Tubulin polymerization disrupts cardiac β-adrenergic regulation of late I_Na

Abstract

Aims

The anticancer drug paclitaxel (TXL) that polymerizes microtubules is associated with arrhythmias and sinus node dysfunction. TXL can alter membrane expression of Na channels (Na_V1.5) and Na current (I_Na), but the mechanisms are unknown. Calcium/calmodulin-dependent protein kinase II (CaMKII) can be activated by β-adrenergic stimulation and regulates I_Na gating. We tested whether TXL interferes with isoproterenol (ISO)-induced activation of CaMKII and consequent I_Na regulation.

Methods and results

In wild-type mouse myocytes, the addition of ISO (1 µmol/L) resulted in increased CaMKII auto-phosphorylation (western blotting). This increase was completely abolished after pre-treatment with TXL (100 µmol/L, 1.5 h). The mechanism was further investigated in human embryonic kidney cells. TXL inhibited the ISO-induced β-arrestin translocation. Interestingly, both knockdown of β-arrestin2 expression using small interfering RNA and inhibition of exchange protein directly activated by cAMP (Epac) blocked the ISO-induced CaMKII auto-phosphorylation similar to TXL. The generation of cAMP, however, was unaltered (Epac1-camps). CaMKII-dependent Na channel function was measured using patch-clamp technique in isolated cardiomyoctes. ISO stimulation failed to induce CaMKII-dependent enhancement of late I_Na and Na channel inactivation (negative voltage shift in steady-state activation and enhanced intermediate inactivation) after pre-incubation with TXL. Consistent with this, TXL also inhibited ISO-induced CaMKII-specific Na channel phosphorylation (at serine 571 of Na_V1.5).

Conclusion

Pre-incubation with TXL disrupts the ISO-dependent CaMKII activation and consequent Na channel regulation. This may be important for patients receiving TXL treatments, but also relevant for conditions of increased CaMKII expression and enhanced β-adrenergic stimulation like in heart failure.

1. Introduction

Paclitaxel (TXL), which causes polymerization of cytoskeleton protein tubulin, is used in the treatment of ovarian and breast cancers. Cardiac side effects of TXL treatment have been reported including bradycardia, cardiac conduction abnormalities, ventricular tachyarrhythmias, and sudden cardiac death.^1,2 Voltage-gated Na channels are responsible for the action potential upstroke and are important for sinus node function and normal conduction of electrical impulses in the heart. Interestingly, changes in cardiac Na current (I_Na) can lead to sinus node dysfunction,³ life-threatening arrhythmias, and contractile dysfunction.^4,5 Moreover, it was shown that pre-incubation with TXL can lead to disturbed function of cardiac voltage-gated Na channels.^1,2,6 Thus, cardiac side effects of TXL treatments may be mediated by disturbed Na channel function. The mechanisms, however, are poorly understood. It was shown that Na current depends on intracellular trafficking of Na_V1.5, the pore-forming subunit generating I_Na,⁷ suggesting that an intact cytoskeleton is important for correct function of Na channels. The role of the cytoskeleton for regulation of cardiac Na channels has been the subject of recent studies. Both cytochalasin D, which disrupts actin polymerization, and also TXL can reduce peak I_Na.^6,8 Interestingly, it was recently shown that TXL also slows I_Na intermediate inactivation, which could be pro-arrhythmogenic and cannot be merely explained by reduced microtubule-dependent targeting of Na channels to the sarcolemma.⁶ Thus, TXL may also interfere with Na channel regulation. We and others have shown that calcium/calmodulin-dependent protein kinase II (CaMKII) can phosphorylate Na_V1.5.^4,9–11 Increased CaMKII-dependent Na channel phosphorylation results in enhanced I_Na intermediate inactivation, slowed recovery from inactivation, and enhanced late I_Na.^4,9–11 CaMKII is a cytosolic kinase that becomes activated by β-adrenergic stimulation,^12,13 and is required for the physiological β-adrenergic response of the sinus node.¹⁴ It seems conceivable that β-adrenergic CaMKII activation and consequent Na channel regulation may depend on an intact cytoskeleton, which could be hampered by TXL.

Thus, we tested in the present study if pre-incubation with TXL disturbs the β-adrenergic activation of CaMKII and if disturbed CaMKII activation underlies the TXL-dependent effects on Na channel gating.

Here, we show that pre-treatment with TXL disrupts the isoproterenol (ISO)-induced activation of CaMKII possibly by interfering with β-arrestin translocation. We also show that the ISO-induced CaMKII-specific Na channel phosphorylation was inhibited by pre-incubation with TXL, resulting in severely altered Na channel function. These results emphasize the role for microtubules in β-adrenergic signalling and may have implications for patients receiving TXL treatment.

2. Methods

A detailed description of Methods can be found in Supplementary material online.

2.1. Myocyte isolation and incubation with microtubule stabilizer TXL

Eleven- to 16-week-old CaMKIIδ knockout mice (CaMKIIδ^−/−, 37 mice)¹⁵ were compared with their age- and sex-matched wild-type (WT) littermates (40 mice). Anaesthesia was initiated by exposing mice to 2% isoflurane (in 100% O₂) under a glass cover. After adequate anaesthesia had been reached, the anaesthetized animal was euthanized by cervical dislocation. All animal procedures were performed in accordance with the Guide for the Care and Use of Laboratory Animals and approved by the Institutional Animal Care and Use Committee. Ventricular myocytes were isolated according to established enzymatic procedures,⁴ incubated with 100 μmol/L of TXL for at least 1.5 h at room temperature, and used for electrophysiological investigation, western blotting, and immunocytochemistry. For some experiments, CaMKII was activated by exposure to ISO (1 μmol/L for 10 min).

2.2. Measurement of ISO-induced β-arrestin translocation

ISO-induced β-arrestin2 translocation was measured in HEK293a cells transfected with β₁-adrenoreceptor, G protein-coupled receptor kinase 2 (GRK2), and β-arrestin2-GFP plasmids using confocal microscopy.

2.3. Measurement of cAMP generation using fluorescence resonance energy transfer

Measurement of cAMP levels was performed in Epac1-camps-transgenic mouse myocytes using fluorescence resonance energy transfer (FRET) imaging experiments.

2.4. Patch-clamp experiments

Ruptured-patch whole-cell voltage-clamp was used to measure I_Na at room temperature as detailed in Supplementary material online.

2.5. Western blots and immunocytochemical staining

General expression levels as well as levels of phosphorylated CaMKII and Na_V1.5 were investigated by western blotting and immunocytochemical staining (see Supplementary material online for details).

2.6. Statistics

All data are expressed as mean ± SEM. Fits were tested for significant difference using F tests. For longitudinal data, two-way repeated-measures ANOVA was performed; where appropriate, one-way ANOVA with multiple comparison test (Newman–Keuls test) was used. Otherwise, Student's unpaired t-test was used. A two-sided P-value of <0.05 was considered significant.

3. Results

Enhanced TXL-dependent tubulin polymerization was confirmed via immunocytochemical analysis (see Supplementary material online, Figure S1). Western blots were used to investigate CaMKII expression and auto-phosphorylation as a measure of CaMKII activity. While global CaMKII expression was unchanged, ISO increased CaMKII auto-phosphorylation (p-CaMKII/CaMKII, vehicle vs. ISO, 1 vs. 1.33 ± 0.11, n = 15 each, P < 0.05, Figure 1A). Pre-incubation with TXL, however, completely abolished the ISO-induced increase in CaMKII auto-phosphorylation (TXL vs. TXL + ISO, 0.70 ± 0.08 vs. 0.88 ± 0.08, n = 11 vs. 15, Figure 1A). In addition, immunocytochemical staining revealed that ISO increased the amount of membrane-localized CaMKII and also p-CaMKII at the intercalated disc region (ICD, placement of region of interest is indicated in Supplementary material online, Figure S3). We have previously shown that CaMKII also co-localizes with Na_V1.5 in membrane regions with a periodic pattern typical for z-line and t-tubule, respectively.⁴ Interestingly, immunocytochemistry analysis also showed that ISO increased CaMKII and p-CaMKII expression in this z-line region (Figure 1B and C). Similar to global activation, TXL abolished the ISO-dependent increase in CaMKII expression at ICD and z-line region (Figure 1B and C). Pre-incubation with TXL also significantly reduced the ISO-induced CaMKII auto-phosphorylation at both regions (ISO vs. TXL + ISO, P < 0.05, Figure 1B and C). However, when compared with TXL alone, ISO still increased p-CaMKII expression (TXL vs. TXL + ISO, P < 0.05, Figure 1B and C).

Figure 1.

CaMKII expression and p-CaMKII (at threonine 286) in WT mouse myocytes. (A) Western blot original data (upper panel) and densitometric mean for p-CaMKII (mid panel) or CaMKII (lower panel). (B and C) Original data for immunocytochemistry (upper panel) and mean fluorescence intensity at intercalated disc (mid panel) and z-line region (lower panel) for CaMKII (B) or p-CaMKII (C). While general expression of CaMKII was unchanged, ISO increased CaMKII expression and auto-phosphorylation which was abolished by TXL. *P < 0.05 vs. vehicle, ^†P < 0.05 vs. ISO. ^#P < 0.05 vs. TXL. Scale bar indicates 10 µm; calibration bar indicates colour-coded fluorescence intensity.

3.1. Mechanism of ISO-induced CaMKII activation

β-Arrestin2 membrane translocation is required for ISO-induced CaMKII activation.¹⁶ To further understand the mechanism of ISO-induced CaMKII activation and the involvement of microtubules, we measured the β-arrestin2 membrane translocation in HEK293a cells transfected with GFP-tagged β-arrestin2. Indeed, pre-incubation with TXL abolished the ISO-induced β-arrestin2 membrane translocation (Figure 2A). Interestingly, also in our model, silencing β-arrestin2 gene expression using small interfering RNA-transfected HEK293a cells completely abolished the ISO-induced increase in CaMKII auto-phosphorylation (p-CaMKII), assessed by western blotting (Figure 2B). To test whether the ISO-induced cAMP generation was also affected by TXL, we measured cAMP levels in mouse myocytes ubiquitously expressing an FRET cAMP sensor (Epac1-camps).¹⁷ Figure 2C and D shows that the ISO-induced cAMP generation was unaffected by TXL pre-treatment, arguing against an important role for cAMP-dependent protein kinase A (PKA). In addition, PKA inhibition with PKA inhibitor (PKI) (5 µmol/L) did not affect the ISO-induced increase in p-CaMKII levels, assessed by western blotting (Figure 2E). On the other hand, direct stimulation of exchange protein directly activated by cAMP (Epac) using 8-(4-chlorophenylthio)-2′-O-methyladenosine-3′,5′-cyclic monophosphate (8-pCPT-2′-O-Me-cAMP, 8-CPT, 5 µmol/L) resulted in an increase in p-CaMKII levels comparable with ISO (Figure 2E) and this increase was also blocked by pre-incubation with TXL, suggesting that both β-arrestin2 and Epac are involved in ISO-dependent activation of CaMKII in a microtubule-dependent manner.

Figure 2.

ISO-dependent activation of CaMKII requires β-arrestin2. (A) Original scans (left panel) of HEK293a cells co-transfected with β₁-adrenoreceptor, GRK2, and β-arrestin2-GFP before and after exposition to ISO (5 min). Arrow indicates membrane localization of β-arrestin2-GFP. Right panel: mean change in membrane fluorescence intensity. Exposition to ISO (1 µmol/L) results in the translocation of β-arrestin2-GFP to the membrane, which was abolished by TXL. *P < 0.05 vs. vehicle. (B) Western blot original data (upper panel) and densitometric mean (lower panel) from HEK293a cells stably expressing β₁-adrenoreceptors and co-transfected with Epac1, CaMKIIδ, and either β-arrestin2 siRNA or control siRNA. Silencing β-arrestin2 expression abolished the ISO-induced increase in p-CaMKII levels similar to pre-incubation with TXL. (C) Original scans of fluorescence emission (CFP at 490–515 nm and FRET at 530–600 nm) of a ventricular myocyte from an Epac1-camps-transgenic mouse at baseline and after exposure to ISO or forskolin. (D) Left panel shows exemplary recordings of cyan fluorescent protein (CFP)/FRET ratio for myocytes pre-treated with TXL or vehicle. Right panel: mean data shows no difference in the ISO or forskolin-induced change in CFP/FRET ratio with TXL. (E) Western blot original data (upper panel) and densitometric mean (lower panel) from HEK293a cells stably expressing β₁-adrenoreceptors and co-transfected with Epac1 and CaMKIIδ. Inhibition of PKA (PKI, 5 µmol/L) did not affect the ISO-induced increase in p-CaMKII levels. Interestingly, direct Epac stimulation with 8-CPT (5 µmol/L) increased p-CaMKII levels, but this was also blocked by pre-incubation with TXL. *P < 0.05 vs. vehicle, ^†P < 0.05 vs. ISO, ^#P < 0.05 vs. 8-CPT (one-way ANOVA). Scale bar indicates 10 µm; calibration bar indicates colour-coded fluorescence intensity.

3.2. CaMKII-dependent Na channel phosphorylation

To test whether TXL interferes with ISO-dependent CaMKII-specific Na channel phosphorylation, we measured Na channel expression and phosphorylation. Western blotting revealed that ISO increased CaMKII-specific Na channel phosphorylation (vehicle vs. ISO, 1 vs. 1.52 ± 0.13, n = 16 vs. n = 15, P < 0.05, Figure 3A). A similar increase in p-Na_V1.5 levels at both the ICD and regions that could correspond to the t-tubular membrane structure (t-tubular region) was shown by immunocytochemical analysis (Figure 3B). This is confirmed by data from CaMKIIδ^−/− myocytes, showing that exposure to ISO did not result in CaMKII-specific Na channel phosphorylation (see Supplementary material online, Figure S2).

Figure 3.

Na channel expression and Na channel phosphorylation (at serine 571, p-Na_V1.5) in WT mouse myocytes. (A) Western blot original data (upper panel) and mean densitometric units for p-Na_V1.5 (mid panel) or Na_V1.5 (lower panel). (B and C) Original data for immunocytochemistry (upper panel) and analysed mean fluorescence intensity at intercalated disc (mid panel) and at a region possibly corresponding to t-tubular membranes (lower panel) for p-Na_V1.5 (B) and Na_V1.5 (C). While general expression of Na_V1.5 was unchanged, ISO increased CaMKII-dependent Na channel phosphorylation and this was abolished upon pre-incubation with TXL. Pre-incubation with TXL alone resulted in a reduced sarcolemmal Na channel expression consistent with reduced Na channel trafficking due to inhibition of microtubular activity. Surprisingly, ISO also reduced sarcolemmal Na channel expression. *P < 0.05 vs. WT, ^†P < 0.05 vs. WT + ISO (one-way ANOVA). Scale bar indicates 10 µm; calibration bar indicates colour-coded fluorescence intensity.

The global cellular Na channel expression was unchanged (vehicle vs. ISO, TXL, and TXL + ISO, respectively, 1 vs. 0.96 ± 0.13, 1.07 ± 0.12, 1.02 ± 0.14, respectively, n = 16 each, Figure 3A). Pre-incubation with TXL, however, abolished the increase in CaMKII-specific Na channel phosphorylation (TXL + ISO, 0.91 ± 0.10, n = 16, P < 0.05 vs. ISO, Figure 3A). In addition, treatment of cardiomyocytes with TXL abolished the increase in p-Na_V1.5 levels at both the ICD and regions possibly corresponding to t-tubular membranes (Figure 3B). This suggests that TXL may interfere with ISO-dependent Na channel regulation by inhibition of CaMKII activation.

Surprisingly, while global Na channel expression was unchanged, ISO reduced membrane expression of Na_V1.5 at ICD (P < 0.05) and also at the region possibly corresponding to t-tubular membranes (P < 0.05, Figure 3C). Thus, the fraction of phosphorylated Na channels, in comparison with total Na_V1.5 in the sarcolemma, may be substantially enhanced in the presence of ISO. Interestingly, pre-incubation with TXL alone also reduced membrane expression of Na_V1.5. Exposure to ISO did not further reduce membrane expression of Na_V1.5 in myocytes pre-incubated with TXL (Figure 3C).

3.3. TXL disrupts ISO-dependent CaMKII-specific Na channel regulation

To test if TXL interferes with ISO-dependent Na channel regulation, I_Na was measured in isolated ventricular myocytes. A hallmark of CaMKII-dependent Na channel regulation is enhanced late I_Na.⁴ In WT myocytes, ISO increased late I_Na integrated from 50 to 450 ms after onset of the depolarizing clamp pulse (Figure 4A and B). Pre-incubation with TXL, however, completely abolished the ISO-dependent increase in late I_Na (P < 0.05 vs. WT + ISO, Figure 4A and B). To confirm the specificity of the measured current, some experiments were conducted in the presence of the selective Na channel blocker tetrodotoxin (TTX; 1 µmol/L, Figure 4A and B). Interestingly, while PKA inhibition did not affect the ISO enhancement of late I_Na (PKI, 5 µmol/L, Figure 4A and B), direct stimulation of Epac with 8-CPT resulted in a significant increase in late I_Na comparable with the ISO effect (Figure 4A and B). This 8-CPT-induced enhancement of late I_Na was abolished upon pre-incubation with TXL (Figure 4A and B). The CaMKII dependency of late I_Na regulation by ISO was confirmed by experiments in myocytes lacking CaMKIIδ (CaMKIIδ^−/−). ISO did not increase late I_Na in CaMKIIδ^−/− myocytes (P < 0.05 vs. WT + ISO, Figure 4B).

Figure 4.

Late I_Na and I_Na steady-state inactivation in mouse myocytes. (A) Original traces for late I_Na elicited using the protocol shown in inset [135 mmol/L [Na]_o, contaminant [Ca]_o, free [Ca]_i 100 nmol/L buffered with 1 mmol/L of ethylene glycol tetraacetic acid (EGTA)]. Late I_Na was leak substracted and normalized to membrane capacitance. Late I_Na was estimated by integrating I_Na between 50 and 450 ms, and mean data are shown in (B). ISO enhanced late I_Na, but this was completely abolished upon pre-incubation with TXL or in the presence of I_Na inhibitor TTX (1 µmol/L). Inhibition of PKA (PKI) did not affect the ISO-induced late I_Na enhancement. Interestingly, direct stimulation of Epac with 8-CPT (5 µmol/L) also enhanced late I_Na and this was also blocked by pre-incubation with TXL. Of note, ISO could not increase late I_Na in myocytes lacking CaMKIIδ (CaMKIIδ^−/−). (C) Normalized original traces and (D) mean data for steady-state inactivation (measured with the protocol in inset) are shown (5 mmol/L [Na]_o, 0.4 mmol/L [Ca]_o, free [Ca]_i 100 nmol/L buffered with 1 mmol/L of EGTA). In WT, ISO left-shifted steady-state inactivation, and this was abolished upon pre-incubation with TXL. (D, right panel) Mean data for half-inactivation (V_1/2) revealed that the ISO effect on steady-state inactivation was less pronounced in CaMKIIδ^−/− myocytes. V_1/2 is derived from the curve by fitting it to standard Boltzmann equation: h_∞ = 1/{1 + exp[(V_1/2− V)/k_∞]}. *P < 0.05 vs. WT, ^†P < 0.05 vs. WT + ISO, ^‡P < 0.05 vs. CaMKIIδ^−/−, ^#P < 0.05 vs. CaMKIIδ^−/−+ ISO (one-way ANOVA). For all recordings, nifedipine 20 µmol/L, niflumic acid 30 µmol/L, and strophanthidine 4 µmol/L were added to the pipette solution.

Beside late I_Na, CaMKII has been shown to enhance I_Na intermediate inactivation (I_IM), resulting in a negative shift in steady-state inactivation and slowed recovery from inactivation.⁴ Figure 4C and D shows I_Na steady-state inactivation as a function of membrane potential (E_m). ISO enhanced steady-state inactivation, equivalent to a negative voltage shift of ∼9 mV in WT myocytes (P < 0.05; see Supplementary material online, Table S1; Figure 4D). Similar to late I_Na, TXL pre-incubation abolished the ISO effect on steady-state inactivation (Figure 4D, P < 0.05 vs. WT + ISO). Also, the ISO-induced negative shift in steady-state inactivation was blunted in CaMKIIδ^−/− myocytes, confirming that CaMKIIδ is involved (Figure 4D, P < 0.05 vs. WT + ISO; see Supplementary material online, Table S1).

ISO enhanced I_IM in WT myocytes (P < 0.05, Figure 5A and B and see Supplementary material online, Table S1). Interestingly, the rate constant for entry into I_IM was also increased (see Supplementary material online, Table S1). Both these effects were abolished with TXL (P < 0.05 vs. WT + ISO, Figure 5A and B, and see Supplementary material online, Table S1). TXL alone significantly reduces the fraction of channels that undergo I_IM (see Supplementary material online, Table S1). The CaMKII dependency of the ISO-induced entry into I_IM was confirmed in CaMKIIδ^−/− myocytes (P < 0.05 vs. WT + ISO, Figure 5B and see Supplementary material online, Table S1).

Figure 5.

I_Na intermediate inactivation (I_IM) and recovery from inactivation (5 mmol/L [Na]_o, 0.4 mmol/L [Ca]_o, free [Ca]_i 100 nmol/L buffered with 1 mmol/L of EGTA). Original traces (A and C) and mean data (B and D) for I_IM and recovery from inactivation, respectively. For I_IM, increasing conditioning pulse duration (P₁) reduced I_Na amplitude assessed with a second pulse (P₂) consistent with entry of Na channels into intermediate inactivation (protocol in the inset of A). Data were fit with a single exponential function (see Supplementary material online, Table S1); the mean data for plateau y₀ are shown in B (lower panel). For recovery from inactivation, increasing duration of the recovery interval between conditioning pulse (P₁, causing I_Na inactivation) and test pulse (P₂) results in an exponential increase in the amplitude of I_Na upon P₂ consistent with I_Na recovery (protocol in the inset of D). Data were fit with a single exponential function (see Supplementary material online, Table S1); the mean data for the rate constant of recovery k_rec are shown in D (lower panel). In WT, ISO enhanced I_IM and slowed recovery from inactivation, but this was completely abolished upon pre-incubation with TXL. In CaMKIIδ^−/− myocytes, external application of ISO neither enhanced I_IM nor slowed recovery from inactivation. *P < 0.05 vs. WT, ^†P < 0.05 vs. WT + ISO (one-way ANOVA). For all recordings, nifedipine 20 µmol/L, niflumic acid 30 µmol/L, and strophanthidine 4 µmol/L were added to the pipette solution.

Inactivation and recovery from inactivation are closely related. Recovery from inactivation was investigated using sustained depolarization (1000 ms), followed by recovery intervals of increasing durations and a subsequent test pulse (see pulse protocol in the inset of Figure 5D). In accordance with enhanced I_IM, ISO slowed recovery from inactivation (Figure 5C). The rate constant for recovery k_rec was 48.2 ± 3.19 vs. 32.35 ± 1.42 s⁻¹, n = 13 vs. 12 cells from seven mice, P < 0.05 (Figure 5C and D, and see Supplementary material online, Table S1). Similar to I_IM, pre-incubation with TXL could completely abolish the ISO effect on I_Na recovery (k_rec: 51.26 ± 3.09 s⁻¹, n = 9 cells from four mice, P < 0.05 vs. WT + ISO, Figure 5D and see Supplementary material online, Table S1). As expected, ISO could not slow I_Na recovery in CaMKIIδ^−/− myocytes confirming the CaMKIIδ dependency (P < 0.05 vs. WT + ISO, Figure 5D and see Supplementary material online, Table S1).

3.4. CaMKII-independent Na channel regulation

We have previously shown that Na conductance and voltage dependence of activation are not influenced by CaMKII.⁴ To test if ISO also regulates the number of functional channels in the sarcolemma, we investigated the current–voltage (I–V) relationship (Figure 6) in mouse myocytes. Interestingly, ISO increased maximal Na conductance (Figure 6A, B, and D). For vehicle vs. ISO, peak I_Na at −30 mV was −39.9 ± 5.1 vs. −89.3 ± 10.5 A/F (n = 13 vs. 6 cells from seven mice, P < 0.05, see Supplementary material online, Table S2, and Figure 6B and D). As expected, the ISO effect on current–voltage relationship was maintained in CaMKIIδ^−/− myocytes (P < 0.05 vs. CaMKIIδ^−/−), suggesting that other pathways independent from CaMKII are involved.^18–22 Interestingly, despite the fact that CaMKII is not relevant, pre-incubation with TXL also blunted the ISO effect on Na channel conductance, suggesting that TXL may also inhibit these pathways. For TXL + ISO, peak I_Na at −30 mV was −50.4 ± 7.6 A/F (P < 0.05 vs. WT + ISO; see Supplementary material online, Table S2). The fact that pre-incubation with TXL alone also reduced maximal Na conductance (P < 0.05 vs. WT; see Supplementary material online, Table S2) suggests that microtubules may be involved in the basal turnover of Na channels in the sarcolemma.⁶

Figure 6.

E_m dependence of I_Na activation in mouse myocytes (5 mmol/L [Na]_o, 0.4 mmol/L [Ca]_o, free [Ca]_i 100 nmol/L buffered with 1 mmol/L of EGTA). Original traces (A) and mean current–voltage relationship (I–V, B) in mouse myocytes measured using the protocol shown in inset. In WT, ISO increased I_Na current density. Enhanced tubulin polymerization (TXL) could blunt the ISO-induced increase in peak I_Na. (C) I_Na activation curve. Relative conductance was derived from maximal chord conductance and reversal potential (E_rev); for each I–V: peak I_Na/(E_m− E_rev). The resulting conductance was normalized to the maximal chord conductance. ISO also enhanced I_Na activation evident as leftward-shift of the I_Na activation curve. Upon pre-incubation with TXL, this effect was blunted but not completely abolished. (D) Mean data for peak I_Na at −30 mV. Interestingly, the ISO-induced increase in I_Na current density was also evident in CaMKIIδ^−/− myocytes, but was less pronounced. (E) Mean data of voltage for half-activation (V_1/2). V_1/2 is derived from the activation curve (see Supplementary material online, Table S2). Similar to I_Na current density, the ISO-dependent enhancement of I_Na activation was maintained in CaMKIIδ^−/−, but was less pronounced. *P < 0.05 vs. WT, ^†P < 0.05 vs. WT + ISO, ^‡P < 0.05 vs. CaMKIIδ^−/−, ^#P < 0.05 vs. CaMKIIδ^−/− + ISO (one-way ANOVA). For all recordings, nifedipine 20 µmol/L, niflumic acid 30 µmol/L, and strophanthidine 4 µmol/L were added to the pipette solution.

ISO also negatively shifted I_Na activation in WT myocytes ∼10 mV (P < 0.05; Figure 6C and E, and see Supplementary material online, Table S2), resulting in increased I_Na at negative membrane potentials. The slope factor k was unaltered (see Supplementary material online, Table S2). This effect was also independent from CaMKII. In myocytes lacking CaMKIIδ, the ISO-induced negative shift in I_Na activation was maintained (P < 0.05, Figure 2E and see Supplementary material online, Table S2).

Surprisingly, TXL did not affect this ISO-induced negative shift in I_Na activation. ISO still negatively shifted I_Na activation by ∼8 mV (P < 0.05 vs. WT + TXL, Figure 2C and E, and see Supplementary material online, Table S2). This suggests that ISO-dependent regulation of I_Na activation is independent from microtubular activity. Also, TXL alone had no influence on voltage for half-activation nor the slope factor (Figure 2E and see Supplementary material online, Table S2), in accordance with previously published data.⁶

We also investigated the effect of ISO on open-state inactivation. Fast I_Na decay (first 20 ms) was fit to a double exponential: f(x) = A_fast exp(−t/τ_fast) + A_slow exp(−t/τ_slow) + y₀. ISO enhanced the late phase of open-state inactivation, which is in accordance with previously published data.²³ This, however, was neither dependent on TXL pre-incubation nor on CaMKIIδ (see Supplementary material online, Table S2).

4. Discussion

The present study shows that correct function of microtubules is critically important for the β-adrenergic activation of CaMKII and its specific Na channel regulation. This is based on our findings that polymerization of microtubules with TXL inhibits the ISO-induced CaMKII auto-phosphorylation and membrane association. As a result, TXL inhibits ISO-dependent CaMKII-specific Na channel phosphorylation and regulation of late I_Na, I_Na steady-state inactivation, intermediate inactivation, and recovery from inactivation, effects known to be mediated by CaMKII.

The identified mechanisms in this study may be important for the understanding of cardiac side effects of TXL. Since increased CaMKII expression levels²⁴ coincide with enhanced β-adrenergic stimulation in heart failure (HF), this may also be relevant for arrhythmias and contractile dysfunction in HF.

4.1. Microtubules are required for ISO-induced CaMKII activation

CaMKII becomes activated by elevated intracellular Ca²⁵ or increased reactive oxygen species generation.^5,26 Also, β-adrenergic stimulation can result in CaMKII activation. Here, we show that exposure of ventricular myocytes to ISO results in an increase in CaMKII auto-phosphorylation, which is consistent with CaMKII activation. Interestingly, we show that the ISO-dependent increase in CaMKII auto-phosphorylation was abolished upon pre-treatment with TXL, which binds to the β-tubulin subunit and arrests microtubules in the polymerized state by rendering them resistant to depolymerization.²⁷ While western blot data showed that TXL completely abolished ISO-induced CaMKII auto-phosphorylation, the latter was only diminished in the immunocytochemistry analysis. Nevertheless, both showed that TXL significantly reduced p-CaMKII levels in the presence of ISO. This strongly suggests that microtubular activity may be required for the β-adrenergic activation of CaMKII. Microtubules, which represent the dynamic component of the cytoskeleton, are cylindrical polymers consisting of heterodimers of α- and β-tubulin. They undergo rapid and stochastic transitions from growing (polymerization) to shrinking (depolymerization) phases, resulting in a dynamic instability²⁸ which drives intracellular trafficking of macromolecules.

It was suggested that β₁-adrenoreceptor phosphorylation by GRK can result in the recruitment of β-arrestin and CaMKII to the receptor at the sarcolemma.¹⁶ As a consequence, Epac interacts with β-arrestin and CaMKII in close proximity to the β₁-adrenoreceptor complex, where cAMP is generated by adenylate cyclase.^16,29,30 Epac is stimulated by cAMP, which can lead to CaMKII activation.¹⁶ If β-arrestin2 membrane translocation is required for CaMKII activation and if this translocation is a process involving the activity of microtubules, then disruption of microtubular activity would also inhibit the β-arrestin2 movement upon β-adrenergic stimulation. Indeed, we show here that pre-treatment with TXL completely abolished the ISO-dependent β-arrestin2 membrane translocation. Moreover, silencing β-arrestin2 gene expression abolished the ISO-induced increase in CaMKII auto-phosphorylation. Thus, β-arrestin2 seems to be crucially involved in this activation process. On the other hand, TXL did not influence the amount of cAMP generated by stimulation of either the β-receptor (using ISO) or adenylate cyclase (forskolin), suggesting that these processes are independent from microtubules. This would also imply that PKA activity may not be affected by pre-treatment with TXL and PKA is not involved in CaMKII activation. Indeed, in the present study, application of the specific PKA inhibitor PKI did not abolish ISO-induced CaMKII auto-phosphorylation. Consistent with this, it was shown that the β-adrenergic stimulation of CaMKII activity was independent from PKA activation.^16,29–31 Moreover, using Epac2 knockout mice, it was recently shown that Epac2-dependent CaMKII activation, but not PKA, is required for the sarcoplasmic reticulum Ca leak that develops after β-adrenergic stimulation.³² Here, we could show that direct stimulation of Epac by 8-CPT resulted in an increase in CaMKII auto-phosphorylation, comparable with stimulation by ISO. Moreover, this effect was abolished by pre-incubation with TXL, suggesting that, besides β-arrestin, Epac may also need to translocate to the sarcolemma. The latter, however, remains to be shown.

4.2. Activated CaMKII mediates the ISO-dependent regulation of I_Na

What are the consequences of disturbed CaMKII activation for important downstream targets like the cardiac Na channel? We have previously shown that CaMKII regulates Na channel gating by direct phosphorylation of Na_V1.5.^4,5 This regulation results in enhanced I_IM, slowed I_Na recovery, negative-shifted steady-state inactivation, and also enhanced late I_Na. Recently, CaMKII-dependent phosphorylation sites on Na_V1.5 including serine 571,^9,10 serine 516, and threonine 594¹¹ have been identified and linked to altered Na channel gating.

Here, we show that exposure to ISO similarly enhanced I_IM, slowed I_Na recovery, negatively shifted steady-state inactivation, and enhanced late I_Na. This is in accordance with previously published data showing that β-adrenergic stimulation enhances steady-state inactivation,^21,33,34 enhances intermediate inactivation,²¹ and slows recovery from inactivation.²¹ Interestingly, we show here for the first time that the ISO effects on Na channel phosphorylation, I_Na inactivation, and late I_Na are abolished in CaMKIIδ knockout mice, highlighting the importance of CaMKIIδ for the β-adrenergic regulation of cardiac excitability. Moreover, we also show that all these CaMKII-dependent effects on I_Na gating are abolished by pre-incubation with TXL, suggesting that TXL treatment interferes with β-adrenergic regulation of I_Na by inhibiting CaMKII activation.

These data are supported by our immunocytochemistry experiments showing that ISO increased CaMKII expression and auto-phosphorylation in cellular regions expected to richly express Na_V1.5 such as at the ICD.³⁵ Interestingly, ISO also increased CaMKII expression at a region which corresponds to the z-line, a region very close to t-tubules.³⁶ Co-localization of CaMKII and Na_V1.5 at this z-line/t-tubular region has been shown before.⁴ Although not directly proven, it seems tempting to speculate that activated CaMKII may also play a role for the regulation of Na channels localized at the t-tubule. Consistent with the current measurements, pre-incubation with TXL also blocked β-adrenergic CaMKII-dependent Na channel phosphorylation.

Upstream of CaMKIIδ, Epac may be an important mediator of the β-adrenergic regulation of late I_Na. Here, we show that direct stimulation of Epac with 8-CPT enhanced late I_Na, comparable with the effect of ISO. Consistent with our data on CaMKII auto-phosphorylation, this 8-CPT-mediated increase of late I_Na was blocked by pre-incubation with TXL. This Epac-dependent CaMKII activation may require Ca/CaM binding to CaMKII. Although ethylene glycol tetraacetic acid (1 mmol/L) was added to the pipette solution to clamp diastolic Ca, this weak Ca buffer cannot prevent short, localized fluctuations in the free Ca concentration. We have previously shown that the strong Ca buffer Br₂-BAPTA, but not ETGA, prevented the H₂O₂-induced Ca-dependent activation of CaMKII.⁵ Thus, it is quite possible that the β-adrenergic CaMKII activation and late I_Na augmentation depend on Ca fluxes and binding of Ca/CaM to the regulatory subunit of CaMKII.

4.3. TXL disturbs regulation of Na channel function

We show here that microtubules are involved in the β-adrenergic activation of CaMKII and CaMKII regulates Na channel gating. Thus, it is not surprising that TXL disturbs the regulation of ISO-induced CaMKII-dependent Na channel. However, we and others have shown that maximal Na conductance and I_Na activation are not regulated by CaMKII.^4,9,10 Moreover, we show here that β-adrenergic stimulation increased maximal Na conductance and enhanced I_Na activation. Both effects were inhibited by pre-incubation with TXL. This indicates that, besides CaMKII, other important regulators of Na channel function may also depend on microtubules. In fact, the β-adrenergic enhancement of Na conductance and I_Na activation is in accordance with many previous publications.^21,23,33,34 Several mechanisms have been attributed to increased Na conductance upon β-adrenergic stimulation.^18,37 It was suggested that PKA-dependent Na channel phosphorylation may play a role.¹⁸ Thus, microtubules may also be important for PKA activation or PKA-dependent Na channel phosphorylation, although this remains to be shown. On the other hand, Na conductance also depends on ion channel trafficking.¹⁰ Therefore, pre-treatment with TXL should also impair ion channel trafficking, resulting in reduced membrane expression. Indeed, we show here that pre-incubation with TXL alone reduced maximal Na conductance and Na_V1.5 membrane expression, whereas the total Na_V1.5 expression was not changed. This is in accordance with previously published data.^6,8 In addition to controlling of the number of Na channels that contribute to I_Na, microtubules may also be involved in the direct regulation of Na channel gating. It was shown that TXL enhanced the spontaneous time-dependent shift in steady-state activation upon whole-cell recordings lasting for several minutes,³⁸ although the physiological relevance of such long-lasting recordings is unclear. Others did not observe changes in voltage dependence of activation or inactivation, but showed a slowed intermediate inactivation.⁶ We also show in the present paper that pre-treatment with TXL slowed intermediate inactivation (Figure 5) at baseline. However, this basal TXL effect was also abolished in myocytes lacking CaMKIIδ, which challenges the concept that microtubules directly regulate Na channel gating and suggests that CaMKII is not only important for the β-adrenergic regulation of I_Na but also important at baseline. In accordance, basal p-CaMKII levels were significantly lower upon pre-treatment with TXL in the present study.

In summary, we show that TXL disrupts the β-adrenergic stimulation of CaMKII activity, resulting in disturbed Na channel regulation. CaMKII and Na channels are required for the physiological β-adrenergic response of the heart.¹⁴ Moreover, Na channel dysregulation is an important contributor to arrhythmias and disturbed contractile function.^4,5 Thus, the present study may be of relevance for the treatment of cardiac side effects in patients receiving TXL treatment and also relevant for conditions of increased CaMKII expression and enhanced β-adrenergic stimulation like in HF.

Supplementary material

Supplementary material is available at Cardiovascular Research online.

Funding

This work was supported by Deutsche Forschungsgemeinschaft (DFG) through an International Research Training Group GRK 1816 to S.W., V.O.N., and L.S.M., through grants (TPA03 SFB 1002) to L.S.M., grants (NI 1301/1-1 and TPA01 SFB 1002) to V.O.N., and grants (BA 2258/2-1) to J.B. This work was also supported by Fondation Leducq ‘Redox and Nitrosative Regulation of Cardiac Remodeling to L.S.M. This work was also supported by Deutsches Zentrum für Herz-Kreislauf-Forschung—German Centre for Cardiovascular Research (DZHK) grants to L.S.M. and J.B. This work was supported by the European Commission (FP7-Health-2010 and MEDIA-261409) to J.B. This work was supported by the National Institute of Health grants (HL084583 and HL083422) to P.J.M. and grants (HL096805 and HL114893) to T.J.H. This work was supported by the American Heart Association Established Investigator Award to P.J.M. Finally, this work was supported by the Saving Tiny Hearts Society through a grant to P.J.M.