Regulation of HERG potassium channel activation by protein kinase C independent of direct phosphorylation of the channel protein

Abstract

Objective: Patients with HERG-associated long QT syndrome typically develop tachyarrhythmias during physical or emotional stress. Previous studies have revealed that activation of the beta-adrenergic system and consecutive elevation of the intracellular cAMP concentration regulate HERG channels via protein kinase A-mediated phosphorylation of the channel protein and via direct interaction with the cAMP binding site of HERG. In contrast, the influence of the alpha-adrenergic signal transduction cascade on HERG currents as suggested by recent reports is less well understood. The aim of the present study was to elucidate the biochemical pathways of the protein kinase C (PKC)-dependent regulation of HERG currents. Methods: HERG channels were heterologously expressed in Xenopus laevis oocytes, and currents were measured using the two-microelectrode voltage clamp technique. Results: Application of the phorbol ester PMA, an unspecific protein kinase activator, shifted the voltage dependence of HERG activation towards more positive potentials. This effect could be mimicked by activation of conventional PKC isoforms with thymeleatoxin. Coexpression of HERG with the beta-subunits minK or hMiRP1 did not alter the effect of PMA. Specific inhibition of PKC abolished the PMA-induced activation shift, suggesting that PKC is required within the regulatory mechanism. The PMA-induced effect could still be observed when the PKC-dependent phosphorylation sites in HERG were deleted by mutagenesis. Cytoskeletal proteins such as actin filaments or microtubules did not affect the HERG activation shift. Conclusion: In addition to the known effects of PKA and cAMP, HERG channels are also modulated by PKC. The molecular mechanisms of this PKC-dependent process are not completely understood but do not depend on direct PKC-dependent phosphorylation of the channel.

Time for primary review 22 days.

1 Introduction

Repolarization of the cardiac action potential is regulated by different potassium currents [1]. The rapid component of the delayed rectifier potassium current, I_Kr, initiates repolarization and terminates the plateau phase of the action potential. The human ether-a-go-go-related gene (HERG) [2] encodes the voltage-gated potassium channel underlying I_Kr[3,4]. HERG channels are a primary target for the pharmacological management of arrhythmias with class III antiarrhythmic agents [5], and several class III antiarrhythmic drugs such as dofetilide, amiodarone, azimilide, or BRL-32872, act via inhibition of HERG potassium channels [4,6–8]. Reduction of HERG currents due to mutations in HERG or via excessive blockade by antiarrhythmic or non-antiarrhythmic drugs produces chromosome 7-linked congenital long QT syndrome (LQTS-2) and acquired long QT syndrome (LQTS), respectively [3,9–12]. Both forms of LQTS are associated with delayed cardiac repolarization, prolonged electrocardiographic QT intervals and a high risk for the development of ventricular ‘torsade de pointes’ arrhythmias and sudden cardiac death [13,14].

The clinical observation that patients with LQTS typically develop arrhythmias during physical or emotional stress [15] suggests a link between adrenergic stimulation and HERG potassium channel activity. Recent studies have revealed that activation of the beta-adrenergic system and consecutive elevation of the intracellular concentration of cAMP as second messenger regulates HERG channels through protein kinase A-mediated phosphorylation of the channel protein and via direct interaction with the cAMP binding site of HERG [16–20]. These effects are, at least in part, mediated by the β1-adrenoceptor, as demonstrated by Karle et al. [21].

In contrast, the influence of the alpha-adrenergic signal transduction cascade on HERG currents is less well understood. Adrenergic activity and subsequent stimulation of the Gαq-protein coupled cardiac α1A-adrenoceptor causes activation of the phosphatidyl inositol-specific phospholipase Cβ (PLCβ). Its substrate phosphatidyl-4,5-bisphosphate (PIP2), a membrane phospholipid, is hydrolyzed to the intracellular second messengers 1,4,5-inositol-trisphosphate (IP3) and diacylglycerol (DAG). IP3 leads to mobilization of intracellular calcium, and DAG is the physiological activator of protein kinase C (PKC), a serine-/threonine-dependent kinase. The biochemical action of PKC is limited by the activity of the protein serine-/threonine-phosphatases (PP) PP1, PP2A, and PP2B. At least 10 isoforms of PKC have been identified and grouped into three subtypes: conventional cPKCs (α, βI, βII, γ) that require calcium and/or DAG for activation; novel nPKCs (δ, ε, η, θ) that can be activated by DAG, but are insensitive to calcium; and atypical aPKCs (ζ, and ι/λ), which are unresponsive to calcium or DAG [22]. Under experimental conditions, phorbol esters can be used to activate cPKC and nPKC isoforms by binding to the cysteine-rich motifs of the protein, but they do not act on aPKC isoforms.

We have previously demonstrated that phorbol 12-myristate-13-acetate (PMA), a potent, but unspecific activator of protein kinase C and other protein kinases, reduces HERG currents in guinea pig cardiomyocytes and Xenopus oocytes via an activation shift [23]. In the same year, Barros et al. reported that thyreotropin releasing hormone (TRH) and consecutive activation of the TRH receptor causes a shift in the HERG activation curve towards more positive potentials, an effect that could be mimicked by PMA [24]. Furthermore, coexpression of the α1A-adrenoceptor (α1C-adrenoceptor according to the old classification) [25] with HERG channels and stimulation with the α1A-adrenergic agonist phenylephrine caused current reduction and a shift in the HERG activation curve in Xenopus oocytes [26].

To elucidate the biochemical pathways of the putative PKC-dependent regulatory process in detail, we generated mutations of all 18 PKC-specific phosphorylation sites in HERG to test for direct PKC-dependent phosphorylation of the HERG protein. The signal transduction mechanism was investigated by application of activators and inhibitors of PKC and PKA.

2 Methods

2.1 Site-directed mutagenesis

Consensus PKC phosphorylation sites with the amino acid sequence [Ser or Thr]-Xaa-[Arg or Lys] in HERG were identified using PROSITE software. The program identified 18 sites in the HERG wild type protein. The serine or threonine residues of the PKC phosphorylation sites (S26, T74, T162, T174, S179, S250, S278, S354, T371, T526, S606, S636, T670, S890, T895, S918, S960, and T1133) were replaced with alanine residues to eliminate PKC-mediated phosphorylation. This was performed by site-directed mutagenesis with the QuikChange kit (Stratagene, La Jolla, CA). The PCR products between unique anchor sites were sequenced (SeqLAB, Göttingen, Germany), and fragments between unique restriction sites were removed by use of the following combinations of restriction enzymes: Hind III, Nco I for S26A, T74A, T162A, T174A, S179A; Nco I, BstE II for S250A, S278A, S354A, T371A; BstE II, Xho I for T526A, S606A, S636A, T670A; Xho I, FseI for S890A, T895A, S918A; and Fse I, BamH I for S960A and T1133A. The mutated fragments were ligated into the original HERG template and sequenced again. To generate multiple residue mutants, fragments with single mutations were cut out and sequentially subcloned into the respective templates, yielding the following four constructs: construct 1 (S26A, T74A, T162A, T174A, and T179A); construct 2 (S250A, S278A, S354A, and T371A); construct 3 (T526A, S606A, S636A, and T670A); construct 4 (S890A, T895A, S918A, S960A and T1133A). The cDNAs of the constructs were verified by sequencing. Electrophysiological measurements were carried out to probe for functional expression of the constructs in Xenopus oocytes. The constructs 2, 3, and 4 displayed typical HERG potassium currents, whereas after expression of construct 1 no currents could be detected. We found that the T74A point mutation prevented the functional expression of construct 1. Thus, we generated a construct 1′, which lacks the T74A mutation. This clone could be expressed similar to the other constructs. The HERG ΔPKC clone lacking all functional PKC phosphorylation sites (except T74) was generated by introducing the restriction fragment with the mutations of construct 1′ into construct 2. This resulted in construct 1′2. Then the fragment containing the point mutations of construct 4 was cut out from its vector and inserted into the vector carrying construct 1′2 using the restriction enzymes named above. Finally, the mutated fragment of construct 3 was cut out and inserted into this intermediate construct. The resulting clone was named HERG ΔPKC. In the same manner we generated the HERG 18 M clone lacking all 18 phosphorylation sites (including T74A). Each of the intermediate constructs and the HERG ΔPKC clone were verified by DNA sequencing.

2.2 Expression of HERG channels in Xenopus laevis oocytes

Procedures for in vitro transcription and oocyte injection have been published previously [16]. Briefly, cRNAs of HERG wild type [2], minK [27], hMiRP1 [28], and the constructs generated in this study were prepared with the mMESSAGE mMACHINE in vitro transcription kit (Ambion, Austin, USA) after linearization with EcoRI (Roche Diagnostics, Mannheim, Germany). Stage V–VI defolliculated Xenopus oocytes were injected with 46 nl of cRNA per oocyte 2–10 days prior to use. The investigation conforms with the Guide for the Care and Use of Laboratory Animals published by the US National Institute of Health (NIH Publication No. 85-23, revised 1996).

2.3 Electrophysiology and data analysis

The two-microelectrode voltage-clamp configuration was used to record currents from Xenopus laevis oocytes as described previously in detail [6]. Statistical data are expressed as mean±standard deviation (n = number of experiments). We used paired and unpaired Student's t-tests to compare statistical significance of the results. A P-value<0.05 was considered to be statistically significant.

2.4 Solutions and drug administration

Voltage clamp measurements of Xenopus oocytes were performed in a solution containing (in mM) 5 KCl, 100 NaCl, 1.5 CaCl₂, 2 MgCl₂, and 10 HEPES (pH 7.4). Current and voltage electrodes were filled with 3 M KCl solution. Bisindolylmaleimide I, cytochalasin B, PMA (phorbol-12-myristate-13-acetate), and wortmannin (all from Calbiochem) were dissolved in DMSO to a stock solution of 10 mM and stored at −20°C. KT 5720 (Calbiochem), thymeleatoxin (Calbiochem), and Ro-32-0432 (Calbiochem) were dissolved in DMSO to a stock solution of 1 mM and stored at −20°C. Genistein (Sigma) was dissolved in DMSO to a stock solution of 100 mM and stored at −20°C. On the day of experiments, aliquots of the stock solution were diluted to the desired test concentrations with the bath solution. All measurements were carried out at room temperature (20–22°C).

3 Results

3.1 The PMA-induced shift of the HERG WT activation curve is mediated by PKC and PKA

To measure HERG channel activation, we recorded HERG currents using a two-step protocol (see Fig. 1A): A variable first step (test pulse) was recorded at different potentials from −100 to +80 mV (increment 10 mV) for 400 ms, followed by a second step to −120 mV (400 ms) to measure inward tail currents. The holding potential was −80 mV. This protocol was used for all electrophysiological measurements performed in this study. The tail current amplitude depends on the preceding test pulse potential and is a measure of channel activation. The activation curves of HERG WT control currents had a mean half maximal activation voltage V_1/2 of −3.5±4.9 mV (n = 29). First, we examined the effects of the phorbol ester PMA, an unspecific protein kinase activator that stimulates cPKCs, nPKS, and PKC-related kinases (PRKs), as well as other protein kinases [22]. Typical current traces are shown in Fig. 1A (control measurements) and Fig. 1B (after application of 100 nM PMA for 30 min). The maximum outward current amplitude during the test pulse (at 0 mV) was reduced significantly by 55.5±23.4% (n = 8). Fig. 1C displays the normalized and inverted tail current amplitudes during the constant pulse as a function of the preceding test pulse potentials, reflecting the activation curves (control measurement and recording after incubation with 100 nM PMA for 30 min). The half maximal activation voltage V_1/2 was shifted by 31.2±8.4 mV from −3.5±4.9 to 27.7±8.7 mV (n = 29), which is the biophysical basis for the outward current reduction [23]. In addition, the peak tail current following the test pulse to 0 mV was reduced by 63.8±24.6% (n = 10), whereas maximum peak tail currents were only reduced by 9.5±9.2% (n = 10; Table 1, row A). The reduction of the maximum peak tail currents was less pronounced compared to the reduction of peak tail currents measured after the 0-mV pulse, since the shift of the activation curve by 31.2 mV still allows pronounced activation at very positive potentials. Furthermore, PMA treatment caused an acceleration of the deactivation time constant, as illustrated by faster tail current decay after PMA application (Fig. 1B). The time constants of deactivation (tau_deact) were analyzed by fitting the tail current decay (after the 80-mV pulse) monoexponentially, revealing that tau_deact was accelerated by 32.0±7.4% (Table 1, row A) in this series of experiments.

Since PMA is a rather unspecific protein kinase activator (Fig. 1D), we examined whether specific inhibition of protein kinases could reduce the effects caused by 100 nM PMA (30 min). Pretreatment of oocytes with the specific PKC inhibitor Ro-32-0432 (3 μM) for at least 4 h significantly reduced the shift of the half-maximal activation voltage (ΔV_1/2) to 3.5±4.1 mV (n = 5; Fig. 2A–C,G), and values for current decrease and acceleration of the deactivation time constant were markedly less pronounced (Table 1, row B). In addition, preincubation with the PKC inhibitor bisindolylmaleimide I (1 and 10 μM) also reduced the shift of the activation curve to 21.5±8.4 mV (n = 7) and −0.2±2.6 mV (n = 4), respectively (Fig. 2G), and current decrease as well as deactivation time constants were significantly reduced or even abolished (Table 1, rows C,D). The drugs used in this series of experiments have been previously shown to inhibit PKC under similar experimental conditions [24,29,30]. These results demonstrate that HERG channels are regulated via protein kinase C-dependent pathways (Fig. 2H). To demonstrate the effect of the PKC inhibitors Ro-32-0432 and bisindolylmaleimide I in the absence of PMA, HERG currents were measured under control conditions (without preincubation with PKC inhibitors) and after incubation with 3 μM Ro-32-0432 or 10 μM bisindolylmaleimide I (the maximum concentration used in this study) for 90 min. A longer incubation time was not tolerated by the cells, which is most likely due to membrane damage by the microelectrodes. This approach revealed that Ro-32-0432 (Table 2, line A) caused weak reduction of HERG tail currents (following a 0-mV pulse), which is in line with other control experiments (Table 2, rows C–E). The (weak) current reduction is most likely due to current rundown after the relatively long incubation period of 90 min. The deactivation time constant and the half-maximal activation voltage (ΔV_1/2=−2.8±2.1 mV; n = 4) were virtually not altered by Ro-32-0432. When bisindolylmaleimide I (10 μM) was applied, the deactivation time constant was not significantly changed (Table 2, row B; n = 4), and the half-maximal activation voltage was only shifted by −5.3±3.7 mV towards more negative potentials (n = 5). However, a pronounced tail current decrease by 51.4±9.4% (tail current after the 0-mV step; 90 min) and 59.1±7.6% (maximum peak tail current; 90 min) occurred rapidly upon drug application, reaching steady-state conditions within 10 min. The rapid time course of this effect (Table 2, row B) indicates that this effect is most likely due to direct pharmacological inhibition of HERG currents. Pharmacological blockade of ether-a-go-go-related gene potassium channels by bisindolylmaleimide I has been described previously by Schledermann et al. [30].

To investigate whether other protein kinases are involved as well, experiments with additional protein kinase inhibitors were carried out. Application of the protein kinase A-inhibitor KT 5720 [31] (2.5 μM) together with PMA (100 nM) for 30 min reduced both the activation shift (ΔV_1/2=1.2±4.7 mV; n = 7) and the tail current reduction (Fig. 2D–F,G; Table 1, row E), which is consistent with known PKA-dependent regulation of HERG activation [16]. However, the deactivation time constant was still accelerated by PMA, indicating that the deactivation process is not PKA-dependent (Table 1, row E). Control experiments, where 2.5 μM KT 5720 was applied for 30 min without PMA show that KT 5720 (Table 2, row C) caused weak reduction of HERG tail currents due to current rundown, as observed in most control experiments (Table 2, rows A,C–E). The deactivation time constant (Table 2, row C) and the half-maximal activation voltage (ΔV_1/2=2.2±0.6 mV; n = 5) were not markedly affected. The influence of protein tyrosine kinases and PI3 kinase was also determined. As shown in Fig. 2G, neither preincubation with the broad range tyrosine kinase inhibitor genistein [32–36] (100 μM) nor pretreatment with the PI3 kinase inhibitor wortmannin [37,38] (10 μM) attenuated the shift of the activation curve [ΔV_1/2=34.4±3.6 mV (n = 4), and ΔV_1/2=31.4±5.5 (n = 3), respectively]. In addition, neither tail current reduction nor deactivation time constant were markedly less reduced compared to PMA alone (Table 1, rows F,G). Control measurements with 100 μM genistein and 10 μM wortmannin (incubation time 90 min) gave similar results in comparison with KT 5720 and Ro-32-0432 (Table 2, rows D,E; ΔV_1/2=1.5±3.2 mV (genistein, n = 4) and ΔV_1/2=4.7±4.9 mV (wortmannin, n = 4), respectively).

3.2 The shift of the HERG activation curve is not mediated by direct PKC-dependent phosphorylation of the HERG channel protein

To differentiate direct effects of protein kinase C on the channel protein from intermediate actions within the signal transduction cascade, we performed site-directed mutagenesis to generate mutated HERG channels that lack consensus PKC phosphorylation sites (S26, T74, T162, T174, S179, S250, S278, S354, T371, T526, S606, S636, T670, S890, T895, S918, S960, and T1133; Fig. 3). Four to five point mutations were combined in four different HERG constructs, and these clones were expressed in Xenopus oocytes to perform electrophysiological measurements (Table 3). The constructs 2, 3, and 4 generated in this study resulted in functional potassium channels with biophysical properties similar to those of HERG WT. Construct 1 showed no HERG potassium currents. However, when the point mutation T74A was removed from the construct (giving construct 1′), typical HERG currents could be recorded. In addition to these intermediate constructs, we generated the HERG ΔPKC clone where all putative PKC-dependent phosphorylation sites in HERG (except T74) were mutated. When we inserted the T74A point mutation into HERG ΔPKC (construct HERG 18M), no HERG currents could be measured. The HERG ΔPKC clone showed biophysical properties almost identical to those of wild type HERG channels (compare Figs. 4A and 1A). The half-maximal activation voltage obtained during control measurements was shifted towards more negative potentials compared to HERG WT (Fig. 4C; Table 3). The average V_1/2 during control experiments yielded −12.2±2.9 mV (n = 10). Application of 100 nM PMA shifted the half-maximal activation voltage of HERG ΔPKC by 28.1±5.6 mV to 16.1±6.7 mV (Fig. 4A–C; Table 3; n = 10). This value was almost identical to ΔV_1/2 obtained with HERG WT channels (31.2 mV). In addition, HERG tail currents (following the 0-mV step) were significantly reduced, as observed with HERG WT channels (compare Table 1, row A and Table 4, row A), and the deactivation time constant was markedly accelerated (Table 4, row A). In the following set of experiments we preincubated the oocytes expressing HERG ΔPKC channels with Ro-32-0432 (3 μM) for at least 4 h prior to use. In these experiments, the HERG activation curve was only shifted by 7.2±5.4 mV after addition of 100 nM PMA (n = 5; Fig. 4D). Simultaneous application of KT 5720 (2.5 μM) and PMA abolished the shift (ΔV_1/2=−1.4±1.4 mV; n = 4; Fig. 4D). Pretreatment with these drugs almost abolished the reduction of HERG tail currents (measured after the 0-mV step) caused by PMA application, and the deactivation time constants were markedly less accelerated (Table 4, rows B,C). The intermediate constructs were investigated as well. As displayed in Table 2, the baseline midpoints in the activation curves were more negative in constructs 3 and 4 compared to HERG WT channels. The activation shift caused by PMA was attenuated in construct 3. Other parameters were not markedly different from HERG WT channels. These results illustrate that PMA does not act via direct PKC-dependent channel phosphorylation. Instead, it is likely that intermediate signal transduction proteins mediate this effect (Fig. 4E).

3.3 Specific activation of conventional PKC-isoforms by thymeleatoxin induces a shift in the HERG activation curve independently of direct channel phosphorylation by PKC

To investigate the role of protein kinase C in the regulation of HERG current activation in more detail, we examined the acute effects of the specific cPKC activator thymeleatoxin (Fig. 5, Table 5). When 100 nM thymeleatoxin was applied into the bath for 60 min, the activation curve of HERG WT was shifted by 24.5±7.9 mV from −3.5±3.6 to 20.9±8.5 mV (n = 11), HERG tail currents following the 0-mV pulse were reduced, and the deactivation time constant was accelerated (Table 5, row A). Thus, it is likely that conventional PKC isoforms act on HERG currents, since the PMA-mediated effects were mimicked by the cPKC activator thymeleatoxin. Furthermore, it was investigated whether thymeleatoxin acts via PKC-dependent HERG channel phosphorylation. Thymeleatoxin caused a HERG ΔPKC activation curve shift of 23.4±3.2 mV (n = 4; Fig. 5C,D), HERG ΔPKC tail current reduction and marked acceleration of the deactivation time constant under the same experimental conditions as described above (Table 5, row B), suggesting that the thymeleatoxin-induced activation shift caused by activation of cPKC isoforms is not mediated by direct PKC-dependent phosphorylation of the HERG channel protein (Fig. 5E).

3.4 Characterization of the intracellular environment required for the PMA-induced HERG activation shift

Cytoskeletal actin filaments or microtubules might modulate the HERG activation shift via localization of PKC and the putative additional factors to their respective substrates (including HERG) or anchoring proteins, as it was observed in the regulation of Kv1.5 channels by protein kinases [40]. To elucidate the role of the cytoskeleton in the PMA-induced modulation of HERG wild type activation kinetics, we preincubated oocytes for 4–9 h with cytoskeletal modifying agents and examined the response to 100 nM PMA. First, we applied cytochalasin B (1 μM), a compound that inhibits the polymerization of actin monomers to actin filaments. The half-maximal activation voltage after the preincubation period was not significantly altered compared to non-incubated oocytes (−3.3±1.6 mV), and 100 nM PMA (30 min) caused a shift of the activation curve of 37.2±6.3 mV (n = 5). In addition, pretreatment of oocytes with colchicine (10 μM), an agent that disrupts microtubules, had no effect on V_1/2 under control conditions (V_1/2=−3.9±4.9 mV). When 100 nM PMA were added for 30 min, V_1/2 was shifted by 25.4±5.2 mV (n = 8). This value was not significantly different from the shift induced by PMA alone (31.2 mV). Moreover, values for tail current reduction (measured after the 0-mV step) and deactivation time constant acceleration obtained after preincubation with cytochalasin B (Table 6, row A) or colchicine (Table 6, row B) were similar compared to experiments where PMA was applied without additional drugs (Table 1, row A). Thus, the regulation of HERG currents by PMA is not significantly modulated by cytoskeleton integrity.

Finally, we coexpressed HERG wild type (WT) channels with the regulatory β-subunits minK (minimal K⁺ channel) [27] and hMiRP1 (human minK-related peptide 1) [28]. Under control conditions, the half maximal activation voltage did not differ markedly from HERG WT channels (−6.3±1.4 mV for minK, and −7.8±1.6 mV for hMiRP1, respectively). After incubation with 100 nM PMA for 30 min, V_1/2 was shifted by 32.4±3.2 mV (n = 4; minK) and 29.4±9.4 mV (n = 7; hMiRP1). Furthermore, tail currents following the 0-mV step were reduced and deactivation time constants were accelerated largely similar to experiments with PMA alone (Table 6, rows C,D; Table 1, row A). Thus, neither minK nor hMiRP1 caused pronounced changes in HERG current regulation by PMA.

4 Discussion

In the present study, we investigated the effects of protein kinase C activation on HERG currents. The results of the present study imply a novel pathway of HERG channel regulation by protein kinase C independently of direct phosphorylation of the channel protein (Fig. 6A), in addition to known PKA-dependent pathways involving direct HERG channel phosphorylation (Fig. 6B) [16,18]. This was demonstrated by comparing the action of protein kinase activators and specific protein kinase inhibitors on HERG wild type channels with their effects on mutant HERG ΔPKC channels lacking PKC-dependent phosphorylation sites.

One important finding of this study is the involvement of PKC in the complex mechanism by which HERG channel activation kinetics are modulated. In the presence of the unspecific protein kinase activator PMA, the HERG wild type activation curve was shifted, HERG tail currents (measured after a 0-mV voltage step) were reduced and the deactivation was accelerated. Preincubation with PKC inhibitors markedly attenuated this effect, suggesting a role of protein kinase C in the regulation of HERG currents. We also investigated the significance of putative direct channel phosphorylation by protein kinase C. The activation and deactivation kinetics as well as current amplitudes of mutant HERG channels lacking PKC-dependent phosphorylation sites (HERG ΔPKC) were still modulated by PMA, indicating that PKC does not act via direct phosphorylation of HERG channel proteins at the PKC consensus sites (Fig. 6A). Furthermore, the PMA-induced effect on HERG channels could be mimicked by the specific cPKC activator thymeleatoxin, demonstrating that conventional PKC isoforms are involved in the regulatory mechanism. It is reasonable to assume that additional intermediate signal transduction proteins mediate the effect, since direct PKC-dependent phosphorylation is not involved. The time course of PMA to cause (ion channel) phosphorylation can be obtained from a study by Karle et al. [41]. Kir2.1b potassium currents expressed in Xenopus oocytes were inhibited by 100 nM PMA with relatively slow onset, and maximum block was reached after 30 min. This effect was due to direct PKC-dependent channel phosphorylation, since after mutagenesis of the PKC phosphorylation sites in Kir2.1b the PMA-induced effect was completely abolished. Thus, we may speculate that PMA exerts its (indirect) effects on HERG currents via PKC-dependent phosphorylation of intermediate factors/proteins, since the time course of the PMA-effects on HERG currents observed in this study are similar (30 min). The putative intermediate factors might consecutively act on HERG channels with rapid time course. Moreover, PMA is likely to act via PKC-dependent pathways, since the effect could be attenuated by PKC inhibitors. However, in contrast to PMA, there are no studies investigating precisely the time course of PKC inhibition by these PKC inhibitors. Therefore, we have to adhere to the indirect conclusions that can be obtained from previous studies [24,29,30]. These data demonstrate that in general preincubation for several hours is necessary to achieve pronounced inhibition of PKC activity by Ro-32-0432 and bisindolylmaleimide I. Furthermore, bisindolylmaleimide I did not significantly block the PMA-induced effect when the drugs were applied simultaneously without preincubation with the PKC inhibitor [23]. Thus, it is likely that the time courses of PKC inhibition and block of the PMA-induced effect do not differ. The hypothesis that PMA as well as the PKC inhibitors act through PKC modulation (although independently of direct HERG protein phosphorylation) is further supported by the following aspects. First, previous studies have shown that PMA activates protein kinase C in Xenopus oocytes [24,37,39,41–44]. Second, specific inhibition of PKC in Xenopus oocytes by bisindolylmaleimide I [24] and by Ro-32-0432 [29] has been demonstrated previously. Third, application of the specific cPKC activator thymeleatoxin mimicked the PMA effects. In summary, it is concluded that the PMA-induced effects are mediated through protein kinase C activation.

Protein kinase C-dependent modulation of multiple cardiac ion channels has been investigated before, revealing that in most cases the regulatory mechanism involves direct phosphorylation of the channel (Table 7). There are, however, few ion currents where PKC does not act directly on the channel protein. First, I_Ks and the underlying minK subunit are regulated by PKC in a dual manner involving initial current increase and subsequent current reduction [45]. This effect can be modulated by protein kinase A in an additive manner, i.e. activation of PKA and subsequent current increase prevents additional PKC dependent effects, and vice versa. To date, the molecular mechanism of minK regulation by PKC is not completely clear, but it is speculated that multiple phosphorylation sites are involved. In the present study, all PKC-dependent phosphorylation sites in HERG have been eliminated except for T74, because the T74A mutation prevented channel expression and functional electrophysiological studies. Thus, although 17 phosphorylation sites were not functionally relevant, we cannot totally exclude that the remaining T74 site is responsible for the PKC-dependent HERG current modulation. Secondly, the investigation of Kv1.5 regulation revealed that PKC primarily phosphorylates the accessory channel subunit Kvβ1.2, thereby reducing Kv1.5 currents [44]. In the present study, neither of the putative regulatory HERG subunits minK and hMiRP1 significantly modulated the PMA effect. Thus, these proteins can be ruled out as intermediate factors mediating the PKC-induced HERG activation shift. In addition, based on our results, neither actin filaments nor microtubules seem to be of functional relevance, as demonstrated for the modulation of Kv1.5 by protein kinase A [40].

The physiological relevance of our findings is supported by the clinical observation that the incidence of cardiac arrhythmias is markedly increased by stress, especially among patients with congenital or acquired long QT syndrome [15]. Thus, the effect of PKC activation on HERG channels observed in this study could be caused by adrenergic stimulation under physiological and pathophysiological conditions in the human heart. Under in vitro conditions, coexpression of HERG channels and α1A-adrenoceptors in Xenopus oocytes and consecutive receptor stimulation with phenylephrine was shown to cause a shift of V_1/2 towards more positive potentials [26]. Furthermore, in HEK 293 cells expressing HERG and α1A-adrenoceptors, α-adrenergic stimulation led to a significant current decrease and a small depolarizing shift in the voltage dependence of activation [52].

In summary, we could demonstrate that HERG channel activation is modulated by activation of protein kinase C in addition to known direct PKA-dependent phosphorylation. Direct phosphorylation of the channel protein by protein kinase C is not involved in this novel PKC-dependent regulatory mechanism (Fig. 6). Further investigations are necessary to elucidate cell type specific factors that modulate this pathway, and to identify putative intermediate proteins mediating the effect.

Acknowledgements

We thank Dr M.T. Keating for providing the HERG clone, Dr A.M. Brown for donating the minK clone, and Dr S.A. Goldstein for providing the hMiRP1 clone. We gratefully acknowledge the excellent technical support of K. Güth, S. Lück, and R. Bloehs. This work was supported by grants from the University of Heidelberg (AiP+F) and from the Novartis-Foundation to D.T., and by grants from the Deutsche Forschungsgemeinschaft (project KI 663/1-1 to J.K.; project KA 1714/1-1 to C.A.K.).

References