Inhibition of HERG1 K⁺ channels by the novel second-generation antihistamine mizolastine

Abstract

1. Ventricular arrhythmias are rare but life-threatening side effects of therapy with the second-generation H₁ receptor antagonists terfenadine and astemizole. Blockade of the K⁺ channels encoded by the Human Ether-à-go-go-Related Gene 1 (HERG1) K⁺ channels, which is the molecular basis of the cardiac repolarizing current I_Kr, by prolonging cardiac repolarization, has been recognized as the mechanism underlying the cardiac toxicity of these compounds.

2. In the present study, the potential blocking ability of the novel second-generation H₁ receptor antagonist mizolastine of the HERG1 K⁺ channels heterologously expressed in Xenopus oocytes and in HEK 293 cells or constitutively present in SH-SY5Y human neuroblastoma cells has been examined and compared to that of astemizole.

3. Mizolastine blocked HERG1 K⁺ channels expressed in Xenopus oocytes with an estimated IC₅₀ of 3.4 μM. Mizolastine blockade was characterized by a fast dissociation rate when compared to that of astemizole; when fitted to a monoexponential function, the time constants for drug dissociation from the K⁺ channel were 72.4±11.9 s for 3 μM mizolastine, and 1361±306 s for 1 μM astemizole.

4. In human embryonic kidney 293 cells (HEK 293 cells) stably transfected with HERG1 cDNA, extracellular application of mizolastine exerted a dose-related inhibitory action on I_HERG1, with an IC₅₀ of 350±76 nM. Furthermore, mizolastine dose-dependently inhibited HERG1 K⁺ channels constitutively expressed in SH-SY5Y human neuroblastoma clonal cells.

5. The results of the present study suggest that the novel second-generation H₁ receptor antagonist mizolastine, in concentrations higher than those achieved in vivo during standard therapy, is able to block in some degree both constitutively and heterologously expressed HERG1 K⁺ channels, and confirm the heterogeneity of molecules belonging to this therapeutical class with respect to their HERG1-inhibitory action.

Introduction

In the last two decades, second-generation H₁ receptor blockers have been developed for the symptomatic treatment of allergic reactions in order to overcome the marked antimuscarinic and sedative properties displayed by first-generation congeners (Sorkin & Heel, 1985). Due to their improved H₁ selectivity, absence of sedation, and possibly, antiallergic properties not deriving from their antihistaminic activity, second-generation H₁ receptor blockers are among the most prescribed drug families (Slater et al., 1999).

Mizolastine (2-[[[1-[1-[(4-fluorophenyl)methyl)-1H-benzimidazol-2-yl]piperidinyl]methylamino]-4(3H)-pyrimidinone; SL 85.0324) is a new potent and selective second-generation H₁ receptor blocker available in Europe, which is effective in relieving the symptoms of allergic rhinitis (Leynadier et al., 1996; Tasman & Weber, 1995) and urticaria (Brostoff et al., 1996). This compound, structurally related to astemizole, is devoid of sedative (Patat et al., 1995; Vuurman et al., 1994) and antimuscarinic (Danjou et al., 1992) properties, and displays a more hydrophilic profile which confers a lower distribution volume, good bioavailability, and limited metabolic conversion by the hepatic CYPs 3A4 and 2A6 (Delauche-Cavallier et al., 1999). These pharmacological properties have represented a distinct advantage over some other antihistamines.

Despite the obvious improvement in the treatment of allergic diseases represented by second-generation antihistamines, concerns over the risk/benefit profile of some members of this pharmacological class have recently been raised (Woosley, 1996). In fact, the rare occurrence of ‘torsade de pointes', a form of polymorphic ventricular arrhythmia, has been described in patients taking intentional overdoses of the second-generation antihistamines astemizole or terfenadine (Craft, 1986; Davies et al., 1989), or suffering from predisposing factors to the development of cardiac arrhythmias and undergoing therapy with these antihistamines (Monahnan et al., 1990). These potentially fatal adverse cardiovascular reactions have increased the level of awareness in the medical community, leading to guidelines established by the European Agency for the Evaluation of Medicinal Products (EMEA) for pre-clinical and clinical studies aimed at reducing the risk for cardiovascular adverse effects by drugs to be introduced in the market (EMEA, 1997).

The QT prolongation and ventricular arrhythmia caused by terfenadine and astemizole have been related to the ability of these drugs to interfere with cardiac potassium channels involved in action potential repolarization (Berul & Morad, 1995), and particularly with the rapid component of the cardiac repolarizing potassium current (I_Kr) (Woosley et al., 1993; Sanguinetti et al., 1995; Salata et al., 1995). The Human Ether-à-go-go-Related Gene 1 (HERG1) (Warmke & Ganetzky, 1994) seems to represent the molecular basis of I_Kr, since the K⁺ currents encoded by HERG1 display biophysical and pharmacological characteristics similar to those of cardiac I_Kr (Trudeau et al., 1995; Sanguinetti et al., 1995).

The aim of the present study has been to investigate the potential interaction of increasing concentrations of mizolastine with HERG1 K⁺ channels heterologously expressed in Xenopus oocytes and in human embryonic kidney 293 cells (HEK 293 cells) (Zhou et al., 1998), or endogenously present in SH-SY5Y human neuroblastoma cells (Arcangeli et al., 1995); and to compare the actions of the novel H₁ receptor blocker with those of astemizole, one of the second-generation antihistamines with well-known inhibitory actions on the K⁺ channels encoded by HERG1 (Suessbrich et al., 1996; Taglialatela et al., 1998b).

Methods

Xenopus oocytes isolation

Ovarian lobes were surgically removed from adult female Xenopus frogs (Rettili di Schneider, Varese, Italy) and placed in 100-mm Petri dishes containing a Ca²⁺-free solution of the following composition (in mM): NaCl 82.5, KCl 2, MgCl₂ 1, HEPES 5, piruvic acid 2.5, 100 u ml⁻¹ penicillin, and 100 μg ml⁻¹ streptomycin, pH 7.5 with NaOH. After four extensive washes, the oocytes (stage V–VI) were dissociated at room temperature by collagenase treatment (type IA, 45–80 min at a concentration of 2 mg ml⁻¹). At the end of the collagenase treatment, the oocytes were placed in a Ca²⁺-containing solution of the following composition (in mM): NaCl 100, KCl 2, CaCl₂ 1.8, MgCl₂ 1, HEPES 5, piruvic acid 2.5, 100 u ml⁻¹ penicillin, and 100 μg ml⁻¹ streptomycin, pH 7.5 with NaOH. Dissociated oocytes were then placed in a 19°C incubator and used for the experiments on the following day.

Molecular biology and oocyte injection

The cloning of HERG1 (Warmke & Ganetzky, 1994), BEAG (Frings et al., 1998), rKv2.1 (Frech et al., 1989), and rKv3.1 (Taglialatela et al., 1991) has already been described. These cDNAs were cloned into the following vectors: pSp64A⁺ for HERG1, pCRII for BEAG, and pBluescript for rKv2.1 and rKv3.1. cDNAs were linearized with HindIII for BEAG, EcoRI for HERG1, and NotI for rKv2.1 and rKv3.1. cRNAs were transcribed in vitro from linearized cDNAs by means of a commercially available kit (mCAP, Stratagene), using T7 RNA polymerase for BEAG, rKv2.1, and rKv3.1, and SP6 RNA polymerase for HERG1. RNAs were quantified by the RiboGreen fluorescent kit (Molecular Probes) and stored in a stock solution (250 ng μl⁻¹) at −20°C in 0.1 M KCl. One day after isolation, Xenopus oocytes were microinjected with 46 nl of the respective cRNA stock solution or appropriate dilution.

Cell culture

Human neuroblastoma SH-SY5Y cells were cultured in Dulbecco's Modified Eagle Medium (DMEM), containing glucose (4.5 g l⁻¹) and 5% foetal calf serum (FCS), and incubated at 37°C in a humidified atmosphere with 10% CO₂ in 100-mm plastic Petri dishes. HEK 293 cells were cultured in modified eagle medium (MEM), supplemented with Earle's salts, nonessential amino acids (0.1 mM), penicillin (50 u ml⁻¹), steptomycin (50 μg ml⁻¹), G418 (0.4 mg ml⁻¹) and 10% foetal calf serum (FCS), and incubated at 37°C in a humidified atmosphere with 5% CO₂ in 100-mm plastic Petri dishes. For electrophysiological experiments, the cells were seeded on glass coverslips (Fisher) coated with poly-L-lysine (30 μg ml⁻¹). All the experiments were performed 1–4 days after seeding at room temperature (22–23°C).

Electrophysiology

Voltage-clamp with two microelectrodes

The oocytes were voltage-clamped with a commercially available amplifier (Warner OC-725A, Warner Instr. Corp.). Current and voltage electrodes were filled with 3 M KCl, 10 mM HEPES (pH 7.4; ≈amp;1 MΩ resistance). The bath solution contained (in mM): NaCl 88, KCl 10, MgCl₂ 2.6, CaCl₂ 0.18, HEPES 5, pH 7.5. This solution was perfused in the recording chamber at a rate of about 0.2 ml min⁻¹. Data were stored on the hard disk of a 486 IBM compatible computer for off-line analysis. The pCLAMP software (version 6.0.2, Axon Instr., CA, U.S.A.) was used for data acquisition and analysis. Currents were recorded at room temperature.

Patch-clamp

Currents from the human neuroblastoma SH-SY5Y and the HERG1-transfected HEK 293 cells were recorded at room temperature using a commercially available amplifier (Axopatch 200A, Axon Instruments, Foster City, CA, U.S.A.). The whole-cell configuration of the patch-clamp technique (Hamill et al., 1981) was adopted using glass micropipettes of 3–7 MΩ resistance. No compensation was performed for pipette resistance and cell capacitance. The cells were perfused with an extracellular solution containing (in mM): KCl 100, EGTA 10 and HEPES 10, pH 7.3 with KOH (for the SH-SY5Y cells), or NaCl 150, KCl 10, CaCl₂ 3, MgCl₂ 1, HEPES 10, pH 7.4 with NaOH (for the HERG1-transfected HEK 293 cells). The pipettes were filled with (in mM) CsCl 110, tetraethylammonium-Cl 10 MgCl₂ 2, EGTA 10, glucose 8, Mg-ATP 2, cyclic AMP 0.25 and HEPES 10, pH 7.3 with NaCl KOH (for the SH-SY5Y cells), or K-Aspartate 130, NaCl 10, CaCl₂ 4, MgCl₂ 2, EGTA 10, Mg-ATP 2, cyclic AMP 0.25 and HEPES 10, pH 7.4 with NaOH (for the HERG1-transfected HEK 293 cells; free Ca²⁺: 51 nM, calculated with MaxChelator Ver. 6.50).

Drugs and statistics

All the reagents were purchased from Sigma Chemicals (Italy). Astemizole was kindly provided by Janssen-Cilag (Rome, Italy). The H₁-receptor antagonists were dissolved in dimethylsulphoxide (DMSO) at concentrations between 5 and 50 mM, and stock solutions were kept at −20°C. Appropriate drug dilutions were prepared daily. The maximal DMSO concentration (0.6%) did not affect HERG1 K⁺ channels recorded in Xenopus oocytes, HEK 293 cells, or SH-SY5Y human neuroblastoma cells. Statistical significance between the data was obtained by means of the Student's t-test or by the ANOVA test followed by the Tukey test. When appropriate, data are expressed as the mean±s.e.mean.

Results

Differential effect of mizolastine on the K⁺ currents carried by four cloned K⁺ channels expressed in Xenopus oocytes

Upon microinjection with HERG1 cRNA, Xenopus oocytes expressed a K⁺ current having biophysical properties which resembled those of the I_Kr current (Trudeau et al., 1995; Sanguinetti et al., 1995). This K⁺ current is activated by depolarization yet displays a pronounced inward rectification of the current-voltage relationship at positive potentials (>0 mV); furthermore, it shows rather slow kinetics of activation, and exhibits a large inward component upon repolarization to −100 mV, a membrane potential value below the equilibrium potential for K⁺ ions. Extracellular perfusion with mizolastine (10 μM) inhibited HERG1 K⁺ channels in Xenopus oocytes (Figure 1); the per cent of inhibition (≈70%), was similar whether measured at the peak of the inward current elicited at −100 mV following long (1–2 s) depolarizing pulses to 0 mV or at the peak of the outward current during the depolarizing pulse. Analogously, 10 μM mizolastine also blocked by ≈amp;70% the outward K⁺ currents carried by the channels encoded by the bovine isoform of the Ether-à-go-go Gene (BEAG; Frings et al., 1998), when heterologously expressed in Xenopus oocytes. By contrast, the K⁺ currents flowing through the channels encoded by K_v2.1 were insensitive to blockade by 10 μM mizolastine; whereas those carried by K_v3.1 were only slightly affected by the same concentration of the second-generation antihistamine (Figure 1).

Figure 1.

Effect of mizolastine on the K⁺ currents carried by four cloned K⁺ channels expressed in Xenopus oocytes. (A) Representative current traces recorded with the two-microelectrode voltage-clamp technique from four different oocytes expressing HERG1, BEAG, Kv3.1, and Kv2.1 recorded in control condition and after 5 min perfusion with 10 μM mizolastine. Holding potential: −90 mV (HERG1 and BEAG), −70 mV (K_v3.1 and K_v2.1); test potentials from −80 mV to +20 mV (HERG1 and BEAG) or from −60 mV to + 20 mV (K_v3.1 and K_v2.1) in 20 mV steps; return potential: −100 mV (HERG1 and BEAG) or −70 mV (K_v3.1 and K_v2.1). When reported, the dashed lines in the panel indicate the zero current level. (B) The per cent variation of the outward K⁺ current induced by 5 min perfusion with 10 μM mizolastine (I_drug/I_control) at voltages which maximally activated the conductance for each K⁺ channel (0 mV for HERG1, +20 mV for BEAG, K_v3.1 and K_v2.1). Each experimental point is the mean±s.e.mean of 3–8 separate experiments. Asterisks denote values significantly different from control values (P<0.05). (C) Chemical structures of mizolastine and astemizole.

Figure 2A shows the effect of different concentrations of mizolastine (1, 10 and 30 μM) on the inward K⁺ currents carried by heterologously expressed HERG1 expressed in Xenopus oocytes recorded upon cell hyperpolarization to −100 mV following depolarization to 0 mV to activate the current. In Figure 2B, the per cent of control HERG1 K⁺ tail currents blockade exerted by the different mizolastine concentrations (0.3–30 μM), calculated from the ratio I_drug/I_control of the peak inward K⁺ currents elicited at −100 mV following depolarization to 0 mV (without subtraction of the steady-state inward current measured at the end of the hyperpolarizing pulse), were normalized and fit to the following binding isotherm: y=max/(1+X/IC₅₀)ⁿ, where X is the drug concentration and n the Hill coefficient. Fitted values for n were 0.93±0.08. The IC₅₀ for HERG1 blockade by mizolastine was 3.4 μM.

Figure 2.

Dose-dependence of the inhibitory action of mizolastine on HERG1 K⁺ currents expressed in Xenopus oocytes. (A) Representative current traces recorded with the two-microelectrode voltage-clamp technique from a single HERG1-expressing oocyte exposed to control condition, and after perfusion with 1, 10 and 30 μM mizolastine (5 min for each concentration). Holding potential: −90 mV; test potential: 0 mV; return potential: −100 mV. Only the tail currents recorded upon return to −100 mV are shown. The dashed line in the panel indicate the zero current level. (B) Dose-dependence of HERG1 K⁺ currents inhibition by mizolastine. The inhibitory effect of the different concentration of the H₁ receptor antagonist on the peak of the inward HERG1 K⁺ currents recorded at −100 mV was calculated for several cells and plotted against drug concentrations. The solid line represents the fit of the experimental data to the following binding isotherm: y=max/(1+X/IC₅₀)ⁿ, where X is the drug concentration and n the Hill coefficient. The fitted value for n was between 0.72 and 0.96. Each point is the mean±s.e.mean of 3–6 determinations.

Since other second-generation antihistamines such as terfenadine and astemizole are known to potently inhibit HERG1 K⁺ channels, in the next series of experiments the kinetics of HERG1 K⁺ channels blockade by astemizole and mizolastine were compared. Figure 3 shows the time-course of the blockade of the inward HERG1 tail currents at −100 mV by increasing concentrations of mizolastine (0.3–3 μM), each perfused for 3 min followed by a washout period of variable duration (3–10 min). After achieving a steady-state block with each mizolastine concentration, the removal of the drug from the solution perfusing the oocytes caused a complete recovery of the K⁺ currents. When the time-course of such recovery was fit to a monoexponential function, the time constants of this monoexponential function did not depend on mizolastine concentration. In the experiment shown in Figure 3, the time constants were 46.6 s, 31.2 s, and 48.4 s for 0.3, 1 and 3 μM mizolastine, respectively. These rates likely reflect the kinetics of drug dissociation from the channel and could not be accounted for by the time-course of solution exchange in the recording bath (≈amp;15 s), as judged by the change in reversal potential of the HERG1 K⁺ current occurring upon switching solutions with different K⁺ concentrations (data not shown). On the other hand, the time-course of HERG1 K⁺ current recovery after 1 μM astemizole was much slower than that occurring with mizolastine, resulting in a value of 2340 s in the experiment shown in Figure 3. When similar experiments and analysis were performed on five different cells, the time constants for HERG1 K⁺ current recovery were 72.4±11.9 s (n=5) for 3 μM mizolastine, and 1361±306 s (n=5) for 1 μM astemizole. These results suggested that the lower potency for HERG1 K⁺ channels blockade exhibited by mizolastine when compared to astemizole (the IC_50s were 3.4 μM vs 480 nM, respectively) (Taglialatela et al., 1998b) could be mainly ascribed to the faster dissociation rate of mizolastine from HERG1 K⁺ channels.

Figure 3.

Kinetic difference between HERG1 K⁺ channel blockade by mizolastine and astemizole in Xenopus oocytes. The peak of the inward tail K⁺ current carried by HERG1 channels, measured upon return to −100 mV after depolarizing pulses of 1 s duration to 0 mV elicited every 15–20 s from an holding potential of −90 mV, is plotted versus time during the exposure to the different experimental conditions indicated by the black boxes. The solid lines are the fits of the experimental points recorded upon washout of each of the drug concentrations indicated to single exponential equations of the following form: y=A1^*exp{−(t-K)/τ}+C, using the fitting routines of the Clampfit module of the pClamp suite (ver. 6.0.3), where A1 is the amplitude of the exponential function, (t−K) is the time range used fot the fitting (taking into consideration the start of the fit region, K), τ is the time constant, and C is an offset to account for the steady-state asymptote.

Effects of mizolastine on HERG1 K⁺ channels heterologously expressed upon stable transfection of HEK 293 cells with HERG1 cDNA or constitutively expressed in SH-SY5Y human neuroblastoma cells

In order to verify that the HERG1 inhibitory action exerted by mizolastine was not a characteristic of the Xenopus oocyte expression system, HEK 293 cells stably transfected with HERG1 cDNA (Zhou et al., 1998) were used. Figure 4 shows representative current traces from the same HERG1-transfected HEK 293 cell subsequently recorded in control conditions and after perfusion with increasing mizolastine concentrations (0.01–1 μM), each lasting about 4 min. From a holding potential of −80 mV, the K⁺ channels were activated by 1-s depolarizing pulses at potentials from −60 mV to +40 mV. At the end of the depolarizing pulses, the cell was repolarized to −100 mV, giving rise to a large inward tail current. In line with the results obtained in Xenopus oocytes, mizolastine displayed a dose-dependent inhibitory action on the HERG1 K⁺ currents heterologously expressed in HEK 293 cells, with an IC₅₀ for such inhibitory action of 350±76 nM (n=5 cells). After perfusing the cells with the highest concentration of mizolastine (1 μM), 1 μM astemizole perfusion caused a further and almost complete blockade of HERG1 inward K⁺ currents (91.8±2.5%; n=8). The small residual outward and inward K⁺ currents recorded after astemizole perfusion are likely carried by endogenous outwardly-rectifying K⁺ channels constitutively expressed in HEK 293 cells, since it was insensitive to blockade by extracellular or intracellular application of astemizole and was identical to the K⁺ current recorded in untransfected HEK 293 cells (data not shown).

Figure 4.

Effect of extracellular perfusion with mizolastine on I_HERG1 heterologously expressed in HEK 293 cells stably transfected with HERG1 cDNA. Representative current traces recorded in the whole-cell configuration of the patch-clamp technique from a single HERG1-transfected HEK 293 cell. Records were obtained in control conditions, after 5 min perfusion with increasing concentrations of extracellular mizolastine (10 nM, 100 nM and 1 μM), and after 5 min exposure to 1 μM extracellular astemizole. The pulse protocol (shown at the bottom of the Figure) was the following: holding potential: −80 mV; test potentials from −80 mV to +40 mV in 20 mV steps (1 s duration); return potential: −100 mV. The bottom panels of the figure (corresponding to the time region indicated by the boxed area in the voltage pulse protocol) show the inward tail currents recorded upon repolarization to −100 mV on an expanded time and current scale.

In addition to the fundamental role of the I_Kr current in regulating action potential repolarization in cardiac cells, recent evidence suggests that HERG1 K⁺ channels are also expressed in other excitable tissues such as the brain (Wymore et al., 1997), in several neuroblastoma cell lines (Arcangeli et al., 1995), as well as in most tumor cell lines of different histogenesis (I_HERG1) (Bianchi et al., 1998). For this reason, the SH-SY5Y clone of human neuroblastoma cells was used to study the effect of mizolastine on mammalian cells constitutively expressing HERG1 K⁺ channels. In order to avoid contamination by other K⁺ currents also expressed in these cells, I_HERG1 was studied by means of a voltage-clamp protocol in which the cell was depolarized for 10 s to 0 mV, a membrane potential which fully activated I_HERG1 and completely inactivated the delayed rectifier K⁺ current (Arcangeli et al., 1995), and then repolarized to increasingly negative voltages (from 0 to −120 mV) for 365 ms. This voltage protocol allows to detect a K⁺-selective inward current displaying the biophysical properties of I_HERG1 (Arcangeli et al., 1995; Bianchi et al., 1998). Figure 5 shows representative current responses from a single SH-SY5Y human neuroblastoma cell exposed to increasing mizolastine concentrations (0.1–1 μM). Similarly to the results obtained in HEK 293 HERG1-transfected cells, 0.1 μM mizolastine causes approximately a 50% blockade of the inward HERG1 K⁺ currents; and a further increase in I_HERG1 blockade was observed by increasing the concentration of the second-generation H₁ receptor blocker to 1 μM. Pooling together the results obtained in five different cells in which experiments similar to that shown in Figure 5 were performed, the per cent of inhibition of the peak inward I_HERG1 of current recorded at −120 mV after 0.1 and 1 μM mizolastine, respectively, were 60.9±9.9% (n=5) and 78.5±2.6% (n=5) of control currents. It should be underlined that the intracellular solution for SH-SY5Y cells contained Cs⁺ and TEA to minimize the contribution of the delayed rectifier current during the long depolarizing pulse (10 s at 0 mV) required to activate I_HERG1, as previously described (Taglialatela et al., 1998b). Due to the intrinsic voltage-dependence and to the fast kinetics of dissociation for both TEA and Cs⁺, the slow inward currents recorded at negative potentials should be minimally affected by the blocking ions. In fact, similar inhibition by mizolastine was obtained using an intracellular solution for SH-SY5Y recording identical to that for HEK 293 cells (data not shown).

Figure 5.

Effect of mizolastine on I_HERG1 constitutively expressed in SH-SY5Y human neuroblastoma cells. Representative current traces recorded in the whole-cell configuration of the patch-clamp technique from an SH-SY5Y human neuroblastoma cell recorded in control conditions and after subsequent perfusion with 100 nM and 1 μM mizolastine (5 min each). The pulse protocol (shown at the bottom of the Figure) was the following: holding potential: −70 mV; depolarizing test potential: 0 mV for 10 s; hyperpolarizing return potentials: from 0 to −120 mV in −20 mV steps for 365 ms; return potential: −70 mV. Only approximately the last 450 ms of each pulses are shown, as indicated by the boxed area in the voltage pulse protocol.

Discussion

The present study aimed to investigate the possible interference of mizolastine, a novel second-generation H₁ receptor antagonist recently available in Europe, with the K⁺ currents carried by the HERG1 gene product. These channels, which provide the molecular basis for the cardiac repolarizing current I_Kr that plays a crucial role in controlling the duration of the cardiac action potential (Sanguinetti & Jurkiewicz, 1990; Sanguinetti et al., 1995), have been focused on since several reports have appeared in the literature of the last decade warning about the potential pro-arrhythmic effects of drugs such as some first- and second-generation antihistamines (astemizole and terfenadine, in particular), prokinetics, antipsychotics, and antidepressants that derive from HERG1 K⁺ channels blockade (Taglialatela et al., 1998a).

The results of this study suggest that mizolastine is able to block in some degree both constitutively and heterologously expressed HERG1 K⁺ channels. In fact, the benzimidazole blocked HERG1 K⁺ channels either heterologously expressed in amphibian (Xenopus oocytes) or in mammalian (HEK 293) cells, as well as constitutively present in SH-SY5Y human neuroblastoma cells. The existence of a 10 fold difference in the IC₅₀ values obtained for mizolastine blockade of HERG1 K⁺ channels expressed in Xenopus oocytes and in HEK 293 cells could be explained by the different properties of the amphybian membrane when compared to that of mammalian cells. In fact, the follicular cell layer and the vitelline membrane surrounding the oocytic membrane might to a certain degree impede the access of compounds at the level of the ion channels expressed in the amphybian membrane. Furthermore, it should be reminded that this difference is known to exist also for other HERG1 K⁺ channel blockers; in fact, the antiarrhythmic methanesulphonanilide dofetilide displayed an IC₅₀ value for HERG1 in whole Xenopus oocytes of 320 nM (Ficker et al., 1998), whereas in HEK-transfected mammalian cells this value is approximately 30 times lower, being 12 nM (Snyders et al., 1996). In addition, the potency of mizolastine in inhibiting HERG1 K⁺ channels appears to be at least 10 times lower than that of the well-known cardiotoxic molecule astemizole. In fact, the IC₅₀s for HERG1 inhibition in Xenopus oocytes were 3 μM for mizolastine and 300 nM for astemizole; furthermore, in HERG1-transfected HEK 293 cells, the IC₅₀s for the two second-generation antihistamines were 300 and 1 nM, respectively (Zhou et al., 1998). The lower potency of HERG1 inhibition by mizolastine could be at least in part ascribed to its faster dissociation rate from its receptor site on the channel when compared to astemizole.

The potential clinical relevance of a certain inhibitory action by mizolastine on HERG1 K⁺ channels is worth considering, since the drug reached a C_MAX of 0.3–1 μM during standard therapy (10–20 mg die⁻¹) (Chosidow et al., 1996; Chaufour et al., 1999), a concentration range which clearly showed significant HERG1 inhibitory action, particularly in eukaryotic cells. However, it should be noted that the steady-state levels achieved by the drug in therapeutic settings are much lower than the C_MAX, and that the potency for HERG1 K⁺ channels blockade by mizolastine is much lower than that exerted by the same drug at the level of H₁ receptors in vitro (the K_D for mizolastine antagonism at H₁ receptors is 1 nM) (Benavides et al., 1995).

In addition, the present study also shows that mizolastine is able to interfere with the K⁺ channels encoded by the Ether-à-go-go (EAG) and by the K_v3.1 genes; the potency of mizolastine in inhibiting EAG channels seems to be comparable to that displayed by the same molecule on the K⁺ channels encoded by HERG1, whereas K_v3.1 was only inhibited by 15% at the maximal drug concentration tested (10 μM). However, it should be underlined that, in several species including humans (Pardo et al., 1999), EAG K⁺ channels seem not to be expressed in cardiac tissue (Ludwig et al., 1994), but limited to the brain. It is therefore possible to speculate that EAG blockade is unlikely to cause an increased risk of arrhythmic events. On the other hand, the K⁺ channels encoded by the K_v3.1 gene also do not seem to play a major role in the control of cardiac excitability, even though one study has suggested the participation of K_v3.1-like channels in the ultrarapid delayed rectifier K⁺ current found in canine atrial myocytes (Yue et al., 1996). More importantly, mizolastine (up to 10 μM) was unable to affect the potassium channels encoded by the K_v2.1 gene, which is expressed in the heart and participates in the control of cardiac repolarization (Xu et al., 1999).

In consideration of the rare, yet serious adverse cardiac effects occurring with terfenadine and astemizole, European and US Health Regulating Authorities have recommended a careful examination of the cardiac safety of newly introduced drugs in preclinical and clinical studies, and in particular the study of their possible interference with ion channels controlling the cardiac action potential (EMEA, 1997). In vitro studies performed on guinea-pig ventricular myocytes have suggested that mizolastine is unable to lengthen cardiac action potential duration, even at concentrations 10–30 times above the standard therapeutic range (Biton et al., 1998). Furthermore, clinical studies in both healthy volunteers (Chaufour et al., 1999) and in allergic patients (Delauche-Cavallier et al., 1999) have shown that mizolastine showed no significant effect on cardiac repolarization. Therefore, the results of both animal and clinical studies argue against the possible occurrence of cardiotoxic events during standard therapy with the novel H₁ receptor blocker. However, it should be underlined that the recognition of the potential for cardiotoxicity exerted by terfenadine and astemizole required more than 10 years of extensive world-wide clinical use of these drugs. Interestingly, the initial warnings issued by the Regulatory Agencies against the potential risks associated to terfenadine and astemizole assumption were based on the reviews of only a couple of dozen clinical cases (among the millions of daily drug doses sold per year) in which the correlation between drug assumption and adverse cardiotoxic effects had been clearly demonstrated (Honig & Baraniuk, 1996).

A direct comparison of HERG1-blocking properties by different H₁ receptor blockers also allows to provide further insight into the structure-activity relationships for these molecules. In fact, it has been suggested that the HERG1 K⁺ channel-blocking properties of terfenadine and its structural analogue ebastine are at least in part related to the tertiary amine substituent of the molecule rather than to the presence of the piperidine ring, or to the aromatic ring structures conferring H₁ receptor blocking activity (Zhang, 1997); no correlation has been recently found between the ability to prolong the cardiac action potential duration, an effect possibly related to HERG1 K⁺ channels blockade, and the H₁ antagonistic activity by several antihistamines (Zhang, 1997; Taglialatela et al., 2000). Lipophilicity and bulkiness appear to be the two crucial parameters in the tertiary amine group substituent conferring HERG1 K⁺ channel-blocking capacity to the antihistaminic molecule. In fact, the second-generation antihistamine cetirizine, which appears to lack of significant HERG1-inhibitory properties (Taglialatela et al., 1998b), has polar and smaller substitutions at the nitrogen atom (amido and carboxyl groups, respectively); on the other hand, terfenadine, astemizole, and ebastine, the H₁ receptor antagonists most effective in inhibiting HERG1 K⁺ channels, have less polar and bulkier phenyl rings in the substituting side chains. Mizolastine, on the other hand, is characterized by a lower lipophylicity and thus by a low tissular penetration and distribution volume (Delauche-Cavallier et al., 1999). The lower lipophilicity of mizolastine could explain the lower potency of this novel congener in inhibiting HERG1 K⁺ channels when compared to astemizole or terfenadine. Furthermore, HERG1 K⁺ channels blockade by terfenadine (Roy et al., 1996) and astemizole (Taglialatela et al., 1998b), as well as by the antiarrhythmic dofetilide (Kiehn et al., 1996), seems to occur at a site located on the cytoplasmic side of the channel. However, the more hydrophilic profile, together with the relatively fast dissociation from the receptor site on the HERG1 K⁺ channels when compared to astemizole, seems to suggest the possible extracellular location of the blocking site for mizolastine on the channel protein.

In conclusion, the results of the present study suggest that the new second-generation H₁ receptor antagonist mizolastine displays a certain degree of inhibitory action on HERG1 K⁺ channels, which underlie the cardiac repolarizing current I_Kr and represent a major target for the arrhythmogenic action exerted by other antihistamines such as terfenadine and astemizole. Although caution should be exercised when extrapolating the possible in vivo consequences of the results achieved from in vitro studies, and in consideration of the relatively high concentrations of mizolastine required to inhibit HERG1 K⁺ channels, which appear to be above the steady-state plasma levels required to exert its therapeutical action, the present results might be of some clinical significance for those patients at risk of developing cardiac arrhythmias undergoing therapy with mizolastine.

Abbreviations

DMEM

Dulbecco's Modified Eagle Medium

DMSO

dimethylsulphoxide

EMEA

European Agency for the Evaluation of Medicinal Products

FCS

foetal calf serum

HEK293 cells

human embryonic kidney 293 cells

HERG1

Human Ether-à-go-go-Related Gene 1

I_Kr

rapid component of the cardiac repolarizing potassium current

MEM

modified eagle medium