Introduction

Top Abstract Introduction Materials Results Discussion References

The anorectic agent dextenfluramine (d-FF), known as Redux, is a serotonin reuptake inhibitor and releasing agent. The importance of serotonin in the control of eating behavior and the mediation of satiety has long been established (Blundell, 1977, 1984; Leibowitz and Shor-Posner, 1986). It was believed that procedures that increase hypothalamic extracellular 5-HT availability or 5-HT receptor activation reduce food consumption, whereas procedures that reduce serotonin activity bring about the opposite effect. However, much of the recently accumulated data were not consistent with d-FF and other related appetite suppressants acting by boosting 5-HT activity, because the anorectic activity was not blocker by either the 5-HT neurotoxin (Grignaschi and Samanin, 1992) or by 5-HT synthesis inhibitor (Caccia et al., 1992; Gibson et al., 1993; Oluyomi et al., 1994; Lightowler et al., 1996). These discordant findings suggest the existence of possible additional mechanism(s) underlying the anorectic effect of these agents.

It has been recently reported that centrally acting K⁺ channel modulators play a regulatory role in food intake in rats (Ghelardini et al., 1997). The study demonstrated that intracerebroventricular administration of K⁺ channel antagonists produced a decrease in food consumption and the opposite was true with K⁺ channel agonists. The K⁺ channel activity in peripheral cells was reported to be associated with macronutrient preference (Liu et al., 1997). Specifically, the delayed rectifier K⁺ (DRK) channels in rat lingual taste cells was found to be inhibited by polyunsaturated fatty acids, and the degree of susceptibility to fatty acids was inversely correlated with fat preference in rats. Consistent with the potential regulatory role of K⁺ channels in feeding behavior, weight gain has been reported in humans in the clinical trial of orally administered pinacidil, a known K⁺ channel opener (Friedel and Brogden, 1990). Collectively, these results suggested an active role for K⁺ channel modulation in the determination of food preference and consumption.

The present study, using the patch-clamp technique, was designed to evaluate the response of the voltage-gated DRK channels in taste cells from fungiform papillae of rat tongues to a number of anorectic agents. Test substances were either 5-HT reuptake inhibitors and releasing agents like d-FF, fenfluramine (FF) and sibutramine (SB) or 5-HT_2C receptor agonist like m-chlorophenylpiperazine (mCPP). The latter was chosen because d-FF appeared to exert an anorectic effect by directly acting on 5-HT_2C receptor (Gibson et al., 1993). Given that the DRK current in taste cells demonstrates physiological and pharmacological characteristics typical of the DRK current in cardiac myocytes, the effect of d-FF was also examined in rat ventricular myocytes. The results revealed a common inhibitory effect on DRK currents from these two distinct types of cells. We propose that the effect may be a contributing mechanism to the beneficial anorectic as well as the recently documented adverse cardiaovascular actions of d-FF.

	Methods and Materials

Top Abstract Introduction Materials Results Discussion References

Isolation of rat taste buds. Individual taste buds were isolated from the tongues of Sprague-Dawley rats (250 to 300 g) with the procedure described earlier (Bene et al., 1990). The rats were killed by CO₂ inhalation. The tongues were cut at the extreme posterior region where vallate papillae are located and immediately placed in ice-cold Tyrode's solution. To remove the lingual epithelium in which the taste buds embedded from the underlying muscle layer, Tyrode's solution containing 0.5 mg/ml collagenase A (Boehringer-Mannheim), 5 mg/ml dispase (grade II, Boehringer-Mannheim) and 1 mg/ml trypsin inhibitor (type I-S, Sigma) was injected subcutaneously into the ventral and dorsal sides of the fungiform papillae and around the vallate and foliate papillae. The injected tongue was incubated in oxygen bubbled Ca⁺⁺- and Mg⁺⁺-free Tyrode's for 35 min. The epithelium around the papillae was gently lifted away from the rest of the tongue, and pinned serosal side up in Ca⁺⁺- and Mg⁺⁺- free Tyrode's. In most cases, a complementary 20 min incubation at room temperature was necessary to loosen the tight junction and therefore the attachment of the buds to the papillae. A glass capillary with a firepolished tip opening of ~200 µm was introduced into the papillae and the taste buds were individually sucked into the capillary and then transferred to a recording chamber filled with Tyrode's solution.

Isolation of rat cardiac ventricular myocytes. Sprague-Dawley rats (250 to 300 g) rats were killed by i.p. injection of 1 ml of Na pentobarbital at 60 mg/ml. The chest was opened, and the heart was quickly excised and immersed in prewarmed (to 34°C) and preoxygenated Tyrode's solution. The left ventricle was dissected and cut into small strips. The muscle strips were then incubated in continuously oxygenated Tyrode's solution containing 2 mg/ml collagenase (type I, Sigma) and 4 mg/ml bovine albumin (fraction V, GIBCO) at 34°C for 1 hr. At the end of incubation, the enzyme solution was removed and muscle strips were rinsed with "KB" solution (Isenberg and Clockner, 1982) for several times. The muscle pieces were placed in a test tube half-filled with KB solution, and single ventricular myocytes were released by gently shaking the test tube.

Electrophysiological recording. K⁺ currents were recorded in rat taste receptor cells maintained in the taste buds or in single ventricular myocytes by using the whole-cell configuration of the patch-clamp technique (Hamill et al., 1981). The recording chamber consisted of a Corning 35 mm culture dish fitted with a Sylgard O-ring of ~3 mm thickness and ~10 mm inner diameter residing on top of a Corning 22-mm² cover glass, resulting in a chamber volume of ~0.3 ml. Taste buds, viewed with Nikon inverted microscope at a magnification of 600×, were attached to the cover glass after a typical 10-min settling time. The experimental chamber was continuously superfused with Tyrode's solution with and without testing drugs at a constant rate of ~1.5 ml/min. The perfusion system was built such that the test solutions would reach the chamber in <15 sec, once the microswitch was turned on. As a standard procedure, recordings were always made at 1 min after the perfusion, which was taken as time zero in all electrophysiological studies. According to the perfusion rate and the volume of the chamber, a time period of 1 min would allow a complete replacement of the previous solution by the new solution. Patch electrodes were pulled from Kimax-51 capillary tubes (Kimble Products, NJ). The resistance of electrodes after firepolishing was around 5 M for taste receptor cells and 2 M for cardiac cells. The junction potential between the electrodes and the bath solution was compensated using the DC offset in the amplifier. Series resistance were typically in the range between 3 and 6 M, of which 70% was electronically compensated. No leak subtraction was applied.

Solutions. Tyrode's solution, also used as bath solution, contains (in mM) 140 NaCl, 5 KCl, 1 CaCl₂, 1 MgCl₂, 10 HEPES, 10 glucose and 10 Na pyruvate, pH 7.4. In Ca⁺⁺- and Mg⁺⁺-free Tyrode, nominally zero CaCl₂ and MgCl₂ was supplemented with 2 mM BAPTA (Sigma). The intracellular pipette solution is composed of (in mM) 140 KCl, 0.1 CaCl₂, 2 MgCl₂, 0.6 EGTA, 3 K₂ATP, 2 Na₂UDP and 10 HEPES, in which the free Ca⁺⁺ concentration was estimated to be 10⁸ M (Imai and Takeda, 1967). The "KB" solution contains (in mM) 85 KCl, 30 K₂HPO₄, 5 MgSO₄, 5 Na₂ATP, 5 pyruvic acid, 5 creatine, 20 taurine, 5 DL--OH butyric acid and 1g/l bovine albumin.

Materials. Collagenase A and dispase were purchased from Boehringer-Mannheim (Indianapolis, IN) and bovine serum albumin was from GIBCO (Gaithersburg, MD). Fenfluramine, serotonin, trypsin inhibitor (type I-S) and all salts were obtained from Sigma (St. Louis, MO). Dexfenfluramine and sibutramine were synthesized in the Research department at Novartis Pharmaceuticals Corp. All drugs, being water soluble, were directly dissolved in Tyrode's solution to form a stock solutions of 50 mM in concentration and then diluted with Tyrode's to the desired concentrations for testing.

	Results

Top Abstract Introduction Materials Results Discussion References

The whole-cell outward K⁺ currents in taste cells were elicited by a series of voltage steps varying between 80 and +80 mV with an increment of 20 mV from a holding potential of 80 mV. The currents were voltage dependent, fast activating and slowly inactivating. We found that the currents were insensitive to dendrotoxin (up to 1 mM), iberiotoxin (up to 100 nM), charybdotoxin (up to 200 nM), apamin (up to 1 mM) and glyburide (up to 5 mM) but were inhibited by TEA, 4-AP, quinine, nifedipine, terfenadine and flecainine in a concentration-dependent manner (fig. 1). The concentrations for a half-maximal inhibition (IC₅₀) of the K⁺ currents elicited by a +60 mV depolarizing step were, respectively, 2.0 mM, 4.7 mM, 9.0 µM, 16.8 µM, 3.9 µM and 29.3 µM for the aforementioned K⁺ channel blockers (fig. 1). The electrophysiological and pharmacological properties of the K⁺ currents in taste cells were consistent with those of the voltage-dependent DRK channels (Hume, 1985; Bokvist et al., 1990; Dreyer, 1990; Lu et al., 1990; Rampe et al., 1993; Daleau et al., 1997) but not the Ca⁺⁺-activated and ATP-sensitive K⁺ channels, although the latter channels have been identified in taste cells (Fujiyama et al., 1994; Lindemann, 1996).

View larger version (24K): [in this window] [in a new window]   Fig. 1.   Pharmacology of DRK currents in taste cells. Concentration-response curves for blockade of the DRK currents by 4-AP, TEA, quinine, nifedipine, terfenadine and flecainine (from left to right, top to bottom). The amplitude of the steady-state whole-cell DRK current at +60 mV was used as an index of response. Points of each plot were the mean ± S.E.M. of 5 to 7 cells. The ordinate is the amplitude of DRK currents in the presence of drugs as fraction of the control. The abscissa denotes drug concentration (mM or µM) in a logarithmic scale. Data were fit to the logistic equation Y = 1  1/[1 + (X/a)b] with a general nonlinear, least-squares analysis using the Gauss-Newton algorithm as modified by Marquardt and Levenburg. X and Y are, respectively, the concentration of drugs and the normalized DRK currents. a is IC50, which is shown in the lower left corner of each plot, and b is the slope coefficient, that is 0.9, 0.8, 0.8, 0.8, 1.5 and 0.4, respectively for 4-AP, TEA, quinine, nifedipine, terfenadine and flecainine.

Typical responses to d-FF of the DRK currents in a taste cell are shown in figure 2A, in which the drug reduced the current amplitude in a concentration-dependent manner. The action had a rapid onset and recovery. Specifically, the response was observed typically within 1 min of d-FF application and reached a steady state in ~4 min. Reversal of the blockade upon removal of d-FF was completed between 4 and 5 min. The inhibitory effect is also reflected in the current-voltage relationship (fig. 2B), in which the blockade appears to be weakly voltage dependent, with more blockade at higher depolarized potentials.

View larger version (25K): [in this window] [in a new window]   Fig. 2.   d-FF inhibits DRK currents in rat taste cells. A, Representative whole-cell DRK currents elicited by voltage steps to 80, 40, 20, +20 and +60 mV in control, in 3, 10, 30, 100 and 300 µM d-FF and 4 min after removal of d-FF. Holding potential = 80 mV. The cell was bathed in physiological K+ solutions ([K+]o 5 mM/[K+]i 140 mM). All current were recorded in the same cell. B, Current-voltage relationship in the absence and presence of d-FF from the cell in A. Ordinate shows the amplitude of steady-state DRK currents (pA) in control (), in 3 (), 10 (), 30 (), 100 () and 300 µM d-FF (). The abscissa denotes membrane potentials (mV) at which the DRK currents were recorded.

Parallel studies were undertaken with the anorectic agents FF, SB and mCPP. These drugs caused an inhibitory effect on DRK currents in a manner that was strikingly similar to that induced by d-FF. Figure 3 displays superimposed records of the DRK current at +60 mV without and with FF at concentrations indicated in the figure. The results with SB are shown in figure 4, A and B, in which the concentration-dependent inhibition of DRK currents was readily visible. SB seemed relatively more potent than d-FF with a minimal effective concentration of 0.3 µM. As d-FF reportedly exerts the anorectic effect by directly acting on 5-HT_2C receptor (Gibson et al., 1993), mCPP, a selective 5-HT_2C receptor agonist, was also found to reversibly block the DRK currents in taste cells (data not shown).

View larger version (23K): [in this window] [in a new window]   Fig. 3.   Concentration-dependent inhibition of DRK current by FF in taste cells. Superimposed whole-cell current traces elicited by a voltage step to +60 mV from a holding potential of 80 mV in control and in the presence of FF at 3, 10, 30, 100 and 300 µM and 1 mM. The cell was bathed in physiological K+ solutions ([K+]o 5 mM/[K+]i 140 mM). Scales of 0.2-nA reference current and 100-msec time are shown at the lower left corner.

View larger version (22K): [in this window] [in a new window]   Fig. 4.   SB inhibits DRK currents in rat taste cells. A: Representative whole-cell DRK currents elicited by voltage steps to 80, 20, +20 and +80 mV in control, in SB at 0.3, 3, 10, 30 and 100 µM. Holding potential = 80 mV. The cell was bathed in physiological K+ solutions ([K+]o 5 mM/[K+]i 140 mM). All current were recorded in the same cell. Scales of 0.2-nA reference current and 100-msec time are shown at the lower left corner. B, Current-voltage relationship in the absence and presence of SB from the taste cell in A. Ordinate shows the amplitude of steady-state DRK currents (nA) in control (), in 0.3 (), 3 (), 10 (), 30 () and 100 µM SB () and abscissa denotes membrane potentials (mV) at which the DRK currents were recorded.

To determine whether d-FF, FF, SB and mCPP acted directly on the channel gating to block the DRK channels or acted indirectly through a 5-HT mediated mechanism to affected the channels, 5-HT between concentrations of 1 µM and 1 mM was directly applied to the taste cells and failed to significantly alter the DRK currents. In addition, the 5-HT synthesis inhibitor p-chlorophenylalanine (PCPA) up to 1 mM neither altered DRK currents (fig. 5B) nor blocked d-FF-induced inhibitory effect (fig. 5C). These results suggest that the blocking effect on DRK channels was independent of 5-HT availability. Another result argued against the involvement of 5-HT activation in the blockade of DRK currents is shown in figure 6, in which metergoline (MTG), a nonselective 5-HT receptor antagonist, acted as an agonist in reducing DRK currents (fig. 6B) with IC₅₀ of ~10 µM (results not shown). At submaximal concentrations its effect was additive to that of d-FF, a result implying an involvement of the 5-HT receptor in the blockade of DRK channels.

View larger version (22K): [in this window] [in a new window]   Fig. 5.   PCPA, a 5-HT synthesis inhibitor, at 1 mM alters neither DRK currents nor d-FF-induced blockade of DRK currents in rat taste cell. Whole-cell DRK currents elicited by voltage steps to 80, 20, 0, +20 and +60 mV in control (A), in 1 mM PCPA (B), 1 mM PCPA with 100 µM d-FF (C), 5 min after recovery (D), 3 min after reapplication of 100 µM d-FF (E) and 5 min after washout of all drugs (F). Holding potential = 80 mV. All currents were recorded in the same cell. Note: (a) PCPA had little effect on DRK currents (6B); (b) PCPA did not prevent d-FF blocking effect (6C); (c) Repeated application of d-FF did not cause desensitization (6C and 6E); (d) all effects were fully reversible (6D and 6F).

View larger version (11K): [in this window] [in a new window]   Fig. 6.   MTG, a nonselective 5-HT receptor antagonist, does not prevent d-FF-induced-blocking effect in taste cell. Whole-cell DRK currents triggered by voltage steps to 80, 20, 0, +20 and +60 mV in control (A), in 10 µM MTG (B) and 10 µM MTG with 100 µM d-FF (C). Holding potential was 80 mV.

Using the amplitude of DRK currents at +60 mV as an index, concentration-response curves for all drugs tested were constructed (fig. 7). The IC₅₀ values obtained from the least square fitting of the data (Fletcher and Shrager, 1973) were 8.6, 30.5, 69.0 and 95.4 µM, respectively, for SB, d-FF, FF and mCPP.

View larger version (21K): [in this window] [in a new window]   Fig. 7.   Concentration-response curves of inhibition of DRK currents. Inhibitory effect was induced by SB (, n = 6), d-FF (, n = 7), FF (, n = 5) and mCPP (, n = 5) but not by 5-HT (, n = 4). The ordinate is the amplitude of DRK currents in the presence of drugs as fraction of the control and the abscissa denotes concentrations of drugs (M) in a logarithmic scale. The amplitude of the DRK currents at depolarizing step to +60 mV was used to construct the concentration-response curves. Data displayed as mean ± S.E.M. were fit with least-square nonlinear regression as those in fig. 1. For SB, d-FF, FF and mCPP, IC50 values are, respectively, 8.6, 30.5, 69.0 and 95.4 µM, and the slope coefficients were, respectively, 0.9, 1.0, 0.6 and 0.9.

Cardiac myocytes are rich in DRK channels and these channels constitute the major component of outward K⁺ current and participate in mediating repolarization of action potential (Noble, 1984). We found that d-FF caused a reversible and concentration-dependent inhibition of the DRK currents in cardiac myocytes (fig. 8A). The effect is also illustrated in the current-voltage relationship in figure 8B and in the concentration-response curve in figure 8C. A least-squares fitting with the percent inhibition of the current at +60 mV resulted an IC₅₀ of 250.9 µM, ~8 times that in the taste cells.

View larger version (30K): [in this window] [in a new window]   Fig. 8.   d-FF inhibits the DRK currents in rat cardiac ventricular myocytes. A, Whole-cell DRK currents elicited by voltage steps to 80, 0, +40, +60 and +80 mV in control, in d-FF at 10, 30, 100 and 300 µM and 1 mM and after a 5-min recovery in Tyrode's solution. Holding potential = 80 mV. All currents were recorded in the same cell. B, Current-voltage relationship in the absence and presence of d-FF from the data shown in A. Ordinate shows the amplitude of steady-state DRK currents (nA) in control (), 10 (), 30 (), 100 (), 300 µM () and 1 mM d-FF (). The abscissa denotes membrane potentials (mV) at which the DRK currents were recorded. C, Concentration-response of inhibition of DRK currents by d-FF. Ordinate is normalized DRK currents (at +60 mV) and abscissa denotes d-FF concentration (M) in a logarithmic scale. Data represent mean ± S.E.M. of 6 experiments.

	Discussion

Top Abstract Introduction Materials Results Discussion References

d-FF, FF and SB have been established as centrally acting anorectic agents, whose mechanism of action is believed to be attributed to their ability to stimulate the activity of 5-HT. Such a traditional concept has been challenged by increasing number of recent publications, which directly contradicted a 5-HT mediated mechanism, because the anorectic activity was not blocker by either the 5-HT neurotoxin (Grignaschi and Samanin, 1992) or by 5-HT synthesis inhibitor (Caccia et al., 1992; Gibson et al., 1993; Oluyomi et al., 1994; Lightowler et al., 1996). Thus, additional mechanism(s), central and/or peripheral, may exist to contribute to the control of appetite by this class of compounds. We speculated a role of K⁺ channels underlying the action of these anorectics in the light of the recent reports revealing a regulatory effect of centrally acting K⁺ channel modulators on food consumption (Ghelardini et al., 1997) and a linkage between the macronutrient preference and K⁺ channel expression in taste cells (Liu et al., 1997). The results of the present study, for the first time, provided evidence that 5-HT boosting anorectics suppressed DRK channels in lingual taste cells. In addition to the well established concept that the DRK channels in taste cells play a pivotal role in the gustatory transduction of basic tastes like sour, bitter and sweet (Avenet and Kinnamon, 1991; Kinnamon, 1992; Gilbertson, 1993), a recent study indicated that the same DRK channels in taste cells may be involved in chemoreception of dietary fat through their direct inhibition by essential fatty acids (Gilbertson et al., 1997). The blockade of DRK channels would be predicted to promote action potential firing leading to neurotransmitter release to gustatory afferent nerves. This sequence of events may contribute to the suppression of fat preference through enhancement of the satiating power of dietary fats (Smith et al., 1998). Given the reduction of food consumption by centrally acting K⁺ channel blockers (Ghelardini et al., 1997), our finding of the inhibitory effect of 5-HT agonists on the DRK channels in taste cells may suggest a peripheral route by which these drug produce anorectic action.

The inability of 5-HT to block the DRK currents in taste cells, in combination with the observation that the inhibitory effect by d-FF was not affected by PCPA (a 5-HT synthesis inhibitor), suggests that the d-FF-induced-blocking effect was independent of 5-HT availability. On the other hand, MTG, a nonselective 5-HT receptor antagonist, inhibited DRK channels, and its effect was additive to that of d-FF at submaximal concentrations, a result implying an involvement of the 5-HT receptor in d-FF effect. To explore whether alteration of 5-HT receptor activity is responsible for the effect of d-FF on DRK current, we compared the effect when the taste cells were dialyzed with G protein inhibitor, GDPS, with that in control, because 5-HT receptors belong to the superfamily of G protein-coupled receptors. The results convincingly showed that the blockade of G protein by GDPS did not at all affect the concentration dependence and the time course of the d-FF effect on the K⁺ current (data not shown). These observations tended to argue against the involvement of G protein and intracellular signaling pathways. Although the coupling between the 5-HT receptor and the DRK channel in response to d-FF and other anorectics has yet to be delineated, our results suggest that these 5-HT boosting agents blocked DRK channels by a mechanism requiring 5-HT receptors but not 5-HT stores. Interestingly, our findings coincide with the reported observation that the anorectic effect of d-FF depends only on the 5-HT receptor but not 5-HT availability (Curzon et al., 1997).

As all anorectics tested were hydrophilic in nature and were applied extracellularly, the rapid onset and washout of the blocking effects on the channels led us to propose a direct action at a site located on or close to the external surface of the cell membrane. To validate the hypothesis, we attempted to record the effect of intracellularly applied d-FF on the single DRK channel activity from inside-out detached membrane patches of taste cells. The time and concentration dependence of the effect could be indicative of the location of the site of action in reference to cell membrane and of the possible involvement of cellular second messenger(s), because in excised membrane patches the intracellular biochemical and biophysical pathways are usually disrupted due to lack of substrates and cellular organelles. However, we were unable to isolate and quantitatively analyze the activity of DRK channel under the influence of the anorectics due to the complexity of coexistence of multiple channels types with varying single-channel conductances in taste cell membrane.

As the knowledge of the molecular structure of the K⁺ channels that contribute to the DRK currents in taste cells is vitally important in determining the selectivity of these 5-HT boosting anorectics, we recently conducted studies to identify the subtype of the DRK channels in taste cells using multiple molecular biology and immunology tools. Preliminary results showed that a Shaker Kv1.5-like channel constituted the major component, whereas a Shab Kv2.1 channel was likely a minor component of the total DRK current in taste cells (Liu et al., 1998).

Cardiac myocytes possess DRK channels, and these channels are the major contributors to the outward K⁺ current mediating repolarization of action potential (Noble, 1984). Modern molecular cloning has revealed the primary structure of the cardiac DRK channels, of which Kv1.5 channel represents a prominent component (Roberds et al., 1993). The inhibition of DRK current in cardiac myocytes by d-FF, albeit at higher concentration than that in taste cells, could inherently lead to the delay of repolarization and hence the prolongation of action potential duration accompanied by positive inotropy in the heart (Tande and Refsum, 1998). In the long run, the sustained prolongation of repolarization and elevation of contractility of cardiac tissues could lead to Ca⁺⁺ overload and energy exhaustion. Chronic administration of d-FF reportedly causes pulmonary hypertension associated with increase in right ventricular systolic pressure and pulmonary arterial pressure (Abenhaim et al., 1996; SoRelle, 1997; Connolly et al., 1997). While the latter effect appears to be in part mediated by its inhibition of K⁺ channels in pulmonary vascular smooth muscle cells (Weir et al., 1996; Curfman, 1997), the blockade of cardiac DRK channels may contribute to the elevation of right ventricular pressure. It remains to be determined whether the cardiac effects resulting from the blockade of DRK channels would contribute to the pathogenesis of the clinical observation of regurgitant valvular heart disease developed after long-term treatment of d-FF (Graham and Green, 1997;. Cannistra et al., 1997), which has been suggested to be related to alterations in the concentration of circulating serotonin (Connolly et al., 1997).

In conclusion, our results indicate an important pharmacological role for d-FF and other 5-HT boosting agents as blockers of DRK channels via mechanism(s) that are independent of 5-HT availability. Although the effect in taste cells may contribute to anorectic action, the ubiquity of DRK channels in cardiac, neuronal, skeletal and smooth muscle cells (Rudy, 1988) could imply more extensive activities produced by this class of agents in the body.