Arsenic has been used by physicians for more than 2000 years to treat diverse diseases ranging from cancer to syphilis and tuberculosis (Zhu et al., 2002). It was abandoned as a therapeutic drug only in the last century because of its broad toxicity, which is responsible for its use and reputation as a poison. Arsenic is believed to cause widespread organ damage by combining with neighboring thiol groups present in many proteins, and exposure may result in severe long- or short-term illness (Miller et al., 2002). Despite its well known toxicity, arsenic was rediscovered recently by modern medicine and is now in use worldwide after arsenic trioxide (As₂O₃) was identified as the active ingredient of traditional Chinese medicines that produced dramatic remissions in patients with relapsed or refractory acute promyelocytic leukemia (APL) (Soignet et al., 2001; Zhang et al., 2001; Barbey et al., 2003).

The therapeutic use of As₂O₃ in the treatment of APL is burdened by its toxicity; multiple adverse effects, including QT prolongation, torsade de pointes tachycardias (TdP), and sudden cardiac death, have been reported (St. Petery et al., 1970; Ohnishi et al., 2000; Unnikrishnan et al., 2001; Westervelt et al., 2001; Zhang et al., 2001). Although the metabolic toxicity of As₂O₃ may be attributed to the exquisite sensitivity of accessible thiol groups in key enzymes or to enzymes such as pyruvate dehydrogenase, which uses the dithiol lipoic acid as a cofactor (Miller et al., 2002), it remains entirely unclear by what mechanisms As₂O₃ induces electrocardiogram abnormalities (Little et al., 1990; Ohnishi et al., 2000; Unnikrishnan et al., 2001; Westervelt et al., 2001; Chiang et al., 2002). More recently, hereditary QT prolongation [long QT syndrome (LQTS)] has been linked to mutations in cardiac ion channels, notably hERG and KvLQT1 (Keating and Sanguinetti, 2001). In addition, drug-induced, acquired LQTS is often caused by direct blockade of the cardiac potassium channel hERG by a wide variety of structurally diverse therapeutic compounds (Redfern et al., 2003). To our surprise, we found that As₂O₃ produced its effects not by direct block of cardiac ion channels, but indirectly, by increasing cardiac calcium currents and by inhibiting the processing of hERG protein in the endoplasmic reticulum (ER). This report represents the first example of drug-induced LQTS in which an acquired trafficking defect of a cardiac ion channel contributes to QT prolongation and TdP.

Materials and Methods

Western Blot Analysis. Stable cell lines and antibodies used in the present study have been described previously (Ficker et al., 2003). Cells were solubilized for1hat4°C in a lysis buffer containing 1% Triton X-100 and protease inhibitors (Complete; Roche Diagnostics, Indianapolis, IN). Protein concentrations were determined by the bicinchoninic acid method (Pierce Chemical, Rockford, IL). Proteins were separated on SDS polyacrylamide gels, transferred to polyvinylidene difluoride membranes, and developed using appropriate antibodies and ECL Plus (Amersham Biosciences, Piscataway, NJ). For quantitative analysis, chemiluminescence signals were captured directly on a Storm PhosphorImager (Amersham Biosciences) (Ficker et al., 2003).

Metabolic Labeling and Immunoprecipitation. HEK/hERG WT cells were starved for 30 min and pulse-labeled for 60 min in 100 to 150 μCi/ml [³⁵S]methionine/cysteine-containing medium. Cells were treated for 5 min at room temperature with 2 mM of the chemical cross-linker dithiobis(succinimidyl propionate) (Pierce Chemical). Cross-linking was quenched by the addition of glycine. Cells were harvested and lysed in a 0.1% Nonidet P-40 buffer in the presence of protease inhibitor (Ficker et al., 2003). Immunoprecipitation reactions were incubated overnight at 4°C and collected with Protein G Dynabeads (Dynal Biotech, Lake Success, NY). Samples were boiled in β-mercaptoethanol/SDS sample buffer to reverse cross-linking and to release immunoprecipitated proteins. Eluted samples were separated by SDS-polyacrylamide gel electrophoresis and analyzed with a Storm PhosphorImager (Amersham Biosciences) (Ficker et al., 2003).

In pulse-chase experiments performed without chemical cross-linking, image densities of fully and core-glycosylated hERG protein bands were determined after different chase periods by integrating pixel densities from equal areas after background subtraction. Image densities were normalized to the signal of the freshly synthesized, core-glycosylated hERG protein band isolated immediately after radiolabeling at t = 0. Pulse-chase data represent measurements from at least three independent experiments and are shown as mean ± S.E. After normalization, the time course of protein degradation as well as generation of fully glycosylated hERG protein can be compared directly between different experiments.

To quantify hERG/chaperone interactions at specific time points, image densities of core-glycosylated hERG protein bands were determined on autoradiograms after immunoprecipitation with anti-hERG, anti-Hsp90, and anti-Hsp/c70 antibodies as described. Image densities found in immunoprecipitations with anti-hERG antibody were used as a measure of total hERG protein present. Image densities of hERG protein bands isolated with antichaperone antibodies were normalized to image densities of hERG protein bands isolated with anti-hERG antibody to compute association quotients, which were used to assess time-dependent changes in hERG-chaperone interactions independently from small variations between different experiments. Data are shown as mean ± S.E. of three independent experiments.

Chemiluminescence Detection of Cell Surface hERG Protein. A hemagglutinin (HA) tag was inserted into the extracellular loop of hERG between transmembrane domains S1 and S2 (Ficker et al., 2003). Stably transfected HEK/hERG WT HA_ex cells were plated at 40,000 cells/well in a 96-well plate. After overnight incubation with As₂O₃, cells were fixed with ice-cold 4% paraformaldehyde, blocked by incubation with 1% goat serum, and incubated for 1 h with rat anti-HA antibody (Roche Diagnostics). After washing, horseradish peroxidase-conjugated goat anti-rat IgG (Jackson ImmunoResearch Laboratories Inc., West Grove, PA) and the double-stranded DNA stain SYBR Green (Molecular Probes, Eugene, OR) were added for 1 h (Myers, 1998; Margeta-Mitrovic et al., 2000). SYBR Green fluorescence was measured to determine cell numbers, after which chemiluminescent signals were developed using Super-Signal (Pierce Chemical) and captured in a luminometer. Luminescence signals in As₂O₃-treated wells were corrected for cell loss as measured by SYBR Green fluorescence with the data presented as normalized surface expression relative to control. For correction of cell loss, a standard curve of SYBR Green fluorescence was generated using four different amounts of cells per well (10, 20, 30, or 40 × 10³, n = 3). Chemiluminescence signals were corrected when cell loss ranged from 10 to 75%.

Patch-Clamp Recordings. HEK/hERG WT and L/hKv1.5 cells were recorded using patch pipettes filled with 100 mM K-aspartate, 20 mM KCl, 2 mM MgCl₂, 1 mM CaCl₂, 10 mM EGTA, and 10 mM HEPES, pH 7.2. The extracellular solution was 140 mM NaCl, 5 mM KCl, 1 mM MgCl₂, 1.8 mM CaCl₂, 10 mM HEPES, and 10 mM glucose, pH 7.4. To study the long-term effects of As₂O₃ on hERG or Kv1.5 currents, drug was added to cell lines for 16 to 20 h (overnight) before recording. I_Kr and I_Ks currents and action potentials were recorded in ventricular guinea pig myocytes cultured overnight in M199 medium using the following intracellular solution: 119 mM K-gluconate, 15 mM KCl, 3.75 mM MgCl₂, 5 mM EGTA, 5 mM HEPES, 4 mM K-ATP, 14 mM phosphocreatine, 0.3 mM Tris-GTP, and 50 U/ml creatine phosphokinase, pH 7.2. The extracellular solution was 132 mM NaCl, 4 mM KCl, 1.2 mM MgCl₂, 1.8 mM CaCl₂, and 5 mM HEPES, pH 7.4. L-type Ca²⁺ currents were blocked with 1 μM nisoldipine. The specific hERG/I_Kr blocker E4031 was used to isolate I_Kr currents. Cardiac sodium currents were recorded using the following extracellular solution: 40 mM NaCl, 55 mM N-methyl-d-glucamine, 20 mM CsCl, 5.4 mM KCl, 2 mM MgCl₂, 0.02 mM CaCl₂, 30 mM TEA Cl, 5 mM 4-aminopyridine, and 10 mM HEPES, pH 7.4. The intracellular solution was 120 mM CsCl, 2 mM MgCl₂, 1 mM CaCl₂, 11 mM EGTA, 10 mM HEPES, 10 mM glucose, and 1 mM Mg-ATP, pH 7.2. Cardiac calcium-current traces were recorded using the following extracellular solution: 137 mM NaCl, 5.4 mM CsCl, 1.8 mM MgCl₂, 1.8 mM CaCl₂, 10 mM glucose, and 10 mM HEPES, pH 7.4. The intracellular solution was 130 mM Cs MeSO₄, 20 mM TEA Cl, 1 mM MgCl₂, 10 mM EGTA, 10 mM HEPES, 4 mM Mg-ATP, 14 mM Tris-phosphocreatine, 0.3 mM Tris-GTP, and 50 U/ml creatine phosphokinase, pH 7.2. To analyze current densities, membrane capacitances were measured using the analog compensation circuit of an Axon 200B patch-clamp amplifier (Axon Instruments Inc., Union City, CA). pCLAMP software (Axon Instruments) was used for generation of voltage-clamp protocols and data acquisition. All current recordings were performed at room temperature (20-22°C).

Results

Arsenic Trioxide Inhibits Maturation of the Cardiac Potassium Channel hERG. During treatment of patients with APL, As₂O₃ concentrations peak rapidly between 5 and 10 μM immediately after drug administration before reaching lower steady-state levels between 0.1 and 1 μM, the therapeutically relevant concentration range (Shen et al., 1997). Neither 10 nor 100 μM As₂O₃ applied extracellularly reduced hERG tail current amplitudes recorded within 10 to 15 min (Fig. 1A). To exclude the possibility that access of As₂O₃ to an intracellular blocking site was hindered, we added 10 μMAs₂O₃ to the intracellular patch solution. hERG tail current amplitudes remained stable for up to 1 h after the start of the intracellular perfusion (data not shown). Rather than short-term block, As₂O₃ significantly increased hERG outward currents and accelerated current deactivation. As₂O₃-induced gating changes became more pronounced with longer exposure, and after 3 h, we found shifts in steady-state activation and inactivation of -8.9 and +21.7 mV, respectively (Fig. 1B). These changes were accompanied by faster activation kinetics. At +20 mV, we measured an activation time constant of 549 ± 107 ms in control myocytes and 301 ± 94 after 3-h exposure to As₂O₃ (n = 5). The accelerated activation time course together with the depolarizing shift of the steady-state inactivation curve are responsible for the increase in hERG outward currents observed during 2-s depolarizations. Current deactivation was fit with two exponential time constants that decreased at -50 mV from 526 ± 54 and 2738 ± 271 ms in control myocytes to 155 ± 32 and 938 ± 80 ms after 3-h exposure, whereas tail current amplitudes were not affected (Fig. 1B; n = 5). Inactivation kinetics remained unchanged (data not shown).

Because As₂O₃ did not block hERG directly and gating changes were complex, we considered the possibility of an As₂O₃-induced trafficking block, as we have reported previously for geldanamycin (Ficker et al., 2003). We exposed stably transfected HEK/hERG cells overnight to increasing concentrations of As₂O₃ and tested for effects on hERG processing. First, we found that hERG current was reduced in a concentration-dependent manner with an IC₅₀ of 1.5 μM as measured by analyzing tail current amplitudes on return to -50 mV after maximal activation with depolarizing voltage steps to +60 mV (Fig. 1, C and D). Second, we performed a Western blot analysis to test for changes in expression of hERG protein (Fig. 2A). It is known that hERG channels are synthesized in two forms: 1) an immature core-glycosylated protein of approximately 135 kDa that is localized to the ER, and 2) a fully glycosylated mature protein of approximately 155 kDa, which is transported to the cell surface (Zhou et al., 1998). Incubation with As₂O₃ for 24 h produced a time- and concentration-dependent decrease in the amount of mature, fully glycosylated hERG protein. Production of the fully glycosylated cell surface form of hERG was suppressed with an IC₅₀ of 1.5 μM, identical with the long-term effect on currents (Fig. 2B). At higher concentrations of As₂O₃, hERG channels were primarily present as the ER resident 135-kDa form.

The equivalent effects on hERG current and fully glycosylated protein suggested that surface expression of hERG was suppressed. To examine changes in surface expression, we used a chemiluminescence assay. HEK293 cells stably expressing hERG with an extracellular HA epitope tag inserted in the extracellular S1 to S2 loop (Ficker et al., 2003) were treated overnight with As₂O₃. As₂O₃ treatment caused a significant concentration-dependent reduction in hERG surface expression in line with the electrophysiological and biochemical experiments (Fig. 2C).

To examine the specificity of the As₂O₃ effects, we performed Western blot experiments with human Kv1.5 channels stably expressed in L cells. HKv1.5 channels are responsible for the ultra-rapid, delayed rectifier current found primarily in atrial myocytes (Nerbonne, 2000). The channel is synthesized as a core-glycosylated protein of approximately 68 kDa and matures into a fully glycoslyated cell surface protein of approximately 75 kDa (Ficker et al., 2003). When L/hKv1.5 cells were treated under steady-state conditions with As₂O₃, we detected a concentration-dependent increase in the fully glycosylated cell surface as well as the core-glycosylated protein form of hKv1.5 on Western blots (Fig. 3, A and B). The increase in surface expression was also reflected in electrophysiological experiments in which current densities increased significantly from 103.3 ± 25.9 pA/pF at 0 mV (n = 8) measured under control conditions to 733.8 ± 246.5 pA/pF (n = 5, p < 0.05, Student's t test) after overnight exposure to 10 μM As₂O₃ (Fig. 3C). In addition, we studied Kv4.3, the channel responsible for the cardiac transient outward current (Nerbonne, 2000) and found that current densities were not altered after incubation with As₂O₃ (data not shown).

Mechanism of hERG Suppression by As₂O₃. To explore the effects of As₂O₃ on hERG processing more directly, we performed pulse-chase experiments (Fig. 4A). Measured immediately after radiolabeling (t = 0 h), similar quantities of ER-resident, 135-kDa hERG were synthesized under control conditions and after preincubation with 3 μM As₂O₃. In control myocytes, within the first 6 h, approximately 50% of the initially synthesized 135-kDa protein progressed to the mature, fully glycosylated 155-kDa form. When HEK/hERG cells were pretreated with 3 μM As₂O₃ for 16 h before the pulse-chase experiment, maturation of the ER-resident form was reduced to approximately 25%, consistent with our steady-state experiments (Fig. 4B).

Because the cytosolic chaperones Hsp70 and Hsp90 are required for the biochemical maturation of hERG (Ficker et al., 2003), we determined whether As₂O₃ reduced steady-state expression levels of endogenous Hsp90 and/or Hsp70. Neither Hsp90 nor Hsp70 levels were significantly altered after overnight exposure to 3 μM As₂O₃ (data not shown). Because the trafficking block observed with As₂O₃ was reminiscent of effects induced by geldanamycin and radicicol, two specific inhibitors of Hsp90 (Roe et al., 1999; Ficker et al., 2003), we tested next whether As₂O₃ modified the association of hERG channels with Hsp70 and/or Hsp90 in coimmunoprecipitation experiments using a stable HEK/hERG WT cell line and radiolabeling to determine the formation of channel/chaperone complexes. After radiolabeling, hERG WT was immunoprecipitated in its core-glycosylated 135-kDa form using anti-hERG antibody. When the formation of hERGHsp70 chaperone complexes was assessed in immunoprecipitations with anti-Hsp/c70 antibody, we found Hsp70 in complex with newly synthesized, 135-kDa hERG protein (Fig. 4C). Likewise, Hsp90 could be isolated under control conditions in complex with the 135-kDa form of hERG. However, in the presence of As₂O₃, the formation of hERG/Hsp90 complexes was inhibited by approximately 50%, whereas the formation of hERG/Hsp70 complexes was reduced by approximately 20% (Fig. 4D). Taken together, these data provide evidence that the arsenic-induced trafficking block of hERG is probably caused by altered channel/chaperone interactions.

Arsenic Trioxide Reduces Native I_Kr Currents in Guinea Pig Ventricular Myocytes. To test whether our analysis in heterologous expression systems may be extended to ventricular cardiomyocytes, we first studied the effects of extracellularly applied As₂O₃ on action potentials evoked in freshly isolated guinea pig ventricular myocytes. We found that action potential duration (APD) was not affected by short-term As₂O₃ application (Fig. 5A). However, when myocytes were cultured overnight under control conditions or in the presence of 3 μM As₂O_3, APD₉₀ was significantly prolonged from 459 ± 15 to 1014 ± 127 ms, respectively (n = 5-6) (Fig. 5, B and C). Because our experiments in heterologous expression systems pointed toward hERG/I_Kr as the target for prolongation of APD₉₀, we used the specific blocker E4031 to evaluate the contribution of hERG/I_Kr currents to cardiac repolarization under control conditions and after overnight treatment with 3 μM As₂O₃. In these experiments, E4031 significantly prolonged APD₉₀ by 179 ± 34 ms under control conditions, whereas in As₂O₃ treated myocytes, APD₉₀ was not affected (Fig. 5C). Similar changes were observed in APD₃₀, a measure of repolarization more geared toward changes in depolarizing inward currents, with APD₃₀ increasing from 301 ± 22 to 774 ± 122 ms upon overnight exposure of myocytes to 3 μM As₂O₃.

Because experiments with E4031 indicated that a reduction of hERG/I_Kr currents may underlie the prolonged action potentials in As₂O₃-treated myocytes, we analyzed current densities of the cardiac rapid, delayed rectifier current I_Kr in voltage-clamp experiments. I_Kr currents were elicited during 650-ms depolarizations from a holding potential (HP) of -40 mV in cultured guinea pig cardiomyocytes treated overnight with 1.5 or 3 μM As₂O₃. I_Kr was isolated as an E4031-sensitive tail current upon return from +60 mV (Fig. 6, A and B1). As₂O₃ (1.5 and 3 μM) reduced tail current densities in guinea pig cardiomyocytes from 0.7 to 0.3 and 0.03 pA/pF, respectively (Fig. 6C). To test for specificity, slow, delayed rectifier currents I_Ks were elicited in the presence of 5 μM E4031 with 2.5-s step depolarizations (Fig. B2) and were not changed by drug exposure (Fig. 6D). Our data show that processing and maturation of cardiac I_Ks channels are not sensitive to As₂O₃ and reveal specificity in the interaction of As₂O₃ with different potassium channels. In addition, current densities of the cardiac inward rectifier current I_K1 did not change upon long-term exposure to As₂O₃. At -100 mV, we measured 12.0 ± 0.8 pA/pF under control conditions and 11.0 ± 0.7 pA/pF after overnight exposure to 3 μM As₂O₃ (n = 5).

Arsenic Trioxide Increases Cardiac Calcium Currents in Ventricular Myocytes. Because our analysis of APD₃₀ pointed toward the participation of depolarizing inward currents in As₂O₃-induced prolongation of cardiac repolarization, we also studied cardiac sodium and calcium currents. Sodium currents were elicited using a short depolarizing pulse protocol after a conditioning prepulse to -140 mV (Fig. 7A). I-V relationships recorded in control myocytes versus myocytes exposed overnight to As₂O₃ showed a shift in the activation of sodium currents to more hyperpolarized potentials and a modest increase in current density (Fig. 7B). To analyze potential changes in inactivation kinetics and late current amplitudes, we applied As₂O₃ to HEK293 cells stably expressing the human cardiac Na⁺ channel gene SCN5A. Neither parameter was affected by As₂O₃ (data not shown).

In contrast to the small changes in cardiac sodium currents, cardiac calcium current density more than doubled, from -5.2 ± 0.7 pA/pF (n = 6) in control myocytes at 0 mV to -17.7 ± 1.3 pA/pF (n = 10) in ventricular myocytes exposed overnight to 3 μM As₂O₃ (Fig. 7, C and D). The increase in current density was accompanied by small changes in the voltage-dependence of activation and an acceleration of the rate of inactivation. The large changes in calcium current are expected to prolong the cardiac action potential independent of the reported reduction of hERG/I_Kr currents. To demonstrate the contribution of calcium currents to the prolongation of cardiac action potentials, we used 25 nM nisoldipine, which shortened APD₉₀ from 880 ± 61 ms in myocytes treated overnight with 3 μM As₂O₃ to 686 ± 36 ms. The latter is close to APD₉₀ values of 495 ± 40 ms measured under control conditions (Fig. 8; n = 4). At a much higher concentration of 1 μM nisoldipine, APD₉₀ was further shortened to 254 ± 15 ms in control myocytes and to 304 ± 12 ms in myocytes incubated overnight with 3 μM As₂O₃ (Fig. 8; n = 4). The longer APD₉₀ values measured in As₂O₃-treated myocytes are consistent with inhibition of hERG/I_Kr trafficking.

Short-Term Effects of Arsenic Trioxide. Modifications of hERG/I_Kr currents by As₂O₃ proceed in two different time windows: 1) gating changes that occur in minutes, and 2) changes in surface expression that occur in hours and are caused by alterations in protein processing. To study time-dependent changes of cardiac calcium currents closely_, we analyzed current densities in myocytes exposed for short times to 3 μM As₂O₃. We found a small increase in cardiac calcium current density after a 1-h exposure (Fig. 9A). At the same time, APD₉₀ was prolonged from 581 ± 18 to731 ± 65 ms (n = 5-6) (Fig. 9, C and D). After 3 h, calcium currents were further increased with APD₉₀ prolonged from 494 ± 78 to 725 ± 68 ms (n = 8-13) (Fig. 9B). Because changes in surface expression of ion channels are a function of slow protein turnover rates, we speculated that the fast changes in cardiac repolarization seen after 1- and 3-h exposure to As₂O₃ may be caused either by fast gating changes in hERG currents and/or by the fast increase in calcium-current densities demonstrated in our experiments. To decide between these possibilities, we assayed the contribution of hERG/I_Kr currents to cardiac repolarization in myocytes exposed for 1 h to As₂O₃ using 5 μM of the specific hERG/I_Kr blocker E4031. We found that hERG/I_Kr contributed to a similar extent in both control and As₂O₃-treated myocytes. E4031 (5 μM) increased APD₉₀ by 211 ± 18 ms in control and by 284 ± 32 ms in As₂O₃-treated myocytes (n = 5-6; ΔAPD₉₀ values were not significantly different, Student's t test, p < 0.05) (Fig. 9D). Likewise, APD₃₀ was increased after 1-h exposure to As₂O₃. However, E4031 increased APD₃₀ by 205 ± 21 ms in As₂O₃-treated myocytes, whereas in control myocytes, APD₃₀ was prolonged only by 85 ± 12 ms (n = 5-6; ΔAPD₃₀ values significantly different at p < 0.05 level, Student's t test) (Fig. 9E). Hence, our analysis suggests that the contribution of hERG/I_Kr currents to cardiac repolarization was not altered by fast gating changes and that the prolongation of cardiac repolarization seen on initial exposure to As₂O₃ is most likely the result of an increase in calcium current amplitudes.

Short-Term Effects of Membrane-Permeable Phenyl Arsine Oxide. Given our results with short-term incubations, it is surprising that As₂O₃ applied with the extracellular perfusate for up to 15 min was not able to produce short-term changes in cardiac action potentials recorded at room temperature (Fig. 5A). One explanation may be that the uptake of As₂O₃ into cardiac myocytes is slow at room temperature. To overcome slow uptake, we added 10 μM As₂O₃ to the intracellular patch solution and recorded cardiac action potentials. We found that 10 μM As₂O₃ applied for 10 min via the patch pipette prolonged APD₉₀ in three of six cells from 616 ± 59 to 866 ± 98 ms and APD₃₀ from 410 ± 31 to 641 ± 49 ms (n = 3; both APD₉₀ and APD₃₀ were significantly prolonged at p < 0.05, Student's t test). In marked contrast, both APD₉₀ and APD₃₀ were stable for more than 15 min after the start of intracellular perfusion with control solution (Fig. 10A). Thus, fast changes in cardiac calcium current seem to depend on the access of As₂O₃ to intracellular sites. In another test of fast modification of cardiac calcium currents, we applied the membrane-permeable As₂O₃ analog phenyl arsine oxide (PAO) with the extracellular perfusate. In electrophysiological experiments, PAO applied at 10 μM with the extracellular perfusate increased calcium-current amplitudes within minutes (Fig. 10B). At the same time, current inactivation kinetics became faster. Both changes in cardiac calcium currents were similar to what has been observed upon long-term exposure to As₂O₃. Taken together, our data suggest that calcium currents are modified in a fast process that is limited only by As₂O₃ gaining access to the sarcoplasm of cardiac ventricular myocytes.

Discussion

In the present study, we provide evidence that As₂O₃ prolongs the action potential of guinea pig ventricular myocytes via two independent molecular mechanisms. As₂O₃ increases cardiac calcium currents, which regulate the plateau phase of the cardiac action potential but also reduces surface expression of the cardiac potassium current hERG/I_Kr, which is crucial to later stages of cardiac repolarization. The effects on current amplitudes were preceded by changes in hERG gating kinetics induced immediately upon exposure to As₂O₃. Enhanced outward currents and accelerated deactivation kinetics have been reported as a hallmark of hERG modulation by radical oxygen species (ROS) (Taglialatela et al., 1997; Berube et al., 2001) and are compatible with the well-documented property of As₂O₃ to induce oxidative stress by increasing ROS production (Liu et al., 2001). However, ROS-induced gating changes of hERG are ambiguous in their effects on cardiac repolarization in that faster activating hERG currents allow for larger outward currents during the plateau phase of the cardiac action potential with a tendency to shorten the cardiac action potential, whereas faster deactivation kinetics are expected to reduce hERG currents during cardiac repolarization and prolong the QT, as observed with As₂O₃. It is important the ROS-induced acceleration of current deactivation was not sufficient to prolong the cardiac action potential as judged from our current clamp experiments and seemed to be compensated by the concomitant ROS-induced increase in hERG outward currents. After 1-h exposure to As₂O₃, we detected an increase in cardiac calcium-current amplitudes accompanied by a prolongation of action potential duration. Changes in calcium current amplitudes became more pronounced with longer exposure times and seemed to be diminished in short-term experiments by slow uptake of As₂O₃ into the sarcoplasm. The changes in calcium currents were paralleled by an As₂O₃-induced block of hERG processing, which reduced hERG export to the cell surface. As expected, changes in cell-surface expression of hERG could be detected only after long-term exposure to As₂O₃ for several hours, because changes in surface expression are limited by the slow turnover rate of hERG channel protein of approximately 11 h (Ficker et al., 2003). Because patients are treated for several weeks with daily injections of As₂O₃, such a dosing regimen may be approximated best by long-term exposure of cardiomyocytes to As₂O₃.

Long-term exposure to As₂O₃ suppresses trafficking of hERG, and we propose that this mechanism underlies the reduction in I_Kr and the prolongation of action potential duration in ventricular myocytes. Trafficking block of hERG was observed at clinically relevant concentrations of 0.1 to 1.5 μM As₂O₃ (Shen et al., 1997). This report is the first to show acquired trafficking block as the basis for QT prolongation and TdP. It is consistent with other reports that electrocardiogram changes develop gradually and are observed most often after multiple rounds of treatment with As₂O₃ (Ohnishi et al., 2000; Unnikrishnan et al., 2001). It has been reported in guinea pig ventricular myocytes that cardiac action potential duration was significantly prolonged only with a delay of hours after feeding or intravenous infusion of As₂O₃; short-term effects were minimal (Chiang et al., 2002). Furthermore, a study in intact rabbits demonstrated long-term effects of As₂O₃ on cardiac action potential duration and QT interval that developed slowly over several weeks with no short-term effects reported (Wu et al., 2003).

Given the clinically available data (Ohnishi et al., 2000; Unnikrishnan et al., 2001), data from two different animal models (Chiang et al., 2002; Wu et al., 2003), and the present data acquired in stably transfected hERG/HEK cells, all of which point toward slow mechanisms of As₂O₃ action, it is surprising that short-term block of hERG and KvLQT/mink currents heterologously expressed in Chinese hamster ovary cells has been reported (Drolet et al., 2004). An explanation for the difference is not at hand. However, it is important to note that short-term application of As₂O₃ did not prolong action potential duration in guinea pig ventricular myocytes. Moreover, long-term exposure to As₂O₃ blocked neither cardiac I_Ks nor I_K1 currents. In addition, hERG/I_Kr current amplitudes were not altered at times when calcium currents were modified.

The slow reduction in hERG/I_Kr processing and surface expression is a direct consequence of the interference of As₂O₃ with the binding of two essential cytosolic chaperones, Hsp70 and Hsp90, to hERG (Ficker et al., 2003). As a result, processing of the ER-resident hERG protein into the mature cell-surface form of hERG is inhibited. Because As₂O₃ interacts preferentially with neighboring thiol groups, it is of interest that many Hsp90 proteins possess a strictly conserved thiol pair in the middle segment of the chaperone protein. It is conceivable that modification of these thiols by As₂O₃ renders Hsp90 nonfunctional (Matsumoto et al., 2002). Direct inhibition of Hsp90 by As₂O₃ is also consistent with the trafficking block of hERG induced by two specific inhibitors of Hsp90 function, geldanamycin and radicicol (Roe et al., 1999). Because Hsp90 is a previously unrecognized target for As₂O₃, our results may also further the understanding of its anticancer activity.

Although the trafficking block of hERG would be sufficient to explain the increased propensity of patients with APL treated with As₂O₃ for QT prolongation and TdP, it is not too surprising that As₂O₃, which not only modifies thiol groups but also increases the production of reactive oxygen species, exerts multiple effects on cardiac ion channels. Most prominent is the large increase in cardiac calcium currents. The modification of calcium currents by As₂O₃ is different from the acquired processing defect of hERG in that the underlying process is faster, possibly the result of a direct enzymatic modification of the calcium channel itself or of an essential accessory protein (e.g., in phosphorylation/dephosphorylation reactions).

As demonstrated with low doses of nisoldipine, the increase in calcium currents has important therapeutic implications because it suggests that calcium-channel blockers may be well-suited to normalize QT prolongation during As₂O₃ therapy. The sodium channel blocker mexiletine has already been used as part of a prophylactic therapy to reduce cardiac complications in patients with APL (Ohnishi et al., 2000). However, the use of sodium-channel blockers in the treatment of arrhythmias is controversial because of their recognized proarrhythmic properties (Balser, 2001). We provide the first experimental evidence that calcium-channel blockers may be more efficacious. In addition, previous proposals to control electrolyte imbalances in patients with APL have now gained special importance because it is well known that hERG currents are strongly modulated by extracellular potassium concentrations (Sanguinetti et al., 1995; Ohnishi et al., 2000). Taken together, the concerted action of the proarrhythmic drug As₂O₃ on two cardiac ion channel systems represents a new paradigm for acquired long QT syndrome, which may help us to understand why QT prolongation and TdP seem to be so prevalent among patients treated for acute promycelocytic leukemia.