptor coupling to distinct GTPbinding proteins.
A large family of cell surface receptors conforming to the predicted topography of seven transmembrane spanning proteins elicit their physiological effect by first coupling to and GM27800 and CA54427 (to Dr. H. R. Bourne), KllHL02555 (to * These studies were funded by National Institutes of Health Grants B. R. C.), HL25182 and HL43671 (to L. E. L.), and resources from the Ministry of Foreign Atrairs and a grant from the University of Grenoble (to 0. C.). The costs of publication of this article were defrayed in part marked "aduertisement" in accordance with 18  activating a population of heterotrimeric GTP-binding proteins that then mediate the responses of a plethora of cellular effectors, including enzymes and ion channels. Recently, it has been appreciated that cell surface receptors often are coupled to a variety of effectors, and this is likely due to the interaction of the receptor with multiple distinct GTP-binding proteins (1-8).
A number of experimental strategies have revealed that a single receptor population is capable of interacting with multiple GTP-binding proteins. For example, reconstitution of purified receptors with differing GTP-binding proteins into lipid vesicles or a mixed detergentflipid milieu restores receptor-Gprotein coupling disrupted by biological detergents, as assessed using agonist-dependent GTPase or GTPv5S1 binding activities. Such strategies have revealed, for example, that the p-adrenergic receptor can couple not only to G, but also to Gi (9) and that the muscarinic receptor can couple both to Gi and to Go (10). The ability of receptors to couple to multiple families of GTP-binding proteins also can be revealed by overexpression of a given receptor or G-protein in heterologous cells. Such studies, for example, have demonstrated at the azAAR can couple not only to pertussis toxin substrates (Gi and Go), as it does in physiological settings, but also to activation of phospholipase C (3), presumably through interaction with Gq/Gall, and to stimulation of adenylyl cyclase through G, (11,12). Similarly, co-expression of aZAAR and G, leads to the pertussis toxininsensitive inhibition of adenylyl cyclase (13). The biochemical reconstitution or heterologous expression strategies of recombinant proteins outlined above have been useful in identifying which receptor-G-protein interactions can occur, and have demonstrated that receptors are capable of interacting with a wide array of GTP-binding proteins when high concentrations of receptor and G-protein are provided in biochemical or recombinant "reconstitution" paradigms. In contrast, a greater fidelity of receptor-G-protein interactions apparently occurs in the physiological setting. For example, virtually all azAAR receptor-mediated physiological effects are sensitive to pertussis toxin, implying a role for the Gi and Go GTP-binding protein family in mediating the physiological consequences of aZAAFt activation in vivo. Similar conclusions about fidelity in receptor-effector interactions in vivo derive from recent studies using antisense strategies which reveal that receptor-G-protein interactions giving rise to a particular cellular response are very stringent in terms of the a (14) as well as the p (15) and y (16) subunits involved.
The present study was undertaken not to measure receptor-GTP-binding interactions that occur in a given physiological setting, but rather to provide a straightforward analysis of the The abbreviations used are: GTPY5S, labeled guanosine 5'-O-(thiotriphosphate); a " , a,-adrenergic receptor; IP, inositol phosphate; LH, luteinizing hormone; hCG, human chorionic gonadotropin. efficacy of coupling of a particular receptor, the a2AAR, to various G-proteins expressed in intact cells and thereby allow a prediction of the selectivity of receptor interaction with given sub-populations of G-proteins in the intact cell. For these studies, the azAAR was introduced into HEK-293 cells, and its coupling to various GTP-binding proteins and subsequent effectors was assessed following transient expression of the receptor and particular G-protein a subunits. The studies revealed that the azAAR preferentially interacts with the Gi protein family and to a lesser extent with G, and G,. Interestingly, the D79N mutation of the a,AR decreases receptor coupling to the Gi protein as well as to G, and G, such that, following this mutation, only receptor-mediated attenuation of adenylyl cyclase through the Gi pathway can be detected, and this inhibition occurs a t higher receptor occupancy than required for inhibition of adenylyl cyclase by the wild-type aZAAR. This transient expression approach, then, provides a reasonably rapid and straightforward analysis of the efficacy of receptor coupling to a particular GTP-binding protein family. Such an analysis could be of value in predicting the preferential signal transduction pathway utilized by a given receptor system or, following intentional structural alterations via mutagenesis, the domains within receptors and/or G-protein a subunits involved in specifying the affinity or efficacy of a given receptor-G-protein interaction.
DNAs were transfected into HEK-293 cells (2 x lo6 eelld100-mm dish) in the presence of DEAE-dextran (250 pg/ml) and chloroquine (100 PM) for 90 min. Cells were shocked with phosphate-buffered saline containing 10% dimethyl sulfoxide for 2 min, then washed once with phosphate-buffered saline and allowed to grow overnight. The next day cells were reseeded in 24-well plates (1-2 x lo5 cells/well) for analysis of receptor binding or for detection CAMP and inositol phosphate (IP) production.
CAMP a n d Z P Assays-cAMP and IP production in HEK-293 cells were measured as described (22,23). For CAMP, data are presented as PHFohimbine Binding Assuy-HEK-293 cells were transfected with the porcine aUAR (3 pg of DNN100-mm dish), and membranes were prepared and assayed as described elsewhere (13). Binding was analyzed in the presence of 100 mM NaCl, to maximally impose regulatory effects of monovalent cations (241, and 100 p~ Gpp(NH)p, to eliminate effects of GTP-binding protein-receptor interactions on 02,AR affinity for agonists (25). Although agonist potency at the D79N mutant a,,AR is not modulated by monovalent cations (26), in contrast to wild-type a,AR, Na' was present in the binding assays for evaluation of receptor affinity for agonist a t both wild-type and D79N aZAAR.

RESULTS AND DISCUSSION
To study the ECSo for a given az-adrenergic agonist, UK14304 (bromoxidine), on second messenger production via a variety of G-proteins, we expressed the azAAR in HEK-293 cells, which lack endogenous a2AAR binding activity, in the absence or presence of DNA coding for G,, and G,,. Regardless of whether the a2AAR was transfected in the absence or presence of G-proteins, the KO for UK14304 at the receptor was 0.67 p~ (data not shown). Fig. 1 demonstrates that introduction of azAAR into HEK 293 cells permits detection of inhibition of endogenous adenylyl cyclase by the azAAR due to interaction with endogenous Gi proteins. To facilitate detection of inhibition of the adenylyl cyclase enzyme by transfected azAAR, DNA coding for the a,m was co-transfected with that coding for the hCG/LH receptor, which also is not expressed in HEK-293 cells. It has been determined empirically that the same fraction of cells that take up the DNA coding for a z A A R also take up that coding for the LH receptor, and thus UK14304-elicited inhibition of CAMP production elevated by hCG provides a sensitive tool for monitoring exogenous azAAR-rnediated inhibition of adenylyl cyclase via a pertussis toxin-sensitive, Gi-coupled pathway (22). As shown in Fig. 1, the EC50 for inhibition of CAMP accumulation by UK14304 was 0.07 2 0.05 (n = 4). The UK14304-elicited inhibition of cyclase was sensitive to pertussis toxin, indicating that Gi mediates this effect. In the absence of introduction of the aZAm, no inhibition of CAMP accumulation by UK14304 could be detected. Fig. 2 demonstrates that co-expression of the aZAAR with G, results in the activation of phospholipase C as measured by production of [3HlIP. Although HEK-293 cells do have endogenous G,, or members of the G, family, activation of phospholipase C by the azAAR required co-expression of G,, (23). Further evidence that LIZAAR activation of phospholipase C was occurring via a G,-dependent pathway was the insensitivity of inositol phosphate production to pertussis toxin. The EC50 of UK14304 for activation of the phospholipase C pathway was 60 * 22 nM (a = 3).
In contrast to inhibition of cyclase, stimulation of endogenous adenylyl cyclase via aZAAR in HEK 383 cells could be detected only following co-expression of ~z A A R with G,, (Fig.  3). As also shown in Fig. 3, UK14304-elicited stimulation of adenylyl cyclase via the a Z A A R could be increased by prior pertussis toxin treatment, presumably due to the abolition of receptor-mediated inhibition of the cyclase. The EC50 of UK14304 for stimulation of cyclase via G, was 70 3 nM.
In all cases, the pathways activated by UK14304 were strictly dependent on the expression of transfected azAAR. In addition, a2AAR receptor density achieved in each transfection, as assessed using Scatchard transformation of [3H]yohimbine saturation binding data, was not significantly different whether the azAAR was transfected alone or with G, or G, and ranged from 2.8 to 3.6 pmol L3H1yohimbine binding per mg of protein. Thus, the dramatic difference in EC50 of UK14304 for inhibition of adenylyl cyclase (0.07 nM) via activation of phospholipase C via G, (60 nM) or stimulation of a adenylyl cyclase versus G,, (70 nM) when compared to the KO of the azAAR for UK14304 expressed in HEK 293 cells, 670 nM, suggests that inhibition of adenylyl cyclase is achieved by occupancy of only a small fraction of the azAAR that are expressed in the HEK 293 cells when compared to effects of the aZAAR on G,or G,-mediated signal transduction pathways. The fact that activation of adenylyl cyclase and of phospholipase C in HEK 293 cells requires both increased fractional occupancy of the aZAm ( E C~O = 60-70 nM, i.e. 1000-fold greater agonist concentrations required than for inhibition of adenylyl cyclase via Gi) and co-expression of heterologous G,, or G,, subunits indicates that encounters among aZAAFt and these signal-transducing molecules are of lesser affinity (lower frequency) and of lower efficacy than among aZAAR, Gi, and cyclase. Said another way, the present findings suggest that aZAAR preferentially associates with GTP-binding proteins of the Gi family in contrast to the G, or G, family. Implicit in the above interpretations is the premise that agonist potency in activating various signal transduction pathways via a single receptor population reflects not only the intrinsic receptor affinity for the agonist but also the efficacy of receptor coupling to a given G-protein population and the degree of amplification in downstream reactions (27). We therefore decided to compare the potency of UK14304 in inhibition of adenylyl cyclase via GI, activation of adenylyl cyclase via G,, and stimulation of phospholipase C via G, when HEK 293 cells were transfected with the D79N mutant of the aZAAR instead of the wild-type a2AAR. Such studies should allow a direct comparison of receptor-G-protein coupling efficiency of the wildtype versus D79 azAAR, as receptor affinity for agonist is not markedly different at the two receptors (KO = 670 nM at the wild-type azAAR and 300 nM at the D79N a2AAR) and the downstream consequences of G-protein activation should be the same for both preparations. Previous studies with other Gprotein-coupled receptors mutated at a topographically homologous position indicate that receptor interaction with Gproteins is perturbed by this mutation (28)(29)(30), and similar findings have been reported for the aZAAR, at least as assessed in broken cell assays of receptor-G-protein coupling (18,31). In contrast, introduction of the D79N aZAAR mutant into AtT2O cells results in detectable aZA-mediated inhibition of CAMP accumulation in intact cells and suppression of voltage-gated Ca2+ currents measured in the cell-attached patch configuration, despite a loss of D79N aZAAR-elicited activation of receptor-operated K+ currents in the AtT20 cells (18). These findings could arise either by selective loss of the ability of the D79N receptor to couple to the G-protein population that mediates K+ channel activation or, alternatively, by perturbation of D79N aZAAR coupling to all GTP-binding proteins, with sufficient coupling remaining to permit inhibition of CAMP accumulation and of voltage-gated Ca+ currents via the D79N aZAAR but not sufficient to activate receptor operated K' channels via this mutant receptor.
To evaluate whether the D79N mutation of the a 2 A A R perturbs receptor-G-protein coupling, as manifest in the transient transfection assays described above, HEK 293 cells were transfected with DNA coding for the D79N ~z A A R mutant in the absence or presence of DNA coding for G,, or Gqa. The receptor density achieved in these transfections (2.5 pmol of t3Hlyohimbine binding per mg of protein) was comparable from that for the wild-type receptor (2.8-4.6 pmol of L3HIyohimbine binding per mg of protein). Nonetheless, as shown in Fig. 4.4, UK14304 is considerably less potent in inhibiting adenylyl cyclase through the D79N receptor (EC50 = 7.2 nM) than through the wild-type aZAAR (EC50 = 0.09 nM), implying that the mutant D79N a 2 A A R is less efficacious than the wild-type L Y Z A A R in interacting with the Gi proteins that mediate inhibition of adenylyl cyclase. This difference in EC50 values for wild-type versus D79N aZAAR in inhibiting adenylyl cyclase is not unique to the UK14304 agonist, but also is detected for the agonist (UK14304) was dependent on which G-protein the wild-type aZAAR was coupled t o in order to elicit its effects on second messenger production. The EC50 of the agonist was not due to changes in intrinsic receptor affinity for UK14304, as receptor affinity for UK14304 was undistinguishable regardless of the G-protein composition of the transfected HEK293 cells evaluated. Instead, the EC50 reflects the efficiency of receptor-Gprotein coupling as well as G-protein-effector interaction to elicit second messenger production. Furthermore, the comparison of agonist potency in eliciting second messenger production via the wild-type uersus the D79N mutant a 2 A A R allowed a direct comparison of receptor-G-protein coupling efficiency for these two receptors, as the downstream consequences of Gprotein activation elicited by wild-type versus D79N a 2 A A R should not be different in these experiments, and the intrinsic affinity of the wild-type (KO = 0.67 p~) and D79N (K, = 0.3 p~) (Y2AAR for UK14304 are essentially indistinguishable.
The lower potency of the D79N a 2 A A R in inhibiting adenylyl cyclase via the endogenous Gi proteins and the inability of the mutant aZAAR to activate phospholipase C via Gq or adenylyl cyclase via G, is consistent with the interpretation that the D79N mutation perturbs the efficiency of agonist-facilitated receptor-G-protein interaction.
It is clear that the experimental paradigm of transient expression of receptors and G-proteins described in the present report should be of considerable value in comparing the efficacy of receptor-G-protein coupling for wild-type and intentionally mutated receptors, and perhaps assist in revealing the preferred signal transduction pathway of newly discovered receptors identified by molecular cloning strategies. clonidine for inhibition of adenylyl cyclase via wild-type ~~A A R at 0.07 nM; EC50 via D79N ~~A A R at 10 nM); data not shown. Furthermore, the mutant D79N ~~A A R receptor is not capable of activating phospholipase C via G, (Fig. 4 B ) or adenylyl cyclase via G, (Fig. 4C). Although these results could be interpreted as a selective effect of this mutation on coupling of a 2 A A R to G, or G,, the finding that agonist potency for inhibition of adenylyl cyclase via Gi was markedly decreased for the D79N mutant when compared to the wild-type azAAR suggests that mutation ofAsp7' to in the aZAAR creates a receptor that, upon agonist occupancy, is less effective in coupling to all G-proteins, with the result of eliminating signal transduction through those G-proteins less efficiently coupled to the aZAAR. Consistent with this interpretation is our observation that transfection of 1/10 of the aZAAR plasmid (using 0.3 pg instead of 3.0 pg of DNA coding for azAAR/lOO-mm dish) results in no reduction in the extent of attenuation of adenylyl cyclase by the wild-type aZAAR, but a loss in readily detectable attenuation by the D79N a,AR.

SUMMARY
A transient co-transfection approach has been used to study the coupling of a single a 2 A A R population to multiple GTPbinding proteins. We observed that the EC50 of an agonist