RYR2 Channel Inhibition Is the Principal Mechanism of Flecainide Action in CPVT

To isolate the specific effects of flecainide on RyR2-mediated Ca release, we synthesized 2 flecainide analogues, a tetraalkylammonium salt derivative of flecainide, which is a quaternary ammonium cation also known as QX-FL,²³ and a single-methylated flecainide derivative, N-methyl flecainide (NM-FL), shown in Figure 1A. We hypothesized that the secondary amine on the piperidine ring in flecainide is necessary for its activity in RyR2 channels but not sodium channels. This idea stems from work in which QX-FL has been shown to not block RyR2 in single-channel studies performed at +40 mV membrane potential difference, where an effect with flecainide is observed.¹⁶ While both QX-FL and NM-FL have similar benzamide-derived structures compared with flecainide, there are key differences in the piperidine moiety. QX-FL, synthesized using a modified literature procedure, is doubly methylated at the piperidine nitrogen, resulting in a quaternary ammonium salt.²⁴ This modification renders QX-FL permanently charged and removes its ability to participate in hydrogen bonding at that nitrogen center. To test the hypothesis that the secondary amine is critical for flecainide RyR2 block as well as make a substrate suitable for in vivo administration, we synthesized NM-FL using an adapted Eschweiler-Clarke reaction to install a single methyl at the piperidine nitrogen.²⁴ This substrate maintains similar physicochemical properties to flecainide yet also has the key N-H donor removed. Flecainide has a pKa of 9.3, resulting in 99% ionization at physiological pH (7.4). Although an exact pKa value for NM-FL has not been experimentally determined, the effect of N-methylation of amines on pKa has been extensively studied and documented. Hall shows that the pKa of 2-methylpiperidine and that of 1,2-dimethyl piperidine are <1 pKa unit apart.^25,26 This can be extended to the case of flecainide and NM-FL. Hence, like flecainide, NM-FL is predicted to be predominately ionized at physiological pH. Sodium channel or RyR2 block by NM-FL has not been characterized previously.

To determine the structure-activity relationship of the N-methylated flecainide analogues, we recorded single-channel RyR2 openings and measured open probability (RyR2 P_o) from sheep RyR2 incorporated into lipid bilayers (representative traces shown in Figure 1B). At end-diastolic cytoplasmic Ca concentrations (0.1 µmol/L), RyR2 open probability had a skewed distribution with a median of 0.042 and 25 to 75 percentiles at 0.009 and 0.18 (n=60). Addition of flecainide to the cytoplasmic bath at a membrane potential of +40 mV inhibited RyR2 with an IC₅₀ of 15.9 µmol/L (CI_95%=[10.9–21.2]). Phosphorylation of single RyR2 channels had no significant effect on the potency of flecainide block (Figure I in the Data Supplement). N-Methylation of flecainide substantially reduced its inhibitory potency on RyR2 channels. The IC₅₀ of the single-methylated NM-FL was 99 µmol/L (CI_95%=[68.7–144], P=3.71×10⁻⁹ versus flecainide). Double methylation further reduced the inhibitory potency, with an IC₅₀ of 182 µmol/L for QX-FL (CI_95%=[128–259], P=2.55×10⁻⁵ versus flecainide, P=4.86×10⁻² versus NM-FL for QX-FL, Figure 1B and 1C).

We next tested flecainide and its analogues on RyR2 at a Ca concentration of 100 µmol/L, which is the cytoplasmic Ca concentration in the dyadic cleft during systole.²⁷ Under those conditions, RyR2 was near maximally activated (median P_o=0.68, 25–75 percentiles at 0.21 and 0.86, n=32). As previously reported,⁹ high cytosolic Ca reduces the potency of RyR2 block by flecainide (IC₅₀=88 µmol/L (CI_95%=[78.5–99.5]), Figure II in the Data Supplement), which may explain why flecainide does not inhibit systolic SR Ca release during normal excitation-contraction coupling.²⁸ N-Methylation significantly reduced the inhibitory potency of flecainide also at elevated Ca concentrations, with an IC₅₀ of 357 µmol/L (CI_95%=[254–502]) for the single-methylated NM-FL (P=2.34×10⁻¹⁴ versus flecainide), and 773 µmol/L (CI_95%=[626–952]) for the double-methylated QX-FL (P=5.19×10⁻²⁴ versus flecainide, Figure II in the Data Supplement).

To test whether reduced activity on single RyR2 channels translates into altered drug activity when the RyR2 channel is incorporated in its biological SR membrane, we recorded spontaneous Ca waves—which are generated by RyR2-mediated SR Ca release—in Casq2−/− cardiomyocytes subjected to saponin to selectively permeabilize the sarcolemma and enable equivalent access of flecainide or its analogues to RyR2. Importantly, treatment with saponin produces large pores in the sarcolemma, effectively clamping the sarcolemmal membrane potential at 0 mV and eliminating any effects of sodium channel block or sarcolemmal polarity on SR Ca release, while leaving the SR membrane intact.²⁹ Representative line scans of Ca waves are shown in Figure 2A. We observed a concentration-dependent decrease in Ca wave frequency (Figure 2B) following treatment with flecainide (IC₅₀=7.0 µmol/L, CI_95%=[4.69–10.3]). QX-FL and NM-FL did not inhibit Ca waves at the highest concentration tested, consistent with their reduced potency in the RyR2 single-channel assay (Figure 1C). Taken together, these results indicate that N-methylation of flecainide reduces its ability to inhibit single RyR2 channels and RyR2-mediated Ca release.

Cardiac sodium channel block by QX-FL has been extensively characterized.²³ QX-FL is permanently charged at any pH, rendering it membrane impermeable. This requires it to be introduced intracellularly to exert any effect. Flecainide, while predominantly ionized at physiological pH, is in equilibrium with a small fraction that is neutral and can diffuse across the cell membrane. Once inside the cell, flecainide again reaches equilibrium with most of the compound being charged. Only the ionized form of flecainide is responsible for its characteristic sodium channel use-dependent block (UDB; ie, the channel inhibition increases with higher channel activity).²³ UDB of sodium channels is an important mechanism of action for class I antiarrhythmic drugs such as flecainide.⁸

Since NM-FL had not previously been characterized biologically, we compared its effect on the cardiac sodium channel (Na_V1.5) to that of flecainide. Experiments were carried out using whole cell patch clamp of human embryonic kidney (HEK293T) cells stably expressing human Na_V1.5.³⁰ Flecainide or NM-FL were applied in the extracellular perfusate and sodium currents were measured using standard protocols. Both flecainide and NM-FL exhibited strong UDB, with a similar concentration dependence (Figure 3A). With an IC₅₀=7.7 µmol/L (CI_95%=[6.31–9.49]) for UDB, NM-FL was modestly more potent than flecainide (IC₅₀=10.6 µmol/L, CI_95%=[8.11–13.9], P=0.07), which is consistent with a slight difference in pKa. Rate of recovery from UDB (Figure III in the Data Supplement) and voltage dependence of UDB (Figure IV in the Data Supplement) were not statistically different between flecainide and NM-FL. To assess the tonic block of sodium channels, we measured the effect of flecainide and NM-FL on the current-voltage relationship of I_Na activation (Figure 3B) and steady-state inactivation (Figure V in the Data Supplement). At 6 µmol/L, there were no statistically significant differences between flecainide and NM-FL (Figure 3B peak current t test, P=0.93). Taken together, these results demonstrate that N-methylation of flecainide’s piperidine moiety does not significantly alter its activity on cardiac sodium channels.

We next examined spontaneous SR Ca release inhibition in intact Casq2−/− cardiomyocytes using whole-cell patch clamp. Because QX-FL is permanently charged and does not cross cell membranes, flecainide, QX-FL, or NM-FL (6 µmol/L) were applied in both the extracellular (perfused solution) and intracellular solutions (dialyzed via patch pipette) for 5 minutes. Cyclic AMP (200 µmol/L) was included in the patch pipette to maximally activate PKA (protein kinase A) and mimic the catecholamine-mediated phosphorylation state that occurs during adrenergic response. Spontaneous Ca waves were recorded during a 48-second test period following termination of a short conditioning train of membrane-depolarizing steps (Figure 4A, Figure VI in the Data Supplement). To abolish any contribution of sodium channel block by flecainide and its analogues and isolate potential effects from RyR2 inhibition, cells were pretreated with 30 µmol/L tetrodotoxin for 15 minutes, which blocked I_Na completely (Figure VII in the Data Supplement). Tetrodotoxin pretreatment (ie, complete sodium channel block) significantly reduced Ca wave frequency by ≈45% compared with vehicle treatment (Figure 4A). In the presence of tetrodotoxin, flecainide further reduced Ca wave frequency (by ≈90%, Figure 4A), which was significantly more than that of tetrodotoxin alone (Figure 4A), indicating an additive effect. QX-FL and NM-FL, on the other hand, did not significantly reduce the frequency of spontaneous Ca waves beyond that observed with tetrodotoxin alone (Figure 4A). Flecainide, QX-FL, and NM-FL had no effect on the SR Ca content, as estimated by a rapid 10 mmol/L caffeine spritz at the end of every recording.

We next compared flecainide and NM-FL in the absence of tetrodotoxin (Figure 4B). The magnitude of Ca wave inhibition by flecainide was significantly greater than that of NM-FL, further supporting the additive effect of RyR2 block on Ca wave inhibition (Figure 4B). Taken together, these results establish that flecainide inhibits arrhythmogenic RyR2-mediated SR Ca release in intact cardiomyocytes independent of its sodium channel blocking properties.

We next sought to determine if and to what extent the RyR2 block by flecainide influences its antiarrhythmic efficacy in vivo. Given its activity by extracellular application (Figure 3), equivalent block of cardiac sodium channels (Figure 3), and lack of direct effect on RyR2-mediated Ca waves (Figures 2 and 4) compared with flecainide, we used NM-FL as a tool to answer this question. Using the well-established Casq2−/− CPVT mouse model, Casq2−/− mice were randomized to a triple crossover study design and pretreated via intraperitoneal injection with vehicle, 15 mg/kg flecainide (Flec.), or 15 mg/kg NM-FL, followed by catecholamine challenge with isoproterenol (3 mg/kg intraperitoneal) 90 minutes later. Informed by our previous studies of flecainide pharmacokinetics in mice,⁹ the dose was chosen to produce flecainide plasma concentrations in mice that are within the therapeutic range (0.4–1 mg/L) used clinically to prevent ventricular arrhythmias in patients with CPVT.¹⁰ Figure 5A shows representative time courses of the heart rate in Casq2−/− mice before and after catecholamine challenge (time=0). Injection of isoproterenol causes development of ventricular arrhythmia (inset), observed as a fluctuating heart rate because of variable R-R intervals with premature ventricular complexes and ventricular tachycardia. In vivo, use-dependent sodium channel block manifests itself as a widening of the QRS complex because of the reduction in depolarization rate and hence slowed conduction velocity.³¹ Figure 5B shows representative QRS waveforms from a Casq2−/− mouse. Both flecainide and NM-FL prolonged the QRS complex to a comparable degree (Figure 5C), indicating that the doses were bioequivalent and both drugs exhibit equivalent potency of use-dependent sodium channel block in vivo. Arrhythmia burden was classified on an ordinal scale from 0 to 4. Only flecainide, but not NM-FL, which lacks RyR2 blocking properties, significantly decreased arrhythmia burden (Figure 5D and 5E, Figure VIIIA in the Data Supplement) and premature ventricular complexes (Figure VIIIB in the Data Supplement), despite both drugs equivalently inhibiting the sodium channel. There were no differences in heart rates between treatment groups (Figure VIIIC through VIIIE in the Data Supplement). Taken together, these results establish the importance of RyR2 inhibition by flecainide for its antiarrhythmic activity in vivo.

To answer this question, one should consider the mechanisms underlying the generation of catecholamine-induced ventricular arrhythmias in CPVT.³⁵ At the level of the cardiomyocyte, hyperactive RyR2 channels cause spontaneous RyR2 openings during diastole that coalesce into a propagated Ca wave, which in turn activates the electrogenic sodium calcium exchanger, resulting in a cell membrane depolarization. Membrane depolarizations of sufficient amplitude will activate voltage-gated sodium channels, which in turn will trigger a full cardiac action potential. The therapeutic benefit of RyR2 block by flecainide stems from its drastic reduction or complete prevention of arrhythmogenic calcium waves (Figure 4 and studies by Watanabe et al,⁹ Galimberti and Knollmann,¹³ and Hilliard et al²⁸), which is the primary cellular event responsible for the generation of the spontaneous action potential in CPVT.³⁵ The therapeutic benefit of sodium channel block stems from 2 following independent actions: (1) the reduction of cellular sodium and hence Ca loading, reducing the likelihood of spontaneous SR Ca release^16,18 and (2) increasing the cell membrane potential threshold that can trigger a spontaneous action potential.¹⁵ Which of these mechanisms predominates depends largely on the experimental conditions, which has contributed to different investigators coming to various conclusions about the relative importance of each mechanism. For example, rapid pacing protocols favor sodium channel block as the primary mechanism because of the strong use-dependence of flecainide, leading to an overestimation of the contribution of sodium channel block.³⁶ To overcome this limitation, here we used a train of voltage-clamp pulses to stimulate cardiomyocytes and elicit arrhythmogenic Ca waves. We find that a complete block of sodium channels with tetrodotoxin reduced the rate of spontaneous Ca waves by ≈45% (Figure 4). A similar reduction was achieved by sodium channel block with NM-FL, which blocks sodium channels with similar potency as flecainide (Figure 4B) but does not inhibit RyR2 at the concentration tested. On the other hand, flecainide reduced spontaneous Ca waves by over 90% (Figure 4), indicating that the additive effect of RyR2 block is substantial. In contrast, Bannister et al¹⁶ reported that sodium channel block by flecainide is solely responsible for its effect on Ca waves, based on a comparison of intracellular application of flecainide and QX-FL, which does not block RyR2 (Figure 1). However, QX-FL is a much more potent sodium channel blocker than flecainide when selectively applied to the intracellular side of the membrane because flecainide rapidly diffuses out of the cell whereas quaternary ammonium QX-FL does not.²³ In our experiments, QX and flecainide were applied to both the intracellular and extracellular side of the sarcolemma at equal concentrations (Figure 4). Taken together, our results indicate that at least half of the therapeutic benefit at the level of the cardiomyocyte is derived from flecainide’s inhibition of RyR2-mediated Ca release, with a contribution by sodium channel inhibition that depends on the experimental conditions.

The collective evidence from experimental and clinical studies suggests that while RyR2 inhibition is beneficial, treatment with pure sodium channel blockers does not provide benefit and may even harm patients with CPVT. In experimental studies, selective RyR2 block with ent-verticilide prevented catecholamine-induced ventricular arrhythmia in a murine CPVT model.³⁷ In clinical studies, based on case-control studies^9,11,12 and a randomized clinical trial,⁹ the dual RyR2 and sodium channel blocker flecainide prevented ventricular arrythmias in patients with CPVT when used as a monotherapy or in combination with β-blockers, and since 2013 is considered the standard of care in the treatment of patients with CPVT.³⁸ Propafenone, another class Ic agent that inhibits RyR2 channels similar to flecainide, is also effective in patients with CPVT.³⁹ Clinical experience with class Ia and Ib sodium channel blockers which lack activity on RyR2¹³ is limited, but in a case series of 29 patients with CPVT, class Ia and Ib agents given alone or in combination with a β blocker were not protective but rather significantly increased the risk of sudden death.⁴⁰ Two of the 3 cases given mexiletine and one case given disopyramide died suddenly. Our data testing the antiarrhythmic efficacy of class I agents in CPVT mouse models also indicate that sodium channel block alone is not effective in CPVT. Compared with flecainide or (R)-propafenone, sodium channel blockers that have less ((S)-propafenone) or no RyR2 blocking properties (lidocaine, procainamide) did not significantly reduce the incidence of exercise or catecholamine-induced ventricular tachycardia in CPVT mice.^9,39 The current study demonstrates that the N-methyl flecainide analogue, which lacks RyR2 block at clinically relevant concentrations (Figures 1 and 2), did not reduce the severity of ventricular tachyarrhythmias in CPVT mice (Figure 5), although the degree of sodium channel block as evidenced by the QRS widening on the ECG was comparable to that of flecainide (Figure 5). Why do sodium channel blockers reduce triggered activity in isolated CPVT cardiomyocytes^15–18 but do not protect against ventricular tachycardia in vivo? The answer is not clear but may relate to the well-established proarrhythmic liability of sodium channel blockers at the tissue level,⁶ which could explain why treatment with pure sodium channel blockers increased the risk of fatal arrhythmias in patients with CPVT.⁴⁰ Collectively, these clinical and experimental data indicate that sodium channel block, by itself, likely confers little or no therapeutic benefit in CPVT. Rather, inhibition of RyR2 is the principal mechanism of antiarrhythmic action of flecainide in CPVT. Whether or not sodium channel block adds any therapeutic benefit when combined with RyR2 block—as is the case for flecainide—remains to be determined.