Ca²⁺-dependent modulation of voltage-gated myocyte sodium channels

Transmembrane action potential initiation and propagation, mediated by surface membrane Na⁺ channel (Nav) proteins, is strategic to activation in excitable cells, of which skeletal and cardiac myocytes constitute important examples. The activation process feeds forward into a ryanodine receptor (RyR) mediated release of sarcoplasmic reticular (SR) store Ca²⁺. The consequent elevation of cytosolic Ca²⁺ concentration [Ca²⁺]_i is central to initiation of myocyte contraction. Ca²⁺ is additionally a strategic second messenger with signalling actions regulating protein activity through widespread cell types. This article addresses recent interest in possible Ca²⁺ feedback signalling on the Na⁺ channel itself, its possible physiological significance, and implications for human disease in skeletal and cardiac muscle. We relate the voltage sensing, and channel opening and inactivation processes in skeletal, Nav1.4 and cardiac Nav1.5 to their potential regulation at direct and indirect Ca²⁺ binding and phosphorylation sites. This includes its III–IV linker region and its interactions with its C-terminal domain, whose different regions are associated with widespread mutations related to skeletal and cardiac muscle disease. We examine in vitro studies in expression systems exploring for direct and indirect effects of Ca²⁺ on channel properties, then extend these to physiological studies in both skeletal and cardiac myocytes in situ, from experimental platforms using normal hearts, and those modelling genetic Ca²⁺ homeostatic disease, broadening these to genetic exemplars for more common human disease types.

Voltage-gated sodium channels (Navs), expressed in excitable cells including neurons and skeletal and cardiac myocytes, initiate action potentials underlying electrical excitation and its propagation. Their principal α-subunits (Mwt ∼220–250 kDa) each include four homologous domains, DI-IV, each containing six transmembrane α-helices, S1–S6, following a S0 helix just preceding the S1 segment (Figure 1A). High-resolution structures obtained by cryo-electron microscopy (cryo-EM) of Nav1.4, Nav1.5 (Figure 1B) and other Nav subtypes demonstrate a highly conserved fourfold pseudosymmetric structure, with voltage sensing helices S1–S4 at the outer rim. Positively charged amino acid residues along one face of each S4 helix permit its outward rotation upon membrane depolarization. Transitions in the DI, DII and DIII S4 helices drive conformational changes in the tethered S5 and S6 helices forming the central pore region within each domain. These open the central, ion-selective pore, transitioning the channel from its resting, closed to an open, activated, state. The latter permits the inward, depolarizing, transmembrane Na⁺ fluxes driving cell excitation.

The slower outward movement of the DIV S4 helix then facilitates binding of a hydrophobic IFM (isoleucine, phenylalanine, and methionine) motif within the cytoplasmic III–IV linker (Figure 2A) to a hydrophobic pocket between domains III and IV (Figs. 1B and 2B) blocking the pore in the channel inactivated state, and restoring the resting membrane potential [1,2]. Protein purification inevitably requires cell lysis, dissipating the cell membrane potential: currently available Nav channel cryo-EM structures likely correspond to the inactivated state [2–6]. Indeed, these structures represent the IFM motif, as expected, engaged with an allosteric intracellular DIII site. In addition however, two separate, short α-helical regions of the DIII-DIV intracellular linker, site A and site B (equivalent to helix 0 of DIV: Figure 1A), make contacts with intracellular sites on DIV, probably further stabilizing the inactivated state (Figure 2A,B) [6]. However, if the engagement of the DIII-DIV linker and IFM motifs with these allosteric sites is indeed critical for promoting the inactivated state, then they must adopt different conformations in the resting and open states.

Nav channels also include a regulatory, globular, intracellular C-terminal domain (CTD), highly conserved amongst Nav subtypes (Figure 3A), connected to the DIV S6 helix via a flexible and disordered linker (Figure 1A). The CTD begins from amino acids 1599 in Nav1.4 and 1773 in Nav1.5, with a sequence of five α-helical regions fitting the consensus sequence for an EF-like hand (EFL) (Figure 3A,B) [7], with the latter part a fibroblast growth factor (FGF) homologous factor (FHF) binding site. It is followed by a sixth α-helical region (Figure 3A,B) and ends with a more disordered and less-well structurally characterized region containing short motifs likely controlling cytoskeletal binding and ubiquitination [8], including a Nedd4-like binding domain, PY motif domain and a syntrophin-anchoring PDZ binding motif (Figure 3A) [9].

NMR analysis of purified EFLs indicates the presence of a prominent cleft in the EFL, bounded by α-helices [10]. This cleft can complex with Site A of the DIII-DIV linker (Figure 2A,B). Modelling of dynamic interactions between the DIII-DIV linker and the CTD through the Nav channel cycle in mammalian Nav1.7 and cockroach NavPas channel structures [11] suggested that in the channel closed state, acidic residues on the CTD EFL domain form salt bridges with basic residues on the DIV S4 helix, whilst Site A of the DIII-DIV linker is held in the CTD EFL cleft. As a consequence, the IFM motif is physically constrained and prevented from prematurely engaging with the inactivated state [11,12]. Upward movement of the DIV S4 helix accompanying channel opening, disrupts these salt bridges. CTD dissociation from the DIII-DIV linker then frees the IFM motif permitting transition into the inactivated state (Figure 4A). Most of the cryo-EM structures thus do not show a resolved CTD [2–5]. This suggests that in the inactivated state, the CTD is free to adopt multiple conformations with respect to the bulk of the Nav channel, constrained only by its tethering to the S6 helix [11].

In skeletal and cardiac muscle, the RyR-mediated SR Ca²⁺ release following Nav1.4 or Nav1.5-mediated depolarization elevates bulk [Ca²⁺]_i from ∼100 nM to 1–10 µM causing contractile activation. In addition, recent reports implicated cytosolic Ca²⁺ in a feedback Nav modulation whether through Ca²⁺ by itself or following its binding to the modulator protein calmodulin [8,13]. The latter nM-low µM K_d Ca²⁺ sensor contains N- and C-lobes, each possessing two Ca²⁺-binding EF hands. In turn, Ca²⁺-free, apo-, calmodulin shows ‘closed’ and ‘semi-open’ states, while Ca²⁺ bound Ca²⁺-calmodulin shows ‘open’ and ‘semi-open’ states. EF hand helix orientations in the ‘open’ and ‘semi-open’ states expose a hydrophobic groove capable of binding distinct α-helical protein sequence motifs. [14–16].

Biophysical studies on isolated protein fragments demonstrate that Site A and Site B of the Nav DIII-DIV linker bind to the C- and N- lobes of Ca²⁺-calmodulin, respectively. But this interaction does not occur with the C- or N-lobes of apo-calmodulin (Figure 2A,C) [17,18]. As noted above, Site A can also bind the CTD EFL cleft. Interestingly, Nav1.5 DIII-DIV linker and CTD co-precipitation occurs in the presence of Ca²⁺-calmodulin, but is inhibited by the Ca²⁺-chelator, EGTA. This could indicate that Ca²⁺-calmodulin acts catalytically to load the DIII-DIV linker onto the CTD [19] (see below).

It had previously been suggested that Nav1.5 CTD EFLs could bind Ca²⁺ directly [7,9,13]. However, in Ca²⁺-binding EF hands, such as those occurring in calmodulin, the Ca²⁺-chelating acidic residues typically lie within turn loops between adjacent α-helices. This pattern is not seen in the CTD-EFL domain [10,20,21]. On the other hand, the CTD, with its significant homologies between Nav subtypes, illustrated for Nav1.4 and Nav1.5 (Figure 3A), binds calmodulin. So, this is the most likely mechanism by which the CTD senses [Ca²⁺]_i. The IQ motif within helix 6 of the Nav1.5 CTD [22] (Figure 3A) can bind the apo-calmodulin C-lobe [14]. Additionally, both the EFL domain and the N-terminal of helix 6 can bind the apo-calmodulin N-lobe (Figure 3Ba). Following Ca²⁺-calmodulin binding, the IQ motif (Figure 3A) can bind the ‘semi-open state’ Ca²⁺-calmodulin C-lobe. But now a downstream, slightly overlapping N-lobe binding motif (NLBM) (Figure 3A) can bind a shifted Ca²⁺-calmodulin N-lobe (Figure 3Ba,b). An alternative structure (PDB: 6mud) for the Nav1.5 CTD Ca²⁺-calmodulin complex is shown in Figure 3Bc [23]. Here, the Ca²⁺-calmodulin N-lobe is untethered to the CTD, and the Ca²⁺-calmodulin C-lobe adopts a strikingly different orientation on helix 6 (Figure 3Bc). However, the CTD construct used in this structure contained a truncated NLBM motif, so that its binding to Ca²⁺-calmodulin N-lobe was likely compromised [23]. Interestingly, a BrS mutation A1924T [24] (Table 1) occurs within the Nav1.5 NLBM site, suggesting that the structure shown in Figure 3Bc could represent an abortive complex, leading to a BrS phenotype. Nav1.4 lacks a functioning NLBM (Figure 3A), whence this shift cannot occur (cf. [23] and it is striking that the rearrangements of calmodulin on the Nav1.4 CTD helix 6 are noticeably less pronounced compared with Nav1.5 (Figure 3Bd,e))

The CTD and DIII-DIV linker of both Nav1.4 or Nav1.5 show mutations associated with specific disease phenotypes. These respectively involve skeletal or cardiac muscle electrophysiological function (Table 1) [25]. Interestingly, within the DIII-DIV linker, gain of Nav1.5 function LQT3 mutations cluster in Site A and affect residues that stabilize DIII-DIV linker binding to the intracellular face of DIV (Figure 2A,B) [26]. In the CTD, the LQT3 mutants tend to occur on helix 6, within and around the IQ motif anchoring apo-calmodulin, as well as contact sufaces between helix 6 and the EFL domain [26] (Figure 3A). These mutations are rescued by overexpressed calmodulin [27].

Contrastingly, loss of Nav1.5 function, Brugada Syndrome (BrS), mutations mainly occur in Site B of the DIII-DIV linker [26] (Figure 2A). One exception, however, is Site A residue Y1494. Mutations in this residue are associated with BrS, not LQT3 (Figure 2A). It may be significant that in the presumed inactivated state structure, residue Y1494 points away from the inactivation site on the intracellular region of DIV (Figure 2B), but in the Ca²⁺-calmodulin C-lobe/Site A complex, it now lies within the protein docking site (Figure 2C) [17]. Thus, BrS and LQT3-associated mutations in Site A, may perturb different molecular contacts. In the CTD, residues associated with BrS cluster particularly within the EFL cleft (Figure 3A,B). This could compromise the capture of the DIII-DIV linker and compromise recovery from inactivation (Figure 4A). In Nav1.4, mutations in two EFL residues, Q1633 and F1705 are associated with myotonia and paramyotonia congenita (PMC), respectively (Figure 3A). In the Nav1.4 EFL structure, these two residues lie suggestively close to each other, where they could help stabilize the EFL cleft (Figure 3Bd,e).

In summary: site A of the DIII-DIV linker can bind to an intracellular site on Nav α-subunit DIV, when the channel is in the inactivated state (Figure 2B). Yet it can also bind to the Ca²⁺-calmodulin C-lobe (Figure 2C) and to the CTD-EFL domain, when the channel is in the closed state [11]. Similarly, site B of the DIII-DIV linker can bind to DIV on the inactivated Nav α-subunit (Figure 2B), but also to the Ca²⁺-calmodulin N-lobe (Figure 2C). Furthermore, in several cases, the same amino acid residues contribute to the different binding states (Figure 2B,C). Thus, within a given channel, these interaction states must be mutually exclusive. Finally, as noted above, the cryo-EM structure (Figure 1B), suggests that the CTD does not bind the DIII-DIV linker when the channel is in the inactivated state [6]. The simplest interpretation is that these different binding states can only take place at specific points during the activation/inactivation/recovery from inactivation cycle of the channel and thus could help impose directionality onto the process.

This idea is outlined in schematic form for the whole Nav activity cycle in Figure 4A and for the role of calmodulin in the recovery from inactivation steps in Figure 4B. One may suggest that immediately after Nav1.5 inactivation, Site A and B, and the IFM motif of the DIII-DIV linker, are all fully engaged with their sites on the α-subunit DIII, and the CTD does not bind the DIII-DIV linker (Figure 4Ba). With an elevated [Ca²⁺]_i, the interaction between Ca²⁺-calmodulin and the CTD is represented by structure PDB: 4jq0 (Figure 3Bb). As the membrane potential hyperpolarizes, the voltage sensing helices of DIII and DIV return to their resting states. Site A and the IFM motif detach from their sites on DIV (Figure 4Bb). The Ca²⁺-calmodulin C-lobe can then bind Site A, adopting the conformation shown in PDB: 4djc (Figure 2C, upper panel). Further rearrangements allow the Ca²⁺-calmodulin N-lobe to bind to Site B as in PDB: 5dbr (Figure 2C lower panel). Together, this could act like a ratchet to prevent the reattachment of Sites A and B and thus the IFM motif to DIV (Figure 4Bc) [18]. There must be further rearrangements to free the calmodulin C-lobe from Site A and the calmodulin N-lobe from site B, so that Site A can reattach to the cleft in the EFL domain of the CTD (Figure 4Bd–f) [21]. Since the affinity of calmodulin for Site A and B is strictly Ca²⁺-dependent, [18], this could take place as [Ca²⁺]_i returns to its resting state, (Figure 4Bf).

Other Nav sites may potentially be involved in Ca²⁺-mediated regulation. Thus, CaMKII-mediated phosphorylation of particular (Ser516, Ser571, and Thr594) residues within the DI-DII intracellular linker region increases late I_Na delaying action potential repolarization, characteristic of LQT3 [28]. However, an existence of calmodulin-KN93 interactions could result in attribution of modified protein function to CaMKII phosphorylation rather than calmodulin action. KN93 may also impair calmodulin-III–IV linker domain interaction and I_Na recovery from inactivation [29]. Phosphorylation at a protein kinase C specific site reduced peak I_Na and shifted (by −15 mV) steady state inactivation V_1/2 [30]. Mutations at a Nav1.5 N-terminal domain calmodulin binding site down-regulated I_Na [31]. Elevated [Ca²⁺] may also up-regulate Nedd4-2 in turn targeting Nav1.5 for degradation via a CTD PY motif [32].

The precise mechanisms of Ca²⁺-mediated channel modification amongst Nav isoforms are thus likely subjects of continued evaluation. Nevertheless, functional assessments confirm regulatory actions of Ca²⁺, Ca²⁺-calmodulin and apo-calmodulin on Nav1.4 and Nav1.5 electrophysiological properties. Table 2 summarizes available in vitro conventional patch-clamp explorations for Ca²⁺-dependent Nav1.4 and Nav1.5 current modulation variously employing heterologous tsA201, HEK293 and CHO expression systems. These quantified steady-state Na⁺ conductance (g_Na) through its maximum currents, I_max, and activation and/or inactivation half-maximal voltages, V_1/2, and slope factors, k. Here, Nav1.4 and Nav1.5 are likely expressed in an absence of other accompanying in vivo proteins. Manoeuvres exploring alterations in [Ca²⁺]_i and calmodulin often used buffered, Ca²⁺-containing (0–10 µM), pipette solutions, to test for Ca²⁺, Ca²⁺-calmodulin or apo-calmodulin-mediated actions, on Nav1.4 and/or Nav1.5 C-terminal EF-hand or IQ domains, with some differences between reports [7,10,13,15,17,33–35].

However, their pipette [Ca²⁺] often significantly exceeded the Ca²⁺ dissociation constant, K_d of either the EGTA (67 nM) or 1,2-bis(2-aminophenoxy)ethane-N,N,N′,N′-tetra-acetic acid (BAPTA) (192 nM) pipette buffer, even as determined in the absence of Mg²⁺ [36]. Possible Ca²⁺-F^- binding (solubility product K_sp ∼ 3.45 × 10⁻¹¹ M³) with use of (often ∼100 mM, giving [Ca²⁺] = 3.45 nM) CsF-containing pipette solutions to stabilize the whole-cell patch-clamp recordings, and intrinsic cell buffering properties, added additional uncertainties to detailed interpretation of their experimental results [37].

Nevertheless, all these studies reported little or no effects on k. Nor did pipette Ca²⁺/EGTA, Ca²⁺/BAPTA or calmodulin manipulations alter I_max. However, experiments instead buffering pipette Ca²⁺ using F^--free N-(2-hydroxyethyl)ethylenediamine-N,N′,N′-triacetic acid (HEDTA), and elevating [Ca²⁺]_i by Nitr-photo-uncaging, or activating co-expressed Cav1.2, contrastingly all reduced I_max in Nav1.4, or Nav1.5 chimeras expressing the Nav1.4 CTD (Figure 5). Contrastingly, they did not do so with Nav1.5 or Nav1.4 chimeras expressing the Nav1.5 CTD [36,38]. Inactivation V_1/2s were unaffected and activation V_1/2s not explored [36]. The remaining studies investigating V_1/2 reported consistently unchanged activation V_1/2s, but either altered or depolarized inactivation V_1/2s, with no trends related to expression platform (Table 2). Nor did inactivation time constants alter, with two exceptions [15,18]. Finally, Ca²⁺ uncaging also revealed that FGF homologous factors (FHF) diminished Ca²⁺-calmodulin-regulation of Nav1.4 expressed in HEK293 cells, possibly involving allosteric sites within upstream CTD regions distinct from the calmodulin-binding interface [39].

Ca²⁺ uncaging investigations extending to skeletal myotubes derived from mouse glt cells similarly demonstrated Ca²⁺-mediated Nav1.4 regulation at sensitivities appropriate for physiological Ca²⁺ transients, but no such Nav1.5 regulation in adult guinea-pig ventricular myocytes [36]. However, in freshly isolated rabbit ventricular myocytes, [Ca²⁺]_i elevations produced by Ca²⁺-BAPTA (0–500 nM)-buffered patch-clamp electrode solutions or caffeine challenge caused parallel reductions in I_Na density, unit channel amplitudes and maximum action potential upstroke rates (dV/dt)_max, without altering steady state voltage dependences of I_Na activation or inactivation [40]. Cultured rat neonatal ventricular cardiomyocytes also showed altered Nav expression with more sustained alterations in intracellular Ca²⁺ homeostasis. Nav1.5 mRNA levels then altered in parallel with decreases or increases in whole cell patch-clamp I_Na with 24 h sustained elevations (10 mM) or BAPTA-AM-mediated reductions of culture media [Ca²⁺]. These also occurred without alterations in single conductance, or activation and inactivation properties [41].

These varied observations could arise from a range of possible Nav Ca²⁺ sensing mechanisms, including direct Ca²⁺ binding to the first EF-like hand [7,15,35], or Ca²⁺-calmodulin or apo-calmodulin binding to, the CTD [34,36]. The latter possibilities were compatible with reported calmodulin binding to peptide channel fragments [42,43]. Finally, structural studies invoked possible Nav regulatory sites alternative to the CTD including the III–IV loop [17]. At all events, this available evidence permits a direct in vivo regulation of Nav-mediated excitable activity by intracellular Ca²⁺, involving mechanisms highly conserved among all nine Nav isoforms. This could complement or replace hypotheses invoking [Ca²⁺]_i-mediated increases in electrogenic Na⁺/Ca²⁺ exchanger (NCX) activity in cardiac muscle under pro-arrhythmic conditions [44]. The latter may mediate delayed after depolarization (DAD) phenomena and is also implicated in altering action potential recovery as opposed to initiation and propagation activity. NCX may also increase [Na⁺]_i thereby influencing transmembrane Na⁺ electrochemical gradients. However, this would involve µM-levels corresponding to the altered [Ca²⁺]_i as opposed to normal background nM-[Na⁺]_i levels. Furthermore, NCX activity is not a prominent normal skeletal as opposed to cardiac muscle feature. Nevertheless, in either event, over the long term, reduced or increased background [Ca²⁺]_i resulting from sustained low or high firing rates could furnish a form of Ca²⁺ memory modifying Nav expression or gating and therefore its availability for driving action potential upstroke and propagation. In skeletal muscle, this could reduce cell excitability permitting recovery from fatiguing stimulation. However, the accompanying conduction velocity (CV) reductions could contribute to pathological cardiac arrhythmic or epileptiform nerve cell phenotypes.

In vivo Ca²⁺-dependent Nav modulation was observed in native cardiac or skeletal myocytes in intact physiological systems and clinical disease models. Use of loose, as opposed to conventional cell-attached, patch-clamp methods, avoided Ca²⁺ perturbations produced by the measurement method itself. I_Na families recorded from voltage steps from resting to sequentially depolarized activating test potentials, followed by further pulses to a fixed depolarized level to evaluate the resulting channel inactivation (Figure 6Aa,b) were compared before and following perturbations of their in vivo Ca²⁺ homeostatic mechanisms. Studies in both skeletal and cardiac myocytes demonstrated potentially physiologically significant negative feedback regulation of Nav1.4 and Nav1.5 by RyR-mediated release of intracellularly stored SR Ca²⁺. In murine skeletal muscle, acute RyR2 activation by the exchange protein directly activated by cAMP (Epac) by the activator 8-(4-chlorophenylthio)adenosine-3′,5′-cyclic monophosphate (8-CPT, 1 µM) [45], reduced maximum I_Na whilst leaving V_1/2 values unchanged, actions abrogated by the RyR-inhibitor dantrolene (10 µM) [46]. The RyR agonist caffeine, at concentrations of 0.5 or 2 mM, produced sustained activation or transient activation followed by inactivation, of RyR-mediated SR Ca²⁺ release and corresponding parallel alterations in [Ca²⁺]_i [47; Figure 6B]. These changes directly paralleled time-dependent decreases or increases in peak I_Na values (Figure 6Cc,d) also abrogated by dantrolene (10 µM). Finally, dantrolene applied by itself produced small increases in I_Na, suggesting inhibitory effects of even background Ca²⁺ release on I_Na ((Figure 6Ca,b) [48], potentially through formation of microdomains localizing [Ca²⁺]_i heterogeneities in junctional regions separating the T-tubular and SR membranes [49]

Elevating [Ca²⁺]_i by applications of high extracellular [Ca²⁺], caffeine, and the SR Ca²⁺ ATPase inhibitor cyclopiazonic acid in murine atria [50], in addition to 8-CPT in murine atria and ventricles, all reduced mean peak inward I_Na. 8-CPT (1 µM) induced Ca²⁺ homeostatic changes manifesting as spectrofluometrically measured spontaneous Ca²⁺ waves in murine atrial myocytes (Figure 7A) [45]. These findings accompanied 30–50% reductions in inward I_Na (Figure 7B,C), abrogated by dantrolene (10 µM), which by itself left I_Na at pre-treatment levels. Inactivation V_1/2 and k (Figure 7D), and time constants for Na⁺ current recovery from inactivation remained unchanged [51]. Intracellular sharp microelectrode membrane potential recordings in intact Langendorff-perfused preparations correspondingly demonstrated reduced maximum atrial and ventricular (dV/dt)_max [51]. Action potential latencies reflecting delayed conduction increased while action potential durations and refractory periods were unchanged. The hearts also showed increased ventricular arrhythmic incidences following rapid pacing or extrasystolic stimuli [52].

A first clinical example of a C-terminal Nav1.4, SCN4A, mutation associated with human disease is cold-aggravated myotonia, which causes transient myotonic stiffness or renders fibres transiently inexcitable resulting in a periodic paralysis (Table 1). The SCN4A mutant concerned contained two predicted amino acid substitutions, a DIS5-S6 loop T323M and an intracellular C-terminus F1705I substitution. Whole cell patch clamp I_Na from transiently transfected HEK293 cells expressing Nav1.4-T323M were indistinguishable from WT, consistent with a benign polymorphism. However, Nav1.4-F1705I channels showed a slowed fast inactivation with a positive 8.6 mV shift in steady-state voltage-dependence often associated with myotonia, but normal activation, recovery from fast inactivation or persistent current [53].

The hereditary pro-arrhythmic condition catecholaminergic polymorphic ventricular tachycardia (CPVT), is associated with gene mutations involving ryanodine receptor type 2 (RYR2), calsequestrin (CASQ2), triadin (TRDN) or calmodulin (CALM1, CALM2 and CALM3) [54]. It clinically presents as potentially fatal bidirectional, and mono and polymorphic ventricular tachycardia (VT) provoked by adrenergic stress. Experimental murine RyR2-P2328S ventricles showed abnormal RyR2-mediated diastolic [Ca²⁺]_i elevations [55]. Homozygotic murine RyR2-P2328S ventricles showed reduced loose patch-clamp I_Na and possible additional evidence for down-regulated Nav1.5 expression [56]. Intrinsically beating murine RyR2-P2328S hearts recapitulated the clinical pro-arrhythmic phenotypes on isoproteronol and caffeine challenge. Intracellular floating microelectrode and multi-electrode array recordings then demonstrated correspondingly reduced (dV/dt)_max, and ventricular epicardial CVs, particularly in homo- as opposed to heterozygotic, RyR2-P2328S/+, hearts, changes not observed in wild-type (WT) controls [57].

CPVT is also associated with atrial fibrillation similarly attributed to abnormal Ca²⁺ homeostasis particularly following increased sympathetic tone [58]. In superfused RyR2-P2328S/P2328S atrial preparations, loose patch clamp measurements also demonstrated reduced peak I_Na with otherwise normal activation and inactivation current–voltage relationships (Figure 8A,Ba) [50]. Floating intracellular microelectrode measurements demonstrated reduced (dV/dt)_max and interatrial CVs though normal action potential duration amplitudes and refractory periods (Figure 8Bb,c) while multi-electrode arrays detected reduced atrial epicardial action potential CVs in RyR2-P2328S/P2328S atria when compared with WT [59]. Intrinsically active and regularly stimulated RyR2-P2328S/P2328S but not wild-type atria correspondingly showed frequent sustained tachyarrhythmias, delayed afterdepolarizations and ectopic action potentials. Extrasystolic S2 stimulation provoked arrhythmia at longer S1S2 intervals in RyR2-P2328S/P2328S than WT atria, nevertheless corresponding to similar (dV/dt)_max, and effective interatrial CVs as in WT [59]. Gain-of-function skeletal muscle RYR1 mutations are associated with a malignant hyperthermia typically following halothane anaesthesia. Reports of increased slowly inactivating inward, tetrodotoxin sensitive current in cultured human malignant hyperthermia skeletal myocytes may prompt further investigations into possible electrophysiological, Nav1.4 phenotypes [60].

The above properties may underpin reported paradoxical pro- and anti-arrhythmic actions of low (1 µM) flecainide concentrations in WT and RyR2-P2328S/P2328S murine atria. Flecainide blocks both Nav1.5 and RyR2 with IC₅₀s of 2–7 µM and 5–11 µM, respectively [61–63]. Either effect could potentially rescue an elevated [Ca²⁺]_i. On the one hand, flecainide's Class Ic Nav1.5 blocking action causes a pro-arrhythmic CV slowing; however, action of a consequently reduced [Na⁺]_i on NCX could reduce pro-arrhythmic [Ca²⁺]_i elevations [64–66]. In intact WT hearts, flecainide (1 µM) exerted atrial pro-arrhythmic effects, accompanying reduced loose patch clamp I_Na and multi-electrode array recorded CV, whilst sparing refractory periods (Figure 9Aa,Ba,Ca). On the other hand, in RyR2-P2328S/P2328S atria, flecainide paradoxically rescued increased arrhythmic frequency. However, in contrast with its Nav1.5 inhibitory action in WT, it rescued I_Na and maintained CV at WT values, leaving refractory periods unchanged (Figure 9Ab,Bb,Cb), effects directly replicated by the RyR blocker dantrolene (Figure 9D) [67]. These findings together suggested a rescue of the arrhythmic phenotype by RyR2 block causing Nav1.5 rescue rather than Nav1.5 block. RyR2 inhibition would reduce the elevated diastolic Ca²⁺ and its pro-arrhythmic inhibition of Nav1.5 [67]. The latter mechanism of action could underlie anti-arrhythmic effects of monotherapeutic low-dose flecainide introduced to treat clinical CPVT [62,68–71].

Ca²⁺-mediated regulation of Nav1.5 may also contribute to commoner pro-arrhythmic cardiac conditions associated with spontaneous SR Ca²⁺ leak. The latter was reported in peroxisome proliferator activated receptor-γ coactivator-1 (PGC-1) transcriptional coactivator deficient (Pgc1-β^−/−) murine models for pro-arrhythmic metabolic changes related to ageing, obesity and diabetes mellitus [72]. Atrial fibrillation, cardiac failure and hypertrophic cardiomyopathies are also accompanied by spontaneous SR Ca²⁺ leak. Classically, SR Ca²⁺ leak is implicated in a pro-arrhythmic activation of inward depolarizing, NCX current [44]. However, the pro-arrhythmic phenotypes in Pgc1-β^−/− atria and ventricles were also associated with reduced I_Na [73,74], (dV/dt)_max and CVs [75,76]. A decreased I_Na in these experimental conditions as well as in clinical heart failure or atrial fibrillation slowing action potential CV could contribute pro-arrhythmic substrate.

• Action potential generation by Na⁺ channel (Nav) activation and the resulting release of intracellular Ca²⁺ stores underly skeletal and cardiac myocyte excitation-contraction coupling abnormalities which underly a wide range of human genetic diseases.

• Nav channels possess sites directly or indirectly binding Ca²⁺ potentially of regulatory importance in their reciprocal Ca²⁺-mediated feedback regulation. Evidence from cell expression systems, native myocytes and normal and disease models demonstrate such Ca²⁺-mediated Nav regulation effects.

• Future studies may correlate this molecular evidence bearing particularly on the Nav C-terminal and III–IV linker domains and biophysical studies of Na⁺ channel function with associated clinical conditions.

The authors declare that there are no competing interests associated with the manuscript.

We thank the British Heart Foundation (PG/14/79/31102, PG/19/59/34582 and Cambridge Centre for Research Excellence, S.C.S., A.P.J., C.L.-H.H.), Medical Research Council (MR/M001288/1, C.L.-H.H.) and Wellcome Trust (105727/Z/14/Z, C.L.-H.H). for their generous support.

Open access for this article was enabled by the participation of University of Cambridge in an all-inclusive Read & Publish pilot with Portland Press and the Biochemical Society under a transformative agreement with JISC.

C.L.-H.H. conceived and drafted the review. A.P.J. provided the biochemical aspects bearing on Nav1.4 and Nav1.5 structure. A.P.J., S.C.S. and Z.H. reviewed the mutations associated with the CTD, particularly Table 1. S.C.S., H.R.M. and C.L.-H.H. collated and reviewed the patch clamp data, particularly the systematic classification in Table 2, and reviewed the physiological findings. S.C.S. and C.L.-H.H. reviewed the experimental studies in disease models.

BAPTA

bis(2-aminophenoxy)ethane-N,N,N′,N′-tetra-acetic acid

CPVT

catecholaminergic polymorphic ventricular tachycardia

CTD

C-terminal domains

CV

conduction velocity

EFL

EF-like hand

NCX

Na⁺/Ca²⁺ exchanger

NLBM

N-lobe binding motif

PMC

paramyotonia congenita

SR

sarcoplasmic reticular