Late sodium current contributes to diastolic cell Ca²⁺ accumulation in chronic heart failure

Abstract

We elucidate the role of late Na⁺ current (I_NaL) for diastolic intracellular Ca²⁺ (DCa) accumulation in chronic heart failure (HF). HF was induced in 19 dogs by multiple coronary artery microembolizations; 6 normal dogs served as control. Ca²⁺ transients were recorded in field-paced (0.25 or 1.5 Hz) fluo-4-loaded ventricular myocytes (VM). I_NaL and action potentials were recorded by patch-clamp. Failing VM, but not normal VM, exhibited (1) prolonged action potentials and Ca²⁺ transients at 0.25 Hz, (2) substantial DCa accumulation at 1.5 Hz, and (3) spontaneous Ca²⁺ releases, which occurred after 1.5 Hz stimulation trains in ~31% cases. Selective I_NaL blocker ranolazine (10 μM) or the prototypical Na⁺ channel blocker tetrodotoxin (2 μM) reversibly improved function of failing VM. The DCa accumulation and the beneficial effect of I_NaL blockade were reproduced in silico using an excitation-contraction coupling model. We conclude that I_NaL contributes to diastolic Ca²⁺ accumulation and spontaneous Ca²⁺ release in HF.

Introduction

Congestive heart failure (HF) is associated with severe abnormalities in both cardiac rhythm and contractile function. While Ca²⁺ directly triggers contractions in cardiac myocytes, both cell Ca²⁺ content and dynamics are modulated by Na⁺ ions [1]. Abnormal Na⁺ handling and Na⁺/Ca²⁺ interplay in HF are being extensively investigated in order to develop new approaches to the treatment of HF [2–4]. While the majority of Na⁺ channels open and inactivate during the action potential (AP) upstroke, some of them reopen during the AP plateau, carrying the so-called late or persistent Na⁺ current (I_NaL) [5]. We found I_NaL in cardiomyocytes of both normal and failing human hearts [6], and further studies showed that I_NaL increases in HF, both in patients and in animal models [5, 7–9]. Augmented I_NaL plays an important role in impaired repolarization of cardiomyocytes of failing hearts [6, 7, 9–13].

Previous studies have suggested that during each beat a relatively small, but long lasting, I_NaL brings a substantial amount of Na⁺ into cardiomyocytes, comparable to that carried by a large but very brief transient Na⁺ current (I_NaT) [9]. The latter is reduced in chronic HF [8, 14, 15], whereas I_NaL is increased; hence, the relative contribution of I_NaL to the regulation of intracellular Na⁺ is expected to increase further in myocytes from failing hearts. Thus I_NaL, by altering AP and providing Na⁺ influx, might be implicated in Na⁺-related Ca²⁺ overload, especially in the context of up-regulated Na⁺/Ca²⁺ exchanger (NCX) in HF [16–18]; reduced sarcoplasmic reticulum (SR), Ca²⁺ ATPase (SERCA), or Ca²⁺-pump function; down-regulated Ca²⁺-release channel (ryanodine receptor, RyR); and increased SR Ca²⁺ leak [18–23]. The increased I_NaL by depolarizing the cell membrane and increasing late Na⁺ influx during AP is likely to facilitate a shift of the NCX operation in HF from the prime forward mode to the reverse mode, i.e., from Ca²⁺ efflux to Ca²⁺ entry [24], thereby contributing to cell Ca²⁺ accumulation in HF. Abnormal cell Ca²⁺ accumulation, in turn, worsens both contractility (via diastolic function) and rhythm (via spontaneous Ca²⁺ releases-triggered delayed after-depolarizations, DADs) [18, 25]. Therefore, inhibition of I_NaL could be a potential target to improve function of the failing heart, but this possibility remains largely unexplored. While some prior studies have indeed shown improved diastolic performance due to inhibition of I_NaL in different experimental conditions (including in vivo settings), those studies were mainly performed in a normal myocardium using pharmacological I_NaL enhancement with toxins or other Na⁺ channel agonists [26–28], as well as expressing LQT-3 syndrome-related mutant Na⁺ channels with retarded inactivation [29]. The properties of I_NaL in HF myocytes are distinctively different [9, 11, 12] from the non-inactivating I_Na produced by Na⁺ channel agonists and mutations in myocytes of normal hearts (NH), not to mention differences in Ca²⁺ handling. Thus, the role of an augmented I_NaL in HF remains to be established.

Using a combination of experimental and numerical modeling approaches, the present study tested the hypothesis that I_NaL is indeed a major contributor to the dynamic, Na⁺-dependent diastolic Ca²⁺ (DCa) accumulation in single ventricular myocytes isolated from failing hearts. We used a canine chronic HF model that causes numerous physiological deficiencies similar to those observed in patients with HF [30]. Isolated single ventricular myocytes from failing hearts exhibited abnormal Ca²⁺ handling, including substantial DCa accumulation at a moderate-high physiological pacing rate of 1.5 Hz, which is deleterious to left ventricular ejection fraction in HF patients [31, 32]. Inhibition of I_NaL greatly reduced DCa and substantially decreased the probability of spontaneous Ca²⁺ releases (SCaRs), a well-known initiation mechanism of DADs and triggered arrhythmia [18, 25]. Our numerical modeling confirmed that I_NaL provides a substantial contribution to diastolic cell Ca²⁺ accumulation observed experimentally. Thus, inhibition of I_NaL may suppress a significant portion of the Ca²⁺ accumulation and decrease occurrence of SCaRs in cardiomyocytes and, thereby, is likely to improve heart contractility and rhythm in HF.

Methods

This study conforms to the guide of the care and use of laboratory animals published by the NIH and was approved by the Institutional Animal Care and Use Committee of the Henry Ford Health System. Our experimental methods, data analysis, theoretical formulations, and calculations are given in the Electronic Supplementary Material (ESM). In brief, chronic HF in dogs was induced by multiple sequential coronary microembolizations. Ca²⁺ signals were measured in field-stimulated, fluo-4-loaded single ventricular myocytes using a photo-multiplier; electrophysiological recordings were performed using the patch-clamp technique. Ca²⁺ signals and APs were measured at 35°C and 1.8 mM [Ca²⁺]_o, I_NaL was measured at room temperature (21–23°C).

We used ranolazine (RAN) and tetrodotoxin (TTX) as pharmacological tools. Whereas TTX blocks both I_NaT and I_NaL equally [9], RAN blocks I_NaL preferentially [13]. RAN is superior to other Class I or III drugs (RAN > amiodarone > flecainide > mexiletine > lidocaine) [13, 33–36] at blocking I_NaL over I_NaT. Furthermore, RAN at the concentration of 10 μM used in the present study (i.e., close to therapeutic range of 2–8 μM) has only minor effects on K⁺ currents [33], unlike amiodarone, which has major effects on K⁺ currents. Concentrations of RAN (10 μM) and TTX (2 μM) were chosen to produce a comparable I_NaL inhibition (~1.5 of IC₅₀) based on our previous studies [9, 13].

Results are expressed as mean ± SEM (if not stated otherwise), and statistical significance was based on P < 0.05, ANOVA followed by Bonferroni’s post-hoc test. To evaluate the probability of the SCaRs we used nonparametric two-tailed Fisher’s exact test.

We simulated the effect of selective inhibition of I_NaL on DCa accumulation and APs in silico using a modified excitation-contraction (E-C) coupling model of failing canine ventricular myocytes developed previously by Winslow et al. [37] (see details in the ESM).

Results

Minor effects of ranolazine and TTX on intracellular Ca²⁺ in myocytes from normal heart

First we tested the effects of 10 μM RAN and 2 μM TTX in field-stimulated cardiomyocytes of normal dogs on the duration (90%) of Ca²⁺ transients (CaT₉₀) and accumulation of DCa during a pulse train (see the ESM for definition) (Fig. 1). Both RAN and TTX only slightly reduced CaT₉₀ at a low stimulation rate of 0.25 Hz, and these effects were not statistically different (Fig. 1c). At a high stimulation rate of 1.5 Hz, accumulation of DCa was negligible (a few percent) both in control and under RAN or TTX (Fig. 1b).

Fig. 1

Representative examples of Ca²⁺ transient recordings in normal dog heart ventricular myocytes at low (a) and high (b) pacing rates before and after infusion of ranolazine (RAN) or tetrodotoxin (TTX). c Data summary (mean ± SEM, n = 10–20) for Ca²⁺ transient duration (CaT₉₀)

Improvement of intracellular Ca²⁺ handling in myocytes from failing heart by TTX and ranolazine

Ca²⁺ transients were substantially different in field-stimulated myocytes of failing compared to normal heart. CaT₉₀ of failing myocytes was significantly longer, and the CaT had a spike-dome configuration when myocytes were stimulated at a frequency of 0.25 Hz (Fig. 2). Some failing myocytes exhibited multiple oscillations on top of the “dome” (Fig. 2a, upper panel). Both RAN and TTX improved CaT, as they significantly and reversibly shortened CaT₉₀ by suppressing the dome phase (Fig. 2c).

Fig. 2

Ranolazine and TTX shorten Ca²⁺ transient duration (CaT₉₀) in failing dog ventricular myocytes at low pacing rates (0.25 Hz). a, b Representative examples of Ca²⁺ transient recordings, c summary of the data (mean ± SEM, n = 10–22) for CaT₉₀

Increasing the stimulation rate to 1.5 Hz markedly increased the DCa level during the pulse train (Fig. 3a, b, upper panels) and induced beat-to-beat CaT alternations in amplitude and duration in all HF myocytes studied (Fig. 3a, upper panel). Both RAN (Fig. 3a) and TTX (Fig. 3b) significantly and reversibly decreased DCa accumulation and eliminated these changes (Fig. 3c).

Fig. 3

Ranolazine and TTX reduce diastolic Ca²⁺ elevation at high pacing rates (1.5 Hz) in failing dog ventricular myocytes. a, b Representative examples of Ca²⁺ transient recordings, c data summary for diastolic Ca²⁺ (CaD) changes (ΔCaD) during a pulse train. Data are mean ± SEM pooled from 13 to 30 cells

Next we determined whether failing myocytes generate SCaRs when subjected to 1.5 Hz trains of stimulation pulses. SCaRs occurred in ~31% of instances with a delay of 2.1 ± 0.2 s (n = 17) after the end of the last pulse of the 1.5 Hz train of pulses (Fig. 4a). SCaRs were absent in myocytes of NH. The SCaR amplitude was comparable to that of CaT within the pulse train (see below) and ranged from 163 to 251 nM (calculation of [Ca²⁺]_i is given in the ESM). Both TTX and RAN greatly (~threefold) and reversibly reduced the probability of SCaR occurrence (Fig. 4).

Fig. 4

Ranolazine prevents spontaneous Ca²⁺ releases (SCaR) in myocytes from failing dog heart. Representative traces of Ca²⁺ signals just after the 1.5 Hz pulse train a in control and b in the presence of 10 μM ranolazine (RAN). Vertical arrows indicate the last pulse in the train. c RAN significantly and reversibly decreases the rate of SCaR. The probability of SCaR occurrence (bars) was evaluated as the percentage of traces that contained SCaR (similar to that shown in a). In control conditions 18 out of total 58 traces contained SCaR, in TTX it was present in 4 out of total 34 traces, in RAN it was present in 3 out of 31 traces, and after washout it was present in 13 out of total 40 traces. Statistical comparison was performed using two-tailed Fisher’s exact tests, and P < 0.05 was considered to be significant. Data were pooled from 31 to 58 cells

Does ranolazine affect SR Ca²⁺ content and NCX function?

The mechanism for the SCaRs, which are precursors for DADs, is the interplay between the SR Ca²⁺ load, cytoplasmic [Ca²⁺], and NCX (see for review [25]). Therefore, in order to further address the mechanism of the beneficial effects of RAN to decrease the probability of SCaR occurrence (Fig. 4), in the next experiments we determined SR Ca²⁺ content (or load) and NCX function by the caffeine-induced Ca²⁺-transients at a basal stimulation rate of 1.5 Hz in the presence and absence of the drug. The amplitude of the caffeine-induced Ca²⁺ transient was used as a measure for SR Ca²⁺ load [28], and the decay time constant was used as an estimate of NCX activity [38]. Figure 5a shows traces of the caffeine-induced Ca²⁺ transients in the absence (left panel) and presence of RAN (10 μM, right panel). Figure 5b and c shows summary data of the caffeine-induced transient amplitude (SR Ca²⁺ content) and decay time constant (NCX function). Our data show that the decay kinetics (Fig. 5c) after RAN application changes insignificantly. Although there was an apparent reduction in SR Ca²⁺ load (Fig. 5b), the statistical analysis does not reveal statistically significant difference in caffeine-induced Ca²⁺ transient amplitude (Fig. 5b). Therefore, we assumed that RAN affects neither SR Ca²⁺ load nor NCX function.

Fig. 5

Ranolazine affects SR Ca²⁺ content or NCX function insignificantly in ventricular myocytes from failing dog hearts. a Original traces of Ca²⁺ transients during the exposure to caffeine (10 mM) in control (left panel) and in the presence of ranolazine (RAN, 10 μM, right panel). b Average data for SR Ca²⁺ load, c average data for Ca²⁺ decay as a measure of NCX function (single exponential fit). Statistical analysis does not reveal a statistical difference between control and RAN for these parameters (ANOVA). Data bars in b and c represent mean ± SEM pooled from 8 to 15 cells

Effects of ranolazine on I_NaL in myocytes from failing heart

In our previous study we have shown that RAN preferentially blocked I_NaL (IC₅₀ = 6.46 μM, for HF) over I_NaT (IC₅₀ = 294 μM, NH; IC₅₀ = 244 μM, HF) [13]. The effect of the drug on I_NaL kinetics has not been previously studied and thus was examined in the present study. The decay of I_NaL was approximated by a double-exponential fit (Eq. 4 in the ESM; see examples in Fig. 5a) as previously suggested [12]. The fast component (τ_BM, tens of ms) represents burst mode (BM) and the slow component (τ_LSM, hundreds of ms) represents late scattered mode (LSM) of late Na⁺ channel openings [12]. The effect of RAN (10 μM) on I_NaL was complex and reversible: the density was reduced (Fig. 6b) and the drug accelerated decay kinetics of both I_NaL components (Fig. 6c, d).

Fig. 6

Effects of RAN (10 μM) on I_NaL in ventricular myocytes from canine failing hearts. a Representative traces along with double-exponential fit of the time course of I_NaL decay (solid lines). b Summary of the data on I_NaL density measured as a mean current within 200–220 ms after the depolarization onset, and decay kinetics (c, d) measured at −30 mV in control, in the presence of RAN, and during the washout. Data are mean ± SEM pooled from 5 to 23 cells. See details in text

Effects of ranolazine on action potentials in myocytes from failing heart

The AP shape and its duration, especially the variability of AP duration, play an important role in abnormalities of the E-C coupling in HF. We previously showed that AP duration of ventricular myocytes is prolonged (at low rates) and AP duration variability is increased in this canine HF model and in HF patients compared to NH, and the role of I_NaL in the AP duration abnormalities has been suggested [6, 7, 10]. Using the specific I_NaL inhibitor RAN allows further investigation of the role of I_NaL in AP regulation in HF. Accordingly, we determined the effects of RAN (10 μM) on AP duration and its variability at low and high pacing rates in myocytes from failing hearts. The results of these experiments are summarized in Fig. 7. At both stimulation rates RAN effectively and reversibly shortened the prolonged APs and abolished the beat-to-beat variability apparent as a large dispersion in the AP duration distribution histograms (Fig. 7c–h). For comparison the AP duration and its dispersion in normal dog ventricular myocytes recorded at 0.25 and 1.5 Hz were 337 ± 52 ms (mean ± SD, n = 8 cells) and 303 ± 40 ms (n = 9), respectively (data pooled from the previous publications [7, 9, 13]).

Fig. 7

RAN reversibly shortens the AP duration (APD) and reduces the dispersion of APD in ventricular myocytes from canine failing hearts. a, b Representative AP traces at low and high pacing rates recorded at the end of a pulse train, respectively. Dotted line indicates −30 mV level; APD values at this potential are indicated at the traces. c–h Histograms of the distribution of the APD measured at 90% of repolarization (APD₉₀). Bin sizes were 50 and 23 ms for c, e, g and for d, f, h, respectively. SD Standard deviation. APs were recorded in control (c, d), in the presence of 10 μM RAN (e, f), and after RAN washout (g, h). Data were pooled from 6 to 12 cells of four failing hearts

Numerical evaluation of I_NaL and late Na⁺ influx during action potential plateau and its importance for the diastolic Ca²⁺ accumulation in HF

Based on our experimental data, we performed several numerical estimates to answer three key important questions:

• 1. What is the time course for I_NaL and late Na⁺ influx (i.e., I_NaL integral) during AP plateau?

• 2. How much of this Na⁺ influx could be blocked by 10 μM of RAN?

• 3. Is late Na⁺ influx sufficient to account for a major portion of diastolic cell Ca²⁺ accumulation observed experimentally during the pulse train at 1.5 Hz?

In the first approximation we used experimentally recorded AP traces to simulate Na⁺ influx dynamics during AP plateau (above −30 mV) at low and high rates of stimulation before and after RAN (Fig. 8a, b) (see formulations in the ESM). Results of these AP-clamp simulations are as follows: Because the model includes RAN-induced steady-state inactivation (SSI) shifts found at the physiological resting potential [13] (see Table 1 in the ESM), it predicts greater inhibition of I_NaL and late Na⁺ influx by RAN in physiological conditions in comparison to the voltage-clamp data obtained from a very low holding potential (compare Figs. 6a, 8c). More specifically, RAN blocks about 77% of the late Na⁺ influx during the single AP (from 64.3 to 14.5 fC/pF at low rate and from 50.4 to 11.5 fC/pF at the high rate).

Fig. 8

AP-clamp numerical model. A substantial inhibition of I_NaL and late Na⁺ influx (77%) during AP plateau by a therapeutic concentration (10 μM) of RAN. a, b Experimental AP recordings at low and high pacing rates in the absence or presence of RAN. c, d Predicted I_NaL and its integral (~late Na⁺ influx) during the corresponding AP shapes (Eq. 9 in the ESM)

A remarkable portion of late Na⁺ influx is likely exchanged by NCX for Ca²⁺ steady state during stimulation at 1.5 Hz (Fig. 3) for the following reasons: (1) NCX function is enhanced in HF [16–18], (2) NCX operates in the reverse mode during AP plateau in canine failing myocytes (see Fig. 1f in [37]) and extrudes 3 Na⁺ for influx of 1 Ca²⁺, and (3) I_NaL blockade by RAN substantially reduces DCa (Fig. 3) but affects neither SR Ca²⁺ loading, nor NCX function (Fig. 5). Based on this reasoning, our numerical estimate demonstrates that the late Na⁺ influx (~I_NaL integral) during the AP plateau might be indeed substantial and sufficient to increase cytosolic [Ca²⁺] by ~72 nM assuming its efflux by the reverse Na⁺/Ca²⁺ exchange at the steady state during a 1.5 Hz stimulation train (see the ESM for detailed formulations and calculations). While our AP clamp simulations, described above, include real AP shapes of myocytes in our specific HF model, they only consider I_NaL, NCX, and cell Ca²⁺ buffering. Thus, the above 72 nM is likely a higher estimate because this simple estimate does not include possible contributions of other Ca²⁺ and Na⁺ transport systems. Therefore, we further addressed the role on I_NaL in CaD accumulation in silico using our modified version of an established model for pacing-induced dog HF ventricular myocytes that includes all essential components of E-C coupling and Na⁺ homeostasis [37]. This pacing-induced HF model produces similar remodeling of AP and ionic currents, including the augmented I_NaL, found in patients with HF and in our dog HF model [6–9, 37, 39]. The details of the model modification to include I_NaL are given in the ESM. The results of our model simulations are shown in Fig. 9. We applied a train of 11 stimulation pulses in the model and recorded intracellular Ca²⁺ dynamics and APs in control and when I_NaL was set to zero. Our model simulations reproduced DCa accumulation (Fig. 9, upper panel) observed experimentally in failing myocytes (Fig. 3a, b, upper panels). Selective blockade of I_NaL in the model (either completely or partially, simulating RAN effect) shortened AP (Fig. 9, lower panel), and substantially reduced this simulated DCa accumulation similarly to that found in experiment (Figs. 7b–f, 3a, b, middle panels).

Fig. 9

In silico demonstration of the role of the augmented I_NaL in AP shape and diastolic Ca²⁺ accumulation in canine failing ventricular myocytes. Numerically simulated dynamics of intracellular Ca²⁺ concentration ([Ca²⁺]_i) (upper panel) in a train of 11 pulses applied with a rate of 1.5 Hz. Complete I_NaL elimination or reduction by RAN (10 μM) substantially reduces the diastolic Ca²⁺ accumulation and shortens AP duration (lower panel). Simulations were performed using a modified Winslow et al. E-C coupling model of failing canine ventricular myocytes [37] (see details in the ESM)

We next calculated the absolute cytosolic free Ca²⁺ dynamics during stimulation with the rate of 1.5 Hz from the respective Ca²⁺ signal traces represented originally by F/F₀ values (ESM). The CaT amplitude was 210–325 nM in myocytes from the HF model, i.e., values similar to those reported previously for myocytes from pacing-induced models of HF in dogs [37] (~200 nM, see also Fig. 7e) or cats [40] (~340 nM). Finally, from our experimental traces (examples in Fig. 3a, b), we found that during 1.5 Hz pulse trains RAN indeed suppresses the major portion of DCa accumulation, which in absolute terms of [Ca²⁺] ranges from 62 to 98 nM. Thus, both of our numerical simulations closely predict this experimental estimate for substantial DCa accumulation suppression by I_NaL inhibition.

Discussion

Using a combination of experimental and numerical modeling studies, we demonstrate at the single cell level that in chronic HF augmented activity of the late Na⁺ channels exerts both direct electrophysiological effects and indirect effects on intracellular Ca²⁺ concentration in ventricular myocytes. As discussed in detail below, our findings thus suggest a novel cellular and molecular mechanism contributing to rate-dependent [31, 32] impaired diastolic function (i.e., poor relaxation) and DAD-mediated triggered activity, two major abnormalities associated with chronic HF [2, 18, 41, 42]. As we noted in the Introduction, the previous suggestions about the I_NaL-mediated mechanism in HF were based on results obtained in normal myocardium using Anemonia sulcata toxin, ATX-II, to produce persistent I_Na [27, 28]. The role of I_NaL in these latter studies could be exaggerated because ATX-II caused a fivefold increase in the persistent I_Na (Fig. 4a in [28]), whereas studies in failing human and dog hearts showed I_NaL amplitude increases of 30–50% [8, 9].

Specific mechanisms to improve Ca²⁺ handling and heart contractility by inhibition of late Na⁺ current

Diastolic Ca²⁺ accumulation

Previous studies in animal models of HF showed that partial blockade of either total Na⁺ current by saxitoxin or I_NaL by RAN improved contractility of myocytes from failing hearts [6, 13]. The results of the present study support the following specific mechanisms of late Na⁺ current contribution to abnormal Ca²⁺ handling and impaired contraction in chronic HF. Increased I_NaL in HF alters AP and simultaneously provides a substantial systolic late Na⁺ influx (during AP plateau). These two effects combined likely cause or modulate Ca²⁺ influx via the reverse mode Na⁺/Ca²⁺ exchange. This additional I_NaL-dependent systolic Ca²⁺ influx contributes to major abnormalities of cell Ca²⁺ handling, including DCa accumulation (compare upper panels in Figs. 1b and 3b), Ca²⁺ alternans (Fig. 3a, upper panel), and SCaR (Fig. 4). While the DCa accumulation and Ca²⁺ alternans can directly impair contractions of individual myocytes, increased rate of SCaRs can further worsen the diastolic cardiac muscle function. Indeed, heterogeneity of diastolic [Ca²⁺]_i among cells within myocardial tissue caused by asynchronous SCaRs leads to heterogeneous myofilament activation, the summation of which produces a Ca²⁺-dependent component to diastolic tone [43].

The idea that I_NaL and its related Na⁺ influx cause a dynamic cell Ca²⁺ accumulation in HF is supported by the following results. We showed that partial blockade of I_NaL by RAN or TTX greatly improves Ca²⁺ handling in failing myocytes specifically; associated with inhibition of I_NaL, the DCa accumulation decreased (Fig. 3), alternans disappeared (Fig. 3a), and the probability of SCaRs decreased substantially (~threefold) (Fig. 4). This idea is also supported by recent studies in ventricular muscle strips isolated from end-stage failing human hearts [28] and in myocytes from dog failing hearts [13], which demonstrated that inhibition of I_NaL by RAN significantly reduces frequency-dependent increase in diastolic tension (i.e., diastolic dysfunction) by approximately 30% (in line with our results). Finally, inhibition of I_NaL and I_NaL-related Na⁺ influx in silico (both an AP clamp and a complete model of E-C coupling) substantially and selectively suppresses DCa accumulation of the myocytes of failing hearts (Figs. 7, 9).

Dynamic Ca²⁺ cycling abnormalities in failing heart; alternans and spontaneous Ca²⁺ releases

At a low stimulation rate of 0.25 Hz we also observed an abnormal tonic or dome component of CaT (compare upper panels in Figs. 1a, 2a, b) (first reported in heart ventricular muscle strips from patients with HF [44]). In addition we found some Ca²⁺ oscillations at low pacing rates (Fig. 2a upper panel) and beat-to-beat variations of Ca²⁺ transients at higher rates of stimulation (Fig. 3a, upper panel) that may be a sign of alternans of intracellular Ca²⁺ cycling [45]. TTX (middle panels in Fig. 2b) or RAN (middle panels in Fig. 2a) reversed these abnormalities.

The threshold for SCaR is lower in HF because of reduced SERCA function and increased SR Ca²⁺ leak [20, 21] in line with observation of SCaRs in our recordings in failing (Fig. 4) but not in normal myocytes (Fig. 1b). Diastolic SCaRs can initiate DADs via activation of the forward mode NCX inward current, and/or nonselective channels [18, 25]. Previous studies demonstrated that when SCaRs within a cardiomyocyte are sufficiently synchronized (e.g., multifocal Ca²⁺ waves), the resultant depolarization summates and can be sufficient to trigger a spontaneous AP [43]. The amplitude of SCaRs (F/F₀ ~1.55) is close to that of CaT during 1.5 Hz pacing (Fig. 4a), which could be pro-arrhythmic, especially in the context of up-regulated NCX in HF [16–18]. Indeed, it has been documented that the threshold amplitude of SCaRs that is sufficient to evoke DADs is almost twofold lower for HF myocytes (~280 nM) compared to that in normal heart (~512 nM) [42]. This threshold is close to the amplitude of SCaRs reported here (~250 nM) (Fig. 4a). Increased Ca²⁺ entry is an established mechanism for SCaR from the SR [46]. Thus, the DCa accumulation represents a reasonable mechanism for SCaR after the pulse train in our chronic HF model. The fact that TTX and/or RAN inhibited SCaRs without significantly affecting SR load and NCX function (Fig. 5) supports this mechanism (Fig. 4b, c). On the other hand, small (albeit insignificant) decrease in SR Ca²⁺ load in the presence of RAN (Fig. 5b) does not completely rule out some involvement of SR load in SCaR occurrence, given the non-linear behavior of the SR (i.e., a relatively small change in load triggers the release) [47]. Also elucidation of possible contributions of changes in AP shape and duration in the improvement of Ca²⁺ handling, caused by the I_NaL reduction shown both experimentally and in silico (Figs. 7, 9), merits consideration in future studies.

Predictions in silico

Our model simulations provide quantitative evidence that I_NaL accounts for a major portion of diastolic cell Ca²⁺ accumulation in myocytes (stimulated at rate of 1.5 Hz) measured experimentally. A simplified numerical modeling of I_NaL during AP clamp shows that RAN (10 μM) almost completely (by 77%) inhibits late Na⁺ influx (Fig. 8d). At the same time we also show that Na⁺ influx via I_NaL can be indeed substantial. Taking into account that NCX operates in the reverse mode during AP in HF [24, 48] and NCX function is enhanced in HF [16–18], we reasoned that a large portion of Na⁺ influx is likely to be the cause of an increase in the dynamic exchanged by the reverse mode NCX for Ca²⁺. It is known, for example, that the reverse mode NCX-mediated Ca²⁺ influx is indeed substantial and may even result in the direct activation of contraction in HF [48]. We estimated that late Na⁺ influx is sufficient to cause a 72 nM increase in cytosolic [Ca²⁺] via this mechanism. We also tested the effect of I_NaL inhibition in in silico simulations of the E-C coupling process including all substantial mechanisms involved in the regulation of the Ca²⁺ and Na⁺ homeostasis in HF (Fig. 9). Our model simulations closely reproduce our experimental results and confirm the importance of I_NaL for both the AP duration and the DCa accumulation in failing myocardium.

Physiological significance, clinical relevance

In a recent clinical trial RAN significantly reduced arrhythmias in patients with non ST-segment elevation acute coronary syndrome [49] pointing to a potential clinical relevance of I_NaL. The beneficial effect of RAN in these patients can be thus explained, in part, by two mechanisms: (1) improvement of repolarization shown in previous studies [13, 33] and in this study (see Fig. 6) and (2) Ca²⁺ handling associated with a decreased probability of spontaneous releases (shown in this study, Fig. 4) and, hence, DADs and their triggered APs (see previous section). Partial inhibition of I_NaL by RAN, TTX, or saxitoxin greatly improves repolarization and decreases intrinsic beat-to-beat AP duration variability of failing cardiomyocytes [7, 9, 13] (Fig. 6). The cellular mechanisms of RAN effects proposed in this study may explain, at least in part, beneficial RAN effects previously reported in whole-animal studies by our group and others. In dogs and rabbits RAN reduced infarct size and Ca²⁺ overload in response to a regional ischemia-reperfusion [50, 51]. In dogs with coronary microembolization-induced HF, acute intravenous administration of RAN improved LV systolic function and LV mechanical efficiency without increasing myocardial oxygen consumption and without any increase in heart rate or reduction of systemic blood pressure [52].

I_NaL as a novel therapeutic target

The emerging paradigm with regard to Na⁺ channels in HF is that I_NaT is decreased [8, 14, 15] but simultaneously I_NaL is increased [5]. Blockers of I_NaT are pro-arrhythmic in HF because they will further slow conduction, thus worsening impulse conduction [53] and thereby facilitating development of re-entry. Hence, transient and late Na⁺ currents must be treated differently. The new type of “smart” drugs should preferentially block I_NaL over I_NaT [5, 54, 55]. The potential benefits (preventing Ca²⁺ overload and arrhythmias) of the preferential I_NaL blockade can be also expected in hypoxia and ischemia, in which I_NaL increase and Na⁺-induced Ca²⁺ overload are major features. In ischemia, accumulation of toxic metabolite lysophosphatidylcholine and reactive oxygen species dramatically increases I_NaL [56–58]. Indeed Na⁺ channels are critically involved in this process because their blockers or activators reduce or increase Ca²⁺ overload in these pathological conditions, respectively [59, 60].

In conclusion, we provide evidence that I_NaL and its systolic Na⁺ influx contribute to the dynamic DCa accumulation and spontaneous release in ventricular myocytes from dog model of chronic HF. Therefore, the results of the present study support the idea that selective blockade of this Na⁺ current represents a plausible strategy to treat Ca²⁺-related diastolic dysfunction and arrhythmia in HF.