Abstract
This study documents differences in ligand binding and signal transduction properties between the human (h) 5-hydroxytryptamine (5-HT)4a and h5-HT4b receptor splice variants stably expressed in human embryonic kidney 293 cells. The fraction of the [3H]5-HT high-affinity site relative to the whole receptor population measured with [3H]GR113808 was higher for the h5-HT4a isoform (around 0.4) than for the 5-HT4b isoform (around 0.2) and was independent of the level of expression. The potency and efficacy of reference compounds tested for the cAMP response differed slightly but significantly between both variants. Most remarkably, 5-methoxytryptamine and prucalopride were found more potent on the 5-HT4b variant, whereas SDZ-HTF 919 and SB204070 were more potent on the 5-HT4a variant. Guanosine-5′-O-(3-[35S]thio)triphosphate binding on membranes and cAMP assays in whole cells revealed that only the h5-HT4b isoform coupled to Gαi/o-proteins in addition to its well-documented Gαs coupling. In contrast, the h5-HT4a receptor coupled only to Gαs-proteins, however, was able to trigger an increase in the intracellular calcium concentration ([Ca2+]i). The observed [Ca2+]i increase did not occur through inositol phosphate formation and was not sensitive to Bordatella pertussis toxin, forskolin, or 3-isobutyl-1-methylxanthine (pre)treatment but was due to Ca2+ influx from the extracellular environment. Interestingly, the Ca2+ pathway was dependent on high receptor expression levels and was compound-specific, because benzamide-like compounds triggered two to three times higher responses than indoleamines. Taken together, these data provide the first evidence for fine functional differences between C-terminal splice variants of the h5-HT4 receptor, which may contribute to a better understanding of the functional diversity of this receptor class.
The ubiquitous neurotransmitter 5-hydroxytryptamine (5-HT, serotonin) has so far been shown to interact with seven receptor classes, classified as 5-HT1 to 5-HT7receptors. Three of them were defined as Gαs-protein-coupled receptors: the 5-HT4, 5-HT6, and 5-HT7 receptors. In this study, we focus on the functional properties of the 5-HT4 receptor class, in which a new level of structural diversity has been highlighted in recent years.
The wide distribution of 5-HT4 receptors, from the central nervous system to the peripheral tissues, suggests a wide range of functional roles. In the central nervous system, 5-HT4 receptors have been localized in rat basal ganglia, hippocampus, and olfactory tubercule (Compan et al., 1996;Vilaro et al., 1996). In human brain, radioligand binding (Domenech et al., 1994; Reynolds et al., 1995) and in situ hybridization studies (Bonaventure et al., 2000) have shown predominant expression of 5-HT4 receptors in basal ganglia and limbic structures. Functionally, central 5-HT4 receptors have been implicated in diverse processes such as anxiety, memory, and cognition (Silvestre et al., 1996; Fontana et al., 1997). 5-HT4 receptors have also been shown to be widely distributed in peripheral organs such as the gastrointestinal tract, the myocardium, the urinary bladder, and the adrenal glands (for review, see Ford and Clarke, 1993). Furthermore, pharmacological differences have been identified in various tissues and species, suggesting a heterogeneity of 5-HT4 receptors (Ford and Clarke, 1993; Leung et al., 1996). Indeed, Gerald et al. (1995) have initially cloned two alternative splice variants coding for short (5-HT4S) and long (5-HT4L) isoforms of the receptor, which could be the structural basis for functional diversity. So far, seven receptor variants that differ in their C termini have been identified: 5-HT4a (previously named 5-HT4S), 5-HT4b (previously named 5-HT4L), 5-HT4c, 5-HT4d, 5-HT4e, 5-HT4f, and 5-HT4g(Claeysen et al., 1997; Blondel et al., 1998; Bender et al., 2000). Moreover, by cloning the human 5-HT4 receptor gene, our group has recently shown the existence of an additional site of alternative splicing (5-HT4h), leading to an extended second extracellular loop that could theoretically combine with each C-terminal variant (Bender et al., 2000). Interestingly, the only experimentally isolated h5-HT4hb-receptor variant showed an altered functional pharmacology because the prototypic 5-HT4 receptor antagonist GR113808 exhibited partial agonistic properties. This finding supported the hypothesis of the existence of a functional diversity among 5-HT4 receptor splice variants. In addition, we have recently shown a differential tissue distribution of the h5-HT4 receptor isoforms (Bender et al., 2000), which may also contribute to tissue-specific functional differences. To date, all isoforms have been shown to activate adenylyl cyclase (AC) in vitro, and no difference in signal transduction between C-terminal 5-HT4 receptor variants has been demonstrated. Although the functional significance for the existence of 5-HT4 receptor splice variants is currently not well understood, studies on other receptors suggest that different C termini may discriminate between signal transduction pathways. G-protein selectivity was shown to be different for the C-terminal splice variants of the prostaglandin E receptor EP3 (Satoh et al., 1999), whereas somatostatin receptor isoforms have been found to differ in the coupling efficiency and desensitization events (Vanetti et al., 1993). To better understand the functional significance of 5-HT4 receptor isoforms, we investigated the pharmacological profile and signal transduction of h5-HT4a and h5-HT4breceptors, two initially cloned C-terminal splice variants.
We have found major differences between the two variants, not only in the pharmacology of the cAMP response and the fractions of agonist high-affinity sites, but also in the selection of the G-proteins activated by the receptors and the second messengers triggered by agonists. Interestingly, this differential signaling was receptor variant and compound specific. These data provide further evidence for functional differences between 5-HT4 receptor isoforms, underscoring a physiological relevance for the existence of 5-HT4 receptor splice variants.
Experimental Procedures
Materials.
[3H]GR113808 (specific activity of 3.7 TBq/mmol), 5-hydroxy[3H]tryptamine triacetate ([3H]5-HT, specific activity of 4.44 TBq/mmol), and [35S]GTPγS (specific activity of 37 MBq/ml) were obtained from Amersham Biosciences (Little Chalfont, Buckinghamshire, UK). The mammalian expression vector pcDNA3 was obtained from Invitrogen (Carlsbad, CA). Dulbecco's modified Eagle's medium (DMEM), phosphate-buffered saline (PBS), sodium butyrate, Geneticin (G-418), and calf serum were from Invitrogen. The Bradford protein assay was performed with the reagent supplied by Bio-Rad (Nazareth Eke, Belgium). The Adenylyl Cyclase Activation FlashPlate kit was supplied by PerkinElmer Life Sciences (Brussels, Belgium). The Bordatella pertussis toxin (PTX) was from Calbiochem (La Jolla, CA). The liquid scintillation spectrometer, the scintillation fluids Ultima Gold MV and Ultimaflo AF, and the Unifilter-96 GF/B plates were from Packard (Meriden, CT). GDP and dilithium salt were from Hoffman-La Roche (Basel, Switzerland). Fluo3-acetoxymethyl ester was purchased from Molecular Probes (Leiden, The Netherlands). Forskolin and probenecid were from Sigma (St Louis, MO). Cisapride and prucalopride are proprietary compounds from Janssen Pharmaceutica (Antwerp, Belgium). Renzapride, SB204070, GR113808, SDZ205070, and SDZ HTF919 were synthesized by Janssen Pharmaceutica for its own purposes. 5-HT and 5-MeOT were from Acros Organics (Geel, Belgium). For pharmacological testing, compounds were dissolved and diluted in dimethyl sulfoxide except for indoleamines, which were dissolved in water and protected from light throughout the experiment. The final dimethyl sulfoxide concentration never exceeded 0.5% (v/v). The GraphPad Prism program was from GraphPad Software, Inc. (San Diego, CA).
Stable Transfections and Selection of Monoclonal Cell Lines.
The h5-HT4a and h5-HT4breceptors were stably expressed in HEK 293 cells. Expression was under the control of the constitutive cytomegalovirus promoter provided by the pcDNA3 vector. The calcium phosphate transfection method was used with modifications as described previously (Lesage et al., 1998). Stable monoclonal cell lines were created by limited dilution and selected by [3H]GR113808 radioligand binding after culture under Geneticin (G-418) selection (800 μg/ml). Receptor expression levels of a range of isolated monoclonal cell lines were investigated with the antagonist [3H]GR113808 in radioligand binding experiments; nontransfected HEK 293 cells did not show specific [3H]GR113808 binding (data not shown). Of the 39 monoclonal cell lines obtained after transfection with the h5-HT4a receptor, six had expression levels of about 2000 fmol/mg of protein after treatment with sodium butyrate (see below). Clone 8 (maximal expression levels ∼3100 fmol/mg) was used for further studies and characterization. Another clone (clone 1) with a lower expression level (maximal ∼1500 fmol/mg) was used to investigate the contribution of different receptor expression levels in the functional tests performed in this study. Of the 33 monoclonal cell lines expressing the h5-HT4b receptor, cell clone 9 had an expression level of about 7000 fmol/mg of protein. Clone 9 was used for further study and characterization, and as for the other receptor splice variant, a clone of lower expression level (clone 4, maximal ∼2000 fmol/mg) was used to investigate the effect of different receptor expression levels in the functional tests performed in this study.
Cell Culture and Treatments.
Cells were grown in DMEM containing 10% dialyzed calf serum, 105 IU/l penicillin G, 0.1 g/l streptomycin, 0.1 g/l pyruvate, and 0.292 g/ll-glutamine, under 5% CO2 at 37°C. Cells were grown under selection (800 μg/ml G418) for 48 h every 2 weeks. Culture plates were coated with PBS containing 10 μg/ml poly(l-lysine) for 45 min at 37°C and washed once with PBS before the cells were seeded.
In certain experiments, receptor expression levels were also manipulated using sodium butyrate, which is a histone deacetylase inhibitor found to arrest growth and to induce differentiation in various cell types by modulating gene expression (Archer et al., 1998). Therefore, sodium butyrate is a “differentiating agent” able to change expression levels of various proteins. The observed and described differences between both isoforms have been obtained using the same conditions in sodium butyrate-treated or nontreated cells. In addition, we provide data from clones exhibiting lower receptor levels whenever appropriate to understand the effects. In our experience, short-term overnight sodium butyrate treatments can be used to boost receptor expression without altering the pharmacological properties of the receptor. For sodium butyrate treatment, cells were incubated in medium containing 5 mM sodium butyrate for 16 to 18 h before experiments to boost receptor expression before membrane preparation for radioligand saturation binding, [35S]GTPγS binding, and calcium measurements. PTX was used to prevent the coupling of Gαi/o-proteins to the receptor and was dissolved in 10 mM sodium phosphate, pH 7.0, and 50 mM NaCl and added to the cell medium overnight before the experiment at a concentration of 100 ng/ml.
Membrane Preparations and Radioligand Binding.
Cells were cultured on 150-mm Petri dishes and washed twice with ice-cold PBS. The cells were then scraped from the plates with a cell scraper, suspended in 50 mM Tris-HCl buffer, pH 7.4, and harvested by centrifugation at 16,000g for 10 min. The pellet was re-suspended in 5 mM Tris-HCl, pH 7.4, and homogenized with an Ultra Turrax homogenizer (IKA Labortecnik, Staufen, Germany). The resulting membranes were collected by centrifugation at 25,000g for 20 min and stored at −70°C in 50 mM Tris-HCl buffer, pH 7.4, at a protein concentration of approximately 1 mg/ml. The Bradford protein assay was used for protein determination with bovine serum albumin as a standard.
For radioligand binding, assay mixtures (0.5 ml) contained 50 μl of the tritiated ligand (either the 5-HT4 antagonist [3H]GR113808 or the agonist [3H]5-HT) and 0.4 ml of membrane preparation (at 0.010 mg/ml protein for [3H]GR113808 binding or 0.1 mg/ml for [3H]5-HT). Furthermore, 50 μl of solvent was added for total binding or either 50 μl of 10 μM SB204070 for [3H]GR113808 or 50 μl of 10 μM GR113808 for [3H]5-HT binding, for determination of nonspecific binding. The [3H]GR113808 assay buffer was 50 mM HEPES-NaOH, pH 7.5. The [3H]5-HT assay buffer was 50 mM Tris-HCl, pH 7.4, containing 10 mM MgCl2, 1 μM pargyline (monoamine oxidase inhibitor), and 1 μM paroxetine (5-HT transporter inhibitor). The mixtures were incubated for 1 h at 25°C. The incubation was terminated by rapid filtration over Whatman GF/B filters presoaked in 0.15% polyethylenimine, followed by three washing steps with 3 ml of 50 mM HEPES-NaOH, pH 7.5, for [3H]GR113808 binding, or 3 ml of 50 mM Tris-HCl, pH 7.4, for [3H]5-HT binding. Ligand concentration isotherms were obtained with eight concentrations of [3H]GR113808 ranging from 0.02 to 1 nM. For [3H]5-HT saturation curves, eight concentrations ranging from 0.2 to 6 nM were used to characterize the high-affinity site, whereas 14 concentrations from 0.2 to 40 nM were chosen for the determination of the low-affinity site.
Ligand concentration binding isotherms (rectangular hyperbola) were calculated by nonlinear regression analysis as described previously (Lesage et al., 1998). The maximal number of binding sites (B max) and equilibrium dissociation constant (K D) of the radioligand were derived from the curve fitting. Graphs were prepared with the GraphPad Prism program.
Measurements of cAMP Formation.
After 2 days of growth, cells were detached from the Petri dishes with 3 ml of EDTA (0.04% w/v in PBS) and resuspended in PBS without Ca2+ and Mg2+. The cells were centrifuged at 500g for 5 min. The pellet was resuspended in the “stimulation buffer” provided with the PerkinElmer kit and diluted to a concentration of 106 cells/ml, and 50 μl was added per well of the Flashplate (50,000 cells/well). Compounds were diluted in PBS without Ca2+ and Mg2+ containing 1 μM pargyline and 1 μM paroxetine. Fifty microliters of the compound solution was added per well, followed by an incubation for 20 min at 37°C. After the incubation period, a direct radioimmunoassay using125I-cAMP was performed by the addition of 100 μl of detection mix per well according to the supplier's instructions. After incubation for 18 h at room temperature, counting was performed in a Topcount HTS9912V counter (Packard).
[35S]GTPγS Binding Assays.
For further functional testing, the [35S]GTPγS binding assay was performed on membrane preparations. Membranes were incubated (105 μg/ml) in 20 mM HEPES-NaOH, pH 7.4, containing 10 mM MgCl2, 1 μM GDP, 1 μM pargyline, and 1 μM paroxetine, in the presence of 0.25 nM [35S]GTPγS and compounds. The final volume was 250 μl and the incubation was performed in polypropylene tubes (PPN-tube-96; Micronic bv, Lelystad, The Netherlands) at 37°C for 20 min. The incubation was stopped by filtration on Unifilter-96 GF/B plates with ice-cold phosphate buffer, 10 mM Na2HPO4, adjusted to pH 7.4 with a solution of 10 mM NaH2PO4. The filter plates were dried at room temperature, 30 μl of Microscint was added per well, and plates were counted in a Topcount HTS9912V counter (Packard).
Measurements of Intracellular Calcium Concentration.
The intracellular [Ca2+]iwas measured with a fluorometric imaging plate reader (FLIPR; Molecular Devices, Crawley, England). Cells were seeded at 40,000 cells/well in 96-well plates (Cluster plates, black with clear bottom; Costar; Merck, Overijse, Belgium) for 2 days until they reached confluence. Cells were loaded with 2 μM Fluo3-acetoxymethyl ester in DMEM supplemented with 20 mM HEPES, pH 7.4, and 2.5 mM probenecid, for 1 h at 37°C under 5% CO2. The plates were washed three times with 5 mM HEPES, pH 7.4, 1.25 mM CaCl2, 140 mM NaCl, 1 mM MgCl2, 5 mM KCl, and 10 mM glucose, containing 2.5 mM probenecid (to prevent extrusion of Fluo-3), 1 μM pargyline, and 1 μM paroxetine. [Ca2+]i was measured by FLIPR, and the peak of the calcium transient was considered as the relevant signal. Treatment of the cells with 10 μM ionomycin (Ca2+-inophore) was performed as a control for dye loading and cell viability.
To investigate the source of the [Ca2+]i increase, we performed the same assay as described above, using several modifications. We performed the assay in the absence of extracellular calcium, where the Fluo3-loaded cells were washed two times with the above buffer without calcium. The final washing step and the measurements were performed in Ca2+-free buffer containing 2 mM EGTA. Ionomycin (10 μM) and 1% Triton X-100 were used as controls to estimate the maximal signals from extracellular and intracellular compartments, respectively. [Ca2+]i measurements were also performed on cells pretreated with 100 ng/ml PTX (overnight), 200 μM bepridil (15 min), 1 mM 3-isobutyl-1-methylxanthine (IBMX; 30 min), and 1 μM forskolin (15 min).
Results
Characterization of Cell Lines in Ligand Concentration Binding.
The two human 5-HT4 receptor isoforms investigated in this study, h5-HT4a and h5-HT4b, were stably transfected into HEK 293 cells. In this study, we have primarily used a high-level expressing clone of each splice variant, the h5-HT4a clone 8 and the h5-HT4b clone 9. In addition, clones with lower expression levels, h5-HT4a clone 1 and h5-HT4b clone 4, were employed in certain experiments to investigate the effect of receptor expression on the functional properties of each receptor variant. The isolated monoclonal cell lines were characterized by saturation binding, using two radioligands, an antagonist, [3H]GR113808, and the natural agonist, [3H]5-HT. The experiments were performed on membrane preparations from cells treated with sodium butyrate and from untreated cells. Independent of expression levels, the concentration binding isotherms of the antagonist [3H]GR113808 to h5-HT4band h5-HT4a receptors expressed in different clones showed rectangular hyperbolae. The resulting linear Scatchard plots revealed a single high-affinity binding site (K D values ranging between 0.05 and 0.1 nM for both 5-HT4 receptor isoforms). Furthermore, whereas treatment with sodium butyrate increased allB max values by almost 70%, it did not alter mean K D values. The calculatedK D and B maxvalues are presented in Table 1. For comparison, the B max values of the two lower level expressing clones, treated with sodium butyrate, were 1500 (±160) and 2087 (±513) fmol/mg of protein for the h5-HT4a clone 1 and the h5-HT4b clone 4, respectively.
The same comparison of the derived K D andB max values with and without sodium butyrate treatment, and between the two receptor splice variants, was performed with the agonist [3H]5-HT at eight concentrations ranging from 0.2 to 6 nM (Table 1). Interestingly, in agreement with the findings of the antagonist binding experiments, sodium butyrate treatment up-regulated B maxvalues for [3H]5-HT also by 60 to 86%, indicating that the ratio of G-protein-coupled to total receptors was not affected. Furthermore, sodium butyrate did not affectK D values for [3H]5-HT, which were found to be in the range of 1.2 to 3.0 nM. However, in contrast to the results obtained with [3H]GR113808, the Scatchard plots of [3H]5-HT binding indicated the existence of a low-affinity site when slightly higher concentrations were used.
When [3H]5-HT concentrations up to 40 nM were employed Scatchard analysis clearly showed two affinity sites for both splice variants (Fig. 1). The low-affinity site K D values computed from the saturation curve were 34 ± 12 nM for the h5-HT4a receptor and 18 ± 5 nM for the h5HT4b receptor in preparations treated with sodium butyrate. Without the treatment, the low-affinity siteK D values were 42 ± 19 and 26 ± 5 nM for the h5-HT4a and 5-HT4b receptor, respectively.
As an indication for the proportion of receptors that are in the conformation to bind an agonist with high affinity, we calculated the fraction [3H]5-HTB max value of the [3H]GR113808 B maxvalue (Table 1). Interestingly, this fraction of [3H]5-HT high-affinity sites was approximately two times higher for the h5-HT4a than for the h5-HT4b receptor splice variant and was not modified by sodium butyrate treatment and thus by receptor expression levels. This different ratio of receptor states observed for the two receptor isoforms could be a consequence of a higher G-protein coupling of h5-HT4a receptors, which could result in functional differences of the two receptor isoforms. These findings prompted us to investigate signal transduction events triggered by the two variants.
Adenylyl Cyclase Activation.
The activity of AC after h5-HT4 receptor stimulation with various reference ligands was estimated on living cells, which were not treated with sodium butyrate to elevate receptor expression levels. An increase of basal cAMP levels, found in the high expressing clones compared with wild-type HEK 293 cells, revealed the constitutive activity of both 5-HT4 receptors, which is consistent with previous reports (Claeysen et al., 1999, 2000). Compared with the basal values measured in the wild-type HEK 293 cells, the measured constitutive activity was 482% (±135) and 881% (±156) (mean ± S.D., n = 3) in 5-HT4a and 5-HT4b receptor-expressing cells, respectively.
The investigated agonists were the natural agonist 5-HT, its natural derivative 5-methoxytryptamine (5-MeOT), the two benzamides cisapride and renzapride, the benzofuran prucalopride, and the indole-derivative SDZ-HTF919 (Appel-Dingemanse et al., 1999). The reference antagonists used were GR113808, an indole carboxylate analog, SDZ205,557, a benzoate analog, and SB204070, an imidazole amide. All compounds were tested for agonistic efficacy, and the results were expressed as a percentage of maximal 5-HT response (5-HTmax); i.e., 100% of stimulation was the maximum of the fitted plots obtained with 5-HT. The derived pEC50 and 5-HTmax values are presented in Table2 and the concentration-response curves are shown in Fig. 2. All tested agonists were found to be full agonists for both h5-HT4aand h5-HT4b receptors. The antagonist GR113808 did not induce cAMP formation via h5-HT4a or via h5-HT4b receptors.
The putative reference antagonist SB204070 acted as partial agonists inducing 41 ± 3 and 50 ± 3% of the 5-HT-induced response on h5-HT4a and h5-HT4breceptors, respectively. SDZ205,557 also exhibited partial agonistic properties leading to 31 ± 7 and 30 ± 11% of maximal stimulation on h5-HT4a and h5-HT4b receptors, respectively. For SB204070 this difference in efficacy and in potency (9.5 ± 0.5 versus 8.8 ± 0.1) was found to be statistically significant. Another compound that showed a significant difference in efficacy of cAMP stimulation was cisapride, which behaved as a more efficacious agonist on the h5-HT4b-receptor isoform. In terms of potency 5-MeOT and prucalopride were significantly more potent on the 5-HT4b-variant, whereas SDZ-HTF919 and SB204070 were more potent on the h5-HT4a receptor isoform. Therefore, the stronger G-protein coupling of the h5-HT4b indicated in the saturation binding assays was not reflected in a general shift in pharmacology of the cAMP assay but was rather a specific property of tested compounds.
Changes in Intracellular Calcium Concentration.
To investigate other potential signal transduction pathways, we measured the modulations of [Ca2+]iafter h5-HT4a and h5-HT4breceptor activation. The same reference compounds as described for the cAMP experiments were employed. The h5-HT4areceptor showed fast increase in [Ca2+]i after exposure to all tested agonists, independent of the expression level found in both investigated clones. Interestingly, the Ca2+response induced by the h5-HT4a receptor was dependent on the chemical nature of the agonists. For the two agonistic benzamide-like compounds, cisapride and prucalopride, the peaks of the calcium transients were two and three times higher than for the natural agonist 5-HT and showed a different time course (Fig.3). Both cisapride- and prucalopride-stimulation needed a longer time (>100 s) to reach maximal [Ca2+]i than was observed with indoleamines (<50 s).
In contrast, the h5-HT4b receptor variant showed a totally different pharmacological response pattern despite a higher expression level. Specifically, only the natural indoles 5-HT and 5-MeOT could induce weak and slow increases in [Ca2+]i (Figs. 3 and 4). The pEC50 values derived from fitted concentration-response curves are presented in Table3. No response could be detected with the antagonist GR113808 and the partial agonist SDZ205,557. However, SB204070 was found to act as a weak partial agonist for the calcium response on the highly expressing h5-HT4a clone only and induced variable responses with a mean of 23% (±24) of the 5-HT-induced maximal response. This indicates that this partial agonist exhibits a lower efficacy in the calcium signaling than in the AC pathway. Several findings provide evidence that unlike in cAMP tests, the intensity of the calcium responses was dependent on the receptor expression level. Specifically, the higher expressing clone of the h5-HT4a receptor (clone 8) showed a higher response than the lower expressing h5-HT4a clone (clone 1). On average clone 8 responded with a 2- to 3-fold higher Ca2+-peak than clone 1, which is consistent with its 2-fold higher B max. In addition, the comparison of cells treated with sodium butyrate versus untreated cells showed a decrease in the signal intensity for h5-HT4a receptor cell lines and a disappearance of the signal for the h5-HT4b receptor variant in untreated cells (Fig. 4). Although sodium butyrate has been reported to up-regulate Na-channels (Kunzelmann et al., 1995), we believe that its enhancing effect on the Ca2+-signaling is rather linked to its up-regulating effect of receptor levels, because the treatment is not a prerequisite to measure a response (Fig. 4, D versus C). Taken together this results indicate that the calcium pathway is relatively h5-HT4a receptor-specific, dependent on receptor expression level, and more responsive to benzamide-like compounds in comparison to indoleamines.
To further investigate the molecular mechanism of the calcium response, we investigated whether the changes in [Ca2+]i were caused by an influx from the extracellular milieu and/or resulting from mobilization of intracellular stores.
Inositol Phosphates or Extracellular Calcium Influx.
Mobilization of calcium from intracellular stores would classically involve the activation of phospholipase C and was investigated by means of IP measurements. Assays were performed after h5-HT4a receptors had been stimulated with the natural agonist 5-HT and the benzofuran prucalopride. No increase of IP concentrations could be detected (data not shown) in contrast to a parallel control performed with CHO cells expressing h5-HT2A receptors. Our results, therefore, indicate that the h5-HT4a-induced calcium mobilization does not involve the phospholipase C/IP3 pathway.
When calcium was removed from the extracellular environment before h5-HT4a receptor stimulation, no response could be triggered with 1 μM 5-HT or 1 μM prucalopride (Fig.5A). These results indicate that the h5-HT4a receptor-induced increase in [Ca2+]i was caused by opening of calcium channels located in the plasma membrane. This hypothesis could be confirmed by pretreating the cells with 200 μM bepridil, which resulted in a complete blockade of the Ca2+-signal (Fig. 5B). In addition, the response was not dependent on the contribution of Gαi-proteins, because it was insensitive to PTX pretreatment (Fig. 5A). IBMX treatment (not shown) and forskolin addition (Fig. 5B) did not modify or induce the Ca2+ transients, which speaks against a direct contribution of the cAMP pathway.
The difference in signal transduction between the two variants, 5-HT4a and 5-HT4b, showing an apparent stronger coupling of the 5-HT4aisoform to the Ca2+-pathway, is in parallel to the higher fraction of high-affinity agonist sites of the 5-HT4a variant. This prompted us to investigate the signal transduction at the direct level of G-protein activation.
[35S]GTPγS Binding.
G-protein coupling to receptors was investigated by measuring [35S]GTPγS binding to membrane preparations from cells expressing either h5-HT4 receptor isoform. Results were expressed as percentage increase in [35S]GTPγS binding over basal levels (Fig.6). The levels of stimulation were found to differ between h5-HT4a and h5-HT4b receptors, and were in the range of 50 and 100% of stimulation, respectively. To investigate potential differences in pharmacology, the same six agonists and three putative antagonists were used as described above. The pEC50 values derived from fitted concentration-response curves are presented in Table4. We were not able to detect any stimulatory effects on [35S]GTPγS binding with the two antagonists GR113808 and SDZ205,557. However, SB204070 triggered a weak response via h5-HT4b receptors. Despite the higher levels of stimulation triggered via the h5-HT4b receptor, no significant differences in the potency of the agonistic compounds were detected between the both receptor variants.
To investigate whether the difference in levels of stimulation was triggered by the involvement of the Gαi/o-proteins, we performed similar [35S]GTPγS binding experiments on membranes from PTX-pretreated cells (Fig. 6; Table 4). We observed, for all agonists used, a decrease in the maximal levels of [35S]GTPγS binding in the h5-HT4b receptor preparation treated with PTX. The percentages of inhibition induced by PTX on the h5-HT4b-receptor are presented in Table5. Therefore, the uncoupling of the receptor from the Gαi/o-proteins by PTX pretreatment results in lower levels of [35S]GTPγS binding stimulation via the h5-HT4b receptor. Interestingly, the pEC50 values were not significantly modified, although they tended to be higher on PTX-treated membranes (Table 4). In contrast, no significant modulation of [35S]GTPγS binding was observed on the h5-HT4a receptor preparation. Because the h5-HT4b cell line used for these experiments had higher receptor expression levels than the h5-HT4a clone, we investigated whether the coupling to Gαi/o-proteins was not a consequence of the high receptor levels. To this end, we performed the same experiments on the lower-expressing h5-HT4b clone (i.e., clone 4, expressing approximately one-third of receptors). We observed the same decrease in [35S]GTPγS binding efficacy after PTX pretreatment (data not shown). Taken together, these results indicate that in addition to Gαs, the h5-HT4breceptor splice variant is also coupled to Gαi/o-proteins and that this is independent of the expression levels. In contrast, the lack of sensitivity to PTX of [35S]GTPγS binding indicates that h5-HT4a receptors do not couple to Gαi/o-proteins. We also investigated the contribution of Gαi/o-proteins to the other h5-HT4breceptor-mediated signal transduction pathways.
Adenylyl Cyclase Activation after PTX Pretreatment.
AC stimulation was investigated on h5-HT4b- and, as a control, on h5-HT4a receptors on cells pretreated with PTX and on untreated cells. Because the basal levels of cAMP were increased after PTX pretreatment the results were expressed as a percentage of the maximal stimulation obtained on nontreated cells. Consistent with the findings in [35S]GTPγS binding experiments, cAMP formation induced by 5-HT or prucalopride was found to be increased by PTX pretreatment in the h5-HT4b-variant indicating a des-inhibition of the cAMP-pathway (Fig.7). The increase in maximally accumulated cAMP was 27% (±5) for 5-HT and 29% (±6) for prucalopride stimulation. In contrast, no significant modulation of the AC response by PTX treatment was found with the h5-HT4areceptor variant (Fig. 7).
Discussion
This study uses the initially cloned and widely distributed 5-HT4a and 5-HT4b receptors to investigate the pharmacological and functional properties of these receptor splice variants in stably transfected HEK 293 cells. Analysis of saturation binding performed with [3H]GR113808 revealedK D values that were consistent with previous findings on transiently transfected cells (Ansanay et al., 1996; Van den Wyngaert et al., 1997) and different brain tissues (Grossman et al., 1993; Waeber et al., 1994; Arranz et al., 1998). We confirm that C-terminal variations of h5-HT4a and h5-HT4b receptors did not modify bindingK D values, consistent with the findings on EP3 and somatostatin receptors (Tsunehisa et al., 1993; Vanetti et al., 1993) and very recently also 5-HT4 receptors (Bach et al., 2001). Detailed saturation binding performed with [3H]5-HT, revealed the presence of two affinity sites for both h5-HT4a and h5-HT4b receptors. Previous studies could only show a single affinity site for [3H]5-HT of 20 (±7) nM for the rat 5-HT4b receptor expressed in COS7 cells, although the shallow competition binding curves and partial GTPγS sensitivity indicated the existence of two affinity sites (Adham et al., 1996). The high concentrations of [3H]5-HT used in that study (5–100 nM) may have precluded the detection of the high-affinity site described in this report (K D values ranging from 1.3 to 3.2 nM).
Comparison of saturation binding B max data using [3H]GR113808 and [3H]5-HT allowed us to estimate the fraction of receptors present in the agonist high-affinity state. This fraction of high-affinity receptors was not dependent on expression levels, either modified by sodium butyrate pretreatment of cells or using different monoclonal cell lines, but characteristic for a given splice variant. The h5-HT4b receptors showed a consistently higher proportion of non–G-protein-coupled or inactive receptors than the h5-HT4a splice variant, based on the assumption that the natural agonist [3H]5-HT labels only active receptors in R*- and R*G-state with high affinity. The independence of the fraction of [3H]5-HT high-affinity sites from expression levels is analogous to the equilibrium J constant calculated in the absence of ligand as recently shown by Claeysen and colleagues (2000). They noticed that in COS-7 cells transfected with human or mouse 5-HT4a receptors, the expression levels had no effect on the equilibrium J constant.
The pharmacological profile obtained for cAMP formation was partially divergent compared with previously reported results. We have shown that benzamide-like compounds in our system were full agonists on both variants, in contrast to reports classifying renzapride and cisapride as partial agonists (Ouadid et al., 1992; Blondel et al., 1998). Moreover, SB204070 and SDZ205,557, described as antagonists, were found to be partial agonists on both h5-HT4a and h5-HT4b receptors. In addition, small but statistically significant differences in efficacy and/or potency of agonists were found between the two splice variants. Direct extrapolation of these results to tissue-based studies should bear in mind that functional responses found in tissues may reflect not only different expression levels but also be the result of the stimulation of several 5-HT4 receptor splice variants expressed in a given tissue.
A very recent study of Bach et al. (2001) investigating the cAMP response in h5-HT4a and h5-HT4b receptors could also not match the pharmacological pattern found on cells to measurement on right atrium membranes, although the 5-HT4a isoform is thought to be predominant in that tissue (Blondel et al., 1997). In contrast to our study, they have found partial agonism of renzapride and cisapride. However, their pharmacological comparison is focused on antagonists, and the study uses [α-32P]ATP-loaded membranes for measurements of the cAMP, whereas our study is performed on intact cells. At similar expression levels, our EC50 for 5-HT is approximately 10-fold lower than the ones reported by Bach et al. (2001). Therefore, the membrane based assay used by Bach et al., which allows a direct comparison to measurements performed on tissue samples, may suffer from impaired coupling and may not be able to reveal the subtle differences seen in this study on intact cells.
In addition to AC activation, 5-HT4 receptors have also been shown to trigger other signal transduction pathways, such as reduction of potassium currents in mouse colliculi neurons (Fagni et al., 1992), calcium influx in adrenocortical cells (Contesse et al., 1996), and atrial myocytes (Ouadid et al., 1992). Our results revealed a clear functional difference between the two h5-HT4 receptor isoforms in their ability and pharmacology to increase [Ca2+]i. The h5-HT4a receptor was able to trigger strong [Ca2+]i increase, in contrast to h5-HT4b receptors, which responded with a weaker signal and only to indoleamines. This signal transduction pathway was dependent on receptor expression levels, however, also detected the lowest expression level of the h5-HT4a receptor [i.e., clone 1 expressing 1500 (±160) fmol/mg]. Interestingly, the calcium response was not only variant-specific but also dependent on compound structure, because benzamide-like compounds triggered higher responses than indoleamines.
The absence of IP pathway activation, the elimination of the response after removal of extracellular calcium, and the inhibition of the response by bepridil suggest that h5-HT4areceptors were able to activate calcium channels. Nevertheless, the question of the cascade events triggering the calcium channel activation has yet to be assessed. The phosphodiesterase inhibitor IBMX and forskolin, a direct AC activator did not enhance the increase of [Ca2+]i (data not shown), suggesting that, in contrast to the h5-HT4-induced [Ca2+]i response in adrenocortical cells (Contesse et al., 1996), the h5-HT4a calcium response in HEK 293 cells was not dependent on the cAMP pathway. Moreover, PTX treatment did not modify the calcium response, which excluded the hypothesis of Gαi/o-protein involvement in the mechanism. However, because it has been shown that after activation of G-protein heterotrimers, the βγ-subunits are able to activate calcium channels (for review, see Holler et al., 1999), such a pathway has to be considered for the 5-HT4a receptor. In addition, opening of Ca2+-channels subsequent to a βγ-subunit-mediated activation of potassium channels could be another mechanism.
In regard to the G-protein coupling to h5-HT4aand h5-HT4b receptors, we have found a difference between the level of G-protein stimulation induced by h5-HT4a and h5-HT4breceptors. PTX pretreatment induced a reduction in h5-HT4b receptor-mediated [35S]GTPγS binding to the level found for the h5-HT4a receptor, suggesting the existence of an additional Gαi/o component in the h5-HT4breceptor G-protein coupling. This was confirmed by an enhanced increase of cAMP formation induced via the h5-HT4b variant after a PTX pretreatment, which was not found for the h5-HT4a receptor. Taken together, these results indicate that the h5-HT4b receptor is able to activate both Gαi/o- and Gαs-proteins, in contrast to the h5-HT4a-receptor, which couples only to Gαs-proteins. The pEC50 values of both PTX-pretreated and nontreated cells in [35S]GTPγS binding and cAMP formation were similar, suggesting that unlike for the D2receptor, 5-HT4b receptor dual coupling was not dependent on the agonist concentration (Chang et al., 1997). Until now, dual G-protein coupling is hypothesized to have two main functions. On the one hand, it has been shown for other receptors that these were sequential events contributing to desensitization mechanisms (Lefkowitz, 1998). On the other hand, Gαs- and Gαi/o-proteins have been shown to couple simultaneously to the α2adrenergic receptor (Eason et al., 1992), the functional net result of cAMP production representing the integration of the two G-protein-mediated effects.
Taken together, these results show that the h5-HT4a and h5-HT4b-receptor variants do activate partially different intracellular processes (Fig.8, A and B). The characteristic sum of these differences may result in pharmacological differences of compounds as described in different tissues (Dumuis et al., 1988;Bockaert et al., 1997). Although possible differences in the localization of 5-HT4a and 5-HT4b receptors in brain tissue are still controversial (Gerald et al., 1995; Bender et al., 2000), saturation binding experiments performed on membranes from different human brain regions have revealed a higher fraction of agonist binding sites in the substantia nigra (0.54) than in other brain regions such as the frontal cortex (0.27), the striatum (0.25), or the hippocampus (0.16) (Bonaventure et al., 2000). These values parallel very closely the findings in the present cellular study. Although further studies on cell- and tissue-specific 5-HT4 receptor expression patterns are required; the above findings together with the reported tissue-specific differences in agonist efficacy provide a physiological context of our study. We have evaluated the influence of receptor levels on the observed responses, however, for the Ca2+-response, promiscuous coupling cannot be completely ruled out. However, we have shown that it is not cAMP-dependent in HEK cells and has different thresholds and pharmacology on both receptor isoforms. It is important to state that a direct comparison of tissue expression levels (guinea pig striatum ∼200 fmol/mg; Van den Wyngaert et al., 1997) to data derived on recombinant cells is problematic, because the physiological concentration of neurotransmitter receptors in the synaptic cleft is certainly higher than the tissue average observed in binding studies.
On the receptor level the differences in the functional profile of the investigated reference compounds in cAMP and Ca2+measurements favor different forms of R* for each receptor isoform and signaling pathway, which can be stabilized to a different degree by the compounds investigated. In general, further understanding of the fine-tuning of G-protein-coupled receptor isoform-associated signaling and its compound specificity could, in the future, lead to G-protein-coupled receptor isoform-specific (i.e., “super”) drugs with potentially broader safety margins.
Footnotes
- Received March 28, 2001.
- Accepted September 25, 2001.
-
↵1 Present address: CNS Discovery Research, Janssen Research Foundation, B-2340 Beerse, Belgium.
Abbreviations
- 5-HT
- 5-hydroxytryptamine, serotonin
- h5-HT4 receptor
- human 5-hydroxytryptamine4receptor
- AC
- adenylyl cyclase
- DMEM
- Dulbecco's modified Eagle's medium
- GTPγS
- guanosine-5′-O-(3-thio)triphosphate
- PBS
- phosphate-buffered saline
- PTX
- Bordatella pertussis toxin
- HEK
- human embryonic kidney
- FLIPR
- fluorometric imaging plate reader
- IBMX
- 3-isobutyl-1-methylxanthine
- 5-MeOT
- 5-methoxytryptamine
- IP
- inositol phosphate
- RFU
- relative fluorescence units
- SB204070
- 1,4-benzodioxin-5-carboxylic acid, 8-amino-7-chloro-2,3-dihydro-(1-butyl-4-piperidinyl)methyl ester
- GR113808
- 1H-indole-3-carboxylic acid, 1-methyl-(1-(2-((methylsulfonyl)-amino)ethyl)-4-piperidinyl)methyl ester
- SDZ205,557
- benzoic acid, 4-amino-5-chloro-2-methoxy-2-diethylamino)ethyl ester
- SDZ-HTF919 hydrazinecarboximidamide
- 2-((5-methoxy-1H-indol-3-yl)methylene)-N-pentyl-(Z)-2-butenedioate
- The American Society for Pharmacology and Experimental Therapeutics