Agonist–receptor interaction. The structural features that define the 5-HT4 ligands are an aromatic moiety, a coplanar hydrogen acceptor group such as the carbonyl function, and a basic nitrogen atom. Despite these common features, it has been demonstrated that triptamines (5-HT, tegaserod) and benzamides (metoclopramide, renzapride, cisapride) have different pharmacophores.52,75 The mutation of the conserved Asp100 generates a 5-HT4 receptor activated solely by synthetic ligands (RASSL). The 5-HT4 RASSL is insensitive to 5-HT and tegaserod, which can be explained by the lack of binding of these compounds to the mutant receptor.76 The 5-HT4 receptor antagonist GR113808 is still able to bind to the mutant receptor, which can also be stimulated by numerous synthetic agonists from different chemical classes, including the benzamides metoclopramide, renzapride and cisapride. These compounds all have a bulky substitute of the basic nitrogen, most of them with the basic nitrogen included in a structured ring (Fig. 1), necessary to interact in a hydrophobic pocket of the receptor.77 In contrast, the highly conserved Asp100 is known to make an ionic bond with the protonated amine of 5-HT derivatives. Taken together, this suggests that tegaserod on the one hand and the benzamides and prucalopride on the other hand, use different binding pockets on the 5-HT4 receptor.
It is becoming increasingly recognized that seven-transmembrane receptors are allosteric proteins that can adopt numerous conformations which can activate different transduction pathways by agonist-directed stimulus trafficking (ligand-directed signalling).78,79 Therefore, a linear view of efficacy whereby a single receptor activation state (two-state model, where the receptor is either on or off) triggers all possible receptor interactions with a cell is no longer supported. The existence of different receptor conformations has been shown for constitutively active mutants of the 5-HT4 receptor80 and for the full, partial and inverse agonist-liganded 5-HT4 receptor.81,82 In addition, the efficacy of a partial 5-HT4 receptor agonist is likely due to the stabilization of an agonist-specific conformational state.82 The efficacy of a certain agonist is then defined as the ability to stabilize these specific conformations. This also implies that some 5-HT4 receptor agonists might only partially activate the library of potential signalling systems in a cell. Indeed, in HEK293 cells, 5-HT4(a) receptor activation triggers a strong [Ca2+]i increase that is dependent on the compound structure because benzamide-related compounds trigger higher responses than indoleamines.83
The existence of many C-terminal splice variants of the 5-HT4 receptor adds even further complexity. Several splice variant-specific alternative signal transduction cascades have been described in recombinant expression systems.83–87 The expression of different splice variants in different tissues (or species) might therefore contribute to differences in overall coupling in the different tissues. These variant and therefore tissue- or species-dependent differences in coupling are expected to influence the response of all agonists unless agonist-directed trafficking of the stimulus response occurs. Different agonists may then be able to activate different alternative signalling cascades through the different splice variants. In support of this, the calcium response observed with benzamidic compounds seems to be splice variant-specific, being only observed in 5-HT4(a) and not in 5-HT4(b).83
Agonist-induced effects. Quite often, prucalopride is referred to as a full agonist while tegaserod is regarded as a partial agonist [see for instance, Refs (88,89)]. This clearly is an oversimplification and one really should avoid extrapolating the degree of agonism of a substance in a given tissue. For a start, there is the inherent difficulty in defining what a full agonist is.90 For reasons of simplicity we refer to full agonism when the agonists elicit the same maximal response as 5-HT, the endogenous ligand of the 5-HT4 receptor. When using this criterium in the literature, it is clear that both prucalopride and tegaserod have been reported as either full or partial agonist [see, e.g. Refs (21,39,69)].
These differences may originate from different grounds. First, a given agonist can behave as a full agonist in certain tissues, while it can express partial agonism in others. Indeed, the expression of partial vs full agonism by a drug largely depends on the system in which it is evaluated as the receptor density or coupling efficiency can differ. Some agonists will thus behave as partial agonists in low efficacy systems, while they behave as full agonists in high efficacy systems (e.g. with high receptor reserve). Therefore, based on differences in receptor density or coupling efficiency (which might be influenced by the nature of the splice variants), the degree of agonism of a given agonist in a given tissue can vary between species. Even within a single species, these properties can lead to a differential degree of agonism for a given substance between tissues.21 This may explain the many differences in benzamide-induced 5-HT4 receptor-mediated responses between different tissues or species.9
One example of a low efficacy system is the heart, which is reflected in the relatively low stimulation potency of 5-HT4 receptor agonists in this tissue. Because of its nature of low efficacy, this tissue provides a useful method for evaluating differences in the pharmacological profile of 5-HT4 receptor ligands. Substituted benzamides like renzapride, cisapride and metoclopramide or dihydrobenzofurancarboxamides like prucalopride, M0003 and M0004 behave as partial agonists at porcine sinoatrial and/or human right atrial 5-HT4 receptors91–93 while most tryptamines are full agonists compared with 5-HT.91,92 Prucalopride also acted as a partial agonist on the L-type Ca2+ current through 5-HT4 receptors in human atrial cells.94 Therefore, different organ-specific behaviour can be expected between these classes of compounds. Recently a porcine in vitro model has been developed to describe and explain organ-specific behaviour of 5-HT4 receptor agonists.21 In this model, the effects of 5-HT4 receptor agonists in a gastric assay (5-HT4 receptors on cholinergic neurones) were compared with their effects on the right and left atrium (low 5-HT4 receptor density 95). The model quantified the tissue-dependent partial agonism of prucalopride, M0003 and M0004 with prucalopride behaving as a full agonist in the gastric assay, and as a partial agonist in the left and right atrium whereas tegaserod was shown to be a full agonist in all assays.21 Tissue-selective expression of partial agonism might thus result in organ selectivity of a certain drug, as illustrated for prucalopride above. Interestingly, because of their low efficacy, benzamides and dihydrobenzofurancarboxamides actually antagonize the effects of exogenously administered 5-HT, as demonstrated for prucalopride, cisapride, renzapride and metoclopramide in human and/or porcine right atrium.91,93,96 A similar antagonizing and possibly protective behaviour might also occur vs endogenously released 5-HT as demonstrated in a porcine in vivo model in which renzapride reduced the 5-HT-induced tachycardia.97
Secondly, the situation is further complicated by the concept of agonist-directed trafficking of receptor signalling. This hampers the interpretation and extrapolation of in vitro tissue experiments since different agonists might couple to different effectors (which might be even more complicated by the differential expression of splice variants between tissues or species, see above) and consequently, the order of efficacy between agonists might be reversed in different tissues (or species). Several examples of reversed efficacy order between 5-HT4 receptor agonists can be found in the literature. In many systems where both agonists were tested, prucalopride has been described as the agonist with a lower efficacy compared with tegaserod,15,21,92,98 while in others prucalopride is classified as a full and tegaserod as a partial agonist.34,39 Interestingly, at the guinea-pig 5-HT4 receptor expressed in HEK293 cells, tegaserod behaved as a full agonist and prucalopride as a partial agonist. This order was reversed in the guinea-pig isolated colon longitudinal muscle myenteric plexus preparation (LMMPP).99 In addition, compared to 5-HT, cisapride has been shown to behave as a superagonist in mouse colliculi neurones, while in most GI preparations it is a partial agonist.9 Furthermore, the effect of prucalopride on substance P release in the guinea-pig colon (NK1-mediated rebound contraction) has not been observed with cisapride (30, personal observation). Although speculative, all these observations might result from the stabilization of a different conformational state of the 5-HT4 receptor by the different agonists.
Taken together, based on mixed drug-specific and tissue-dependent properties (e.g. the expression of partial agonism or agonist-specific signal transduction pathways), different agonists might differentially activate the multiple 5-HT4 receptor populations in the GI tract, thereby resulting in distinct overall effects in different tissues.
The tissue selectivity of agonists could also be affected by the degree of receptor desensitization. It is well established that 5-HT4 receptors can desensitize upon stimulation by an agonist as shown in recombinant expression systems, but also with native neuronal and non-neuronal receptors.100,101 It is often generalized that a partial agonist is less likely to induce desensitization.102 This general paradigm that the activity of an agonist and its ability to cause desensitization are coupled is based on studies on β2-adrenergic and muscarinic receptors demonstrating that partial agonists cause less desensitization than full agonists.103 However, for several 5-HT4 receptor agonists, the desensitization power was shown to be a function of the activation potency of the drugs rather than of their efficacy (partial agonistic character).100 The dependence on potency and not on efficacy reflects the influence of the agonists’ binding affinity. This is underscored by the fact that antagonists can also induce desensitization104 which even challenges the involvement of G protein activation and second messenger generation in the triggering of the desensitization. Indeed, for the motilin receptor it has been shown that desensitization can be decoupled from activation and desensitization is not only determined by potency.105 The existence of agonist-specific receptor conformations (see above), each unfolding and exposing different regions to the intracellular apparatus and each associated with specific function may render different desensitizing properties to two agonists despite similar activation potencies. Furthermore, since the desensitization process depends on the cellular environment of the 5-HT4 receptor,106 different agonists might induce different degrees of desensitization in different tissues, resulting in agonist-specific tissue selectivity; this also complicates extrapolation of data obtained in recombinant receptor systems. Above that, there are also splice variant-specific differences in desensitization.107,108 As shown in heterologous expression systems, 5-HT4(d) and 5-HT4(b) desensitize more rapidly than do 5-HT4(e) and 5-HT4(a). Therefore, different organ systems might exert different degrees of desensitization after stimulation with the same concentration of agonist if they express different 5-HT4 receptor splice variants. Differential effects on desensitization may have important therapeutic implications because desensitization can limit the biological effects of 5-HT4 agonists in vivo.
Differences with 5-HT at cardiac 5-HT4 receptors. The QT prolonging effect of cisapride (see above) together with the cardiac location of 5-HT4 receptors did establish an atmosphere of confusion on the cardiac effects of 5-HT4 receptor agonists. Some authors still associate the cardiac effects of cisapride with cardiac 5-HT4 receptors.109 This is, however, completely wrong. 5-HT4 receptors are expressed in human atria and ventricles, albeit at very low densities.110 Therefore, the effects of many 5-HT4 receptor agonists (benzamides and dihydrobenzofurancarboxamides; see above) in the heart are particularly modest compared to 5-HT. These effects include an increase in contractile force (inotropic effect), a hastening of the onset of muscle relaxation (lusitropic effect) or for sinoatrial receptors an increase in beating rate (chronotropic effect). The very low efficacy compared with 5-HT is particularly relevant for the hypothetical association between 5-HT4 receptor activation and atrial arrhythmias, for which no evidence was found using renzapride and prucalopride96,111 but only using 5-HT at high concentrations and in specific in vitro conditions or preconditioned tissues. Prior treatment with ß-adrenoreceptor antagonists enhances the inotropic and arrhythmogenic potential of 5-HT by increasing the maximal response of ICaL to 5-HT112 without affecting expression levels of 5-HT4 receptors or L-type Ca2+ channel subunits.113 Interestingly, even in cells from β-adrenoreceptor antagonist-treated patients, prucalopride was devoid of any effect on arrhythmic activity.94 Proarrhythmic effects of 5-HT have been suggested to be related to Ca2+ overload. Prucalopride induces a smaller increase in L-type Ca2+ current compared with 5-HT and it would therefore appear unlikely that prucalopride could elicit atrial arrhythmias in patients with constipation.94,96 Prucalopride would actually block cardiac 5-HT4 receptors (see above), thereby even preventing arrhythmogenic effects of 5-HT.96
Cardiac 5-HT4 receptor-mediated responses quickly fade because of the action of phosphodiesterases (PDEs) which catalyse cAMP breakdown. Their inhibition is often needed to unmask the effects of 5-HT4 receptor agonists and 5-HT, as illustrated for the left atrium and the ventricles respectively.92,114 In addition to quenching of the 5-HT4 receptor-mediated effect, PDEs play a role in the compartmentation and targeting of the cAMP signal in the myocytes. Because of this mechanism, activation of different Gs-coupled receptors (e.g. β-adrenoreceptors and 5-HT4 receptors), both increasing cAMP levels, may lead to different effects on cardiac function. Because PDE inhibition greatly reduces cardiac compartmentation, the inotropic effect of the 5-HT4 receptor agonists in the presence of PDE inhibitors could be a direct consequence of the inhibition of PDE action (less cAMP breakdown) and/or, as compartmentation is lost, result from the more diffuse activation of cAMP substrates. Therefore, the physiological relevance of the experimental 5-HT4 receptor-mediated responses in the presence of a PDE inhibitor is questionable.
The large species-dependent heterogeneity in 5-HT receptor subtypes expressed in the heart limits the options to study all of this in a laboratory animal. Only pigs and monkeys have functional cardiac 5-HT4 receptors under physiological conditions. In addition, 5-HT4 receptor mRNA was detected in rat hearts, but a functional inotropic response only manifested after its upregulation following the induction of post-infarction congestive heart failure.115,116 Up as well as down regulation of 5-HT4 receptor mRNA expression levels associated with cardiac dysfunction has been demonstrated in human hearts.114,117