Contribution of sodium channel neuronal isoform Na_v1.1 to late sodium current in ventricular myocytes from failing hearts

Abstract

Late Na⁺ current (I_NaL) contributes to action potential (AP) duration and Ca²⁺ handling in cardiac cells. Augmented I_NaL was implicated in delayed repolarization and impaired Ca²⁺ handling in heart failure (HF). We tested if Na⁺ channel (Na_v) neuronal isoforms contribute to I_NaL and Ca²⁺ cycling defects in HF in 17 dogs in which HF was achieved via sequential coronary artery embolizations. Six normal dogs served as control. Transient Na⁺ current (I_NaT) and I_NaL in left ventricular cardiomyocytes (VCMs) were recorded by patch clamp while Ca²⁺ dynamics was monitored using Fluo-4. Virally delivered short interfering RNA (siRNA) ensured Na_v1.1 and Na_v1.5 post-transcriptional silencing. The expression of six Na_vs was observed in failing VCMs as follows: Na_v1.5 (57.3%) > Na_v1.2 (15.3%) > Na_v1.1 (11.6%) > Na_v2.1 (10.7%) > Na_v1.3 (4.6%) > Na_v1.6 (0.5%). Failing VCMs showed up-regulation of Na_v1.1 expression, but reduction of Na_v1.6 mRNA. A similar Na_v expression pattern was found in samples from human hearts with ischaemic HF. VCMs with silenced Na_v1.5 exhibited residual I_NaT and I_NaL (∼30% of control) with rightwardly shifted steady-state activation and inactivation. These currents were tetrodotoxin sensitive but resistant to MTSEA, a specific Na_v1.5 blocker. The amplitude of the tetrodotoxin-sensitive I_NaL was 0.1709 ± 0.0299 pA pF^–1 (n = 7 cells) and the decay time constant was τ = 790 ± 76 ms (n = 5). This I_NaL component was lacking in VCMs with a silenced Na_v1.1 gene, indicating that, among neuronal isoforms, Na_v1.1 provides the largest contribution to I_NaL. At –10 mV this contribution is ∼60% of total I_NaL. Our further experimental and in silico examinations showed that this new Na_v1.1 I_NaL component contributes to Ca²⁺ accumulation in failing VCMs and modulates AP shape and duration. In conclusion, we have discovered an Na_v1.1-originated I_NaL component in dog heart ventricular cells. This component is physiologically relevant to controlling AP shape and duration, as well as to cell Ca²⁺ dynamics.

Key points

• Late Na⁺ current (I_NaL) contributes to action potential remodelling and Ca²⁺/Na⁺ changes in heart failure.

• The molecular identity of I_NaL remains unclear.

• The contributions of different Na⁺ channel isoforms, apart from the cardiac isoform, remain unknown.

• We discovered and characterized a substantial contribution of neuronal isoform Na_v1.1 to I_NaL.

• This new component is physiologically relevant to the control of action potential shape and duration, as well as to cell Ca²⁺ dynamics, especially in heart failure.

Introduction

A physiological role of neuronal isoforms of voltage-sensitive Na⁺ channels (Na_vs) in heart has been suggested many years ago by Corabeouf (Coraboeuf et al. 1979) based on their result that low concentrations (33 nm) of tetrodotoxin (TTX) significantly shortened action potential (AP) duration recorded in cardiac Purkinje fibres. At this low concentration TTX blocks only so-called ‘TTX-sensitive’ Na_vs which are neuronal isoforms (Haufe et al. 2005a2005a). Indeed, during the past decade, besides the dominating cardiac isoform main α-subunit, Na_v1.5, different transcripts of highly TTX-sensitive neuronal Na_v isoforms have been found in mouse (Na_v1.1, 1.3, 1.6) and dog (Na_v1.1, 1.2, 1.3) heart (Maier et al. 2004; Haufe et al. 2005a2005a,b). These neuronal Na_vs are responsible for 10% and 20% of the peak transient Na⁺ current, I_NaT, in dog myocardial and Purkinje cells, respectively (Haufe et al. 2005b2005b). The contribution of neuronal Na_vs to the late Na⁺ current, I_NaL, has not been studied in cardiac cells, therefore the molecular identity of I_NaL needs further careful consideration. This problem is especially important in HF where the gene expression profile changes dramatically. Thus an intriguing possibility could be that the differential expression of neuronal Na_vs can contribute, at least in part, to I_NaL alterations (augmentation) (Valdivia et al. 2005; Maltsev et al. 2007) and to I_NaL-related electrophysiological abnormalities observed in HF (for review see Maltsev & Undrovinas, 2008).

In the present study we used a combination of molecular and pharmacological approaches, as well as numerical modelling, to test a hypothesis that neuronal Na_v isoforms contribute to I_NaL in a physiologically significant manner in ventricular myocytes (VCMs) isolated from failing adult dog hearts. We performed post-transcriptional silencing of Na_v1.1 and Na_v1.5 genes by sequence-directed RNA using virally delivered short interfering RNA (siRNA). We found the expression of six Na_vs in both human heart and adult dog VCMs. Analysis of the expression revealed up-regulation of Na_v1.1 and down-regulation of Na_v1.6 in failing heart compared to normal heart. We took advantage of the method of culturing adult dog VCMs, established by our group, that preserves I_NaT and I_NaL over 5 days, i.e. sufficient time for gene silencing and membrane protein turnover (Maltsev et al. 2008a2008a; Mishra et al. 2011). In VCMs with silenced Na_v1.5 we found a substantial residual I_NaT and I_NaL. Our biophysical and pharmacological examinations confirmed its neuronal isoform origin. Diastolic Ca²⁺ accumulation was reduced via blockade of the neuronal I_NaL component, indicating its physiological importance. Our numerical model simulations also demonstrated the physiological importance of the novel I_NaL component: the Na_v1.1- associated I_NaL effectively modulates AP shape and duration, as well as intracellular Ca²⁺ homeostasis.

Methods

HF model, human heart samples

The Henry Ford Health System Institutional Review Board approved all studies using human tissue and written informed consent was obtained. Studies conducted in animals as well as the methods of killing the animals were in accordance with the Guidelines for the Care and Use of Laboratory Animals (NIH publication No. 85-23), and were approved by the Henry Ford Health System Institutional Animal Care and Use Committee.

The dog model of chronic HF is similar by a vast array of functional and pathophysiological parameters (Sabbah et al. 1992) to that in humans. HF was produced in 17 dogs by multiple sequential coronary artery microsphere embolizations as previously described (Sabbah et al. 1991). All haemodynamic data were collected during cardiac catheterization performed under general anaesthesia and sterile conditions. Anaesthesia was induced using a combination of intravenous injections of hydromorphone (0.22 mg kg⁻¹) and diazepam (0.2–0.6 mg kg⁻¹), and maintained throughout the procedure with 1–2% isoflurane. All dogs were intubated and ventilated with room air supplemented with oxygen. At the end of data collection, and while the animal was still under general anaesthesia, the chest was opened via a left thoracotomy and the heart rapidly removed and tissue prepared for further studies. After each coronary microembolization session, dogs were weaned off of the respirator and allowed to recover. Animals were closely monitored until the endotracheal tube was removed and the dog was in a sternal position. Animals were then placed in post-operative recovery cages with incubated cloth blankets for warmth. Animals were given analgesics for pain and/or medications for cardiovascular complications only if needed. The animals were monitored until pain medications and/or cardiovascular drugs were no longer needed. HF state was confirmed by haemodynamic and angiographic measurements. At the time of harvesting the hearts (∼3 months after last embolization), left ventricular (LV) ejection fraction was 25 ± 1.2% (n = 17).

Human heart samples were obtained from explanted hearts from patients with chronic HF.

Cell culture and transfection

VCMs were enzymatically isolated from the apical LV mid-myocardial slices as previously reported (Maltsev et al. 1998a). The yield of viable rod-shaped, Ca²⁺-tolerant VCMs varied from 40 to 70%. VCMs were cultured for 5 days (the time frame required for gene silencing; Maltsev et al. 2008a), as previously described (Maltsev et al. 2002). We added 10 μl of 5 × 10⁸ virus particles ml^–1 directly to the medium of the culture dish containing ∼15 × 10⁴ cells and cultured them further during the next 4 days (5 days total). Effectiveness of transfection was visually monitored in individual cardiomyocytes (Nikon Diaphot, 200) by green fluorescent protein (GFP) fluorescence (Fig.3A inset) and by the occurrence of mRNA encoding GFP.

Figure 3.

Post-transcriptional SCN5A gene silencing results in functional reduction of the transient Na⁺ current, I_NaT, and reveals non-cardiac Na⁺ channel component in LV cardiomyocytes from dog with chronic heart failure

A, representative raw traces of I_NaT were recorded at different membrane potentials (V_m) in cells 5 days after infection with virus containing control non-silencing siRNA and cDNA for GFP (control GFP, left), and siRNA-Na_v1.5 (right). Cell images expressing GFP are shown in insets. B, average data from peak I_NaT–voltage relationship obtained in cultured cells for 5 days with virally delivered control siRNA (• control GFP, n = 6), and Na_v1.5-siRNA (▴, n = 7). We found a statistically significant reduction of I_NaT in response to Na_v1.5-siRNA in a wide range of potentials (*). Continuous lines show theoretical curves fitted to experimental current–voltage (I–V) data points to evaluate the steady-state activation (SSA) parameters which are given in the panel. •, control GFP, control siRNA (n = 18); ▴, Na_v1.5-siRNA (n = 16). C, average experimental data of steady-state inactivation (SSI) data points in control (•, n = 17) and after SCN5A gene silencing by the siRNA (▴, n = 16) along with the fit to a Boltzmann function. The F test revealed a statistically significant (P < 0.05) difference between theoretical fits (B and C), and statistical significance, P < 0.05, between experimental data points presented as mean ± SEM (B) was evaluated by ANOVA followed by Bonferroni's post hoc test. Voltage clamp protocols are shown in B and C insets; V_h, holding and V_p, pre-pulse potentials, respectively.

siRNA design

Double stranded hairpin siRNAs corresponding to the previously sequenced dog Na_v1.5 and Na_v1.1 (GenBank DQ665939 and AY126477, respectively) and other Na_vs were designed as recommended (Elbashir et al. 2001) and published by our group (Mishra et al. 2011). The details and sequences of silencing and non-silencing (control) siRNAs used in this study are given in Table1. A BLAST search for these sequences against the GenBank database was performed and did not reveal matches with any other gene.

Table 1.

DNA sequences for coding siRNA designed to silence different genes that encode the Na⁺ channel isoform main α- and auxiliary β-subunit used in this study to make the adenovirus

Nav1.1-A3906F	5′-GATCATCTCTCAGGACACTAAGAGCGGAATTCGGCTCTTAGTGTCCTGAGAGATTTTTT-3′
Nav1.2A-1683F	5′-GATCCCGTAAATTGGTTCCTATCGTCTGTTGAATTCCGCAGACGATAGGAACCAATTTATTTTTTCCAAA-3′
Nav1.3-B5191F	5′-GATCCGTCAGAGCAAAGAGCAGTGTGCTTGAATTCCGGCACACTGCTCTTTGCTCTGATTTTTTCCAA-3′
Nav1.5–190F	5′-GATCAAAGCTGCCAGATCTCTATGGGATCGATGCCATAGAGATCTGGCAGCTTTTTTTT-3′
Nav1.6-A1271F	5′-GATCCGTGAGTCGGTCTAACATCTGCTTTGAATTCCGAGCAGATGTTAGACCGACTCATTTTTTCCAA-3′
Nav2.1-A88F	5′-GATCGAGTGTGGATCTTCTGGAGAAGGAATTCGTTCTCCAGAAGATCCACACTCTTTTT-3′
NEG-CTRL-F	5′-GATCCCGTCGCTTACCGATTCAGAATGGTTGATATCCGCCATTCTGAATCGGTAAGCGATTTTTTCGA-3′
Navβ1–202F	5′-GATCGAAGATCCACATTGAGGTGGTGATCGATGACCACCTCAATGTGGATCTTCTTTTT-3′
Navβ2–216F	5′-GATCCAACTGCTCTGAGGAGATGTTGATCGATGAACATCTCCTCAGAGCAGTTGTTTTT-3′

Real-time PCR and Western blot analysis

To quantify mRNA levels, we used both conventional gels and the fluorescence-based (SYBR green) kinetic real-time PCR performed using 7500 Fast Real-Time PCR Applied Biosystems (Life Technologies, Grand Island, NY, USA) sequence detection system. For each sodium channel isoform, several transcript-specific perfect match primer sets were designed and custom synthesized on the basis of sequences available in the NIH database. Primers were tested for production of predicted band size by RT-PCR. PCR products of correct band sizes were sequenced to check specificity. Each specific primer, which was selected after sequencing, was tested for use in real-time PCR before using in silencing experiments. Primers were also tested for amplification efficiency by standard curve analysis. Efficiencies of all the primers used in the study were between 90 and 99%. The quality of the amplified fragments was monitored using melting curve analysis and agarose gel electrophoresis. Quantitative values were obtained from the PCR quantification cycle number (Cq) at which point the increase in signal for the PCR product was exponential. The sets of primers are given in Table2.

Table 2.

Sequences of primers used in this study for the real-time PCR

Isoform		Sequences	Product size (bp)
Nav1.1A	F	5′-CCTGGTGTTCATTGTGTTGTTC	491
	R	5′-AATCGGGTGGTTTACTGTTGAG
Nav1.2A	F	5′-TCAAGAGAGTTCAGTGGTGCTG	418
	R	5′-AGTCTCTCCGGCTGTCATTGT
Nav1.3A	F	5′-CTACCACATCTCCTCCTTCCT	189
	R	5′-TCCAGTTGACACATAGACTTTACCA
Nav1.5A	F	5′-CCCAGAAGCAGGATGAGAAG	516
	R	5′-CGGTGAAGGTGTACTCGACA
Nav1.6A	F	5′-CAGCAGATGTTAGACCGACTCA	155
	R	5′-CTCCTTGGCACTCTTTGAGC
Nav2.1A	F	5′-TATTTGCTGGTTGGGATGG	586
	R	5′-ACTGCCTCTTGTTTTCGTTTCA
Navβ1	F	5′-TATGAGAATGAGGTGTTGCAGCTGG-3′	482
	R	5′-ATACGGCTGGCTCTTCCTTGAGG-3′
Navβ2	F	5′-CCGACTAACATCTCAGTCTC-3′	211
	R	5′-CTGTTTGTGGTTCACTGTG-3′

The relative mRNA abundance expressed as arbitrary units was calculated using expression levels of all transcripts normalized to glyceraldehyde 3-phosphate dehydrogenase (GAPDH) mRNA. This value was then normalized to mRNA levels measured from VCMs transfected with the non-silencing siRNA for the appropriate transcript (2^−ΔΔCt method). Membrane protein preparations were obtained from approximately 3 × 10⁶ myocytes as previously described (Zicha et al. 2004; Mishra et al. 2011). Antibody was detected with Western blot chemo-luminescence reagent (NEN Life Science Products, Boston, MA, USA). Western blots were considered specific if the peptide epitope reduced the band intensity. The resulting images of Western blots and PCR gels (See Fig.2A and B) were scanned and the relative densities of bands were quantified using SigmaGel (Sigma-Aldrich, USA) or ImageJ (National Institute of Health free software) software. The primary polyclonal anti-Na_v1.5 and anti-Na_v1.1 antibodies were obtained from Alomone Labs (Jerusalem, Israel). Polyclonal calsequestrin antibody (Abcam Inc., Cambridge, MA, USA) was used for the protein loading control.

Figure 2.

Silencing of SCN5A gene that encodes Na_v1.5 in LV cardiomyocytes of dog with chronic HF by virally delivered siRNA

A, RT-PCR analysis of mRNA abundance in control (fresh isolated cells), in cells infected with virus that contains cDNA for Na_v1.5 siRNA + GFP (Ad-GFP-Na_v1.5) and GFP alone (Ad-GFP). Upper panel: representative gel of such analysis performed at different times after the virus infection. Lower panel: summary data with statistical analysis of experiments similar to that shown in the upper panel obtained in 5–7 cells. Relative Na_v1.5 mRNA is given as relative to control density of bands. *P < 0.05, vs. control and GFP. B, Western blot analysis of Na_v1.5 protein abundance in control, and in Ad-GFP-Na_v1.5- and Ad-GFP-treated cells. *P < 0.001, vs. control and Ad-GFP. C, RT-PCR analysis of different sodium channel isoforms after transfection with Na_v1.5-siRNA (from 4 to 8 samples). *P < 0.05. Statistically significant changes in A, B and C, were evaluated by the ANOVA test followed by Bonferroni's post hoc test. Data are mean ± SEM.

Electrophysiology

I_Na was recorded using conventional whole-cell patch-clamp techniques. Experimental and theoretical evaluation of the decay kinetics of I_Na and the steady-state activation (SSA) and inactivation (SSI) voltage dependency of I_Na were the same as we reported previously (Mishra et al. 2011). The decays of both I_NaT and I_NaL were approximated by double exponential fits, but the SSA and SSI were approximated by respective Boltzmann functions. Ca²⁺ signals were measured at 35 °C and 1.8 mm [Ca²⁺]_o using a photo-multiplier in field-stimulated single VCMs loaded with the acetoxymethyl ester form of Fluo-4, Fluo-4 AM (Undrovinas et al. 2010). I_NaT and I_NaL were measured at room temperature (21–23 °C).

Numerical excitation–contraction coupling model

To address the physiological significance of the neuronal I_NaL component, we used a modified excitation–contraction (E–C) coupling model of a failing canine ventricular myocyte developed previously by Winslow et al. (1999). The original model was modified to include I_NaL equations as we previously reported (Undrovinas et al. 2010; Mishra et al. 2011). In this study we further modified I_NaL formulations to include two separate I_NaL components for cardiac and neuronal isoforms, respectively. Each component has its own activation and inactivation gating variables.

Statistical analysis

Statistics are reported as mean ± SEM with ‘n’ representing the number of cells. Multiple comparisons between treatment groups were made using one-way analysis of variance (ANOVA) followed by Bonferroni's post hoc test. The significance of changes of theoretical SSA and SSI fits was evaluated with an F test (StatMost, Data-Most, Salt Lake City, UT, USA) for tabulated values predicted by the Boltzmann model at a confidence level of 0.95. Differences for both experimental data and model predictions were considered statistically significant for P < 0.05.

Results

Molecular biology data

We observed HF-related changes in mRNA levels of Na_v1.1 and Na_v1.6 in human and dog VCMs, whereas bands corresponding to Na_v1.2 were strong in both samples shown in Fig.1A and B. The relative level of main α-subunit Na_v expression was compared by real-time PCR. We found transcripts of six genes encoding the Na⁺ channel main α-subunit expressed in failing VCMs in the following order: Na_v1.5 (57.3%) > Na_v1.2 (15.3%) > Na_v1.1 (11.6%) > Na_v2.1 (10.7%) > Na_v1.3 (4.6%) > Na_v1.6 (0.5%). Statistical analysis in normal and failing hearts of both human and dog revealed up-regulation of Na_v1.1 expression, but reduction of Na_v1.6 mRNA levels (Fig.1C and D).

Figure 1.

Expression of genes encoding different Na⁺ channel isoforms Na_vx

RT-PCR analysis of Na_v transcripts in normal and failing human (A) and dog (B) left ventricular tissue samples. A, from left to right: Na_v1.1 isoform obtained in normal heart (NH), in explanted heart from patients with ischaemic (IHF), and non-ischaemic (NIHF) heart failure, respectively. The expected 491 bp band was not detected in NH and NIHF, Na_v1.2 isoform (417 bp), Na_v1.3 isoform (189 bp), Na_v1.6 isoform. A specific band of 155 bp is missing in IHF; a non-specific smaller band is present instead. Na_v2.1 isoform (586 bp). B, the same Na⁺ channel isoforms as shown in A were obtained in dog normal (NH) and failing (HF) hearts. Last gel lane on right represents standard bp marker. C, summary data with statistics for Na_v isoforms expressed in human heart. RT-PCR gels were analysed in 5–7 tissue samples from left ventricles, D, summary data with statistics for Na_v isoforms expressed in dog heart. RT-PCR gels were analysed in 5 samples of ventricular cardiac myocytes from left ventricles of normal dog hearts and failing dog hearts, respectively. *P < 0.05, ANOVA, data are mean ± SEM for both C and D.

The adenovirus-delivered Na_v1.5-siRNA reduced (in time-dependent fashion) the numbers of transcripts of the main α-subunit of the cardiac Na⁺ channel isoform (Fig.2A), as well as its protein level (Fig.2B). This inhibition of gene expression is specific to Na_v1.5, without a cross-reaction with the genes expressing auxiliary β-subunits (not shown) or other Na_vs (Fig.2C).

Patch-clamp data

Patch-clamp data experiments were performed in HF dog VCMs infected with adenovirus-Na_v1.5-siRNA. Our previous studies with specific oligonucleotides (Maltsev et al. 2008a) have shown that Na_v1.5 provides a dominant contribution to I_NaL. Indeed, silencing of the Na_v1.5 gene in the present study also resulted in a significant and substantial reduction (by about 70%) of the peak transient Na⁺ current, I_NaT, shown in Fig.3. Furthermore, the results of our patch clamp experiments in Fig.3B and C show that there was a prominent and statistically significant rightward shift of the SSA (V_1/2G = −40.9 ± 1.7 mV, K_G = 6.5 mV, n = 8 in control vs. 36.8 ± 1.6 mV, K_G = 6.9 mV, n = 5 in Na_v1.5-siRNA, P < 0.05 when V_1/2 G was compared) and SSI (V_1/2A = 81.3 ± 1.8 mV, K_A = −6.6 mV, n = 8 in control vs. V_1/2A = 70.3 ± 2.3 mV, K_A = −6.4 mV, n = 5, P < 0.05 when V_1/2A was compared). G and A stands for activation and inactivation, respectively. These results illustrate that the remaining pool (whole or part) of Na⁺ channels demonstrates activation and inactivation characteristics different from that for Na_v1.5. It is known that the neuronal Na⁺ channels activate at more depolarized potentials (Haufe et al. 2005b). Therefore these results support the idea that neuronal channels may contribute to the remaining I_NaT in these failing VCMs.

To further explore the idea of the neuronal isoforms’ contribution to I_NaT, we used 2-aminoethyl methanethiosulfonate hydrobromide (MTSEA), which is known to block only cardiac and not neuronal isoforms of the Na⁺ channel (Haufe et al. 2005b). In the freshly isolated failing VCMs we found a substantial portion of I_NaT (∼20%) that was MTSEA resistant (Fig.4). While this component was smaller than the remaining I_NaT after Na_v1.5 post-transcriptional silencing (∼30%) (Fig.4C), it can still be physiologically significant (as we show below).

Figure 4.

Contribution of non-cardiac Na⁺ channel isoforms to I_NaT in LV cardiomyocytes of dogs with HF

A, consecutive raw I_NaT traces recorded before (control) and during MTSEA application in freshly isolated cells. B, time course of MTSEA effect on the peak I_NaT recorded in the same cell as shown in A. C, summary data and statistical analysis of I_NaT density in freshly isolated control and MTSEA-treated cells and infected cells with virus containing non-silencing siRNA or Na_v1.5-siRNA. Statistically significant changes, P < 0.05, were found in both groups evaluated by the ANOVA test followed by Bonferroni's post hoc test (*,** Na_v1.5-siRNA vs. freshly isolated control and GFP control, respectively, § MTSEA vs. freshly isolated control) I_NaT was recorded at –10 mV. Data are mean ± SEM.

In the next series of experiments we tested if these channels with different SSA and SSI characteristics contribute to I_NaL. To unmask the contribution of neuronal Na_vs to I_NaL we deliberately silenced Na_v1.5 gene expression by the virally delivered Na_v1.5-siRNA in these cells. Na_v1.5-siRNA caused a substantial I_NaL reduction with an apparent slower decay of the remaining I_NaL (Fig.5A). The density of I_NaL evaluated at –30 mV significantly decreased in cells infected with the virus containing Na_v1.5-siRNA compared to freshly isolated cells and to cells infected with the non-silencing siRNA expressing GFP (Fig.5B). Analysis of SSA and SSI in these cells clearly shows a rightward shift of the V_1/2 (Fig.5C and D) as we found for I_NaT. We then fitted this I_NaL component to a two exponential decay function (yielding ‘slow’ and ‘fast’ sub-components, respectively). The slow sub-component of I_NaL decay (probably corresponding to the late scattered mode; Maltsev & Undrovinas, 2006) of the remaining I_NaL was significantly slower (Fig.5F). The first (i.e. fast) exponent that corresponds to the burst mode (Maltsev & Undrovinas, 2006) remained unchanged (not shown). Therefore these data reveal a different Na_v contribution to I_NaL in failing VCMs.

Figure 5.

Post-transcriptional SCN5A gene silencing results in functional reduction of the late Na⁺ current, I_NaL, and unmasks non-cardiac Na⁺ channel component in LV cardiomyocytes from dog with chronic heart failure

A, representative raw traces of I_NaL were recorded at membrane potentials V_m = −10 mV in freshly isolated cells, and in cells 5 days after infection with virus containing control non-silencing siRNA and cDNA for GFP and siRNA-Na_v1.5 (indicated by the arrows). Exponential fits to I_NaL decay are shown by continuous lines and are superimposed with traces. Values of slow exponent time constant, τ,  are given in the panel. B, summary data with statistics for I_NaL density. C, representative I_NaL–voltage relationship obtained in freshly isolated (•) and cultured cells for 5 days with virally delivered control siRNA (▴ control GFP), and Na_v1.5-siRNA (▪). Continuous lines show theoretical curves fitted to experimental current–voltage (I–V) data points to evaluate steady-state activation (SSA) parameters which are given in the panel. D, average experimental data of steady-state inactivation (SSI) data points in control (•, n = 17) and after SCN5A gene silencing by the siRNA (▪, n = 16) along with the fit to a Boltzmann function. E and F, statistical data for I_NaL decay time course evaluated by the double exponent fit. Difference between theoretical fits (C and D) was statistically significant (P < 0.05) evaluated by F test. Statistical significance, P < 0.05, between experimental data points presented as mean ± SEM (B, E and F) were evaluated by ANOVA followed by Bonferroni's post hoc test.

To further study the neuronal isoform-generated I_NaL component, we used low concentrations (100 nm) of TTX in VCMs with silenced Na_v1.5 expression (results of these experiments are shown in Fig.6A and B). Under these conditions the contribution of the cardiac (TTX-resistant) Na⁺ channel must remain almost unchanged. However, application of TTX to these myocytes caused a marked reduction of I_NaL (Fig.6A). We then evaluated the revealed TTX-sensitive current as the difference between the recordings before and after TTX application (Fig.6B). This TTX-sensitive current, presumably carried by the neuronal Na_vs, was slowly inactivating in a single-exponential manner. The amplitude of the TTX-sensitive I_NaL was 0.1709 ± 0.0299 pA pF⁻¹ (n = 7 cells) and the decay time constant was τ = 790 ± 76 ms (n = 5 cells) recorded at V_m = −10 mV. These values we used in our numerical model simulations to evaluate the physiological significance of the neuronal I_NaL component (see below). The experimental current-to-voltage relation (I–V curve) for such experiments and theoretical (Boltzmann function) fits are shown in Fig.6C. The I–V for the TTX-sensitive current was calculated (squares) as a subtraction of two I–V curves (circles and triangles). It is clearly seen that the difference I–V is shifted rightwards and we found the average SSA parameters for this component to be as follows: V_1/2G = −24.1 ± 1.0 mV, and slope K_G = 5.7 ± 0.3 mV vs. control V_1/2G = –38.5 ± 1.2 mV, K_G = 5.6 ± 0.9 mV. The I–V relationship analysis revealed a significant neuronal I_NaL component within the range of the AP plateau (Fig.6C). These parameters of TTX-sensitive I_NaL were also used in our numerical modelling.

Figure 6.

Contribution of tetrodotoxin (TTX)-sensitive neuronal Na_vs to I_NaL in failing dog ventricular cardiomyocytes

A, raw current traces recorded in cultured cells in which SCN5A was silenced by the virally delivered Na_v1.5-siRNA before and after TTX (100 nm). B, difference current (Control-TTX), or neuronal I_NaL component along with the best single-exponent fit. Zero current represents difference between control recording and after TTX washout. Fit parameters and voltage-clamp protocol (same for A) is given in the panel. C, normalized I–V relationship of I_NaL before and after TTX recorded in freshly isolated cells. Current density was evaluated as an average within 200–220 ms after depolarization pulse for 2 s, V_h = −120 mV. Continuous lines are theoretical curves of steady-state activation. Note that TTX produced SSA shift to left (V_1/2 = −43 mV in control vs. –45 mV with TTX). Data are mean ± SEM.

To further study the molecular identity of the TTX-sensitive I_NaL, we made adenovirus-based constructs containing cDNA templates for endogenous transcription of the GFP reporter gene, and cognitive siRNA, targeting different Na_v isoforms that are expressed in human and dog LV cardiomyocytes according to our data (Fig.1). As shown in Fig.7, transfection of failing dog VCMs with adenovirus containing Na_v1.1-siRNA significantly reduced both mRNA abundance (Fig.7A) and Na_v1.1 protein level (Fig.7B) without cross-reacting with other Na⁺ channel isoforms (Fig.7C). Furthermore, we found that application of MTSEA completely blocked I_NaL in failing VCMs (shown in Fig.8A and B) with silenced Na_v1.1. At the same time, application of 100 nm TTX was not effective in VCMs with the silenced expression of Na_v1.1 (Fig.8C). Figure8D shows evaluated I_NaL densities measured at –10 mV in GFP control (0.3387 ± 0.06 pA pF⁻¹), Na_v1.5-siRNA (0.2123 ± 0.023 pA pF⁻¹) and Na_v1.1-siRNA (0.1737 ± 0.018 pA pF⁻¹). Therefore at this membrane potential the Na_v1.1-originated I_NaL component (∼60% of total I_NaL) is comparable with Na_v1.5-originated I_NaL (i.e. remaining ∼40% of total I_NaL). Furthermore, in our in silico simulations of APs we compared the dynamic contributions of these components and confirmed that Na_v1.1 provides a major contribution to total I_NaL during the AP plateau (see below). Based on these data, further studies with other Na_vs contributions to I_NaL seemed to be irrelevant and results are not shown. These data support the idea that Na_v1.1 is the major Na⁺ channel isoform underlying the TTX-sensitive component of I_NaL in failing VCMs.

Figure 7.

Silencing of SCN1A gene that encodes Na_v1.1 with adenovirus containing coding sequences for siRNA in cardiomyocytes from dogs with chronic HF

A, summary data of RT-PCR analysis of mRNA abundance (representative gels are shown in the inset at the upper right corner of the panel) in cultured dog cardiomyocytes. AdNC, adenovirus expressing non-specific siRNA as a control, Ad1.1siRNA, adenoviral construct expressing Na_v1.1 silencing sequences. B, summary data of Western blot analysis of Na_v1.1 protein abundance in cells treated with control virus (lanes 2–5 and 10 in the representative gels shown in the inset at the right upper corner of the panel) or Na_v1.1-siRNA (lanes 1 and 6–9, see inset). C, summary data of RT-PCR analysis of different Na⁺ channel isoforms after transfection with Na_v1.1-siRNA (from 4 to 8 samples). *Statistically significant changes in A, B and C were evaluated by the ANOVA test followed by Bonferroni's post hoc test. Data are mean ± SEM.

Figure 8.

Contribution of neuronal Na_v1.1 to I_NaL in failing dog ventricular cardiomyocytes

In these cells expression of Na_v1.1 was silenced by virally delivered Na_v1.1-siRNA. A, raw current traces recorded before (Na_v1.1-siRNA, control) and after MTSEA application. Continuous line represents single exponent fit to I_NaL decay, and value of time constant, τ, is given at the trace. B, time course of I_NaL amplitude, evaluated between 200 and 220 ms after depolarization onset, during MTSEA application. C, raw current traces recorded before (Na_v1.1-siRNA, control) and after 100 nm TTX application (Na_v1.1-siRNA + TTX). The difference current was obtained by subtracting currents before and after TTX. D, summary data and statistical analysis of I_NaL current densities obtained in cells infected with virus that contains cDNA for GFP (GFP Control) for Na_v1.5-siRNA + GFP (Na_v1.5-siRNA) and for Na_v1.1-siRNA + GFP (Na_v1.1-siRNA). Current densities were measured 200 ms after depolarization to –10 mV onset. Statistical significance, P < 0.01, between experimental data points presented as mean ± SEM were evaluated by ANOVA followed by Bonferroni's post hoc test.

Intracellular Ca²⁺ dynamics data

In the next series of experiments we addressed the question of the physiological importance of the neuronal I_NaL component in failing heart. Indeed, we found that blockade of neuronal I_NaL by low concentrations of TTX reduced diastolic Ca²⁺ accumulation in response to a 1.5 Hz pulse train in freshly isolated failing VCMs (Fig.9).

Figure 9.

Blockade of neuronal Na⁺ channels by 100 nm TTX reversibly attenuates diastolic intracellular Ca²⁺ elevation in ventricular cardiomyocytes of dogs with chronic heart failure

Intracellular Ca²⁺ dynamics recorded during 1.5 Hz pulse train in control (A), in the presence of 100 nm TTX (B) and after the toxin washout (C). Dotted lines in A, B and C refer to average diastolic Ca²⁺ levels before and during pulse train. D, data summary for diastolic Ca²⁺ (CaD) changes (ΔCaD) in response to the pulse train. Data are mean ± SEM, cell numbers are given above the bars. *P < 0.05, ANOVA followed by Bonferroni's post hoc test.

Numerical model simulations of the neuronal component of I_NaL

To further address the physiological significance of the neuronal I_NaL component, we used a modified E–C coupling model in which we carefully introduced experimentally obtained I_NaL density, decay kinetics, as well as SSA and SSI parameters. The model predicted I_NaL traces and I–V relationship in good agreement with those obtained in our voltage-clamp experiments (compare Figs5,6 and8 with Fig.10A). We tested the physiological importance of the novel neuronal type I_NaL by numerical simulations of I_NaL dynamics during APs (Fig.10B). Our simulations showed that the neuronal I_NaL lasts during the whole AP plateau and its amplitude (especially at the late phase of the plateau where it amounted to 60% of cardiac I_NaL), indicating that the neuronal component of I_NaL is a major player in the ion current balance during the AP plateau. Indeed, this new I_NaL neuronal component can significantly modulate (prolong) AP. In concert with the cardiac I_NaL, neuronal I_NaL leads to AP prolongation and to a higher and longer plateau of the AP (Fig.10C), i.e. reproducing well-known features of failing myocardium. Next we numerically evaluated cytosolic Ca²⁺ dynamics during a 1.5 Hz pulse train. Introduction of the neuronal I_NaL component significantly increased diastolic Ca²⁺ accumulation at the end of the pulse train (Fig.10D), which is in a good agreement with the experimental data (Fig.9).

Figure 10.

In silico investigation on neuronal and cardiac I_NaL components in HF dog ventricular cell

A, predicted current–voltage relationships. The upper panels show families of predicted current traces at different membrane voltages for the cardiac and neuronal Na_v components. The lower panels show plotted amplitude values of these predictions. Left lower sub-panel shows the peak I_NaL (at zero time after the depolarization onset) and right lower sub-panel shows the current amplitudes 200 ms after depolarization onset. Holding potential was V_h = −120 mV, test pulses were applied from –70 mV to +70 mV. B, predicted profile of I_NaL generated by cardiac and neuronal Na_vs in numerical model simulations of AP. C and D, predicted differential contribution of I_NaL by the neuronal and cardiac Na_vs to shape and duration of AP and intracellular Ca²⁺ dynamics, respectively, at the pacing rate of 1 Hz. Complete I_NaL elimination from simulations (I_NaL = 0) reduces AP duration and eliminates diastolic Ca²⁺ accumulation. Introduction of only the cardiac or neuronal component of I_NaL increases AP duration and leads to diastolic Ca²⁺ accumulation, whereas incorporation of both I_NaL components, which reflects the situation in situ, affects these physiological responses dramatically. All numerical simulations were performed at a physiological temperature of 37 °C using our modification of the Winslow E–C coupling model of canine failing VCMs (Winslow et al. 1999) with formulations for I_NaL which were previously developed based on our experimental voltage-clamp data (Undrovinas et al. 2010; Mishra et al. 2011). In the present study I_NaL formulations included two separate I_NaL components for cardiac and neuronal isoforms, respectively. Each component has its own activation and inactivation gating variables evaluated from patch-clamp data obtained in the present study (see section ‘Patch-clamp data’).

Discussion

Using a combination of post-transcriptional gene silencing by siRNA interference with the pharmacological approaches, cell electrophysiology (patch-clamping), intracellular Ca²⁺ monitoring, and numerical modelling, we discovered and characterized a novel neuronal component of I_NaL in dog ventricular cardiomyocytes. We also found that expression of the Na_v1.1 gene is responsible for the molecular identity of the neuronal I_NaL component. This component seems to be involved in E–C coupling by affecting AP shape and duration as well as intracellular Ca²⁺ dynamics. Furthermore, as discussed below in detail, the neuronal I_NaL component may provide a novel cellular and molecular mechanism contributing to major abnormalities associated with chronic HF, such as the impaired, rate-dependent (Rao et al. 2007; Logeart et al. 2009; Undrovinas et al. 2010) diastolic function and repolarization (for review see Maltsev & Undrovinas, 2008).

After I_NaL had been discovered in human hearts (Maltsev et al. 1998a) it was implicated in HF-related rhythm and contractility deficiencies (for review see Maltsev & Undrovinas, 2008). The importance of the augmented I_NaL contribution to HF mechanisms has been demonstrated in a series of experiments in which ‘correction’ of I_NaL in failing cardiomyocytes resulted in: (1) rescue of normal repolarization, (2) improvement (decreasing) of repolarization beat-to-beat APD variability, and (3) improvement of Ca²⁺ handling and contractility (Maltsev et al. 1998b1998b; Undrovinas et al. 1999, 2006; Maltsev et al. 2007). Accordingly I_NaL has emerged as a novel possible target for cardioprotection to treat the failing heart (Maltsev et al. 2001, 2007; Noble & Noble, 2006; Maltsev & Undrovinas, 2008; Undrovinas & Undrovinas & Maltsev, 2008).

Therefore, the study of modulatory factors and the molecular identity of I_NaL is of obvious importance from both therapeutic and scientific standpoints. Modulatory mechanisms leading to I_NaL augmentation and decay slowing include: (1) the effects of the Na⁺ channel protein microenvironment, such as sub-sarcolemmal cytoskeleton (Undrovinas et al. 1995; Maltsev & Undrovinas, 1997), lipid bilayer composition (Undrovinas et al. 1992) and auxiliary β-subunits (Mishra et al. 2011); and (2) the Ca²⁺/calmodulin/calmodulin kinase signalling pathway (Maltsev et al. 2008a). Recent discoveries of the expression of array of different Na⁺ channel isoforms in hearts indicate a possibility for new molecular mechanisms for I_NaL modulation, i.e. the different Na⁺ channel isoform contributions to I_NaL that were the main objective of the present study.

Molecular identity

We found transcripts of six genes encoding the Na⁺ channel main α-subunit expressed in both human and dog failing hearts and the abundance of mRNA was in the following order: Na_v1.5 (57.3%) > Na_v1.2 (15.3%) > Na_v1.1 (11.6%) > Na_v2.1 (10.7%) > Na_v1.3 (4.6%) > Na_v1.6 (0.5%). It is, in part, in line with the results of previous studies. In rodents, Na_v1.1, Na_v1.3 and Na_v1.6 were initially found in the T tubules (Maier et al. 2002). In dog VCMs and Purkinje fibres, Na_v1.1 and both Na_v1.2 and Na_v1.6 were localized in the intercalated discs and at Z-lines, respectively (Haufe et al. 2005a2005a,b). Therefore these studies suggest species-specific expression and protein localization within compartments of the sarcolemma. Immunoblot data exclude a membrane location for Na_v1.3, despite an abundant amount of mRNA in dog heart (Haufe et al. 2005b). We found expression of Na_v2.1 that has been previously discovered in uterus and human hearts (hNa_v2.1) and may represent a secondary Na_v subfamily (George et al. 1992; Felipe et al. 1994). Its function as a voltage-gated Na⁺ channel has not been confirmed; therefore it is not likely to be a candidate for the I_NaL (Trimmer & Agnew, 1989; Watanabe et al. 2000). Further analysis of expression of other genes as possible candidates for I_NaL revealed HF-related changes in the expression of Na_v1.1 (up-regulation) and Na_v1.6 (down-regulation): at the same time no significant changes in transcripts of other Na⁺ channel isoforms were detected. Therefore our focus was on Na_v1.1 as a plausible contributor to I_NaL. Our dog model is an ischaemic model of heart failure. We show that increased expression of Na_v1.1 in failing VCMs from dogs correlates with human ischaemic HF (IHF) (Fig.1A). At the same time we did not find increased Na_v1.1 expression in ventricle tissue samples from human hearts with non-ischaemic HF.

The contributions of these isoforms to I_NaL under normal conditions and the changes of these contributions in different pathological conditions remain unknown. The molecular identity of I_NaL needs careful consideration especially for failing hearts where the gene expression profile changes dramatically. Using antisense inhibition by oligonucleotides, we have previously demonstrated that Na_v1.5 is a major contributor to I_NaL in normal dog heart (Maltsev et al. 2008a) and this finding was confirmed in the present study. At least 70% of I_NaT and I_NaL are related to this cardiac Na⁺ channel isoform (see Figs6). The late Na⁺ channel openings are expected for at least Na_v1.6 (Welch et al. 2008) and have been reported also for Na_v1.1 (Aman et al. 2009). Therefore, the hypothesis of possible roles of the neuronal isoforms of Na⁺ channel in Na⁺-related Ca²⁺ signals in HF makes sense based on these previous data and is supported, in part, by our findings reported here. A contribution of Na_v1.6 to the mechanism of augmented I_NaL in HF is unlikely because it is down-regulated as we have discussed above. A separate study focusing on Na_v1.6 is required to understand its changes in failing hearts. The Na_v1.1 has been widely studied because of its involvement in the genetic epilepsy syndrome (Catterall et al. 2010). Not only mutant but also wild-type (to a lesser extent though) forms of these channels exhibit slow inactivation and persistent I_Na, which can be modulated by auxiliary β-subunits (Spampanato et al. 2004) similar to what we reported for Na_v1.5 (Maltsev et al. 2009; Mishra et al. 2011).

We demonstrated that β₁ and β₂ subunits act in opposite directions by enhancing or reducing I_NaL in normal and failing dog hearts. It is interesting to note that, similar to Na_v1.5 (Undrovinas et al. 2002; Maltsev & Undrovinas, 2006), the Na_v1.1 also demonstrates modal gating behaviour and can enter a slow mode that generates late openings (Vanoye et al. 2006). This modal shift can be enhanced or attenuated by β-auxiliary subunits (Spampanato et al. 2004).

Physiological importance of neuronal Na_vs in heart

The neuronal component of I_NaL is a significant contributor to AP duration and shape because of its biophysical properties such as a relatively high level of midpoint for steady-state activation and inactivation and a relatively slow decay time course (vs. cardiac type I_NaL). Our numerical model simulations demonstrated that, despite a smaller initial density of the neuronal component (after membrane depolarization) vs. cardiac I_NaL, its ratio to the cardiac I_NaL increases during AP plateau and becomes comparable to the cardiac component at the end of the AP plateau (see Fig.10B). Therefore targeting the neuronal component of I_NaL can be an effective strategy to reduce AP duration and its variability in HF (Fig.10C). We demonstrate that blockade of the neuronal I_NaL component by a low TTX concentration reduces diastolic Ca²⁺ accumulation in both experiments (Fig.9) and in silico simulations (Fig.10D). Therefore this neuronal I_NaL component is probably involved in abnormalities of E–C coupling in failing heart, such as diastolic dysfunction, i.e. the reduced ability of the myocardium to relax during diastole. In contrast, TTX-sensitive I_Na was not implicated in triggered Ca²⁺ release from sarcoplasmic reticulum (SR) in rodent VCMs and the authors claimed no apparent commitment for neuronal Na_vs in E–C coupling (Brette & Orchard, 2006). This discrepancy can be explained in two ways: (1) E–C coupling in rodents versus large animals, such as dog, as well as in humans, is different in the way that the source of Ca²⁺ involved in contraction is mainly stored in SR and is less (or almost not) dependent on transmembrane ion fluxes during AP development, and (2) in rodents, heart rate is very high and APs are very short, therefore I_NaL does not play the same role as in humans or dogs. Since late openings of Na⁺ channels generate both electric current and Na⁺ influx during the AP plateau, I_NaL is expected to contribute to at least two known HF cellular mechanisms: (1) electrophysiological remodelling, and (2) altered cell Na⁺ cycling. The latter mechanism is tightly integrated with Ca²⁺ cycling, as Na⁺ modulates the Na⁺–Ca²⁺ exchanger operation (Bers et al. 2006). Therefore the neuronal component of I_NaL can also play a role in the impaired E–C coupling in HF via Na⁺ balance and its interplay with Ca²⁺ handling and signalling. Haufe et al. (2005b) suggested that neuronal Na_vs might act as a safety mechanism. Because of the biophysical properties of neuronal Na_vs (SSI and SSA are shifted to more depolarized potentials compared to that for Na_v1.5) this safety mechanism can be recruited under pathological conditions, when the resting membrane potential is depolarized.

Study limitations

The relative mRNA abundance expressed as arbitrary units was calculated using the expression levels of all transcripts normalized to GAPDH mRNA. There are some indications that GAPDH expression can change in atria vs. ventricles in patients with HF. Since in this study we did not compare atria vs. ventricles in humans we assume that GAPDH can serve as a valid marker. According to previous studies by our group and others, levels of mRNA for this marker gene do not change in the left ventricle dog or mini pig HF model (Mishra et al. 2005; Imai et al. 2007; Martino et al. 2011). Therefore we are confident that our data are not affected because of this marker choice.

The human data are based on ventricular tissue mRNA analysis, rather than isolated myocytes. It could be, in part, contaminated with mRNA from neurons that can be found in ventricles. We found a very similar pattern of Na_v expression in these samples when compared with samples from isolated myocytes of dog. Therefore we believe that our conclusions are valid.

New strategy to treat HF-related deficiencies

The paradigm in HF for the Na⁺ current is that I_NaT, responsible for the AP upstroke and propagation, is down-regulated, whereas I_NaL is up-regulated and involved in repolarization and Ca²⁺-handling abnormalities. A direct blockade of Na⁺ channels by Class I drugs is not acceptable as a therapeutic strategy in HF, because it leads to further reduction in propagation velocity and increases the probability of excitation propagation blockade or re-entrant arrhythmias. The discovery of a new neuronal component of I_NaL allows a new differential treatment of rhythm abnormalities and contractility deficiency as discussed above. The treatment can include a specific neuronal channel blocker or genetic approaches to specifically reduce the neuronal I_NaL component in heart that is especially important for the late AP plateau phase (Fig.10C) (due to its relatively slow decay, see Fig.6A) but is probably less important for cell excitability (due to its relatively higher activation voltage).

In conclusion we have discovered an SCN1A-related Na_v1.1 component of I_NaL in ventricular myocytes of dog with heart failure. This component is physiologically relevant to controlling AP shape and duration as well as Ca²⁺ dynamics.

Glossary

AP

action potential

E–C

excitation–contraction

GFP

green fluorescent protein

HF

heart failure

I_NaL

late Na⁺ current

I_NaT

transient Na⁺ current

LV

left ventricular

MTSEA

2-aminoethyl methanethiosulfonate hydrobromide

Na_v

sodium channel

siRNA

short interfering RNA

SSA

steady-state activation

SSI

steady-state inactivation

VCM

ventricular cardiomyocyte

Additional information

Competing interests

The authors have no competing interests to declare.

Author contributions

Conception, analysis and design of the experiments: A.U. Collection, analysis and interpretation of data: S.M., V.R., V.A.M., N.A.U. and A.U. Dog chronic heart failure model and mathematical modelling: H.N.S. and V.A.M. Drafting the article or reviewing it critically for important intellectual content: A.U., V.A.M., H.N.S. and S.M. All authors approved the final version of the manuscript.

Funding

This study was supported by National Heart, Lung, and Blood Institute Grants HL-53819 and HL-074238, by a grant-in-aid from the American Heart Association (0350472Z; to A. Undrovinas), and, in part, by the Intramural Research Program of the National Institute on Aging (to V. A. Maltsev; the numerical modelling part).

Author's present address

S. Mishra: Department of Translational Science and Molecular Medicine, MSU College of Human Medicine, Grand Rapids, MI, USA.