Bidirectional Allosteric Effects of Agonists and GTP at α_2A/D-Adrenoceptors1

Experimental Procedures

Materials.

PC12 cells stably expressing the cloned α_2A/D-adrenergic receptor at a density of 3 to 4 pmol/mg membrane protein were generously provided by Dr. Stephen M. Lanier (Medical University of South Carolina). [³⁵S]GTPγS (1255 Ci/mmol) and [³H]rauwolscine (78 Ci/mmol) were purchased from DuPont-NEN (Boston, MA). UK14304 [5-bromo-N-(4,5-dihydro-1H-imidazol-2-yl)-6-quinoxalinamine] and clonidine were obtained from Research Biochemicals International (Natick, MA); R-(−)- andS-(+)-hydroxytolazoline, 3,4-dihydroxytolazoline, andR-(−)-OH-3,4-dihydroxytolazoline were synthesized as previously described (Miller et al., 1983; Sengupta et al., 1987). Other reagents were purchased from Sigma Chemical Co. (St. Louis, MO).

Cell Culture and Membrane Preparation.

Transfected PC12 cells were grown as monolayers in Dulbecco's modified Eagle's medium with high glucose supplemented with 10% fetal calf serum, 5% horse serum, penicillin, streptomycin, and fungizone as described previously (Tian et al., 1994). Cells were washed twice with PBS, harvested with a rubber policeman, and pelleted. The pellet was resuspended in 5 ml/dish of lysis buffer (5 mM Tris-HCl, pH 7.5, 5 mM EDTA, 5 mM EGTA, 0.1 mM phenylmethylsulfonyl fluoride) at 4^oC and homogenized with a Dounce homogenizer. The lysate was then centrifuged at 34,000g for 15 min and the pellet was resuspended in membrane buffer (50 mM Tris-HCl, pH 7.5, 0.6 mM EDTA, 5 mM MgCl₂, 0.1 mM phenylmethylsulfonyl fluoride). Aliquots were frozen in liquid nitrogen and stored at −80^oC until used.

[³⁵S]GTPγS Binding.

As described previously (Tian et al., 1994), binding was initiated by addition of reaction mixture (50 mM Tris-HCl, pH 7.5, 5 mM MgCl₂, 1 mM EDTA, 1 mM dithiothreitol, 100 mM NaCl, 2 mM GDP, 1 mM propranolol, and 2 to 3 nM [³⁵S]GTPγS) to 4 to 8 μg of membranes in a total volume of 0.1 ml. Most experiments were carried out in triplicate at 25°C for a 10-min incubation period. Filters were washed four times (with 4 ml of 50 mM Tris-HCl, pH 7.5, containing 5 mM MgCl₂ and 100 mM NaCl) and then counted. Nonspecific binding was determined in the presence of 10 μM GTPγS and subtracted from total bound radioactivity.

Radioligand Binding Assay.

Assays were performed in a total volume of 0.1 ml and contained 50 mM Tris-HCl (pH 7.5), 5 mM MgCl₂, 1 mM EDTA, 1 mM dithiothreitol, 1 μM propranolol, 100 mM NaCl, and 20 to 50 μg of membrane fraction. Displacing ligands and GTPγS (10 μM) were added when indicated. Nonspecific binding was determined in presence of 0.1 mM phentolamine. Competition binding curves were analyzed with LIGAND (Munson and Rodbard, 1980) and GraphPad Prism. A model with two classes of binding sites was tested for its statistical superiority over a model with a single class of binding site with an F test based on the residual variance between the actual and predicted data points. The size of the GTPγS-induced shift in agonist affinity was quantitated by calculating the area between individual displacement curves and averaging the data.

Results

In an earlier study (Tian et al., 1994), we described the ability of epinephrine (EPI) to stimulate [³⁵S]GTPγS binding in membranes from PC12 cells expressing cloned α_2A/D-receptors. EPI caused an increase of ∼3-fold in [³⁵S]GTPγS binding when measured after a 10-min incubation period and exhibited an EC₅₀ of 0.14 ± 0.01 μM (Fig.1A; Table1). EPI produced the largest increase of binding among all the phenethylamines or imidazolines tested and its activity was assigned a value of 1.0 for comparative purposes.

Five additional phenethylamines, which differed in one or more of the potential binding features of EPI, as well as three imidazoline derivatives (UK14304, clonidine, and oxymetazoline; Fig.2) were evaluated for their effects on [³⁵S]GTPγS binding. As shown in Fig. 1 and Table 1, the maximum stimulation caused by three of these agonists (norepinephrine, deoxyepinephrine, and UK14304) approached but did not equal the level of EPI response. The remaining five agonists produced considerably smaller increases of 40 to 60% above control levels. A rank order for efficacy of EPI > norepinephrine > deoxyepinephrine ≅ UK14304 ≫ synephrine > clonidine > oxymetazoline ≥ phenylephrine = norphenylephrine was determined. Differences in agonist potency were also evident with a potency order of oxymetazoline > clonidine ≥ UK14304 > EPI > norepinephrine = norphenylephrine > deoxyepinephrine = phenylephrine > synephrine. Clearly, the efficacy and potency orders are distinct from each other.

A comparison of EC₅₀ and relative maximal activity values from GTPγS-binding studies with values previously reported from intact tissue functional responses shows a high degree of correlation for both relative efficacy (r = 0.97) and relative potency, expressed as K_Avalues (r = 0.94) (Fig.3). Thus, agonist-induced stimulation of [³⁵S]GTPγS binding in this isolated membrane preparation is predictive of tissue response.

In the absence of GTPγS, agonists competed for [³H]rauwolscine-binding sites in isolated membranes in a biphasic concentration-dependent manner (Fig.4), yielding computed estimatedK_D values forK_DH andK_DL as well as their percentage contribution to total binding (Table 2). In the presence of 10 μM GTPγS, binding was monophasic, yielding an estimated K_D that was generally, but not always, similar to K_DL values. The net influence of GTPγS on agonist binding was quantitated by measuring the total area between displacement curves in its absence and presence and normalizing to the value for EPI. Notably, this approach includes contributions of both the amplitude of theK_D change and the percentage of high- versus low-affinity sites.

The relative ability of agonists to maximally stimulate [³⁵S]GTPγS binding was compared with three different receptor-binding parameters to determine the extent of correlation. As shown in Fig. 5, plots of relative efficacy versus the percentage of high-affinity sites (Fig.5B) or versus the ratio ofK_DH/K_DL(Fig. 5C) each exhibited a low degree of correlation (r= 0.46 and 0.40, respectively). In contrast, the size of the GTPγS-induced shift in binding was highly correlated with efficacy (Fig. 5A; r = 0.97).

From the perspective of structure-activity relationships, the above-mentioned pattern of graded agonism reveals the critical importance of catechol hydroxyl groups for efficacy at α_2D-adrenergic receptors. Thus, the three phenethylamines that lack either a 3- (meta) or 4- (para) position hydroxyl group (phenylephrine, norphenylephrine, and synephrine) each exhibited less than half the efficacy of EPI (relative efficacies of 0.29, 0.28, and 0.47, respectively) and were also less potent. Absence of the β-hydroxyl group (deoxyepinephrine), however, resulted in a more modest decrease in efficacy (0.78) accompanied by a 14-fold lower potency.

To further examine the importance of these hydroxyl groups, we measured the ability of the antagonist tolazoline and several of its hydroxyl-containing derivatives to alter [³⁵S]GTPγS binding and to bind to α_2A/D-adrenergic receptors. Tolazoline itself failed to either stimulate or inhibit GTP binding (Fig.6), thus behaving as a “null” antagonist, as defined by Costa et al. (1992), under these assay conditions. The addition of a β-hydroxyl group [R-(−)-OH-tolazoline] did not alter the activity of tolazoline. However, the addition of catechol 3,4-dihydroxy groups to tolazoline resulted in the remarkable expression of either agonist or inverse agonist activities, depending on whether an additional hydroxyl group was present on the benzylic carbon or not (Fig. 6). Thus, bothR-(−)-OH-3,4-dihydroxytolazoline andS-(+)-OH-3,4-dihydroxytolazoline increased [³⁵S]GTPγS binding, with the former yielding a relative efficacy of 0.72 and the latter 0.31 at the highest concentrations tested. In marked contrast, 3,4-dihydroxytolazoline reduced basal [³⁵S]GTPγS binding by 55% under standard experimental conditions, which included 100 mM NaCl.

The inhibitory effect of 3,4-dihydroxytolazoline was further examined under Na⁺-free conditions that have been shown previously to favor spontaneous activity of the α_2D-adrenergic receptor, thereby facilitating observation of inverse agonist properties (Tian et al., 1994). 3,4-Dihydroxytolazoline progressively reduced [³⁵S]GTPγS binding up to a maximum of 68% at 1 mM, with an IC₅₀ of 30 μM (Fig.7A), similar to its IC₅₀ in the presence of 100 mM NaCl. When the concentration of NaCl was raised stepwise from 0 to 200 mM, the percentage of inhibition produced by 3,4-dihydroxytolazoline was gradually diminished from 55 to 24%, although inhibition remained significant (P < .05) in the presence of 200 mM NaCl (Fig. 7B).

Receptor binding of R-(−)-OH-tolazoline,R-(−)-OH-3,4-dihydroxytolazoline, and 3,4-dihydroxytolazoline was examined in [³H]rauwolscine displacement studies. As shown in Fig. 8 and Table 2, only the agonistR-(−)-OH-3,4-dihydroxytolazoline showed a biphasic displacement pattern. Tolazoline itself had a significantly higher affinity than any of its hydroxyl-substituted derivatives. Notably, the estimated K_D for 3,4-dihydroxytolazoline (3 μM) was 10-fold lower than its IC₅₀ for inhibition of basal [³⁵S]GTPγS binding.

Discussion

Adrenergic receptor agonists provide efficacy by allosterically modifying receptor conformation such that interaction with cognate G proteins is more fruitful than is the case for the unoccupied receptor. Because unoccupied α₂-adrenergic receptors can activate G proteins (Tian et al., 1994), antagonists can either interfere with agonist binding (null antagonists) or can additionally act to reduce the extent of basal receptor activity (inverse agonists), according to the predictions of a ternary complex model (Costa et al., 1992). In the current study, we have used the combination of radioligand binding and G protein activation studies to examine the relationship between receptor binding and agonist efficacy. Our results indicate that for both phenethylamine and imidazoline agonists, differences in relative efficacy are highly correlated with the extent to which GTP alters their own receptor binding. This implies the bidirectional transmission of allosteric effects by agonists and GTP across the receptor/G protein interface.

Measurement of agonist efficacy at the level of G protein activation rather than at subsequent coupling steps has several theoretical advantages and the use of a cloned receptor expression system in a native membrane environment also provides practical advantages. Thus, a single receptor may couple simultaneously to several different G proteins, initiating multiple response pathways. For example, agonist stimulation of α₂-adrenergic receptors in certain intact cells, including PC12 cells, has been shown to activate both G_i and G_S (Jones et al., 1991; Duzic and Lanier, 1992; Eason et al., 1994), although in isolated PC12/α_2D-membranes EPI stimulation of [³⁵S]GTPγS binding is completely blocked by pertussis toxin pretreatment (Tian et al., 1994). Indeed, immunoprecipitation studies with PC12/α_2D-membranes used in the current studies indicated that both G_i and G_o are activated (data not shown). Because we currently know of no adrenergic pathway that does not operate via G protein activation, measurement of [³⁵S]GTPγS binding should capture all avenues of agonist efficacy in a quantitatively reliable manner.

The relationship between binding of agonist to adrenergic receptors and their ability to provide efficacy has been the subject of a number of previous studies, yielding conflicting results (Kahn et al., 1982;Hoffman et al., 1988; Paris et al., 1989). Several studies have reported a positive correlation between agonist efficacy (measured as regulation of adenylate cyclase activity) and the percentage of high-affinity binding for both β- (Hoffman et al., 1988) and α₂-adrenergic receptors (Paris et al., 1989). However, Hoffman et al. (1988) failed to find such a correlation for α₂-adrenergic receptors in platelet membranes but did find a correlation between efficacy and the ratio of low- to high-affinity K_D values. Similarly,Galitsky et al. (1989) found a higher value ofK_DL/K_DHfor the full agonist UK14304 versus the partial agonist clonidine at α₂-receptors in adipocytes. Minneman and Abel (1987) did not find a good correlation for either the percentage of high-affinity sites orK_DL/K_DHand agonist efficacy at α₁-adrenergic receptors.

Rauwolscine is an inverse agonist at α₂-adrenergic receptors (Tian et al., 1994), and this property may have facilitated our observation of a correlation between agonist efficacy and GTP-induced shifts in [³H]rauwolscine displacement. Inverse agonists exhibit a strong preference for the inactive R conformation of the receptor, whereas agonists, even weaker partial agonists, preferentially bind to the active R^* state. The effect of GTP to convert all receptors to the R state should therefore be more critical for an inverse agonist radioligand such as rauwolscine.

In the current studies, the relative efficacy of agonists in promoting [³⁵S]GTPγS binding was well correlated with values reported from functional assays, such as inhibition of neurotransmitter release or platelet aggregation (Fig. 3B). This indicates that despite the potential caveats, these systems do provide useful indications of the extent of G protein activation by full and partial agonists. If the G protein involved in the functional response is the dominant contributor to [³⁵S]GTPγS binding (e.g., G_i), this would be the case. The close correlation between agonist potency andK_A values (Fig. 3A) may seem unexpected because numerous studies have found that agonist potency in functional assays does not correlate with eitherK_DH orK_DL values. However,K_A values, commonly determined via the “Furchgott Method” of partial receptor inactivation, typically are intermediate between K_DH andK_DL values and thus represent a functional composite of high- and low-affinity receptor states.

The tight correlation between relative agonist efficacy and GTP-dependent decrease in agonist affinity that we found is consistent with bidirectional transfer of allosteric effects between two interacting regulatory proteins. Thus, α₂-receptor binding of agonists promotes formation of a productive complex with cognate G proteins, whereas GTP binding promotes dissociation of this complex. The critical surface of the receptor for complexation with G_i includes a segment at the cytoplasmic terminus of transmembrane helix 6 (Liu et al., 1995). In spin label studies with bovine rhodopsin, this region was shown to exhibit helical structure and its orientation was highly responsive to receptor activation (Farrens et al., 1996). By analogy, conformational changes induced by agonist binding may reposition this segment so that it can participate in G protein complexation and activation in the R^* state. At the same time, complexation with the G protein also would serve to stabilize this region of the receptor in the high-affinity R^*, mediated in particular by constraint of helix 6. In short, the ability of agonists to reposition helix 6 may be a critical aspect of their efficacy, and G proteins may stabilize the repositioned helix in the R^* state of the receptor.

Crystallographic studies of G proteins have identified critical “switch” regions in α-subunits that differentially respond to the presence of guanyl nucleotides and are thought to participate in receptor complexation and transmission of allosteric influences to the nucleotide-binding pocket (Coleman et al., 1994; Mixon et al., 1995). Specifically, the surface formed by carboxyl and amino termini has been proposed to be critical for receptor interaction (Sullivan et al., 1987). Binding of GTP introduces disorder in this region (Mixon et al., 1995) that could serve to destabilize the ternary complex with the agonist-occupied receptor, accounting for loss of the high-affinity agonist binding state. Because the same structural features are involved in both antegrade (AR^*G formation) and retrograde (AR^*G dissociation) events, a close correlation between stimulated GTP binding and GTP-dependent loss of affinity would therefore be expected.

Based on the above-mentioned analysis, differences in agonist efficacy may be related to a differential ability to reposition transmembrane helix 6. In the catecholamine ligand-binding pocket, agonists are thought to bind to helix 6 via pi electron bonding interactions with a phenylalanine residue (Phe-391 in the α_2D-adrenergic receptor) (Dixon et al., 1988). Being small, diffusable molecules, agonists are intrinsically limited in their ability to effect the movement of a protein motif, such as a helical element. However, once they are initially bound to the receptor in its R state, further bonding opportunities provided by the now immobile ligand can be much more effective. In the case of phenylethylamines (and imidazolines), initial ionic and hydrogen bond interactions can serve the role of immobilizing the phenyl moiety. Variations in both the strength of these interactions and in their spatial outcomes (i.e., positioning of the phenyl ring) will determine the effectiveness of immobilization on receptor activation.

We therefore propose that differences in agonist efficacy observed among hydroxyl-substituted phenethylamines at α_2D-adrenergic receptors may reflect variations in their ability to immobilize their phenyl ring for binding to Phe-391. Among the compounds we examined, the β-hydroxyl group was less critical for efficacy than either of the catechol hydroxyl groups, suggesting that the latter may be more important for determining phenyl ring orientation, as might be expected. However, the same was not true for binding potency. Such a differential contribution of hydroxyl substituents to agonist efficacy versus potency is well established (Ruffolo et al., 1979; Ruffolo, 1984), and phenolic hydroxyls have previously been shown to be critical for efficacy at α₂-receptors, whereas primarily affecting agonist potency at α₁-receptors (Ruffolo et al., 1984).

UK14304 is a full agonist at α_2D-adrenergic receptors but lacks hydroxyl groups, suggesting the possibility that its phenyl ring may be stably positioned by alternative means. As illustrated in Fig. 2, UK14304 possesses two aromatic nitrogens in a quinoxaline ring with positions equivalent to catechol hydroxyl groups. In their ring-stabilized position, these nitrogens may afford H-bonding or van der Waals bonding to residues in helix 5, thereby serving a role analogous to hydroxyl substituents.

Our studies with hydroxyl-substituted tolazoline derivatives also suggest that hydroxyl groups on the phenyl ring may be important in determining its orientation and presentation for binding to Phe-391. Thus, in the absence of both catechol hydroxyls no agonism was detected [tolazoline and R-(−)-hydroxytolazoline], whereas 3,4-dihydroxytolazoline was a strong inverse agonist (Figs. 6 and 7) and R-(−)-OH-3,4-dihydroxytolazoline was a full agonist (Fig. 6). This is consistent with a specific positioning of the phenyl ring in the latter compounds, leading to either a negative or positive influence on receptor/G protein complex formation. Moreover,R-(−)-OH-3,4-dihydroxytolazoline binding was biphasic (Table 2) whereas 3,4-dihydroxytolazoline binding was not, indicative of the presence and absence of G protein complexation, respectively.

The remarkable inverse agonism of 3,4-dihydroxytolazoline suggests that it induces a receptor conformation that is less capable of activating G proteins than the unoccupied receptor. Its maximum reduction in GTP binding of 68% is approximately twice that produced by rauwolscine under the same conditions (Tian et al., 1994). Thus, either 3,4-dihydroxytolazoline is a more efficacious inverse agonist at α_2D-adrenergic receptors, or it acts on additional receptors. The bound position of the inverse agonist may physically block helix 6 from even its basal probability of moving to an active R^* location. Because the addition of anR-(−)-hydroxyl group to the benzylic carbon converts 3,4-dihydroxytolazoline to a full agonist, the difference between producing a highly favorable versus a highly unfavorable influence on receptor/G protein interaction clearly can be subtle.

Although our results indicate that hydroxyl group-dependent repositioning of helix 6 may be an important aspect of agonist-induced α₂-adrenergic receptor activation, overall conformational events are likely to be more complex, involving additional regions of the receptor. This is especially true in light of the ability of G_i-coupled receptors such as the α₂-receptor to also couple to G_S in certain cellular environments (Duzic and Lanier, 1992; Jones et al., 1991; Eason et al., 1994). Thus, it appears that α₂-receptors may exist in multiple agonist-induced active conformations that display unique G protein selectivity. Additional studies will be needed to determine how the structural features of agonist ligands can direct receptor signaling among these alternative G protein pathways.