Oxidative Stress-Induced Afterdepolarizations and Calmodulin Kinase II Signaling

Abstract

In the heart, oxidative stress caused by exogenous H₂O₂ has been shown to induce early afterdepolarizations (EADs) and triggered activity by impairing Na current (I_Na) inactivation. Since H₂O₂ activates Ca²⁺/calmodulin kinase II (CaMKII), which also impairs I_Na inactivation and promotes EADs, we hypothesized that CaMKII activation may be an important factor in EADs caused by oxidative stress. Using the patch clamp and intracellular Ca (Ca_i) imaging in Fluo-4 AM-loaded rabbit ventricular myocytes, we found that exposure to H₂O₂ (0.2–1mM) for 5-15 min consistently induced EADs that were suppressed by the I_Na blocker tetrodoxtin (TTX, 10 μM), as well as the I_Ca,L blocker nifedipine. H₂O₂ enhanced both peak and late I_Ca,L, consistent with CaMKII-mediated facilitation. By prolonging the AP plateau and increasing Ca influx via I_Ca,L, H₂O₂-induced EADs were also frequently followed by DADs in response to spontaneous (i.e. non I_Ca,L-gated) SR Ca release after repolarization. The CaMKII inhibitor KN-93 (1 μM, n=4), but not its inactive analog KN-92 (1 μM, n=5), prevented H₂O₂-induced EADs and DADs, and the selective CaMKII peptide inhibitor AIP (2 μM) significantly delayed their onset. In conclusion, H₂O₂-induced afterdepolarizations depend on both impaired I_Na inactivation to reduce repolarization reserve, and enhancement of I_Ca,L to reverse repolarization, both facilitated by CaMKII activation. Our observations support a link between increased oxidative stress, CaMKII activation and afterdepolarizations as triggers of lethal ventricular arrhythmias in diseased hearts.

Introduction

Reactive oxygen species (ROS) are generated as natural byproducts of normal oxygen metabolism and play important roles in cell signaling. However, under pathological conditions, such as heart failure and ischemia-reperfusion, ROS levels can become elevated and predispose the heart to arrhythmias^{1 2}. Oxidative stress induced by exposure to hydrogen peroxide (H₂O₂) and other agents has been shown to induce early afterdepolarizations (EADs), delayed afterdepolarizations (DADs) and triggered activity (TA), ^3-6, primarily attributed in previous studies to impaired inactivation of Na current (late I_Na) ^{4, 5}. However, H₂O₂ also affects other ion channel and transporters, including the L-type Ca current (I_Ca,L) ^7-9, K currents ¹⁰, ryanodine receptors (RyR) ^{11, 12}, sarcoplasmic reticulum (SR) Ca pump SERCA2a ¹³, and the Na-Ca exchanger (I_NCX) ^{14, 15}, which also may influence EADs and TA.

Recently, H₂O₂ has been shown to activate Ca/calmodulin (CaM) kinase II (CaMKII) in cardiac myocytes and other cells ^16-18, by direct oxidation of paired Met residues (Met281/282) in the regulatory domain. It is notable that binding of Ca²⁺/CaM is required to expose the redox sites in the regulatory domain in order for oxidation to persistently activate CaMKII ¹⁸. In addition, ventricular myocytes isolated from transgenic mice overexpressing CaMKIV have been shown to exhibit EADs and TA. Increased endogenous CaMKII activity accompanied EADs and VT/VF observed during telemetry in this mouse model¹⁹. Alterations in I_CaL properties have been implicated as a major factor by which CaMKII activation promotes EADs ^{19, 20}, but a recent study demonstrated that CaMKII activation also increases the late I_Na ²¹, similar to H₂O₂. CaMKII inhibition has been shown to suppress EADs and triggered activity. In a genetic heart failure model, for example, CaMKII inhibition prevented EADs, triggered activity and improved mortality¹⁹. The ability of pharmacologic agents such as clofilium to induce EADs in isolated rabbit ventricular myocytes ²² was also suppressed by CaMKII inhibition.

Taken together, this evidence led us to hypothesize that activation of CaMKII by oxidative stress may be an important factor in arrhythmogenic effects of H₂O₂. Accordingly, we used isolated patch-clamped rabbit ventricular myocytes and Ca_i imaging to analyze the ionic mechanism(s) underlying H₂O₂-induced afterdepolarizations in the absence and presence of CaMKII inhibition. Our findings support a direct link between oxidative stress, CaMKII activation and the genesis of TA due to afterdepolarizations.

Methods

Cell isolation

Ventricular myocytes were enzymatically isolated from adult rabbit hearts. Briefly, hearts were removed from adult New Zealand white rabbits (2-3 kg) anesthetized with intravenous pentobarbital, and perfused retrogradely in Langendorff fashion at 37°C with nominally Ca-free Tyrode's solution containing ∼1.4 mg/ml collagenase (Type II; Worthington) and 0.1 mg/ml protease (type XIV, Sigma) for 25-30 minutes. After washing out the enzyme solution, the hearts were removed from the perfusion apparatus, and swirled in a culture dish. The Ca concentration was slowly increased to 1.8 mM and the cells were stored at room temperature and used within 8 hours.

Intracellular Ca measuremen

Myocytes were loaded with the Ca indicator Fluo-4 by incubating them for ∼30 minutes in bath solution containing 4 μM Fluo-4 AM (Molecular Probes) and 0.016% (wt/wt) pluronic (Molecular Probes), after which the cells were washed and placed in a heated chamber on an inverted microscope. Ca_i fluorescence was recorded using an Andor Ixon Charge Coupled Device (CCD) camera (Andor Technology) operating at ∼100 frames per second with a spatial resolution of 512 × 180 pixels. The fluorescence intensity was recorded in arbitrary units.

Patch Clamp Methods

Myocytes were patch-clamped using the whole-cell configuration of the patch clamp technique. For AP recordings, patch pipettes (resistance 2–4 MΩ) were filled with pipette solution containing (in mM): 110 K-Aspartate, 30 KCl, 5 NaCl, 10 HEPES, 0.1 EGTA, 5 MgATP, 5 creatine phosphate, 0.05 cAMP, pH 7.2 with KOH. In some experiments, the CaMKII inhibitor peptide AIP (2-10 μM) (Biomol, or Sigma) was added directly to the pipette solution. The cells were superfused with standard Tyrode's solution containing (in mM): 136 NaCl, 5.4 KCl, 0.33 Na₂PO₄, 1.8 CaCl₂, 1 MgCl₂, 10 glucose and 10 HEPES, pH 7.4 adjusted with NaOH. To isolate I_Ca,L, patch pipettes (resistance 2–4 MΩ) were filled with pipette solution containing (in mM): 110 Cs-Aspartate, 30 CsCl, 5 NaCl, 10 HEPES, 0.1 EGTA, 5 MgATP, 5 creatine phosphate, 0.05 cAMP, pH 7.2 with KOH, and the cells were superperfused with a modified Tyrode's solution in which KCl was replaced by CsCl. H₂O₂ (0.2-1 mM) was added directly to the bath superfusate. Nifedipine (Sigma), KN-93 and KN-92 (Biomol) were dissolved in dimethyl sulfoxide (DMSO) as stock solution before diluting into the superfusate solution at the final concentration. The maximum DMSO concentration was <1/500 (v/v). Chemicals and reagents were purchased from Sigma unless indicated. Voltage or current signals were measured with an Axopatch 200B patch-clamp amplifier controlled by a personal computer using a Digidata 1200 acquisition board driven by pCLAMP 8.0 software (Axon Instruments, Foster City, CA). Action potentials were elicited with 2 ms, 2-4 nA square pulses at pacing cycle lengths (PCL) of 6 sec.

All experiments were performed at 34-36°C. Data are presented as mean ± SEM. Statistical significance was assessed using unpaired Student's t-tests, with p < 0.05 considered significant.

Results

H₂O₂-induced EADs and DADs in adult rabbit ventricular myocytes

APs were recorded from isolated rabbit ventricular myocytes using whole-cell current clamp mode. After AP duration (APD) and morphology reached steady state, cells were superfused with 0.2-1 mM H₂O₂. The generation of EADs by H₂O₂ was highly dependent on the pacing cycle length (PCL). Typically, at long PCLs (e.g. 6 sec), EADs occurred with every AP, and at short PCL (e.g. 1 s), no EADs occurred. In the intermediate range, EADs occurred irregularly. Therefore, to elicit EADs most reliably, we used a PCL of 6 sec throughout. EADs appeared after an average exposure time of 6.9 ± 1.1 min with 0.2 mM H₂O₂ (n=10), and 6.6 ± 0.8 min with 1 mM H₂O₂ (n=12). As shown in Fig. 1A, EADs could be irregular, single or multiple with an oscillating membrane potential before repolarization (Fig. 1 A & B). DADs were also observed after prolonged perfusion of H₂O₂ and occasionally triggered APs (Fig 1B, right panel). DADs often occurred following a previous AP with an EAD (Fig. 1B middle panel and 1C), presumably because the prolonged APD allowed maintained I_Ca,L to overload the SR with Ca. Prolonged perfusion (15-20 min) of H₂O₂ eventually caused gradual depolarization of the resting membrane potential to less than -40 mV and the cells became inexcitable.

Fig. 1. Afterdepolarizations and triggered activity during H₂O₂ exposure in isolated rabbit ventricular myocytes.

A. Action potentials (APs) were elicited with 2 ms-2nA pulses under current clamp conditions at pacing cycle length (PCL) of 6 sec. Values of consecutive APD₉₀ are plotted over time. H₂O₂ (200 μM) was perfused continuously as indicated by the horizontal bar above the plot. EADs are indicated by the sudden and dramatic increases of APD₉₀ (at ∼ 6 min after H₂O₂ application). Insets show APs under control condition, and after H₂O₂ with an EAD. B. Examples of afterdepolarizations and TA during exposure to 1 mM H₂O₂, including multiple oscillatory EADs (left panel), an EAD followed by a DAD (middle panel), and an EAD followed by a DAD causing TA (right panel). C. Successive APs recorded during H₂O₂ exposure, illustrating the irregular occurrence of EADs (asterisks) and associated DADs (arrows).

Ionic mechanisms of H₂O₂-induced EADs and DADs

We next used pharmacologic interventions to analyze the ionic and cellular mechanism(s) underlying H₂O₂-induced aftedepolarizations. As shown in Fig. 2A, H₂O₂-induced EADs were reversibly suppressed by the selective Na current blocker TTX (10 μM), which also shortened APD. Ranolazine (20 μM), a more selective blocker of late I_Na, also eliminated H₂O₂-incuded EADs and shortened APD (data not shown), consistent with previous reports by Song et al ^{4, 5} implicating late I_Na as playing a key role. Unlike H₂O₂, anemone toxin II (ATX), an agent that selectively delays the late phase inactivation of I_Na ²³, failed to induce frank EADs, producing instead only small (1-2 mV) irregular oscillations during phase 2 of the AP. Note that APD was prolonged by ATX to a greater extent than after H₂O₂ (Fig. 2B), yet no EADs were observed. This finding indicates that activation of late I_Na is not, by itself, sufficient to produce EADs under these experimental conditions, implying that other actions of H₂O₂ are also required.

Fig. 2. The involvement of both I_Na and I_Ca,L in H₂O₂-induced EAD.

A. The specific I_Na blocker TTX (10 μM) reversibly suppressed EADs and shortened APD. B. ATX (30 nM), a selective activator of persistent I_Na, prolonged APD but did not generate cause frank EADs with a distinct upstroke. C. The amplitude of H₂O₂-induced EADs depended on their take-off potentials (arrows). D. The I_Ca,L blocker nifedipine (Nif, 10 μM) suppressed the EAD upstroke, although AP plateau remained prolonged, unless TTX (10 μM) was also added.

Fig. 2C shows that EAD amplitude after H₂O₂ depended on its take-off potential during repolarization, such that the more repolarized the take-off potential, the larger the EAD amplitude. This relationship parallels the voltage dependence of I_Ca,L reactivation and the I_Ca,L window current ^{24, 25}. Consistent with this interpretation, the selective I_Ca,L blocker nifedipine (10 μM) eliminated EADs (Fig. 2D), even though APD remained markedly prolonged due to the effect of H₂O₂ on late I_Na (as indicated by shortening of APD with subsequent application of 10 μM TTX or 20 μM ranolazine). Taken together, these results suggest that at least two actions of H₂O₂ are required for EAD formation: activation of late I_Na to reduce repolarization reserve (i.e. prolong APD) and modification of I_Ca,L to enhance its reactivation properties so as to generate the EAD upstroke.

Whereas the effects of H₂O₂ on late I_Na have been well documented in previous studies ^{4, 5}, the reported effects of H₂O₂ on I_Ca,L have been variable. Therefore, we examine how H₂O₂ affected I_Ca,L in rabbit ventricular myocytes under our experimental conditions. Fig. 3A and B show the time course of both peak I_Ca,L and the residual pedestal I_Ca,L at the end of a 300ms voltage clamp to 0 mV in a representative myocyte during exposure to 1 mM H₂O₂. Both peak and pedestal I_Ca,L increased, from 7.3 ± 0.8 to 12.1 ± 1.8 pA/pF (n=5) and from 0.5 ± 0.1 to 3.2 ± 0.3 pA/pF (n=5), respectively. To examine how these changes affect I_Ca,L during the repolarization phase of the AP, we performed AP clamp experiments before and after exposure to 1 mM H₂O₂. Fig. 3C shows that H₂O₂ induced a prominent I_Ca,L hump during the late phase of the AP plateau, consistent with the timing of the EAD (compare to Fig. 1 &2). The recordings shown in Fig. 3 B & C represent the nifedipine-sensitive (subtracted) current, indicating that the increase of the inward current (including the hump in the late phase) is mostly due to I_Ca,L, although minor contamination by the Na-Ca exchange current (I_NCX) cannot be excluded.

Fig. 3. Enhancement of I_Ca,L by H₂O₂.

A. Time course of the peak and residual pedestal (ped) I_Ca,L during a 300 ms voltage clamp pulse to 0 mV (voltage protocol shown in B). B. Voltage clamp pulse (above) and superimposed current traces showing I_Ca,L before (black) and ∼5 min after perfusion of 1mM H₂O₂ (red). The difference current is shown in the bottom trace. C. Same as B, but with an AP clamp waveform replacing the square voltage clamp pulse.

The role of Ca_i cycling

After H₂O₂ exposure, we often observed a transient inward current (I_ti) following repolarization to -80 mV (Fig. 4A and inset), consistent with spontaneous SR Ca release as a result of the enhanced Ca influx via I_Ca,L causing SR Ca overload. To investigate the relationship to the EAD-induced DADs observed in Fig. 1, we simultaneously imaged Ca_i in Fluo-4-AM loaded myocytes (Fig. 4B-D). During H₂O₂-induced EADs, Ca_i remained elevated (Fig. 4C) or increased (Fig. 4D), consistent with additional SR Ca release due to reactivation of I_Ca,L. The line scan in Fig. 4D shows that the rise in Ca_i during the second EAD was not spatially homogeneous. This may indicate that Ca-induced Ca release from the SR, in addition to I_Ca,L enhancement, contributed to EAD formation by enhancing I_ti (since H₂O₂ has been reported to enhance Na-Ca exchange ¹⁴). Fig. 4D documents that after repolarization, a spontaneous Ca_i wave originating from the center of the cell and propagating towards both ends was associated with a DAD. This can be attributed to increased SR Ca loading during the preceding AP, which was markedly prolonged by EADs.

Fig. 4. Transient inward current (I_ti) and Ca_i transients during H₂O₂–induced EADs and DADs.

A. Similar to Fig. 3B, but with an I_ti following repolarization to -80 mV in the presence of H₂O₂ (expanded in bottom trace). B. AP (top), simultaneous whole-cell Ca_i transient (middle), and a line-scan image along the long axis of the myocyte (bottom) before H₂O₂. C & D. Same following H₂O₂ exposure. In C, two EADs occur, leading to persistent elevation in Ca_i and additional Ca²⁺ release (r) during the second EAD. In D (different cell), two EADs (associated with additional I_Ca,L-gated SR Ca²⁺-release) are followed by a DAD due to a non-voltage gated spontaneous Ca_i wave (w) starting from the middle of the cell (*).

To further investigate the role of Ca_i cycling in H₂O₂-induced afterdepolarizations, we examined the effects of pre-loading myocytes with BAPTA-AM to buffer changes in Ca_i (Fig. 5A). We also pre-treated myocytes with thapsigargin (0.2 μM) and ryanodine (10 μM) to suppress SR Ca cycling (Fig. 5B). Both interventions completely prevented EADs and DADs during exposure to 1 mM H₂O₂ (Fig. 5), indicating the intact Ca_i cycling plays a critical role. However, APD remained prolonged under these conditions, probably due to reduced Ca-induced inactivation of I_Ca,L when the intracellular Ca_i transient was suppressed ²⁶.

Fig. 5. Dependence of H₂O₂-induced EADs on intact Ca_i cycling.

A. Time course of APD₉₀ in a myocyte preloaded with BAPTA-AM (4 μM) for 30 min exposure to 1 mM H₂O₂. Representative APs at points a (control), b (∼10 min in H₂O₂), and c (H₂O₂ + TTX) are shown below. B. Same, but in a myocyte pre-treated with 10 μM Ryanodine (Ry) + 200 nM Thapsigargin (Th) prior to H₂O₂.

The role of CaMKII signaling

Similar to the effects of H₂O₂, Ca²⁺-dependent CaMKII activation has been shown to modify I_Na inactivation, enhance I_Ca,L, and promote EADs ^19-21. CaMKII could either be activated by Ca_i, due to the effects of H₂O₂ on Ca_i cycling proteins, or directly by oxidative stress, as shown recently in cardiac myocytes ^{16, 18}. The Ca_i-dependence is still consistent H₂O₂-mediated activation of CaMKII, since persistent activation of CaMKII by oxidation requires Ca-CaM binding to CaMKII to expose its redox sites ¹⁸. We examined the effects of CaMKII inhibition to test these possibilities. Pretreatment with the CaMK inhibitor KN-93 (1 μM) prevented the emergence of EADs during exposure to H₂O₂ (1 mM for up to 20-30 min) in 4 myocytes (Fig. 6A), compared to non-pretreated myocytes in which H₂O₂ (1 mM) consistently induced EADs within 5-10 min of H₂O₂ exposure (Fig. 1A, 6B). In addition, KN-93 also suppressed EADs after they were induced by H₂O₂ perfusion (Fig. 6B).

Fig. 6. Suppression of H₂O₂–induced EADs by the CaMKII inhibitor KN-93 and AIP.

A. Time course of APD₉₀ in a myocyte treated with the CaMKII inhibitor KN-93 prior to exposure to 1 mM H₂O₂. APs under control (a), in the presence of KN-93 and after perfusion with KN93 + H₂O₂ for 9 min (c) and 19 min (d) are shown in the lower panels. B. Same, but with KN-93 applied after EADs were induced by H₂O₂. In both protocols, KN-93 suppressed EADs. C. Bar graph showing the average H₂O₂ (200 μM) perfusion time to cause EAD appearance in the absence and presence of 2 μM AIP.

In contrast, 1μM KN-92, an inactive analog of KN-93, was ineffective (Fig. 7A and B), with EADs appearing after an average of 7.5 ± 1.2 min (n = 5) after exposure to 1 mM H₂O₂ (n=5). BAPTA-AM, however, remained effective at suppressing H₂O₂-induced EADs in the presence of KN-92 (Fig. 7B).

Fig. 7. Failure of KN-92, an inactive analog of KN-93, to prevent H₂O₂–induced EADs.

A. Same protocol as in Fig. 6A, but with KN-92 initially in place of KN-93. KN-92 failed to prevent EADs during exposure to 1 mM H₂O₂, which were subsequently partially suppressed by KN-93. APs at points a (control), b (∼ 2 min after KN-92), c (KN-92 + H₂O₂) and d (KN-93 + H₂O₂) are shown below. B. Same as A, but applying BAPTA-AM to suppress EADs after KN-92 + H₂O₂.

Since both KN-93 and KN-92 have nonspecific effects, such as blockade of Ca and K currents ^{22, 27, 28}, we also examined the effects of the selective CaMKII inhibitor peptide AIP (2 μM). In 8 myocytes, AIP added to patch pipette dialyzing the cytoplasm delayed the appearance of EADs during exposure to 200 μM H₂O₂, from 6.6 ± 0.8 (n=10) to 13.9 ± 2.7 mins (Fig 6C) (p<0.05). Moreover, AIP dialysis also prolonged APD in a concentration-dependent manner, presumably due to nonspecific peptide effects, such that higher AIP concentrations were ineffective at preventing H₂O₂-induced EADs.

Discussion

In the present study, we investigated the mechanisms by which oxidative stress caused by exposure to H₂O₂ induces afterdepolarization and triggered activity in rabbit ventricular myocytes. The novel findings are: 1) I_Ca,L modification, in addition to I_Na modification, plays a crucial role in EAD generation by H₂O₂; 2) CaMKII activation, either directly via oxidative stress, or indirectly via elevated Ca_i, is critical for these effects; 3) By enhancing SR Ca loading and promoting spontaneous Ca_i waves, H₂O₂–induced EADs also cause DADs, providing a direct link between these two types of afterdepolarizations which compounds the arrhythmogenic potential of oxidative stress.

Ionic mechanisms of EADs and DADs induced by H₂O₂

Two conditions are required to generate an EAD. First, repolarization reserve must be reduced during phases 2 & 3 of the AP, by an increase in inward current, a decrease in outward current, or both. Second, once repolarization reserve has been compromised, reactivation of I_Ca,L window current, and/or additional SR Ca release augmenting Ca-sensitive inward currents such as I_NCX, must be sufficiently powerful to reverse repolarization and generate the EAD upstroke. In ventricular muscle, inward currents influencing repolarization reserve include I_Na, I_CaL, and I_NCX, while outward currents include the rapid and slow delayed rectifier K⁺ currents (I_Kr, and I_Ks), the transient outward current (I_to) and the inward rectifier potassium current (I_K1). Previous studies analyzing the mechanisms of H₂O₂-induced EADs have implicated impaired I_Na inactivation as the primary mechanism reducing repolarization reserve ^{4, 5}. However, H₂O₂ has also been reported to affect other currents and transporters. In the present study, we show that in addition to modification of I_Na, modification of I_Ca,L by H₂O₂ plays a key role in EAD genesis. Not only did H₂O₂ substantially increase the peak amplitude of I_Ca,L, but it also impaired I_Ca,L inactivation (increase of pedestal current as shown in Fig. 3A & B). The consequence was a large increase in the late phase of I_Ca,L, as seen during the AP clamp in Fig. 3C, which was not successfully overcome by outward currents, accounting for the genesis of the EAD. The experiments using multiple ion channel blockers (Fig, 2) showed that despite prolonging APD, I_Na modification was not by itself sufficient to induce EADs when I_Ca,L was blocked. Both I_Na and I_Ca,L modification could be attributed to CaMKII activation, since, as shown in Fig. 6A, H₂O₂ failed to prolong APD significantly or cause EADs when CaMKII was blocked by KN-93.

Oxidative stress also modulates the properties of other Ca_i-cycling proteins, including ryanodine receptors ^{11, 12} and SERCA2a ¹³, which could potentially promote EAD genesis by modulating SR Ca release during repolarization. Indeed, some recent studies have shown that EADs and DADs can share a common mechanism under Ca_i overload conditions ^29-31. Non-I_Ca,L-gated SR Ca²⁺-release during repolarization has been proposed to contribute to EAD genesis by activating Ca-sensitive inward currents ²⁹, but it is difficult to unequivocally distinguish from SR Ca release triggered by reactivated L-type Ca channels.

It has been reported that voltage can directly activate SR Ca²⁺-release in cardiac myocytes, which is enhanced by cAMP ^{32, 33}. However, EADs were readily induced by H₂O₂ using cAMP-free pipette solution (data not shown, also see Song et al⁵), excluding an essential role of a cAMP-sensitive voltage-activated Ca release in EAD generation.

Contribution of SR Ca handling to the EADs and DADs induced by H₂O₂

CaMKII can phosphorylate RyR2, which may enhance SR Ca release dependent triggered activity ³⁴. However, while it is well accepted that CaMKII phosphorylates RyR2, the functional consequences are controversial. Bers' group found increased SR Ca leak from CaMKII-mediated RyR2 phosphorylation ^{34, 35}, but Yang et al ³⁶ reported a suppression of Ca sparks and Ca waves. The cause of this discrepancy remains to be determined. In addition, oxidative stress has direct effects on RyR and SR function. The interdependencies between SR Ca cycling, I_Ca,L, I_NCX, and repolarization reserve are complex and nonlinear, making it very challenging, if not impossible, to assign a simple mechanistic role of SR Ca cycling to EAD generation. For example, SR depletion could suppress EADs by any of the following interactive mechanisms: 1) preventing CaMKII activation by oxidative stress (since the Ca-CaM interaction is still required); 2) increasing repolarization reserve by suppressing I_NCX; 3) preventing Ca waves during the repolarization phase; 4) altering activation/inactivation properties of other ionic currents.

In the presence of BAPTA, the AP has a higher rectangular plateau (Fig. 5), which may also hinder the reactivation of I_Ca,L and EAD formation, consistent with a recent study showing that triangulation of the AP favors I_Ca,L reactivation ³⁷.

The Role of CaMKII Signaling in H₂O₂-induced EADs

The enhancement of I_Ca,L by H₂O₂ in the present study is consistent with activation of CaMKII by H₂O₂, although direct redox modifications of L-type Ca channel subunits may also contribute. CaMKII is known to mediate I_Ca,L facilitation ³⁸, and CaMKII activation has been shown to promote EADs in a variety of settings ^{19, 22}. Moreover, CaMKII activation has recently been reported to impair I_Na inactivation ²¹, consistent with a common underlying pathogenesis of H₂O₂-induced and CaMKII-induced EADs. Our finding that the CaMKII inhibitor KN-93, but not its inactive analog KN-92, suppressed EAD formation by H₂O₂, generally supports a common mechanism. On the other hand, both KN-93 and KN-92 have nonspecific effects, including substantial block of I_Ca,L²⁷ which is critical in EAD formation. However, the selective CaMKII peptide inhibitor AIP also significantly delayed the onset of EADs during exposure to 0.2 mM H₂O₂. Theoretically, AIP should be equally effective at preventing CaMKII activation by oxidative stress as by elevated Ca_i, since both activation modes require initial interaction of CaMKII's regulatory domain with Ca-CaM complexes, to expose the redox and autophosphorylation sites on CaMKII's catalytic domain ¹⁸. This also explains why BAPTA-AM prevents EADs during H₂O₂ exposure, by preventing Ca_i from rising sufficiently for Ca-CaM to interact with CaMKII and trigger the subsequent persistent activation by redox modification.

Based on these observations, the role of CaMKII activation in H₂O₂-induced EADs can be summarized as follows: 1) CaMKII activation by H₂O₂ contributes to both impaired I_Na inactivation and I_Ca,L facilitation; 2) Both factors enhance cellular Ca loading by prolonging APD and increasing Ca influx via I_Ca,L, promoting further CaMKII activation via Ca-CaM; 3) KN-93 synergistically suppresses EADs by preventing CaMKII activation by H₂O₂ and reducing Ca influx by directly blocking I_Ca,L. Without concomitantly inhibiting CaMKII activation, partial block of I_Ca,L by KN-92 is not sufficient to suppress EAD formation; 4) Direct redox modification of ion channel proteins may also contribute, since selective CaMKII inhibition with AIP delayed the onset of, but did not entirely suppress, H₂O₂-induced EADs. We do not have a ready explanation for why AIP showed less effectiveness than KN-93 in our present study. KN-93 also blocks the L-type Ca current upon which EAD generation depends, which may account for its more complete potency. Alternatively, AIP may have had nonspecific effects which reduced repolarization reserve, since it modestly prolonged APD.

EAD-associated DADs

Associations between EADs and DADs have been noted previously ^{6, 31, 39, 40}, but the underlying mechanisms have not been analyzed in detail. For example, in atrial myocytes, Song et al ⁴⁰ found that ATX-induced EADs were also associated with DADs. They postulated that DADs were due to excessive Ca-loading as a result of APD prolongation by EADs, leading to spontaneous SR Ca release following repolarization. Here we directly documented this postulated mechanism for H₂O₂-induced EADs, by using optical Ca_i imaging to detect the spontaneous Ca_i wave inducing a transient inward current causing the DAD (Fig. 4). It is also conceivable that oxidative stress directly sensitizes the SR to non I_Ca,L-gated SR Ca²⁺-release through direct effects on Ca_i cycling proteins such as RyR ^{11, 12}, but this remains to be established. In either case, our findings suggest that both EADs and EAD-induced DADs may both contribute to TA during oxidative stress, amplifying the arrhythmogenic consequences.

Clinical Implications

Increased oxidative stress is believed to be an important factor predisposing the diseased heart to lethal arrhythmias. Here we have shown how oxidative stress caused by exogenous H₂O₂ predisposes the heart to both EADs and DADs, and demonstrated a link to CaMKII activation, which is also involved in other aspects of heart failure ^{41, 42}. Other free radical sources have also been reported to produce similar results, i.e. inducing EADs, DADs and triggered activities ⁶. Although it is still under debate, H₂O₂ levels up to ∼35 μM have been reported in human blood plasma ⁴³. It is also well known that ROS levels can increase during ischemia and reperfusion by as much as 100-fold ⁴⁴, and also with age by 7.5 fold ⁴⁵. Since ROS are short-lived radicals, it is conceivable that the local concentrations near sites of production are considerably higher than circulating concentrations. Therefore, the concentrations (100-200 μM) used in most of our experiments are reasonable approximations of the pathophysiologically relevant range. Targeting CaMKII signaling may have therapeutic potential to prevent arrhythmias in the failing heart, although negative inotropic actions ⁴⁶ may limit the use of CaMKII inhibitors.