Main

Classical theories of G-protein-coupled receptor (GPCR) activation assume that functional responses are induced by the action of agonists at otherwise quiescent receptors (Milligan et al. 1995). Accordingly, the induction of an active state of the receptor, capable of stimulating signal transduction pathways, is considered to depend upon the presence of an agonist. Antagonists exert no effects themselves on the state of receptor activation, and merely interfere with the actions of agonists. However, over the past two decades, several observations have accumulated to suggest that certain “antagonists” may act in a fashion opposite to agonists. One early observation was that lesions of dopaminergic pathways and dopaminergic antagonists can induce additive effects on striatal levels of acetylcholine (Fibiger and Grewaal 1974). Nevertheless, in this study, it could not be excluded that antagonists might be blocking the actions of residual pools of dopamine. Creese et al. (1975) reported that agonists had higher affinity for [3H]-dopamine-labelled dopamine receptors whereas antagonists had greater affinity for [3H]-haloperidol-labelled sites. Further, at muscarinic receptors labelled with [3H]-quinuclidyl benzylate, Burgisser et al. (1982) proposed that antagonists have higher affinity for receptors uncoupled from G-proteins. Costa and Herz (1989) proposed that certain antagonists act as “inverse agonists” at δ-opioid receptors endogenously expressed by NG 108-15 cells . Indeed, whereas the agonist, DADLE, increased GTPase activity at these δ-opioid receptors, the “antagonist,” ICI174864, inhibited basal activity. Both actions were blocked by a neutral antagonist. Although agonists and antagonists are classically considered to down- and up-regulate receptors, respectively, the above observations led to the suggestion (Morris and Millan 1991) that “inverse agonist” rather than “antagonist” properties may favor receptor up-regulation in vitro and/or in vivo (Milligan and Bond 1997; Smit et al. 1996).

More recent studies employing homogeneous populations of cloned (wild-type) receptors in heterologous expression systems free of endogenous agonist have underpinned the concept of “negative intrinsic efficacy” and “inverse agonism” (see Costa et al. 1992; Kenakin 1996; Milligan et al. 1995; Schütz and Freissmuth 1992). Thus, many types of wild-type GPCR have been observed to display an agonist-independent constitutive activation of intracellular transduction mechanisms inhibited by inverse agonists. In contrast, genuinely neutral antagonists block the actions of both agonists and inverse agonists, without themselves altering activity.

EXPERIMENTAL STRATEGY: CHARACTERIZATION OF POTENTIAL INVERSE AGONIST ACTIONS AT 5-HT1B RECEPTORS

Several types of recombinant, heterologously-expressed 5-HT receptor (5-HT1A, 5-HT1B, 5-HT1D, 5-HT2A, 5-HT2C, and 5-HT7) display constitutive activity and, correspondingly, reveal inverse agonist properties of certain antagonists (Barker et al. 1994; Egan et al. 1998; Newman-Tancredi et al. 1997b; Thomas et al. 1995, 1998). The 5-HT1A, 5-HT1B, and 5-HT1D receptors are of particular interest inasmuch as they are localized both postsynaptically to serotonergic neurones as well as presynaptically, as inhibitory autoreceptors on their dendrites (5-HT1A and a minor population of 5-HT1B/1D sites) and terminals (5-HT1B sites) (Piñeyro et al. 1995; Starkey and Skingle 1994). It might, thus, be hypothesized that inverse agonist actions at 5-HT1A, 5-HT1B, and/or 5-HT1D autoreceptors may facilitate serotonergic transmission (Gaster et al. 1998; Moret and Briley 1993; Roberts et al. 1997). In addition, postsynaptic structures enriched in these receptor types, such as the substantia nigra and striatum (5-HT1B receptors), are also potential targets for the detection of inverse agonist actions (Boess and Martin 1994).

However, while constitutive activity may be readibly detectable in recombinant systems in vitro, it remains to be determined whether CNS populations of 5-HT receptors (and other classes of GPCR) display constitutive activity in situ under normal and/or pathological conditions. It is also unclear whether drugs express physiologically and therapeutically relevant inverse agonist properties at CNS populations of receptors in vivo. The present commentary summarizes the findings from several complementary approaches which we have utilized to examine the potential existence of inverse agonist actions at 5-HT1B receptors in vitro and in vivo. First, we examine binding of the GTP analogue, [35S]-GTPγS, to G-proteins activated by recombinant h5-HT1B receptors expressed in CHO cells. Whereas agonists increase [35S]-GTPγS binding, inverse agonists inhibit the basal [35S]-GTPγS binding observed in the absence of ligand(s) (Newman-Tancredi et al. 1997b; Thomas et al. 1995). Second, we discuss the binding of [35S]-GTPγS to native rat brain h5-HT1B receptors visualized by quantitative autoradiography (Sim et al. 1995). Third, we consider the modulation of extracellular 5-HT levels as determined by dialysis of the FCX of freely moving rats (Gobert et al. 1998). Fourth, we evaluate the induction of hypothermia in the guinea pig (Hagan et al. 1997; Skingle et al. 1996).

Cloned h5-HT1B Receptors: G-Protein Activation Quantified by [35S]-GTPγS Binding

The influence of ligands upon [35S]-GTPγS binding at h5-HT1B receptors was evaluated using essentially the same procedure as described for h5-HT1A receptors (Newman-Tancredi et al. 1997a). 5-HT and GR46611 behaved as agonists while GR127,935, originally described as an antagonist (Skingle et al. 1996), displayed partial agonist actions (Price et al. 1996; Watson et al. 1996). The non-selective 5-HT1 receptor antagonist, methiothepin and the selective antagonist, SB224,289 (Gaster et al. 1998), inhibited basal activation, suggestive of inverse agonist actions at 5-HT1B receptors. Importantly, the novel and selective 5-HT1B ligand, S18127 (Millan et al. 1998), behaved as a neutral antagonist inasmuch as it was inactive alone and it blocked the actions of both GR46611 and SB224,289 (Figure 1).

Figure 1
figure 1

Influence of 5-HT1B ligands upon [35S]GTPγS binding to CHO-h5-HT1B receptors.

Full size image

Native Rat 5-HT1B Receptors: Functional Autoradiography of G-Protein Activation by [35S]-GTPγS Binding to Cerebral Tissue Sections

In the substantia nigra, employing a procedure based on that detailed by Sim et al. (1995), a pronounced stimulation of [35S]GTPγS was seen with the agonist at rat 5-HT1B receptors, CP93129 (10 μM, 70.5 ± 7.1%). The action of CP93129 was inhibited by S18127 and SB224,289 (10 μM, 12 ± 13% and 11.4 ± 5.9%, respectively). However, neither of these ligands modifed basal [35S]-GTPγS binding alone. S18127 and SB224,289 alone: 4.4 ± 6.7% and 4.3 ± 4.3%, respectively. In a further experiment, shown in Figure 2, SB224,289 (10 μM) also abolished stimulation of [35S]-GTPγS binding by 5-HT (10 μM) in substantia nigra. In contrast to the present study, in which the sections were extensively washed to ensure removal of endogenous 5-HT, Stanton et al. (1997) reported that, the non-selective ligand, methiothepin lowered basal [35S]-GTPγS binding to rat brain sections. However, the present data concur with results on guinea pig brain sections where, similarly, no action of SB224,289 was detected on basal [35S]-GTPγS binding to hippocampus (Dupuis et al. 1998).

Figure 2
figure 2

Influence of 5-HT1B ligands upon [35S]GTPγS binding to substantia nigra in situ. Upper left panel: Binding of [35S]GTPγS in the presence of 5-HT (10 μM) which activates 5-HT1B receptor-coupled G-proteins. Upper right panel: Basal binding observed in the absence of ligand. Lower left panel: Binding of [35S]GTPγS in the presence of SB224,289 (10 μM). No difference from basal values was observed. Lower right panel: Binding of [35S]GTPγS in the presence of 5-HT (10 μM) and SB224,289 (10 μM). SB224,289 blocked the activation induced by 5-HT. Images are pseudo-color representations of [35S]GTPγS labelling observed under the following conditions. Sections were incubated for 60 minutes at 37 °C in a buffer containing HEPES 50 mm (pH 7.5); NaCl 150 mM, EGTA 0.2 mM, DTT 0.2 mM, GDP 2.5 mM, MgCl2 10 mM, and [35S]GTPγS 0.05 nM. Sections were then rinsed and exposed to Hyperfilm. Calibration was according to commercially available C 14 standards.

Full size image

Modulation of Serotonergic Transmission by 5-HT1B Autoreceptors

As reported in Gobert et al. (1998), GR46611 markedly suppresses dialysate levels of 5-HT in the FCX. Notwithstanding their contrasting actions at cloned h5-HT1B receptors (vide supra), GR127,935, S18127 and SB224,289 abolished the actions of GR46611 without themselves modifying 5-HT release (Figure 3). Because blockade of 5-HT1B autoreceptors may result in compensatory activation of 5-HT1A autoreceptors by 5-HT, the actions of SB224,289 and GR127,935 were re-determined in the presence of the 5-HT1A antagonist, WAY100,635, but they still failed to enhance 5-HT levels (Figure 3). GR127,935, S18127, and SB224,289 likewise all facilitated the increase in FCX levels of 5-HT elicited by fluoxetine. These actions of WAY100,635 and SB224,289 were exerted synergistically.

Figure 3
figure 3

Influence of 5-HT1B ligands upon extracellular, dialysate 5-HT levels in frontal cortex of freely moving rats. n = 5–8 per group. Data are means ± SEM. Asterisks (p < .05) indicate significance in ANOVA of following differences; (A) GR127,935/GR46611 vs. vehicle/GR46611; (B) S18127/GR46611 vs. vehicle/GR46611; (C) SB224,289/GR46611 vs. vehicle/GR46611; (D) no statistical differences; (E) GR127,935/ FLX, S18127/FLX and SB224,289/FLX vs. vehicle/FLX; and (F) WAY100,635/ SB224,289/FLX vs. vehicle/vehicle/FLX. 5-HT levels were determined by HPLC followed by colormetric detection. Drugs were given SC. In F, vehicle/vehicle, WAY100,635/vehicle, vehicle/SB224,289 and WAY100,635/ SB224,289 combinations were injected at arrow 1, followed by fluoxetine at arrow 2.

Full size image

Modulation of Core Temperature

It has been shown that activation of 5-HT1B receptors elicits hypothermia in guinea pigs (Hagan et al. 1997). Accordingly, GR46611 decreased core temperature in guinea pigs, and this action was inhibited by GR127,935, S18127 and SB224,289 (Figure 4). However, administered in the absence of GR 46611, none of these drugs modified core temperature (Hagan et al. 1997) (Figure 4).

Figure 4
figure 4

Influence of 5-HT1B ligands upon core temperature in guinea pigs. Data are means ± SEM. n = 5–7 per group. *p < .05 to vehicle/GR46611 values in Newman-Keuls test. Note lack of significance of vehicle/SB224,289, vehicle/S18127, and vehicle/GR127,935 versus vehicle/vehicle values. Core temperature was determined via a digital thermistoprobe placed into the ear for 30 seconds. Vehicle or antagonist were given (SC) 60 minutes, and vehicle or GR46,611 (5.0 mg/kg, IP), 30 minutes prior to testing.

Full size image

GENERAL DISCUSSION: ABSENCE OF INVERSE AGONIST ACTIONS IN VIVO

In agreement with previous reports (Gaster et al. 1998; Pauwels et al. 1998; Thomas et al. 1995), the present study detected negative intrinsic activity actions of both methiothepin and the selective inverse agonist, SB224,289, at h5-HT1B receptors in vitro. Further, the discovery of a selective neutral antagonist, S18127 (Millan et al. 1998), permits an examination of this issue both in vitro and in vivo. However, although S18127 abolished the actions of SB224,289 in CHO cells, no evidence was obtained for inverse agonist actions of SB224,289 at native 5-HT1B receptors in vivo. Indeed, although SB224,289 slightly increased resting 5-HT levels in the guinea pig dentate gyrus in vitro (Gaster et al. 1998), this effect could not unambiguously be attributed to inverse agonism rather than interruption of the actions of spontaneously-released 5-HT. Further, no evidence that SB224,289 increases dialysate levels of 5-HT was obtained herein. There are several possible explanations for the absence of evidence for inverse agonist actions in vivo.

First, SB224,289 may be metabolized in vivo to structurally related neutral antagonists. However, this is unlikely by the SC route and does not account for the lack of suppression of [35S]-GTPγS binding in autoradiographic studies. Second, methiothepin and SB224,289 may act as inverse agonists at recombinant human 5-HT1B/1D receptors but not at their native rat and guinea pig counterparts. Indeed, marked species differences have been documented for the pharmacological profiles of 5-HT1B receptors, reflecting changes as minimal as a single amino acid substitution (Boess and Martin 1994; Price et al. 1996). However, in a direct comparison of recombinant guinea pig and human 5-HT1B receptors, Pauwels et al. (1998) showed that SB224,289 and methiothepin acted as inverse agonists in each case. Third, the intracellular and extracellular enviroment plays an important role in determining ligand-receptor–G-protein interrelationships (Kenakin 1996). For example, the precise complement of G-protein subtypes, protein kinases and other substrates differs between transfected cell lines and native neurones. Thus, conditions used to demonstrate inverse agonist activity in vitro by [35S]-GTPγS binding may not be present in the CNS. Fourth, the efficacy of agonists and inverse agonists is a function of receptor and G-protein expression levels and we have shown that, in CHO-h5-HT1A cell membranes, inverse agonist activity can be enhanced by an augmentation of receptor: G-protein ratios (Newman-Tancredi et al. 1997a). Thus, the receptor and/or G-protein level may be too low at native 5-HT1B receptors to permit inverse agonist actions to be detected. However, modulation of 5-HT release reflects actions at autoreceptors which, in analogy to 5-HT1A sites, likely show a high receptor reserve/density (Boess and Martin 1994). Fifth, the techniques employed may not be sufficiently sensitive to detect modest inverse agonist actions in vivo. This may be the case for the [35S]-GTPγS autoradiography procedure, in which actions of SB224,289 upon 5-HT1B- coupled G-proteins might be diluted by other pools of G-proteins. Sixth, compensatory mechanisms in vivo may mask inverse agonist actions. For example, core temperature is homeostatically controlled by multiple factors. Further, blockade of dendritic 5-HT1D (or 5-HT1B) sites may increase concentrations of 5-HT at colocalized 5-HT1A autoreceptors, thereby masking this action (Starkey and Skingle 1994). However, combined administration of WAY100,635 with GR127,935 or SB224,289 for simultaneous blockade of 5-HT1A and 5-HT1B/1D autoreceptors also failed to enhance dialysate 5-HT levels. Seventh, for inverse agonist actions to be detected, the receptor must be present in a constitutively active state and, ideally; endogenous agonist(s) should not be present. However, in vivo, 5-HT is spontaneously released both at the dendritic and terminal level (Piñeyro et al. 1995). Indeed, basal dialysate levels of 5-HT are reliably measurable, suggesting that 5-HT is available to pre- and postsynaptic receptors (Gobert et al. 1998). Thus, spontaneous release of 5-HT may impede the detection of inverse agonist actions. Indeed, as alluded to above, it is difficult to distinguish a neutral antagonist blockade of the actions of spontaneously released 5-HT from an inverse agonist action opposite to that of 5-HT. For this purpose, studies showing that neutral antagonists block inverse agonist actions are essential (Kenakin 1996; Newman-Tancredi et al. 1997a).

In the light of these comments, an evaluation of the potential inverse agonist actions of SB224,289 upon 5-HT1B receptor would be of interest to perform in rats deprived of 5-HT. Indeed, such 5-HT-depleted conditions might resemble depressive states in which serotonergic transmission has been proposed to be deficient (Maes and Meltzer 1995). Finally, support for constitutively active receptors would be provided by the demonstration that inverse agonists elicit actions which are not evoked by antisense probes, and which are not mimicked by effects observed in gene knock-out mice.

CONCLUSIONS

The above comments illustrate the difficulties faced in demonstrating inverse agonist actions at 5-HT1A and 5-HT1B/1D receptors in vivo and little evidence has been forthcoming from other CNS-localized receptor types that inverse agonist effects of antagonists underlie their functional profiles of activity. For example, although the extrapyramidal actions of neuroleptics might reflect negative intrinsic activity at D2 receptors, both cloza–pine (non cataleptogenic) and haloperidol (cataleptogenic) display inverse agonist properties at cloned hD2 sites (Nilsson et al. 1996; Hall and Strange 1997). Nevertheless, the potential relevance of constitutive activity at GPRCs and of inverse agonist actions in vivo cannot be dismissed, as illustrated by a transgenic system of β2-adrenoceptor overexpression (Bond et al. 1995). As emphasized above, the availability of a highly selective neutral antagonist and inverse agonist is imperative for the rigorous demonstration of inverse agonist actions in vitro and in vivo. The present article describes the availability of such ligands for 5-HT1B sites and further studies will be required for a more precise evaluation of the concept of inverse agonism in vivo at diverse GPCRs. Irrespective of such considerations, the present study strongly suggests that extrapolation of data from in vitro expression models of cloned receptors should be made cautiously to intact physiological systems.