Agonist-independent Activation of Gz by the 5-Hydroxytryptamine1A Receptor Co-expressed inSpodoptera frugiperda Cells

The human 5-hydroxytryptamine1Areceptor, when expressed in Spodoptera frugiperda (Sf9) cells, facilitates the binding of [35S]GTPγS to a co-expressed GTP-binding regulatory protein, Gz, consistent with constitutive activity. The antagonists 4-(2′-methoxyphenyl)-1-[2′(n-2"-pyridinyl)-p-iodobenzamido]ethyl-piperazine (p-MPPI) and the related fluorobenzamido analoguep-MPPF had little (p-MPPI) or no (p-MPPF) effect on this activity. In contrast, a third antagonist, the neuroleptic spiperone, produced an almost complete suppression. Thus, using G protein activation as an index of receptor activity, p-MPPF was classified as a neutral antagonist,p-MPPI as a partial inverse agonist, and spiperone as essentially a full inverse agonist. As predicted, spiperone displayed properties consistent with a special form of noncompetitive antagonism when used to displace the agonist [125I]R-(+)-trans-8-hydroxy-2-[N-n-propyl-N-(3′-iodo-2′-propenyl)amino]tetralin. Our data profile Sf9 cells as a unique vehicle for the characterization of inverse agonists, as these cells support a systematic pairing of mammalian receptors and G proteins, quantitative assays of G protein activation, and unambiguously labeled populations of coupled and uncoupled receptors.

The human 5-hydroxytryptamine 1A receptor, when expressed in Spodoptera frugiperda (Sf9) cells, facilitates the binding of [ 35 S]GTP␥S to a co-expressed GTP-binding regulatory protein, G z , consistent with constitutive activity. The antagonists 4-(2-methoxyphenyl)-1-[2-(n-2؆-pyridinyl)-p-iodobenzamido]ethyl-piperazine (p-MPPI) and the related fluorobenzamido analogue p-MPPF had little (p-MPPI) or no (p-MPPF) effect on this activity. In contrast, a third antagonist, the neuroleptic spiperone, produced an almost complete suppression. Thus, using G protein activation as an index of receptor activity, p-MPPF was classified as a neutral antagonist, p-MPPI as a partial inverse agonist, and spiperone as essentially a full inverse agonist. As predicted, spiperone displayed properties consistent with a special form of noncompetitive antagonism when used to displace the agonist [ 125

I]R-(؉)-trans-8-hydroxy-2-[N-n-propyl-N-(3-iodo-2-propenyl)amino]tetralin. Our data profile Sf9 cells as a unique vehicle for the characterization of inverse agonists, as these cells support a systematic pairing of mammalian receptors and G proteins, quantitative assays of G protein activation, and unambiguously labeled populations of coupled and uncoupled receptors.
The 5-hydroxytryptamine 1A (5-HT 1A ) 1 receptor is one of several subtypes of receptors for serotonin (5-HT) that mediate the actions of this agonist on neuronal activity. The 5-HT 1A receptor exists at particularly high densities in the hippocampus and components of the limbic system and can be divided functionally into two populations -somatodendritic autoreceptors in the dorsal raphe nucleus, which mediate the stimulation of K ϩ channels, and postsynaptic receptors elsewhere, which mediate both the stimulation of K ϩ channels and inhibition of adenylyl cyclase (reviewed by Hoyer et al. (1)). Other activities ascribed to the 5-HT 1A receptor include the activation of the mitogenactivated protein kinases ERK1 and ERK2 (2), activation of nuclear factor-B (3), and stimulation of Na ϩ /H ϩ exchange (4). All of these actions appear to be exerted through one or more members of the G i family of GTP-binding regulatory proteins (G proteins), which comprises G i , G o , and G z (5). The 5-HT 1A receptor can also interact to some extent with members of the G q family to achieve activation of a phosphoinositide-specific phospholipase C depending on the cell and/or concentration of agonist (6,7). The deduced structure of the 5-HT 1A receptor resembles that of the many other G protein-coupled receptors (GPCRs) now identified (8,9).
The classical model of GPCR action implies that the binding of an agonist to a receptor, e.g. serotonin to the 5-HT 1A receptor, is essential for activation of the receptor and transmission of the biological signal across the plasma membrane. However, studies of several GPCRs operating through distinct G proteins have revealed that receptor activation can occur to some extent spontaneously in the absence of an agonist. Agonist-independent, or constitutive, activity has been reported for the ␦-opioid, ␣ 2 -adrenergic, and muscarinic cholinergic receptors expressed normally (26 -28) and for the dopamine D5 (29), ␤ 2 -adrenergic (30), 5-HT 1B (31), 5-HT 1D (31), and 5-HT 2C (32) receptors overexpressed in mammalian cells following transfection. Overexpression of the ␤ 2 -adrenergic receptor in a transgenic mouse model also elicits a marked set of agonist-independent responses (33). Mutations in GPCRs, introduced near the junction of the third cytoplasmic loop and transmembrane domain 6 (30,34) or occurring naturally throughout the receptor (35,36), can dramatically enhance constitutive activity. In part to account for constitutive activity but also to explain changes in ligand affinity as a function of conformations underlying this activity, the ternary complex model has been revised to include an equilibrium between inactive (R) and active (R*) conformations of the receptor (37,38). The active conformation, which through the established equilibrium can exist in some small amount without agonist, is capable of binding to and activating a relevant G protein. Agonists bind preferentially to R* and/or R*G, and thereby shift the equilibrium toward R* as represented by agonist⅐R* and agonist⅐R*⅐G complexes.
The discovery of constitutive receptor activity has opened the way for reclassification of "antagonists" as neutral antagonists and inverse agonists (sometimes called reverse antagonists). Neutral antagonists, through competition with agonists for binding to the receptor, block the actions of agonists but have no effect on constitutive activity. Neutral antagonists are thought to bind R and R* in an equivalent fashion, thereby preserving the existing equilibrium between these two forms of receptor. Inverse agonists not only block the actions of agonists but suppress constitutive activity. Just as agonists shift the equilibrium toward R* and R*G by selectively binding R* forms of the receptor, inverse agonists may shift the equilibrium toward R by binding the R form of receptor in preference to R* or R*G. Inverse agonism was first recognized as a property of ␤-carbolines at the unrelated GABA A (␥-aminobutyric acid) receptor (39) and has subsequently been identified for a number of GPCRs (26, 32, 40 -43).
Spodoptera frugiperda (Sf9) cells represent a powerful model with which to characterize the activity of mammalian GPCRs (44 -49). Receptors, together with selected G protein ␣, ␤, and ␥ subunits, can be co-expressed in intact Sf9 cells through infection with appropriate recombinant baculoviruses. Receptor activity (with or without agonist) can then be characterized in subsequently isolated membranes as a facilitation of [ 35 S]GTP␥S binding to the G protein ␣ subunit (49). The protocol of reconstitution and assay using Sf9 cells is particularly advantageous for the reasons that the pairing of receptor with G protein can be defined precisely, that little or no interference is imposed by endogenous receptors and G proteins, and that G protein activation is a more direct correlate of intrinsic efficacy than effector activity. We demonstrated previously the expected selectivity in receptor⅐G protein coupling for several receptors working through one or more of the four families of G proteins (49). Of interest in the previous study was the observation that the 5-HT 1A receptor alone of the receptors examined expressed a perceptible level of what appeared to be constitutive activity. We confirm here that the 5-HT 1A receptor exhibits constitutive activity and, using this activity as a starting point, demonstrate that the Sf9 cell reconstitution model can provide a direct and quantifiable assay to distinguish neutral antagonists from inverse agonists. We find that p-MPPF is a neutral antagonist, whereas p-MPPI and spiperone, respectively, are partial and full inverse agonists. We also describe the means by which unambiguously defined populations of uncoupled and coupled receptor can be constructed in Sf9 cells, and we provide definitions of efficacy based on G protein activation and ligand displacement analysis.

EXPERIMENTAL PROCEDURES
Baculoviruses-Recombinant baculoviruses encoding the G protein subunits ␤ 1 and ␥ 2 (50) were kindly provided by Drs. T. Kozasa and A. Gilman at Southwestern Medical Center (Dallas, TX). Those encoding the 5-HT 1A receptor and ␣ z were constructed in our laboratory (47).
Cell Culture, Membrane Preparation, and Assay of [ 35 S]GTP␥S Binding-The procedures for handling Sf9 cells and assaying [ 35 S]GTP␥S binding have been described previously (47,49). Essentially, Sf9 cells were propagated in suspension culture with TNM-FH medium containing charcoal-treated serum. For infection with recombinant baculoviruses, the cells were subcultured in monolayer and infected with one or more viruses at a multiplicity of infection of at least one for each virus. The medium was replaced 16 h following infection with Sf900II opti-mized serum-free medium (Life Technologies, Inc.). The cells were harvested at 48 h, and membranes were prepared by differential pelleting following lysis of the cells under hypotonic conditions. [ 35 S]GTP␥S binding was assayed by incubation of the membranes (20 g of protein per assay point) with or without agonists and/or antagonists for 5 min at 30°C with 30 nM [ 35 S]GTP␥S (1300 Ci/mmol) in the presence of 1 M GDP and 3 mM free Mg 2ϩ . In experiments with antagonists, membranes were preincubated for 10 min at room temperature with the antagonists prior to initiating the binding assay. Immunoprecipitation of ␣ z was accomplished with antiserum 6354 (directed toward residues 24 -33 (51)) following extraction of the membranes with 0.5% Nonidet P-40. Bound radioactivity was quantitated by scintillation spectrometry.
Measurement of 5-HT by High Pressure Liquid Chromatography-The procedures of analysis were those of Kreiss et al. (52). Sf9 cell membranes were resuspended in 0.1% ice-cold perchloric acid and sonicated briefly. Membrane protein was pelleted by centrifugation at 16,000 ϫ g for 5 min, and the supernatant was filtered through a 0.2 m filter. Samples were applied to a Bioanalytical Systems 480 reversephase microbore column (1 ϫ 100 mm), and fractions were analyzed with an LC-4C electrochemical detector. The mobile phase consisted of 0.67 mM EDTA, 0.43 mM sodium octyl sulfate, 10 mM NaCl, 32 mM NaH 2 PO 4 , and 11% acetonitrile, pH 4.0. 5-HT, which had a retention time of 8 min, was used as a standard. The limit of detection was 0.1 nM. [

RESULTS
Sf9 cells expressing the human 5-HT 1A receptor and G z (i.e. ␣ z , ␤ 1 , and ␥ 2 ) were used to explore the properties of compounds that bind to the 5-HT 1A receptor. G z is a member of the G i family and has been used extensively by us to study the coupling of the receptor to G proteins (47,49). The chief advantage of G z is that it does not bind significant levels of [ 35 S]GTP␥S when expressed alone and thus affords a clear picture of activation achieved through the receptor. Activation is equated here with the binding of [ 35 S]GTP␥S to ␣ z in Sf9 cell membranes as quantitated following immunoprecipitation.
A fraction of G z co-expressed with the 5-HT 1A receptor was found to assume an activated state in the presence of [ 35 S]GTP␥S but absence of added agonist (Fig. 1, hatched column). As implied above, the activation was not achieved if G z was expressed without the receptor (49). Moreover, the activation was not the consequence of 5-HT carried over from the medium used in cell culture. Serum-free conditions were employed during the time at which the 5-HT 1A receptor was expressed, and no 5-HT was detected in membranes by direct analysis. These data demonstrate that the 5-HT 1A receptor expressed in Sf9 cells exhibits a discernible level of constitutive activity. The receptor is not fully active, however, as 5-HT promoted further activation of G z (Fig. 1, curve). The activation by 5-HT was dose-dependent, characterized by an EC 50 (14 Ϯ 6 nM) comparable to that determined in mammalian cells.
The compounds p-MPPI, p-MPPF, and spiperone are known to be antagonists of 5-HT at the 5-HT 1A receptor (13)(14)(15)(16). As shown in Fig. 2A, all three compounds completely inhibited the activation of G z evoked by 5-HT. IC 50 values of p-MPPI, p-MPPF, and spiperone (at 100 nM 5-HT) were about 30, 40, and 310 nM, respectively. Of interest was the observation in these experiments that spiperone, and possibly to some extent p-MPPI, might actually suppress the activation of G z to a level below that attributed to constitutive receptor activity. Such an action would represent a property of inverse agonism. The possibility was explored further, as shown for the receptor incubated without 5-HT in Fig. 2B. p-MPPF had no effect on constitutive activity at a concentration shown above to be sufficient to block the response to 100 nM 5-HT, and p-MPPI suppressed activity by only a small extent (16 Ϯ 5%, which is significant (p Ͻ 0.01)). An almost complete suppression of constitutive activity, however, was achieved with spiperone. The IC 50 for this action was 68 nM. These data imply that p-MPPF is a neutral antagonist, p-MPPI is a weak inverse agonist, and spiperone is a full inverse agonist.
The effect of spiperone on constitutive receptor activity was completely reversed by p-MPPF (Fig. 3). A similar reversal was achieved with p-MPPI, with the response returning to that observed in the presence of p-MPPI alone. These results confirm that the 5-HT 1A receptor is the site of action for spiperone and that the constitutive activity of the receptor is not attributable to 5-HT carried over from the cell culture medium. Importantly, the data as a whole demonstrate that the direct activation of a G protein as reconstituted in Sf9 cells can serve as a monitor of constitutive activity and can thereby be used to distinguish inverse agonists from neutral antagonists.
The ternary complex model predicts that spiperone, as an inverse agonist, would have a higher affinity for the receptor in the absence of G z than in its presence, whereas a neutral antagonist would have an equal affinity in both cases. Initial binding experiments with [ 3 H]spiperone were unsuccessful due to the high level of nonspecific binding at the high concentrations required. We therefore carried out competition assays using [ 125 I]p-MPPI as the radioligand, on the basis that this compound is a nearly neutral antagonist and might bind to coupled and uncoupled forms of the receptor equivalently (the ready availability of p-MPPI over p-MPPF as a radioiodinated ligand was conducive to this and subsequent assays). Scatchard analysis of [ 125 I]p-MPPI binding to Sf9 cell membranes expressing the 5-HT 1A receptor alone and together with G z  the presence of G z was best fit by a two-site model, wherein about 30% of the receptor exhibited high affinity for 5-HT (K i ϭ 6 Ϯ 3 nM), and the remainder exhibited low affinity (K i ϭ 590 Ϯ 60 nM) (Fig. 4A). Despite the high level of expression of receptor and G protein, therefore, only one-third of the receptor was found to preexist in a coupled state or to enter into a coupled state consequent to binding 5-HT. In the absence of G z , 5-HT displaced [ 125 I]p-MPPI binding in a monophasic manner with uniformly low affinity (K i ϭ 640 Ϯ 100 nM). These data represent the first measurement of the "K H /K L " ratio, a commonly employed index of efficacy, for 5-HT using a single G protein (the ratio is about 100). Displacement of [ 125 I]p-MPPI by p-MPPF, as anticipated, was about the same regardless of G protein (K i ϭ 3.5 Ϯ 0.4 and 3.1 Ϯ 0.3 nM in the absence and presence of G protein, respectively) (Fig.  4B). Similar results were obtained using p-MPPI to displace [ 125 I]p-MPPI (K i ϭ 2.4 Ϯ 0.3 and 2.4 Ϯ 0.1 nM, respectively) (data not shown). To our initial surprise, displacement curves with spiperone were essentially unaffected by G protein (K i ϭ 60 Ϯ 6 nM and 83 Ϯ 9 nM, respectively) (Fig. 4C).
Assuming that G z couples to only 30% of the receptor in the absence of agonist (it may well be less), a difference in affinity of spiperone for uncoupled and coupled forms of the receptor greater than ϳ20 -30-fold would be required for a biphasic displacement to be clearly resolved. An assay to selectively label the G protein-coupled form of receptor was therefore designed. For labeling, we used the agonist [ 125 I]8-OH-PIPAT.
[ 125 I]8-OH-PIPAT binds the 5-HT 1A receptor only in the presence of G protein (47). It binds the uncoupled form of receptor poorly if at all. The K d for [ 125 I]8-OH-PIPAT using G z as the G protein is 0.2 Ϯ 0.02 nM (see below). [ 125 I]p-MPPI, used to label the receptor expressed alone without G z , provided the reference for the uncoupled form of receptor. The validity of this method is established in Fig. 5A, which shows the displacement analysis for 5-HT. 5-HT clearly displaced [ 125 I]8-OH-PIPAT from membranes expressing receptor and G z (i.e. coupled receptor) more easily than it did [ 125 I]p-MPPI from membranes expressing receptor alone (uncoupled), and it did so in a monophasic manner. The K i for 5-HT calculated from displacement of [ 125 I]8-OH-PIPAT was 7 Ϯ 1 nM, in good agreement with that established in the previous experiment for the high affinity form of receptor evident from [ 125 I]p-MPPI displacement in the presence of G z (see above). The K i calculated for p-MPPF from displacement of [ 125 I]8-OH-PIPAT in Fig. 5B (1.9 Ϯ 0.1), moreover, was in close agreement with that calculated previously from displacement of [ 125 I]p-MPPI regardless of G protein. A similar result was obtained with p-MPPI (K i ϭ 1.0 Ϯ 0.2 for displacement of [ 125 I]8-OH-PIPAT; data not shown). In contrast to the almost superimposable displacement curves established by p-MPPF and p-MPPI for coupled and uncoupled forms of receptor, displacement analysis with spiperone as the competing ligand generated inhibition curves with markedly different characteristics (Fig. 5C). Displacement of [ 125 I]8-OH-PIPAT binding by spiperone did not fit well to a one-site competition curve with a Hill coefficient of 1. The fit could be statistically improved, however, by either of two models: a two-site model wherein about 35% of receptors represent a high affinity state (K i ϭ 10 Ϯ 1 nM) and the remainder represent a low-affinity state (K i ϭ 420 Ϯ 50 nM), or a one-site model with a shallow slope (IC 50 ϭ 576 Ϯ 54; Hill coefficient ϭ 0.55) indicative of "noncompetitive" competition. Statistically, either model would be appropriate; however, we favor the latter because displacement of [ 125 I]8-OH-PIPAT binding by 5-HT indicates that only one form of the receptor is labeled by [ 125 I]8-OH-PIPAT. Noncompetitive inhibition is consistent with the predictions of Wreggett and De Léan (53), in which the radioligand and competing ligand would be binding to distinct but interconvertible forms of the receptor.
Further experiments were carried out to determine the validity of the one-site model involving noncompetitive inhibition. The binding of [ 125 I]8-OH-PIPAT was analyzed by Scatchard transformation of saturation binding data (54) in the absence and presence of spiperone. Spiperone (300 nM) was found to cause a parallel shift in the Scatchard plot for [ 125 I]8-OH-PIPAT; i.e. the K d was unchanged, but the B max was reduced by approximately one-half (Fig. 6A). This is consistent with noncompetitive antagonism. p-MPPI (10 nM) showed a characteristic competitive inhibition in the same assay (data not shown). Spiperone (100 nM) caused a change in the slope of the plot for [ 125 I]p-MPPI in a manner implying competitive antagonism (Fig. 6B). The K d was decreased, but the B max was unchanged. Thus, the hypothesis of noncompetitive antagonism pertaining to the inhibition by spiperone of [ 125 I]8-OH-PIPAT binding is consistent with the parallel shift in the Scatchard plot for [ 125 I]8-OH-PIPAT and the single-site curve fit with a Hill coefficient of 0.55. DISCUSSION We have demonstrated using a novel reconstitution technique that the human 5-HT 1A receptor exhibits constitutive activity and that this activity forms the basis by which neutral antagonists and inverse agonists can be defined. The use of Sf9 cells as a medium for reconstitution is attractive because Sf9 cells provide a virtually null background upon which the expression of individual receptors and G proteins can be superimposed. A given G protein is not subject to the actions of multiple and often ill-defined receptors, as it is in mammalian cells, and the equilibrium between R and R* is not complicated by the heterogeneity in G proteins also existing in mammalian cells. The high level of expression of receptors and G proteins that can be achieved in Sf9 cells, moreover, lends itself to the assay of receptor activity by means of G protein activation directly. This kind of assay has two advantages. First, as constructed here by [ 35 S]GTP␥S binding in tandem with an immunoprecipitation technique, the assay is quantitative. Second, the assay eliminates the traditional reliance on effectors and/or second messengers as readouts for receptor activity in calculations of efficacy. Effectors and second messengers are subject to multiple inputs and forms of cross-regulation that compromise their utility in this regard.
In our studies here, we used the G protein G z to monitor the activity of the 5-HT 1A receptor. The choice of G z was based primarily on technical factors. G z alone of the G i family does not bind [ 35 S]GTP␥S under the conditions of the assay unless it is stimulated to do so by a receptor, whether through constitutive or agonist-promoted receptor activity. The resistance of G z to binding [ 35 S]GTP␥S when expressed alone results in a very low background that serves to better highlight activation by the receptor. Sf9 cells do not contain G z (47), and the antibody used to achieve immunoprecipitation of the subunit under the nondenaturing conditions required of the assay is particularly well characterized (49). We recognize that some differences may exist between G z and other members of the G i family. G z nevertheless communicates well with the 5-HT 1A receptor and with many other receptors that utilize the G i family (44,49,55,56), and it is therefore suitable for the purposes described here.
Ours is the first demonstration of constitutive activity as exerted on any individual member of the G i family. The activity exhibited by the 5-HT 1A receptor is commensurate with spontaneous isomerization of a subpopulation of receptor from an inactive (R) to an active (R*) state. In our studies, we have effectively discounted the possibility that the constitutive ac- Binding of [ 125 I]p-MPPI is expressed as a percentage of that achieved in the absence of competing compounds (typically corresponding to about 8000 cpm). The data are the means Ϯ S.E. of three or four experiments performed in triplicate. All curves except that in A referring to both receptor and G z (f) were generated by fitting the data to a one-site model with a Hill coefficient of 1.
tivity is due to endogenous 5-HT carried over from cell culture. Carry-over can be a very real problem, and must be considered in any study dealing with 5-HT receptors, as our preliminary experiments had revealed carry-over of 5-HT for cells exposed to serum at times greater than 16 h following baculovirus infection; i.e. for cells that express the receptor, even when cultured with charcoal-treated serum (data not shown). We have addressed this problem by removal of the cells to serum-free medium prior to expression of the receptor and by extensive washing of the cells and membranes prior to analysis. Using these precautions, we detected no 5-HT in the preparations of membrane used for assay. The possible contribution of carry-over 5-HT to constitutive activity is also ruled out by the inability of p-MPPF, which antagonizes activation by agonist, to suppress this activity.
The constitutive activity of the 5-HT 1A receptor provided the  Fig. 4. All curves were generated by fitting the data to one-site models with a Hill coefficient of 1 except for that in panel C describing the displacement of [ 125 I]8-OH-PIPAT by spiperone, which was fit to a one-site model with a Hill coefficient of 0.55.
basis for testing what had been generically termed antagonists for inverse agonism. Spiperone was considered at the outset to be a good candidate because it had been shown to exhibit inverse efficacy at the 5-HT 2C (G q -coupled) receptor (32) and because GTP was reported to increase the B max for [ 3 H]spiperone in Chinese hamster ovary cells transfected with the 5-HT 1A receptor (by 70%) (24). Indeed, in our studies here with the 5-HT 1A receptor and G z reconstituted in Sf9 cells, spiperone produced an almost complete suppression of constitutive receptor activity. p-MPPI and p-MPPF, in contrast, had little (p-MPPI) or no (p-MPPF) effect. All three compounds blocked the activation of G z by 5-HT, validating the assumption of receptor binding. Thus, spiperone is an almost fully efficacious inverse agonist, p-MPPI is at best a partial inverse agonist, and p-MPPF is unquestionably a neutral antagonist. As expected, the suppression of activity achieved by spiperone was blocked by p-MPPI and p-MPPF. The concept of gradations in inverse efficacy clearly applies to the relationship of spiperone to p-MPPI. That p-MPPI exhibits some degree of inverse activity is consistent with the observation that guanosine 5Ј-(␤,␥-imino) triphosphate increases the B max for [ 125 I]p-MPPI by 18% in hippocampal membranes (15). Of interest, p-MPPI and p-MPPF are structurally almost identical, containing iodine and fluorine atoms, respectively, on the benzamido moiety.
Current models suggest that the mode of action of inverse agonists, such as spiperone at the 5-HT 1A receptor, can be explained by a higher affinity of these compounds for the inactive (R) isomer of the receptor. Binding of a compound preferentially to R would lead to a reduction in constitutive activity by shifting the equilibrium toward R from R* (present as R* and R*G). Two groups in fact report that spiperone displays a higher affinity for the 5-HT 1A receptor in competition assays with radiolabeled antagonists, where the reported K i for spiperone is about 15 nM, than in competition assays with radiolabeled agonists, where the K i is about 500 nM (15,24). The difference in K i values may well reflect the different populations of receptor being analyzed, although the identity of these populations can only be implied due to the presence of a heterogeneous population of G proteins in mammalian cells. Antagonists would label R, R*, and R*G, where R would be the predominant form of receptor. Agonists would label R* and R*G relatively selectively. Whereas the two groups above specifically compared the ability of spiperone to compete against agonists versus antagonists, K i values reported elsewhere for spiperone at the 5-HT 1A receptor varied considerably (9, 15, 19 -23, 25). Given the variation in reported estimates of affinity, we determined the affinity of spiperone (and other ligands) for G proteincoupled and uncoupled forms of the 5-HT 1A receptor explicitly. Our construction of these two forms of receptor was unique, again highlighting the utility of Sf9 cells. The G protein-coupled form of the receptor was fashioned by coexpression of the receptor with G z and was labeled selectively with [ 125 I]8-OH-PIPAT. The uncoupled form of receptor was fashioned by expression of the receptor without G z and was labeled with [ 125 I]p-MPPI. The validity of our method was confirmed by displacement analysis with 5-HT. Our results were unambiguous: 5-HT recognized the coupled form of receptor with high affinity (K i ϭ 7 nM) and the uncoupled form of receptor with low affinity (K i ϭ 640 nM). As expected, essentially the opposite was observed for spiperone. However, the comparisons of displacement in the case of spiperone were more complex. Spiperone acted as a competitive antagonist (K i ϭ 60 nM) when tested with the uncoupled form of receptor but not necessarily so when tested with the coupled receptor. For competitive antagonism to hold in the latter instance, two binding sites (in which the majority of receptor nevertheless displays low affinity for spiperone (K i ϭ 420 nM)) must be postulated. Although this is not implausible, the molecular basis for two-site binding is unclear. The more likely alternative is a special form of noncompetitive antagonism at one site. Noncompetitive behavior may be accounted for by the predictions of Wreggett and De Léan (53) from their simulations of the ternary complex model. When a compound has a higher affinity for the uncoupled receptor and is displacing a radioligand having a higher affinity for coupled receptor, the displacing compound would exhibit a form of noncompetitive inhibition because the compound and radioligand would effectively be binding to distinct though interconvertible forms of the receptor. If the radioligand had equal affinity for both forms of receptor, competitive inhibition would be achieved. Consistent with noncompetitive behavior is the spiperone-effected depression in B max , with no change in K d , for [ 125 I]8-OH-PIPAT. To our knowledge, this is the first experimental evidence of both competitive and noncompetitive antagonism with the same ligand, spiperone, as predicted by Wreggett and De Léan (53). Regardless, our observations are consistent with the conclusion that spiperone has a higher affinity for the uncoupled form of receptor than for the coupled form. p-MPPF, as predicted for a neutral antagonist, did not differentiate the two forms of receptor, further validating the assay. Spiperone is often viewed to be a competitive antagonist even when assayed against an agonist in functional or displacement assays. Detection of noncompetitive antagonism in functional assays, however, may well be hampered by receptor reserve (57). Some binding assays, moreover, are not inconsistent with our results. For example, displacement curves in some reports are shallower than expected for a strictly competitive antagonist, and in one instance, a displacement curve was interpreted with a two-site fit (58). Differences may otherwise lie in the coupling of the 5-HT 1A receptor to multiple G proteins in mammalian systems, where the heterogeneity in R/R*/R*G equilibria may obscure noncompetitive behavior, or in the types of agonists used for analysis, which may distinguish R and R* less well. Here, we examined only one kind of receptor/G protein interaction and the displacement of an agonist that is highly selective for the coupled receptor.
Constitutive activity as monitored by [ 35 S]GTP␥S binding has been studied in several other instances. Among the most relevant to the work here is that of Newman-Tancredi et al. (59), who measured [ 35 S]GTP␥S binding to membranes of 5-HT 1A receptor-transfected Chinese hamster ovary cells. Consistent with our results, this group identified spiperone as an inverse agonist by virtue of its ability to suppress binding of [ 35 S]GTP␥S in the absence of agonist. The binding of [ 35 S]GTP␥S to a G protein was not formally defined, however, and the degree of suppression was less than what we observed (30% versus 95%). We suspect that the difference in the degree of suppression is related to [ 35 S]GTP␥S binding to Chinese hamster ovary cell membranes unrelated to the presence of receptor (and therefore not suppressible by spiperone) or to a form(s) of receptor more refractory to a transition from R* to R. The latter could result from a greater affinity of R* for one or more of the G proteins present in Chinese hamster ovary cells and/or a greater sensitivity of some of these G proteins to activation. Importantly, our studies here deal only with a single receptor and G protein and therefore obviate problems of interpretation resulting from the heterogeneity of receptors and G proteins in mammalian cell expression systems. Moreover, as a result of the null background and isolation of the G protein through immunoprecipitation, we were able to define spiperone as a full inverse agonist, p-MPPI as a partial inverse agonist, and p-MPPF unequivocally as a neutral antagonist. A second report of relevance is that of Hartman and Northup (48), who measured [ 35 S]GTP␥S binding to membranes of Sf9 cells expressing the 5-HT 2C receptor. Constitutive activity was evident for membranes reconstituted with purified G q following extraction of the membranes with urea. Suppression of the constitutive activity was achieved with ketanserin and mianserin. The 5-HT 2C receptor, however, is quite different from the 5-HT 1A receptor in many respects, and the two assays are quite different from each other. Our technique does not involve the in vitro reconstitution of a G protein with membranes but rather the co-expression of receptor and G protein in the intact cell. Targeting and coupling are achieved in a biosynthetic context, and the need for purification of the G protein does not exist. Indeed, our protocol lends itself to assaying various G proteins simply by using different recombinant baculoviruses.
We have confirmed that the 5-HT 1A receptor exhibits constitutive activity in Sf9 cells by direct measurement of G protein activation, and using this activity, we have demonstrated that spiperone is an almost fully efficacious inverse agonist and that p-MPPF and p-MPPI are not. We have provided evidence using both G protein activation and radioligand displacement assays that Sf9 cells can provide a useful tool for establishing ordered inverse efficacy. Assays of G protein activation eliminated the need for effector readouts. We have also developed a unique paradigm for examining the behavior of homogeneous populations of coupled and uncoupled receptors using a defined pairing of receptor and G protein. The techniques employed here can be applied easily to other receptors, perhaps mutated to induce constitutive activity or to assume additional conformations, and can be extended to other G proteins.