ROS-activated Ca/calmodulin kinase IIδ is required for late I_Na augmentation leading to cellular Na and Ca overload

Abstract

Rationale

In heart failure (HF), CaMKII expression and reactive oxygen species (ROS) are increased. Both ROS and CaMKII can increase late I_Na leading to intracellular Na accumulation and arrhythmias. It has been shown that ROS can activate CaMKII via oxidation.

Objective

We tested whether CaMKIIδ is required for ROS-dependent late I_Na regulation and if ROS-induced Ca released from the sarcoplasmic reticulum (SR) is involved.

Methods and Results

40 µmol/L H₂O₂ significantly increased CaMKII oxidation and autophosphorylation in permeabilized rabbit cardiomyocytes. Without free [Ca]_i (5 mmol/L BAPTA/1 mmol/L Br₂-BAPTA) or after SR depletion (caffeine 10 mmol/L, thapsigargin 5 µmol/L) the H₂O₂-dependent CaMKII oxidation and autophosphorylation was abolished. H₂O₂ significantly increased SR Ca spark frequency (confocal microscopy) but reduced SR Ca load. In wildtype (WT) mouse myocytes, H₂O₂ increased late I_Na (whole cell patch-clamp). This increase was abolished in CaMKIIδ^−/− myocytes. H₂O₂-induced [Na]_i and [Ca]_i accumulation (SBFI and Indo-1 epifluorescence) was significantly slowed in CaMKIIδ^−/− myocytes (vs. WT). CaMKIIδ^−/− myocytes developed significantly less H₂O₂-induced arrhythmias, and were more resistant to hypercontracture. Opposite results (increased late I_Na, [Na]_i and [Ca]_i accumulation) were obtained by overexpression of CaMKIIδ in rabbit myocytes (adenoviral gene transfer) reversible with CaMKII inhibition (10 µmol/L KN93 or 0.1 µmol/L AIP).

Conclusion

Free [Ca]_i and a functional SR are required for ROS activation of CaMKII. ROS-activated CaMKIIδ enhances late I_Na, which may lead to cellular Na and Ca overload. This may be of relevance in HF, where enhanced ROS production meets increased CaMKII expression.

Introduction

Reactive oxygen species (ROS) are generated during myocardial infarction and contribute to left ventricular remodelling and adverse outcomes.^{1, 2} In human failing myocardium increased oxidative stress is associated with reduced left ventricular function.³ These detrimental effects involve impairment of Na and Ca homeostasis^4–8 resulting in contractile dysfunction, electrical instability and cell death.⁹ We and others have shown that H₂O₂, which generates ROS,⁷ enhances the late, slowly-inactivating current through cardiac Na channels (late I_Na), thereby leading to action potential (AP) prolongation and arrhythmias.^{6, 10} Na entering the cell via late I_Na leads to cytosolic Na accumulation,⁶ favouring Ca entry via sarcolemmal Na-Ca-exchange (NCX) and cellular Ca overload, contractile dysfunction, and hypercontrature.^{5, 11, 12}

The mechanisms by which ROS increase late I_Na and initiate this detrimental cascade are not clear. We have shown that Ca/calmodulin-dependent protein kinase IIδ (CaMKIIδ) can phosphorylate cardiac Na channels leading to enhanced late I_Na.¹³ The binding of Ca/calmodulin (Ca/CaM) to its regulatory domain results in conformational changes that expose the catalytic domain,¹⁴ which can phosphorylate numerous targets including threonine 287 at its own regulatory domain. This autophosphorylation renders the protein to remain active even after Ca/CaM dissociation.¹⁴ It is known that expression and activity of CaMKII are increased in heart failure (HF), and we have shown that transgenic CaMKIIδ overexpression results in left ventricular dysfunction¹⁵ and arrhythmias,¹³ whereas CaMKIIδ inhibition/knockout was shown to prevent cardiac remodelling and heart failure.^16–18

It has been recently shown that ROS oxidize methionine 281/282 at the regulatory domain of CaMKII, and oxidized CaMKII has properties equivalent to autophosphorylated CaMKII⁹ in mediating angiotensin-induced apoptosis⁹ and ROS-induced afterdepolarizations.¹⁹ Moreover, there is evidence that Ca/CaM-CaMKII interaction is a prerequisite for ROS-dependent oxidation of CaMKII,⁹ and ROS have been shown to directly induce sarcoplasmic reticulum (SR) Ca release via thiol oxidation of the ryanodine receptor 2 (RyR2).²⁰

Thus, we hypothesized that 1) ROS-activated CaMKIIδ is responsible for the observed ROS effects on late I_Na, and 2) Ca released from SR Ca stores is involved in the ROS-dependent CaMKII activation process.

The results of the present study show that H₂O₂-induced augmentation of late I_Na requires CaMKIIδ, and the consequent cellular Na and Ca overload was significantly slowed in myocytes lacking CaMKIIδ. Moreover, we show that the H₂O₂-dependent CaMKII activation requires a functional SR. Finally, we establish CaMKIIδ as an important mediator of H₂O₂-dependent arrhythmogenesis and cellular death injury.

Methods

CaMKIIδ knockout mice and CaMKIIδ overexpression in rabbit myocytes

Ventricular myocytes were isolated from CaMKIIδ knockout mice (CaMKIIδ^−/−)¹⁷ and wild-type (WT) littermates. For CaMKIIδ overexpression, isolated rabbit ventricular myocytes were transfected with CaMKIIδ_C adenovirus (β-galactosidase, βGal, as control).^{5, 13, 21}

Patch-clamp experiments

Ruptured-patch whole-cell voltage-clamp or current clamp was used to measure I_Na or membrane potential, respectively. Myocytes were held at −120 mV and I_Na was elicited using a train of pulses to −20 mV. APs were continuously elicited by square current pulses of 1–2 nA amplitude and 1–5 ms duration at basic cycle length of 2 s.

Data analysis and statistics

All data are expressed as mean±S.E.M. For longitudinal data, two-way repeated measures analysis of variance (ANOVA) was run; else Student’s unpaired t test or one-way ANOVA with Student-Newman-Keuls multiple comparison was used. Double sided P-values of <0.05 were considered significant.

A detailed description of methods, including numerical simulations, can be found in the online supplement.

Results

H₂O₂-dependent CaMKII activation requires functional SR Ca stores

CaMKII autophosphorylation at serine 287 (p-CaMKII) and oxidation at methionine 281/282 (ox-CaMKII) were assessed using Western blotting in lysates of myocytes exposed to H₂O₂ (Fig. 1, Online Table I). Increasing cytosolic [Ca] to 1 µmol/L significantly increased p-CaMKII levels and inhibition of catalytic CaMKII activity using autocamtide 2-related inhibitory peptide (AIP) reversed this activation. In myocytes exposed to H₂O₂, p-CaMKII and ox-CaMKII levels increased in a concentration-dependent manner (40 vs. 200 µmol/L H₂O₂) and AIP significantly reduced H₂O₂-induced autophosphorylation but not oxidation of CaMKII. This suggests that CaMKII oxidation is accompanied by a significant autophosphorylation of CaMKII. Interestingly, when SR Ca stores were depleted using caffeine and thapsigargin (THA), both H₂O₂-dependent oxidation as well as autophosphorylation of CaMKII were prevented. This suggests that Ca derived from SR Ca stores is required for substantial H₂O₂-dependent CaMKII activation to occur. To further investigate the Ca-dependency of CaMKII oxidation, permeabilized myocytes were bathed in a mock intracellular solution, in which Ca was heavily buffered (5 mmol/L BAPTA/1 mmol/L Br₂-BAPTA). Figure 2 and Online Table I show that no significant H₂O₂-dependent CaMKII oxidation or autophosphorylation were observed in the absence of intracellular free [Ca]. Additionally, heavily buffered low free [Ca] of 20 nmol/L prevented a substantial increase in p-CaMKII and ox-CaMKII levels after H₂O₂ exposure. This suggests that albeit the higher dose of H₂O₂ may oxidize a small amount of Ca/CaM-bound CaMKII at clamped free [Ca] of 20 nmol/L, large fluctuations of cytosolic free [Ca] are necessary for substantial oxidation to occur.

Figure 1.

ROS can activate CaMKII in permeabilized myocytes (free [Ca]_i 20 nmol/L) Western blots of p-CaMKII (A) or ox-CaMKII (B) vs. CaMKII and GAPDH are shown. Mean densitometric values for p-CaMKII vs. CaMKII (C) or ox-CaMKII vs. CaMKII (D) are shown. One way ANOVA was used for statistical comparison. All densitometric values are summarized in Online Table I. CaMKII was stimulated with 1 µmol/L [Ca], 200 or 40 µmol/L H₂O₂. For some experiments, caffeine (10 mmol/L) and THA (5 µmol/L) were used to unload the SR or AIP (1 µmol/L) was present.

Figure 2.

ROS activation of CaMKII is Ca-dependent. Permeabilized myocytes bathed in heavy Ca buffers (5 mmol/L BAPTA/1 mmol/L Br₂-BAPTA) were used for Western blots of p-CaMKII (A) or ox-CaMKII (B). All densitometric values are shown in Online Table I.

H₂O₂ effects on RyR2 function

We determined the effects of H₂O₂ on RyR2 function by analyzing the characteristics of SR Ca sparks. In permeabilized myocytes, the addition of 200 µmol/L H₂O₂ led to an immediate and massive SR Ca release throughout the whole cell, comparable to a caffeine transient, followed by an irreversible inhibition of RyR2 function (data not shown). Because intracellular H₂O₂ may reach only 1–15% of the applied exogenous concentration in case of an intact sarcolemma,²² we reduced the H₂O₂ concentration to 40 µmol/L and set the exposure time to 30 s for all experiments in permeabilized myocytes. Addition of H₂O₂ caused an immediate significant increase in CaSpF and FDHM, and a slight reduction in Ca spark amplitude, leading to a 3.5-fold increase in SR Ca leak (Fig. 3A–C). This ROS effect appeared to be transient. Indeed, 2.5 min after onset of H₂O₂ exposure CaSpF and amplitude had returned to baseline values, and SR Ca leak was significantly reduced despite a persistent prolongation of FDHM. Interestingly, at the time of maximal increased SR Ca leak, H₂O₂ exposure caused a significant reduction of SR Ca load that was not prevented by AIP (Fig. 3D, 30 s H₂O₂).

Figure 3.

Ca sparks were measured in permeabilized rabbit myocytes. A) Line scan images and typical corresponding Ca sparks. Mean data for CaSpF, Ca spark amplitude and FDHM (B) or calculated SR Ca leak (C) are shown. D) SR Ca content was estimated using caffeine (10 mmol/L).

SR Ca sparks were also measured in intact mouse myocytes during electrical field stimulation. Under baseline conditions, very few Ca sparks were detected (Fig. 4). Similar to permeabilized myocytes, 200 µmol/L H₂O₂ increased CaSpF but reduced Ca spark amplitude. The resulting SR Ca leak was markedly (about 15-fold) increased (Fig. 4 A–C). This was accompanied by reduced SR Ca load (Fig 4D). Notably, inhibition of CaMKII (KN93) did not influence the H₂O₂-dependent changes in CaSpF, SR Ca leak or SR Ca load, although Ca spark amplitude was slightly reduced (Fig. 4A–D). These results indicate that the H₂O₂-induced SR Ca leak does not require CaMKII.

Figure 4.

Ca sparks were measured in intact twitching mouse myocytes (BCL 2 s). A) Line scan images at baseline and 4 min after onset of H₂O₂ (200 µmol/L) without any pharmacological intervention (left) and in the presence of KN93 (10 µmol/L, right). Mean data for CaSpF, amplitude and FDHM (B) and calculated SR Ca leak (C) are shown. D) SR Ca load (caffeine 10 mmol/L) is shown.

CaMKII is required for H₂O₂-enhanced late I_Na

Our group and others have shown previously^{6, 10, 23, 24} that ROS enhance late I_Na, and CaMKII exerts a strikingly similar effect.^6,13 To test whether H₂O₂-activated CaMKII is required for the effects of ROS on late I_Na, the current was measured in myocytes from WT and CaMKIIδ^−/− mice (Fig. 5A). At baseline, no significant difference in late I_Na integrals was detected between the two groups (Fig. 5B, Table 1). When WT myocytes were exposed to H₂O₂, a significant time-dependent increase in the late I_Na integral, nearly doubling at 12 min after H₂O₂ treatment, was observed. This increase was absent in myocytes lacking CaMKIIδ (Fig 5 A+B, Table 1). Consistent with this, adenoviral CaMKIIδc overexpression in rabbit myocytes exhibit significantly more late I_Na at baseline and larger H₂O₂-dependent late I_Na increase (vs. βGal), both of which are prevented by AIP (Fig. 5 C+D, Table 1). In control myocytes, baseline late I_Na was not affected by AIP, which suggests that baseline CaMKII-induced late I_Na is negligible.

Figure 5.

Late I_Na after exposure to H₂O₂. Late I_Na was measured at BCL 2 s (protocol in inset) in myocytes from CaMKIIδ^−/− mice (vs. WT littermates) and in isolated rabbit myocytes after CaMKIIδ_C overexpression (vs. βGal). Late I_Na was measured as the integral between 50 and 500 ms after onset of depolarisation. Original traces (A) and mean data (B) showed that H₂O₂ significantly increased late I_Na in WT myocytes. This increase was completely abolished in myocytes lacking CaMKIIδ. Similarly, H₂O₂ significantly increased late I_Na in rabbit myocytes (βGal, C&D). CaMKIIδ_C overexpression markedly enhanced H₂O₂-induced late I_Na, and AIP (100 nmol/L) significantly blunted this increase. *P<0.05 (two-way ANOVA) vs. WT or βGal, respectively; †P<0.05 (two-way ANOVA) vs. CaMKIIδ_C.

Table 1.

Late I_Na integral before (baseline) and during exposure to H₂O₂

AmsF−1 [N]	Baseline	12 min H2O2
WT [15]	−203.5±21	−380.1±101.1*
CaMKIIδ−/− [7]	−155.4±43.3	−204.1±93.1†
βGal [17]	−99.3±7.4	−331.8±77.2*
βGal+AIP [18]	−105.9±7.8	−200.5±32.3†
βGal+TTX [4]	−66.4±22.6*	−54.5±32.1†
CaMKIIδC [18]	−249.1±23.3*	−719.7±192.2‡†
CaMKIIδC+AIP [14]	−158.7±12.4‡	−277.0±37.4#
CaMKIIδC+TTX [8]	−101.9±29.4‡	−121.4±63.5#

Late I_Na integrals (50–500 ms; AmsF⁻¹) at baseline and after 12 min of H₂O₂ exposure. For some experiments, AIP (100 nmol/L) or TTX (1 µmol/L) were used.

^*P<0.05 vs. WT (baseline) or βGal (baseline);

^†P<0.05 vs. WT (12 min) or βGal (12 min);

^‡P<0.05 vs. CaMKIIδ_C (baseline);

^#P<0.05 vs. CaMKIIδ_C (12 min)

Low concentrations of the Na channel blocker tetrodotoxin (TTX; 1 µmol/L), which only slightly reduce peak myocyte I_Na, significantly reduced late I_Na at baseline (Table 1). Moreover, this TTX concentration abolished the H₂O₂-induced increase in late I_Na.

H₂O₂-induced Na and Ca overload is mediated via CaMKII

In WT myocytes (stimulated at BCL 2s), H₂O₂ increased [Na]_i time-dependently, but this increase was much lower in CaMKIIδ^−/− myocytes (Fig. 6A,B, Online Table II). Similarly, the H₂O₂-induced increase in diastolic [Ca]_i was markedly reduced in CaMKIIδ^−/− myocytes (vs. WT, Fig. 6F, Online Table II). In WT myocytes, the addition of reverse-mode NCX inhibitor KB-R7943 significantly blunted the H₂O₂-induced increase in diastolic [Ca]_i (Fig. 6F, Online Table II). Likewise, CaMKIIδ_C overexpression enhanced H₂O₂-induced increases in [Na]_i and diastolic [Ca]_i significantly (vs. βGal), and KN93 slowed [Na]_i and [Ca]_i rise in both CaMKIIδ_C overexpressing and control myocytes (Fig. 6A,C&G). KN93 can also reduce I_Ca,L availability. The analogue KN92 (which does not inhibit CaMKII) shares this effect,²⁵ but does not protect from H₂O₂-induced arrhythmogenesis.¹⁹ To control for possible I_Ca,L effects of KN93, experiments were also done with KN92, which did not alter the H₂O₂-induced increase of [Na]_i and diastolic [Ca]_i (Online Table II). Of note, at baseline there was no difference in [Na]_i and diastolic [Ca]_i between WT and CaMKIIδ^−/− myocytes, or in control-transfected rabbit myocytes in the presence of KN93. This is consistent with the late I_Na data, arguing against an important role for CaMKII in the basal regulation of cellular Na and Ca homeostasis (Online Table II). Interestingly, at baseline, CaMKIIδ_C overexpression significantly increased [Na]_i (reversible with KN93) but did not significantly alter diastolic [Ca]_i (Online Table II).

Figure 6.

[Na]_i and [Ca]_i upon exposure to H₂O₂. A) Original traces of SBFI fluorescence in mouse (left panel) and rabbit ventricular myocytes (mid and right panel). Data is shown for baseline (0 min), 6 and 12 min after onset of H₂O₂ exposition (200 µmol/L). WT–black, CaMKIIδ^−/−–red, CaMKIIδ_C overexpression–blue, βGal–black, KN93–blue, RAN–green. Mean data after conversion to [Na]_i is shown for mouse (B) and rabbit myocytes (C). D) Original traces of Na influx (via [Na]_i-converted SBFI fluorescence) in mouse ventricular myocytes (BCL 2s) exposed to 100 µmol/L strophanthidin (Stroph) (with and without 1 µmol/L TTX and/or 200 µmol/L H₂O₂). E) Mean data for the rate of rise of [Na]_i (baseline [Na]_i in inset). (one-way ANOVA). F) Original traces of Indo-1 fluorescence at baseline (0 min), 6 and 12 min after onset of exposure to H₂O₂ (myocytes from WT, left panel and CaMKIIδ^−/−, mid panel). Vertical bars indicate the electrical stimulus. Right panel shows mean data after conversion to [Ca]_i. G) Mean data for rabbit myocytes. CaMKIIδ_C overexpression significantly enhanced (vs. βGal) and CaMKII inhibition (10 µmol/L KN93) significantly slowed the increase in diastolic [Ca]_i after H₂O₂. *P<0.05 (two-way ANOVA) vs. WT or βGal, respectively, †P<0.05 (two-way ANOVA) vs. CaMKIIδ_C

To further investigate the importance of late I_Na for H₂O₂-induced increase in [Na]_i we conducted quantitative Na influx measurements in the presence of the NKA inhibitor strophanthidin (Fig. 6D+E). TTX (1 µmol/L) was used to specifically block Na influx via I_Na. At baseline, the TTX-sensitive Na influx was not different between WT and CaMKIIδ^−/− myocytes, in accordance with our late I_Na data. The addition of 200 µmol/L H₂O₂ increased the TTX-sensitive Na influx nearly threefold in WT but not in CaMKIIδ^−/− myocytes (Fig. 6E) again consistent with our late I_Na measurements (Table 1). These data is further supported by experiments using the late I_Na inhibitor ranolazine (RAN). RAN significantly reduced the H₂O₂-induced gain in [Na]_i and diastolic [Ca]_i after CaMKIIδ_C overexpression, but also in control (βGal, Fig. 6A, Online Table II). Interestingly, RAN also reduced [Na]_i and diastolic [Ca]_i before H₂O₂ exposure (Online Table II) consistent with a relevant TTX-sensitive Na influx at baseline (Fig. 6E).

To test the hypothesis that ROS-mediated SR Ca release is required for CaMKII activation leading to [Na]_i and diastolic [Ca]_i overload, SR Ca stores were depleted with THA. THA greatly slowed the H₂O₂-induced increase in [Na]_i and diastolic [Ca]_i for both CaMKIIδ_C overexpressing and control myocytes (Online Table II), completely eliminating the difference between CaMKIIδ_C and βGal (at 12 min, Online Table II). Interestingly, THA significantly reduced [Na]_i in myocytes overexpressing CaMKIIδ_C not only after H₂O₂ exposure but also at baseline. THA may possibly interfere with a positive feedback loop involving CaMKII-dependent Ca spark augmentation²⁶ leading to Ca-dependent CaMKII activation. On the other hand, THA did not alter [Na]_i in control-transfected myocytes, in line with our data in permeabilized myocytes showing no significant reduction of CaMKII activity with caffeine/THA at baseline. This supports the concept that baseline CaMKII activity is low and reduction of SR Ca release does not further reduce its activity.

H₂O₂-induced arrhythmias and CaMKII

To test, whether CaMKII inhibition could prevent ROS-induced arrhythmias, mouse myocytes were exposed to H₂O₂ and membrane potential or sarcomer length were continuously monitored. H₂O₂ significantly prolonged action potential duration and depolarized the diastolic membrane potential in WT but not in CaMKIIδ^−/− myocytes (Fig. 7A–C). At baseline, there was regular rhythmic activity for both WT and CaMKIIδ^−/−. The addition of H₂O₂ induced EADs and DADs (Fig. 7D) as well as irregular contractile activity (Fig. 7E). These arrhythmias were significantly more frequent and more severe in WT vs. CaMKIIδ^−/− (Fig. 7D,F). Cellular Ca overload results in the development of hypercontracture,⁵ which is a permanent reduction (<50%) of longitudinal cell length due to a persistent activation of myofilaments. Kaplan-Meier analysis (Fig. 7G) revealed that myocytes lacking CaMKIIδ were more resistant to ROS-induced hypercontracture.

Figure 7.

ROS-induced arrhythmias. A) Original traces of stimulated APs before and after H₂O₂ treatment. Mean data shows a significant H₂O₂-dependent increase in AP duration (APD 90, B), and a significant positive shift in the diastolic action potential in WT (C), but no change in APD and no positive shift diastolic action potential in CaMKIIδ_C^−/− myocytes. Insets show baseline APD 90 and diastolic potential, respectively. D) Original traces of consecutive APs (left) in myocytes from WT or CaMKIIδ_C^−/− mice. Mean data (right) shows that H₂O₂-induces EADs+DADs were significantly more frequent in WT vs. CaMKIIδ_C^−/−. E) Original traces of sarcomere length in intact twitching myocytes from WT or CaMKIIδ_C^−/− mice. F) Mean data. The propensity for H₂O₂-induced arrhythmias was significantly reduced in CaMKIIδ_C^−/−. G) Kaplan-Meier analysis of hypercontracture development. *P<0.05 vs. WT. BCL 2 s, vertical bars indicate stimuli.

Discussion

We investigated the mechanisms by which the exposure of H₂O₂ enhances late I_Na in isolated ventricular myocytes. We found that 1) ROS-activated CaMKIIδ is required for late I_Na augmentation, which was abolished in the absence of CaMKIIδ expression, and 2) ROS-induced SR Ca release is a prerequisite for ROS-dependent CaMKII activation, based on the observation that the latter was inhibited after SR depletion with THA and caffeine.

These novel findings add to the growing body of evidence that CaMKII is activated under pathophysiological conditions being involved in signaling cascades that lead to dysregulation of cellular Na and Ca homeostasis, increased arrhythmogenesis, and ultimately cell death.

The role of the SR Ca stores in ROS-dependent CaMKII activation

In a fundamental study Erickson and colleagues have shown that ROS oxidize CaMKII at methionine 281/282, and this oxidation renders CaMKII autonomous of Ca/CaM levels similar to autophosphorylation at threonine 287.⁹ Direct oxidation of the kinase also produces a secondary increase in the fraction of autophosphorylated subunits.^{27, 28} We showed here that exogenous H₂O₂ applied to permeabilized myocytes increased CaMKII oxidation and autophosphorylation. We also showed that oxidation and autophosphorylation critically depend on free cytosolic Ca levels. Without free [Ca]_i no significant H₂O₂-dependent CaMKII oxidation or autophosphorylation was observed (Fig. 2). This is in accordance with the concept that H₂O₂ requires initial interaction of Ca/CaM with CaMKII to expose the oxidation and autophosphorylations sites of CaMKII.⁹

Moreover, we showed that a free [Ca]_i of 20 nmol/L in the presence of a strong and fast Ca buffer (BAPTA/Br₂-BAPTA) is not sufficient to result in H₂O₂-dependent CaMKII activation. In contrast, if fast localized increases in free [Ca] (during diastolic SR Ca leak or Ca sparks) are allowed by a slower buffer (EGTA), the same free [Ca]_i of 20 nmol/L facilitates a robust H₂O₂-dependent CaMKII activation (Fig. 1). This result is in accordance with Song et al.,²⁷ who showed that high concentrations of BAPTA but not EGTA could prevent H₂O₂-induced CaMKII-dependent Ca current facilitation. Further evidence comes from Xie and colleagues showing that BAPTA loaded myocytes were resistant to H₂O₂-induced CaMKII-dependent formation of early afterdepolarizations (EADs).¹⁹ On the other hand, Palomeque at al. showed that neither extracellular application of 1 µmol/L BAPTA-AM in cultured myocytes nor zero Ca-EGTA used for an in vitro phosphorylation assay were sufficient to inhibit H₂O₂-induced CaMKII autophosphorylation.²⁸ Methodological differences may account for this discrepancy: although it is likely that BAPTA was properly loaded into their cultured myocytes, its concentration may not be sufficient to avoid very fast localized increases of [Ca]_i. Also, the isoosmotic homogenization buffer of their in vitro assay contained no detergent, thus leaving functional SR vesicles in the phosphorylation reaction. This suggests that SR Ca release may be involved in the rapid localized increases of free [Ca]_i required for H₂O₂-dependent CaMKII activation. Here we show that unloading the SR with caffeine and THA abolished the H₂O₂-induced CaMKII oxidation and autophosphorylation. Moreover, in the presence of THA, the H₂O₂-dependent cytosolic Na and Ca overload was markedly reduced. This relates mechanistically to work showing that pre-incubation with caffeine and THA abolished the H₂O₂-activated CaMKII-dependent Ca current facilitation.²⁷ Therefore, an SR-dependent mechanism appears to be required for ROS-dependent oxidation of CaMKII.

It has been shown that thiol-oxiziding agents like H₂O₂ activate RyR Ca release after oxidation of more than seven thiols per subunit.^{20, 29–32} In line with this, we show here that H₂O₂ immediately increased CaSpF and SR Ca leak. The SR Ca load was reduced, which may be a consequence of an increased SR Ca leak and/or inhibition of SERCA,^{33, 34} and may account for the observed reduction in Ca spark amplitude. Because activated CaMKII is known to phosphorylate RyR2 thereby increasing CaSpF,²⁶ we cannot exclude that the observed H₂O₂-induced changes in CaSpF and SR Ca leak are secondary to CaMKII activation. However, acute CaMKII inhibition with KN93 did not affect the H₂O₂-induced increases in CaSpF and SR Ca leak indicating that direct H₂O₂-induced thiol-oxidation of RyR2 may be more important.

ROS enhance late I_Na via CaMKII activation

ROS have been shown to acutely enhance late I_Na leading to AP prolongation and EADs.^{6, 24, 35} To date, the mechanism of ROS action on Na channels is not fully understood. Changes in the lipid environment²⁴ or oxidation of multiple methionine residues of the channel may be responsible.³⁶ Here we show that the acute H₂O₂-dependent increase in late I_Na was absent in myocytes with genetic knock out of CaMKIIδ. Furthermore, subacute adenoviral CaMKIIδ_C overexpression enhanced whereas acute CaMKII inhibition with AIP slowed the H₂O₂-induced increase of late I_Na. Therefore, although redox modification of the Na channel or the lipid environment cannot be completely excluded, the dominating mechanism here involves acute CaMKIIδ activation.

We have shown that CaMKII can phosphorylate cardiac Na channels α isoform (Na_V1.5)¹³. Aiba et al. showed that CaMKII phosphorylates the I–II linker but also to a lesser extent the C-terminus.³⁷ Hund et al. found that CaMKII phosphorylates serine 571 in the I–II linker contributing to altered I_Na gating.³⁸ Although phosphorylation does occur, ROS-activated CaMKII may also regulate late I_Na independent of its catalytic activity, as it was shown for oxidation-dependent facilitation of I_Ca,L.²⁷ Because the cardiac Na channel is a macromolecular complex, CaMKII interaction with proteins of this complex could result in altered Na channel gating. For example, co-expression of Na_V1.5 with cardiac β₁ but not β₂ subunit increased late I_Na,³⁹ and it may be possible that oxidized CaMKII is involved in the interaction of α and β subunits.

Beside CaMKII, protein kinase A or C (PKA or PKC) are known to regulate Na channel gating. ROS have been shown to oxidize PKA regulatory subunit I, resulting in PKA translocation and activation.⁴⁰ PKA increases peak I_Na current density via acceleration of channel trafficking to the plasma membrane,⁴¹ but does not change inactivation kinetics.^{42, 43} Mild oxidative stress has also been shown to activate PKC.⁴⁴ The redox regulation of PKC is very specific for different isoforms and different cell types. Ward and Giles showed that the H₂O₂-dependent slowing of I_Na open state inactivation was blocked in the presence of the PKC inhibitor Bis-indolylmaleimide.¹⁰. In heterologeous expression systems (rat and human Na_V1.5), however, PKC activation did not change I_Na inactivation.^{45, 46} The most consistent effect of PKC on cardiac Na channels is a reduction in peak I_Na current density⁴⁷

Cellular Na and Ca accumulation

The present study shows that ROS-induced Na and Ca overload are blunted in myocytes lacking CaMKIIδ. We also show that SR Ca unloading with THA similarly reduces ROS-induced Na and Ca overload, suggesting that the CaMKII activation depends on SR function. These data suggest that CaMKIIδ is critically involved in ROS-induced [Na]_i and [Ca]_i overload. Quantitative Na influx measurements showed that TTX-sensitive Na entry is markedly increased by ROS exposure in WT but not in CaMKIIδ^−/− myocytes and the late I_Na blocker RAN slowed the development of Na and Ca overload, thus suggesting that ROS-induced Na entry is mediated via Na channels. This is in line with data from us and others.^{6, 8, 48} We propose that the increase in [Na]_i precedes the major rise in [Ca]_i because reduction of extracellular [Na] or application of Na channel blockers slows the rise in [Ca],⁸ and inhibition of reverse mode NCX activity with KB-R7943 significantly blunts Ca overload in the present study. However, because RAN did not completely abolish the increase in [Na]_i and [Ca]_i, other mechanisms may also be involved. For example, ROS can significantly impair sodium-potassium ATPase (NKA) function,⁴⁹ reduce SERCA2 activity and L-type Ca current (I_CaL),⁵⁰ and increase NCX activity.^{51, 52} In order to further understand these mechanisms we utilized numerical simulations of the cardiac action potential and Na and Ca handling.⁵³ Incorporation of the measured ROS-induced late I_Na into the model led to a rise in [Na]_i less than 1 mmol/L (Online Fig. IA: a vs. b), much less than measured experimentally. ROS may also (via CaMKII) increase diastolic TTX-sensitive Na influx, as reported in HF myocytes.⁵⁴ Increasing background Na conductance can bring the model to match the measured [Na]_i (Online Fig. IA, c)), but inclusion of ROS-dependent NKA inhibition⁴⁹ limits the extent of background Na current that is required to explain the measured [Na]_i increase (Online Fig. IA: c vs. d). If we also incorporate the above mentioned ROS-dependent effects on SERCA2a, RyR2, I_CaL, and NCX function,^50–52 the model reasonably mimics the ROS-induced alterations observed here in [Na]_i and SR Ca content (Online Fig. I A, f, B and C). Three aspects merit comment. First, the elevated [Na]_i and reduced Ca transient amplitude is predicted to cause substantial Ca influx via NCX during the action potential (i.e. reverse mode NCX; Online Fig. IB–C) and less Ca extrusion during diastole. This is also seen in human and rabbit HF myocytes which also exhibit elevated [Na]_i and reduced Ca transients.^{54, 55} This agrees with our results that the NCX inhibitor KB-R7943 suppressed the ROS-induced increase in diastolic [Ca]_i (Online Table II, Fig. 6F), that ROS-induced Ca overload depends on shifts of NCX activity,⁵⁶ and that late I_Na is responsible for diastolic Ca accumulation in HF where ROS production is augmented.⁵⁷ Second, turning off only the late and diastolic I_Na simulates reasonably the effects of ranolazine on [Na]_i (Online Fig. IA, g). Third, the model does not predict the extent of diastolic [Ca]_i elevation observed. Because intracellular Na and Ca are compartmentalized, part of the progressive measured rise in [Na]_i and [Ca]_i may be attributable to a consequent gradual rise in mitochondrial [Na] and [Ca].

Arrhythmogenesis

Here we show that the propensity for ROS-induced EADs/DADs and cellular arrhythmias were significantly reduced in myocytes lacking CaMKIIδ_C, although they were not completely abolished. ROS effects in myocytes are clearly multifactorial, and so are arrhythmogenic mechanisms. Therefore, activated CaMKII may be an important factor for ROS-induced arrhythmias but does not explain all arrhythmias that develop upon ROS or in HF. Possible mechanisms that contribute to the electrical instability are EADs and DADs. It was shown previously that enhanced late I_Na could prolong APD leading to EADs.⁶ Moreover, transient inward I_NCX (I_ti), which leads to DADs may results from ROS-induced cellular Ca overload and increased RyR2 open probability. Here we show that both EADs and DADs develop less often after H₂O₂ exposure in myocytes lacking CaMKIIδ_C. While the reduction in the propensity for EADs in CaMKIIδ^−/− myocytes may result from the lack in APD prolongation, less frequent DADs possibly indicate less cellular Na and Ca overload as shown here. The CaMKIIδ-dependent increase in late I_Na may be responsible for the observed H₂O₂-induced APD prolongation. Although ROS can decrease outward currents like I_to,⁵⁸ CaMKII activation tends to increase I_to.²¹ In the absence of confounding CaMKII activity (CaMKIIδ^−/− myocytes), we found that H₂O₂ did not increase APD, which argues against a significant role for inhibition of I_to in APD prolongation caused by H₂O₂. Interestingly, H₂O₂ rendered the diastolic membrane potential more positive increasing the likelihood that depolarizing currents would induce afterdepolarizations. This was not the case in CaMKIIδ^−/− myocytes. Nevertheless, although we show here evidence that CaMKII may be involved in ROS-induced arrhythmias there was still H₂O₂-induced arrhythmogenic activity detectable in CaMKIIδ^−/− myocytes suggesting that there are CaMKII-independent mechanisms as well. For instance ROS can directly oxidize RyR2 leading to increased CaSpF.³¹ While the present paper establishes a clear mechanistic working hypothesis for the role of CaMKIIδ in ROS-dependent regulation of late I_Na and its consequences, further studies are needed that aim at testing these mechanisms in HF.

In summary, we show that CaMKIIδ is required for the H₂O₂-induced augmentation of late I_Na and secondary changes of [Na]_i and [Ca]_i. The present results also suggest that ROS-induced SR Ca release may be a prerequisite for ROS-dependent CaMKII activation. In addition, CaMKIIδ appears to be involved in H₂O₂-induced arrhythmogenesis. These results are important, because expression and activity of CaMKII have been shown to be increased in HF,^59–61 where ROS generation is enhanced.³ In addition, HF is associated with enhanced late I_Na⁶² leading to dysregulation of intracellular Na and Ca homeostasis, electrical instability and cell death.

Novelty and Significance.

What is known?

• Heart failure (HF) is associated with increased reactive oxygen species (ROS) and Ca²⁺/calmodulin-dependent protein kinase II (CaMKII) expression.

• ROS can activate CaMKII by oxidation.

• ROS and CaMKII can increase late I_Na and intracellular Na concentration.

What new information does this article contribute?

• Ca²⁺ release from the sarcoplasmic reticulum (SR) is required for ROS-dependent CaMKII oxidation and autophosphorylation.

• ROS-activated CaMKII enhances late I_Na leading to cellular Na⁺ and Ca²⁺ overload.

• ROS-activated CaMKII is arrhythmogenic.

In HF, intracellular ROS are increased and CaMKII expression and activity is upregulated, contributing to electrical and functional remodeling. CaMKII was previously shown to be activated by ROS through Ca²⁺-dependent processes but the origin of Ca²⁺ remained unclear. In addition, ROS and CaMKII have both been shown to increase late I_Na leading to proarhythmogenic events but a mechanistic link was missing. In this study, we show that SR Ca release in required for ROS-dependent CaMKII oxidation and autophosphorylation, which leads to increased late I_Na, intracellular Na accumulation and Ca²⁺ overload. Moreover, ROS-activated CaMKII leads to action potential prolongation and cellular arrhythmias. These results provide novel insights into the Ca²⁺-dependence of the ROS activation process for CaMKII. Moreover, they highlight the importance of CaMKII for the ROS-induced disturbance of excitation-contraction coupling and arrhythmogenesis. Taken together, the present study suggests ROS-activated CaMKII as a promising new target for the treatment of heart failure.

Non-standard Abbreviations and Acronyms

CaMKII

Ca/calmodulin-dependent protein kinase II

CaMKIIδ^−/−

homozygous knockout of Ca/calmodulin-dependent protein kinase IIδ

APD

action potential duration

THA

thapsigargin

RAN

ranolazine

TTX

tetrodotoxin

HF

heart failure

ROS

reactive oxygen species

SR

sarcoplasmic reticulum

I_Na

Na current

NAPDH

nicotinamide adenine dinucleotide phosphate oxidase

NCX

sodium/calcium exchanger

NKA

sodium/potassium ATPase

SERCA

sarcoplasmic reticulum calcium ATPase

RyR2

ryanodine receptor type 2

CaSpF

calcium spark frequency

AIP

autocamtide 2-related inhibitory peptide

FDHM

full-duration-half-maximum

[Na]_i

intracellular sodium concentration

[Ca]_i

intracellular calcium concentration

PKA

protein kinase A

SBFI

sodium-binding benzofuran isophthalate

AM

acetoxymethylester

βGal

β-galactosidase

FKBP

FK-binding protein

BCL

basic cycle length

EGTA

ethylene glycol tetraacetic acid

GAPDH

Glyceraldehyde 3-phosphate dehydrogenase

HEPES

4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid