High affinity interaction of mibefradil with voltage-gated calcium and sodium channels

Abstract

1. Mibefradil is a novel Ca²⁺ antagonist which blocks both high-voltage activated and low voltage-activated Ca²⁺ channels. Although L-type Ca²⁺ channel block was demonstrated in functional experiments its molecular interaction with the channel has not yet been studied. We therefore investigated the binding of [³H]-mibefradil and a series of mibefradil analogues to L-type Ca²⁺ channels in different tissues.

2. [³H]-Mibefradil labelled a single class of high affinity sites on skeletal muscle L-type Ca²⁺ channels (K_D of 2.5±0.4 nM, B_max=56.4±2.3 pmol mg⁻¹ of protein).

3. Mibefradil (and a series of analogues) partially inhibited (+)-[³H]-isradipine binding to skeletal muscle membranes but stimulated binding to brain L-type Ca²⁺ channels and α1C-subunits expressed in tsA201 cells indicating a tissue-specific, non-competitive interaction between the dihydropyridine and mibefradil binding domain.

4. [³H]-Mibefradil also labelled a heterogenous population of high affinity sites in rabbit brain which was inhibited by a series of nonspecific Ca²⁺ and Na⁺-channel blockers.

5. Mibefradil and its analogue RO40-6040 had high affinity for neuronal voltage-gated Na⁺-channels as confirmed in binding (apparent K_i values of 17 and 1.0 nM, respectively) and functional experiments (40% use-dependent inhibition of Na⁺-channel current by 1 μM mibefradil in GH3 cells).

6. Our data demonstrate that mibefradil binds to voltage-gated L-type Ca²⁺ channels with very high affinity and is also a potent blocker of voltage-gated neuronal Na⁺-channels. More lipophilic mibefradil analogues may possess neuroprotective properties like other nonselective Ca²⁺-/Na⁺-channel blockers.

Introduction

The benzimidazolyl-substituted tetraline derivative mibefradil is a novel Ca²⁺ channel blocker with several interesting pharmacological properties. It blocks high-voltage-activated L-type Ca²⁺ channels in the cardiovascular system resulting in antihypertensive activity and slowing of heart rate without decreasing cardiac contractility (Clozel et al., 1997). Although mibefradil was withdrawn from the market due to pharmacokinetic interactions it remains an interesting pharmacological probe to investigate the contribution of individual ion channel subtypes for cardiovascular functions. In vascular and atrial myocytes it preferentially blocks low voltage-activated T-type Ca²⁺ channels suggesting that part of the bradycardic and vasodilating properties are mediated by block of these channels (Bezprozvanny & Tsien, 1995; Clozel et al., 1997). Heterologous expression of cloned Ca²⁺ channel α1 subunits confirmed block of both high (α1A, α1B, α1C, α1E; Bezprozvanny & Tsien, 1995) and low voltage-activated (α1G, α1H; Cribbs et al., 1998) Ca²⁺ channels by low micromolar concentrations of mibefradil. Block of L- (α1C) and non-L-type Ca²⁺ channels (α1A, α1B, α1E) occurred with similar IC₅₀ values. It is therefore tempting to speculate that, if high affinity binding of tritiated mibefradil to L-type Ca²⁺ channels can be demonstrated, this radioligand should also be suitable for the reversible labelling of non-L-type Ca²⁺ channel complexes as well.

So far the binding of mibefradil to L-type Ca²⁺ channels has not been quantified in much detail. In preliminary studies the affinity of mibefradil for cardiac and smooth muscle channels was derived from its potency to displace (−)-[³H]-devapamil from the phenylalkylamine (PAA) binding domain of the channel. IC₅₀ estimates varied between 13 nM (Rutledge & Triggle, 1995) and 39 nM (Osterrieder & Holck, 1989). Direct labelling studies with [³H]-mibefradil were hampered by the presence of low affinity sites (Ratner et al., 1996). Mibefradil failed to modulate (+)-[³H]-isradipine binding to cardiac membranes and to polarized and depolarized cardiomyocytes (Rutledge & Triggle, 1995). This suggests that it binds to a site not coupled to the dihydropyridine (DHP) binding domain.

In order to establish the usefulness of [³H]-mibefradil as a biochemical probe for voltage gated Ca²⁺ channels, we determined its binding properties in different excitable tissues using tritiated mibefradil and a series of unlabelled mibefradil analogues. We provide evidence that mibefradil labels a site on skeletal muscle and brain L-type Ca²⁺ channels which is tightly coupled to the DHP domain by a non-competitive binding mechanism. In addition, we found that mibefradil and its analogues also display high affinity for neuronal voltage-gated Na⁺-channels and potently block Na⁺-currents in GH3 cells.

Methods

Materials

(+)-[³H]-isradipine (≈79 Ci mmol⁻¹) was purchased from New England Nuclear (Vienna, Austria), (−)-[³H]-devapamil (≈85 Ci mmol⁻¹) from Amersham (Vienna, Austria). [³H]-SQ32133 (85 Ci mmol⁻¹) was synthesized by radioactive methylation of the corresponding phenol precursor. [³H]-mibefradil (≈11.47 Ci mmol⁻¹) and [³H]-WIN17317-3 (≈28.2 Ci mmol⁻¹) were provided by Hoffmann-La Roche (Basel, Switzerland) and by Merck (Rahway, NY, U.S.A.), respectively. Chemicals were obtained from the following sources: flunarizine, bovine serum albumin, phenylmethylsulphonyl fluoride and polyethylenimine from Sigma (Vienna, Austria), the stereoisomeres of verapamil, gallopamil and devapamil from Knoll (Ludwigshafen, Germany), the stereoisomeres of isradipine from Sandoz (Basel, Switzerland), fluspirilene and pimozide from Janssen (Beerse, Belgium), niguldipine from BYK Gulden (Konstanz, Germany), and diltiazem from Gödecke (Freiburg, Germany). Mibefradil (RO40-5967) and the mibefradil analogues (RO18-5881, RO19-6945, RO19-8287, RO19-8531, RO19-9495, RO40-0293, RO40-0713, RO40-6040, RO40-6088) (for chemical structures see Figure 1) were generously provided by Hoffmann-La Roche (Basel, Switzerland).

Figure 1.

Chemical structure of mibefradil and mibefradil analogues. For stereochemistry of R2 see Table 1. R2 is isopropyl in all compounds. Note that RO-40-6088 contains a methoxyacetate group on the tetrahydronaphtyl ring instead of a methoxy group.

Membrane preparations and radioligand binding studies

Partially purified t-tubule membranes from rabbit skeletal muscle, rabbit cardiac membranes (Glossmann & Ferry, 1985), guinea-pig synaptic plasma membranes (Feigenbaum et al., 1988) and from transfected tsA201 cells were prepared as described and stored in liquid nitrogen until use. TsA201 cells were transfected with cDNA from α1C (Mitterdorfer et al., 1995) together with β1a and α2-δ (Mitterdorfer et al., 1995) (all cDNAs in plasmid pcDNA3) using the Ca²⁺ phosphate precipitation method. Membrane protein concentrations were determined according to Lowry (Lowry et al., 1951) with bovine serum albumin as a standard.

[³H]-mibefradil equilibrium binding assay

Binding assays were carried out at 25°C in 50 mM Tris-HCl, pH 7.4, 0.1 mM phenylmethylsulphonyl fluoride, 0.2 mg ml⁻¹ bovine serum albumin (0.5–1 ml final assay volume). Nonspecific binding was defined in the presence of 1 μM mibefradil. Incubation lasted for 60–120 min. Bound ligand was determined by filtration of the assay mixture over GF/C Whatman filters pretreated with 0.25% (v v⁻¹) polyethyleneimine for about 30 min (Glossmann & Ferry, 1985). Filters were washed five times with ice-cold buffer (50 mM Tris-HCl, pH 7.4, 150 mM NaCl) and then counted for radioactivity by liquid scintillation counting. All drug solutions were prepared as 10 mM drug stock solutions in DMSO and stored at −25°C. The final DMSO concentration in the incubation mixture was ⩽1% (v v⁻¹) and did not affect radioligand binding. All experiments were done in duplicate.

(−)-[³H]-devapamil, (+)-[³H]-isradipine, [³H]-SQ32133 and [³H]-WIN17317-3 binding assays were performed as previously described (Kraus et al., 1998b; Prinz & Striessnig, 1993; Wanner et al., 1999). Nonspecific binding was defined by the presence of 1 μM (±)-isradipine, 3 μM (±)-devapamil, 10 μM (+)-cis-diltiazem and 1 μM WIN17317-3, respectively. Protein and radioligand concentrations are given in the figure legends.

To determine dissociation rate constants, radioligand and membranes were incubated until equilibrium was reached. The association reaction was blocked by an excess of unlabelled ligand or by 40 fold dilution of the assay mixture with assay buffer followed by filtration at the indicated times. Dissociation rate constants were determined as the slope of the regression line from a plot of 1n (fractional binding) versus time. Association (k₊₁) and dissociation (k₋₁) rate constants were also calculated from association experiments by fitting the data to the differential equations describing a bimolecular reversible binding reaction using the software package SCIENTIST^R (MicroMath Inc.). The dissociation constants (K_D) and the maximal density of binding sites (B_max) were obtained by linear regression analysis after Scatchard transformation of the equilibrium saturation binding data.

Electrophysiological recordings

Inward Na⁺ currents (I_Na) in GH3-cells (ATCC CCL82.1) were measured with the whole-cell configuration of the patch clamp technique (Hamill et al., 1981). Recordings were carried out at room temperature in an extracellular bath solution containing (mM) NaCl 140, CaCl₂ 1.8, MgCl₂ 2, 4-aminopyridine 5, tetraethylammonium chloride 5, HEPES 10 and glucose 10 (adjusted to pH 7.4 with NaOH). Data analysis and acquisition was performed by using the pClamp software package (version 6.0, Axon Instruments). The recording chamber (150 μl total volume) was continuously perfused at a flow rate of 1 ml min⁻¹ with control and drug-containing solutions. I_Na was recorded after depolarization from a holding potential of −80 mV to a test potential of +5 mV. Initial ‘tonic' block (resting state-dependent block) was defined as peak I_Na inhibition during the first pulse 2 min after drug application as compared with I_Na in the absence of drug. ‘Use (frequency-)-dependent' block of I_Na was measured during trains of 5 or 50 ms test pulses (3 Hz) applied from −80 mV to a test potential +5 mV after a 2-min equilibrium period in the drug-containing solution. Use-dependent block was expressed as the per cent decrease of peak I_Na during the last pulse of the train as compared with I_Na during the first pulse.

Data analysis

Nonlinear least square fitting and statistical calculations were performed using ORIGIN 5.0. Data are given as means± s.e.mean for the indicated number of experiments.

Results

High affinity binding of [³H]-mibefradil to L-type Ca²⁺ channels in skeletal muscle

The basic binding properties of [³H]-mibefradil (11.5 Ci mmol⁻¹) for L-type Ca²⁺ channels were determined in partially purified rabbit skeletal muscle transverse tubule membranes. After incubation for 60 min a saturable binding component was detected which increased linearly with membrane protein concentrations up to about 0.04 mg ml⁻¹ (Figure 2A). Saturation analysis revealed a K_D of 2.5±0.4 nM and a B_max of 56.4±2.3 pmol mg⁻¹ (n=8). Scatchard transformation of the saturation data was linear indicating labelling of a homogenous class of sites (Figure 2B, inset). The B_max was 2.6 fold higher than the B_max determined in parallel experiments for the DHP (+)-[³H]-isradipine (21.8±0.6 pmol mg⁻¹, n=3) in the same membrane preparations. The high affinity was confirmed in kinetic experiments (Figure 2C). Mibefradil exhibited slow binding kinetics, with association and dissociation rate constants of 0.0104±0.0019 nM⁻¹ min⁻¹ (n=4) and 0.0226±0.0024 min⁻¹ (n=7), respectively, yielding a kinetically derived K_D (k₋₁/k₊₁) of 2.2 nM. Dissociation was monophasic confirming the absence of site heterogeneity and increased with the concentration of cold ligand (Figure 2C). This ligand-induced accelerated dissociation has been described before for other cationic amphiphilic Ca²⁺ antagonists (Prinz & Striessnig, 1993).

Figure 2.

[³H]-Mibefradil binding to partially purified rabbit skeletal muscle transverse tubule membranes. (A) Protein dependent binding: radioligand concentration was 2.9 nM. (B) Saturation analysis: Increasing concentrations of radioligand were incubated with 8.8 μg ml⁻¹ of membrane protein for 60 min at 25°C before specific binding was determined. The following binding parameters were obtained after Scatchard transformation (inset) by linear regression analysis: r=0.98, K_D=1.6 nM; B_max=52.4 pmol mg⁻¹ membrane protein. (C) Association kinetics. 2.3 nM [³H]-mibefradil was incubated with 4.4 μg ml⁻¹ membrane protein for the indicated times before specific binding was measured. The following binding parameters resulted from non-linear curve fitting to a bimolecular binding reaction as described in the Methods section: k₊₁=0.0087 min⁻¹ nM⁻¹; k₋₁=0.0327 min⁻¹; calculated K_D=3.8 nM. Dissociation kinetics (inset). After specific binding equilibrium was reached (B₀) dissociation was initiated by 40 fold dilution or by addition of 100 nM or 1 μM mibefradil. The following dissociation rate constants were calculated by linear regression from semilogarithmic plots: k₋₁=0.034 min⁻¹ (▪); k₋₁=0.166 min⁻¹ (▵); k₋₁=0.99 min⁻¹ (○). (D) Inhibition of [³H]-mibefradil by phenylalkylamines; membrane protein concentration=4.4 μg ml⁻¹; radioligand concentration=1–2.1 nM.

[³H]-Mibefradil binding was modulated by other chemical classes of Ca²⁺ antagonists. As shown in Figure 2D the enantiomers of the phenylakylamines verapamil, devapamil and gallopamil caused complete inhibition with IC₅₀ values almost identical to those obtained after competitive displacement of (−)-[³H]-devapamil from the PAA binding domain. The following IC₅₀ values (and apparant Hill-slopes, n_H) for [³H]-mibefradil inhibition were obtained by fitting the data of three independent experiments to the general-dose response equation: (−)-verapamil: 98±2.5 nM (0.93±0.05); (+)-verapamil: 66±6.3 nM (0.76±0.03); (−)-gallopamil: 25.3±0.8 (0.86±0.02); (+)-gallopamil: 122±20 nM (0.88±0.03); (−)-devapamil: 8.65±0.22 nM (0.89±0.02); (+)-devapamil: 35.5±5.3 nM (1.01±0.03). Almost identical binding parameters were obtained for (−)-[³H]-devapamil labelled channels in parallel experiments (range, n=2): (−)-verapamil: 50–51 nM (0.83–0.88); (+) verapamil: 43–45 nM (0.96–1.00); (−)-gallopamil: 12–13 nM (0.99–1.08); (+)-gallopamil: 71–94 nM (0.85–1.20); (−)-devapamil: 2.9–5.7 nM (0.97–1.07); (+)-devapamil: 21–28 nM (1.2–1.5) (see also Goll et al., 1984a,1984b).

(−)-Isradipine also completely inhibited [³H]-mibefradil binding, whereas (+)-isradipine was a partial inhibitor (Figure 3A).

Figure 3.

Modulation of [³H]-mibefradil binding to skeletal muscle L-type Ca²⁺ channels by DHPs and cations. Membrane concentration was 8.8 μg ml⁻¹; radioligand concentration was 2.0 nM. (A) The following parameters were calculated by non-linear curve fitting: (+)-isradipine, IC₅₀=1.6 nM, n_H=0.94; inhibition to 42% of control; (−)-isradipine, IC₅₀=1.4 μM, n_H=0.77, complete inhibition. (B) Inhibition of [³H]-mibefradil by cations: All cations were employed as chloride salts. The following binding parameters (IC₅₀, maximal inhibition) were obtained by linear curve fitting to the general-dose response equation: Cd²⁺: 45 μM; 0% of control; Ba²⁺: 310 μM; 12% of control; Ca²⁺: 136 μM; 11% of control; La³⁺: 1.3 mM; 0% of control; Sr²⁺: 1.4 mM; 21% of control; Na⁺: no inhibition. One of 2–3 similar experiments is shown.

As expected for L-type Ca²⁺ channels, [³H]-mibefradil binding was also inhibited by divalent cations but not by Na⁺ (Figure 3B) in a concentration range previously described for other Ca²⁺ channel ligands (Glossman et al., 1983; Goll et al., 1984b).

A series of mibefradil analogues with substituents of different length on the methylaminoethyl nitrogen revealed that replacement of the 3-(2-benzimidazoyl) propyl group by other substitutents (Table 1), or increase of its distance from the methylaminoethyl nitrogen (RO40-6040, RO40-6088) or introduction of one (RO19-8531) or two methyl groups (RO40-293) in this side chain was well tolerated. Replacement by a dimethoxyphenylethyl group (compound RO18-5881) even increased affinity (Table 1).

Table 1.

Modulation of [3H]-mibfradil, (-)-[³H]-devapamil and [³H]-SQ32133 labelled L-type Ca²⁺ channels by mibefradil and its analogues

Figure 4A and Table 1 illustrate that unlabeled mibefradil also modulated the binding of Ca²⁺ antagonists to the DHP, PAA and benzothiazepine binding domain specifically labelled with (+)-[³H]-isradipine, (−)-[³H]-devapamil and [³H]-SQ32,133 (Hering et al., 1993), respectively. Mibefradil was a partial inhibitor of (+)-[³H]-isradipine binding. This inhibitory effect was converted to a stimulatory effect after side-chain replacement with a dimethoxyphenylisopropyl group (RO19-8531, Figure 4A). Stimulation was critically dependent on the presence of a single methyl substituent on the ethyl linker (RO18-5881, complete inhibition; RO19-8531, stimulation; RO40-293, partial inhibition) indicating that this substituent sensitively effects the non-competitive interaction of mibefradil with the DHP binding domain. All other analogues were incomplete inhibitors of (+)-[³H]-isradipine binding (not shown).

Figure 4.

Modulation of drug binding to L-type Ca²⁺ channels by mibefradil and mibefradil analogues. (A) Modulation of (+)-[³H]-isradipine binding to rabbit skeletal muscle Ca²⁺ channels. Radioligand concentration 0.3–1 nM; membrane protein concentration 8.8 μg ml⁻¹. The following binding parameters were determined by non-linear curve fitting (IC₅₀ or EC₅₀; n_H; % maximal inhibition or stimulation: mibefradil: 4.2 nM, 0.94, 59% of control; RO18-5881: 1.34 nM, 0.91, 3% of control; RO19-8531: 14 nM, 0.7, 175% of control; RO40-0293: 22 nM 1.43, 81% of control. (B) Comparison of the apparent affinity of mibefradil and analogues for [³H]-mibefradil, (−)-[³H]-devapamil and [³H]-SQ32133 labelled L-type Ca²⁺ channels. The IC₅₀ values of mibefradil and its analogues for [³H]-mibefradil labelled skeletal muscle Ca²⁺ channels are plotted against the corresponding IC₅₀ values for (−)-[³H]-devapamil (r=0.95, n_H=1.01) and [³H]-SQ32133 (r=0.92, n_H=0.97) labelled channels, respectively. (C) Modulation of (+)-[³H]-isradipine binding to L-type Ca²⁺ channels by mibefradil in cerebral cortex membranes (114 μg ml⁻¹), α1C/α2-δ/β1 expressing tsA201 cell membranes (60 μg ml⁻¹), rabbit heart membranes (82 μg ml⁻¹) and partially purified rabbit skeletal muscle transverse-tubule membranes (82 μg ml⁻¹). Radioligand concentration was 0.05–0.3 nM.

Mibefradil and its analogues completely inhibited (−)-[³H]-devapamil and [³H]-SQ32,133 binding with almost the same affinity as [³H]-mibefradil-labeled channels and with Hill slopes close to unity (Table 1, Figure 4B). Taken together these data are consistent with a competitive interaction of mibefradil with the PAA and the BTZ binding domain.

High affinity of unlabelled mibefradil for neuronal L-type Ca²⁺ channels

Our finding that mibefradil inhibits DHP binding to skeletal muscle L-type Ca²⁺ channels (α1S subunit, Birnbaumer et al., 1994) is in contrast to the previously reported absence of DHP binding modulation in cardiac membranes (Rutledge & Triggle, 1995). Figure 4C illustrates that mibefradil indeed modulated DHP binding in a tissue-dependent manner. Whereas no concentration-dependent mibefradil effect was observed in rabbit heart membranes, the stimulatory effect on (+)-[³H]-isradipine binding to brain L-type Ca²⁺ channels (EC₅₀=15.2±5.6 nM, n=6) and recombinant cardiac α1C subunits expressed in tsA201 cells (EC₅₀=156±17 nM, n=5) unmasked the high affinity of this drug for brain and cardiac L-type channels. Accordingly, mibefradil and its analogues also inhibited (−)-[³H]-devapamil binding to guinea-pig brain membranes with IC₅₀ values similar to those found in skeletal muscle (Table 1). Our data show that the non-competitive binding interaction between mibefradil and the DHP binding domain discriminates structural differences between α1 subunit isoforms (α1S in skeletal muscle vs α1C in heart and brain) and/or splice variants of α1C subunits in heart and brain (Snutch et al., 1991; Welling et al., 1997).

High affinity binding site for [³H]-mibefradil in rabbit brain

To test if mibefradil also binds to neuronal Ca²⁺ channels other than L-type we performed direct labelling studies with [³H]-mibefradil in rabbit brain membranes. A saturable binding component was detected which increased linearly with protein concentrations up to about 0.04 mg ml⁻¹ (not shown). Scatchard transformation of the specific binding data revealed the existence of a single high affinity binding component (K_D=10.1±1.5 nM, B_max=21.3±2.9 pmol mg⁻¹) (Figure 5A, inset). Binding was not modulated by (+)- or (−)-isradipine (n=3, not shown). Inhibition by (−)-devapamil (Figure 5A, see below) occurred with an IC₅₀ two orders of magnitude higher than expected for L-type channels (Figure 2D, Goll et al., 1984a). Therefore L-type Ca²⁺ channels do not account for this binding component in accordance with their much lower density in mammalian brain (<1 pmol mg⁻¹, Striessnig et al., 1988).

Figure 5.

[³H]-mibefradil and [³H]-WIN17317-3 binding to cerebral cortex membranes. (A) Inhibition of [³H]-mibefradil binding to rabbit cerebral cortex membranes. The following binding parameters were determined by nonlinear curve-fitting (IC₅₀; n_H): pimozide: 25 nM, 0.64; flunarizine: 33 nM, 0.73; devapamil: 310 nM; 0.86; DPI 201–106: 23 nM, 0.93. Radioligand and membrane protein concentrations were 1–1.9 nM and 14–21 μg ml⁻¹, respectively. Inset: Saturation analysis of [³H]-mibefradil binding to cerebral cortex membranes: r=0.96; K_D=8.4. nM, B_max=17 pmol mg⁻¹ of protein. (B) Inhibition of [³H]-WIN17317-3 binding to guinea-pig synaptic plasma membranes. The following binding parameters were determined by nonlinear curve fitting (IC₅₀; n_H): WIN17317-3: 4.6 nM, 0.89; mibefradil: 29 nM, 0.93; RO40-0293: 92 nM, 0.92; RO40-6040: 15 nM, 1.04. Radioligand and membrane protein concentrations were 1.4–1.8 nM and 26 μg ml⁻¹, respectively.

To test the possible relationship of the [³H]-mibefradil binding sites in brain membranes with non-L-type Ca²⁺ channels we tested the affinity of fluspirilene, pimozide and flunarizine which are blockers of non-L-type Ca²⁺ channels (Geer et al., 1993; Sah & Bean, 1993; Uneyama et al., 1997). All of them inhibited mibefradil binding with low nanomolar IC₅₀ values (fluspirilene, IC₅₀=16.2±1.7 nM, n=3; pimozide, IC₅₀=19.6±4.6 nM, n=3; flunarizine, IC₅₀=32–36 nM, range, n=2) (Figure 5A). The IC₅₀ of (−)-devapamil (215±52 nM, n=3) also agreed well the known 10–100 fold lower potency of PAAs for block of non-L-type as compared to L-type Ca²⁺ channels (Hockerman et al., 1997; Uneyama et al., 1997) in electrophysiological studies. Taken together these results do not rule out the possibility that [³H]-mibefradil labels non-L-type ca²⁺ channels in mammalian brain but the lack of selectivity of these drugs for Ca²⁺ channels and their shallow inhibition curves prevented a pharmacological proof of this hypothesis.

We were unable to obtain other direct biochemical evidence for the interaction of [³H]-mibefradil with non-L-type Ca²⁺ channels. After solubilization of prelabelled channels in digitonin for immunoprecipitation with anti-α1A (P/Q-type) and anti-β antibodies solubilized binding activity was too unstable for further analysis (not shown). We therefore expressed α1A subunits together with α2-δ and β subunits in tsA201 cells to directly measure binding of [³H]-mibefradil to recombinant channels. A high affinity binding site for [³H]-mibefradil was present at similar densities in membrane preparations of both α1A/α2-δ/β but also of mock transfected cells (K_D=1.02±0.16 nM, B_max=1.08±0.37 pmol mg⁻¹ of protein, n=4). This endogenous site was not characterized further but prevented sensitive detection of [³H]-mibefradil interaction with expressed Ca²⁺ channels in tsA201 and HEK 293 cells. It seemed not to be associated with metabolic enzymes as it was absent in liver membrane preparations (not shown).

Since flunarizine also blocks neuronal voltage-gated Na⁺-channels (Kiskin et al., 1993) and the Na⁺-channel blocker DPI201-106 displayed also high affinity (IC₅₀=28±2.9 nM, n=3; Figure 5A) for the [³H]-mibefradil labelled sites, we tested the possibility that at least part of the [³H]-mibefradil binding occurs to voltage-gated Na⁺-channels. To this end we labelled brain Na⁺-channels with the high affinity Na⁺-channel-selective probe [³H]-WIN17317-3 (Wanner et al., 1999). Its binding properties in guinea-pig synaptic plasma membranes were consistent with high affinity labelling of voltage-gated Na⁺-channels (K_D=4.4±1.0 nM, B_max=10.0±2.9 pmol mg⁻¹ of protein, n=3; Figure 5B). The typical binding stimulation by 1 μM phenytoin was also observed (n=2; Wanner et al., 1999). Mibefradil (K_i=17.2±1.8 nM, n=3) as well as its analogues completely inhibited [³H]-WIN17317-3 binding with low nanomolar IC₅₀ values (Figure 5B; Table 2). The affinity of the mibefradil analogues for [³H]-WIN17317-3-labelled Na⁺-channels did not correlate with their affinity for skeletal muscle Ca²⁺ channels (Table 2) suggesting that binding to Na⁺-channels occurs with a different structure-activity relationship. This was especially evident for compound RO18-5881, which has 127 fold lower affinity for Na⁺- than for Ca²⁺ channels (Table 2).

Table 2.

Inhibition of [³H]-WIN17317-3 binding to voltage-gated Na⁺ channels by mibefradil and its analogues

To prove the Na⁺-channel activity of mibefradil in functional experiments the effect of increasing concentrations of mibefradil on Na⁺-currents in GH3 cells was measured using the whole-cell patch clamp technique. Na⁺-inward current (I_Na) was elicited by 5 ms depolarizations from a holding potential of −80 mV to +5 mV at a frequency of 3 Hz. The concentration-dependent tonic (measured as block of I_Na during the first pulse after incubation with the drug) and use-dependent block (observed at the end of the pulse train) are shown in Figure 6A,B. Under these experimental conditions half-maximal use-dependent block occurred around 1 μM. This concentration is similar to those reported for half-maximal block of native (Liu et al., 1999; Mishra & Hermsmeyer, 1994) and recombinant (Bezprozvanny & Tsien, 1995; Cribbs et al., 1998; Mehrke et al., 1994; Williams et al., 1999) T- and L-type Ca²⁺ channels in muscle cells.

Figure 6.

Inhibition of I_Na by mibefradil in GH3 cells. (A) Inward currents through voltage-dependent Na⁺-channels were elicited by 3 Hz trains of 5 ms depolarizations from a holding potential of −80 mV to +5 mV before and after incubation of the cell in 1 (left), 10 μM (middle) and 100 μM (right) mibefradil. The first, 25th and/or 100th pulse of the train in the absence (C1, C100) and presence (D1, D25, D100) of drug are indicated. (B) Tonic (black) and use-dependent (white) block induced by increasing concentrations of mibefradil during a train of 100 test pulses (5 ms, 3 Hz) are shown as means±s.e.mean for the indicated number of experiments. (C) Use-dependent I_Na inhibition during a train of 100 test pulses (5 or 50 ms) in the absence (white) or presence of 10 μM mibefradil (black). Data are shown as means±s.e.mean for the indicated number of experiments. (D) Use-dependent I_Na inhibition during 3 Hz-pulse trains from −80 to 5 mV of 5 or 50 ms test pulses. The kinetics of use-dependent block were fitted to single exponential functions with corresponding time constants of peak I_Na decay of 7.4±0.7 and 7.9±0.7 ms (n=4, P>0.05). Data are shown as means±s.e.mean for n=4. Inset: Representative I_Na (10 μM mibefradil present) during the first and 100th 50 ms-pulse of a 3 Hz-pulse train.

To evaluate if use-dependent inhibition (Figure 6) reflects preferential block of open or inactivated Na⁺ channels we increased the time which the channel spends in the inactivated state by increasing the duration of individual pulses during the train from 5 to 50 ms. Neither the extent (Figure 6C) nor the time course of block development (Figure 6D) were changed during 50 ms pulse trains arguing against a role of inactivation for mibefradil action.

Discussion

Our experiments show that mibefradil interacts with skeletal muscle and brain L-type Ca²⁺ channels with high affinity. In skeletal muscle its binding properties are compatible with competitive interaction at the PAA/BTZ binding domain as implicated by the following properties: like for the PAA [³H]-verapamil (Goll et al., 1984b) its binding density was higher than for DHPs, [³H]-mibefradil binding was inhibited by unlabelled PAAs in an apparently competitive manner; a series of mibefradil analogues blocked (−)-[³H]-devapamil and [³H]-SQ32,133 interaction with IC₅₀ values corresponding to their affinity for the [³H]-mibefradil-labelled channel. In addition, the dissociation rate increased as a function of unlabelled ligand concentration as previously described for PAAs (Goll et al., 1984b) and BTZs (Prinz & Striessnig, 1993). This ‘ligand-induced accelerated dissociation' indicates that more than one mibefradil molecule can bind to the channel at the same time (Prinz & Striessnig, 1993). Like PAAs and BTZs, mibefradil and its derivatives non-competitively inhibit or stimulate DHP binding (Glossmann et al., 1985). Taken together these results strongly indicate that mibefradil binds to the same multisubsite domain on the α1 subunit of L-type Ca²⁺ channels that contains the closely associated interaction sites for DHPs, PAAs and BTZs (Hockerman et al., 1997; Striessnig et al., 1998). This is further supported by our recent finding that mibefradil not only stimulates DHP binding to recombinant α1C and brain L-type channels (this study) but also to a recombinant DHP binding domain transferred into previously DHP insensitive α1A subunits by site-directed mutagenesis (Huber et al., 2000).

We further demonstrate that mibefradil and some of its analogues display high affinity for neuronal voltage-gated Na⁺-channels. Like [³H]-WIN17317-3 (Wanner et al., 1999) mibefradil must be selective for neuronal vs cardiac Na⁺-channels because of the reported weak Na⁺-channel blocking activity in mammalian heart (Hoffmann-La Roche Inc., 1994). Enhanced inactivation did not affect use-dependent Na⁺ channel block of GH3 cells (Figure 6C,D). This result is in line with previous reports of Aczél et al. (1998) on mibefradil block of P/Q-type Ca²⁺ channels and suggests an open state- but not inactivated state-dependent block mechanism in both channel types.

The potential of mibefradil to block both neuronal Na⁺-channels as well as a wide spectrum of voltage-gated Ca²⁺ channels is of high pharmacological interest. Block of Na⁺- (e.g. phenytoin, carbamazepine, lamotrigine) as well as T-type Ca²⁺ channels (e.g. phenytoin, ethosuximide) is believed to cause antiepileptic effects (MacDonald & Kelly, 1995; Todorovic & Lingle, 1998). Likewise, block of Na⁺-channels (e.g. riluzole, lifarizine; Bryson et al., 1996; McGivern et al., 1995) as well as N-type Ca²⁺ channels (SNX-111; Miljanich & Ramachandran, 1995) results in neuroprotective effects. Therefore lipophilic mibefradil derivatives with high brain penetration may represent interesting anticonvulsant as well as neuroprotective drugs.

Mibefradil has recently been shown to block L- and non-L-type Ca²⁺ channels, including low voltage activated T-type channels, expressed in Xenopus oocytes with similar affinity (Bezprozvanny & Tsien, 1995). This suggests that high affinity binding domains should also be located on non-L-type Ca²⁺ channel α1 subunits such as α1A (P/Q-type), α1B (N-type) and α1E (R-type). Although we demonstrated the existence of high affinity binding sites (K_D=10 nM, B_max=21 pmol mg⁻¹) for [³H]-mibefradil in mammalian brain we were unable to prove their association with non-L-type Ca²⁺ channels. It is difficult to relate the B_max of the [³H]-mibefradil labelled sites to the density of all cloned Ca²⁺ channel types taken together. L-type (0.5–0.9 pmol mg⁻¹ of (+)-[³H]-isradipine or (−)-[³H]-devapamil binding; Striessnig et al., 1988), P/Q-type (up to 3 pmol mg⁻¹ of [¹²⁵I]-omega-conotoxin MVIIC binding; Pichler et al., 1996) and N-type (up to 1 pmol mg⁻¹ of [¹²⁵I]-omega-conotoxin GVIA binding; Knaus et al., 1987) channels account for about 5 pmol of sites per mg of protein. However, due to the lack of specific radioligands the density of R-type (α1E) and T-type channels (α1G, α1I), which together with Na⁺-channels may also contribute to [³H]-mibefradil binding in brain, is still unknown.

The presence of a high affinity binding site in tsA201 cells was unexpected and prevented the further analysis of [³H]-mibefradil interaction with recombinant non-L-type Ca²⁺ channel α1 subunits. As for [³H]-mibefradil binding in brain we lack specific pharmacological tools to prove if these sites are associated with the endogenous Ca²⁺ channels previously identified in HEK 293 cells (Berjukow et al., 1996).

Taken together our data demonstrate that structural modification of mibefradil results in compounds with even higher affinity for Na⁺- (e.g. RO40-6040) or Ca²⁺ channels (e.g. RO18-5881) than mibefradil itself. Thus our study emphasizes the need to screen mibefradil analogues for compounds with higher affinity for non-L-type Ca²⁺ channels which could help to develop the first generation of non-peptide radioligands for non-L-type Ca²⁺ channels. Such ligand would provide us with radioligand-based screening assays to develop non-L-type Ca²⁺ channel-active compounds. Such compounds are believed to represent important therapeutics for the treatment of severe pain (Miljanich & Ramachandran, 1995), migraine (Kraus et al., 1998a) and cerebral ischaemia (Miljanich & Ramachandran, 1995).

Abbreviations

BTZ

benzothiazepine

DHP

dihydropyridine

K_D

dissociation constant

PAA

phenylalkylamine