Epac‐induced ryanodine receptor type 2 activation inhibits sodium currents in atrial and ventricular murine cardiomyocytes

Summary Acute RyR2 activation by exchange protein directly activated by cAMP (Epac) reversibly perturbs myocyte Ca2+ homeostasis, slows myocardial action potential conduction, and exerts pro‐arrhythmic effects. Loose patch‐clamp studies, preserving in vivo extracellular and intracellular conditions, investigated Na+ current in intact cardiomyocytes in murine atrial and ventricular preparations following Epac activation. Depolarising steps to varying test voltages activated typical voltage‐dependent Na+ currents. Plots of peak current against depolarisation from resting potential gave pretreatment maximum atrial and ventricular currents of −20.23 ± 1.48 (17) and −29.8 ± 2.4 (10) pA/μm2 (mean ± SEM [n]). Challenge by 8‐CPT (1 μmol/L) reduced these currents to −11.21 ± 0.91 (12) (P < .004) and −19.3 ± 1.6 (11) pA/μm2 (P < .04) respectively. Currents following further addition of the RyR2 inhibitor dantrolene (10 μmol/L) (−19.91 ± 2.84 (13) and −26.6 ± 1.7 (17)), and dantrolene whether alone (−19.53 ± 1.97 (8) and −27.6 ± 1.9 (14)) or combined with 8‐CPT (−19.93 ± 2.59 (12) and −29.9 ± 2.5(11)), were indistinguishable from pretreatment values (all P >> .05). Assessment of the inactivation that followed by applying subsequent steps to a fixed voltage 100 mV positive to resting potential gave concordant results. Half‐maximal inactivation voltages and steepness factors, and time constants for Na+ current recovery from inactivation in double‐pulse experiments, were similar through all the pharmacological conditions. Intracellular sharp microelectrode membrane potential recordings in intact Langendorff‐perfused preparations demonstrated concordant variations in maximum rates of atrial and ventricular action potential upstroke, (dV/dt)max. We thus demonstrate an acute, reversible, Na+ channel inhibition offering a possible mechanism for previously reported pro‐arrhythmic slowing of AP propagation following modifications of Ca2+ homeostasis, complementing earlier findings from chronic alterations in Ca2+ homeostasis in genetically‐modified RyR2‐P2328S hearts.


| INTRODUCTION
Cardiac arrhythmias result from disruptions in the normal excitable activity propagating through successive structures in the heart. They arise from altered function in surface membrane ionic channels whose successive activation and inactivation underlies the production and propagation of cardiac action potentials (AP). Sustained arrhythmia may require both triggering events and arrhythmic substrate to maintain the resulting abnormal electrical activity. [1][2][3] Arrhythmic substrate arises either from slowed myocardial AP conduction or activation, exemplified by Brugada Syndrome, 4,5 or altered AP recovery, reflected in altered AP duration and/or refractoriness, exemplified by long QT syndrome. 6 Arrhythmic risk is also associated with dysregulated intracellular Ca 2+ homeostasis. This may arise from abnormal ryanodine receptor-2 (RyR2)-mediated sarcoplasmic reticular Ca 2+ release. The resulting elevations in diastolic [Ca 2+ ] and spontaneous propagated intracellular Ca 2+ waves increase electrogenic Na + -Ca 2+ exchanger activity thereby driving delayed after-depolarisations that may trigger premature ventricular beats. 7 Murine hearts carrying genetically altered RyR2-Ca 2+ release channel and SR Ca 2+ storage protein calsequestrin-2 have successfully modelled such triggering events. They recapitulate mutations associated with the chronic pro-arrhythmic condition of human catecholaminergic polymorphic ventricular tachycardia (CPVT). 8,9 Triggering events have also been reported following acute adrenergic activation produced by modifications of RyR2-mediated SR Ca 2+ release 10 and surface Ca 2+ channel properties in wild-type hearts expressing normal RyR2 and calsequestrin-2. 11,12 More recent studies selectively activated RyR2-Ca 2+ release channels using the phosphokinase A (PKA)-independent exchange protein directly activated by cAMP (Epac) pathway. [13][14][15] This increased Ca 2+ spark frequencies in adult rat cardiac myocytes 16 and amplitudes of Ca 2+ -dependent Ca 2+ release after isoproterenol treatment 17 in murine ventricular cardiomyocytes. They also increased the amplitudes and frequencies of spontaneous Ca 2+ release. 18 These changes correlated with increases in triggered activity and ventricular tachycardia (VT) in murine hearts. 18 Fewer studies have explored arrhythmic substrate under conditions of altered Ca 2+ homeostasis. Neither chronic modifications in Ca 2+ homeostasis in RyR2-P2328S models nor acute manipulations of Ca 2+ homeostasis in WT hearts altered AP recovery characteristics as reflected in AP durations (APD), refractory periods (ERP), or the relationships between these. 8,12,19 However, murine RyR2-P2328S CPVT cardiac models showed reduced atrial 20 and ventricular conduction velocities in common with Na v 1.5-haploinsufficient Scn5a+/− hearts modelling the Brugada Syndrome. 21 Pharmacological inhibition of RyR2-mediated Ca 2+ release with flecainide partly rescued these effects. [22][23][24][25] Furthermore, selective, acute RyR2 activation through the Epac pathway produced parallel pro-arrhythmic effects. 18 It correspondingly produced decreases in AP conduction velocities that were partially reversed by the RyR2 antagonist dantrolene, with an absence of alterations in AP recovery characteristics. 19 The mechanism for the conduction velocity changes in RyR2-P2328S hearts was identified as the direct action of intracellular Ca 2+ on Na v 1.5 function 20,21,26 and/or Na v 1.5 membrane expression. 27 However, the mechanisms by which acute manipulations of intracellular Ca 2+ homeostasis, particularly Epac activation, alter AP conduction have not been investigated. The present experiments assessed Nav1.5 activation, inactivation, and recovery from inactivation following acute rather than chronic manipulations of Ca 2+ homeostasis, and in WT rather than genetically-modified hearts.
They employed the loose patch technique for voltage-clamping of Na + current. This apposes an electrode containing extracellular solution against an intact cell surface membrane without accessing intracellular space. Studies were thus performed in cardiomyocytes in intact murine atrial and ventricular preparations without perturbing extracellular [Na + ] and intracellular Ca 2+ homeostasis 21,28,29 as opposed to following cardiomyocyte isolation necessitated by conventional whole-cell patch clamp techniques. 30,31 Recent cardiomyocyte studies involving reversible manipulations of loose patch pipette [Na + ] had identified early inward currents in response to step depolarisations with Na + currents responsible for AP conduction and the maximum upstroke rate, (dV/dt) max , of the cardiac action potential. 21 The corresponding changes in such (dV/dt) max were accordingly determined by independent experiments performing intracellular sharp electrode recordings of membrane potential in intact atrial and ventricular preparations. These explored the extent to which changes in observed (dV/dt) max , paralleled corresponding changes in Na + currents in the loose-patch experiments with similar pharmacological manipulations. Such correlations would be consistent with previously reported relationships between (dV/dt) max and peak Na + currents (I Na ). 32 We thus studied electrophysiological effects of pharmacological manipulations of Ca 2+ homeostasis through Epac activation on Na + currents in a near physiological environment for the first time. The Epac activator (8-pCPT-2′-O-Me-cAMP: 8-(4-chlorophe nylthio)-2′-O-methyladenosine-3′,5′-cyclic monophosphate) 13 at approximately 1 μmol/L offered 300-fold preferential selectivity for Epac over PKA action. 13 Figure 1) and ventricular ( Figure 2) preparations subject to a combined pulse procedure (Panel Aa) designed to explore both activation and inactivation properties of Na + currents. Cells were first held at the resting membrane potential (RMP) for 5 ms from the beginning of the recording period. A prepulse of duration 5 ms to V 0 = (RMP−40) mV was then applied to remove any residual Na + current inactivation and standardise the initial activation state of Na + channels within the patch. Na + current activation properties were then investigated using the initial depolarising test voltage steps made to a voltage varied with the 13 successive sweeps between V 1 = RMP to (RMP + 120) mV in +10 mV increments. These currents, following correction for residual leakage by a P/4 protocol, provided a family of records reflecting the voltage dependence of Na + channel activation (Panels Ab and Ba).
Inward currents are represented as downward, negative deflections.
Records typically began with a consistent small upward deflection in response to the −40 mV prepulse. The subsequent voltage steps to level V 1 yielded a family of inward currents initially increasing with time to a peak value that increased as V 1 was made more positive. This was followed by a decay reflecting channel inactivation whose extent and kinetics was similarly determined by the voltage V 1 .
The subsequent voltage steps in the pulse protocol were used to assess properties of the Na + current inactivation resulting from the voltage step to V 1 . Thus, 5 ms following imposition of the latter step, an additional step was applied to a fixed voltage of V 2 = (RMP + 100) mV that would result in a peak Na + current reflecting the extent of the preceding channel inactivation (Panels Ab and Ca). The step to voltage V 2 thus elicited a second family of currents which decreased in amplitude with increasing V 1 . Both sets of responses showed nonlinear F I G U R E 1 Ionic currents from an isolated atrial preparation studied under loose-patch clamp. (Aa) Pulse protocol designed to investigate Na + channel activation and inactivation. (Ab) Typical recordings obtained using with a pipette of 28 μm tip diameter. Inward currents in nA are represented as negative. All currents shown in subsequent panels converted to current density in units of (pA/μm 2 ) to account for pipette diameters, typically 28-32 μm. (B, C) Na + currents observed upon (B) activation by depolarisation to a level V 1 , before and following pharmacological challenge and (C) in response to the voltage step from levels V 1 to the final level V 2 . The latter permitted assessment of Na + channel inactivation produced by the voltage step to level V 1 . Recordings made

| Currents reflecting Na + channel inactivation
Records demonstrating typical families of Na + current inactivation from atrial and ventricular preparations are shown in Figures 1C and   2C. Each family of records was obtained from an individual patch in response to the steps between the voltages V 1 and V 2 (panel Aa).
These illustrate the extent of Na + channel inactivation resulting from the step to voltage V 1 . Only the fraction of channels spared such inactivation would then be activated by the step to the fixed voltage V 2 .
The Na + current amplitudes would therefore permit quantification of the voltage dependence of Na + channel inactivation at the voltage V 1 .
Records were obtained before 8-CPT challenge (pretreatment) (Ca), following 8-CPT challenge (Cb), and following challenge by 8-CPT combined with dantrolene (Cc). They were compared with results of transferring pretreatment preparations directly to solutions containing dantrolene alone (Cd) or dantrolene combined with 8-CPT (Ce).
The resulting Na + currents showed typical activation and inactivation time courses, but these decreased in amplitude the more depolarised the voltage V 1 (Ca). Again, 8-CPT markedly decreased Na + current amplitudes at all voltages tested (Cb), and further inclusion of dantrolene restored the Na + currents (Cc). Na + currents following challenge by dantrolene alone (Cd) or dantrolene in combination with 8-CPT (Ce) were similar in amplitude and waveform as pretreatment records.
Thus, the pulse procedures assessing current inactivation similarly demonstrated that 8-CPT reduced Na + current amplitude and this action was rescued by further addition of dantrolene. V 1 towards the Na + current reversal potential. Table 1  Ventricular acƟvaƟon and inacƟvaƟon

| Voltage dependences of ventricular Na + current activation and inactivation
Current-voltage curves describing voltage dependences of ventricular Na + current activation ( Figure 4A) similarly corroborate results in Table 1

| Time course of Na + channel recovery from inactivation
The subsequent protocols investigated timecourse of Na + channel recovery from inactivation following atrial or ventricular repolarisation ( Figures 5A and 6A) under the same set of pharmacological conditions as described above. The voltage was held at the RMP for 1 ms from the beginning of the recording period before imposition of a hyperpolarising prepulse to voltage V 0 = (RMP−40) mV for 4 ms. This established a consistent baseline level of Na + current inactivation as in the previous protocol (Figures 5 and 6; panel Aa).
A P1 conditioning step between V 0 and V 1 = (RMP + 80) mV of 5 ms duration then elicited activation of an initial Na + current followed by its inactivation decay (Figures 5 and 6, panel Ab, B). Depolarising P2 steps of 5 ms duration to voltage V 3 = (RMP + 80) mV were then imposed following different time intervals, T, which varied between 5 to 65 ms in 5 ms increments through the 12 successive sweeps making up the protocol. These P2 steps could be used to assess the recovery with time of the peak Na + current from inactivation (Ab). The peak currents are normalized to their values obtained in the P1 step ( Figures 5C and 6C).

| Microelectrode recordings of AP characteristics
Reductions of Na + current of the kind described above have been as-  Figures 7C and 8C). The similar findings in both atrial and ventricular preparations thus paralleled corresponding observations in loose patch clamp Na + currents. In contrast, resting membrane F I G U R E 7 Analysis of atrial action potential waveforms. Intracellular recording of atrial action potential waveforms during regular 6 Hz pacing (A) and their first derivatives reflecting (dV/dt) (B) in (i) untreated cardiomyocytes, (ii) with 8-CPT challenge followed by (iii) further addition of dantrolene, as well as (iv) dantrolene alone. These recordings yielded values of (dV/dt) max (C) that could be compared with remaining features of the resting potential (RMP) (D) and APD 90 (E). Single, double and triple symbols denote significantly different pairs of values, to a significance value of P < .05, .01 and .001 respectively potentials showed no significant differences between groups (atria: F = 1.24, P = .30; ventricles: F = 1.18, P = .32), falling close to −75 mV in both atrial and ventricular cardiomyocytes (Figures 7D and 8D).

| DISCUSSION
The experiments here investigated for acute effects of Epac activation, known to acutely perturb cellular Ca 2+ homeostasis, 18  The experiments followed directly from findings that acute Epac (exchange protein directly activated by cAMP) activation: (i) produced phosphokinase A (PKA)-independent activation of RyR2-mediated Ca 2+ release, [13][14][15] increasing Ca 2+ spark frequencies in adult rat cardiomyocytes, 16 Ca 2+ -dependent Ca 2+ release amplitudes after isoproterenol treatment, 17 and amplitudes and frequencies of spontaneous Ca 2+ release in mouse ventricular cardiomyocytes. 18 It correspondingly, (ii) increased incidences of both triggered activity and ventricular tachycardia (VT). 18 However, it also (iii) associated such pro-arrhythmic effects with decreased conduction velocities. 19 Furthermore, (iv) these electrophysiological characteristics were rescued by the RyR2-Ca 2+ release channel blocker dantrolene, despite (v) all these changes being associated with unaltered recovery characteristics as reflected in the cardiac action potential (AP) durations and ventricular effective refractory periods. 18,19 Such findings suggest that acutely altered Ca 2+ homeostasis might result in arrhythmic substrate arising from delayed conduction, as previously suggested for the pro-arrhythmic Brugada syndrome [1][2][3][4][5] in contrast to the recovery abnormalities associated with the LQT syndromes. 6,45 The pro-arrhythmic effects of acute abnormalities in cardiomyocyte Ca 2+ homeostasis provoked by adrenergic stimulation, caffeine-mediated RyR2 stimulation 10 or modified extracellular Ca 2+ entry, 11,12 have hitherto been primarily associated with arrhythmic triggering by delayed afterdepolarisation effects arising from the consequently altered Na + -Ca 2+ exchanger activity.
Early Na + current is a primary determinant of cardiac AP conduction velocity. The present experiments are the first time Na + current measurements have been made during Epac activation using an experimental design that permits comparisons with those previous studies. 20,21,26 Thus, in addition to employing similar methods of Na + current measurement: (i) They also (cf. 18 ) investigated effects of Epac activation using the agonist 8-CPT 13 at ~1 μmol/L, a concentration range known to provide a 300-fold preferential selectivity for Epac over PKA pathways; 13,33,34 8-CPT only inhibits phosphodiesterase isoforms at considerably higher concentrations. 35  sparks in cardiomyocyte models for CPVT 36 and cardiac failure. 37,38 (iii) Effects of dantrolene were further controlled for in experiments involving direct exposure to dantrolene, both by itself and in combination with 8-CPT (cf. 19 ), and (iv) The experiments similarly measured the time courses and steady-state voltage-dependences of Na + current activation and inactivation (cf. 21 ), in both atrial and ventricular preparations from WT murine hearts.
The variation of the test step to voltage V 1 allowed measurement of the peak amplitudes of Na + current and yielded current-voltage activation relationships. These peaks were followed by decays to a steady-state inactivation level whose dependence upon the voltage The present findings thus implicate conduction velocity secondary to compromised Na + current as a source for arrhythmic substrate under conditions of acutely perturbed cytosolic Ca 2+ homeostasis, reconstructing the altered conduction previously reported in Na v 1.5 haplo-insufficient, Scn5a +/− murine models for Brugada Syndrome. [1][2][3][4][5] They also complement previous findings in murine hearts chronically modelling catecholaminergic polymorphic ventricular tachycardia (CPVT). Cardiomyocytes in the latter systems similarly showed diastolic episodes, or propagating waves, of RyR2-mediated Ca 2+ release, as well as afterdepolarisation and triggering phenomena. 8,9 RyR2-P2328S hearts additionally showed parallel reductions in atrial 21 and ventricular action potential conduction velocities, the latter particularly following catcholaminergic challenge. 20 These changes accompanied chronically downregulated Na v 1.5 expression. 21,27 Furthermore, WT rat cardiomyocytes increased their expression of functionally active surface membrane Na v 1.5, Na v 1.5 mRNA and total Na v 1.5 protein following verapamil F I G U R E 8 Analysis of ventricular action potential waveforms. Intracellular recording of ventricular action potential waveforms during regular 6 Hz pacing (A) and their first derivatives reflecting (dV/dt) (B) in (i) untreated cardiomyocytes, (ii) with 8-CPT challenge followed by (iii) further addition of dantrolene, as well as (iv) dantrolene alone. These recordings yielded values of (dV/dt) max (C) that could be compared with remaining features of the resting potential (RMP) (D) and APD 90 (E). Single, double and triple symbols denote significantly different pairs of values, to a significance value of P < .05, .01 and .001 respectively challenge and decreased their surface membrane Na v 1.5 expression following calcimycin challenge. 47,48 In parallel with the present findings, RyR2-P2328S cardiomyocytes also demonstrated acutely reduced Na v 1.5 function. 20,21,26

| Solutions
Krebs-Henseleit (KH) solution was prepared (mmol/L: NaCl, 119;   Loose patch-clamp experiments were performed as previously described. 21,24,27 The pipette was lowered onto the membrane surface and gentle suction applied to allow seal formation around the patch of membrane. Voltage-clamp steps were delivered under computer control relative to the resting membrane potential. The loose patch clamp method applies voltage steps to the extracellular surface of membrane within the seal. Positive and negative voltage steps applied through the pipette respectively hyperpolarise and depolarise the membrane potential relative to the initial cell resting membrane potential (RMP).

| Loose patch clamp recording
In the text membrane potentials are therefore stated in terms of their displacement from the RMP in the same convention adopted by earlier studies using this technique. 29,57 The loose patch clamp configuration results in larger leakage currents than the conventional patch clamp due to the comparatively low seal resistance. A custom-built amplifier was used to compensate for the majority of the leakage current, series resistance errors and the displacement current through the pipette capacitance. 28 Residual leakage and capacitative currents were corrected for using reference records from subsequent P/4 control protocols applying steps whose amplitudes were scaled down by a factor of four and of the opposite sign relative to the test steps, as fully described previously. 29,57 Experiments were first performed under pretreatment condi- Patch-clamp studies took place under standardised electrophysiological conditions following similar intervals after each solution change.
Each patch was subject only to a single application of the pulse protocols. These made differences between results attributable to prolonged changes in the patch such as bleb formation unlikely. 58

| Whole heart intracellular microelectrode recordings
Microelectrode studies were performed using a modified horizontal were obtained from the first time differential of the intracellular AP waveform. These were measured along with cardiomyocyte resting potentials, RMP, with AP amplitude measured from the RMP to the peak AP voltage. Action potential duration was measured as the time from the AP peak to 90% repolarisation to baseline.

| Analysis of results
The currents obtained from the loose patch procedure are dimensioned in units of current (nA). Subsequent analysis converted the units of current (nA) to current densities (pA/μm 2 ) using the formula: Current-voltage curves describing Na + current activation and inactivation and current-time relationships plotting the recovery timecourse of Na + current from inactivation were obtained from the corresponding measurements of peak Na + currents (means ± standard error of

| Statistical analysis of results
Owing to the limited recording time permitted following establishment of each loose patch, 58  nist.gov/div898/handbook/).