Elevated local [Ca²⁺] and CaMKII promote spontaneous Ca²⁺ release in ankyrin-B-deficient hearts

Abstract

Aims

Loss-of-function mutations in the cytoskeletal protein ankyrin-B (AnkB) cause ventricular tachyarrhythmias in humans. Previously, we found that a larger fraction of the sarcoplasmic reticulum (SR) Ca²⁺ leak occurs through Ca²⁺ sparks in AnkB-deficient (AnkB^+/−) mice, which may contribute to arrhythmogenicity via Ca²⁺ waves. Here, we investigated the mechanisms responsible for increased Ca²⁺ spark frequency in AnkB^+/− hearts.

Methods and results

Using immunoblots and phospho-specific antibodies, we found that phosphorylation of ryanodine receptors (RyRs) by CaMKII is enhanced in AnkB^+/− hearts. In contrast, the PKA-mediated RyR phosphorylation was comparable in AnkB^+/− and wild-type (WT) mice. CaMKII inhibition greatly reduced Ca²⁺ spark frequency in myocytes from AnkB^+/− mice but had little effect in the WT. Global activities of the major phosphatases PP1 and PP2A were similar in AnkB^+/− and WT hearts, while CaMKII autophosphorylation, a marker of CaMKII activation, was increased in AnkB^+/− hearts. Thus, CaMKII-dependent RyR hyperphosphorylation in AnkB^+/− hearts is caused by augmented CaMKII activity. Intriguingly, CaMKII activation is limited to the sarcolemma–SR junctions since non-junctional CaMKII targets (phospholamban, HDAC4) are not hyperphosphorylated in AnkB^+/− myocytes. This local CaMKII activation may be the consequence of elevated [Ca²⁺] in the junctional cleft caused by reduced Na⁺/Ca²⁺ exchange activity. Indeed, using the RyR-targeted Ca²⁺ sensor GCaMP2.2-FBKP12.6, we found that local junctional [Ca²⁺] is significantly elevated in AnkB^+/− myocytes.

Conclusions

The increased incidence of pro-arrhythmogenic Ca²⁺ sparks and waves in AnkB^+/− hearts is due to enhanced CaMKII-mediated RyR phosphorylation, which is caused by higher junctional [Ca²⁺] and consequent local CaMKII activation.

1. Introduction

Ankyrin-B (AnkB) is an adaptor protein that targets and tethers to the cytoskeleton select membrane proteins, including the Na⁺/Ca²⁺ exchanger (NCX),^1–6 the main route for Ca²⁺ extrusion in cardiac myocytes, and the Na⁺/K⁺ pump (NKA).^1–3,6,7 Loss-of-function mutations in AnkB cause a complex arrhythmogenic phenotype in humans that may include bradycardia, atrial fibrillation, conduction defects, long QT syndrome, and ventricular arrhythmias.^1–3,8–10 Several AnkB variants (including E1425G, V1516D, E1813K, L1622I, and R1788W) are associated with stress-induced ventricular tachyarrhythmias and sudden cardiac death.^1–3,8 However, the mechanisms that link deficient AnkB function to altered electrical signalling in the ventricle are largely unknown.

Spontaneous sarcoplasmic reticulum (SR) Ca²⁺ release during diastole, or SR Ca²⁺ leak, can cause Ca²⁺ waves that activate the Na⁺/Ca²⁺ exchanger and thus generate an inward current. This current depolarizes the membrane and, if large enough, triggers a spontaneous action potential.¹¹ When such depolarizations arise simultaneo­usly in a large number of neighbouring myocytes, they cause spontaneous excitation of the entire heart and may trigger ventricular tachy­arrhythmias or fibrillation.

SR Ca²⁺ leak is mediated by ryanodine receptors (RyRs) and occurs either in the form of Ca²⁺ sparks, when multiple (6–20) RyRs within a SR–sarcolemma junction are activated, or as smaller, invisible, events.^12–14 Using mice heterozygous for a null mutation in AnkB (AnkB^+/− mice), we previously found that AnkB deficiency leads to more frequent pro-arrhythmogenic Ca²⁺ sparks and waves during diastole.¹⁵ Despite higher Ca²⁺ spark frequency, the total SR Ca²⁺ leak was similar in myocytes from AnkB^+/− and wild-type (WT) mice.¹⁵ These results suggest that reduced AnkB function modifies the RyR gating to favour large Ca²⁺ release events (i.e. sparks) at the expense of smaller, non-spark releases. Notably, RyR open probability is augmented in AnkB^+/− vs. WT myocytes.¹⁶ We further found that membrane permeabilization with saponin equalizes Ca²⁺ spark frequency in AnkB^+/− and WT myocytes,¹⁵ which suggests that decreased AnkB function does not produce major changes in RyR expression and cluster organization. On the basis of these data, we concluded that the more coordinated RyR openings in AnkB^+/− myocytes are likely an indirect effect of AnkB reduction, mediated by altered RyR regulation at the cytosolic side.¹⁵

The RyR open probability, and thus Ca²⁺ spark frequency, is increased by several post-translational modifications of RyRs as well as by direct Ca²⁺ sensitization. While we found that diastolic [Ca²⁺]_i is similar in myocytes from AnkB^+/− and WT mice,¹⁵ RyRs respond to the local Ca²⁺ level in the junctional cleft ([Ca²⁺]_Cleft), which is generally different from [Ca²⁺]_i in bulk cytosol.^17–19 Here, we combine the assessment of RyR post-translational modifications, kinase and phosphatase activity assays, and direct measurements of [Ca²⁺]_Cleft to uncover the mechanisms that link deficient AnkB function to a higher frequency of pro-arrhythmogenic Ca²⁺ sparks and waves.

2. Methods

Detailed methods are included in the Supplementary material online.

2.1. Ventricular myocyte isolation

All animal experiments conform to the NIH guide for the care and use of laboratory animals and were approved by the Institutional Animal Care and Use Committee at University of Kentucky or the Ohio State University. Single ventricular myocytes were enzymatically isolated from hearts of mice heterozygous for AnkB null allele (AnkB^+/−) and WT littermates as previously described.¹⁵ Briefly, mice were anaesthetized with 3–5% isoflurane and hearts were excised quickly, mounted on a gravity-driven Langendorff perfusion apparatus and perfused with 1 mg/mL collagenase. When the heart became flaccid (11–15 min), the tissue was cut into small pieces, dispersed, and filtered, and the myocyte suspension was rinsed several times. A total of 20 AnkB^+/− and 20 WT mice were used for this study. All experiments were done at room temperature (23–25°C).

2.2. Ca²⁺ spark measurements in intact cardiac myocyte

Fluo-4-loaded myocytes were imaged in line-scan mode with a laser-scanning confocal microscope. Myocytes were perfused with 1 mmol/L Ca²⁺ Tyrode's solution and stimulated at various pacing frequencies (0.5, 1, or 2 Hz) until Ca²⁺ transients reached steady state. Stimulation was then stopped and spontaneous Ca²⁺ sparks were recorded in a 0Na⁺/0Ca²⁺ Tyrode's solution. Finally, the SR Ca²⁺ content was assessed from the amplitude of the Ca²⁺ transient generated by rapid application of 10 mmol/L caffeine. Ca²⁺ sparks were detected, counted, and characterized using SparkMaster.²⁰

2.3 Immunoblot

Proteins from left ventricle homogenates were separated by SDS–PAGE electrophoresis and transferred on PVDF membranes, which were then blocked and probed with primary antibodies against total RyR2 (Abcam, San Francisco, CA, USA), phospho(Ser2814)-RyR2, phospho(Ser2808)-RyR2, total phospholamban (PLB), phospho(Thr17)-specific PLB (Badrilla, Leeds, UK), total CaMKIIδ (Santa Cruz, Dallas, TX, USA), phospho(Thr286)-CaMKII (Abcam), oxidized CaMKII (ox-CaMKII(Met281/2); Genetex, Irvine, CA, USA), total HDAC4 (H-92, Santa Cruz), and phospho(Ser632)-HDAC4 (Abcam). Equal loading was verified by re-probing with anti-GAPDH. Bands were detected by chemiluminescence. Protein band intensity was quantified using ImageJ software (NIH, Bethesda, MD, USA).

2.4. Protein phosphatase assays

The activity of protein phosphatases 1 (PP1) and 2A (PP2A) was measured with a fluorescence-based assay kit (RediPlate™ EnzChek^R serine/threonine phosphatase assay kit, Molecular Probes, USA), according to manufacturer's instructions. Briefly, hearts were homogenized in homogenization buffer containing 150 mM NaCl, 50 mM Tris–HCl, 50 mM NaF, 2% Triton X-100, 0.1% SDS, and 1% (vol/vol) protease inhibitor cocktail (Millipore, Billerica, MA, USA), but no phosphatase inhibitors. For PP1 measurements, 25 µg protein from each sample was diluted in PP1 reaction buffer (optimized with 2 mM DTT and 200 µM MnCl₂) to a total volume of 100 µL, and incubated for 30 min in the absence and in the presence of the PP1-specific inhibitor tautomycin (30 nM). For PP2A measurements, 50 µg sample protein was diluted in PP2A reaction buffer (optimized with 1 mM NiCl₂) to a total volume of 100 µl, and incubated 30 min with or without the PP2A inhibitor okadaic acid (1 nM). Fluorescence intensity was assayed with a microplate reader using an excitation filter centred at 358 nm and an emission filter at 452 nm. Tautomycin and okadaic acid greatly reduced the fluorescence signal in homogenates from both WT and Ank+/− hearts (see Supplementary material online, Figure S1). For each sample, PP1 and PP2A activities were derived by subtracting the fluorescence intensity in the presence of the inhibitor from the signal obtained in its absence.

2.5. Measurement of local Ca²⁺ concentration in the junctional cleft

Local [Ca²⁺]_Cleft was measured using the GCaMP2.2-FKBP12.6 Ca²⁺ sensor as previously described.¹⁷ Isolated mouse cardiomyocytes were plated on laminin-coated cover slips, infected with an adenoviral construct expressing the GCaMP2.2-FKBP12.6 sensor (MOI = 100), and cultured overnight. Membrane staining with Di-8-ANNEPS indicated that mouse myocytes retain the rod shape and the T-tubules network after 24 h in culture (see Supplementary material online, Figure S2). GCaMP2.2 fluorescence was recorded with a wide-field fluorescence microscope coupled to a CCD camera (excitation = 490 nm, emission = 540 nm). Two-dimensional images were taken 1 s apart.

2.6. Statistical analysis

Statistical differences between groups were determined using the Student's t-test or two-way ANOVA, as appropriate. The data are presented as mean ± standard error. Differences were considered statistically significant when P < 0.05. Statistical analysis for single-cell experiments in Figures 2, 3, and 5 were done at the cell level. Data in Figures 2 and 3 were collected from 3 to 5 myocytes/mouse, while 6–8 cells/mouse were measured for data in Figure 5. Since the analysis included a comparable number of myocytes from each mouse, any clustering effects should be minimal.

Figure 2.

CaMKII inhibition greatly reduces Ca²⁺ spark frequency in AnkB^+/− mice. (A) Representative Ca²⁺ sparks recordings in myocytes from WT and AnkB^+/− mice. Cells were pre-conditioned by steady-state pacing at 1 Hz. (B) Mean Ca²⁺ spark frequency in WT and AnkB^+/− myocytes pre-conditioned by steady-state pacing at 0.5, 1, and 2 Hz. Data were collected from 17 myocytes (five different mice) for WT and 19 cells (four mice) for AnkB^+/−. (C) Ca²⁺ spark frequency normalized to the mean value in cells pre-conditioned by pacing at 0.5 Hz. (D and E) Effect of CaMKII inhibition with KN-93 (1 μmol/L) on Ca²⁺ spark frequency in myocytes from WT (D) and AnkB^+/− (E) mice. KN-93 data were collected from 20 myocytes (five mice) for the WT and 19 cells (four mice) for AnkB^+/− mice. (F) KN-93 suppresses the occurrence of Ca²⁺ waves in AnkB^+/− myocytes. In panels B–E, statistical differences between groups (AnkB^+/− vs. WT in panels B and C, control vs. KN-93 in panels D and E) were determined using two-way ANOVA with Bonferroni post-test to compare spark frequency for each rate of pre-conditioning pulses. *P < 0.05, **P < 0.01, and ***P < 0.001.

Figure 3.

The full width at half-maximum (FWHM) of Ca²⁺ sparks is larger in AnkB^+/− myocytes, while spark amplitude and duration are similar in cells from AnkB^+/− and WT mice. Statistical analysis was done using the two-way ANOVA test. P = 0.0067 for the AnkB-dependence of FWHM. Data are from the same cells as in Figure 2.

Figure 5.

[Ca²⁺]_Cleft is increased in myocytes from AnkB^+/− mice. (A) Working hypothesis. NCX and NKA density in the T-tubules is reduced in myocytes from AnkB^+/− mice, which may result in slower Ca²⁺ extrusion from the junctional cleft and consequently elevated [Ca²⁺]_Cleft compared with WT myocytes. This higher [Ca²⁺]_Cleft may activate RyR through direct sensitization and indirectly by activating local CaMKII. (B) Example of [Ca²⁺]_Cleft measurements with GCaMP2.2-FKBP12.6. (C) Mean decrease in fluorescence intensity upon application of 0Na⁺/0Ca²⁺ external solution and tetracaine for AnkB^+/− and WT myocytes. (D) [Ca²⁺]_Cleft calculated from data in panel C using the in situ characteristics of GCaMP2.2-FKBP12.6. (E) Mean decrease in GCaMP2.2-FKBP12.6 fluorescence upon inhibition of sarcolemmal Ca²⁺ transport in 0Na⁺/0Ca²⁺ solution (Sarc) and blockade of SR Ca²⁺ leak with tetracaine (SR). Data in panels C–E are from 24 cells from three different AnkB^+/− mice and 20 cells from three WT mice. Statistical differences between groups were determined using Student's t-test. *P < 0.05 and ***P < 0.001.

3. Results

3.1. Ca²⁺ spark frequency is higher in AnkB^+/− myocytes due to enhanced CaMKII-dependent phosphorylation of RyRs

We previously demonstrated that RyR gating is altered in myocytes from AnkB^+/− mice so that it favours larger Ca²⁺ release events (i.e. Ca²⁺ sparks) at the expense of smaller, non-spark releases.¹⁵ Several post-translational modifications of RyRs, including phosphorylation, are known to increase Ca²⁺ spark frequency. Here, we used immunoblot and phospho-specific antibodies to assess the phosphorylation status of RyR at Ser2814 and Ser2808, which are targets for phosphorylation by CaMKII and PKA, respectively (Figure 1). We found that RyR phosphorylation at the CaMKII site is enhanced in hearts from AnkB^+/− vs. WT mice (Figure 1A and B; P = 0.007), while phosphorylation at the PKA site is similar (Figure 1A and C; P = 1.0). These data agree well with previously published data.¹⁶

Figure 1.

RyR phosphorylation at the CaMKII-specific site is increased in hearts from AnkB^+/− mice. (A) Representative immunoblots on heart homogenates from WT and AnkB^+/− mice using antibodies that recognize RyR phosphorylated at the CaMKII (S2814) and PKA (S2808) sites and total RyR. (B and C) Ratio of phosphorylated-to-total RyR in hearts from AnkB^+/− and WT mice calculated from the relative band intensities. GAPDH was used as loading control. We investigated five hearts/group and immunoblots were repeated four times. Student's t-test was used for statistical analysis. **P < 0.01.

Thus, the higher Ca²⁺ spark frequency in AnkB^+/− myocytes may be due to elevated CaMKII-dependent RyR phosphorylation. To test this hypothesis, we measured Ca²⁺ sparks in the absence and in the presence of CaMKII inhibition (Figure 2). In these experiments, myocytes from WT and AnkB^+/− mice were pre-incubated in Tyrode's solution with or without the CaMKII inhibitor KN-93 (1 µM). After 20 min, cells were superfused with the same solution (±KN-93) and stimulated at various pacing frequencies (0.5, 1, and 2 Hz) until Ca²⁺ transients reached steady state. Stimulation was then stopped and spontaneous Ca²⁺ sparks were recorded in a 0Na⁺/0Ca²⁺ Tyrode's solution. As in our previous study, the frequency of Ca²⁺ sparks was significantly larger in AnkB^+/− myocytes vs. WT (Figure 2A and B; P = 0.002 using two-way ANOVA). Moreover, Ca²⁺ spark frequency raised rather steeply with the rate of pre-conditioning pulses in AnkB^+/− myocytes, while it was practically independent of pacing in the WT cells (Figure 2B and C). This characteristic was maintained after normalizing to the SR Ca²⁺ load (see Supplementary material online, Figure S3). This distinct dependence on pre-conditioning may thus be the result of increased diastolic Ca²⁺ accumulation, and consequent RyR activation, with higher pacing rates in AnkB^+/− myocytes. The width at half-maximum of Ca²⁺ sparks was higher in AnkB^+/− myocytes, while the spark amplitude and duration were similar in cells from AnkB^+/− and WT mice (Figure 3). Increased spark width with unchanged spark amplitude and duration implies a larger ‘spark mass’ as a result of an augmented Ca²⁺ flux per release event.

Inhibition of CaMKII with KN-93 greatly decreased Ca²⁺ spark frequency in myocytes from AnkB^+/− mice (P < 0.0001 using two-way ANOVA) but had little effect in the WT (P = 0.27; Figure 2D and E). KN-92, an inactive derivative of KN-93 that is ineffective at inhibiting CaMKII, did not significantly affect Ca²⁺ spark frequency in AnkB^+/− myocytes (see Supplementary material online, Figure S4). This result suggests that the effect of KN-93 is mediated specifically by CaMKII inhibition. With CaMKII blocked, Ca²⁺ spark frequency in AnkB^+/− myocytes was lower than in WT cells, despite a larger SR Ca²⁺ content (caffeine ΔF/F₀ = 9.4 ± 0.3 in AnkB^+/− vs. 8.4 ± 0.3 in WT myocytes treated with KN-93). It is thus possible that CaMKII inhibition unmasks a secondary mechanism through which AnkB affects RyR function that results in lower Ca²⁺ spark frequency in AnkB^+/− myocytes. This is conceivable since RyR activity is regulated by interaction with several proteins and various post-translational modifications, including oxidation, nitrosylation, and glutahionylation. However, the CaMKII-dependent activation of RyR is the prevalent physiological effect of AnkB reduction since Ca²⁺ spark frequency is increased in AnkB^+/− myocytes under control conditions.

CaMKII inhibition also abolished the occurrence of Ca²⁺ waves in AnkB^+/− myocytes (Figure 2F). Combined, the findings in Figures 1 and 2 suggest that the higher Ca²⁺ spark frequency in AnkB^+/− myocytes is due to enhanced RyR phosphorylation by CaMKII.

3.2. CaMKII-mediated protein hyperphosphorylation in AnkB^+/− myocytes is a local rather than global effect

Protein phosphorylation is the net result of a delicate balance between the activity of kinases and phosphatases. AnkB is essential for targeting the regulatory subunit B56α of protein phosphatase 2A (PP2A) and B56α localization is altered in AnkB^+/− myocytes.²¹ Abnormal B56α distribution may modify the PP2A activity and, since PP2A is associated with RyR, may alter RyR phosphorylation. However, we found that the phosphatase activity of PP2A was similar in hearts from AnkB^+/− and WT mice (Figure 4A; P = 0.14). Moreover, the activity of protein phosphatase 1 (PP1), the other phosphatase associated with RyR, was also unaffected in AnkB^+/− hearts (Figure 4A; P = 0.72). While we cannot exclude the possibility that ‘local’ activities of PP1 or PP2A are altered, these results indicate that the increased CaMKII-dependent phosphorylation of RyRs in AnkB^+/− hearts is not caused by decreased ‘global’ phosphatase activity. Therefore, we next tested whether the CaMKII activity is increased in AnkB-deficient hearts.

Figure 4.

CaMKII activity is increased but CaMKII-mediated phosphorylation of phospholamban and HDAC4 is not modified in AnkB^+/− hearts. (A) Relative activity of protein phosphatases PP2A and PP1 in hearts from AnkB^+/− vs. WT mice. n = 5 hearts/group. Experiments were performed in triplicate. (B) Ratio of CaMKII autophosphorylated at T287 to total CaMKII from AnkB^+/− and WT hearts calculated from the relative band intensities. The bottom panel shows a representative example. n = 5 hearts/group, and immunoblots were repeated six times. (C) Oxidized-to-total CaMKII in AnkB^+/− vs. WT hearts. n = 5 hearts/group, and immunoblots repeated four times. (D) PLB phosphorylated at Thr17 vs. total PLB in hearts from WT and AnkB^+/− mice. n = 5 hearts/group, and immunoblots were repeated six times. (E) Ratio of phosphorylated-to-total HDAC4 in AnkB^+/− and WT hearts. n = 5 hearts/group. Experiments were performed in triplicate. Student's t-test was used for statistical analysis. *P < 0.5.

Activation of CaMKII is initiated by the binding of Ca²⁺/calmodulin, which disrupts the association between the catalytic and regulatory domains of the kinase and exposes the catalytic domain for substrate binding. This conformational change also facilitates post-translational modifications of CaMKII, particularly autophosphorylation at the T287 site and oxidation at M281/M282, that result in autonomous activation of the kinase.²² We found significantly increased CaMKII phosphorylation at T287 in hearts from AnkB^+/− mice (Figure 4B; P = 0.017), while CaMKII oxidation was similar in AnkB^+/− and WT hearts (Figure 4C; P = 1.0). Thus, CaMKII is activated in AnkB^+/− hearts vs. the WT.

To determine whether this increase in CaMKII function is a global effect, we assessed the phosphorylation status of two other well-known CaMKII targets: phospholamban (PLB, at Thr17) and HDAC4 (at Ser632). In contrast to RyRs, the CaMKII-dependent phosphorylation of PLB and HDAC was not significantly enhanced in AnkB^+/− hearts (Figure 4D and E; P = 0.32 and 0.26 for PLB and HDAC4, respectively). This result raises the possibility that CaMKII is activated only locally, near RyR, and not within the entire myocyte. Indeed, RyRs are localized mainly in the junctional SR membrane, where the SR comes in very close proximity to sarcolemma,^23–25 whereas PLB and HDAC4 are located away from the junctions.

3.3. Local [Ca²⁺] in the junctional cleft is elevated in myocytes from AnkB^+/− mice

The activity of CaMKII in the junctional cleft could be elevated in myocytes from AnkB^+/− mice compared with the WT if local [Ca²⁺]_Cleft is higher in AnkB^+/− myocytes. This is conceivable, since AnkB directly associates with NCX,^1–6 and myocytes from AnkB^+/− mice show lower levels of NCX, particularly at the T-tubules, where the junctions are mainly located.¹ A reduction in junctionally localized NCX slows down Ca²⁺ extrusion from the cleft and may result in elevated [Ca²⁺]_Cleft (see cartoon in Figure 5A). This hypothesis is supported by prior studies demonstrating that NCX inhibition results in increased Ca²⁺ spark frequency without significant changes in the SR Ca²⁺ content.^26,27

We tested this hypothesis by measuring the diastolic [Ca²⁺]_Cleft in AnkB^+/− and WT myocytes using our newly developed GCaMP2.2-FKBP12.6 [Ca²⁺]_Cleft sensor.¹⁷ The sensor was created by attaching the genetically encoded Ca²⁺ sensor GCaMP2.2 to FKBP12.6, a protein that binds with high affinity and specificity to RyR monomers, but does not greatly influence RyR function.²⁸ GCaMP2.2-FKBP12.6 was expressed in intact myocytes by adenoviral infection and cells were used for experiments 20–24 h later. We previously demonstrated rigorously that expressed this way, the sensor reports local [Ca²⁺]_Cleft.¹⁷ To measure diastolic [Ca²⁺]_Cleft, myocytes were first field-stimulated (at 0.5 Hz) to reach steady state. Then stimulation was stopped and we blocked Ca²⁺ fluxes into the cleft by switching to a 0Na⁺/0Ca²⁺ external solution, which prevents Ca²⁺ movement through both l-type Ca²⁺ channels and NCX, followed by application of 1 mM tetracaine, a RyR inhibitor (Figure 5B). Under these conditions, [Ca²⁺]_Cleft declines, and in the new steady state (F₀), any [Ca²⁺] gradient between the cleft and bulk cytosol dissipates.¹⁷ Since diastolic [Ca²⁺]_i in the bulk cytosol is similar in AnkB^+/− and WT myocytes (85 ± 6 vs. 87 ± 5 nM, measured by us¹⁵), at the F₀ steady state, the sensor detects the same Ca²⁺ concentration in WT and AnkB-deficient myocytes. Therefore, the F₀ state was used to normalize the GCaMP-FKBP12.6 fluorescence intensity in different cells (Figure 5B). The decrease in GCaMP2.2-FKBP12.6 fluorescence induced by inhibition of Ca²⁺ fluxes into the cleft was significantly greater in myocytes from AnkB^+/− mice compared with the WT (Figure 5C; P = 0.014), which indicates that [Ca²⁺]_Cleft is higher in the AnkB^+/− myocytes. Using the in situ GCaMP2.2-FKBP12.6 characteristics that we previously measured¹⁷ (F_max/F_min = 6, K_d = 1100 nM, and n_Hill = 1), we calculated that [Ca²⁺]_Cleft is 299 ± 17 nM in AnkB^+/− myocytes and 234 ± 20 nM in WT cells (Figure 5D; P = 0.017). Thus, AnkB deficiency leads to a local increase of diastolic [Ca²⁺] in the junctional cleft.

For both WT and AnkB^+/− myocytes, the decrease in GCaMP2.2-FKBP12.6 signal was significantly greater upon inhibition of the SR Ca²⁺ leak compared with the blockade of sarcolemmal Ca²⁺ fluxes (Figure 5B and E; P < 0.0001). This means that SR Ca²⁺ leak rather than Ca²⁺ entry across sarcolemma is the main source of Ca²⁺ in the junctional cleft in myocytes from WT and AnkB^+/− mice, similar to what we previously found in rat myocytes.¹⁷

4. Discussion

We previously found that reduced AnkB function modifies the RyR gating so that a larger fraction of the SR Ca²⁺ leak occurs through Ca²⁺ sparks.¹⁵ These more coordinated RyRs openings result in a higher propensity for Ca²⁺ waves,^15,29 which may contribute to the increased arrhythmogenicity in AnkB-deficient hearts. Here, we found that the increased frequency of Ca²⁺ sparks and waves in AnkB^+/− myocytes is due to enhanced phosphorylation of RyRs by CaMKII. This result agrees well with prior observations that CaMKII inhibition normalizes the RyR open probability and rescues the abnormal electrical activity in AnkB^+/− hearts.¹⁶ We further demonstrated that the CaMKII-dependent RyR hyperphosphorylation is caused by augmented CaMKII activity rather than reduced function of protein phosphatases. Intriguingly, CaMKII activation seems to be limited to the junctional space where the sarcolemma and SR membrane come in close proximity since CaMKII targets located outside the junctions (phospholamban and HDAC4) are not hyperphosphorylated in AnkB^+/− myocytes. The activity of junctionally located CaMKII is likely increased in AnkB-deficient hearts by a higher local [Ca²⁺] in the cleft. Indeed, we found that although the global diastolic [Ca²⁺]_i is similar in myocytes from AnkB^+/− and WT mice, the local [Ca²⁺]_Cleft is significantly elevated in AnkB^+/− myocytes.

A prior study¹⁶ reported enhanced RyR phosphorylation at the CaMKII site in myocytes from AnkB^+/− mice. However, the cause for this increase (higher CaMKII activity vs. reduced phosphatase function) was not previously determined. AnkB binds to and targets the regulatory subunit B56α of PP2A,²¹ which raises the possibility that PP2A activity is altered in AnkB^+/− myocytes. However, we found similar PP2A function in WT and AnkB-deficient myocytes. The B56α subunit was recently shown to have an autoinhibitory role that suppresses PP2A activity, so that lower B56α levels resulted in fewer Ca²⁺ waves and sparks and decreased RyR phosphorylation.³⁰ This further supports our conclusion that the RyR hyperphosphorylation in AnkB^+/− myocytes is due to increased CaMKII activity rather than lower phosphatase activity.

[Ca²⁺]_i is regulated and signals differently in various subcellular microdomains, which greatly enhances its versatility as a secondary messenger.^31–32 In the heart, the SR Ca²⁺ release is controlled locally by the [Ca²⁺] in the restricted space between the sarcolemma and junctional SR. It is increasingly recognized that [Ca²⁺]_Cleft is regulated differently from the bulk [Ca²⁺]_i, both during electrical excitation and at rest.^{17–19,24,33} While the mechanisms leading to higher [Ca²⁺]_Cleft in AnkB-deficient myocytes compared with WT are not fully elucidated, we hypothesize that the reduced expression and altered localization of NCX is a major player. AnkB directly associates with NCX,^1–6 and AnkB mutants that lose the ability to target NCX to the membrane result in congenital arrhythmias.^1,3 Cardiac myocytes from AnkB^+/− mice show drastically reduced NCX localization at the T-tubules.¹ In control myocytes, ∼25% of the T-tubular NCX is co-localized with RyR at the junctions.²⁵ During diastole, Ca²⁺ removal by the junctionally located NCX keeps local [Ca²⁺]_Cleft low (Figure 5A) and limits Ca²⁺ release within a given RyRs cluster.³⁴ Reduced NCX expression in AnkB^+/− myocytes slows down Ca²⁺ extrusion from the cleft, which may result in elevated [Ca²⁺]_Cleft and therefore enhanced spark-mediated SR Ca²⁺ release. Studies showing that NCX inhibition leads to increased Ca²⁺ spark frequency in the absence of significant changes in the SR Ca²⁺ content^26,27 support this hypothesis. Moreover, functional evidence (reduced availability of l-type Ca²⁺ channels and enhanced ability of the non-inactivated channels to trigger SR Ca²⁺ release) indicates that [Ca²⁺]_Cleft is elevated in myocytes from cardiac-specific NCX knockout mice.³⁵ Besides activating local CaMKII and consequently increasing the CaMKII-mediated RyR phosphorylation, elevated [Ca²⁺]_Cleft may also promote the occurrence of Ca²⁺ sparks and waves through direct RyR sensitization.

In summary, we found that the higher incidence of pro-arrhythmogenic Ca²⁺ sparks and waves in AnkB-deficient myocytes is due to enhanced phosphorylation of RyRs by CaMKII. This CaMKII-dependent RyR hyperphosphorylation is the result of activation of CaMKII located at the sarcolemma–SR junctions by an elevated [Ca²⁺] in the junctional cleft.

Supplementary material

Supplementary material is available at Cardiovascular Research online.

Funding

This work was supported by the National Institutes of Health (grants HL109501 to S.D. and HL084583 to P.J.M.).