Calmodulin/CaMKII inhibition improves intercellular communication and impulse propagation in the heart and is antiarrhythmic under conditions when fibrosis is absent

Abstract

Aim In healthy hearts, ventricular gap junctions are mainly composed by connexin43 (Cx43) and localize in the intercalated disc, enabling appropriate electrical coupling. In diseased hearts, Cx43 is heterogeneously down-regulated, whereas activity of calmodulin/calcium-calmodulin protein kinase II (CaM/CaMKII) signalling increases. It is unclear if CaM/CaMKII affects Cx43 expression/localization or impulse propagation. We analysed different models to assess this.

Methods and results AC3-I mice with CaMKII genetically inhibited were subjected to pressure overload (16 weeks, TAC vs. sham). Optical and epicardial mapping was performed on Langendorff-perfused rabbit and AC3-I hearts, respectively. Cx43 subcellular distribution from rabbit/mouse ventricles was evaluated by immunoblot after Triton X-100-based fractionation. In mice with constitutively reduced CaMKII activity (AC3-I), conduction velocity (CV) was augmented (n = 11, P < 0.01 vs. WT); in AC3-I, CV was preserved after TAC, in contrast to a reduction seen in TAC-WT mice (−20%). Cx43 expression was preserved after TAC in AC3-I mice, though arrhythmias and fibrosis were still present. In rabbits, W7 (CaM inhibitor, 10 µM) increased CV (6–13%, n= 6, P< 0.05), while susceptibility to arrhythmias decreased. Immunoconfocal microscopy revealed enlarged Cx43 cluster sizes at intercalated discs of those hearts. Total Cx43 did not change by W7 (n= 4), whereas Triton X-100 insoluble Cx43 increased (+21%, n= 4, P< 0.01). Similar findings were obtained in AC3-I mouse hearts when compared with control, and in cultured dog cardiomyocytes. Functional implication was shown through increased intercellular coupling in cultured neonatal rat cardiomyocytes.

Conclusion Both acute and chronic CaM/CaMKII inhibition improves conduction characteristics and enhances localization of Cx43 in the intercalated disc. In the absence of fibrosis, this reduced the susceptibility for arrhythmias.

1. Introduction

Normal cardiac conduction of electrical excitation is essential for the synchronous beating of the heart. This electrical conductivity includes active conduction due to sodium channel activation (depolarization) and passive conduction through gap junction channels, which propagates the electrical charge from one cell to the other. In ventricular tissue, connexin43 (Cx43) is the major component of gap junction channels, and sodium channels mainly consist of the alpha-subunit protein Na_V1.5. Reduced expression and heterogeneous redistribution of Cx43 channels in the heart, as well as lateralization, negatively modify the conductivity of cardiac tissue in diseased hearts and enhance susceptibility for arrhythmias.^1–3 Besides Cx43, disturbed Na_V1.5 expression and localization, and increased cardiac fibrosis are contributing factors,^2,4 which often act in combination.

Calmodulin (CaM) and the multifunctional Ca²⁺ and calmodulin-dependent protein kinase II (CaMKII) are involved in cardiac arrhythmias, both in drug-induced arrhythmia models,^5,6 and in CaMKII overexpression animal models.^7,8 The available evidence suggests that CaM/CaMKII activation leads to arrhythmias via three potential mechanisms: (i) it can directly phosphorylate ion channels, thereby altering their function in a proarrhythmic manner;⁹ (ii) prolonged CaMKII activation gives rise to changes in ion channel expression patterns, called electrical remodelling;¹⁰ and (iii) long-term CaMKII activation ultimately culminates in cardiac hypertrophy, heart failure,^11–13 apoptosis of cardiomyocytes, and fibrosis formation,¹⁴ which may provide the substrate of re-entry-based arrhythmias.

There are several studies showing an interaction between CaM/CaMKII and Cx43: (i) Cx43 has a CaM binding site to regulate the gating of Cx43-mediated gap junction channels;^15,16 (ii) CaM inhibition by W7 prevented Ca–CaM complex-induced impairment of gap-junctional conductance;¹⁷ and (iii) bepridil, which has CaM inhibiting activity, increased the intercellular electrical coupling, and terminated re-entrant arrhythmias.¹⁸ These studies suggested, but did not directly test the hypothesis that CaM and CaMKII can influence the conductive properties of myocardium via modulation of Cx43.

We hypothesized that CaM/CaMKII has not only a direct effect on Cx43 channels by regulating its gating properties but also has an indirect effect by modifying their expression, or subcellular localization in the intercalated disc. To approach this issue, we applied several methodologies in different animal models to determine the effects of both acute and chronic inhibition of CaM/CaMKII on Cx43 localization, conduction velocity (CV), and arrhythmogenesis.

2. Methods

2.1 Ethics

All acute and chronic experiments were governed by the guidelines in each facility (Institutional Animal Care and Use Committees at Nagoya University, Japan and Utrecht University, The Netherlands), which were based on the NIH guidelines (guide for the care and use of laboratory animals) and the guidelines from Directive 2010/63/EU of the European Parliament on the protection of animals used for scientific purposes.

2.2 Acute experiments

The effect of acute CaM inhibition on cardiac Cx43-based conduction properties (molecular and electrophysiological) was investigated using rabbit hearts. Rat neonatal and canine cardiomyocytes were also used to determine gap-junctional intercellular communication and Cx43 subcellular distribution.

2.2.1 Optical mapping of Langendorff-perfused rabbit hearts

The experimental model and procedures of optical mapping have been described before.^18,19 For the electrophysiological experiments, Japanese white rabbits (Chubu Kagaku Shizai Co., Ltd, Japan) were anaesthetized by intravenous application of pentobarbital (30–40 mg/kg) and exsanguination performed by excision of the heart. Six rabbit hearts were Langendorff perfused with modified Krebs-Ringer solution, equilibrated with 95% O₂–5% CO₂ at 37°C. Complete atrioventricular block was produced by His bundle destruction. Cryoablation of the left ventricular (LV) endomyocardium was applied to make a thin epicardial layer of ventricular myocardium (∼1 mm thick). The hearts were stained with voltage-sensitive dye, di-4-ANEPPS. Motion artefacts due to heartbeat were eliminated by 100 mM of 2,3-butandione monoxime (BDM). The effect of acute CaM inhibition was tested through application of 10 µM W7. Hearts were illuminated by green LED lights, and the fluorescent image was recorded with a high-speed video camera at a sampling rate of 1000 frames/s. Isochrone maps of 4 ms intervals were generated from the filtered image, then CV was measured in a square of 18 × 18 mm around the stimulation site from the centre of the LV free wall. CV in longitudinal (CV_L) and transverse directions (CV_T) was calculated from the slope of a linear least-squares fit of the activation time plotted against the distance.

The space constant (λ), an index of intercellular electrical coupling, was estimated from the exponential decay of the electrotonic membrane depolarization in response to subthreshold stimulus as described.^18,20,21 A magnified image covering a square of 5 × 5 mm was obtained through a photographic lens with a longer focal distance (Micro-Nikkor 105 mm, f/2.8D, Nikon). A single subthreshold stimulus (20 ms, ∼0.8 times threshold) was delivered during electrical diastole after regular (basic cycle length: BCL 400 ms) suprathreshold stimuli (2.0 ms, ∼1.2 times threshold) through a Teflon-coated platinum wire electrode placed at the centre of the LV free wall. Myocardial excitability was reduced by increasing the extracellular K⁺ from 4 to 8 mM in order to induce subthreshold membrane potential responses.

Ventricular tachycardia or ventricular fibrillation (VT/VF) was induced by modified S1–S2 cross-field stimulation and classified by the length of arrhythmias into non-sustained (<10 s) and sustained (≥10 s) VT/VF. When VT/VF sustained for more than 60 s, it was terminated by DC shocks. Wave propagation patterns during VT/VF were also analysed by the phase mapping to detect the movement of phase singularity as the centre of spiral-wave re-entry.

2.2.2 Western blot and immunohistochemistry

Rabbit hearts were perfused on a Langendorff apparatus for 1 h with (n= 4) or without (n= 4) 10 µM W7 to see the effect of sub-acute CaM inhibition on Cx43 distribution in the hearts. Then, the hearts were removed from the set-up and immediately frozen by liquid nitrogen. For protein isolation (n= 8), the frozen tissue was treated with modified radioimmunoprecipitation buffer (10 mM Tris–HCl, 150 mM NaCl, 1% Triton X-100, 0.1% sodium dodecyl sulfate (SDS), and 0.5% sodium deoxycholate, pH 7.4). Total protein samples were loaded on 10% SDS gel and transferred on nitrocellulose membrane. The membrane was blocked by 5% milk and then incubated with mouse monoclonal anti-Cx43 antibody (BD Transduction Laboratories, 610062) to detect total Cx43. Detection was performed using ECL detection kit (GE Healthcare). Fractionation of total protein in Triton X-100 soluble proteins (non-junctional fraction: cytoplasmic Cx43 en route to the plasma membrane, Cx43 in the plasma membrane as undocked connexons, and Cx43 en route to degradation), and Triton X-100 insoluble proteins (junctional fraction: Cx43 in the docked connexons of gap-junctional plaques), was based on the Triton X-100 extraction method as previously described.²²

Immunoconfocal microscopy was performed on cryosections of frozen rabbit hearts, which were fixed by 4% paraformaldehyde (PFA). Cryosections were made in two different planes, longitudinal and transverse, relative to the apical-base axis of the heart. After permeabilization by 0.3% Triton X-100, Cx43 was labelled with mouse monoclonal anti-Cx43 antibody (Millipore, mab3068) and stained with green fluorescence Alexa-Fluor 488 (Molecular Probes, A11006). At least five cryosections were analysed from each of the three hearts per group.

Complete anaesthesia was induced in an adult normal sinus rhythm Mongrel dog (Marshall Bioresources, USA) by intravenous injection of pentobarbital (25 mg/kg) and maintained by isoflurane inhalation (1.5% in O₂ and N₂O, 1:2). Euthanasia was performed with exsanguination by excision of the heart. The heart was then washed with ice-cold Ca²⁺-free buffer and perfused with enzymatic solutions as previously described.²³ Isolated cardiomyocytes were cultured overnight, and then stimulated for 4 h with either 5 µM of W7 (CaM inhibitor) or 1–10 µM of KN93 (CaMKII inhibitor) to test acute CaM/CaMKII inhibition. Total protein was then isolated and prepared for western blot using the same procedure as described above. Cx43 was detected using mouse monoclonal anti-Cx43 (BD Transduction Laboratories, The Netherlands, 610062).

2.2.3 Scrape loading assay on cultured rat neonatal cardiomyocytes

The communication through gap junctions was assessed by a scrape loading assay on primary cultures of neonatal rat cardiomyocytes (NRCMs) that were prepared as reported previously.²⁴ Briefly, 1-day-old Wistar rats (Chubu Kagaku Shizai Co., Ltd) were sacrificed by cervical dislocation, thereafter ventricles were excised with enzymatic solution.²⁴ Preplating was applied to remove fibroblasts. Forty-eight hours after generating a confluent monolayer of NRCMs, either normal culture medium (n= 4) or medium with 10 µM W7 (n= 4) was added. Thirty minutes later, the monolayer was washed twice with normal phosphate-buffered saline (PBS). Then, PBS containing 0.05% of Lucifer yellow CH was applied to the dishes, and the monolayer was scraped using a plastic pipette tip. Ten minutes after scraping, the monolayer was washed again with PBS to remove background fluorescence. Confocal images were obtained by laser scanning microscopy (TCS SP8, Leica), which was controlled by LAS-X software. The distance between the scrape edge and the end of the dye-stained area was quantified to assess the permeability of Lucifer yellow through gap junctions.

2.3 Chronic experiments

Mice expressing cardiospecific autocamtide-3-related peptide (AC3-I)²⁵ were bred into a C57BL/6 background; we recently confirmed that the AC3-I mice have reduced CaMKII activity at multiple substrate proteins by comparative phosphoproteomic analysis.²⁶

2.3.1 Experimental set-up

To evaluate the effect of chronic CaMKII inhibition on Cx43 expression and conduction properties under basic conditions and during chronic pressure overload, 11 AC3-I mice (10–12 weeks of age) underwent sham surgery and another 11 AC3-I mice were subjected to transverse constriction of the aorta (TAC). All mice were followed for 4 months. For comparative reasons, historical data were obtained from TAC experiments in wild-type (WT) animals performed in the same strain of mice, for the same duration of 16 weeks.^3,4

AC3-I mice were TAC operated under inhaled isoflurane anaesthesia (2.5% in O₂), as previously described.³ Effectiveness of the constriction was confirmed by Doppler echocardiography (pressure gradient: 66 ± 4 mmHg). Sham animals received the same treatment, but without aortic constriction. Sixteen weeks after the surgery, mice were anaesthetized by 2.5% isoflurane in O₂ and subjected to the following experiments.

2.3.2 Electrocardiogram and echocardiography

A three-lead electrocardiogram (ECG) was recorded using PowerLab 4/30 and Dual Bio Amp and analysed offline using LabChart 7 Pro (all from ADInstruments Ltd., UK). Echocardiography was performed immediately after ECG analysis, to determine functional and structural characteristics (SONOS 5500, Philips Medical Systems).

2.3.3 Epicardial mapping of Langendorff-perfused mouse hearts

After the in vivo measurements, mice were exsanguinated by excision of the heart under deep-inhaled isoflurane anaesthesia (4–5% in O₂). Hearts were quickly mounted on a Langendorff apparatus for epicardial mapping during retrograde perfusion of the hearts. Perfusion buffer contained (mM) NaCl 116, KCl 5, MgSO₄ 1.1, NaH₂PO₄ 0.35, NaHCO₃ 27, glucose 10, mannitol 16, and CaCl₂ 1.8, which was carbogen-gassed at 37°C. A multielectrode 19 × 13 grid was placed on the epicardial surface of the heart. Stimulation was performed from the centre of the grid (two times stimulation threshold). Analysis of CV was done offline. The maximal negative dV/dt on the unipolar electrogram was defined as the time of local activation. The combined activation times of the electrodes allowed generation of an activation map, which subsequently could be used to determine CVs. All measurements were analysed with custom made software based on Matlab (The MathWorks, Inc.).

Assessment of arrhythmias was performed in three steps: (i) spontaneous arrhythmias, (ii) arrhythmias induced by a 16-paced train (BCL 100 ms) followed by 1–3 premature stimuli close to the effective refractory period, and (iii) arrhythmias induced by 2 s of burst pacing at the shortest possible cycle length.

2.3.4 Western blot and histological examination

Hearts were removed from the Langendorff apparatus and immediately frozen in liquid nitrogen. Protein lysate samples were made in a similar fashion as described for rabbit heart tissue. Mouse monoclonal anti-Cx43 antibody (BD Transduction Laboratories, 610062) was used to detect Cx43. To detect Na_V1.5, protein samples were loaded on 7% SDS gel and incubated with rabbit polyclonal anti-Na_V1.5 antibody (Sigma-Aldrich Corp., USA, S0819). Cryosections were generated (in four-chamber view) for histological analysis and immunohistochemistry. Immunolabelling using rabbit polyclonal anti-Cx43 (Invitrogen, 71–0700) antibodies and rabbit polyclonal anti-Na_V1.5 (Sigma-Aldrich Corp., S0819) antibodies was performed to assess subcellular distribution of Cx43 and Na_V1.5, as described previously.²⁷ For fibrosis detection, cryosections were fixed with 4% PFA and stained by Picrosirius Red, as previously described.²⁸ Microscopy was performed using an Eclipse 80i (Nikon, Tokyo, Japan), and analysis was done in ImageJ on four pictures per heart, from at least four hearts per group.

2.4 Statistical analysis

Data were shown as means ± standard error of the mean (SEM). Statistics were performed by one- or two-way ANOVA with Tukey's HSD post hoc test, repeated measures ANOVA with Sidak's post hoc test, Student's t-test, or Fisher's exact test when appropriate. All analyses were performed using GraphPad Prism 6.0 (GraphPad Software, La Jolla, CA, USA) or Origin 8 (Microcal Software, Inc., Northampton, MA, USA). Differences were considered significant if P< 0.05.

3. Results

3.1 Acute CaM inhibition

3.1.1 Improved conduction velocity and tempered arrhythmogenicity in rabbit hearts

In perfused rabbit hearts, W7 (10 µM) significantly increased CV_L and CV_T by 6.0 and 12.6%, respectively (BCL 800 ms, Figure 1A). The space constant (λ), an index of intercellular electrical coupling, was significantly increased by W7 in both longitudinal and transversal directions (λ_L, 1.52 ± 0.16 vs. 1.09 ± 0.08 mm, n= 6, P< 0.05; λ_T, 1.31 ± 0.07 vs. 0.93 ± 0.05 mm, n= 6, P< 0.01, Figure 1B and C).

Figure 1.

Electrophysiological changes by CaM inhibitor W7 in perfused rabbit hearts. (A) CV in longitudinal (CV_L) and transverse (CV_T) directions as measured by high-resolution optical mapping. BCL, basic cycle length. (B) Representative measurement of space constant in the absence (open circle) or presence (closed circle) of 10 µM W7. The amplitude of membrane voltage (Vm) in response to subthreshold stimulation decreased exponentially depending on the distance from the stimulation site. (C) Summarized data of space constant measured in longitudinal (λ_L) and transverse (λ_T) directions. *P< 0.05, **P< 0.01 vs. control (n= 6).

Sustained ventricular arrhythmias (VT or VF) were induced in 6/6 control hearts. With subsequent exposure to 10 µM W7, incidence of sustained VT/VF decreased to 4/6 hearts. Total incidence of VT/VF decreased from 53 to 35% by W7 (Figure 2A), and the percentage of sustained VT/VF over total incidence was significantly reduced from 82 to 33% (n= 6, P< 0.01). In 5 hearts, spiral-wave re-entry before and after application of 10 µM W7 was clearly seen (Figure 2B and C). Under control conditions (30 s after VT initiation), a stable rotor circulating around a line of functional block (∼4 mm in length) was observed (Figure 2B). The distant bipolar electrogram showed a monomorphic VT pattern. In the phase map, a phase singularity, the rotational centre of spiral wave, moved back and forth along the functional block line (Figure 2D, left). On the other hand, in the presence of 10 µM W7, spiral waves changed their circuits in a beat-to-beat manner with a polymorphic VT pattern (Figure 2C). The shape and length of the functional block line changed with every beat. Consequently, the rotation centre drifted upwards and collided with the atrioventricular groove, resulting in self-termination of the spiral wave. The phase singularity trajectory during these beats was illustrated in Figure 2D (right). Mean length of functional block line extended from 8.2 ± 2.1 to 16.7 ± 2.3 mm by 10 µM of W7 (P< 0.01, Figure 2E), revealing a larger spiral-wave meandering in the presence of W7.

Figure 2.

Characteristics of ventricular tachyarrhythmias (VT) induced in perfused rabbit hearts by cross-field stimulation. (A) (Non) sustained VT incidence (white and black bars, respectively) before and after application of 10 µM W7 (n= 6). (B, top) Electrogram of sustained VT induced in the absence of W7. (bottom) Isochrone map of VT recorded by high-resolution optical mapping. (C, top) Electrogram of self-terminated VT induced in the presence of 10 µM W7. (bottom) Three isochronal maps recorded at three different periods to show the meandering of the circuit of spiral-wave re-entry during non-sustained VT. (D) The trajectory of phase singularity during VTs shown in B and C. (E) Mean length of functional block line (FBL) in the absence and presence of 10 µM W7. **P< 0.01 vs. control (n= 6).

3.1.2 Augmented localization of Cx43 in the intercalated disc

As expected because of the acute nature of our study, the total amount of Cx43 was not significantly changed by W7 (Figure 3A). Surprisingly, Triton X-100 soluble Cx43 was significantly decreased (14%) by a 1 h W7 treatment, whereas Triton X-100 insoluble Cx43 level was significantly increased by 21% (Figure 3B). Using immunoconfocal imaging, no apparent changes in Cx43 distribution by W7 were observed in longitudinal sections, whereas Cx43 was well localized to the intercalated disc in both W7-treated and W7-untreated hearts (Figure 3C, top). Examining transverse cryosections, we noticed that the total area of intercalated disc was not changed (3696 ± 690 vs. 3933 ± 293 pixels in W7, n= 3, P= 0.77), while the fluorescent signal of Cx43 increased in the intercalated disc (30.0 ± 2.2 vs. 50.9 ± 0.6% in W7, n= 3, P< 0.01; Figure 3C, bottom). These results suggest that acute CaM inhibition increased the localization of Cx43 to the intercalated disc.

Figure 3.

The expression and localization of Cx43 in isolated rabbit hearts perfused for 1 h with or without 10 µM W7 (n= 4). (A) Representative blot of Cx43 in total lysate (left), Triton X-100 soluble (middle), or insoluble (right) fractions. The intensity of the blot signal was normalized by Ponceau staining. { indicates region of Cx43 immunoreactivity. (B) Summarized data of Cx43 expression [in total lysate (left), Triton X-100 soluble (middle), or insoluble (right) fractions]. **P< 0.01 vs. control. (C) Representative images of immunohistochemistry labelled for Cx43 (green). Frozen sections of perfused rabbit hearts were made in longitudinal (top) and transverse (bottom) directions. Scale bars represent 50 and 10 µm, respectively. Please note the increase in Cx43 labelling.

With respect to a physiological meaning of the increased accumulation of junctional Cx43, dye-coupled distance was increased after W7 treatment of scraped monolayers of cultured NRCMs, as depicted in Supplementary material online, Figure S1.

A different and more specific inhibitor of CaMKII (KN93) was also tested in a third model of acute inhibition to generalize the effects observed by W7 on Cx43. In isolated and cultured adult dog cardiomyocytes, the effect of W7 on total and Triton X-100 insoluble Cx43 level was compared with KN93. In Supplementary material online, Figure S2, it is depicted that both drugs increased the level of Triton X-100 insoluble Cx43 in a comparable fashion, without affecting the total Cx43 expression.

3.2 Chronic CaMKII inhibition in (diseased) mouse hearts

3.2.1 Electrical and structural remodelling upon chronic pressure overload

The basic data of sham- and TAC-operated AC3-I mice are depicted in Tables 1 and 2. Aortic constriction of 16 weeks in TAC AC3-I mice resulted in increased heart and lung weights when compared with sham AC3-I, also when heart weight was corrected for body weight (HW/BW, 8.3 ± 0.4 vs. 6.0 ± 0.4 mg/g, n= 11, P< 0.01, Table 1) or tibia length. Likewise, all echocardiographic parameters were significantly increased with the exception of fractional shortening (FS) which was significantly decreased in TAC AC3-I (FS, 35.8 ± 1.7 vs. 48.4 ± 0.8%, n= 11, P< 0.01, Table 1). Moreover, heart rate and other electrocardiographic parameters remained unchanged after TAC, except for QRS width, which was increased in TAC AC3-I compared with sham AC3-I (QRS, 11.4 ± 0.2 vs. 10.2 ± 0.3 ms, n= 11, P< 0.01, Table 2).

Table 1.

Tissue characteristics and echocardiographic measurements of sham- and TAC-operated AC3-I mice

	Sham	TAC
N	11	11
Organ weights
Body weight (g)	27.9 ± 1.1	28.6 ± 1.3
Heart weight (mg)	163 ± 10	236 ± 17†
HW/BW (mg/g)	5.97 ± 0.40	8.29 ± 0.42†
Tibia length (cm)	1.84 ± 0.02	1.76 ± 0.09
HW/TL (mg/cm)	89 ± 5	142 ± 17*
Liver weight (g)	1.21 ± 0.06	1.21 ± 0.08
Lungs weight (g)	0.183 ± 0.006	0.206 ± 0.005†
Kidneys weight (g)	0.176 ± 0.010	0.169 ± 0.010
Echocardiography
IVSd (mm)	0.0497 ± 0.0011	0.0714 ± 0.0013†
LVIDd (mm)	0.318 ± 0.013	0.373 ± 0.018*
LVPWd (mm)	0.0486 ± 0.0011	0.0715 ± 0.0013†
IVSs (mm)	0.0678 ± 0.0013	0.0889 ± 0.0010†
LVIDs (mm)	0.164 ± 0.007	0.242 ± 0.018†
LVPWs (mm)	0.067 ± 0.002	0.087 ± 0.002†
Aortic root diameter (mm)	0.123 ± 0.005	0.149 ± 0.005†
LVOT diameter (mm)	0.108 ± 0.004	0.131 ± 0.003†
LA dimension (mm)	0.145 ± 0.007	0.182 ± 0.007†
Pressure gradient (mmHg)	2.4 ± 0.2	65.7 ± 4.1†
FS (%)	48.4 ± 0.8	35.8 ± 1.7†

Values are mean ± SEM.

TAC, transverse aortic constriction; N, number of animals; HW/BW, heart weight to body weight ratio; HW/TL, heart weight to tibia length ratio; IVSd and IVSs, diastolic and systolic interventricular septal thicknesses, respectively; LV, left ventricle; LVIDd and LVIDs, LV diastolic and systolic internal diameters, respectively; LVPWd and LVPWs diastolic and systolic posterior wall thicknesses, respectively; LVOT, LV outflow tract; LA, left atrium.

*P< 0.05, ^†P< 0.01 vs. sham.

Table 2.

Electrocardiographic and electrophysiological measurements of sham- and TAC-operated AC3-I mice

	Sham	TAC
N	11	11
Electrocardiography
Heart rate (beats/min)	504 ± 26	537 ± 20
RR interval (ms)	123 ± 7	114 ± 5
PR interval (ms)	44.5 ± 1.1	47.1 ± 1.5
P duration (ms)	11.2 ± 0.2	11.2 ± 0.6
QRS interval (ms)	10.2 ± 0.3	11.4 ± 0.2†
QT duration (ms)	49.4 ± 0.6	50.6 ± 2.0
QTc duration (ms)	45.0 ± 1.0	47.5 ± 1.7
Electrophysiology
LV CVL (cm/s)	79.7 ± 4.4	86.2 ± 2.5
LV CVT (cm/s)	60.0 ± 2.5	53.5 ± 3.5
RV CVL (cm/s)	74.6 ± 3.8	80.7 ± 4.3
RV CVT (cm/s)	55.1 ± 3.7	49.6 ± 7.1
LV AR (CVL/CVT)	1.34 ± 0.08	1.67 ± 0.17
RV AR (CVL/CVT)	1.37 ± 0.05	1.75 ± 0.20
LV ERP (ms)	57.8 ± 3.6	50.0 ± 3.0
RV ERP (ms)	44.5 ± 3.9	41.8 ± 3.0
LV CV disp index	1.56 ± 0.18	1.59 ± 0.14
RV CV disp index	1.28 ± 0.22	1.40 ± 0.22

Values are mean ± SEM.

TAC, transverse aortic constriction; N, number of hearts; QTc, corrected QT interval using Bazett's formula; LV and RV, left and right ventricles, respectively; CV_L and CV_T, longitudinal and transversal conduction velocity, respectively; AR, anisotropic ratio; ERP, effective refractory period; CV disp, dispersion of conduction velocity.

^†P< 0.01 vs. sham.

3.2.2 Preservation of conduction velocity does not prevent arrhythmia induction

Electrophysiological data obtained from epicardial mapping are shown in Table 2 (bottom). In transgenic AC3-I mice, CV, both in longitudinal and transverse directions, and both in left and right ventricles, was clearly increased compared with WT mice (P< 0.01, Figure 4A and B). Interestingly, the increases in CV seen in sham AC3-I mice were preserved after TAC. A major finding was that the decreases in longitudinal CV normally seen in TAC-WT are completely prevented. These measurements confirmed that CaMKII inhibition resulted in enhanced CV in the TAC model of structural heart disease.

Figure 4.

CV and VTs induced in perfused TAC-operated mouse hearts. (A and B) CV in longitudinal and transverse directions measured by epicardial mapping on the left ventricle (LV) and right ventricle (RV), respectively. Sham and TAC AC3-I mice were compared with historical data of sham and TAC-WT mice.⁴ Please note that the CV reduction in WT TAC is not present in AC3-I TAC mice. *P< 0.05, **P< 0.01 (general experiments per group: 6–7). (C) Representative epicardial electrogram of a polymorphic VT induced by pacing in TAC AC3-I mice. Asterisks (*) indicate the last 5 burst-paced (cycle length 40 ms) complexes. (D) Comparison of the VT incidence in TAC AC3-I mice with TAC-WT mice from our previous studies.^3,4 Numbers in the bars indicate the amount of positive VT hearts over group size. (E) Activation maps from four numbered VT complexes indicated in the epicardial electrogram in C. Black isochronal lines of activation are 1 ms apart. Red colour represents the earliest activation time and blue colour the latest.

Despite the preservation of CV, AC3-I mice still showed arrhythmias on the Langendorff set-up after TAC (4/11, sham mice 0/11). Arrhythmias were burst pacing-induced, polymorphic, and sustained (>15 beats), as exampled in Figure 4C. The percentage of TAC AC3-I mice with arrhythmias (36.4%) was comparable with TAC-WT populations (historical data in identical C57BL/6J strain), as shown in the studies of Jansen et al.⁴ (18.2%) or Boulaksil et al.³ (44.4%) (Figure 4D). Importantly, the epicardial activation maps generated during the arrhythmias showed no signs of functional block, and CV was preserved in general (Figure 4E).

3.2.3 Chronic pressure overload does not change the levels of Cx43 and Na_V1.5 but induces fibrosis formation

The total amount of Cx43 expression was not changed in TAC AC3-I hearts when compared with sham AC3-I hearts as shown both by immunohistochemistry (Figure 5A and B) and western blot (Figure 5C and D). Additionally, there was no significant difference in total Cx43 from chronic CaMKII inhibition of sham AC3-I hearts compared with control mouse hearts from the same C57BL/6 background (Figure 5C and D). Interestingly, Triton X-100 insoluble Cx43 level was significantly increased after CaMKII inhibition in sham AC3-I when compared with control hearts (2.5 ± 0.2 vs. 1.0 ± 0.2, P< 0.05), and further increased after TAC AC3-I (2.5 ± 0.2 vs. 3.7 ± 0.3, P< 0.05, Figure 5E and F).

Figure 5.

Cx43 expression, Na_V1.5 expression, and fibrosis in isolated sham- and TAC-operated AC3-I mouse hearts compared with control mouse hearts from the same C57BL/6 background. (A) Representative pictures of Cx43 expression (red) in sham and TAC AC3-I hearts. Scale bar represents 100 µm. (B) Quantification of Cx43 immunolabelling exemplified in A. Western blots of Cx43 in total lysate (C) and in Triton X-100 insoluble fraction (E) with their respective Ponceau stainings below. { indicates region of Cx43 immunoreactivity. (D and F) Quantification of the blots (Cx43/Ponceau) represented in C and E, respectively. *P< 0.05; **P< 0.01 (n= 3–4 per group). (G) Representative pictures of Na_V1.5 expression (red) in sham and TAC AC3-I hearts. Scale bar represents 100 µm. (H) Quantification of Na_V1.5 immunolabelling exemplified in G. (I) Western blot of Na_V1.5 in total lysate and respective Ponceau staining below. (J) Quantification of the blots (Na_V1.5/Ponceau) represented in I (n= 3–4 per group). Control samples (C, E, and I) were loaded on the same membrane as respective AC3-I sham and TAC but were not adjoining (indicated by white line in between). (K) Representative pictures of collagen in sham and TAC AC3-I hearts. Scale bar represents 100 µm. (L) Quantification of collagen exemplified in K. *P< 0.05 vs. sham AC3-I. For quantification of (immuno)histological data, four pictures were taken from each heart, at least four hearts per group (B, H, and L).

Similar to our measurements of total Cx43, there was no change in total Na_V1.5 expression between sham and TAC AC3-I hearts or when compared with control hearts as assessed by immunohistochemistry (Figure 5G and H) and western blot (Figure 5I and J).

Fibrosis did show profound changes when assessed through Picrosirius Red staining. AC3-I mice displayed areas of patchy fibrosis after TAC, and total fibrosis levels were significantly increased, by a factor of four, when compared with sham (P< 0.05, Figure 5K and L).

4. Discussion

The central question of this research project was to directly determine whether the CaM/CaMKII pathway regulated CV, subcellular distribution of Cx43 levels and arrhythmogenesis. We discovered that (i) conduction characteristics improved and the level of Triton X-100 insoluble Cx43 (in the intercalated disc) was increased by acute CaM/CaMKII inhibition in rabbit hearts and dog cardiomyocytes, and by chronic CaMKII inhibition in AC3-I transgenic mice; (ii) acute CaM/CaMKII inhibition improved intercellular communication in NRCMs and decreased the incidence of re-entrant arrhythmic episodes in rabbit hearts; (iii) chronic CaMKII inhibition prevented TAC-induced reduction of CV and (Triton X-100 insoluble) Cx43 expression in AC3-I mice but failed to prevent structural and contractile remodelling, and had no suppressive effect on ventricular arrhythmias induced by programmed electrical stimulation.

4.1 CaM/CaMKII inhibition improves intercellular coupling and conduction velocity

In normal ventricles, Cx43 protein localizes to the intercalated disc and forms gap junctions that enable cardiomyocytes to transfer excitability and small metabolic solutes. Total Cx43 protein expression was neither affected by acute CaM inhibition in rabbit hearts nor by chronic CaMKII inhibition in transgenic AC3-I mice hearts. However, the level of Triton X-100 insoluble Cx43 was increased in both models (Figures 3, 5, and Supplementary material online, Figure S2), while the cluster size of Cx43-mediated gap junction was enlarged by treatment with W7 (Figure 3). In line with these observations, similar effects were shown in canine cardiomyocytes using KN93 and W7. We recognize that W7 and KN93 have off target, CaMKII-independent, actions at a number of ion channels that could affect CV.^5,29,30 However, our findings were confirmed using three mechanistically distinct antagonists, including transgenic AC3-I that has no known CaMKII-independent actions on ion channels. Thus, taken together, our findings support a view that the CaM/CaMKII pathway is an important signal for controlling Cx43 localization and CV in myocardium. Because we did not find a change in total Cx43 expression after W7, KN93 or in mice with transgenic myocardial expression of AC3-I, we speculate that CaMKII is involved in post-translational modification of Cx43 leading to an increased junctional positioning of this gap junction protein.

CaMKII phosphorylation sites have been identified on Cx43,¹⁶ although the physiological implication of this is not yet known. In our chronic inhibition of CaMKII, we found no obvious differences between phosphorylated and non-phosphorylated Cx43 levels (as suggested by the upper and lower Cx43 bands, respectively, in AC3-I sham vs. control groups of Figure 5C and E). Beyond phosphorylation, it is possible that other post-translational modifications of Cx43 may be involved (e.g. ubiquitination, SUMOylation, nitrosylation, hydroxylation, acetylation, methylation, and γ-carboxyglutamation).³¹

That the increased presence of Cx43 signals in the intercalated disc has functional implications was indicated in several ways. The dye transfer assay in NRCMs, indicative for coupling, revealed a higher number of coupled cells after acute CaM inhibition with W7 (Supplementary material online, Figure S1). Support for this observation comes from a recent study, demonstrating that Cx43-mediated gap junction channels are closed by a CaM-dependent mechanism.¹⁵

To speculate, the observed increased coupling and junctional presence of Cx43 could possibly explain the increase in CV observed in both studied models of CaM/CaMKII inhibition. In perfused rabbit hearts, acute treatment with W7 increased CV_L and CV_T (Figure 1). In a similar way, we have demonstrated that chronic CaMKII inhibition in AC3-I transgenic mice improved CV_L and CV_T under baseline conditions (Figure 4). When chronic pressure overload is induced through TAC surgery, CV is generally decreased.^3,4 Interestingly, upon 16 weeks of pressure overload in our transgenic AC3-I mice of chronic CaMKII inhibition, CV and the Triton X-100 insoluble Cx43 levels were preserved. To the best of our knowledge, this is the first time CaM/CaMKII is experimentally linked to regulation of cardiac CV.

Loss of Cx43 distribution to the intercalated disc is an important characteristic of heart failure that may contribute to abnormal excitation–contraction coupling and arrhythmia susceptibility. Our findings suggest that the CaM/CaMKII pathway, which is augmented in structural heart disease,³² may contribute to this pathological phenotype.

4.2 CaM/CaMKII inhibition does not alter Na_V1.5 protein levels

Another factor that importantly adds to intercellular impulse propagation is the functionality of the sodium channel Na_V1.5. Na_V1.5 can be directly phosphorylated by CaMKII, which alters its gating properties resulting in increased sodium current.^33,34 Although this suggests that CaMKII inhibition could decrease sodium channel function, we found no difference in Na_V1.5 protein expression between normal and CaMKII-inhibited mouse hearts (Figure 5J). It is thus tempting to speculate that Na_V1.5 expression is not contributing to the improved CV seen in AC3-I mice at baseline and that enhanced CV is mainly dependent on increased gap-junctional coupling.

Expression of Na_V1.5 remained also unchanged in TAC AC3-I hearts when analysed after 16 weeks of pressure overload. A recent study that used the same intervention to induce pressure overload showed an increase in Na_V1.5 after 1 and 5 weeks of TAC.³⁵ Furthermore, they showed an increase in late sodium current that could be reversed by inhibiting CaMKII. This, however, most likely reflects an earlier phase of cardiac remodelling that is not comparable with the phase studied in our approach.

4.3 Antiarrhythmic effect of acute and chronic CaM/CaMKII inhibition

We demonstrated with the ex vivo rabbit model that acute inhibition of CaM by W7 reduces arrhythmias, especially sustained VT/VF. The antiarrhythmic effect of W7 was previously shown in an in vivo rabbit model.³⁶ Additionally, we observed that spiral-wave re-entry arrhythmias in the presence of W7 were self-terminated. In a previous study, we demonstrated that early termination of spiral-wave re-entry could be explained by destabilization of the spiral waves due to increased intercellular coupling triggered by application of the gap junction opener Rotigaptide.²¹ Here, we show that CaM inhibition results in increased coupling between cultured NRCMs and increased Triton X-100 insoluble levels of Cx43 in dog cardiomyocytes and rabbits. In addition, our perfused rabbit hearts showed an increase in space constant indicative of increased coupling but also an increase in length of the functional block line. Both factors likely contribute to the destabilization of the spiral waves and therefore to its self-termination.

Previously, we showed that enhancement of gap junction conductance caused a large reduction in the incidence of sustained VT by 58%, with an increase in CV of only 10%.²¹ Additionally, we also showed that reduced (and heterogeneous) expression of Cx43 increases the susceptibility to cardiac arrhythmias.⁴ Here, acute CaM inhibition reduced sustained VT by 60%, while increasing CV by 6–9%. Therefore, the reduction in sustained VT observed in acute CaM inhibition could be explained by the increase in CV due to the increased gap-junctional Cx43, although a contribution of additional factors that might have been of additive influence for the observed antiarrhythmic effect of W7 should not be excluded given the rather small effect on CV. The antiarrhythmic effect in this special preparation was, however, not robust: a number of non-sustained VT/VF remained.

We also investigated potential antiarrhythmic effects of chronic CaMKII inhibition during pathological pressure overload by TAC surgery, as it has been shown that CaMKII activity is already increased after 1 week of TAC.³⁷ Many studies have demonstrated that acute inhibition of CaMKII through application of the pharmacological agent KN93 has antiarrhythmic effects in different animal models.^5,38,39 Although CV and Triton X-100 insoluble levels of Cx43 were increased in AC3-I TAC hearts, against our expectations, chronic CaMKII inhibition did not reduce susceptibility to arrhythmias. We speculate that this discrepancy is due to the unattenuated presence of fibrosis, which is a major proarrhythmic substrate in diseased hearts. Besides fibrosis, mechanical remodelling due to increased LV wall stress and poor mechanical function may also contribute to the presence of arrhythmias.

Although in our previous studies on WT TAC hearts the arrhythmias were re-entry based,^3,4 in AC3-I TAC mice, we found no evidence of re-entry-like activity. In fact, arrhythmias had a polymorphic nature suggesting focal activity possibly starting from early afterdepolarizations.

4.4 Structural hypertrophy and fibrosis upon chronic pressure overload

TAC surgery induces an increase in HW/BW, dilatation, an increased wall thickness and a reduced FS (Supplementary material online, Figure S3). In contrast to the effects observed with regard to intercellular coupling, this mechanical and structural remodelling after TAC was not prevented by chronic CaMKII inhibition (Supplementary material online, Figure S3). In several studies, CaMKII has been reported as an important factor for the progression of cardiac hypertrophy and that CaM/CaMKII inhibition could rescue hearts from hypertrophy after 3–8 weeks of TAC.^14,40,41 However, two recent independent studies using CaMKII knockout mice reported a similar level of hypertrophy compared with WT mice after 3 weeks of TAC.^42,43 Nonetheless, these studies had opposite findings regarding fibrosis; Kreusser et al.⁴³ showed preservation of fibrosis, while Cheng et al.⁴² showed a substantial increase in fibrosis compared with WT. We found that fibrosis remained present in our CaMKII inhibition model of 16 weeks of TAC but did not compare hearts with CaMKII inhibition to WT controls. This paradox separating preservation of Cx43/Na_V1.5 vs. increased fibrosis is new and of great interest. The most striking finding is that CV was improved compared with WT hearts, even in the presence of fibrosis after TAC surgery. The QRS interval prolongation after TAC was also not prevented by CaMKII inhibition, but this most likely can be explained by the increased cell size due to the observed hypertrophy, as was previously confirmed in a rabbit model.⁴⁴

4.5 Study limitations

In our study, we investigated acute CaM/CaMKII inhibition and chronic CaMKII inhibition. CaM activates not only CaMKII but also several Ca²⁺ handling proteins like calcineurin. We have to pay attention to the fact that CaM inhibition and CaMKII inhibition might not be the same and, as discussed above, that the currently available antagonists are suboptimal. Despite these limitations, the results of acute CaM inhibition and chronic CaMKII inhibition are similar: both increased CV and also the level of Triton X-100 insoluble Cx43 in the intercalated disc. These results suggest a common signal transduction pathway between CaM and CaMKII inhibition that could affect Cx43 and conduction properties.

We speculated about Na_V1.5 not contributing to the increased CV. However, this remark should be taken with caution since we only show results for total Na_V1.5 protein level. We attempted to evaluate the levels of Na_V1.5 in Triton X-100 soluble and insoluble fractions by western blot, but unfortunately were not successful.

Since we took advantage of historical data for WT mice, we acknowledge the possibility that slight differences in the actual execution of the experiments, the equipment used, and the fact that the ones that performed the experiments were different might potentially have been of minor influence.

5. Conclusions

Both acute and chronic CaM/CaMKII inhibition improves conduction characteristics and enhances localization of Cx43 in the intercalated disc. In the absence of fibrosis, this reduced the susceptibility for arrhythmias. These data suggest that CaM and CaMKII are potential therapeutic targets to reduce arrhythmogenic susceptibility via enhancement of intercellular electrical coupling.

Supplementary material

Supplementary material is available at Cardiovascular Research online.

Funding

We acknowledge the support from the Netherlands CardioVascular Research Initiative: the Dutch Heart Foundation, Dutch Federation of University Medical Centres, the Netherlands Organisation for Health Research and Development, and the Royal Netherlands Academy of Sciences (CVON-PREDICT), Dutch Heart Foundation Grant 2009B072, Foundation LeDucq: The Alliance for CaMKII signaling in heart disease, Grant-in-Aid for Scientific Research (C) 20590860 and Grant-in-Aid for Scientific Research (C) 26860215 from the Japanese Society for Promotion of Sciences, and Grant-in-Aid for Scientific Research on Innovative Areas 22136010 from the Ministry of Education, Culture, Sports, Science and Technology, Japan. We also thank SENSHIN Medical Research Foundation and the Japanese Society of Electrocardiology, and National Institutes of Health (NIH) Grants R01-HL079031, R01-HL096652, R01-HL070250, and R01-HL071140 for their financial support.