The progression and desensitization of 5-HT4 receptor-mediated responses were evaluated in two clinically relevant in vitro models: a porcine GI model of cholinergic neurotransmission (De Maeyer et al., 2006b), and the porcine atrium. Summarized, our results show that exposure of the gastric tissue to 5-HT or selective 5-HT4 receptor agonists results in a sustained, non-transient effect during exposure; after washout this does not result in a clear desensitization of the response to a subsequent challenge with 5-HT. Incubation of left atrial tissue with 5-HT results in a transient response, after washout leading to a potent desensitization of the subsequent response to 5-HT. The selective 5-HT4 receptor agonists prucalopride and M0003 induced only very weak atrial responses but appeared very effective in desensitizing the response to 5-HT. Our observations also suggest that the properties of prucalopride and M0003 to bind and activate the 5-HT4 receptor differ from those of 5-HT, and this might have contributed to the observed desensitization.
In the gastric muscle preparations, 5-HT4 receptors are presumably located presynaptically on myenteric cholinergic nerves (De Maeyer et al., 2006b). By facilitating the release of acetylcholine, their activation results in a reinforcement of muscle contraction through activation of M3 receptors on the smooth muscle cells (Leclere and Lefebvre, 2002b). 5-HT, prucalopride and M0003 all persistently increased the cholinergic contraction of the gastric muscle strips, confirming our previous observations (De Maeyer et al., 2006b). This suggests either the absence of receptor desensitization or the sustained activation of a mechanism downstream of the receptor. In the latter scenario, the 5-HT4 receptors themselves could well be desensitized while the response initiated by their activation is not. Indeed, in spite of a rapidly desensitizing adenylyl cyclase response to 5-HT4 receptor activation in mouse colliculi neurons (Ansanay et al., 1992), their activation does result in long-term effects. In these cells, a 1 min application and washout of 5-HT causes a 5-HT4 receptor-mediated transient protein kinase A (PKA) response and a long-lasting inhibition of phosphatases resulting in a progressive and long-lasting blockade of K+ channels (Ansanay et al., 1995). Similarly, the effect of prucalopride appeared to be wash-resistant in rat hippocampal slices (Spencer et al., 2004). In our experiments, the effects mediated by prucalopride and M0003 were also resistant to drug washout, theoretically corresponding with the possibility of sustained facilitation of transmitter release by these two drugs via a mechanism downstream of the receptor. However, two observations cast doubt on the validity of this hypothesis in the gastric preparations. First, the phenomenon is absent for 5-HT as the effects induced by 5-HT were reversed after washout of the drug (regardless of a small trend for 0.3 and 1 µmol·L−1 5-HT). Second, the 5-HT4 receptor antagonist GR113808 was able to nullify the sensitized cholinergic contractions before and after (results not shown) washout of the drugs, suggesting that the receptors are still active. In colliculi neurons, the effect of 5-HT was abolished by pre-incubation with a 5-HT4 receptor antagonist, but its effect was not reversed when the antagonist was applied during washout of the drug (Ansanay et al., 1995). In our experiments, an activated state of the receptor is thus required to sustain the response and no long-term effect on a downstream effector has occurred. There may be a mechanistic analogy with the colon because Cellek et al. (2006) have described the non-transient effect of prucalopride on cholinergic contractions and nitrergic relaxations in the human colon, both effects being reversed upon administration of a 5-HT4 receptor antagonist. The observation that application of 5-HT, 15 min after a conditioning application of 5-HT, prucalopride or M0003 had an additional facilitatory effect on the EFS-induced contractions resulting in similar EFS-induced contractions as in solvent-treated strips also argues against the occurrence of desensitization; in other words, the maximal response was always reached, whatever the level of the basal value. It is also possible that because of the high receptor reserve for 5-HT in this preparation (De Maeyer et al., 2006b), 5-HT4 receptor desensitization has only moderate effects on the agonist-induced responses. Similarly, we cannot exclude the possibility that receptors already have been (partially) resensitized, because in colliculi neurons and CHO cells, partial recovery of the 5-HT4 receptor-induced cAMP production was already observed after 30 min (Ansanay et al., 1992; Mialet et al., 2003); because of the high receptor reserve for 5-HT in the stomach, only part of the receptors need to be operational again to achieve the same tissue response.
The experiments with 10 µmol·L−1 5-HT show that at this concentration, a receptor that is resistant to high concentrations of GR113808 is involved in the EFS-induced responses. Because our experiments were performed in the presence of granisetron and methysergide, 10 µmol·L−1 5-HT either surmounts antagonism by these antagonists, or an unknown 5-HT receptor is involved. The tonic contractile response to 5-HT in the same preparation as used in this study has previously been shown to be mediated by 5-HT2A and a receptor that could not be characterized (Janssen et al., 2002).
It appears that prucalopride and M0003 are unique 5-HT4 receptor agonists in that they interact with the receptor both in an antagonist-reversible but wash-resistant manner. Although we cannot exclude the possibility that the compounds partition into biological membranes, resulting in the development of a substantial depot of agonist that exhibits relative resistance to washing, we question this explanation because the hydrophobicity of prucalopride, M0003 and 5-HT is comparable (data on file, Movetis NV). Interestingly, the persistent nature of the interaction of the two benzofurancarboxamides, prucalopride and M0003, shows many similarities to xanomeline, a muscarinic M1 acetylcholine receptor (mAChR) agonist (Christopoulos et al., 1998; De Lorme et al., 2007): (i) resistance to extensive washout and persistent agonistic effect; (ii) concentration dependency of this effect; and (iii) susceptibility to reversal by the addition of the antagonist atropine. For xanomeline, a model was proposed that incorporated two possible modes of interaction for this ligand with the M1 receptor (Christopoulos et al., 1998). The first mode involves a reversible, syntopic interaction with the classic binding site on the receptor, shared by other agonists or antagonists. The second mode involves the subsequent development of a persistent attachment with specific receptor regions that allowed xanomeline to continue to activate the M1 mAChR via the classic binding site but did not allow it to be readily removed from the receptor compartment. In contrast to 5-HT, prucalopride and M0003 have the structural features to interact in a hydrophobic binding pocket on the 5-HT4 receptor (Rivail et al., 2004; De Maeyer et al., 2008). This pocket might fulfil the role of the wash-resistant binding site. However, whether prucalopride and M0003 do, indeed, interact with the 5-HT4 receptor in a way similar to that of xanomeline with M1 receptors, needs further investigation.
In order to evaluate the long-term off-target effects of 5-HT, prucalopride and M0003, these compounds were also studied in the porcine left atrium. Next to human, the pig is one of the few animals expressing functional 5-HT4 receptors in the heart. In contrast to the response in gastric tissue, the increase in contraction force upon application of 5-HT rapidly faded and was not reproducible after washout. Because we only monitored changes in the EFS-induced contraction-relaxation cycle, we cannot deduce the exact underlying mechanism of this response desensitization. Several mechanisms might be involved, each contributing to the observed diminishing response and these mechanisms will be discussed below.
We previously described that the transient nature of the influence of 5-HT on electrically induced atrial contractions does not result from ligand breakdown but involves the recruitment of PDEs, limiting the magnitude and duration of cAMP signals elicited by a 5-HT4 receptor agonist in the atrial tissue (De Maeyer et al., 2006a). In a recent abstract, using paced left atrium of newborn piglets, Vargas et al. (2006) showed that after a 2 min incubation period with 10 µmol·L−1 5-HT, left atrial contraction force was increased but not the (global) left atrial cAMP levels. After 20 min, the inotropic response had faded while cAMP levels were markedly increased. In the presence of PDE3 and PDE4 inhibitors, the inotropic response as well as the tissue cAMP levels were increased after 2 min and still increased after 20 min. This confirms the involvement of PDEs in the fading response and suggests that PDEs contribute to the compartmentation of the 5-HT4 receptor-mediated cAMP signalling pathway in the left atrium by limiting the diffusion of the second messenger. PDE activation might provide a framework for a negative feedback that controls global cAMP homeostasis beneath the membrane. The inhibition of PDEs therefore generates a condition that is very different from the physiological situation, with no control over subsarcolemal cAMP levels (Fischmeister et al., 2006).
Still, it is conceivable that the 5-HT4 receptor-mediated inotropic response is affected not only by PDE recruitment, but also by receptor desensitization (inactivation), by analogy with β2-adrenoreceptors for which the transient cAMP response in HEK-293 cells could be explained by a GRK-β-arrestin-mediated receptor inactivation as well as a PKA-mediated induction of PDE activity (Rochais et al., 2004; Violin et al., 2008). In recombinant systems, 5-HT4 receptor activation has indeed been shown to be followed by rapid receptor desensitization through a process involving GRK and β-arrestin (Barthet et al., 2005; Ponimaskin et al., 2005). The results obtained in protocol C showed that the pEC50 and the maximal response of the concentration-response curve for the inotropic effect of 5-HT were decreased after pre-treatment with 5-HT, also in the presence of the PDE inhibitor IBMX. This also implies that the decreased inotropic response upon repeated administration of 5-HT is not a purely PDE-mediated phenomenon, although inhibition of PDE activity by IBMX might not be complete.
The expression level of 5-HT4 receptors in the atrium is very low, i.e. ∼0.3 and ∼4 fmol·mg−1 protein in newborn piglets and adult man, respectively (Kaumann et al., 1995; 1996), compared with ∼ 225 fmol·mg−1 protein in human brain (Domenech et al., 1994). Additionally, in contrast to gastric tissue, there is no 5-HT4 receptor reserve for 5-HT in atrial tissue (De Maeyer et al., 2006b). This might explain why, in atria in contrast to gastric tissue, there is a robust desensitization of the response to 5-HT.
We do acknowledge two additional factors that might have added to the observed diminishing response to 5-HT. One factor is the incomplete washout of prucalopride and M0003. As described above, our data obtained with gastric tissue indeed suggest an antagonist-reversible but wash-resistant receptor interaction of these agonists. Both agonists, as well as GR113808, have the structural features to interact in a hydrophobic binding pocket on the 5-HT4 receptor, putatively fulfilling the role of a wash-resistant binding site. In the atrium, the response to these agonists already fades during the exposure period because of PDE recruitment and this could hide the fact that they are still present after the washing step. If the agonists are bound in a wash-resistant way, they could act as an antagonist for subsequently administered 5-HT; antagonism of 5-HT was previously shown for prucalopride at human and porcine atrial 5-HT4 receptors (Krobert et al., 2005). Beyond these considerations, because the agonists are still bound to the receptor after the washing step, PDE recruitment would persist at the time of 5-HT administration. In the experiments with 5-HT itself as desensitizing agent in the presence of IBMX, the 5-HT-induced response was maintained during the desensitizing exposure period and after washout, theoretically suggesting that wash-resistant binding might also occur for 5-HT. We are, however, not aware of any reports on wash-resistant binding of 5-HT to the 5-HT4 receptor.
The second contributing factor can be found in the activation of the lusitropic apparatus. We previously demonstrated that 5-HT4 receptor activation caused a non-transient lusitropic effect which is resistant to inhibition by PDEs (De Maeyer et al., 2006a). We now show that this enhancement of cardiac diastolic muscle relaxation is also resistant to drug washout. The effect of 5-HT4 stimulation in accelerating relaxation is believed to involve a combination of an increased rate of SR Ca2+ uptake due to phospholamban (PLB) phosphorylation, and a reduction in myofilament Ca2+ sensitivity and an increase in cross-bridge cycling rate due to troponin I (TnI) phosphorylation (Li et al., 2000; Birkeland et al., 2007). Our results suggest that neither PDEs nor drug washout interferes with the different intracellular processes involved in the positive lusitropic effect of 5-HT4 receptor stimulation. This can be explained by the slow dephosphorylation rates for PLB and TnI (Garvey et al., 1988). The relative contribution of both proteins to the lusitropic effect is highly dependent on the loading conditions (Li et al., 2000). It has been suggested that under isometric conditions, because of an increased sensitivity of myofilament for Ca2+ in such heavy loading conditions, the relaxation time course is mainly determined by Ca2+ dissociating from troponin C, and that the lusitropic coefficient R2 indirectly reflects myofilament Ca2+ sensitivity (Hanouz et al., 1998). A primary role for TnI is indirectly supported by our results because the lusitropic response was still maintained after fade of the inotropic response. Because phosphorylation of PLB also has an inotropic effect by increasing Ca2+ load of the SR (Birkeland et al., 2007), one would not expect fading to basal values. The situation might be different when blocking PDEs, because this disrupts spatially compartmentalized cAMP production (Rochais et al., 2004). The involvement of long-term phosphorylation of PLB in the wash-resistant positive inotropic effect of high concentrations of 5-HT in the presence of IBMX can be ruled out by the ability of GR113808 to reverse this maintained response to 5-HT (preliminary results not shown). The long-term activation of the lusitropic machinery implies that the effect of 5-HT4 receptor activation on the subsequent response to 5-HT was assessed under conditions of reduced Ca2+ sensitivity of the myofilaments, which might in part explain the desensitized inotropic responses to 5-HT.
Because the inotropic effects of prucalopride and M0003 were very small, these compounds allowed us to examine further the desensitization properties of left atrial 5-HT4 receptor-mediated responses. For many GPCRs there is a good correlation between agonist efficacy and the ability to induce receptor desensitization (Clark et al., 1999). However, pre-treatment with prucalopride or M0003 very effectively blunted the response to 5-HT, arguing against this generalization. This finding can be interpreted in three ways. First, in colliculi neurons, a good correlation was found between the potencies of several 5-HT4 receptor agonists and their abilities to desensitize the adenylyl cyclase response (Ansanay et al., 1992). According to the authors, the dependence on potency, and not on efficacy, reflects the influence of the agonists' binding affinity or its mean occupancy time of the receptor. Our results fit into this theory because pre-treatment with prucalopride or M0003 very effectively blunted the response to 5-HT with a potency proportional to their 5-HT4 receptor affinity. Second, the correlation with affinity is also in accordance with antagonism by residually bound agonist (see above). Finally, this can also be interpreted in light of ‘functional selectivity’. The idea that different agonists are able to produce different response profiles by stabilizing different receptor conformations and activating different signal transduction pathways or mechanisms of desensitization is extending (Kelly et al., 2008). This has clearly been shown for the µ-opioid receptor for which it has been demonstrated that different agonists induce different mechanisms of µ-opioid receptor desensitization (Kelly et al., 2008).
In conclusion, our results show a clear-cut difference between the desensitization of the 5-HT4 receptor-mediated responses in stomach and left atrium. In the stomach, the responses were only apparently desensitized while in the atrium the responses to 5-HT were clearly not reproducible. In addition to the efficacy differences of 5-HT4 receptor agonists such as prucalopride and M0003 in stomach versus atrium (De Maeyer et al., 2006b), the phenomenon might contribute to tissue selectivity of 5-HT4 receptor agonists.