The human Nav1.5 F1486 deletion associated with long QT syndrome leads to impaired sodium channel inactivation and reduced lidocaine sensitivity

Abstract

The deletion of phenylalanine 1486 (F1486del) in the human cardiac voltage-gated sodium channel (hNav1.5) is associated with fatal long QT (LQT) syndrome. In this study we determined how F1486del impairs the functional properties of hNav1.5 and alters action potential firing in heterologous expression systems (human embryonic kidney (HEK) 293 cells) and their native cardiomyocyte background. Cells expressing hNav1.5-F1486del exhibited a loss-of-function alteration, reflected by an 80% reduction of peak current density, and several gain-of-function alterations, including reduced channel inactivation, enlarged window current, substantial augmentation of persistent late sodium current and an increase in ramp current. We also observed substantial action potential duration (APD) prolongation and prominent early afterdepolarizations (EADs) in neonatal cardiomyocytes expressing the F1486del channels, as well as in computer simulations of myocyte activity. In addition, lidocaine sensitivity was dramatically reduced, which probably contributed to the poor therapeutic outcome observed in the patient carrying the hNav1.5-F1486del mutation. Therefore, despite the significant reduction in peak current density, the F1486del mutation also leads to substantial gain-of-function alterations that are sufficient to cause APD prolongation and EADs, the predominant characteristic of LQTs. These data demonstrate that hNav1.5 mutations can have complex functional consequences and highlight the importance of identifying the specific molecular defect when evaluating potential treatments for individuals with prolonged QT intervals.

Key points

• We investigated how the F1486 deletion LQT3 mutation impairs the functional properties of the human cardiac voltage-gated sodium channel (hNav1.5) and alters action potential firing.

• Voltage-clamp recordings from HEK 293 cells and cardiomyocytes expressing recombinant channels demonstrated that the F1486del mutation reduces peak current density but also impairs inactivation and increases late current density.

• Current-clamp recordings from cardiomyocytes indicated that the increase in late current density would result in prolonged action potential duration and this was confirmed using computer simulations.

• The deletion of F1486 abolished the ability of lidocaine to stabilize the inactivated state and eliminated the high-affinity binding of lidocaine to inactivated channels.

• Our data show that the hNav1.5-F1486del mutation has complex functional consequences and indicate that knowledge of the specific molecular defect is critical when developing potential treatments for individuals with prolonged QT intervals.

Introduction

As lethal inherited cardiovascular disorders, congenital LQT syndromes (LQTs) have a hallmark of prolonged QTc (corrected QT interval) measured by surface electrocardiograph (ECG). Episodic ventricular tachycardia (VT) such as ventricular fibrillation and torsade de points can consequently be triggered by LQTs. So far, congenital LQTs have been linked to mutations in 12 different genes including SCN5A, the gene that encodes the cardiac sodium channel Nav1.5 (Hedley et al. 2009).

Nav1.5 is the major voltage-gated sodium channel in the heart (Rogart et al. 1989) and is essential for the initiation and propagation of action potentials throughout the heart. As with other types of voltage-gated sodium channels, Nav1.5 can undergo conformational changes between closed non-conducting, open ion-conducting, and inactivated non-conducting states in response to the changes in the local electrical potential across the cell membrane. Among the 12 types of LQTs (Hedley et al. 2009), gain-of-function SCN5A mutations (mutations that result in increased channel activity) underlie congenital LQT3 syndrome (Wang et al. 1995). These mutations increase late sodium current and are predicted to prolong the APD (Bankston et al. 2007). By contrast, with Brugada syndrome, which is characterized by coved-type ST segment elevation in the ECG and polymorphic ventricular tachycardia, the associated mutations in hNav1.5 are predominantly classified as loss-of-function mutations because they typically result in reduced sodium current densities (Amin et al. 2010). However, a few SCN5A mutations that underlie mixed disease phenotypes have been identified (Remme & Wilde, 2008; Makita, 2009).

The α subunit of the sodium channel consists of four structurally homologous domains (DI–DIV), each containing six transmembrane segments. The intracellular linker between domains III and IV plays an important role in channel fast inactivation (Supplemental Fig. S1). Within this region, an isoleucine–phenylalanine–methionine (IFM) motif was demonstrated to be critical for channel inactivation (West et al. 1992). Triple mutation of IFM to QQQ (glutamine) and single point mutation of F (phenylalanine) to Q have been shown to slow the inactivation of Nav1.2 (West et al. 1992), a brain sodium channel, and hNav1.5 (Bennett et al. 1995a; Grant et al. 2000). Recently, Yamamura and colleagues (2010) described an infant patient with a sodium channel mutation and severe cardiac arrhythmia that resulted in death in the first day of life. The infant exhibited atrioventricular block, ventricular tachycardia and, after administration of intravenous amiodarone, an extremely prolonged QT interval. Mutational analysis identified the de novo deletion of phenylalanine 1486 (F1486del) within the III–IV linker domain of hNav1.5 and it was concluded that the patient had fatal long QT syndrome (Yamamura et al. 2010). Although F1486 is part of the IFM motif, it is not known how specific deletion of this key phenylalanine residue alters the detailed biophysical properties of hNav1.5 or affects cardiac action potential generation.

Interestingly, the patient harbouring the F1486del mutation did not respond positively to lidocaine, raising the possibility that the hNav1.5 mutation reduced sensitivity to lidocaine (Yamamura et al. 2010). Previous studies have reported that while the IFM/QQQ triple mutation can substantially reduce the ability of lidocaine to induce use-dependent block of hNav1.5 (Bennett et al. 1995a), mutating only the F of the IFM motif to Q does not alter lidocaine sensitivity, even though this substitution substantially impairs sodium channel inactivation (Balser et al. 1996). Furthermore, the abnormal channel activity generated by the LQT3 ΔKPQ deletion, also in the III–IV linker, is preferentially inhibited by anti-arrhythmic drugs (Wang et al. 1997). Thus, because III–IV linker mutations can have variable impact on hNav1.5 lidocaine sensitivity and LQT3 mutations can have variable impact on anti-arrhythmic sensitivity (Ruan et al. 2007), it is difficult to predict the degree to which the F1486del mutation might increase or decrease lidocaine sensitivity.

In the present study we sought to determine the electrophysiological consequence of deleting F1486 and to investigate the role of this particular residue in contributing to lidocaine sensitivity. We used heterologous expression of hNav1.5 channels in HEK 293 cells and computer modelling to predict the impact of functional alterations on action potential generation. In addition to these standard techniques, we also overexpressed F1486del channels in cardiomyocytes in order to further characterize the biophysical properties and impact on action potential generation in their native cell background. This is the first report to study these properties following overexpression of a LQT3 mutation in transfected native cardiomyocytes.

Although disease-causing channelopathies are relatively rare, investigating the biophysical and functional consequence of these mutations increases our knowledge of the roles that sodium channels play in controlling cardiac cellular excitability. Our data show that despite substantially reducing hNav1.5 current density, the F1486del mutation also induces substantial gain-of-function effects that can underlie extremely prolonged action potential durations. Furthermore, our data show that although some LQT3 mutant channels may respond to lidocaine administration, others like F1486del can display reduced (or even essentially abolished) sensitivity to lidocaine.

Methods

Sodium channel constructs and mutagenesis

Heterologous expression constructs encoding the hNav1.5 channel (Gellens et al. 1992) were introduced into pcDNA3.1(+) with the CMV promoter. Mutagenesis primers were designed to introduce the correct base pair changes into wild-type (WT) hNav1.5 channel cDNA using Vector NTI software (Invitrogen). Site-directed mutagenesis was carried out using Quick Change XL site-directed mutagenesis kit (Stratagene) according to the manufacturer's protocol. The full-length sequence was analysed to verify the specific mutation at the target sites. A tetrodotoxin (TTX)-resistant version of hNav1.5 (C373S) was used to allow isolation and characterization of F1486del and WT currents in cardiomyocytes (hereafter referred to as ‘WT-R’ and ‘F1486del-R’). C373S endows the channel with higher resistance to TTX (K_i approximately 100 μm) (Leffler et al. 2005), but does not alter the gating properties of hNav1.5 channels.

HEK 293 cell culture and transient transfection of recombinant hNav1.5

HEK 293 cells were ordered from ATCC (Manassas, VA, USA). Use of HEK 293 cells was approved by the Institutional Biosafety Committee and conformed to the ethical guidelines for the National Institutes of Health for the use of human-derived cell lines. Cells were grown on 35 mm tissue culture dishes under standard tissue culture conditions (37°C; 5% CO₂) in Dulbecco's modified Eagle's medium containing high glucose (DMEM/H) supplemented with 10% fetal bovine serum (FBS). Equal amount of cDNA encoding human voltage-gated sodium channel Nav1.5 α subunit and auxiliary β₁ subunit were transiently cotransfected into HEK 293 cells using the calcium phosphate precipitation method. Plasmid DNA encoding N₂-EGFP (Clontech Laboratories Inc., CA, USA) was used as a transfection control in order to visualize the transfected cells. The mixture of calcium phosphate and DNA was added to the culture medium and maintained for 4–6 h, after which cells were washed with fresh culture medium and replated onto 12 mm glass coverslips in 24-well plates. Typically, patch-clamp recordings were performed 24–48 h after transfection.

Whole cell patch-clamp recordings on HEK 293 cells

Whole cell patch-clamp recordings were conducted at room temperature (∼22°C) using a HEKA EPC-10 amplifier as described previously (Jarecki et al. 2008). Data were collected using a computer and the Pulse program (v. 8.65, HEKA Electronic, Germany). Fire-polished glass electrodes were fabricated from capillary glasses (VWR International), which were pulled using a Sutter P-97 puller (Novato, CA, USA) to have a resistance of 0.9–1.4 MΩ. The standard external bathing solution for sodium current recording from HEK 293 cells contained (in mm): 140 NaCl, 1 MgCl₂, 3 KCl, 1 CaCl₂ and 10 Hepes. pH was adjusted to 7.3 with 1 n NaOH. The standard pipette solution for sodium current recording from HEK 293 cells contained (in mm): 140 CsF, 10 NaCl, 1.1 EGTA and 10 Hepes, pH 7.3 (adjusted with CsOH). Cells on glass coverslips were transferred to the recording chamber containing 300 μl of the external bathing solution. Series resistance was compensated to reduce the voltage error to less than 3 mV. Recordings were started 5 min after forming whole cell configuration to allow sufficient equilibration of the intracellular solution and the pipette solution. Lidocaine solution was added to the recording chamber by first withdrawing 30 μl of bath solution, and then adding 30 μl of 10-fold concentrated lidocaine and mixing 10 to 15 times using a 100 μl pipettor. Cells were kept in the external bathing solution for no more than 1 h.

Isolation and culture of neonatal rat cardiomyocytes

Neonatal Sprague–Dawley rat pups of postnatal day 1 (P1 pups) were obtained from Harlan Sprague Dawley Inc., Indianapolis, IN, USA). The use of animals was in compliance with the Guide for the Care and Use of Laboratory Animals published by the US National Institutes of Health and was approved by the Indiana University School of Medicine Animal Care and Use Committee (no. 0000003129-A1). Rat pups were killed by decapitation and hearts were immediately collected for enzymatic digestion. Briefly, hearts were taken out from the P1 rats and only ventricles were saved for the subsequent dissociation procedure. Heart tissues were digested in HBSS solution containing type II collagenase (0.5 mg ml⁻¹, Worthington) and pancreatine (0.4 mg ml⁻¹, Sigma). Non-myocyte contaminants were removed by pre-plating on 10 cm tissue culture plates and culturing in serum-containing medium for 2 h. Cardiomyocytes were cultured using DMEM/F12 media containing 10% fetal bovine serum, 100 units ml⁻¹ penicillin, 100 mg ml⁻¹ streptomycin and 1 mm l-glutamine. Following a 24 h incubation period, the serum-containing medium was replaced with a low serum medium to prevent hypertrophy.

Transfection of primary neonatal rat cardiomyocytes

hNav1.5 TTX-resistant ‘WT-R’ or ‘F1486del-R’ cDNA and pmaxGFP cDNA (at a ratio of 5:1) were transfected into neonatal rat cardiomyocytes using the Amaxa (Lonza, Switzerland) kit according to the manufacture's protocol. Briefly, 2 × 10⁶ cells were used for each transfection reaction. After transfection, cells were plated onto 12 mm glass coverslips coated with 0.1% gelatin for electrophysiology recordings.

Whole cell current-clamp recordings on cardiomyocytes

The Tyrode solution used for action potential studies contained (in mm): NaCl 126, KCl 5.4, Hepes 10, NaH₂PO₄ 0.33, MgCl₂ 1.0, CaCl₂ 1.8 and glucose 10. pH was adjusted with NaOH to 7.4. The pipette solution used in current-clamp studies contained (mm): KCl 20, potassium aspartate 110, MgCl₂ 1.0, Hepes 5.0, EGTA 10 and Na₂-ATP 5.0. pH was adjusted to 7.2 with KOH. Action potentials (APs) were elicited by 2 ms stimulus pulses at a frequency of 1 kHz. AP measurements were started 5 min after establishing the whole cell configuration. In this study, the early phase of the action potential repolarization was indicated by APD₃₀ (30% of full repolarization), while the steady- state repolarization of the action potential repolarization was indicated by APD₅₀, APD₇₀ and APD₉₀ (50, 70 and 90% of full action potential repolarization). The data sampling rate varied from 2 to 2.5 kHz. Cells generating oscillations in the membrane potential during the plateau phase of the action potential or during the terminal repolarization phase were defined to have early afterdepolarizations (EADs) (Volders et al. 2000).

Whole cell voltage-clamp recordings on cardiomyocytes

The bath solution contained (mm): 120 NaCl, 1.2 MgCl₂, 1.5 CaCl₂, 10 Hepes, 10 tetraethylammoniumchloride, 5.0 sucrose, 5.0 glucose and 5.0 CoCl₂, pH 7.4 with NaOH. The pipette solution contained: 120 cesium aspartate, 5.0 NaCl, 2.0 MgCl₂, 10 Hepes and 10 EGTA. pH was adjusted to 7.3 with CsOH. TTX was added to the external solution at a final concentration of 20 μm to eliminate the contamination of endogenous sodium currents in neonatal cardiomyocytes. Pipette resistance was 1.5–2.5 MΩ when filled with the internal solution.

Drugs

Lidocaine (Sigma Aldrich) and tetrodotoxin (Alomone Lab) were used in the current study. Lidocaine hydrochloride monohydrate was dissolved in the external recording solution to make a stock solution. The stock solution was prepared shortly before the experiment and pH was adjusted to 7.3 using NaOH. Citrate-free tetrodotoxin powder was dissolved in 50 mm acetic acid solution to make the stock solution. Subsequent dilutions were performed using the standard external bathing solution.

Cardiac myocyte simulation

We modified mathematical models of cardiac AP firing (Luo & Rudy, 1994; Courtemanche et al. 1998) to simulate the impact of the F1486L LQT3/SIDS mutation. The cardiac myocyte model was previously implemented by Ingemar Jacobson in the NEURON simulation environment (Hines & Carnevale, 1997) and is available at http://senselab.med.yale.edu/ModelDB/ShowModel.asp?model=3800. The only changes made to the model were to the voltage-dependent sodium current (Naf in the original model). Simulations were run with: (1) 100% hNav1.5 current density, (2) 50% hNav1.5 current density, (3) 50% hNav1.5 and 12.5% hNav1.5-F1486del current density, (4) 50% hNav1.5 and 10% hNav1.5-F1486del current density and (5) 50% hNav1.5 and 7.5% hNav1.5-F1486del current density.

We used a Markov model of Nav1.5 (Jarecki et al. 2010) that was based on the Hodgkin–Huxley formulation of Nav1.5 in the original cardiac model of AP firing (Luo & Rudy, 1994; Courtemanche et al. 1998). The Markov formulation is more amenable to implementation of the F1486del effects. Slow inactivation states were included in this formulation. The diagram for the Markov model used for the simulated voltage-gated sodium conductances is shown in Supplemental Fig. S2 and the transition rate expressions for hNav1.5 and F1486del currents are provided in Supplemental Table S1.

Data analysis

Voltage-clamp data were analysed using the Pulsefit (v. 8.65, HEKA electronic), Microsoft Excel and Prism (GraphPad Software Inc.) software programs. Normalized conductance–voltage relationships (G–V) were derived using the function:

(1)

G_Na is conductance of sodium channel, I_max is the peak current density in response to the test pulse, V_m is the test pulse potential and E_Na is the measured sodium channel equilibrium potential. The conductance of activation, steady-state fast inactivation (h_∞) and slow inactivation curves were fitted using the general Boltzmann function:

Slope factors and V_1/2 were also calculated based on Boltzmann function analysis. Data are presented as mean ± SEM. n is presented as the number of the separate experimental cells. The equations used for the concentration–inhibition relationships for lidocaine and the Nav channels were as follows:

(2)

Statistical data analysis

Data are presented as mean ± SEM. Statistical difference was assessed using either a Student's unpaired t test or a one-way ANOVA followed by the Bonferroni post hoc analysis to compare selected pairs of columns as appropriate. P≤ 0.05 was considered statistically significant.

Results

Deletion of F1486 within intracellular loop between DIII–DIV of hNav1.5 reduced peak current density but augmented late sodium current

The initial comparison of current traces recorded from HEK 293 cells revealed striking differences both in current amplitudes and kinetics between wild-type (WT) and F1486del channels. Currents generated by F1486del channels were significantly smaller (Fig. 1A; also see Fig. 1E for the protocol). The averaged peak current density at −20 mV was 514.2 ± 76.3 pA pF⁻¹ (n = 14) for WT channels and 100.2 ± 24.1 pA pF⁻¹ (n = 14) for F1486del channels (P < 0.001, Fig. 1B). This observation was somewhat surprising as LQT3 mutations are typically reported to be associated with gain-of-function, not loss-of-function effects (Wang et al. 2007; Amin et al. 2010; Wilde & Brugada, 2011). However, Nav1.5 channels with the IFM/QQQ substitution that severely impairs inactivation are known to express poorly in HEK 293 cells. This can be rescued in part by co-expression of potassium channels that hyperpolarize the membrane potential of HEK 293 cells, suggesting that the reduced current density observed with some inactivation-deficient channels in HEK 293 cells might result from the relatively depolarized resting membrane potential of these cells (Grant et al. 2000).

Figure 1. The sodium channel F1486del causes decreased peak current density.

A, representative sodium current traces recorded from HEK 293 cells elicited using an incremental depolarizing protocol. Both channels were co-expressed with human auxiliary β₁ subunits. B, mean current–voltage relationships in response to a series of 50 ms stimulation pulses from −80 mV to +60 mV (5 mV increment) for channels expressed in HEK 293 cells (WT n = 14, filled squares; F1486del n = 14, open circles). C, representative current traces recorded from neonatal rat cardiomyocytes (WT-R n = 12, filled squares; F1486del-R n = 8, open circles). D, mean current–voltage relationships for cardiomyocytes in response to the same stimulation as in B. E, test protocols used for current density measurements.

To investigate whether the decrease in current density might be an artifact that was caused by the heterologous expression system, we also expressed the channel in neonatal rat cardiomyocytes. In order to be able to isolate the current generated by recombinant channels from endogenous myocyte sodium currents, the recombinant channels were made highly resistant to TTX (using the C373S substitution) and the endogenous currents were blocked with 20 μm TTX (Supplemental Fig. S3). The modified TTX-resistant channels are henceforth referred to as WT-R and F1486del-R. It should be noted that the C373S mutation did not alter the voltage dependence of activation and inactivation; WT and WT-R channels expressed in HEK 293 cells exhibited similar voltage-dependent properties (Supplemental Fig. S4 and Table S2). It should also be noted that we used an aspartate-based intracellular solution with the cardiac myocytes, rather than the fluoride-based intracellular solution used with HEK 293 cells, to avoid the complications of fluoride-based intracellular solutions (Meadows et al. 2002).

We found that F1486del-R channels expressed in cardiac myocytes have significantly reduced current density compared with WT-R channels (Fig. 1C and D). At −20 mV, cardiomyocytes expressing the WT-R channels had a current density of 232.0 ± 52.0 pA pF⁻¹ (n = 12), while the F1486del-R channels carried a current density of only 64.4 ± 14.8 pA pF⁻¹ (n = 8, P < 0.05, Fig. 1D). Thus, the reduced current density is probably not an artifact of the HEK 293 heterologous expression system.

We next examined the effect of the F1486del mutation on channel kinetics. Sodium current (I_Na) recorded from WT and WT-R channels activated and decayed rapidly (Fig. 1A and C). In contrast, the F1486del and F1486del-R channels had incomplete decay (Fig. 1B and D). We quantified the late I_Na by normalizing the current remaining at the end of 200 ms depolarizing pulses to the peak amplitude derived from the same depolarizing stimulus. At the voltage of −10 mV, only 0.7% of late I_Na remained in HEK 293 cells expressing WT channels, and only 0.2% remained in cardiomyocytes expressing WT-R channels. In comparison, cells expressing the F1486del and F1486del-R channels exhibited substantial late currents (∼20.7% late I_Na remained in HEK 293 cells; 19.8% late I_Na in cardiomyocytes) (Fig. 2A and C). At each test voltage, the remaining late I_Na was significantly greater in cells expressing F1486del channels (Fig. 2B and D). Furthermore, not only was the relative amount of late current increased by the deletion of F1486, but also the absolute amplitude of the late current was greater for F1486del channels than for WT channels. At 0 mV, the density of the late current in HEK 293 cells was 1.6 ± 4.5 pA pF⁻¹ for WT channels and 20.4 ± 3.3 pA pF⁻¹ for F1486del channels (P < 0.01), and in neonatal cardiomyocytes the late current density was 2.3 ± 0.6 pA pF⁻¹ for WT-R channels and 11.7 ± 2.5 pA pF⁻¹ for F1486del-R channels (P < 0.01). Thus, despite the decrease in peak current density induced by the F1486del mutation, the late current density was significantly increased.

Figure 2. Deletion of F1486 causes increased late I_Na.

A, representative traces of sodium current recorded at −10 mV from HEK 293 cells. Lines indicate the averaged raw data from F1486del and WT channels. Augmented late I_Na was shown by normalizing the trace recorded from F1486del channels to the trace recorded from WT channels. B, bar graph indicates the summarized percentage of late I_Na (normalized to the peak current amplitude) in HEK 293 cells recorded at each tested voltage. *P < 0.0001, F1486del (n = 12) vs. WT (n = 6). C, representative traces of averaged sodium current recorded at −10 mV from neonatal cardiomyocytes in the presence of 20 μm TTX. The traces for TTX-resistant hNav1.5 (WT-R) and F1486del-R channels (see Methods) are labelled. The normalized data from F1486del-R channels are also shown. D, bar graph indicates averaged late I_Na recorded from neonatal rat cardiomyocytes in the presence of 20 μm TTX. *P < 0.0001, F1486del-R (n = 7) vs. WT-R (n = 6).

Deletion of F1486 alters the voltage dependence of inactivation but not activation

The IFM motif has been demonstrated to be critical for channel inactivation in the neuronal sodium channel Nav1.2 (West et al. 1992). Therefore, we tested whether deletion of F1486 on hNav1.5 could alter the voltage dependence of channel inactivation. We observed a depolarizing shift of the steady-state fast inactivation curve and the V_1/2 for F1486del channels expressed in HEK 293 cells and F1486del-R channels expressed in cardiac myocytes (Fig. 3C and E; Supplemental Table S3). The shift in the voltage dependence of inactivation, along with the incomplete inactivation, is consistent with the proposal that the IFM motif is a crucial determinant of fast inactivation in voltage-gated sodium channels.

Figure 3. Deletion of F1486 does not affect the voltage dependence of Na⁺ channel activation but causes a depolarizing shift of the voltage dependence of inactivation and increases window current.

A, mean steady-state activation curves generated under the 50 ms stimulation pulses from −80 mV to +40 mV with 5 mV increments for WT (n = 14, open circles) and F1486del channels (n = 13, filled circles) expressed in HEK 293 cells. Both curves were fitted with the Boltzmann function. Data indicate that F1486del does not affect the voltage dependence of activation. B, mean steady-state activation curve generated from neonatal rat cardiomyocytes using the same protocol as in A (WT-R n = 12, open circles; F1486del-R n = 8, filled circles). There is no statistical difference observed for V_1/2 between these two groups. C, the steady-state fast inactivation curve of F1486del channels (filled squares, n = 16) expressed in HEK 293 cells is significantly shifted toward more positive potentials compared with that of WT channel (open squares, n = 11). Cells were held at −100 mV and stepped to an inactivating prepulse (−150 mV to −10 mV) for 500 ms. The channels that remain available after each inactivating prepulse were evaluated by the peak current produced during a test pulse to −10 mV for 20 ms. D, steady-state fast inactivation curve from neonatal cardiomyocytes expressing TTX-resistant hNav1.5 channels (WT-R or F1486del-R, see Methods). Cells were applied with the same protocol as in C. F1486del-R shift the voltage dependence of inactivation to the right. E and F highlight the overlap of the activation and steady-state inactivation for HEK 293 cells and cardiomyocytes expressing the recombinant channels, respectively. The overlap indicates the region where generation of window currents are predicted. The grey area indicates the window current for F1486del (E) and F1486del-R (F) channels, while the black area indicates the window current for WT (E) and WT-R (F) channels. Both F1486del and F1486del-R channels are predicted to generate substantially enhanced window currents compared with WT and WT-R channels due to the greater overlap of the activation and steady-state inactivation curves.

Next, we characterized the voltage dependence of activation for the WT and F1486del channels. Conductance–voltage (G–V) curve analysis revealed similar activation profiles (Fig. 3A and B). The macroscopic voltage dependence of activation was not statistically different (Fig. 3A and B, Supplemental Table S3). These results are consistent with the previous studies showing that mutation of amino acids within the IFM motif does not alter the voltage dependence of channel activation in Xenopus oocytes (West et al. 1992).

Although the voltage dependence of activation and inactivation differed between channels transfected into HEK 293 cells and cardiac myocytes (due in part to the different pipette solutions used), our data clearly show that F1486del shifts the voltage dependence of inactivation, but not activation, in both HEK 293 cells and cardiac myocytes. As a result, there is a larger overlap of the voltage dependence of activation and the voltage dependence of fast-inactivation curves for F1486del channels than for WT channels (Fig. 3E and F). Based on this, we predict a marked increase in window currents for the F1486del channel. To examine this possibility, we used a slow depolarizing ramp stimulus from −100 mV to +40 mV. In response to the ramp depolarization, WT channels carried inward sodium current that was only observed over a narrow voltage range (Fig. 4A and C). By contrast, the F1486del channels conducted ramp currents over a broader range of voltages (Fig. 4A and C). This is consistent with the voltage range where window currents are predicted to occur. The peak ramp current generated by the F1486del channel increased by 5.24-fold in HEK 293 cells and 1.48-fold in cardiomyocytes compared with WT channels (Fig. 4B and D). In addition, the F1486del channels did not completely inactivate at the end of the voltage ramp (Fig. 4A and C). This increased ramp current might reflect a combination of window currents and the incomplete inactivation that contributes to the increase of late I_Na.

Figure 4. The F1486del increases ramp current in HEK 293 and neonatal rat cardiomyocytes.

A, averaged inward ramp current traces for WT (grey, n = 10) and F1486del (black, n = 10) in response to a slow depolarizing (0.28 mV ms⁻¹) ramp protocol from a holding potential of –120 mV to +40 mV during a 500 ms stimulation. Ramp current amplitude is presented as a percentage of the peak transient current elicited with a standard I–V protocol, then yielding the percentage of peak current for each recording. The presented traces demonstrate that F1486del causes enhanced inward ramp current in HEK 293 cells. B, bar graph interpretation of the percentage of peak ramp current to the transient peak current elicited with a standard I–V protocol (WT, n = 10; F1486del, n = 10) in HEK 293 cells. *P < 0.0001, WT vs. F1486del group. C, averaged ramp current recorded in neonatal cardiomyocytes expressing the recombinant channels in the presence of TTX. Traces were compiled from 4–6 individual recordings and then plotted vs. time course. D, bar graph interpretation of the percentage of peak ramp current to the transient peak current in cardiomyocytes (WT-R, n = 6; F1486del-R, n = 4). *P < 0.05 WT-R vs. F1486del-R group.

Deletion of F1486 from hNav.15 results in APD prolongation and EAD in neonatal rat cardiomyocytes and computer simulations

The patient with the F1486del mutation identified by Yamamura and colleagues presented a severe cardiac arrhythmia with an extreme prolonged QTc interval (as long as 860 ms) (Yamamura et al. 2010). The slow inactivating component of the sodium current, late I_Na, is a depolarizing current that can increase the ventricular APD (Kiyosue & Arita, 1989; Bennett et al. 1995b). However, mutations that substantially reduce peak current amplitude are typically associated with Brugada syndrome and generally do not prolong the QT interval (Amin et al. 2010). Based on our experimental results showing that deletion of F1486 reduces peak current density but also enhances late current density and shifts the voltage dependence of inactivation in the depolarizing direction in both HEK 293 cells and cardiac myocytes, it is unclear how F1486del channels would alter the cardiac action potentials. Therefore we chose to study the effects of the mutant channel on the action potential profile following expression in neonatal rat cardiomyocytes. As the proband carried the F1486del mutation on only one allele, we compared the impact of the overexpressed WT and F1486del channels on action potential properties in the presence of the endogenous sodium channels. Representative action potential traces are shown in Fig. 5A. Cardiomyocytes carrying the WT channels had an averaged APD of 88.9 ms. This is consistent with several previous studies in neonatal rat cardiomyocytes (Craelius et al. 1990; Gaughan et al. 1998). By contrast, cells expressing F1486del channels had a significant APD prolongation (Fig. 5B, averaged APD 1001.0 ms, P < 0.0001 F1486del vs. WT). During each stage of the action potential repolarization phase, there were significant differences between the cells expressing WT and F1486del channels (Fig. 5C). An increase in late I_Na has been shown to be followed by sequential appearance of EADs (Song et al. 2008), afterdepolarizations that usually happen during the early phase of the action potential repolarization. We were interested in exploring whether increased late I_Na induced by the F1486del channels might facilitate the occurrence of EADs, which are commonly associated with bradyarrhythmia-triggered arrhythmic activity (Clancy et al. 2002). In cardiomyocytes expressing the F1486del channel, we observed spontaneous EADs (Fig. 5E and F) in 33.3% of the cells we studied (4/12). By contrast, cells expressing the WT channel exhibited normal action potential profiles, without observable EADs (0/22) (Fig. 5D). We predict that the increased late I_Na results in the APD prolongation and induces the occurrence of EADs, which will lead to the subsequent bradycardia-induced arrhythmia.

Figure 5. APD was significantly prolonged in neonatal rat cardiomyocytes expressing the F1486del-R channels.

A, representative action potential traces for cardiomyocytes expressing WT-R and F1486del-R channels in the absence of TTX. B, averaged APD measurement. Cardiomyocytes expressing the WT-R channel have an averaged APD of 88.87 ± 28.63 ms (n = 12), whereas the averaged APD for cells expressing the F1486del-R channel is 1001 ± 167.1 (n = 11, *P < 0.0001). C, averaged APDs (APD at 20, 30, 50, 70 and 90% repolarization – APD20, APD30, APD50 and APD90, respectively) indicate that cells expressing the mutant channel take significantly longer to repolarize to the resting membrane potential. Data are mean ± SEM. *P < 0.001, F1486del-R (n = 7) vs. WT-R (n = 15) and non-transfected cells (n = 5). Data were analysed by one-way ANOVA. D, a representative recording from a neonatal cardiomyocyte expressing the WT-R channel. One single current injection was given to the cell for 2.5 ms. There was no EAD. E and F, representative traces from two individual cells expressing the F1486del-R channels. Cells present not only prolonged APD, but also prominent EADs.

We also conducted computer simulations of myocyte electrical activity as an alternative strategy to examine the probable impact of disease mutations on cellular excitability. Although computer modelling can have limitations (e.g. the model may not accurately reproduce the full array of ionic currents in myocytes), it allows the levels of WT and F1486del channels to be precisely controlled. We used an existing model of cardiac myocyte excitability (Courtemanche et al. 1998) that had been implemented in the NEURON simulation environment (Hines & Carnevale, 1997). By modifying the sodium channel formulation, we explored the ability of the F1486del mutation to alter action potential properties. The simulated currents are shown in Fig. 6A. Note that the kinetics are faster than those shown in Fig. 1 for the actual recombinant currents because the simulated currents are generated with the simulation temperature set at 37°C and the currents in Fig. 1 were collected at room temperature. Simulations were run with 100% WT current and with mixtures of WT and F1486del currents. For the action potential train shown in Fig. 6B, the mixed simulation contained WT currents at 50% of normal current density and F1486del currents at one-fifth of this (10% of normal WT current density). This ratio reflects the relative current density levels obtained from expressing WT and F1486del channels in HEK 293 cells and cardiomyocytes (∼5:1; see Fig. 1). Note that the action potentials in the model containing the mixture of WT and F1486del currents exhibited prolonged ADPs, sometimes with EADs (Fig. 6B, lower traces). Figure 6C compares single action potentials from four separate simulations. When the model cell contained either 100% WT current or just 50% WT current, the APD was ∼280 ms. In the simulation with 50% WT current and 7.5% F1486del current the APD was 550 ms. In the simulation with 50% WT current and 12.5% F1486del current the ADP was 980 ms with a pronounced EAD. If the model cell was paced at 1 s intervals, the simulation with 7.5% F1486del current never generated an EAD but the simulation with 12.5% F1486del current consistently generated EADs. These data demonstrate that the gain-of-function alterations induced by the F1486del mutation are sufficient to substantially prolong the APD despite the associated reduction in current density.

Figure 6. Computer simulations illustrate the impact of F1486del mutations on cardiomyocyte excitability.

A, simulated WT hNav1.5 (left panel) and F1486del (right panel) currents elicited by step depolarizations. The F1486del current density was set to be equal to the peak WT current density for comparison of kinetic differences. B, simulated action potentials from a modelled cardiac myocyte. The top panel shows action potentials from a model cell containing only WT conductance set at 100% of the peak conductance. The bottom panel shows action potentials from a model cell containing 50% of the WT peak conductance and F1486del channels set at 10% of the peak conductance (to mimic the 5:1 current density ratio observed in voltage-clamp experiments). Action potentials were paced at 1 Hz in these simulations. C, single action potentials elicited in the model cell containing different amounts of WT and F1486del conductances. No difference was observed between simulations with only WT channels set at 100% (black trace) and 50% (blue trace) of the peak conductance (note: the traces mostly overlap). By contrast, in simulations containing 50% of the WT peak conductance and F1486del channels set at either 7.5% (red trace) or 12.5% (green trace) of the peak conductance the action potentials were substantially prolonged. With the higher level of F1486del channels (blue trace) pronounced early after depolarizations were consistently produced.

F1486del leads to reduced response to the anti-arrhythmic drug lidocaine

As a local anaesthetic and anti-arrhythmic drug, lidocaine interferes with impulse conduction and action potential generation by binding to the inner pore of the voltage-gated sodium channels, and blocking the sodium current (Hille, 1977, 2001; McNulty et al. 2007). Experimental evidence suggests that lidocaine can bind to sodium channels in both resting and inactivated states, but preferentially binds to the channels in an inactivated state (Bennett et al. 1995a; Balser et al. 1996). Indeed, it has been proposed that lidocaine is able to preferentially block late currents associated with LQT3 mutants due to its ability to stabilize the inactivated state (Dumaine & Kirsch, 1998). The poor response of the infant patient carrying F1486del channels to lidocaine led us to ask whether deletion of F1486 alters lidocaine sensitivity. In order to examine the ability of lidocaine to inhibit resting channels, HEK 293 cells expressing WT and F1486del hNav1.5 channels were held at −120 mV for 10 s (to allow all channels to move to the resting state) and then depolarized to 0 mV for 20 ms to elicit the current. Different concentrations of lidocaine were applied to test its inhibitory effect. Peak current amplitude was evaluated before and after the application of lidocaine hydrochloride (10–3000 μm; n = 3–6 cells). The IC₅₀ was determined by fitting the concentration–response curve (Fig. 7A). The resting IC₅₀ was 96.16 μm for WT channels and 526.02 μm for F1486del channels, indicating that deletion of F1486 decreases the sensitivity of the resting channel to lidocaine. Figure 7B shows representative current traces for WT and F1486del channels before and after application of 100 μm lidocaine obtained with the protocol used to test the inhibitory effect of lidocaine on resting channels. We next compared the inhibitory effects of lidocaine on inactivated channels. The ability of lidocaine to interact with inactivated channels was examined by depolarizing cells expressing hNav1.5 channels to −50 mV for 10 s (to allow channels to move into inactivated states) and then stepping the voltage back to −120 mV for 100 ms (allowing channels without drug bound to recover from fast inactivation). Finally, cells were depolarized to 0 mV for 20 ms to elicit currents and channel availability as measured. Peak current was measured before and after application of lidocaine (10–1000 μm; n = 3–8). Fitting the concentration–response curve indicated that the IC₅₀ for lidocaine on inactivated WT channel is 12.50 μm. By contrast, the mutant channel exhibited a flat concentration–response curve (Fig. 7C), indicating that deletion of F1486 eliminates the high-affinity binding of lidocaine to inactivated channels and disrupts the ability of lidocaine to stabilize the inactivated state. Figure 7D shows representative current traces obtained with the protocol used to test the inhibitory effect of lidocaine on channels in the inactivated state. For comparison, we also examined the impact on lidocaine sensitivity of another LQT3 mutation, F1486L. While F1486L channels did show slightly reduced resting affinity for lidocaine, the lidocaine sensitivity of inactivated F1486L channels was similar to that of WT channels (Supplemental Fig. S5). Based on these data, it seems likely that the poor cardiac response of the patient to lidocaine resulted, at least in part, from the deletion of F1486 on hNav1.5.

Figure 7. F1486del dramatically reduces pharmacological effects of lidocaine on hNav1.5 channels expressed in HEK cells.

A, concentration–response curve for the inhibitory effect of lidocaine on resting hNav1.5 WT and F1486del channels. B, representative traces recorded when cells expressing the WT and F1486del channels were treated with 100 μm lidocaine under the protocol used for testing the inhibitory effect of lidocaine on resting-state channels. C, concentration–response curve for the inhibitory effect of lidocaine on inactivated hNav1.5 channels. Only the data recorded from WT channel can be fitted with a concentration–response curve. The dotted line manually connecting the F1486del data serves as the visual guide. D, representative traces from cells expressing WT and F1486del channels in the presence of 300 μm lidocaine under the protocol for testing the effect of lidocaine on inactivated channels.

Discussion

As an inherited cardiac disorder, congenital LQTs are characterized by the prolongation of the QTc interval. Mutations of SCN5A are associated with LQT3, Brugada syndrome, atrial fibrillation and sudden infant death (Chen et al. 1998; Wang et al. 2007; Makiyama et al. 2008). Our current study follows up on a report by Yamamura and colleagues regarding an infant patient with severe cardiac arrhythmia who was identified as carrying a heterozygous deletion of F1486 in the cardiac sodium channel hNav1.5 (Yamamura et al. 2010). ECG recordings from this patient indicated QTc intervals as long as 860 ms, 2:1 atria-ventricular block and polymorphic VT. Surprisingly, the patient's VT was refractory to intravenous administration of lidocaine. The mutation site is intriguing because F1486 is within the IFM motif which is crucial for channel inactivation (Vassilev et al. 1988; Stuhmer et al. 1989; Patton et al. 1992). We wanted to test the hypothesis that deletion of F1486 in hNav1.5 would alter functional properties of the channel and prolong the APD, thus contributing to the tachycardia observed in the patient. In addition, we also speculated that the decreased sensitivity to the anti-arrhythmic drug lidocaine might result from the deletion of F1486.

We found that HEK 293 cells expressing F1486del channels had substantially reduced peak current density compared with HEK 293 cells expressing WT channels. Sodium channel constructs with impaired inactivation often (Grant et al. 2000), but not always (Wang et al. 2004), exhibit reduced sodium current densities when expressed in HEK 293 cells. Therefore, in order to investigate whether this reduction in current density might be an artifact of the expression system, we also determined the biophysical consequences of the F1486del mutation in neonatal rat cardiomyocytes. Our data show that the F1486del-R channels also exhibited substantially reduced peak sodium current density in myocytes compared with WT-R hNav1.5 channels expressed in myocytes. This was somewhat surprising as substantial reductions in current density are typically associated with Brugada syndrome (Amin et al. 2010; Wilde & Brugada, 2011), but the patient with the F1486del mutation was diagnosed as having fatal long QT syndrome. Although in the current study we did not determine the mechanism for the decreased current density, Brugada mutations can decrease current density through multiple mechanisms (Amin et al. 2010), including enhanced slow inactivation and impaired channel trafficking. Interestingly, the SCN5A F1473S mutation, within the same intracellular loop as F1486del, has been reported to cause a channel trafficking defect (Ruan et al. 2010). By contrast, the F1486L mutation did not alter current density in HEK 293 cells (Wang et al. 2007), even though F1486L occurs at the same position as the F1486del mutation and both mutations are associated with long QT intervals and sudden infant death.

Our data demonstrate that the deletion of F1486 does not affect steady-state activation and these data are consistent with the previous finding demonstrating that activation was not affected by other mutations within the III–IV linker region (Vassilev et al. 1988; Bankston et al. 2007). The III–IV linker, along with the intracellular C-terminal domain, is important for inactivation (Vassilev et al. 1988; Kass, 2006) and mutations associated with LQT3 and sudden infant death generally disrupt the voltage-dependent transition to the inactivated state (Wang et al. 2007; Amin et al. 2010). In our study, the F1486del mutation caused a rightward depolarizing shift of the voltage dependence of steady-state inactivation, suggesting that deletion of F1486 can destabilize the inactivated channel configuration and increase the percentage of channels that retain the capability to open (Wang et al. 1996). We also observed increased persistent, or so called ‘late’I_Na, with the F1486del channels. The percentage of late I_Na persisting at 200 ms for the WT channel was less than 1%, consistent with previous work (Saint et al. 1992). By contrast, our data from both HEK 293 cells and cardiomyocytes clearly demonstrate a ∼20% increase of late I_Na in cells expressing the F1486del channels. It is interesting to note that this increase of late I_Na is substantially greater than what is typically observed with previously described LQT3 mutations (Wang et al. 1996; Bankston et al. 2007). For example, the ΔKPQ mutation only increases the late I_Na to ∼4% of the peak transient current (An et al. 1996) and in a study of seven distinct missense variants identified in a Norwegian sudden infant death syndrome cohort the largest persistent current was 1.7% of the peak transient current (Wang et al. 2007).

As an inward depolarizing current, late I_Na can increase the APD of ventricular cardiomyocytes, facilitate the onset of EADs and cause sustained triggered activity (Song et al. 2008). However, because the F1486del mutation also substantially decreased peak current density, it was not initially clear if the relative increase in late current amplitude was sufficient to increase the duration of the cardiac action potential. Therefore we examined this question using over-expression of recombinant channels in cultured rat neonatal myocytes as well as computer simulations of myocyte electrical activity. Each approach has specific advantages and disadvantages. Although it is difficult to precisely control the current amplitudes when over-expressing channels in actual cardiomyocytes, our current-clamp recordings from these cells demonstrate that while over-expression of hNav1.5 WT-R channels does not prolong APD or induce EADs, over-expression of hNav1.5-F1485del-R channels substantially prolonged the action potential and even induced what appeared to be EAD events in some myocytes. In contrast, while computer modelling necessitates that numerous assumptions are made regarding ionic conductances, it does allow for precise control of current amplitudes. We used an established myocyte model and observed that introducing the F1486del current along with WT current at a ratio of 1:5 was sufficient to substantially prolong the APD in simulated myocytes. Slightly higher levels of F1486del current in the model cell consistently generated not only prolonged APDs but also pronounced EADs. Together these data indicate that, despite the loss of peak amplitude associated with the F1486del mutation, the large relative increase in late current is sufficient to substantially increase action potential duration. This is consistent with the severely prolonged QT segment observed in the patient with the F1486del mutation. Although mutations associated with SCN5A are often classified as being associated with distinct clinical syndromes (e.g. LQT3 syndrome, Brugada syndrome, conduction defect), increasing evidence has indicated that some SCN5A mutations can underlie mixed disease phenotypes (Remme & Wilde, 2008; Makita, 2009). Our data suggest that the F1486del mutation may belong to the disease entity that some have called the ‘overlap syndrome of cardiac sodium channelopathy’ (Remme et al. 2008).

The ventricular tachycardia in the patient with the F1486del mutation was refractory to treatment with intravenous lidocaine (Yamamura et al. 2010). Lidocaine, widely used as an anti-arrhythmic drug, inhibits Na⁺ channels with voltage- and frequency-dependent properties. We tested the hypothesis that the F1486del mutation may contribute to the lack of response to lidocaine. We show that lidocaine was potent at inhibiting inactivated WT channels, which is consistent with the previous findings demonstrating that lidocaine preferentially binds to the sodium channel in the inactivated state, and inhibits the channel (Butterworth & Strichartz, 1990; Ragsdale et al. 1996). By contrast, we found that lidocaine has a significantly smaller effect on the F1486del mutant channels in the resting state and virtually no additional inhibition of inactivated channels. This is very different from what has been observed with previously described long QT mutations. For example, inactivated ΔKPQ channels, along with N1325S and R1644H LQT3 mutant channels, have been reported to have similar sensitivities to lidocaine as inactivated WT channels (An et al. 1996; Dumaine & Kirsch, 1998). Since the F1486L mutation induces a mild impairment of inactivation and does not alter sensitivity of inactivated channels to lidocaine, our data suggest that lidocaine sensitivity may not be significantly affected by mutations that moderately impair inactivation but can be substantially reduced by mutations that severely impair inactivation. Indeed, as previously mentioned, others have shown that while the mutation of F1486 to Q did not alter the peak current concentration–response curve for lidocaine (Balser et al. 1996), the IFM/QQQ triple mutation substantially reduced the ability of lidocaine to induce use-dependent block of hNav1.5 (Bennett et al. 1995a). However, not all sodium channel mutations that produce inactivation-deficient channels seem to affect lidocaine sensitivity. A set of three mutations in the S6 region of domain I in rat Nav1.4 (L435W/L437C/A438W) have been identified that severely impair inactivation but do not impact high-affinity lidocaine binding to the sodium channels (Wang et al. 2004). These findings suggest that both the degree to which inactivation is impaired and the mechanism by which it is impaired are important in determining the extent to which lidocaine sensitivity is affected. Although much has been learned over the last several decades regarding the molecular interactions between lidocaine and different conformations of the sodium channel (Sheets & Hanck, 2007; Hanck et al. 2009), the role of fast inactivation in high-affinity lidocaine block is still not fully understood (Sheets et al. 2010). Our data reveal that deletion of F1486 removes a critical element mediating the ability of lidocaine to stabilize the inactivated state of hNav1.5.

Taken together, our data show that mutation of F1486 has complex effects, including loss-of-function (decreased sodium current density and decreased lidocaine sensitivity) and gain-of-function alterations (impaired inactivation, augmented late I_Na and increased ramp current). Despite the decrease in peak current density, the substantial gain-of-function changes induced by the F1486del are sufficient to cause APD prolongation. Our results demonstrating the profound insensitivity of inactivated F1486del channels to lidocaine suggest that this probably contributed to the patient's lack of response to lidocaine and illustrates that the sensitivity of patients with presumed LQT3 to lidocaine can be critically dependent on the specific hNav1.5 mutation.

Glossary

APD

action potential duration

EAD

early afterdepolarization

F1486del

deletion of phenylalanine 1486

F1486del-R

TTX-resistant F1486del channels

HEK 293

human embryonic kidney 293

hNav1.5

human Nav1.5

IFM motif

isoleucine–phenylalanine–methionine motif

LQT3

long QT 3 syndrome

TTX

tetrodotoxin

WT

wild-type

WT-R

TTX-resistant wild-type channels

Author contributions

W.So. performed and analysed experiments, contributed to study design and writing of the manuscript. Y.X. contributed to collection and analysis of the data. H.C., N.M.A. and P.M. contributed to collection of the data. A.D.P. and A.H. contributed to design of the experiments and interpretation of the data. T.R.C. and W.Sh. contributed to the design of the study, interpretation of the data and writing of the manuscript. All authors approved the final version.

Supplementary material

Supplementary Figure S1

Supplementary Figure S2

Supplementary Figure S3

Supplementary Figure S4

Supplementary Figure S5

Supplementary Table S1

Supplementary Table S2