The kinetics of drug-receptor interactions can profoundly influence in vivo and in vitro pharmacology. In vitro, the potencies of slowly associating agonists may be underestimated in assays capturing transient signaling events. When divergent receptor-mediated signaling pathways are evaluated using combinations of equilibrium and transient assays, potency differences driven by kinetics may be erroneously interpreted as biased signaling. In vivo, drugs with slow dissociation rates may display prolonged physiologic effects inconsistent with their pharmacokinetic profiles. We evaluated a panel of 5-hydroxytryptamine2B (5-HT2B) receptor agonists in kinetic radioligand binding assays and in transient, calcium flux assays, and inositol phosphate accumulation assays; two functional readouts emanating from Gαq-mediated activation of phospholipase C. In binding studies, ergot derivatives demonstrated slow receptor association and dissociation rates, resulting in significantly reduced potency in calcium assays relative to inositol phosphate accumulation assays. Ergot potencies for activation of extracellular signal-regulated kinases 1 and 2 were also highly time-dependent. A number of ergots produced wash-resistant 5-HT2B signaling that persisted for many hours without appreciable loss of potency, which was not explained simply by slow receptor-dissociation kinetics. Mechanistic studies indicated that persistent signaling originated from internalized or sequestered receptors. This study provides a mechanistic basis for the long durations of action in vivo and wash-resistant effects in ex vivo tissue models often observed for ergots. The 5-HT2B agonist activity of a number of ergot-derived therapeutics has been implicated in development of cardiac valvulopathy in man. The novel, sustained nature of ergot signaling reported here may represent an additional mechanism contributing to the valvulopathic potential of these compounds.
The 5-hydroxytryptamine2B (5-HT2B) receptor is broadly expressed in the cardiovascular system (Kaumann and Levy, 2006). Activation of the receptor on heart valve interstitial cells is believed to result in abnormal deposition of extracellular matrix components, resulting in thickening of valve leaflets, ultimately producing valvular insufficiency or valvulopathy (Fitzgerald et al., 2000; Rothman et al., 2000; Huang et al., 2009; Rothman and Baumann, 2009; Elangbam, 2010; Hutcheson et al., 2011). The appetite suppressant fenfluramine, perhaps the most notorious valvulopathogen, was withdrawn from the US market in 1997. Norfenfluramine, the primary fenfluramine metabolite, is a potent 5-HT2B receptor agonist. Benfluorex, a related drug sharing norfenfluramine as a metabolite, was withdrawn in Europe in 2009. A number of ergot-derived drugs have also been associated with increased incidence of cardiac valvulopathy. Both ergotamine and dihydroergotamine have been associated with case reports of fibrotic responses, including valvulopathy (Elangbam, 2010). More recently, pergolide, an ergot-derived dopaminergic agonist prescribed to alleviate symptoms of Parkinson's disease, was associated with the development of cardiac valvulopathy (Zanettini et al., 2007; Hutcheson et al., 2011), leading to its withdrawal from the U.S. market in 2007. Similarly, cabergoline (CAB) has been implicated in the development of a smaller number of cases of valvular insufficiency, although these findings are still subject to some debate and may be more prevalent in patients taking high doses (Lafeber et al., 2010; Tan et al., 2010; Vallette et al., 2010; Gu et al., 2011; Steffensen et al., 2012).
Ergot derivatives typically exhibit complex pharmacology deriving from agonist and antagonist interactions with numerous monoamine receptors, including serotonin, dopamine, and α-adrenergic receptors (Millan et al., 2002). In vivo, many ergot derivatives have extraordinarily long durations of action that are not easily attributed to circulating levels of parent drug (de Marées et al., 1986; de Hoon et al., 2001). For some derivatives this may be partially explained by the presence of circulating active metabolites with extended plasma half-lives. However, a number of ex vivo tissue studies suggest that ergots may be tightly bound to their receptor targets in a wash-resistant manner. Ergotamine and dihydroergotamine induce slow-onset, prolonged contractions of human coronary and superficial temporal arterial preparations that are resistant to compound washout (Ostergaard et al., 1981; MaassenVanDenBrink et al., 1998). Similarly, dihydroergotamine-induced contractions of canine saphenous and femoral vein strips (Muller-Schweinitzer, 1980) and ergovaline-induced contraction of rat and guinea pig arterial preparations (Schoning et al., 2001) are prolonged and resistant to washout. Despite these reports, there are few published in vitro studies documenting the kinetics of ergot-receptor interactions. The dissociation of [3H]dihydroergocryptine from dopamine D2 receptors in bovine anterior pituitary membranes has been reported to be 20 times slower than observed for the nonergot antagonist [3H]spiroperidol (Sibley and Creese, 1983). However, there appear to be no literature reports characterizing the kinetics of ergot interactions with other receptors.
Accurate measurement of functional potencies and efficacies for compounds with 5-HT2B receptor affinity is critical in assessing their valvulopathic potential (Huang, 2009). We investigated human 5-HT2B receptor interactions for a panel of ergot and nonergot drugs and reference agonists along with some previously uncharacterized metabolites. These studies reveal unusually slow receptor association and dissociation rates for most ergot derivatives, which result in significant underestimation of their potencies in functional assays capturing transient signaling events immediately following receptor stimulation. Thus, careful assay selection is critical in assessing the valvulopathic potential of therapeutics with 5-HT2B agonist activity. Furthermore, for a number of the ergots, persistent, wash-resistant receptor signaling is characterized, and these signals appear to emanate from internalized or sequestered receptors. This novel, persistent mode of signaling may represent an additional mechanism contributing to the valvulopathic potential of some compounds.
Materials and Methods
Data in this study were obtained in a clonal human embryonic kidney HEK293 5-HT2B receptor–expressing stable cell line or membranes derived from these cells. The 5-HT2B receptor density in the cell line ranged from 50–300 fmol/mg. Test compounds were purchased from Sigma-Aldrich (St. Louis, MO) or Tocris Bioscience (Bristol, UK). Single-use 20 mM aliquots in dimethylsulfoxide were stored at −20°C. Enzymatic synthesis of 8'-OH-dihydroergotamine (8'-OH-DHE) from DHE was performed according to published methods (Maurer and Frick, 1984a; Peyronneau et al., 1994). Test incubations were conducted with dexamethasone pretreated rat and human liver microsomes. The incubation mixture contained 1 mg/ml microsomal protein, 100 mM potassium phosphate buffer containing 3 mM magnesium chloride and 1 mM EDTA (pH 7.4), and 0.5 mM DHE. After preincubation for 3 minutes at 37°C, the reaction was initiated by the addition of 1 mM β-NADPH. An aliquot of reaction mixture was taken out at 0-, 5-, 10-, 30-, and 60-minute time points and added to an equal volume of methanol to stop the reaction. After centrifugation at 3000g for 20 minutes, the supernatant was analyzed by liquid chromatography/mass spectrometry. Formation of 8'-OH-DHE was linear up to 30 minutes in both rat and human liver microsomes but was more efficient in rat liver microsomes.
Milligram quantities of 8'-OH-DHE were generated using a larger incubation (17.49 mg of DHE/60 ml) with dexamethasone pretreated rat liver microsomes for 30 minutes under optimized conditions. Samples were diluted 1:1 with methanol (MeOH) and microsomal proteins separated by centrifugation. The MeOH solution was reduced to half-volume and diluted with excess dichloromethane (DCM). Saturated aqueous sodium carbonate (20 ml) was added, and the mixture was briefly shaken. The DCM layer was removed and the aqueous layer was extracted one additional time with DCM. The combined extracts were washed with saturated aqueous sodium bicarbonate and brine and then dried over sodium sulfate, filtered, and concentrated. 8'-OH-DHE was isolated from the concentrate utilizing thin-layer chromatography (Rf = 0.487, Analtech 1000 Micron Silica Gel GF Preparative Layer UV254 Plates, 10% MeOH in DCM) to afford 2.5 mg as a 29:71 mixture of 8'α- and 8'β-OH-DHE isomers.
HEK293 cells stably expressing the 5-HT2B receptor were treated with phenoxybenzamine (PBZ) to alkylate cell-surface receptors. Growth medium was removed and the cells were washed with phosphate-buffered saline (PBS). After removal of PBS, Opti-MEM (Invitrogen, Carlsbad, CA) containing PBZ (Sigma-Aldrich Cat no. B-019, 0–8 μM final concentration) was added and cells were incubated at 37°C for 1 hour. Cells were then washed with Opti-MEM and used for functional assays and/or membrane preparation.
Radioligand Binding Assays.
Binding assays were performed following standard procedures with [125I](±)-2,5-dimethoxy-4-iodoamphetamine (DOI) as radioligand. Competition experiments used 5-HT2B-expressing HEK293 cell membranes (15–25 µg/well membrane protein) and radioligand at final assay concentrations of 0.4–0.6 nM. Assay incubations were terminated by rapid filtration through PerkinElmer GF/C plates (PerkinElmer Life and Analytical Sciences, Boston, MA) followed by a 3× wash with ice-cold assay buffer. Plates were dried at 45°C and BetaScint scintillation cocktail (25 µl/well) was added prior to counting on a Packard TopCount scintillation counter.
The kinetics of [125I]DOI association/dissociation at the 5-HT2B receptor were determined using standard methods. In brief, receptor-expressing membranes were incubated with [125I]DOI (0.6 nM) and the equilibration process was terminated by rapid filtration at varying time points. Receptor dissociation was measured using membranes exposed to radioligand for 2 hours followed by addition of a large excess of unlabeled DOI. Data were analyzed by nonlinear regression in GraphPad Prism (GraphPad Software, Inc., San Diego, CA). Association and dissociation rates for DOI using this method were 3 × 108 M−1 min−1 and 0.196 minute−1, respectively.
Kinetic binding assays were performed by simultaneous addition of [125I]DOI (0.57 nM) and multiple, fixed concentrations of test compounds to 5-HT2B receptor–expressing membrane preparations. Assay incubations were terminated at times ranging from 1 to 150 minutes, with an emphasis on earlier time points, by rapid filtration. Binding of radioligand was quantitated as described above. Data were analyzed according to the method of Motulsky and Mahan (1984) in GraphPad Prism 5 (GraphPad Software, La Jolla, CA) using the manufacturer’s supplied algorithm for kinetics of competitive binding. To calculate association and dissociation rates for nonlabeled test compounds, the concentration of [125I]DOI and its association/dissociation rates were fixed to the values determined above.
IP Accumulation Assays.
5-HT2B-expressing cells were added to sterile poly(d-lysine)-coated 96-well microtiter plates (35,000 cells/well) and labeled with 0.6 µCi/well [3H]inositol in myoinositol-free Dulbecco’s modified Eagle’s medium for 18 hours. Unincorporated [3H]inositol was removed by aspiration and media replaced with fresh myoinositol-free Dulbecco’s modified Eagle’s medium containing pargyline (10 µM). Test compound incubations were conducted at 37°C. Inositol phosphate accumulation was captured by addition of lithium chloride (10 mM) to the assay medium. The timing of lithium addition varied depending on the goal of each assay. Assays were terminated by lysing cells with ice-cold 0.1 M formic acid followed by freezing at −80°C. After thawing, total [3H]inositol phosphates were resolved from [3H]inositol using AG1 X8 ion-exchange resin (Bio-Rad, Hercules, CA). Eluted [3H]inositol phosphates were dried down in 96-well plates at 50°C overnight and then quantitated using BetaScint (35 μl/well) on a PerkinElmer TopCount scintillation counter (PerkinElmer Life and Analytical Sciences).
Cells were plated in 25 μl of assay buffer consisting of 1× Hanks’ balanced salt solution (with calcium and magnesium) containing 20 mM HEPES at pH 7.4 at a density of 30,000 cells/well in standard tissue culture grade, clear-bottom, black 384-well plates. Molecular Devices Calcium 4 dye (Molecular Devices, Sunnyvale, CA) was used per the manufacturer’s instructions and was loaded into cells for 1 hour at 37°C. Test compounds (in 25 µl assay buffer) were then added to assay plates using a Molecular Devices FLIPR instrument. Plates were read every 2 seconds for 1.5 minutes and the peak height was determined for each well.
Extracellular Signal-Regulated Kinase Phosphorylation Assays.
Extracellular signal-regulated kinase 1 and 2 (ERK1/2) phosphorylation studies used Cellul'erk Phospho-Erk Homogeneous Time-Resolved Fluorescence Detection Kit (Cisbio US, Bedford, MA). Cells were plated in 96-well plates at 45,000 cells/well and placed in serum-free media the night before the assay. The following day, the serum-free medium was replaced with 80 µl of fresh medium (DMEM containing 100 mM sodium pyruvate and 100 mM GlutaMax). Serially diluted test compounds (20 µl) were added and the plates incubated at 37°C. Assays were stopped by addition of lysis buffer included in the kit, supplemented with 1 mM phenylmethylsulfonyl fluoride and a protease inhibitor cocktail. Cell lysates were analyzed following the manufacturer’s instructions and read on EnVision Multilabel Plate Reader (PerkinElmer Life and Analytical Sciences).
β-Arrestin Recruitment Assays.
5-HT2B receptor arrestin recruitment assays were performed using the PathHunter arrestin assay platform from DiscoveRx (Fremont, CA) using a 5-HT2B receptor stable cell line generated in PathHunter HEK293 parental cells. Arrestin recruitment assays were performed per the manufacturer’s instructions. In brief, cells stably expressing recombinant receptor were seeded into 384-well microtiter plates (8000 cells per well/20 μl) in Opti-MEM and incubated overnight in a humidified incubator. Plates were removed from the incubator and equilibrated to room temperature for 1 hour. Serially diluted test compounds (5 μl) were added to the plates, which were then incubated at room temperature for 3 hours. Lysis/detection reagents (12 μl total) were added and plates sealed and incubated for an additional 2 hours at room temperature. Plates were read on an EnVision (PerkinElmer Life and Analytical Sciences) or PheraStar (BMG Labtech, Durham, NC) plate reader.
Detection of 5-HT2B Internalization.
Receptor internalization was characterized using standard methods (Richman et al., 2007; Calebiro et al., 2009). In brief, Chinese hamster ovary cells were transiently transfected with NH2-terminal HA epitope-tagged 5-HT2B and plated on glass coverslips in 24-well plates (1 × 105 cells/well). Receptor internalization was assayed at 48-hours post-transfection. Cells were washed three times with PBS and incubated on ice for 10 minutes with Alexa Fluor 488-conjugated anti-HA-antibody 16B12 (Invitrogen) at 5 μg/ml and Alexa Fluor 694-labeled transferrin (Invitrogen) at 300 μg/ml in Ham’s F12 media supplemented with 0.1% bovine serum albumin. Compounds at 1 μM were added to the cells and incubated at 37°C for 1 hour. Cells were washed three times with PBS and fixed with formaldehyde. Coverslips were mounted on slides for visualization.
Dose-response curves were fit to a nonlinear least squares curve-fitting program in GraphPad Prism 5 (GraphPad Software, La Jolla, CA) to obtain EC50 and IC50 values. Binding Ki values were determined from IC50 values using the Cheng-Prusoff equation and the radioligand Kd value for the 5-HT2B receptor of 5.15 nM. Mean EC50/IC50/Ki values were calculated from the mean of the respective log values.
The potencies and efficacies of a panel of 21 common serotonergic agonists were determined in both calcium and inositol phosphate (IP) accumulation assays using a clonal cell line expressing the recombinant human 5-HT2B receptor at a density of 50–300 fmol/mg to avoid receptor reserve effects (Table 1). The absence of receptor reserve in this cell line was confirmed by elimination of receptor binding sites using phenoxybenzamine, which eliminated cell-surface binding sites and reduced agonist responses without changing agonist potencies or relative efficacies (Fig. 1). Agonists with less than 40 to 50% efficacy in the IP accumulation assay were typically inactive in the calcium assay, perhaps indicating that levels of IP3 produced by receptor stimulation were below a threshold required to elicit robust intracellular calcium release. The potencies of compounds that were active in both assays are compared in Fig. 2. While potencies in the two assays generally correlated, ergot derivatives clearly displayed unusually low potencies in calcium assays compared with IP accumulation assays. IP accumulation assays were performed using a 2-hour compound stimulation at 37°C and presumably allowed test compounds to equilibrate at receptor agonist binding sites. Since calcium assays capture transient signals following agonist addition and are not performed at equilibrium, we hypothesized that the potencies of ergot derivatives were perhaps underestimated in the calcium assay due to slow receptor association kinetics.
Receptor association was initially explored in [125I]DOI radioligand binding assays. 5-HT2B receptor-expressing membranes were allowed to equilibrate with radioligand and then challenged with serial dilutions of serotonin or dihydroergotamine, which displayed low potency in calcium assays relative to IP accumulation assays. Assays were terminated by rapid filtration following incubation times ranging from 4 to 256 minutes, and IC50 values were determined at each time point (Fig. 3). While serotonin IC50 values were almost identical at all measured time points, consistent with rapid equilibration with [125I]DOI at receptor binding sites, the potency of DHE was highly time-dependent, suggesting that the equilibration process for the ergot derivative was considerably slower. A larger set of agonists was then evaluated in the same assay format employing incubations of 5 minutes, 5 hours, and 20 hours (Table 2). For the typical small-molecule 5-HT2B agonists, serotonin, DOI, nordexfenfluramine, and Ro 60-0175, resulting Ki values were not appreciably time-dependent. However, with the exception of pergolide, ergot Ki values at the 5-minute time point were significantly different from values determined after a 5-hour equilibration. In general, Ki values for ergot derivatives at the 5-hour and 20-hour time points did not differ, with the exception of ergotamine and DHE.
Since intracellular calcium flux assays capture transient signals that typically peak within 1 minute following agonist addition, the time-dependence of agonist potencies could not be directly studied in this system. However, the affinities of agonists with slow receptor association rates may be estimated in pseudo-antagonist assays where a dose response of agonist is followed at a later time point by a fixed concentration of a full agonist. The panel of agonists evaluated in binding assays was therefore evaluated in 5-HT2B calcium assays using this method. Thus serially diluted agonists were applied to 5-HT2B-expressing cells, which were subsequently stimulated with serotonin (10 nM) following compound incubations ranging from 1 minute to 1 hour, and IC50 values were determined at each time point. As predicted by the radioligand binding studies, IC50 values for serotonin and typical small-molecule agonists were not appreciably time-dependent in this assay (Fig. 4A) while, with the exception of pergolide, ergot derivatives produced time-dependent IC50 values (Fig. 4B).
Since the mitogenic effects of 5-HT2B agonists are often evaluated in ERK1/2 phosphorylation assays, the impact of receptor association kinetics was also investigated in these assays. Stimulation of human 5-HT2B-expressing HEK293 cells with 100 nM serotonin, an approximately EC90 concentration in both IP accumulation and ERK1/2 phosphorylation, produced a rapid and transient elevation of pERK1/2, which peaked at approximately 3 minutes and rapidly diminished. Stimulation with 100 nM DHE produced a response with much slower onset and decay compared with serotonin, which directly reflects its slower receptor association rate (Fig. 5A). As expected based upon binding kinetics, the potency of serotonin in this assay was not appreciably time-dependent (Fig. 5, B and C), while the potency of DHE increased continually and had not stabilized by the end of the experiment (Fig. 5, B and D).
The IP accumulation data in Table 1 were generated using 2-hour agonist incubations at 37°C. Since nonergot compounds did not display time-dependent binding affinities or IP accumulation potencies in assays that used longer incubation times, these data likely reflect the true functional activities of the compounds at or close to equilibrium with the 5-HT2B receptor. However, having established that some ergot derivatives may have extremely slow receptor association rates, we investigated whether ergots would display even greater potencies if assay incubation times were further extended. In general, incubation times greater than 2 hours in the IP accumulation assay did not yield appreciably greater potencies. However, consistent with binding data, the potencies of ergotamine and DHE, which appeared to have the slowest receptor-association rates, increased from 44 and 7.6 nM to 6 and 0.4 nM (pEC50 7.16 and 8.32 increased to 8.22 and 9.40) in assays using 2- and 20-hour compound incubation times, respectively. Having established that the 2-hour IP accumulation assay was more appropriate for determining 5-HT2B agonist potencies and efficacies at or closely approaching equilibrium, we sought to characterize an extended panel of compounds (Table 3). This panel focused on drugs and their metabolites and included a number of therapeutics recently identified as modestly active 5-HT2B agonists (Huang et al., 2009).
Since most ergot derivatives displayed slow receptor association rates, we also evaluated their dissociation rates. A small panel of ergot derivatives was evaluated in kinetic binding assays following the method of Motulsky and Mahan (Motulsky and Mahan, 1984; Dowling and Charlton, 2006). This method uses simultaneous addition of fixed doses of radioligand and test compounds to receptor-expressing membranes followed by measurements of specific radioligand binding at numerous time points. In these experiments serotonin produced a typical profile with binding rapidly increasing to a plateau at equilibrium (Fig. 6A). In contrast, the profile for cabergoline (Fig. 6B) and other ergots showed the characteristic overshoot in binding levels at early time points typically observed for ligands with slow receptor association/dissociation kinetics. Receptor on/off-rates calculated from these data (Table 4) clearly demonstrated significantly slower association and dissociation rates for ergot derivatives relative to serotonin.
To explore the functional consequences of slow receptor dissociation kinetics for the ergot derivatives, we used IP accumulation assays. In the absence of lithium ions, inositol phosphates resulting from receptor activation do not accumulate to a measurable degree in these assays due to rapid degradation. As a result, the timing of lithium addition can be varied to capture IP accumulation occurring during specific time windows following or during compound exposures. Serially diluted serotonin, cabergoline, and DHE were incubated on three sets of [3H]myoinositol-loaded 5-HT2B cells for 2 hours in the absence of lithium. An additional set of cells plated in lithium-containing assay buffer was also included to provide control curves. Following the 2-hour compound incubation, the control cells and one set of cells incubated without lithium were assayed for [3H]IP accumulation. As expected, no IP accumulation was detected in cells incubated with agonists in the absence of lithium, while control cells produced normal agonist dose responses. The two remaining sets of assay plates were repeatedly washed to remove excess agonists and allowed to equilibrate for an additional 2 or 4 hours, at which point the cells were again extensively washed before addition of fresh assay buffer containing lithium to begin capture of any subsequently formed inositol phosphates. After a 2-hour incubation in the presence of lithium, cells were assayed for [3H]IP accumulation. While the serotonin response was almost completely eliminated following the washing protocol, dose-dependent, wash-resistant IP accumulation responses were observed for cabergoline and DHE in both sets of washed cells (Fig. 7, A–C). In further experiments, an additional five ergot derivatives were evaluated (Fig. 8). Remarkably, despite extensive compound washout, the potencies of a number of agonists, including cabergoline, ergotamine, and DHE, matched those determined in the control, unwashed cells. For other ergots, including pergolide, 8-hydroxy-DHE, and des-carbamoyl-cabergoline, potencies after washing were within 10-fold of control values, while ergonovine and methylergonovine appeared to lack wash-resistance. Persistent signaling was not observed for any nonergot agonists. This experiment was also repeated for serotonin, cabergoline, and DHE, using an overnight incubation between the first and second wash steps. Although the assay signal was significantly reduced, perhaps because of a depleted [3H]myoinositol load in the cells, clear dose responses were observed with potencies again similar to those observed in control cells (Fig. 7D).
A series of mechanistic studies was performed to investigate the origin of persistent signaling. Since it produced one of the more potent persistent signals, these experiments primarily used cabergoline. In membrane radioligand binding assays, the half-life for cabergoline dissociation from the 5-HT2B receptor was 100 minutes (Table 4). Since the duration of CAB-mediated persistent signaling was far greater than expected from this receptor-dissociation rate, we hypothesized that in the whole-cell context the CAB-receptor complex might be sequestered in a microenvironment, either at the cell surface or in an internalized endosomal compartment that restricted the dissociation of CAB. Since PBZ is an effective tool to alkylate and eliminate signaling mediated by cell-surface 5-HT2B receptors (Fig. 1), we investigated its effects on CAB-induced signaling. As expected, in cells exposed to increasing doses of PBZ prior to application of serially diluted CAB, responses were progressively eliminated without significant reductions in CAB potency (Fig. 9A). However, the same PBZ treatments performed on washed cells previously exposed to serially diluted CAB had no effect on the persistent CAB dose response (Fig. 9B), suggesting either that persistent CAB signaling did not emanate from normal cell-surface receptors that were easily accessible to PBZ, or that CAB was bound to the 5-HT2B receptor in a noncompetitive manner preventing PBZ access to the receptor binding pocket.
We next explored the ability of antagonists to block persistent signaling. In classic antagonist experiments where serially diluted antagonists were applied to cells prior to stimulation with fixed-dose CAB in the presence of lithium, antagonist IC50 values were consistent with their published binding affinities, suggesting normal competitive interactions between CAB and antagonists at the cell-surface 5-HT2B receptor (Fig. 10, open circles). Next, cells were exposed to 10 nM CAB for 2 hours in the absence of lithium to establish persistent signaling. The cells were washed, allowed to equilibrate for 1 hour, and then rewashed to ensure that all traces of exogenous CAB were removed from the assay system. Serially diluted antagonists were then applied, along with lithium chloride to initiate inositol phosphate capture. Following a 2-hour incubation, cells were assayed for [3H]IP accumulation. The antagonists blocked persistent CAB signaling with potencies similar to those observed in the classic antagonist experiments, suggesting they were able to access the persistently signaling receptor and interact in a competitive manner to displace CAB (Fig. 10, closed circles). In a further modification we explored signaling after the antagonists were themselves washed from the assay system. Thus persistent signaling was again established in cells using 10 nM CAB in the absence of lithium. After washing CAB from the system, serially diluted antagonists were applied, this time in the absence of lithium, and again allowed to incubate with the cells for 2 hours. Antagonists were then extensively washed from the system and lithium added to initiate inositol phosphate capture. Since antagonists had been shown to interact with the persistently signaling receptor in a competitive manner, our expectation was that, due to the absence of an exogenous pool of CAB during the antagonist incubation, CAB displaced from the receptor would diffuse into the surrounding medium, effectively experiencing infinite dilution, and thus be unable to rebind to the receptor when it and the antagonists were subsequently washed from the system. As expected, antagonist effects on persistent CAB signaling were still evident following this procedure (Fig. 10, open squares). However, with the exception of ritanserin, which has been shown to be an irreversible antagonist at 5-HT2 receptors (Leysen et al., 1985), apparent antagonist potencies following washout were reduced, suggesting that some rebinding of displaced CAB occurred as antagonists dissociated from the receptor.
In a series of experiments analogous to those used to characterize antagonist effects, we also investigated the ability of serotonin to modulate the activity of persistently signaling 5-HT2B receptors. In contrast to the antagonists, which tended to be lipophilic and therefore more likely to be membrane permeable, serotonin is a smaller, more polar molecule that is less likely to permeate cell membranes. To determine whether serotonin could compete with CAB at the cell-surface 5-HT2B receptor, serially diluted CAB was added, with or without serotonin (200 nM), to 5-HT2B expressing cells in the absence of lithium. After a 2-hour incubation followed by extensive washing, recovery, and further washing, lithium was added to the assay for 2 hours before assaying for IP accumulation. The simultaneous addition of serotonin in the initial incubation medium resulted in a 20-fold reduction in potency for the resulting persistent CAB signal relative to control cells (Fig. 11), consistent with competition of serotonin and CAB at the cell-surface receptor. Next, cells in which persistent signaling was established by exposure to serially diluted CAB for 2 hours, followed by extensive washing, recovery, and further washing, were exposed to a fixed concentration of serotonin (500 nM) for 2 hours to allow serotonin to compete with CAB bound to persistently signaling receptors. Serotonin was then washed from the assay system and lithium chloride added for 2 hours before assaying for IP accumulation. In contrast to the antagonists, serotonin did not affect the potency of pre-established CAB persistent signaling, suggesting it was unable to access persistently signaling receptors (Fig. 11). To explore the ability of even higher concentrations of serotonin to disrupt persistent CAB signaling, cells exposed to CAB (10 nM) and extensively washed were exposed to serially diluted serotonin with a starting concentration of 10 μM for 2 hours. After extensive washing to remove excess serotonin and a 2-hour incubation in lithium-containing medium to capture inositol phosphates, no reduction in the magnitude of the CAB signal was observed at any serotonin treatment concentration (Fig. 11), confirming that serotonin was unable to disrupt persistent CAB signaling even at very high concentrations.
To provide further insights into the interaction of CAB with the 5-HT2B receptor, we investigated agonist-induced β-arrestin recruitment by the receptor and monitored receptor internalization using fluorescence microscopy. Arrestin recruitment was evaluated using the PathHunter assay (DiscoveRx, Fremont, CA). The potencies and efficacies of arrestin recruitment in response to a number of agonists were determined in PathHunter HEK293 cells stably expressing the human 5-HT2B receptor using 3-hour compound incubations. Agonist potencies in the PathHunter assay generally matched receptor binding affinities, while efficacies were somewhat greater than observed in IP accumulation assays. Thus, 1-(3-chlorophenyl)piperazine retained partial agonist activity in the arrestin assay while DHE appeared to behave as a full agonist (Fig. 12). Thus, CAB and ergot agonists with very slow receptor-association rates (DHE) are able to induce β-arrestin recruitment to the 5-HT2B receptor in a normal manner. Since cells used in functional studies expressed an untagged 5-HT2B receptor, we employed Chinese hamster ovary cells transiently transfected with HA-tagged receptor to determine whether 5-HT2B undergoes internalization upon activation with CAB and/or 5-HT. Receptor internalization was detected by monitoring redistribution of fluorescently tagged anti-HA antibody prebound to the receptor upon treatment with agonists. Interestingly, a significant level of constitutive 5-HT2B internalization was detected in vehicle-treated cells as judged by the appearance of punctate intracellular receptor staining colocalized with endosomal compartments visualized by internalized fluorescently labeled transferrin (Fig. 13). Both CAB and 5-HT further promoted 5-HT2B internalization, as a majority of the detectable receptor staining appeared in transferrin-labeled endocytic compartments upon treatment with agonists.
An unusually poor correlation between agonist potencies in 5-HT2B IP accumulation and calcium assays for ergot derivatives prompted us to hypothesize that ergots might have unusually slow receptor-association kinetics. Reduced on-rates were readily detected by radioligand binding, where, in contrast to serotonin and typical small-molecule agonists, most ergots required a number of hours to equilibrate with radioligand. The effects of slower receptor-association kinetics were also apparent in pERK1/2 assays, in which ergots produced time-dependent potencies that did not stabilize during the 30-minute window in which ERK phosphorylation was detectable. Additionally, in pseudo-antagonist calcium assays, time-dependent IC50 values for ergots were readily apparent. To our knowledge, the present study is the first to characterize the kinetics of receptor association and dissociation for ergot agonists at any recombinant receptor. The data indicate that most ergots have slow on/off kinetics at the 5-HT2B receptor, resulting in significant underestimates of receptor potency in widely used calcium assays.
In the present study, IP accumulation assays provided the most appropriate method for evaluation of diverse 5-HT2B agonists and also proved more sensitive in detecting partial agonists, particularly for lower-efficacy ergots. While calcium responses for partial agonists may be more easily quantified in cells expressing higher receptor densities, a key goal of the present study was to define agonist potencies and efficacies in cells devoid of receptor reserve to provide more informative pharmacological data.
Ergots also tended to have slower receptor dissociation rates, which were clearly detected in kinetic membrane radioligand binding studies. Furthermore, many of the ergots produced sustained, wash-resistant receptor signaling in IP accumulation assays with little loss of potency, consistent with literature reports of wash-resistant effects in tissue bath experiments. Critically, the duration of persistent signaling greatly exceeded what might be expected based on rates of receptor dissociation measured in membrane binding assays, suggesting that persistent signaling was not driven simply by irreversible agonist binding.
When compared with serotonin, mechanistic studies demonstrated no significant differences in the ability of CAB to recruit β-arrestin to the 5-HT2B receptor or to induce receptor internalization. Further experiments suggested that persistently signaling 5-HT2B receptors were either internalized or sequestered in a microenvironment not always accessible to other ligands. PBZ, an efficient alkylator of cell-surface 5-HT2B receptors, was unable to disrupt established, persistent signaling, indicating that persistently activated receptors were either not readily accessible at the cell surface or that CAB bound to the cell-surface receptor in a noncompetitive manner that precluded PBZ binding. The latter possibility is difficult to reconcile with the observation that numerous, structurally diverse antagonists were able to compete with CAB at the persistently activated receptor. While relatively hydrophobic antagonists were capable of disrupting established, persistent signaling with potencies similar to those observed in traditional antagonist assays, serotonin, a more polar molecule that is less likely to cross cell membranes, had no effect.
A significant number of G-protein-coupled receptor antagonists are known to act in an insurmountable manner. In contrast, there are far fewer reports documenting persistent agonist binding and signaling (Calebiro et al., 2009; Mullershausen et al., 2009; Neumann et al., 2010; Boutin et al., 2011; Werthmann et al., 2012). In the present study, the persistent signaling for ergots such as CAB appears to be established in an essentially quantitative manner, since measured potencies did not decrease following extensive compound washout. Since persistent signaling was associated with agonists having slower receptor dissociation rates, it is possible that these agonists are more efficiently internalized with the activated receptor and may not dissociate appropriately in endosomal sorting compartments thus trapping signaling receptors. The mechanisms underlying both constitutive and agonist induced 5-HT2B internalization, the postendocytic trafficking of the receptor and the dependence of these processes on the nature of the agonist will be the subjects of further investigation. Additionally, it is unclear whether the components of the signaling machinery are internalized together with the CAB-receptor complex, or are recruited de novo to endosomes similar to recent reports for the β2-adrenergic receptor (Irannejad et al., 2013).
Since IP accumulation proved to be a more appropriate method for measuring the 5-HT2B agonist potencies of ergots, we have sought to understand whether the increased potencies measured in this assay relative to calcium assays provide a clearer understanding of the valvulopathic potential of 5-HT2B agonists at normal therapeutic plasma concentrations. Table 5 summarizes a number of pharmacokinetic studies for a range of 5-HT2B active drugs with and without valvulopathic potential in humans. Typical plasma concentrations of drugs or relevant 5-HT2B-active metabolites, corrected for plasma protein binding, are presented along with their potencies measured in the 5-HT2B IP accumulation assay. For valvulopathic drugs, free plasma concentrations are very close to or in excess of measured 5-HT2B receptor potencies. In retrospective studies, fenfluramine was associated with risk ratios for development of aortic and combined aortic/mitral regurgitation of 2.32 (95% confidence interval 1.79–3.01) (Loke et al., 2002) and 2.2 (1.7–2.7), respectively (Sachdev et al., 2002). In clinical use, both peak and trough plasma concentrations of the 5-HT2B active metabolite norfenfluramine exceeded its potency at the receptor, suggesting valvulopathy driven by continuous 5-HT2B receptor activation. Although benfluorex (relative risk 2.97 [1.91–4.63]) (Derumeaux et al., 2012), is metabolized less efficiently than fenfluramine, peak norfenfluramine plasma levels were almost 7-fold higher than its 5-HT2B EC50. In contrast, in human clinical studies of lorcaserin, the unbound plasma concentrations at Tmax and Tmin were approximately 80 and 40 nM, respectively, almost 30- and 60-fold below lorcaserin’s 5-HT2B receptor potency (2.38 μM), and the associated relative risk, measured across multiple clinical studies was 1.16 (0.81–1.67) (Colman et al., 2012). Similarly, maximal plasma levels of additional drugs possessing modest 5-HT2B agonist activity that have not been associated with valvulopathy (Huang et al., 2009), (e.g., extended release guanfacine) consistently fall below measured IP accumulation potencies.
Interpretation of ergot alkaloid pharmacology is often confounded by production of numerous pharmacologically active metabolites, which may circulate at concentrations exceeding those of the parent drugs (Maurer and Frick, 1984b; Silberstein, 1997; Blin, 2003). In pharmacokinetic studies of pergolide, plasma concentrations at higher doses appear to approach the 5-HT2B receptor potency measured in the current study. However, in patients receiving a single oral dose of 14C pergolide, as a result of extensive first-pass metabolism, none of the plasma radioactivity was attributable to parent compound (Rubin et al., 1981). Thus numerous pergolide metabolites may be present in plasma at significantly greater concentrations than pergolide itself. The identification of these metabolites and their plasma concentrations remains incomplete. In the present study, sulfoxide and N-despropyl metabolites of pergolide were evaluated but showed reduced 5-HT2B activity.
Similar to pergolide, cabergoline is extensively metabolized in man (Battaglia et al., 1993). The valvulopathic potential of cabergoline is the subject of considerable debate. In a retrospective study, risk ratios for valvulopathy of 50.3 (95% confidence interval 6.6, 381.4) and 2.6 (0.5, 12.8) were reported in patients exposed to doses >3 mg/day or <3 mg/day, respectively (Schade et al., 2007). More recently, studies in hyperprolactinemic patients prescribed lower doses (typically 0.5 to 1 mg dosed once or twice weekly) have reported little or no valvulopathic risk (Zanettini et al., 2007; Lafeber et al., 2010; Tan et al., 2010; Vallette et al., 2010; Steffensen et al., 2012). Pharmacokinetic studies suggest that high doses of cabergoline result in plasma levels that can exceed its IP accumulation potency at the 5-HT2B receptor while a single, 0.5-mg dose produces plasma concentrations >10-fold below its 5-HT2B potency (Table 5).
Similar analyses for methylergonovine (after dosing of methysergide) (Bredberg et al., 1986), ergotamine (Tfelt-Hansen and Paalzow, 1985; Saper and Silberstein, 2006), and DHE (Schran et al., 1979; Schran and Tse, 1985; Silberstein, 1997; Saper and Silberstein, 2006) indicate that plasma concentrations following typical dosing regimens approach or exceed those required for receptor activation. Both ergotamine and DHE are subject to extensive metabolism in humans with numerous metabolites retaining biologic activity (Muller-Schweinitzer, 1984; de Marées et al., 1986; Silberstein, 1997). 8-Hydroxy-DHE, the major circulating metabolite of DHE, which typically circulates at levels 5–7-fold higher than DHE (Saper and Silberstein, 2006), was found to be slightly more potent than DHE in the current study.
In summary, IP accumulation assays with longer compound incubations and in cells devoid of receptor reserve appear to reveal a clearer relationship between 5-HT2B potencies and plasma exposures that drive valvulopathic changes for both small-molecule and ergot agonists. Additionally, the persistent signaling of some clinically used ergots, if translated in vivo, could represent an additional mechanism by which these compounds might drive valvular interstitial cell dysfunction. Our data also highlight the need to interpret functional data obtained using kinetic versus equilibrium assays with great caution (Charlton and Vauquelin, 2010). It is noteworthy that the crystal structure of the human 5-HT2B receptor bound to ergotamine has recently been published (Wang et al., 2013) and ergotamine described as a biased agonist favoring β-arrestin signaling over G-protein activation, based on differential potencies for arrestin recruitment and intracellular calcium flux (Wacker et al., 2013). Since ergotamine has very slow receptor association kinetics, it seems possible that its differential calcium and arrestin potencies, which were measured within 1 minute and overnight exposures, respectively, may have been driven by receptor association kinetics. In the present study, likely because of the low expression of the 5-HT2B receptor used to avoid receptor reserve effects, ergotamine did not yield a measurable calcium response. However, in IP accumulation assays performed with an overnight compound incubation, ergotamine behaved as a partial agonist with efficacy of 45% relative to serotonin and a potency of 6 nM, even more potent than the 89-nM value for β-arrestin recruitment reported by Wacker et al, 2013. The reported 5-HT2B-ergotamine structure displayed conformational characteristics of both the active and inactive state, compared with the classic agonist-induced state in the structure of the 5-HT1B-ergotamine complex. These structural features are perhaps consistent with those expected from a receptor bound to a partial agonist.
Participated in research design: Unett, Gaidarov, Gatlin, Frazer, Sadeque, Anthony.
Conducted experiments: Dang, Le, Xing, Chen, Anthony, Chang, Chen, Buzard, Frazer.
Performed data analysis: Unett, Gatlin, Frazer, Anthony, Gaidarov.
Wrote or contributed to the writing of the manuscript: Unett, Gaidarov.
- Received July 1, 2012.
- Accepted September 17, 2013.
- extracellular signal-regulated kinases 1 and 2
- human embryonic kidney
- inositol phosphate
- phosphate-buffered saline
- Copyright © 2013 by The American Society for Pharmacology and Experimental Therapeutics