Ca/Calmodulin Kinase II Differentially Modulates Potassium Currents

Abstract

Background

Potassium currents contribute to action potential duration (APD) and arrhythmogenesis. In heart failure, Ca/calmodulin-dependent protein kinase II (CaMKII) is upregulated and can alter ion channel regulation and expression.

Methods and Results

We examine the influence of overexpressing cytoplasmic CaMKIIδ_C, both acutely in rabbit ventricular myocytes (24 h adenoviral gene transfer) and chronically in CaMKIIδ_C-transgenic mice, on transient outward potassium current (I_to), and inward rectifying current (I_K1). Acute and chronic CaMKII overexpression increases I_to,slow amplitude and expression of the underlying channel protein K_V1.4. Chronic, but not acute CaMKII overexpression, causes down-regulation of I_to,fast, as well as K_V4.2 and KChIP2, suggesting that K_V1.4 expression responds faster and oppositely to K_V4.2 upon CaMKII activation. These amplitude changes were not reversed by CaMKII inhibition, consistent with CaMKII-dependent regulation of channel expression and/or trafficking. CaMKII (acute and chronic) greatly accelerated recovery from inactivation for both I_to components, but these effects were acutely reversed by AIP (CaMKII inhibitor), suggesting that CaMKII activity directly accelerates I_to recovery. Expression levels of I_K1 and Kir2.1 mRNA were downregulated by CaMKII overexpression. CaMKII acutely increased I_K1, based on inhibition by AIP (in both models). CaMKII overexpression in mouse prolonged APD (consistent with reduced I_to,fast and I_K1), while CaMKII overexpression in rabbit shortened APD (consistent with enhanced I_K1 and I_to,slow and faster I_to recovery). Computational models allowed discrimination of contributions of different channel effects on APD.

Conclusion

CaMKII has both acute regulatory effects and chronic expression level effects on I_to and I_K1 with complex consequences on APD.

Introduction

Heart failure (HF) is accompanied by arrhythmogenic changes related to electrical remodeling. This is associated with prolongation of action potential duration (APD)¹ and down-regulation of transient outward K-current (I_to) and inward rectifying K-current (I_K1). I_K1 is responsible for stabilizing the diastolic membrane potential (E_m), such that decreased I_K1 increases the propensity for triggered arrhythmias.² I_to is important in early repolarization and influences the effects of other currents and transporters by affecting AP voltage-time trajectory. There are at least two components of I_to generated by different K-channel isoforms, which can be distinguished according to their recovery and inactivation kinetics.^{3, 4} The fast component (I_to,fast) recovers and inactivates with time constants (τ) of <100 ms, whereas the slow component (I_to,slow) recovers with τ of hundreds of milliseconds up to several seconds and inactivates with τ of ~200 ms.^3–5 Downregulation of I_to has been described in animal models of hypertrophy and human HF, ^{2, 6, 7} is associated with APD prolongation⁸ and predisposes to early after-depolarizations.

In HF, expression and activity of Ca/calmodulin-dependent kinase II (CaMKII) are enhanced.^9–11 CaMKIIδ is the predominant cardiac isoform¹² and transgenic mice (TG) overexpressing cytosolic CaMKIIδ_C develop HF with increased APD¹³ and are prone to ventricular arrhythmias.¹⁴ Recent evidence suggests that chronic inhibition of CaMKII results in APD shortening¹⁵ and prevents remodeling after myocardial infarction and excessive β-adrenergic stimulation.¹⁶ Moreover, an enhancement of I_to and inward rectifying I_K1 after chronic CaMKII inhibition was described, whereas acute CaMKII-inhibition did not increase I_to and I_K1¹⁵.

We studied how CaMKII alters I_to and I_K1, both acutely by adenoviral CaMKII overexpression in rabbit myocytes and chronically in CaMKII TG mice. We found that CaMKII activation exerts an acute regulatory effect on I_to and I_K1, and that CaMKII causes opposite effects on functional expression of I_to,slow and I_to,fast, whereas functional expression of I_K1 was only downregulated after TG CaMKII overexpression.

Methods

CaMKIIδ_C TG mice and overexpression of CaMKIIδ_C in rabbit myocytes

CaMKIIδ_C TG mice were used at 17.6±2.3-weeks of age¹⁷ and compared to their age- and sex-matched wild-type (WT) littermates. Ventricular myocytes were isolated¹⁸ and kept in modified Tyrode solution containing (mmol/l) 137 NaCl, 5.4 KCl, 1.2 MgSO₄, 1.2 Na₂HPO₄, 20 HEPES, 15 glucose, 1 CaCl₂ (pH 7.4). Ventricular myocytes were isolated from rabbits (1.3–2.0 kg).¹⁹ Transfection with CaMKIIδ_C adenovirus (Ad-CaMKIIδ_C) was performed and compared to β-galactosidase (Ad-βGal) as a control at a MOI of 100.^{18, 20} Cells were cultured for 24 h with M199 and washed prior to the experiment.¹⁸ All animal procedures were approved by the Institutional Animal Care and Use Committee.

An expanded Materials and Methods section can be found in the online data supplement.

Results

Functional expression and inactivation kinetics of I_to

To measure CaMKIIδ_C-dependent regulation of I_to, the E_m-dependence of activation and the kinetics of inactivation were analyzed (Figure 1–2). I_to,fast and I_to,slow were separated based on inactivation rate. Original traces were fit to bi-exponentials to obtain amplitudes and time constants τ_fast and τ_slow of inactivation.^{3, 21} In CaMKIIδ_C TG mice, total I_to was significantly reduced (vs. WT, Fig 1A), due to a large reduction in I_to,fast with a slightly smaller increase in I_to,slow (Fig 1C–D).

Figure 1.

I_to current-voltage relation in mouse myocytes. (A) Mean data of total I_to which was significantly reduced in TG. (B) Original records. (C) I_to,slow amplitude was enhanced in TG but not reversed with AIP. Time constant of I_to,slow inactivation was slowed (right), but not reversed by AIP. (D) The reduction of total I_to was based on a smaller I_to,fast in TG and not influenced by AIP. The time constant of I_to,fast inactivation was not altered (right). (E) Steady-state currents were not affected. Average maximum I_to: 1.85±0.15 nA, 7.82±0.77 pA/pF, C_m 257±18.6 pF, R_S 8.5±0.6 MΩ. *P<0.05 vs. WT (ANOVA).

Figure 2.

I_to current-voltage relation in rabbit myocytes. (A) Mean data of total I_to. (B) Original records. (C) The increase of total I_to appears to be due to an increase of I_to,slow amplitude and not reversible by AIP, (D) while I_to,fast was not altered. Unchanged time constants of I_to,slow and I_to,fast inactivation (right panels). (E) Steady-state currents were not affected. Average maximum I_to: 0.32±0.03 nA, 2.12±0.16 pA/pF, C_m 151±4.8 pF, R_S 8.6±0.4 MΩ. *P<0.05 vs. βGal (ANOVA).

In CaMKIIδ_C mice, neither the E_m-dependence of I_to activation (both I_to components), nor the inactivation kinetics for I_to,fast nor the steady-state current were altered (Fig 1C–E). However, I_to,slow inactivated more slowly in the CaMKIIδ_C mice (Figure 1C, right). Neither the I_to amplitude effects nor the inactivation kinetic effects in CaMKIIδ_C mice were reversed by CaMKII-inhibition using AIP, implying that CaMKII alters functional channel (or regulator) expression.

In rabbit myocytes, CaMKIIδ_C-overexpression significantly enhanced total I_to (Fig 2A). This was the result of significant up-regulation of I_to,slow amplitude, without detectable alteration in I_to,fast (Fig 2C–D). No CaMKII-dependent shifts were detected in either the E_m-dependence or kinetics of inactivation. Again, the CaMKII-induced changes in total I_to and I_to,slow amplitude were not reversible using AIP, suggesting that functional upregulation of I_to,slow may occur after only 24 hr of CaMKIIδ_C adenoviral infection.

I_to recovery from inactivation

I_to recovery from inactivation was assessed using a two pulse protocol (Fig 3A). Compared to WT and β-gal control myocytes, I_to recovery from inactivation was significantly faster in CaMKIIδ_C-overexpressing mouse (Fig 3B–C) and rabbit (Fig 4A) myocytes. This accelerated recovery was completely reversed by AIP (Fig 4B). Recovery was bi-exponential and CaMKII accelerated both fast (k_fast) and slow (k_slow) recovery from inactivation, an effect reversed by AIP (Fig 4B, Table 1). Thus, CaMKII appears to acutely speed recovery of I_to,fast and I_to,slow, and this would increase I_to availability at high heart rates.

Figure 3.

I_to recovery from inactivation in mouse myocytes. (A) Protocol. (B) Mean data for total I_to. Inset shows I_to recovery within the first 2 s. Data was fit to a double exponential function (Table 1). (C) Original records. I_to recovery from inactivation was enhanced in TG vs. WT (*P<0.05, F-test) which was reversible with AIP (†P<0.05, F-test).

Figure 4.

(A) I_to recovery from inactivation in rabbit myocytes. Mean data for total I_to. *P<0.05 vs. βGal; †P<0.05 vs. CaMKIIδ_C (F-test) (B+C) Recovery rate constants and amplitudes of I_to,fast and I_to,slow. (B) I_to,fast and I_to,slow where significantly enhanced after CaMKII overexpression in rabbit and mouse myocytes and could be slowed using AIP. (C) After adenoviral CaMKIIδ_c overexpression, the amplitude of I_to,slow was significantly increased. I_to,fast was unchanged. In TG mice, I_to,slow amplitude was increased, but I_to,fast was significantly reduced. All effects were unaffected upon CaMKII inhibition. *P<0.05 vs. WT or βGal; †P<0.05 vs. corresponding vehicle (one-way ANOVA).

Table 1.

Fitting parameters for I_to recovery from inactivation

	Mouse [n]		Rabbit [n]
	TG CaMKIIδC [6] (WT) [7]	TG CaMKIIδC+AIP [8]	Ad-CaMKIIδC [13] (Ad-βGal) [14]	Ad-CaMKIIδC+AIP [8] (Ad-βGal+AIP) [8]
Afast (A/F)	1.92±0.38* (4.95±1.12)	2.10±0.65	0.51±0.06 (0.64±0.07)	0.60±0.09 (0.76±0.22)
Afast/y0	0.71±0.07* (0.89±0.04)	0.59±0.09	0.35±0.03* (0.54±0.05)	0.45±0.04 (0.62±0.08)
kfast (s−1)	42.8±5.2* (12.3±2.1)	26.9±1.6†	80.4±15.0* (42.9±3.8)	42.4±6.6† (38.3±10.2)
Aslow (A/F)	1.41±0.28* (0.24±0.12)	1.52±0.38	0.99±0.13* (0.60±0.11)	0.76±0.15 (0.48±0.18)
Aslow/y0	0.29±0.07* (0.11±0.05)	0.41±0.09	0.63±0.03* (0.46±0.05)	0.55±0.04 (0.38±0.08)
kslow (s−1)	7.9±1.5* (1.9±0.8)	1.8±0.9†	2.51±0.75* (0.85±0.17)	0.29±0.06† (0.63±0.11)
y0 (A/F)	3.75±0.48 (4.90±1.29)	4.14±0.87	1.46±0.15 (1.24±0.14)	1.36±0.21 (1.24±0.3)

Fit parameters measured in mouse (average maximum I_to 1.04±0.13 nA, 4.16±0.57 pA/pF, C_m 260±17.5 pF, R_S 8.6±0.6 MΩ) and rabbit (average maximum I_to 0.21±0.02 nA, 1.3±0.1 pA/pF, C_m 160.4±5.3 pF, R_S 8.7±0.5 MΩ) myocytes.. C_m=membrane capacitance, R_S=series resistance.

^*P<0.05 vs. βGal and WT.

^†P<0.05 vs. corresponding vehicle.

Since I_to,fast typically inactivates and recovers faster than I_to,slow, we compared the amplitudes of I_to recovery. CaMKII significantly increased the I_to,slow proportion (A_slow/(A_slow+A_fast)) and I_to,slow density in both rabbit and mouse, but this effect was not reversed by AIP (Fig 4C, Table 1). This may reflect a CaMKII-dependent increase in functional expression of I_to,slow. In contrast, I_to,fast density was not altered by CaMKII overexpression in rabbit, suggesting unaltered functional expression of I_to,fast. Conversely, CaMKIIδ_C mice exhibit decreased I_to,fast (Fig 4C, Table 1) which was not reversible by AIP. These conclusions match those from the inactivation kinetics (Figs 1–2) and strengthen the conclusions regarding I_to assignment to I_to,fast and I_to,slow.

K-channel expression and co-localization of CaMKII with K_V1.4, 4.2, and 4.3

To clarify whether there is a combination of altered channel expression changes (insensitive to AIP) as well as acute CaMKII modulation (reversed by AIP), we assessed mRNA and protein expression levels of the channel forming subunits thought to underlie I_to,slow (K_V1.4) and I_to,fast (K_V4.2/4.3) and the β-subunit KChIP2 that associates with Kv4.2/Kv4.3.

K_V1.4 expression was significantly enhanced by CaMKIIδ_C expression in mice and rabbits overexpressing CaMKIIδ_C (Fig 5A–B). This agrees with the observed increase of I_to,slow that is not acutely blockable by AIP. Notably, mRNA measurements (Suppl. Fig. 1A+B) roughly paralleled this data, consistent with CaMKII-dependent transcriptional regulation.

Figure 5.

K-channel expression. (A) Western blots of K_V1.4 and K_V4.2 from mouse heart. (B) Western blots of K_V1.4 and K_V4.3 in rabbit myocytes. (C+D) Confocal microscopy using immunocytological stainings of CaMKII, K_V1.4 (C), K_V4.3 (D). Merge indicates the close relationship of CaMKII with K_V1.4 and K_V4.3. *P<0.05 (t-test)

In contrast to the enhanced K_V1.4, expression, K_V4.2 was significantly down-regulated in TG mice vs. WT at both protein and mRNA levels (Fig 5A and Suppl. Fig. 1), while K_V4.3 expression was unaltered. The reduction of Kv4.2 protein expression was relatively small compared to the greatly reduced I_to,fast density. However, mRNA levels of KChIP2 were also reduced to about 50% in TG vs. WT (Suppl. Fig. 1A). Thus both Kv4.2 and KChIP2 decreases may contribute to the net reduction of I_to,fast density. This reduced expression of K_V4.2/4.3 agrees with previously published data from failing myocardium.^{6, 22–24} Reduced I_to,fast generated via K_V4.2/4.3 may be a general feature of cardiac remodeling in HF,²⁵ and this might be partly caused by CaMKII.

In rabbit myocytes, CaMKIIδ_C failed to alter I_to,fast (Fig 2D&4C) and there was likewise no significant change in K_V4.3 protein or Kv4.2 or Kv4.3 mRNA expression (Fig 5B and Suppl. Fig. 1B). There was an increase in KChIP2 mRNA levels (Suppl. Fig. 1B), but that may be unrelated to the unaltered I_to,fast amplitude.

Subcellular immunolocalization of CaMKII and K-channels (Fig 5C–D) show K_V1.4, K_V4.3 and CaMKII in rabbit myocytes. All proteins show a striated pattern, consistent with a co-localization in the transverse tubules.

Inward rectifying current I_K1

CaMKII effects on I_K1 were assessed using 500 ms pulses to varying E_m (I–V curves in Figure 6A). We compared I_K1 (which reversed near −80 mV) using the outward current amplitudes at negative E_m. Measuring I_K1 in the presence of AIP showed 30–40% I_K1 down-regulation upon both acute and chronic CaMKIIδ_C overexpression (▲ vs. △ in Fig 6). In accordance with this, Kir2.1 gene expression was downregulated by ~40% in TG mice, and the mean value was lower in CaMKIIδ_C rabbit myocytes (but not significantly; Suppl. Fig. 1). Thus, CaMKII downregulates I_K1 expression within 24 h.

Figure 6.

I_K1 current-voltage relation. (A) Mean data for peak I_K1 in myocytes from TG mice (vs. WT) and rabbit CaMKIIδ_C myocytes (vs. βGal). *P<0.05 vs. WT and βGal; †P<0.05 vs. TG CaMKIIδ_C and CaMKIIδ_C; #P<0.05 vs. WT+AIP and βGal+AIP (ANOVA). (B) Original traces. Each upper trace is the raw current, each mid trace shows the remaining current after BaCl₂, and each lower trace depicts the BaCl₂ sensitive current used for analysis.

In all cases, the measured I_K1 was reduced by AIP (squares vs. triangles, Fig 6A), from which we infer that there may be some basal level of CaMKII-dependent I_K1 activation. In the TG mouse the AIP-sensitive component of I_K1 was reduced, implying unaltered basal CaMKII-dependent regulation. However, in the rabbit, CaMKIIδ_C overexpression increased the AIP-sensitive I_K1 by >100% despite a lower baseline. This raises the intriguing possibility that short-term enhancement of CaMKII can activate I_K1. These changes were seen both in the inward component as well as the small but physiologically relevant outward component of I_K1 (not shown).

APD and computational modeling

Figure 7A–B show that in CaMKIIδ_C TG mice, APD was significantly prolonged over a broad range of frequencies, and that AIP did not reverse the effect. This result is consistent with the reductions in I_to,fast, total I_to and I_K1, which would all be expected to prolong APD. The lack of effect of AIP is consistent with the APD prolongation being due to the reduced expression of K_V4.2, KChIP2 and Kir2.1, rather than the acute CaMKII-dependent acceleration of I_to recovery.

Figure 7.

AP measurements. (A) Original traces at 0.25 Hz in mouse and rabbit myocytes. Insets show simulated traces at 0.25 Hz. (B+C) Mean data of APD at 60% (APD60) and 80% (APD 80) for 0.25–2 Hz. (B) Compared to WT, APDs were prolonged in TG; AIP did not influence APDs. (C) APDs were shorter in CaMKIIδ_C-overexpressing rabbit myocytes vs.βGal; AIP re-lengthened APDs.

In striking contrast, rabbit APD was significantly shortened by acute CaMKIIδ_C over-expression (vs. β-Gal), and this effect was largely reversed by AIP (Fig 7A&C). The APD shortening could be partly due to the CaMKII-induced enhancement of total I_to and I_to,slow amplitude (Figs 2C&4C) and higher K_V1.4 expression (Fig 5B), but that should not be AIP-sensitive. A more likely dominant explanation could be the CaMKII-induced acceleration of I_to recovery from inactivation (Fig 4B), which would increase I_to density during the AP. This would be especially the case for I_to,slow which takes several seconds to recover in the absence of CaMKII but where CaMKII doubles the rate of recovery (Fig 4A–B). A third explanation is the CaMKII-dependent increase of I_K1 density (Fig 6A), although this would come in to play late in the AP. CaMKII may also acutely modulate other ion channels.^{14, 26}

Because of the multiple currents during the AP, computational modeling can help identify the relative contributions of various CaMKII-sensitive currents. We made use of our rabbit ventricular AP model²⁷ and a mouse ventricular AP model²⁸ with some modifications to include CaMKII-dependent alterations on I_Ca and I_Na^{13, 14, 18} and the novel I_to and I_K1 data (see online supplement). Figure 7A (insets) shows the calculated APs and Suppl. Table 2 shows the predicted APD60 and 80. Over the full range of stimulation frequencies explored APD60 and 80 are increased in the TG-CaMKIIδ_C myocytes vs. WT (Suppl. Table 2). In the TG mouse, CaMKII effects on I_Na and I_Ca do not significantly impact the APD, whereas reduced I_to,fast and I_K1 conspire to prolong the AP. AP prolongation is not reversed at fast rates where TG-CaMKIIδ_C I_to recovery is faster.

Using our rabbit ventricular myocyte computer model, we showed that the increased rate of I_to recovery and enhancement of I_to,slow are essential for the observed APD shortening²⁷. In the present study the data on I_K1 was added to the model, but the resulting change was minimal. Therefore, CaMKII-dependent enhancement of I_to recovery from inactivation and I_to,slow amplitude appear to be the main determinant of AP duration changes in the rabbit myocytes.

Discussion

The present study shows that CaMKIIδ_C expression in ventricular myocytes induces alterations in both functional expression levels of I_to,fast, I_to,slow, and I_K1, as well as acute alterations in the gating of I_to and I_K1. Only the latter effects are reversed by inclusion of CaMKII inhibitors. CaMKIIδ_C reduces expression of I_to,fast (Kv4.2/4.3) and I_K1 (Kir2.1), but increases expression of I_to,slow (Kv1.4). CaMKIIδ_C activity accelerates the recovery of I_to,fast and I_to,slow from inactivation and activates I_K1. There are quantitative differences in these effects in the TG mouse model vs. the rabbit overexpression model and these changes can have complex effects on the AP (prolonging the mouse APD, shortening the rabbit APD). Since CaMKII expression levels are enhanced in HF,^{10, 11} these effects may predispose to ventricular arrhythmias.

Separation of I_to components and acute vs. long-term CaMKII effects

To separate I_to,fast and I_to,slow components of I_to we used the differences in time constants of recovery from inactivation ~10–30-fold)²⁹ and time constants of inactivation (5–10-fold).³ While K_V1.4 and K_V4.2/4.3 might inactivate and recover in a multi-exponential fashion,³⁰ both methods gave very similar results and the pedestal component was unchanged. This gives us confidence in the component assignments. This is further supported by the Kv1.4 and Kv4.2/4.3 expression data, which largely parallel the CaMKIIδ_C-induced changes in I_to,slow and I_to,fast amplitudes, respectively.

To distinguish between acute dynamic CaMKII-dependent regulation of I_to and longer term effects due to altered protein expression, we used the selective CaMKII inhibitor AIP. Effects that were reversed by AIP were deemed acute (potentially mediated by CaMKII-dependent phosphorylation of channel subunits), whereas CaMKII effects that were not prevented by AIP were assumed to be longer term alterations (e.g. in protein expression). These are functional distinctions, and complementary work is needed to define which specific CaMKII target sites are responsible for acute gating changes and longer term effects.

CaMKII regulates I_to functional expression

The CaMKIIδ_C-induced changes in amplitudes of I_to,fast and I_to,slow were insensitive to AIP. This was true for mouse and rabbit myocytes and for both methods of separating I_to components, and also for the slowing of I_to,slow inactivation in the mouse. CaMKIIδ_C overexpression enhanced I_to,slow amplitude and Kv1.4 expression in the TG CaMKIIδ_C mouse and after CaMKIIδ_C overexpression in rabbit myocytes (Table 2). This suggests that the up-regulation of Kv1.4 and I_to,slow is a relatively rapid consequence of CaMKIIδ_C overexpression. While Kv1.4 protein expression was significantly enhanced, Kv1.4 mRNA changes were less significant. This raises the possibility that CaMKII might enhance trafficking and/or stability of Kv1.4 in the sarcolemma.

Table 2.

Effects of CaMKIIδ_C overexpression on K-currents

	Ito,fast	Ito,slow	IK1*
Rabbit
amplitude (from recov.)	Ø	↑
amplitude (from inact.)	Ø	↑
amplitude (peak)			↓
Protein	=Kv4.3	↑Kv1.4
mRNA	=Kv4.2/3	=Kv1.4	↓Kir2.1
Mouse
amplitude (from recov.)	↓	↑
amplitude (from inact.)	↓	↑
amplitude (peak)			↓
Protein	↓Kv4.2/3	↑Kv1.4
mRNA	↓Kv4.2/3	↑Kv1.4	↓Kir2.1
Both
Ito amplitudes unaltered by AIP
Ito inactivation kinetics unaltered
Ito,fast and Ito,slow recover faster from inactivation (acutely AIP-sensitive)

Legend: Maximal currents, protein data, and mRNA levels were derived from I–V-curves, Western blots, and RT-PCR results, respectively. ↑=upregulation; ↓=downregulation; Ø=unaltered.

^*with AIP, which reduces I_K1 in all cases

I_to,fast density and K_V4.2 mRNA and protein were reduced in CaMKIIδ_C TG mice, but not during short-term CaMKIIδ_C overexpression in rabbit. There are several possible reasons: 1) the CaMKIIδ_C-dependent downregulation of I_to,fast takes longer than 24 h to develop, 2) I_to,fast functional downregulation depends on hypertrophy and/or HF-dependent pathways caused by CaMKIIδ_C overexpression, 3) rabbit and mouse regulate functional I_to,fast differentially. I_to and K_V4.2/4.3 downregulation are typical in many models of hypertrophy or HF,^{6, 22–24} and the fetal K_V1.4 subunit increases during hypertrophy²³ and in HF,²⁴ where CaMKII activity is increased.^{10, 11} A remaining question is how CaMKII mediates the altered expression of channel proteins responsible for I_to.

One potential pathway by which CaMKII may regulate transcription of transient outward K-channels involves phosphorylation of type II histone deacetylases.^{31, 32} Indeed, this system is more activated in rabbit and human HF where CaMKII is upregulated.^{9–11, 33}

The extent of Kv4.2 downregulation in CaMKIIδ_C mice (~20%) was less than the reduction of I_to,fast (~50%). KChIP2 is expressed in heart, and when it is co-expressed with K_V4.2 (but not with K_V1.4), it increases I_to amplitude, slows inactivation and enhances recovery from inactivation.³⁴ Here KChIP2 gene expression was reduced (by ~50%) in CaMKIIδ_C-TG mice. Thus, the I_to,fast density reduction may be due to both decreases in Kv4.2 and KChIP2. Notably, chronic CaMKII inhibition in mouse was reported to downregulate KChIP2,¹⁵ implying that CaMKII increases KChIP2 expression, which we do see in the rabbit experiments (Suppl Fig 1B). It is possible that the KChIP2 downregulation in TG CaMKIIδ_C-overexpression is secondary to the development of hypertrophic/HF signaling. This may also help explain why acute CaMKIIδ_C overexpression in rabbit failed to reduce I_to,fast.

CaMKII acutely alters I_to gating

The lack of AIP effects on I_to amplitude indicates that CaMKII does not exert acute regulation on I_to amplitude or E_m-dependence of activation. However, we found that CaMKII prominently accelerates I_to,fast and I_to,slow recovery from inactivation. This suggests that CaMKII acts directly on I_to.

In HEK293 cells CaMKII-dependent phosphorylation of Ser123 of K_V1.4 channels modulates the inactivation of I_to,slow leading to an accelerated recovery from inactivation.³⁵ Our study suggests that CaMKII enhances I_to,slow recovery also in rabbit and mouse myocytes.

Initial evidence for a CaMKII-dependent regulation of I_to,fast came from human atrial myocytes showing that CaMKII inhibition accelerated I_to,fast inactivation.³⁶ Similar results were obtained from rat ventricular myocytes³⁷ and in heterologeous expression of K_V4.2/K_V4.3.^{37, 38} It was suggested that CaMKII acts on K_V4.3 by a direct effect at Ser550, thereby prolonging open-state inactivation and accelerating the rate of recovery from inactivation.³⁸ At least in vitro, CaMKII phosphorylation sites at Ser438 and Ser459 of K_V4.2 have been identified.³⁹ Therefore, the CaMKII-dependent effect on I_to,fast recovery observed here may be mediated via phosphorylation of one or more of those sites.

CaMKII regulates functional expression and regulation of I_K1

We show that the AIP-insensitive I_K1 and Kir2.1 expression, were significantly downregulated in CaMKIIδ_C-TG mice and rabbit myocyte upon short-term CaMKIIδ_C overexpression (by 30–40%). So, CaMKIIδ_C reduces functional expression of I_K1. This may partly explain the downregulation of I_K1 in HF, where I_K1 is down-regulated ~40–50%,^{2, 22} and may be a consequence of chronically enhanced CaMKII activity in HF.

Acute modulation of I_K1 by CaMKII seems more complex. In all cases, acute CaMKII inhibition reduces I_K1 density, suggesting a constitutive CaMKII-dependent activating effect on I_K1 amplitude. The overall effect of CaMKIIδ_C overexpression was to increase I_K1 in rabbits, but to decrease I_K1 in the TG mouse. We infer that in the rabbit the acute activating effect of CaMKII on I_K1 exceeds the moderate decrease in expression induced by CaMKIIδ_C overnight. In the TG mice, the reduction in I_K1 expression may predominate over the acute activating effect of CaMKII, explaining the overall reduced I_K1.

Heterologous Kir2.X channels are regulated by PKA, PKC and tyrosine kinases and some phosphorylation sites in the C-terminal of Kir2.1 and Kir2.3 have been confirmed biochemically.⁴⁰ Therefore, it is possible that CaMKII may alter I_K1 through direct channel phosphorylation.

Cell hypertrophy

Myocyte hypertrophy seen in CaMKIIδ_C-TG mice¹³ could dilute K-channel current density even if no changes in channel transcription per cell occurred. There was no change in rabbit ventricular myocyte membrane capacitance during the 24 h of exposure to adenovirus (Suppl. Fig. 2), ruling out this complication. For the mouse, the cellular hypertrophy (Suppl. Fig 2) would modestly decrease (<25%) K-current density (if expression/cell is unaltered). Since I_to,fast density was reduced by ~50% and I_to,slow density was increased in the CaMKIIδ_C TG mice, this does not influence our conclusions.

Influence of altered K-currents on APD

CaMKIIδ_C overexpression prolonged APD in mouse, but shortened APD in rabbit. This disparity may depend on differences in CaMKII-induced changes and species differences in the ionic currents. The mouse APD is short and I_to critically determines APD.^{5, 41} CaMKIIδ_C overexpression in mouse myocytes decreased total I_to (Fig 1A), attributed to a large decrease of I_to,fast. At physiological heart rates in mouse I_to,slow will have less impact than implied by Fig 1. The net reduction in I_to may explain the prolonged APD in TG mice, although reduced I_K1 (Fig 6A) could also contribute. CaMKII overexpression increases I_Na and I_Ca in mice¹⁸ which would also prolong APD, so distinguishing the relative contributions may require computer modeling.

I_to can influence APD in the rabbit myocyte, but its influence on APD is weaker than for mouse. In rabbit myocytes overall I_to was slightly (Fig 2A), due to an acute up-regulation of I_to,slow (Fig 2C). This would shorten APD, and for the rabbit I_to,slow is more likely to be relevant. In addition, the accelerated recovery from inactivation would enhance I_to,slow availability. This would shorten APD, as observed. The increased I_K1 could also contribute to APD shortening, while the increases in I_Na and I_Ca^{14, 18} would prolong APD.

Computational modeling

To delineate the relative contributions of I_to and I_K1 to the APD we used computational modeling. We incorporated the measured CaMKII-dependent alterations on I_Ca, I_Na, I_to and I_K1 in a mouse myocyte model^{28, 42} (see supplemental data) similar to our previous study.²⁷ Equations for the fast and slow component of I_to and a Markovian formulation of I_Na^27,43 were introduced.⁴⁴ In TG CaMKIIδ_C mouse myocytes, the reduced I_to,fast is the dominant cause of the observed AP prolongation. While the greater I_Ca and I_Na did not significantly alter APD, the decreased I_K1 in the CaMKIIδ_C mouse slightly slows the terminal repolarization making a slight contribution to the APD increase.

For rabbit, we used our previously published model incorporating CaMKII-dependent alterations on I_Ca, I_Na and I_to.²⁷ The increased rates of I_to recovery and I_to,slow amplitude are essential for APD shortening, whereas the changes in I_Na and I_Ca gating alone tend to prolong the AP duration.²⁷ The addition of the increased I_K1 density caused only a few milliseconds further reduction in APD. The fact that AIP reversed the effects of CaMKIIδ_C on APD in rabbit myocytes indicates that the accelerated I_to recovery is the main determinant of the APD changes. The effect of CaMKII could potentially serve as a Ca-dependent physiologic feedback mechanism. That is, increased [Ca]_i (via CaMKII) could enhance I_to recovery, resulting in APD shortening and reduced Ca influx and [Ca]_i.