Introduction

Top Summary Introduction Materials and Methods Results and Discussion References

Mechanisms of partial agonism for receptors coupled to heterotrimeric G proteins remain unresolved. Explanations of such ligand behavior must incorporate concepts of inverse agonism and receptor G protein precoupling, the "energy landscape" generated by multiple conformations of the receptor and the cell-specific manifestation of this phenomenon. For example, a specific drug can behave as a partial agonist in one tissue, a full agonist in another, and an antagonist in a third system (Steer and Atlas, 1982; Kenakin, 1984; North and Surprenant, 1985; Surprenant et al., 1990; Gollasch et al., 1991; Hoyer and Boddeke, 1993; Eason et al., 1994). The cellular responses mediated by the great majority of G protein-coupled receptors are also cell specific. The realization that a single receptor molecule functions in a cell-specific manner is a simple point but has broad implications. First, these observations indicate that the signaling system is dynamic and likely developmentally regulated and responsive to physiological and nonphysiological challenges. Second, the action of an agonist/antagonist at a receptor in one cell may be different from that observed for the same receptor in another cell type. Third, the action of an agonist/antagonist in a normal cell can be quite different from its action in the same cell after initiation of a disease process.

View larger version (18K): [in this window] [in a new window]   Fig. 1.   Receptor-mediated activation of G proteins in NIH-3T3 fibroblasts expressing 2-AR. G protein activation was determined by measuring agonist-induced increases in the binding of GTP35S in membrane preparations. A, G protein activation was determined in the presence of vehicle (basal), epinephrine (100 µM), or clonidine (100 µM) for various incubation times at 24°C. B, G protein activation was determined in the presence and absence of increasing concentrations of epinephrine, UK14304, clonidine, and oxymetazoline (24°C for 30 min). Data are expressed as the percentage of maximal response elicited by epinephrine (100 µM). C, influence of pertussis toxin pretreatment (100 ng/ml, 18 h) of cells on G protein activation by agonist. Receptor density (2-AR no. 1) = 5.3 ± 0.5 pmol/mg membrane protein. Data are expressed as means ± S.E. generated from three (A and B) or four separate experiments (C).

View larger version (38K): [in this window] [in a new window]   Fig. 2.   Receptor-mediated activation of G proteins in 2-AR/Go1 and 2-AR/Gi1 transfectants. A, Transfectants were evaluated for Go1 or Gi1 expression by immunoblotting with selective antiserum (left panel: membrane protein-rat brain = 10 µg, NIH-3T3 = 50 µg; right panel: membrane protein-rat brain = 20 µg, NIH-3T3 = 100 µg). G protein activation by 2-AR agonists was determined by measuring agonist-induced increases in GTP35S binding to membranes prepared from 2-AR/Go1 no. 2 (receptor density = 8.7 ± 0.48 pmol/mg membrane protein) or 2-AR/Gi1 (receptor density = 36 ± 0.2 pmol/mg membrane protein) transfectants. B, GTP35S binding was determined in the presence of epinephrine (100 µM) or clonidine (100 µM) for various incubation times at 24°C. C, GTP35S binding (24°C for 30 min) was determined in the presence of increasing concentrations of agonists. Data in B are presented as means ± S.E. derived from two to three experiments. Inset, comparison of the time course of G protein activation by clonidine and epinephrine in 2-AR (), 2-AR/Go1 (), and 2-AR/Gi1 () transfectants. Data from the first 10 min of the time course presented in Figs. 1A and 2B are compared for the three model systems. Data are expressed as the percentage of the maximal response achieved by each individual receptor agonist. Data in C are expressed as the percentage of the maximal response achieved with the selective 2-AR agonist UK14304 or the adrenergic agonist epinephrine and are presented as means ± S.E. derived from two experiments.

View this table: [in this window] [in a new window]   TABLE 1 Influence of receptor density and G protein expression on relative efficacy of clonidine G protein activation was determined by measuring agonist-induced increases in the binding of GTP35S in membrane preparations. Data are presented as mean ± S.E. derived from three to six experiments with duplicate determinations. Refer to Figs. 1B and 2C for the full dose-response curve for 2 AR no. 1, 2-AR/Gi1, and 2-AR/Go1 no. 2 transfectants. Level of Go1 expression was similar in both 2-AR + Go1 no. 2 and 2-AR + Go1 no. 3 transfectants.

View larger version (22K): [in this window] [in a new window]   Fig. 3.   Receptor-mediated activation of G proteins in 2-AR and 2-AR/Go1 transfectants after cell pretreatment with pertussis toxin. Progressive reduction in the amount of G protein available for receptor coupling was achieved by cell pretreatment (18 h) with increasing concentrations of pertussis toxin. GTP35S binding was determined in the absence and presence of epinephrine (100 µM) or clonidine (100 µM) at 30-min incubation time points (24°C). Receptor density: 2-AR no. 1 = 5.3 ± 0.5 pmol/mg membrane protein, 2-AR/Go1 no. 2 = 8.7 ± 0.48 pmol/mg membrane protein. Data are presented as means ± S.E. of two experiments with duplicate determinations.

View larger version (31K): [in this window] [in a new window]   Fig. 4.   Gpp(NH)p-sensitive binding of [3H]clonidine and [3H]UK14304 in 2-AR, 2-AR/Go1, and 2-AR/Gi1 transfectants. Saturation binding isotherms were performed on five cell lines using the 2-AR-selective antagonist [3H]RX821002 (0.025-20 nM) and the 2-AR-selective agonists [3H]UK14304 and [3H]clonidine (0.025-12 nM). For each radioligand, parallel binding studies were conducted in the absence and presence of the nonhydrolyzable GTP analog Gpp(NH)p (100 µM). [3H]RX821002 binding was not influenced by addition of Gpp(NH)p. In contrast, ~80 to 85% of the specific binding of [3H]clonidine and [3H]UK14304 (at ligand concentrations of 2 nM) was Gpp(NH)p-sensitive. For both [3H]clonidine and [3H]UK14304, the specific binding obtained in the presence of Gpp(NH)p was subtracted from that obtained in the absence of Gpp(NH)p at each radioligand concentration to generate the presented data. The data are representative of two to three experiments using different preparations of cell membranes. Kd and Bmax values are presented in Table 2.

View this table: [in this window] [in a new window]   TABLE 2 Ligand recognition properties of 2-AR in different genetic backgrounds For each radioligand, parallel binding studies were conducted in the absence and presence of the nonhydrolyzable GTP analog Gpp(NH)p (100 µM). The binding of [3H]RX821002 was not influenced by the addition of Gpp(NH)p. In contrast, ~80 to 85% of the specific binding of [3H]clonidine and [3H]UK14304 (at ligand concentrations of 2 nM) was Gpp(NH)p-sensitive. For both [3H]clonidine and [3H]UK14304, the specific binding obtained in the presence of Gpp(NH)p was subtracted from that obtained in the absence of Gpp(NH)p at each radioligand concentration to generate a saturation binding isotherm (see Fig. 4). The Gpp(NH)p-sensitive specific binding was evaluated by RADLIG to determine radioligand affinities and binding site densities. "No." refers to independently isolated transfectants. Results are expressed as mean ± S.E. derived from two or three experiments with duplicate determinations.

The realization that there are multiple receptor subtypes at which drug behavior is system-dependent necessitated thought revision on several issues related to concepts of agonism and the basic tenants of receptor theory (Ariëns, 1954; Kenakin and Morgan, 1989; Black, 1989; Weiss et al., 1996; Kenakin, 1997). The classification of partial versus full agonists at G protein-coupled receptors often depends on whether the readout is proximal [i.e., [³⁵S]guanosine-5'-O-(3-thio)triphosphate (GTP³⁵S) binding] or distal (i.e., adenylyl cyclase activity and contraction/relaxation of smooth muscle) within the signal transduction cascade. The degree of receptor reserve for a specific signal transduction pathway also influences data interpretation, and several observations suggest selective activation of different signal transduction pathways by ligand-induced stabilization of specific conformations of the receptor protein (Kenakin, 1995; Perez et al., 1996; Berg et al., 1998; Bonhaus et al., 1998). For receptors capable of coupling to multiple G proteins, the relationships between partial and full agonists may be influenced by the type of G proteins expressed in the cell, stoichiometric considerations, and/or by accessory proteins that regulate the transfer of signal from receptor to G protein and further downstream to various effectors (Coupry et al., 1992; Kataoka et al., 1993; Sato et al., 1995, 1996; Watson et al., 1996; McLatchie et al., 1998; Cismowski et al., 1999; A. Takesono, M. Cismowski, M. Bernard, C. Ribas, P. Chung, E. Duzic, and S.M.L., unpublished observations). As an initial approach to this issue, we determined the relationship between full and partial agonists for a typical G protein-coupled receptor before and after altering the population of G proteins expressed in the cell.

	Materials and Methods

Top Summary Introduction Materials and Methods Results and Discussion References

Experimental Procedures. Techniques used for cell culture, membrane preparations, and radioligand binding studies were conducted as described previously (Coupry et al., 1992; Kataoka et al., 1993; Sato et al., 1996). For [³⁵S]GTPS binding, cell membrane preparations were resuspended in assay buffer (5 mM MgCl₂, 1 mM EDTA, 1 mM dithiothreitol, 100 mM NaCl, 1 µM guanosine diphosphate, 1 µM propranolol, 50 mM Tris-HCl, pH 7.4). The reaction was initiated by adding membranes (10 µg in 10 µl) to tubes containing 90 µl of assay buffer containing [³⁵S]-GTPS (0.2 nM; 1250 Ci/mmol) and agonist or vehicle. Samples were incubated at 24°C for various times and the reactions terminated by rapid filtration through nitrocellulose filters with 4 × 4 ml of wash buffer (100 mM NaCl, 50 mM Tris-HCl, 5 mM MgCl₂, pH 7.4, 4°C). Radioactivity bound to the filters was determined by liquid scintillation counting. Nonspecific binding was defined by 100 µM GTPS.

	Results and Discussion

Top Summary Introduction Materials and Methods Results and Discussion References

Agonism was quantitated at the level of G protein itself. We used a cellular system engineered to express a typical G protein-coupled receptor (₂-AR) at varying receptor densities. We specifically focused on a transfected cell model so that we could control the relative amounts of receptor and provide the same microenvironment for analysis of receptor-mediated activation of G protein subunits. Such an approach overcomes the difficulties presented by cell-specific signaling events and the heterogeneity of signaling proteins when one attempts to address specific events at the receptor-G protein interface. NIH-3T3 fibroblasts were stably transfected with an AR subtype, rat _2A/D-AR, and G protein activation was determined by measuring agonist-induced increases in the binding of the nonhydrolyzable GTP analog GTP³⁵S. In membranes prepared from ₂-AR transfectants, the adrenergic agonist epinephrine and the selective ₂-AR agonist clonidine activated G proteins in a time- and concentration-dependent manner (Fig. 1A and B). The action of both ligands required expression of the ₂-AR. It is important to realize that under these incubation conditions in membrane preparations, a significant component of GTP³⁵S binding is reversible and thus an equilibrium is achieved (Q.Y. and S.M.L., unpublished observations; also refer to discussion in Breivogel et al., 1998). In this system, the major rate-limiting step for GTP³⁵S binding to membrane G proteins in response to agonist is an agonist-induced decrease in the affinity of G for GDP (Lorenzen et al., 1996; Selley et al., 1997; Breivogel et al., 1998). The ability of an agonist to promote the latter event is intimately related to its efficacy (Lorenzen et al., 1996; Selley et al., 1997).

In NIH-3T3 ₂-AR transfectants, the ₂-AR agonists clonidine and oxymetazoline behaved as partial agonists, whereas the selective ₂-AR agonist UK-14304 and the adrenergic agonist epinephrine behaved as full agonists. Receptor-mediated activation of G proteins in membrane preparations was blocked by cell pretreatment with pertussis toxin, indicating receptor coupling to a subgroup of pertussis toxin-sensitive G protein (Gi1-3 and Go1,2; Fig. 1C). Immunoblotting indicated that within the subgroup of pertussis toxin-sensitive G proteins, NIH-3T3 cells expressed Gi2 and Gi3, but not Gi1 or Go (Duzic et al., 1992), and thus the signal evaluated must be mediated through Gi2/Gi3.

Although clonidine and oxymetazoline are generally classified as partial agonists at ₂-AR, their relative efficacy is somewhat tissue and effector specific (Steer and Atlas, 1982; North and Surprenant, 1985; Surprenant et al., 1990; Gollasch et al., 1991; Eason et al., 1994). To determine the influence of G protein type on the relationship between partial and full agonists, we re-evaluated ₂-AR-mediated activation of G proteins following cotransfection of NIH-3T3 fibroblasts with the receptor and two functionally distinct G protein  subunits (Gi1 or Go1). In ₂-AR/Go1 cotransfectants, the amount of G protein activated by the ligands clonidine and oxymetazoline was identical with that elicited by the full agonists epinephrine and UK-14304 (Fig. 2). The enhanced agonist-induced activation of G protein in ₂-AR/Go1 cotransfectants was apparent after only short incubation times (Fig. 2B, inset). Go1 expression also increased the potency of each ₂-AR agonist [EC₅₀: epinephrine, 2.0 ± 0.3 versus 0.3 ± 0.02 µM ( p < .01); UK14304, 1.7 ± 0.2 versus 0.38 ± 0.02 µM ( p < .01); clonidine, 1.1 ± 0.1 versus 0.19 ± 0.01 µM ( p < .01); oxymetazoline, 0.4 ± 0.03 versus 0.05 ± 0.01 µM ( p < .01); p values refer to ₂-AR/Go1 (8.74 ± 0.48 pmol receptor/mg membrane protein) versus ₂-AR transfectants (5.3 ± 0.5 pmol receptor/mg membrane protein]. In contrast to the results obtained in ₂-AR/Go1 cotransfectants, clonidine and oxymetazoline still behaved as partial agonists in cells stably transfected with the ₂-AR and the G protein Gi1, even though receptor expression was 3 to 15 times greater than that achieved in the receptor/Go1 transfectants (Fig. 2 and Table 1). (Expression of G subunits in NIH-3T3 fibroblasts did not alter basal GTP³⁵S binding and did not increase the maximal response elicited by the full agonist epinephrine under these assay conditions. This likely reflects several factors including the use of subsaturating concentrations of GTP³⁵S, the inclusion of GDP in the assay system and the reversible binding of GTP³⁵S in this system.) Previous studies indicated that the ₂-AR is capable of coupling to Gi1, and that the expressed Gi1 protein in NIH-3T3 fibroblasts is functional based on its ability to alter cell responsiveness to transforming growth factor- (Kataoka and Lanier, 1993; Bahia et al., 1998). The differences in the relationship of partial and full agonists in cells expressing ₂-AR/Go1 versus ₂-AR alone were also not altered by a progressive reduction in the amount of G protein available for coupling to the receptor (Fig. 3). These data indicated that the increased efficacy of clonidine and oxymetazoline following the expression of Go1 was not simply an issue of G protein or receptor levels, but was related to the type of G protein available for coupling to receptor. Indeed, the influence of Go1 on the relative efficacy of clonidine and epinephrine in our system parallels the behavior of clonidine as a full agonist at ₂-ARs in systems where the receptor likely couples to Go heterotrimer (e.g., receptor-mediated inhibition of calcium channels; Surprenant et al., 1990; Gollasch et al., 1991; Trombley, 1992).

These observations have significance relative to concepts of conformational induction versus conformational selection in terms of agonist efficacy and the development of cell-specific drug efficacy. Relative to the data reported in this article, three mechanistic concepts of partial agonism should be considered. First, full and partial agonists may stabilize different receptor conformations varying in their affinity for G protein or in their ability to activate G protein (a type of conformational induction by agonists). In this situation, receptor conformations stabilized by the partial agonist clonidine or the full agonist epinephrine would differ in their ability to activate Gi, but would be functionally equivalent in terms of activating Go. Second, full and partial agonists may stabilize the receptor in the same active conformation, but the two types of compounds generate quantitatively different amounts of this receptor conformation. In the Gi, but not Go1 genetic background, the relative amount of such a complex would then be rate limiting in terms of the total amount of G protein that can be activated.

Third, for receptors that are precoupled, the receptor that is precoupled to G protein X may be stabilized in a conformation that is different from that of a receptor that is precoupled to G protein Y (a type of conformational induction by G proteins). Full and partial agonists may differ in their affinity for these two conformations of receptor. Alternatively, these two receptor conformations may differ in the efficiency in which they mediate signal transfer from receptor to G protein when the agonist site is occupied. In one scenario, full agonists would interact with all precoupled R-G complexes (e.g., both G protein X and G protein Y), whereas a partial agonist would only interact with a subgroup of precoupled R-G complexes (e.g., G protein X or G protein Y). This concept can be further extended if one assumes that the precoupled receptor G protein complex is totally responsible for the readout of agonist activation. In such a case, the differences between clonidine and epinephrine observed in the two genetic backgrounds may simply reflect a greater propensity of Go1 to generate the precoupled complex as compared with Gi.

As one approach to these issues, we determined whether expression of Go1 or Gi1 influenced the population of receptors interacting with agonist. The conformation of receptors for which agonists exhibit high affinity is stabilized by interaction with G proteins and this complex is disrupted in radioligand binding assays by the addition of a nonhydrolyzable analog of GTP. We simultaneously evaluated the radioligand binding properties of the antagonist ([³H]RX821002) and agonist ([³H]clonidine and [³H]UK14304) radioligands in the three genetic backgrounds (control and Gi1 and Go1 transfectants). Both [³H]clonidine and [³H]UK14304 exhibited high affinity, Gpp(NH)p-sensitive binding in each cell line and the amount of high-affinity, Gpp(NH)p-sensitive binding was related to the receptor densities (as determined with the antagonist radioligand [³H]RX821002) of the individual cell lines (₂-AR no. 1 versus ₂-AR no. 2; ₂-AR/Go1 no. 1 versus ₂-AR/Go1 no. 2; Fig. 4 and Table 2). In each genetic background, the number of Gpp(NH)p-sensitive binding sites for [³H]UK14304 was greater than that observed with [³H]clonidine. Thus, the conversion of clonidine from a partial to full agonist by expression of Go1 was not due to an increase in the Gpp(NH)p-sensitive, high agonist-affinity receptor population. There was also no change in the affinities exhibited by UK14304 and clonidine at this Gpp(NH)p-sensitive binding site identified in the radioligand binding studies in the different genetic backgrounds (Table 2).

These data suggest that there is little difference in the ability of clonidine to interact with or stabilize ₂-AR-Gi2/Gi3 versus ₂-AR-Go1 complexes, but that the subsequent step of signal transfer from receptor to G protein is more readily achieved for the clonidine/₂-AR/Go1 complex. This interpretation may reflect structural aspects of the interaction between the ₂-AR and Gi2/Gi3 versus Go1 or perhaps the differences in the nucleotide binding properties of the G proteins themselves such that Go1 is "easier" to activate. Relative to Gi2,3, Go1 has a higher basal rate of nucleotide exchange (Ferguson et al., 1986; Carty et al., 1990), and in this sense one might conclude that Go1 is actually "primed" for activation by receptors.

The influence of G protein type on the relationship between full and partial agonism indicates that agonism, or efficacy, is a dynamic phenomenon that is intimately related to the array of proteins expressed by a cell at any given stage of development or in specific disease states. Thus, the effects of a particular drug would be maximized in a tissue where it behaves as a full agonist and diminished in tissues where it behaved as a partial agonist. The same thoughts may apply to inverse agonists. Such information may lead to the development of drugs that display a spectrum of agonistic behavior in a cell type-specific manner such that these distinctions work to therapeutic advantage.