Introduction

Top Abstract Introduction Experimental Procedures Results Discussion References

Some anorexinogen agents have been associated with the development of primary pulmonary hypertension (PPH). In the late 1960s an epidemic of PPH occurred after the introduction in Europe of aminorex fumarate (Follath et al., 1971). More recently, PPH has been strongly associated with the use of fenfluramines [the most prevalent of which was dexfenfluramine (DFF)] in Europe and North America (Abenhaim et al., 1996; Fishman, 1997). The mechanism by which aminorex, fenfluramines, and possibly phentermine increase the risk of PPH is not known. This class of drugs acts as indirect 5-HT receptor agonists by inhibiting 5-HT reuptake, inhibiting 5-HT metabolism, and causing the release of 5-HT from platelets, which all result in an elevation in plasma 5-HT levels (for review, see MacLean et al., 2000). It has been shown that abnormal handling of serotonin by platelets leading to an increase in plasma serotonin occurs in PPH (Hervé et al., 1995). Some data indicate that human pulmonary artery has a mixed functional population of 5-HT1B/1D and 5-HT2A receptors, which mediate the contractile response to 5-HT and could be involved in the development of PPH (Cortijo et al., 1997). In human studies, 5-HT caused only limited increases in tension of isolated rings of pulmonary artery (Higenbottam et al., 1999), but DFF caused concentration-dependent contractions in isolated human pulmonary arteries suspended in organ baths for isometric tension recording (Patnaude et al., 2000).

The underlying defect in PPH might be an abnormality of one or more voltage-gated potassium (Kv) channels in pulmonary artery smooth muscle cells (PASMCs). Indeed, in PASMCs, the activity of Kv channels governs membrane potential and regulates cytosolic free Ca²⁺ concentration (Yuan, 1995). A rise in Ca²⁺ is a trigger of vasoconstriction and a stimulus of smooth muscle proliferation. It has been shown that hypoxic pulmonary vasoconstriction is at least in part due to the inhibition of oxygen-sensitive outward potassium channels in PASMCs (for review, see MacCulloch et al., 1999). In smooth muscle cells, taken from the pulmonary arteries of PPH patients, Kv channels appear to be dysfunctional and Kv currents are significantly diminished in PPH-PASMCs compared with nonpulmonary hypertensive PASMCs (Yuan et al., 1998a). Moreover, it has been found that aminorex, fenfluramine, and dexfenfluramine, like hypoxia, inhibit delayed rectifier potassium current in smooth muscle cells taken from the small resistance pulmonary arteries of the rat lung, stimulate pulmonary vasoconstriction (Weir et al., 1996), and increase intracellular Ca²⁺ (Reeve et al., 1999a). More recently, fluoxetine and phentermine, used as alternative anorexic combination, were shown to cause a dose-dependent inhibition, at depolarized potential, of Kv current in PASMCs from rat (Reeve et al., 1999b).

Some progress has been made in identifying which Kv channels are expressed in PASMCs and what is the molecular identity of the delayed rectifier outward potassium current recorded from these cells. Polymerase chain reaction data suggest that the rat native delayed rectifier K⁺ current from PASMCs may be generated by the Kv1.1, Kv1.2, Kv1.5, Kv1.6, and/or Kv2.1/Kv9.3 -subunit channels (Turner et al., 1996; Yuan et al., 1998c). Antibodies were also used to immunolocalize and functionally characterize the contribution of Kv1.5 and Kv2.1 to rat PASMCs electrophysiology and vascular tone. Both subunits were shown to be expressed and are postulated to underlie the native current and contribute to the initiation of hypoxic pulmonary vasoconstriction (Archer et al., 1998). Also, prolonged hypoxia significantly decreased the mRNA levels of Kv1.5 and Kv2.1 in rat cultured PASMCs, as well as the protein levels of both subunits (Wang et al., 1997). Importantly, data obtained from human PPH-PASMCs suggest that inhibited gene transcription and decreased mRNA stability of Kv1.5 subunit could have an etiological role in the development of PPH (Yuan et al., 1998b).

Therefore, the present study was undertaken to characterize the direct effects of different anorexinogen agents, aminorex, dexfenfluramine, phentermine, fluoxetine, and sibutramine, on hKv1.5 channels expressed in a stable mammalian Chinese hamster ovary (CHO) cell line.

	Experimental Procedures

Top Abstract Introduction Experimental Procedures Results Discussion References

Cell Preparation. Effects of anorexinogen agents have been determined using a stable mammalian cell line expressing human Kv1.5 voltage-gated potassium current (Tamkun et al., 1991). Endogenous voltage-gated potassium channels are expressed in human embryonic kidney 293 cells (Yu and Kerchner, 1998), but little or no voltage-gated-like current was detected in CHO cell line (Philipson et al., 1993). Thus, we chose to express hKv1.5 in a CHO cell line. CHO cells were maintained and passaged as described previously (Perchenet and Clément-Chomienne, 2000). For electrical recordings, cells were split, passed on acid-washed and poly-D-lysine-coated coverslips, and used within 48 h. Nontransfected CHO cells did not display any voltage- or time-dependent current.

Electrical Recording. CHO cells were placed in a perfused recording chamber RCP-10T (Dagan Co., Minneapolis, MN) on the stage of an inverted microscope (Eclipse; Nikon, Basel, Switzerland). Potassium outward current was measured, at 23-25°C, in the whole cell configuration of the patch-clamp technique. A quartz micromanifold (ALA Scientific Instrument Inc., New York, NY) was positioned above individual cells to apply quickly various external drug-containing solutions. Pipettes were prepared from borosilicate capillary glass (Hilgenberg, Malsfeld, Germany) with a Sutter P-97 puller (Sutter Instrument Co., Novato, CA) and CPM-2 ALA microforge (ALA Scientific Instruments Inc.). Resistances of the patch pipettes were 2 to 3 M and 4 to 5 M for whole cell and single channel recordings, respectively. Recordings were performed using an EPC 9 double patch-clamp amplifier (HEKA Elektronik, Lambrecht/Pfalz, Germany). Pipette potential and capacitance were nulled and a 5- to 15-G seal formed with the cell membrane. Voltage-clamp protocols were applied using Pulse software (Heka Elektronik), whole cell and single channel current recordings were displayed and analyzed using Pulse, Pulsefit software (Heka Elektronik), and TAC and TACfit software (Bruxton Corporation, Seattle, WA). The amplitude of the hKv1.5 current expressed in CHO cell line was assessed at the end of 250-ms command pulse voltages applied every 15 s between 70 and +70 mV from a holding potential of 70 mV. Deactivation of tail current was recorded upon repolarization to 40 mV. Tail current amplitude was calculated as the difference between the peak amplitude of the tail and the sustained level of current after 200 ms of repolarization to 40 mV. Membrane potentials were corrected for a liquid junction potential between the pipette and the bath solutions. Capacitance and series resistance were optimized and ~80% compensation was usually obtained.

Analysis. A first order blocking scheme was used to describe drug-channel interaction. Apparent affinity constant, K_D, and Hill coefficient, n_H, were obtained from fitting of the fractional block Y at various drug concentrations [D]:

(1)

Apparent rate constants for binding (k) and unbinding (l) were obtain from solving

(2a)

(2b)

where _B is the time constant of the fast initial drug-induced current decay after activation from the holding potential to +50 mV.

(3)

(4a)

Statistical Method. All values in the text in figures are presented as means ± S.E.M. Direct comparisons between mean values in control conditions and in the presence of drug for a single variable were performed by paired Student's t test. A level of P < 0.05 was considered statistically significant.

	Results

Top Abstract Introduction Experimental Procedures Results Discussion References

Aminorex, Dexfenfluramine, Phentermine, Fluoxetine, and Sibutramine Dose-Dependent Inhibition of hKv1.5. Stably hKv1.5 transfected cells generated outward currents that rapidly rose to a peak and displayed very little inactivation during 250-ms depolarizing steps to +50 mV in control conditions (Fig. 1A). Outward currents were followed by decaying outward tail current upon repolarization to 40 mV. No contaminating current was observed on control nontransfected cells under the experimental conditions used in this work (Fig. 1A). For the study, the cells were exposed to five different anorexinogen compounds: aminorex, dexfenfluramine, phentermine, sibutramine, and fluoxetine. As an example, Fig. 1A shows representative results from a cell exposed to 300 µM dexfenfluramine. All drug effects were stable after a period of 4 to 6 min and reversible upon washout. Dose-response relations (10-300 µM) were constructed for the five compounds and compared with each other in their ability the inhibit hKv1.5 current expressed in CHO cells under whole cell recording conditions (Fig. 1B). The residual current in the presence of each drug was compared with that recorded in control conditions and the percentage of inhibition was calculated. Because aminorex, phentermine, and dexfenfluramine had minimal effect on hKv1.5 currents even at 300 µM, full dose-response curves were not constructed for these compounds, and the mechanism of block was not studied. Indeed, aminorex, phentermine, and dexfenfluramine at 300 µM, respectively, blocked 4 ± 2% (n = 7), 21 ± 5% (n = 8), and 27 ± 5% (n = 5) of the hKv1.5 current at + 50 mV. In contrast, fluoxetine and sibutramine inhibited a significantly greater percentage of hKv1.5 current and almost completely eliminated the current at 300 µM (91 ± 2% inhibition, n = 3 and 88 ± 2% inhibition, n = 9, respectively). Full dose-response curves were determined for these two compounds and the mechanism of block characterized as indicated below.

View larger version (13K): [in this window] [in a new window]   Fig. 1.   Inhibition by anorexinogen agents of hKv1.5 expressed in mammalian CHO cell line. A, hKv1.5 currents were elicited by 250-ms pulses applied from 70 mV to +50 mV in control conditions and in the presence of 300 µM DFF. No endogenous current was recorded from non-transfected CHO cells. B, mean data of end-pulse current amplitude inhibition obtained from five to nine cells using the same protocol as shown in A (+50 mV) were averaged and plotted against anorexinogen agent concentration.

Voltage Dependence of Aminorex, Phentermine, and Dexfenfluramine Blocks of hKv1.5. Figure 2 shows representative whole cell currents evoked by a series of 250-ms depolarizing voltage steps ranging from 70 to +70 mV in 10-mV increments from a holding potential of 70 mV before and after treatment with 1 mM aminorex, dexfenfluramine, or 300 µM phentermine. Families of tail currents were recorded after repolarization of the membrane to 40 mV. Representative current-voltage relations for the current measured at the end of the depolarizing steps before and after treatment with the drugs are also shown in Fig. 2. All end-pulse I-V relationships were linear for depolarizations positive to 10 mV, in control conditions. Aminorex, dexfenfluramine, and phentermine significantly reduced the amplitude of hKv1.5 at all voltages positive to 20 mV, respectively.

View larger version (30K): [in this window] [in a new window]   Fig. 2.   Effect of aminorex, dexfenfluramine, and phentermine on hKv1.5 currents elicited by depolarizing voltage pulses. Representative families of current traces elicited by depolarizing voltage pulses of 250 ms in 10-mV steps from a holding potential of 70 to +70 mV. The current amplitude measured at the end of the depolarizing pulses was plotted against the pulse potential in control conditions and in the presence of 1 mM aminorex (A), 1 mM dexfenfluramine (B), and 300 µM phentermine (C).

Concentration-Dependent Block of hKv1.5 by Sibutramine with One Binding Site. Extracellular application of 30 µM sibutramine resulted in a statistically significant reduction of end-pulse hKv1.5 current at +50 mV (41 ± 6%, n = 8) with a marked increase in the rate of outward current relaxation in the presence of sibutramine. At 100 µM, sibutramine induced a significant reduction of peak and end-pulse current at +50 mV, by 44 ± 4% and 65 ± 5%, respectively (n = 9) (Fig. 3A). This means that steady-state currents recorded at the end of the 250-ms pulses were more reduced by sibutramine than the peak currents, in a concentration-dependent manner. The effect of sibutramine was completely reversed upon washout of the drug (data not shown). Figure 3B shows the concentration dependence of sibutramine block of hKv1.5 in a range of concentration between 1 and 300 µM. The fractional block of steady-sate current measured at the end of 250-ms pulses was plotted against sibutramine concentration. A nonlinear least-squares fit of the concentration-response equation (eq. 1 under Experimental Procedures) to the individual data points yielded a K_D of 39 ± 5 µM and a Hill coefficient of 0.96 ± 0.08. When the data were fitted with the Hill coefficient constrained to 1, a similar apparent affinity was obtained (K_D = 41 ± 5 µM). These results suggested that binding of one molecule per channel was sufficient to inhibit potassium permeation.

View larger version (11K): [in this window] [in a new window]   Fig. 3.   Inhibition by sibutramine of hKv1.5 current expressed in mammalian CHO cell line. A, hKv1.5 current were elicited by 250-ms pulse applied from 70 to +50 mV in control conditions and in the presence of 10, 30, or 300 µM sibutramine. B, mean data of end-pulse current amplitude obtained from five to nine cells using the same protocol as shown in A were averaged and plotted against sibutramine concentration. A nonlinear least-squares fit of the concentration-response equation (continuous line) to the individual data points yielded an apparent KD of 39 ± 5 µM and a Hill coefficient of 0.96 ± 0.08.

Voltage-Dependent Block of hKv1.5 by Sibutramine Representative families of whole cell currents were recorded before and after applications of 100 µM sibutramine (Fig. 4A). Steady-state inhibition was obtained after 5 min of sibutramine application. Current-voltage relations in control conditions or in the presence of sibutramine for end-pulse current are represented in Fig. 4B. Sibutramine (100 µM) reduced the amplitude of hKv1.5 at all voltages positive to 20 mV, as is evident from the change in end-pulse amplitude currents in the families of whole cell current. The peak hKv1.5 current was also reduced by sibutramine, but about 2-fold less than end-pulse current at +70 mV. Figure 4B also shows that the degree of inhibition induced by 100 µM sibutramine was not linear but, displayed evidence of voltage dependence.

View larger version (25K): [in this window] [in a new window]   Fig. 4.   Voltage dependence of hKv1.5 block induced by sibutramine. A, representative families of current traces elicited by depolarizing voltage pulses of 250 ms in 10-mV steps from a holding potential of 70 to + 70 mV in control conditions and in the presence of 100 µM sibutramine. B, current-voltage relationships for control and sibutramine conditions. C, relative current expressed as Isibutramine/Icontrol from data obtained in the absence and in the presence of 100 µM sibutramine was plotted against the pulse potential. Individual data points are mean ± S.E.M. of five experiments. The dotted line shows the activation curve of the hKv1.5 channel. The block steeply increased between 20 and +10 mV. For membrane potentials positive to +10 mV, a continued but more shallow voltage dependence was also observed. This voltage dependence was fitted (continuous line) to eq. 3 (see Experimental Procedures) and yielded a  value of 0.17.

Mechanism Underlying the Block of hKv1.5 Current by Sibutramine. Figure 5 contains results that, together with our previous data, strongly suggest that sibutramine inhibits the hKv1.5 channel through a preferential interaction with the open state. Indeed, sibutramine inhibition of the hKv1.5 current develops following channel activation as is suggested by the development of current inhibition during the depolarizing steps represented in Fig. 3A. The inactivation time course displayed an additional rapid exponential component superimposed on the slow inactivation of the current. The decay of the current was then best fitted by a biexponential function. The time constant of the rapid component was concentration dependent and it was considered to represent the time constant of development of block (_Block). In Fig. 5B, the inverse, 1/_Block, of this time constant was then plotted as a function of sibutramine concentration (eq. 2a under Experimental Procedures) and the least-squares analysis to the experimental data yielded apparent k and l rates. Assuming a first order reaction for the drug-channel interaction, the ratio between l and k values would give the apparent K_D (eq. 2b under Experimental Procedures). This estimate was independent of the apparent K_D obtained from the curve in Fig. 3B. Nevertheless, the derived value (49 µM) was in the same range with that calculated from the fit of the concentration-response curve. This correlation between both independent methods to estimate the affinity supports the open-channel block model used to derive the rate constants. Binding rates can also be derived from the apparent K_D and the average value of _Block at 30 µM (0.015 ± 0.005 s). Using this procedure, the calculated k and l values were 0.94 × 10⁶ (M¹ s¹) and 38.5 s¹, respectively.

View larger version (15K): [in this window] [in a new window]   Fig. 5.   Open channel block mechanism of sibutramine inhibition. A, deactivation tail current crossover caused by sibutramine. Tail currents were recorded under control conditions and in the presence of 10, 30, and 100 µM sibutramine and then superimposed. B, the inverse of sibutramine-induced fast time constant at +50 mV has been plotted against the sibutramine concentration. The least-squares fit to the data (solid line) using the eq. 2a (see Experimental Procedures) resulted in apparent association (k) and dissociation (l) rate constants of 0.83 ± 0.10 × 106 M1 s1) and 41.2 ± 4.2 s1, respectively. The calculated KD value was found to be 49 µM (n = 5-9 cells). C, activation curves were drawn by fitting normalized tails current to the Boltzmann equation (eq. 4a, see Experimental Procedures). Data are mean ± S.E.M. from five cells. V0.5 and slope factor k are 9.4 ± 1.1 mV and 12.9 ± 1.0 and 23.9 ± 0.3 mV and 6.8 ± 0.3 in control conditions, and in the presence of 30 µM sibutramine, respectively.

Concentration-Dependent Block of hKv1.5 Current by Fluoxetine. In the presence of 3, 30, or 300 µM fluoxetine, the peak amplitude of hKv1.5 currents was reduced and there was acceleration in the apparent rate of current decay, resulting in a larger reduction of end-pulse hKv1.5 current at +50 mV. At 30 µM, fluoxetine induced a significant reduction of peak and end-pulse current amplitude of 39 ± 4 and 57 ± 3%, respectively (n = 6) (Fig. 6A). This means that steady-state currents recorded at the end of the 250-ms pulses were more reduced by fluoxetine than the peak currents, in a concentration-dependent manner. Therefore, the reduction in hKv1.5 currents at the end of the pulse was used as an index of inhibition. The concentration dependence of the inhibition of hKv1.5 induced by fluoxetine, in a range of concentrations between 1 and 300 µM, in presented in Fig. 6B. Nonlinear least-squares fit of the concentration-response equation (eq. 1 under Experimental Procedures) to the individual data yielded an apparent K_D of 21 ± 3 µM and a Hill coefficient of 0.90 ± 0.04. When the data were fitted with the Hill coefficient constrained to 1, an identical apparent affinity was obtained (K_D = 21 ± 2 µM). These results suggested that binding of one molecule per channel was sufficient to block potassium permeation.

View larger version (12K): [in this window] [in a new window]   Fig. 6.   Inhibition by fluoxetine of hKv1.5 expressed in mammalian CHO cell line. A, hKv1.5 currents were elicited by 250-ms pulse applied from 70 to +50 mV in control conditions and in the presence of 3, 30, or 300 µM fluoxetine. B, dose-response relationship for fluoxetine block obtained from three to eight cells using the same protocol as shown in A. A nonlinear least-squares fit of the concentration-response equation (continuous line) to the individual data points yielded an apparent affinity constant (KD) of 21 ± 3 µM and a Hill coefficient of 0.90 ± 0.04.

Voltage Dependence of Fluoxetine-Channel Interaction. Current-voltage relations for hKv1.5 currents at the end of the voltage step, as in Fig. 7, indicated that fluoxetine reduced hKv1.5 currents within the entire voltage range over which the current was activated (Fig. 7B). Steady-state inhibition was obtained after 3 to 5 min of fluoxetine application. As depicted in Fig. 7A, application of 10 µM fluoxetine resulted in a statistically significant reduction of end-pulse hKv1.5 current at +70 mV (57 ± 7%, n = 8) with a marked increase in the rate of outward current relaxation in the presence of fluoxetine. Since Fig. 7B shows that fluoxetine induced a downward curvature of the current-voltage relationship, our data indicated a higher current inhibition at more positive potentials. In the presence of 10 µM fluoxetine, the inhibition increased steeply between +10 and 20 mV, which corresponds to the voltage range for channel opening (Fig. 7C). This indicated that the ionized form of fluoxetine bound primarily to the open channel. By fitting the data according to the Woodhull model (eq. 3 under Experimental Procedures), the average electrical distance obtained from four cells was  = 0.22 ± 0.04. 

View larger version (25K): [in this window] [in a new window]   Fig. 7.   Voltage dependence of hKv1.5 block induced by fluoxetine. A, representative families of current traces elicited by depolarizing voltage pulses of 250 ms in 10-mV steps from a holding potential of 70 to +70 mV in control conditions and in the presence of 10 µM fluoxetine. B, plot of the current amplitude measured at the end of depolarizing pulses against the pulse potential in control and in the presence of 10 µM fluoxetine. C, relative current expressed as Ifluoxetine/Icontrol from data obtained in the absence and in the presence of 10 µM fluoxetine. Individual data points are mean ± S.E.M. of four experiments. The dotted line shows the activation curve of the hKv1.5 channel. Block steeply increased between 20 and +10 mV. For membrane potential positive to +10 mV, a continued but shallower voltage dependence was observed. This voltage dependence was fitted (continuous line) to eq. 3 (see Experimental Procedures) and yielded a  value of 0.22.

Time Courses of Inactivation and Deactivation in Presence of Fluoxetine. If fluoxetine can access its receptor only when the channel is in is open state then inhibition of the current can be expected to develop as the channels start to open, and development of block should be visible if the blocking rate is slower than the opening rate. As shown in Fig. 6A, in control conditions, the time course of partial inactivation of the current, elicited by a step to +50 mV, was well fitted by a monoexponential function. In the presence of fluoxetine, however, the current decay displayed a combination of fast and slow phases that were best fitted with a biexponential function. The fast time constant (_Block) was considered to be a reasonable approximation of the drug-channel interaction kinetics because it is a new component that was sufficiently faster than the slow component and was clearly apparent. Figure 8B shows the plot of 1/_Block versus the concentration of fluoxetine at test potential of +50 mV. The resultant K_D value from eq. 2b under Experimental Procedures was 25 µM, which is consistent with the K_D value of 21 µM from the dose-response curve (Fig. 6B).

View larger version (14K): [in this window] [in a new window]   Fig. 8.   Open channel block mechanism of fluoxetine inhibition. A, deactivation tail current crossover caused by fluoxetine. Tail currents were obtained at 40 mV after a 250-ms depolarizing pulse to +50 mV. B, kinetics of fluoxetine block of hKv1.5. The inverse of fluoxetine-induced fast time constant at +50 mV has been plotted against the fluoxetine concentration. The best fit to the data (solid line) using eq. 2a (see Experimental Procedures) resulted in an apparent association (k) and dissociation (l) rate constants of 2.4 ± 0.3 × 106 M1 s1) and 58.9 ± 11.1 s1, respectively. The calculated KD value was found to be 25 µM (n = 3-6 cells). C, activation curves were drawn by fitting normalized tails current to the Boltzmann equation (eq. 4a, see Experimental Procedures). Data are mean ± S.E.M. from four cells. V0.5 and slope factor k are 13.3 ± 0.9 mV and 10.7 ± 0.8 and 22.0 ± 0.7 mV and 7.0 ± 0.7 in control conditions, and in the presence of 10 µM fluoxetine, respectively.

Voltage Dependence of Activation Shift Due to Fluoxetine. The voltage dependence of activation in the presence of fluoxetine was evaluated to investigate whether the inhibition was correlated with a shift in channel activation (Fig. 8C). The data were best fitted with a single Boltzmann function (eq. 4a under Experimental Procedures). There was a clear concentration-dependent shift of the maximum potential to the left in the presence of fluoxetine, and this effect seemed to be greater as the potential became more positive and reached 0 mV. This point is illustrated by the statistically significant differences in the values for the half-activation (V_0.5) and slope factor (k) of the Boltzmann functions in legend of Fig. 8.

Single Channel Recordings in Presence of Sibutramine and Fluoxetine. Figure 9, A and C, show representative recordings and amplitude histograms of single hKv1.5 channel activity of two cell-attached membrane patches at + 60 mV, in the absence and the presence of sibutramine or fluoxetine. Unitary current amplitude was neither affected by 300 µM sibutramine or fluoxetine but both drugs significantly reduced the open probability of the channel. Open dwell-time histograms (Fig. 9, B and D) indicate that this was due at least in part to a significant decrease in mean open time as is suggested by the increase in rate of channel closure indicated by the open  values shown. In the presence of sibutramine, mean open duration was reduced from 65 ± 20 to 9 ± 2 ms (n = 106 observations) and it was reduced from 80 ± 10 to 54 ± 5 ms (n = 65 observations) during the treatment with fluoxetine.

View larger version (25K): [in this window] [in a new window]   Fig. 9.   Inhibition of hKv1.5 single channel activity by sibutramine and fluoxetine. A, representative recordings of hKv1.5 activity in a cell-attached membrane patch under asymmetrical KCl recording conditions at +60 mV before (control) and during treatment with 300 µM sibutramine and amplitude histograms for 50-s recordings of hKv1.5 channel activity in the same cell. C, representative recordings of hKv1.5 activity in an other cell-attached membrane patch before (control) and during treatment with 300 µM fluoxetine and corresponding amplitude histograms. B and D, open dwell-time histograms for representative data before and after treatment with sibutramine or fluoxetine. Open probability and time constant for each condition are indicated.

	Discussion

Top Abstract Introduction Experimental Procedures Results Discussion References

Anorexinogen agents have been associated with PPH and some of them have been shown to block the delayed rectifier potassium currents in PASMCs (Weir et al., 1996; Reeve et al., 1999b). Additionally, based on data obtained from human PASMCs, it has been suggested that the hKv1.5 subunit could have an etiological role in the development of PPH (Yuan et al., 1998b). For these reasons, we have determined the interactions of anorexinogen agents with hKv1.5 channels. Our patch-clamp experiments show that aminorex, phentermine, dexfenfluramine, sibutramine, and fluoxetine cause a dose-dependent inhibition of hKv1.5 channels stably expressed in CHO cell line. Aminorex, phentermine, and dexfenfluramine were found to have a K_D of inhibition greater than 300 µM and are not very potent inhibitors of hKv1.5. On the other hand, we show here that sibutramine and fluoxetine blocked hKv1.5 current with much lower K_D values of 41 and 21 µM, respectively. These values are in the range of other well known blockers of hKv1.5 channel, such as erythromycin (Rampe and Murawsky, 1997), 4-aminopyridine (Bouchard and Fedida, 1995), quinine (Snyders and Yeola, 1995), and descaboethoxyloratadine (Caballero et al., 1997). Either sibutramine or fluoxetine (both at 300 µM) decreased the single channel open probability and mean open time obtained from cell-attached patches. In previous studies, fluoxetine was shown to be a potent blocker of Kv1.3 channel expressed in CHO cells with an IC₅₀ of 6 µM (Choi et al., 1999), but blocked Kv1.1 channel expressed in Xenopus laevis oocytes with an IC₅₀ greater than 500 µM (Tytgat et al., 1997). This difference of potency could be explained by the different expression systems.

The fast time course of current activation was unaffected, either by sibutramine or fluoxetine, indicating that block did not occur until the hKv1.5 channels open. Indeed, the main effect of sibutramine and fluoxetine was to accelerate hKv1.5 current decay during step depolarizations, thus reducing current amplitude at the end of the test pulse. Possible explanations are that sibutramine and fluoxetine bind to the open state or accelerate current inactivation. The sibutramine- and fluoxetine-induced extra components of relaxation in current amplitude had time constants that were much faster than that of slow inactivation; therefore, these fast time constants (_block) can be considered to be the result of the interaction of sibutramine and fluoxetine with the open state of the channel. From the analysis of the concentration dependence of block due to both drugs, it was evident that the potency of fluoxetine to block hKv1.5 channels is 2-fold higher than the potency of sibutramine at +50 mV. Differences in potency between both drugs could be attributed to differences in k and l rate constants. Our results indicate that dissociation rate constants were similar. Thus, the different potency to block hKv1.5 channel can be attributed to the faster association rate constant observed for fluoxetine. In addition to their concentration-dependent effects, the interactions of both compounds with hKv1.5 channel are also voltage-dependent. The inhibition increased steeply in the voltage range of channel activation, thus providing further evidence that the channel must open before sibutramine or fluoxetine can bind to their receptor site on the channel and block permeation. Consistent with the predominantly cationic forms of these drugs at physiological pH, sibutramine and fluoxetine display greater block at more depolarized membrane potentials. Fractional electrical distances  of 0.17 and 0.22 were obtained in the experiments with sibutramine and fluoxetine, respectively. This  value for fluoxetine is similar to those (0.29, 0.24) obtained in previous reports for the drug inhibition of Kv1.3 and Kv1.1 channels (Tytgat et al., 1997; Choi et al., 1999). Open channel block should also affect the time course of tails currents. In our experiments, sibutramine and fluoxetine both decreased the peak tail current amplitudes and slowed the decay of tail currents, producing a crossover of tail phenomenon. This phenomenon was previously reported for a number of open state hKv1.5 channel blockers (Zhang et al., 1997; Franqueza et al., 1998; Perchenet and Clément-Chomienne, 2000). A displacement of the activation curves toward negative voltages was observed in the presence of sibutramine and fluoxetine. These shifts could be explained by the drugs binding preferentially to the channel open state, thus limiting an increase of hKv1.5 channel conductance at more depolarized test potentials. These data are also consistent with the decrease in channel open duration observed in the presence of sibutramine and fluoxetine.

We would like here to emphasize the fact that the percentages of hKv1.5 inhibition at +50 mV, induced by fluoxetine, dexfenfluramine, phentermine, and aminorex, are very similar to the values reported for voltage-gated potassium current inhibition in rat PASMCs (Weir et al., 1996; Reeve et al., 1999b). The homology between human and rat Kv1.5 sequences is known to be high, with the greatest identity apparent in the pore domain of the subunits (Swanson et al., 1990; Fedida et al., 1993). It is possible that fluoxetine binds the rat Kv1.5 subunit expressed in rat PASMCs (Yuan et al., 1998c). Moreover, fluoxetine causes inhibition of the potassium current of this cell type only at membrane potentials positive to 30 mV, supporting an open channel block mechanism (Reeve et al., 1999b). Since voltage-gated potassium currents regulate resting membrane potential in rat pulmonary arterial myocytes (Yuan, 1995), it was speculated that aminorex- and dexfenfluramine-induced PPH, could be due in part, in part, to the inhibition of PASMC delayed rectifier potassium current (Weir et al., 1996; Michelakis et al., 1999). However, the steady-state levels of dexfenfluramine in human are known to be lower than the micromolar range (Haritos et al., 1998) and are far from the concentrations reported here to block hKv1.5. Nevertheless, the degree of primary pulmonary hypertension associated with dexfenfluramine treatment is very high. Low therapeutic plasma concentrations of phentermine ranging from 0.5 to 0.7 µM have also been reported in human (Rothman et al., 2000), but again this does not correlate with the low potency of this compound for inhibition of hKv1.5. We have shown that sibutramine and fluoxetine are more potent blockers than aminorex, dexfenfluramine, and phentermine. However, no epidemic of PPH has been associated with the therapeutic use of Prozac (fluoxetine) or Meridia (sibutramine). The absence of a role for inhibition of hKv1.5 could be explained by the therapeutic plasma concentration of fluoxetine around 0.3 µM, which is 100-fold lower than the K_D for inhibition of the hKv1.5 channel. When administrated to humans, sibutramine is rapidly demethylated to form two active metabolites with very low maximal plasma concentrations between 10 and 30 nM (Hind et al., 1999). It is known that other Kvs subunits are abundantly expressed in rat pulmonary artery such as Kv2.1 (Yuan et al., 1998c). This subunit is also an important determinant of resting membrane potential in PASMCs and could be linked to initiation of PPH (Archer et al., 1998). Patel et al. (1997) have shown that dexfenfluramine blocks Kv2.1 expressed in X. laevis oocytes and in COS mammalian cells with an IC₅₀ of 300 µM. Additional studies would be needed to determine whether other anorexic agents could block Kv2.1, or other Kv subunits expressed in PASMCs.