SCN10A-short gene therapy to restore conduction and protect against malignant cardiac arrhythmias

Abstract

Background and Aims

Life-threatening arrhythmias are a well-established consequence of reduced cardiac sodium current (I_Na). Gene therapy approaches to increase I_Na have demonstrated potential benefits to prevent arrhythmias. However, the development of such therapies is hampered by the large size of sodium channels. In this study, SCN10A-short (S10s), a short transcript encoding the carboxy-terminal domain of the human neuronal sodium channel, was evaluated as a gene therapy target to increase I_Na and prevent arrhythmias.

Methods

Adeno-associated viral vector overexpressing S10s was injected into wild type and Scn5a-haploinsufficient mice on which patch-clamp studies, optical mapping, electrocardiogram analyses, and ischaemia reperfusion were performed. In vitro and in silico studies were conducted to further explore the effect of S10s gene therapy in the context of human hearts.

Results

Cardiac S10s overexpression increased cellular I_Na, maximal action potential upstroke velocity, and action potential amplitude in Scn5a-haploinsufficient cardiomyocytes. S10s gene therapy rescues conduction slowing in Scn5a-haploinsufficient mice and prevented ventricular tachycardia induced by ischaemia-reperfusion in wild type mice. S10s overexpression increased maximal action potential upstroke velocity in human inducible pluripotent stem cell-derived cardiomyocytes and prevented inducible arrhythmias in simulated human heart models.

Conclusions

S10s gene therapy may be effective to treat cardiac conduction abnormalities and associated arrhythmias.

Structured Graphical Abstract

SCN10A-short (S10s) overexpression in cardiomyocytes increases sodium current (I_Na) density and maximal action potential upstroke velocity (dV/dt_max), thereby improving cardiac conduction and preventing inducible arrhythmias. Left panel: Cardiomyocytes isolated from mice injected with S10s showed increased I_Na-driven dV/dt_max and increased I_Na density compared to control. Human induced pluripotent stem cell-derived cardiomyocyte (iPSC-CMs) transduced with S10s also showed increased dV/dt_max compared to control. Middle panel: S10s gene therapy increased ventricular conduction velocity in Scn5a-haploinsufficient mice. S10s gene therapy prevented conduction block in a simulated SCN5A-haploinsufficient human cardiac tissue model. Right panel: S10s gene therapy protected against ventricular arrhythmias in an ischaemia-reperfusion (I/R) mouse model and in a simulated human SCN5A-haploinsufficient whole-heart model. 3D, three-dimensional; WT, wild type.

Gene therapy, Cardiac arrhythmia, Sodium current, SCN5A, SCN10A, AAV

See the editorial comment for this article ‘Gene therapy targeting I_Na to treat life-threatening arrhythmias: beyond proof-of-concept?’, by P. Lugenbiel, https://doi.org10.1093/eurheartj/ehae930.

Translational perspective

Cardiac sodium current (I_Na) reduction is implicated in various arrhythmias including Brugada syndrome and ventricular tachycardia or fibrillation. Overexpression of SCN10A-short (S10s), a short transcript encoding the carboxy-terminal domain of the human neuronal sodium channel, leads to increased I_Na and increased maximal action potential upstroke velocity in both healthy and Scn5a-haploinsufficent cardiomyocytes. Gene therapy based on S10s expression improves cardiac conduction and prevents cardiac arrhythmias in both mouse and simulated human heart models. These results suggest that S10s gene therapy has the potential to be broadly applicable for the treatment of cardiac arrhythmias.

Introduction

Ion channels play an important role in all aspects of heart function including rhythmicity and contractility.¹ The SCN5A-encoded α-subunit of the cardiac sodium channel (Na_V1.5) largely determines the cardiac sodium current (I_Na), which is responsible for the initiation and propagation of the cardiac electrical excitation wavefront. A reduction of I_Na impairs the action potential (AP) upstroke and cardiac conduction, and is implicated in both acquired and inherited arrhythmia syndromes such as Brugada syndrome (BrS), progressive cardiac conduction disease (PCCD), sick sinus syndrome (SSS), atrial fibrillation (AF), and ventricular tachycardia or fibrillation (VT/VF).^2–5

In recent years, gene therapy has emerged as a powerful tool to restore the function of damaged or dysfunctional cells and tissues.^6–8 Restoration of cardiac I_Na by gene transfer of SCN5A represents a logical approach with the potential to provide a curative treatment. However, the large size of the SCN5A coding sequence is beyond the capacity of adeno-associated viral (AAV) vectors, complicating its clinical applicability in gene therapy. In 2021, Doisne et al. reported efficient expression of human SCN5A in neonatal mice utilizing a dual AAV-vector system.⁹ Yet the drawbacks of the dual AAV-vector system, such as its relatively low efficiency, the expression of unwanted products, and the need of two vector preparations, hamper clinical translation of this approach.¹⁰ Similarly, base editing has been successfully implemented to correct SCN5A mutations, but still requires the use of dual AAV-vector system due to the large size of the base editor,¹¹ thus also harbouring the aforementioned drawbacks.

The challenges with dual AAV-vector system inspired the search for alternative strategies involving the expression of smaller transgenes. Overexpression of bacterial sodium channels effectively improved cardiac conduction in vitro and in silico.^12–14 Yet the distinctly different gating properties could introduce pro-arrhythmic AP duration (APD) heterogeneity when expressed in vivo. Besides, anticipated higher immunogenicity poses important translational barriers. A different angle was taken in a recent study that demonstrated the potential value of the chaperone protein MOG1 as a means to suppress a BrS phenotype in Na_V1.5-trafficking defect mouse models.¹⁵ While expression of MOG1 may represent a more translatable approach, the potential pleiotropic effects of MOG1 remain an important concern that could be difficult to address on an individual basis when considering eventual patient treatment.^16–18 Moreover, MOG1 gene therapy was specifically designed to counter SCN5A trafficking mutations, which limits its application to a relatively small subset of genetic defects.

Recently, a naturally occurring cardiac-specific short transcript of SCN10A (SCN10A-short, here designated S10s) was discovered, which modulates the density of the Na_V1.5-mediated I_Na.¹⁹  S10s is expressed in the sinus node, atria, and ventricular conduction system of the heart, and comprises the last seven exons of the neuronal sodium channel gene SCN10A. The predicted coding product of S10s contains the C-terminal portion of the full channel, including part of domain III, the entire domain IV, and the cytosolic C-terminus. Loss of S10s expression was found to slow cardiac conduction in mice and reduce I_Na in isolated cardiomyocytes, while overexpression of S10s in HEK293 cells stably expressing SCN5A increased I_Na.¹⁹

In the current study, we evaluated the potential of S10s as a gene therapy target for treating cardiac conduction abnormalities and associated arrhythmias (Structured Graphical Abstract). First, we studied the effect of S10s on cellular electrophysiology (EP) and found that S10s overexpression increased I_Na density without changing the gating properties. Consequently, it rescued the phenotypes of Scn5a-haploinsufficient cardiomyocytes. Second, we observed that S10s gene therapy improved ventricular conduction and rescued conduction slowing in Scn5a-haploinsufficient mice. Third, we showed that S10s gene therapy reduced susceptibility to cardiac arrhythmias in a reperfusion arrhythmia mouse model. Finally, in an effort to work towards human application, we also demonstrated S10s-induced augmentation of I_Na in human induced pluripotent stem cell-derived cardiomyocytes (hiPSC-CMs), and its anti-arrhythmic effects using in silico human cardiac tissue and whole-heart models.^20–22

Methods

A detailed description of the methods is provided in the Supplementary material.

Results

S10s gene therapy rescues peak I_Na in Scn5a^+/Δ7bp mouse cardiomyocytes

We started our evaluation of S10s overexpression with cellular studies. In order to deliver S10s to cardiomyocytes, we cloned human S10s into a bicistronic expression cassette, which contains a self-cleaving P2A-GFP as a fluorescent marker, allowing identification and isolation of transduced cells for cellular EP studies (see Supplementary data online, Figure S1). The control vector expresses GFP alone. AAV serotype 6-S10s-P2A-GFP (AAV6-S10s) and AAV6-GFP vectors were produced (see Supplementary data online, Figure S1C, top) and injected in the apex of mouse hearts via intramyocardial injection, at a dose of 1 × 10¹¹ viral genome (VG)/mouse. Immunofluorescence staining confirmed the successful overexpression of S10s and GFP in the injected mice (Figure 1A). Haematoxylin and eosin (H&E) staining images showed moderate (immune) cell accumulation at the injection sites in both control and S10s groups (see Supplementary data online, Figure S2), suggesting a transgene-independent immune response.

Figure 1

S10s gene therapy increases sodium current density in cardiomyocytes isolated from wild type and Scn5a^+/Δ7bp mice. (A) Immunofluorescence staining images of P2A-tagged S10s in mouse left ventricles. GFP expression was observed in mice injected with AAV6-GFP and AAV6-S10s while S10s expression was only observed in mice injected with AAV6-S10s. Remote non-injected site was used as negative control. Scale bars represent 200 µm. (B) Typical sodium currents (I_Na) in cardiomyocytes isolated from mice injected with AAV6-GFP or AAV6-S10s. Each group contains 8–12 cells isolated from 3 to 4 mice. (C) Average current–voltage relationships of I_Na density. (D) Peak I_Na at −20 mV. (E) Activation curves of I_Na. (F) Inactivation curves of I_Na. Data are presented as mean ± SEM. Data were compared using two-way ANOVA with post hoc Fisher’s Least Significant Difference (LSD) test. *P < .05; **P < .01

We then performed whole-cell patch-clamp studies on GFP-positive ventricular cardiomyocytes isolated from AAV-injected mice 2 weeks post-injection to measure I_Na. Both wild type (WT) mice and mice heterozygous for a 7-bp deletion in SCN5A exon 3 causing haploinsufficiency (Scn5a^+/Δ7bp) were used.²³ Typical I_Na recordings are shown in Figure 1B and the average current–voltage relationships are shown in Figure 1C. Scn5a^+/Δ7bp mice showed significantly smaller peak I_Na density compared to WT mice [at −20 mV: WT −40.1 pA/pF vs. Scn5a^+/Δ7bp −20.8 pA/pF; predicted mean difference (PMD) −19.3; 95% confidence interval (CI) −31.2 to −7.36; P = .0022], when both injected with control vectors (Figure 1C and D), confirming the phenotype of Scn5a-haploinsufficiency. In WT mice, S10s increased peak I_Na density at −20 mV by 48.4% (GFP −40.1 pA/pF vs. S10s −59.5 pA/pF; PMD 19.4; 95% CI 7.79 to 31.1; P = .0017) (Figure 1D). In Scn5a^+/Δ7bp mice, S10s increased peak I_Na density at −20 mV by 80.4% (GFP −20.8 pA/pF vs. S10s −37.5 pA/pF; PMD 16.7; 95% CI 3.49 to 29.9; P = .015) (Figure 1D), restoring peak I_Na density to WT levels. In both WT and Scn5a^+/Δ7bp mice, the time constant of the fast and slow inactivating components and the amplitude ratio of the slow inactivation components were not significantly different (see Supplementary data online, Figure S3A–C). Moreover, overexpression of S10s did not affect voltage-dependency of activation or inactivation, as indicated by the overlapping curves that were fitted to the individual (in)activation data (Figure 1E and F), and by the absence of significant differences in their half-(in)activation voltages and slope factors (see Supplementary data online, Figure S3D). These findings indicated that S10s overexpression increases peak I_Na in Scn5a^+/Δ7bp mouse cardiomyocytes and successfully restores it to WT level.

S10s gene therapy restores action potential properties in Scn5a^+/Δ7bp mouse cardiomyocytes

To further characterize the cellular EP effect of S10s overexpression, we recorded APs (typical examples are shown in Figure 2A). Compared to WT cells, Scn5a^+/Δ7bp cells showed significantly slower maximal AP upstroke velocity (dV/dt_max; WT 214 V/s vs. Scn5a^+/Δ7bp 135 V/s; PMD 79.4; 95% CI 25.5 to 133; P = .0052) and smaller AP amplitude (APA) (WT 104 mV vs. Scn5a^+/Δ7bp 88.4 mV; PMD 15.6; 95% CI 4.88 to 26.3; P = .0057), when injected with control vectors (Figure 2B and C). No significant differences were observed in resting membrane potential (RMP; Figure 2D) or APDs (Figure 2E and F). In WT cells, S10s overexpression significantly increased dV/dt_max (Figure 2B). In Scn5a^+/Δ7bp cells, S10s overexpression rescued the phenotypes by increasing both dV/dt_max (GFP 135 V/s vs. S10s 211 V/s; PMD −76.1; 95% CI −137 to −15.3; P = .016) and APA (GFP 88.4 mV vs. S10s 104 mV; PMD −16.1; 95% CI −28.2 to −4.08; P = .010). S10s overexpression did not change the RMP or APDs in either WT or Scn5a^+/Δ7bp cells (Figure 2D–F), indicating that ionic currents other than I_Na were virtually unaffected. To study the potential side-effect of AAV transduction on cellular EP, we also recorded APs from cells of untreated WT mice and found no significant differences in all measured parameters, when compared to those injected with control vectors (see Supplementary data online, Figure S4). These data demonstrated that Scn5a^+/Δ7bp cardiomyocytes exhibited phenotypes of slower dV/dt_max and smaller APA, which S10s gene therapy successfully rescued by increasing I_Na without affecting AP parameters other than dV/dt_max and APA.

Figure 2

S10s gene therapy rescues the phenotypes of Scn5a^+/Δ7bp cardiomyocytes. (A) Typical examples of action potentials (APs) elicited at 6 Hz stimulation and their time derivatives near the upstroke (insets) of cardiomyocytes isolated from mice injected with AAV6-GFP or AAV6-S10s. Arrows indicate the maximal action potential upstroke velocity (dV/dt_max). (B–F) Action potential parameters of cardiomyocytes isolated from mice injected with AAV6-GFP or AAV6-S10s. Each group contains 5–14 cells isolated from 3 to 4 mice. (B) dV/dt_max. (C) AP amplitude (APA). (D) Resting membrane potential (RMP). (E) AP duration at 50% of repolarization (APD₅₀). (F) AP duration at 90% of repolarization (APD₉₀). Data are presented as mean ± SEM. Data were compared using two-way ANOVA with post hoc Fisher’s LSD test. *P < .05, **P < .01; ns, not significant

S10s gene therapy increases cardiac conduction velocity in WT mice

Given the importance of I_Na to cardiac impulse propagation,²⁴ we hypothesized that the S10s-induced increase in I_Na density would enhance cardiac conduction velocity (CV) at the whole organ level.

In order to achieve homogenous S10s overexpression throughout the heart, we adapted our existing AAV delivery vector in two ways: (i) the ubiquitous CMV promoter was replaced by a cardiac-specific troponin T (cTnT) promoter and (ii) we made use of the AAV cardio-tropic serotype 9 (AAV9), instead of AAV6 (see Supplementary data online, Figure S1C, bottom). We first administered the AAV vectors to 6-week-old WT mice via intravenous injection at a dose of 8 × 10¹¹ VG/mouse (Figure 3A). Overexpression of S10s in left ventricular tissue of AAV9-S10s mice was detected by RT-qPCR (Figure 3B). Two weeks post-injection, surface ECG analyses showed no significant differences in RR, PR, QRS, QT, and QTc intervals between AAV9-S10s mice and AAV9-GFP mice (Figure 3C and D). Optical mapping on isolated hearts revealed significantly increased longitudinal CV (CV_L) in the left ventricle of AAV9-S10s mice, when compared to AAV9-GFP mice (Figure 3E and F). Transversal CV (CV_T) in the left ventricle was not affected by S10s (see Supplementary data online, Figure S5A), and consequently the anisotropic ratio was significantly increased by S10s (see Supplementary data online, Figure S5D). S10s did not affect CV_L or CV_T in the right ventricle (Figure 3F and Supplementary data online, Figure S5A). H&E staining did not reveal structural differences between hearts of control and S10s treated mice (see Supplementary data online, Figure S2B). These data support the concept that S10s overexpression increases ventricular CV_L in WT mice by augmenting I_Na.

Figure 3

S10s gene therapy increases cardiac conduction velocity in wild type mice. (A) Schematic diagram of the experimental design. (B) mRNA expression level of S10s in left ventricles of wild type mice injected with AAV9-GFP or AAV9-S10s. (C) Typical ECG traces. (D) Average ECG parameters. (E) Representative activation maps of ventricles. Scale bars represent 1 mm. (F) Epicardial longitudinal conduction velocity (CV_L) measured from ventricles stimulated at 8 Hz. Data are presented as mean ± SEM. Data were compared using Student’s t-test (D) or two-way ANOVA with post hoc Fisher’s LSD test (F). *P < .05; ns, not significant

S10s gene therapy rescues conduction slowing in Scn5a^+/Δ7bp mice

We next tested if S10s overexpression could rescue the conduction slowing observed in Scn5a^+/Δ7bp mice. AAV9-GFP or AAV9-S10s were administered to Scn5a^+/Δ7bp mice via intravenous injection at an intermediate dose of 8 × 10¹¹ VG/mouse (Figure 4A). Overexpression of S10s in left ventricular tissue of AAV9-S10s mice was detected by RT-qPCR (Figure 4B). S10s protein was not detected by immunofluorescence staining (Figure 4C), probably because the expression level was below the detection threshold of the anti-P2A antibody. In order to increase the expression level, we injected Scn5a^+/Δ7bp mice with AAV9-S10s at a high dose of 8 × 10¹² VG/mouse (Figure 4A). S10s expression was detected in heart and liver by RT-qPCR (Figure 4B and Supplementary data online, Figure S6A). Cardiac S10s expression was approximately 20-fold higher in mice of the high dose group compared to that in mice of the intermediate dose group (Figure 4B). Immunofluorescence staining showed clear S10s protein expression in the left ventricles (Figure 4C). Mice injected with empty vectors (AAV9-MCS) at the same dose were used as control (Figure 4A and Supplementary data online, Figure S1C). This was to avoid the potential adverse effect of high GFP expression on cardiac conduction since GFP has previously been shown to negatively affect CV in neonatal rat cardiomyocytes cultures.²⁵ H&E staining did not reveal structural differences between hearts of control, intermediate and high dose-treated S10s mice (see Supplementary data online, Figure S2C and D).

Figure 4

S10s gene therapy improves cardiac conduction in Scn5a^+/Δ7bp mice. (A) Schematic diagram of the experimental design. (B) mRNA expression level of S10s in left ventricles of Scn5a^+/Δ7bp mice injected with AAV vectors. (C) Immunofluorescence staining images of P2A-tagged S10s. S10s expression was detected in HD_AAV9-S10s injected hearts. Scale bars represent 200 µm. (D) Average ECG parameters from Scn5a^+/Δ7bp mice injected with AAV9-GFP or AAV9-S10s. (E) Representative activation maps of ventricles. Scale bars represent 1 mm. (F) Epicardial longitudinal conduction velocity (CV_L) measured from ventricles stimulated at 8 Hz. (G) Average ECG parameters from Scn5a^+/Δ7bp mice injected with HD_AAV9-MCS or HD_AAV9-S10s. (H) Representative activation maps of ventricles. Scale bars represent 1 mm. (I) Epicardial CV_L measured from ventricles stimulated at 8 Hz. Data are presented as mean ± SEM. Data were compared using Student’s t-test (D and G) or two-way ANOVA with post hoc Fisher’s LSD test (F and I). *P < .05; ns, not significant

Two weeks post-injection, electrocardiograms (ECGs) were recorded in vivo and hearts were isolated for optical mapping. In the intermediate dose groups, surface ECG analyses showed no significant changes in RR, PR, QRS, QT, and QTc intervals (Figure 4D). Optical mapping revealed that AAV9-S10s gene therapy significantly increased CV_L in the left ventricle compared to control (Figure 4E and F). CV_T and CV_L/CV_T ratio were not changed by S10s (see Supplementary data online, Figure S5B and E). In the high dose groups, surface ECG analyses revealed significantly shortened QRS interval in high dose AAV9-S10s mice comparing to control (MCS 10.2 ms vs. S10s 9.01 ms; PMD −1.19; 95% CI −2.15 to −0.235; P = .018) (Figure 4G). No significant changes were observed between the high dose groups in regards of RR, PR, QT, and QTc intervals (Figure 4G). Optical mapping showed that AAV9-S10s gene therapy at high dose significantly increased CV_L in both left (MCS 59.1 cm/s vs. S10s 68.5 cm/s; PMD −9.37; 95% CI −17.4 to −1.31; P = .025) and right ventricles (MCS 64.7 cm/s vs. S10s 72.9 cm/s; PMD −8.21; 95% CI −16.3 to −0.154; P = .046) (Figure 4H and I), when compared to control. CV_T and CV_L/CV_T ratio were not significantly different (see Supplementary data online, Figure S5C and F). Moreover, AAV9-S10s gene therapy at high dose rescued the conduction slowing phenotype of Scn5a^+/Δ7bp mice. Scn5a^+/Δ7bp mice injected with high dose AAV9-S10s were at similar levels as WT mice injected with AAV9-GFP in QRS interval and ventricular CV_L. These data showed that S10s gene therapy improved cardiac conduction and rescued the conduction slowing phenotype in Scn5a^+/Δ7bp mice.

S10s gene therapy protects against VT in mouse hearts

Conduction abnormalities predispose the heart to unidirectional block, which is a prerequisite for re-entrant arrhythmias, the predominant mechanism of malignant ventricular arrhythmias.^26,27 We therefore hypothesized that S10s gene therapy can be employed to normalize conduction and thereby reduce susceptibility to re-entrant arrhythmias. We tested our hypothesis in an ischaemia–reperfusion (I/R) arrhythmia mouse model, in which re-entry activity was suggested as a key factor for arrhythmia genesis and sodium channel overexpression has been shown to suppress VT.²⁸ Before the controlled application of I/R injury, mice were either untreated, injected with AAV9-MCS, or injected with AAV9-S10s at high dose. I/R-induced arrhythmia testing was subsequently performed 4 weeks later. Figure 5A illustrates representative tracings from each group. Untreated and AAV9-MCS injected mice had a VT incidence of 80.0% (n = 10) and 81.8% (n = 11), respectively. Because no significant differences were expected and detected between these two groups (UT 80.0% vs. MCS 81.8%; odds ratio 0.889; 95% CI 0.119 to 6.74, P > .99), we bundled them into one control group (81.0%, n = 21) and used this for our comparison with the AAV9-S10s group. In AAV9-S10s injected mice, VT inducibility was reduced to 44.4% (n = 18), significantly lower than that in control mice (control 81.0% vs. S10s 44.4%; odds ratio 5.31; 95% CI 1.27 to 18.3; P = .024) (Figure 5B). No differences were observed in the duration of VT or the rate of VT (Figure 5C and D).

Figure 5

S10s gene therapy protects against ventricular tachycardia in mice. (A) Typical ECG tracing of ventricular tachycardia induction following ischaemia–reperfusion injury. Higher magnification tracings are provided on the right. (B) Incidence of ventricular tachycardia. Number of animals with ventricular tachycardia and total number of animals studied are presented within the associated bar charts. (C) Total ventricular tachycardia duration per animal. (D) Average ventricular tachycardia rate per animal. Data are presented as percentage (B) or mean ± SEM (C and D). Data were compared using Fisher’s exact test (B) or Student’s t-test (C and D). Untreated and HD_AAV9-MCS groups were first compared and later combined into one control group, as no significant differences were found between untreated and HD_AAV9-MCS groups.*P < .05; ns, not significant

We also examined heart weight, blood chemistry and gene expression profile from mice injected with AAV9-S10s at high dose to study the safety profile of S10s gene therapy at 4 weeks of follow-up. Heart weight, blood AST, ALT, CK-MB, and CRP levels were not changed by S10s gene therapy (see Supplementary data online, Table S1). Blood ALP levels of AAV9-S10s injected mice were significantly lower than those of untreated mice, but not different from AAV9-MCS injected mice (see Supplementary data online, Table S1). Gene expression of selected ion channels and connexins also remained unchanged 4 weeks post-injection (see Supplementary data online, Figure S7). These data demonstrated the anti-arrhythmic effects and long-term safety of S10s gene therapy in mice.

S10s overexpression increases dV/dt_max in hiPSC-CMs

In order to study the impact of S10s on EP properties of human cardiomyocytes, we made use of hiPSC-CMs. For optimal transduction efficiency, we transduced hiPSC-CMs with lentiviral vectors containing either the S10s-GFP bicistronic expression cassette (Lenti-S10s) or GFP alone (Lenti-GFP) (see Supplementary data online, Figure S1C, top). Successful delivery of S10s was confirmed by immunofluorescence staining on cells (Figure 6A). Transduced hiPSC-CMs were used for AP measurements with the patch-clamp methodology, 5–7 days post-transduction (Figure 6B). Dynamic clamp was used to inject a 2 pA/pF Kir2.1-like current to overcome the depolarized and spontaneous state of the hiPSC-CMs which hampers the functional availability of Na_V1.5 channels (for reviews see ref. 29,30, and primary references cited therein).^29–32 Similar to our results in mouse cardiomyocytes (Figure 2), S10s overexpression led to significantly increased dV/dt_max with a 73% increase compared to the GFP group (GFP 113 V/s vs. S10s 196 V/s; PMD 82.5; 95% CI 19.1 to 146; P = .013) (Figure 6C). No significant differences were observed in RMP, APA, and APDs between the groups (Figure 6D–G).

Figure 6

S10s overexpression increases action potential upstroke velocity in human iPSC-derived cardiomyocytes. (A) Immunofluorescence staining images of GFP and P2A-tagged S10s in hiPSC-CMs. S10s expression was detected in hiPSC-CMs transduced with Lenti-S10s but not in hiPSC-CMs transduced with Lenti-GFP. Scale bars represent 50 µm. (B) Typical examples of action potentials elicited at 1 Hz (left) and their time derivatives near the upstroke (right) from hiPSC-CMs transduced with Lenti-GFP or Lenti-S10s. Arrows indicate the dV/dt_max. (C–G) dV/dt_max, RMP, APA, APD₅₀) and APD₉₀ from hiPSC-CMs transduced with Lenti-GFP or Lenti-S10s. Each group contains 13–14 cells from three differentiations. Data are presented as mean ± SEM. Data were compared using Student’s t-test. *P < .05; ns, not significant

S10s overexpression increases excitability and CV in simulated human tissues

Based on our results that S10s overexpression increases CV in Scn5a^+/Δ7bp mice, and that it increases dV/dt_max in hiPSC-CMs, we hypothesized that introduction of S10s into human ventricular tissues could increase CV in SCN5A-haploinsufficient background and prevent conduction block.

We performed in silico experiments using computer-simulated 1D strands of human left ventricular cardiomyocytes to test our hypothesis. Four hundred cells with a myoplasmic resistivity of 150 Ω·cm were electrically coupled through a gap junctional conductance g_j and stimulated at one end of the strand with a square stimulus of 2 ms duration at a rate of 1 Hz (Figure 7A, inset).^20,21 In this in silico tissue model, a heterozygous loss-of-function mutation in SCN5A was simulated by a 50% decrease in Na_V1.5 conductance (labelled SCN5A^+/−), representing an entirely non-functional SCN5A mutation. We modelled the effect of S10s at ‘normal’ (1×) overexpression level with an 80.4% increase in the residual Na_V1.5 conductance to 90.2% of the WT Na_V1.5 conductance in each of the cardiomyocytes (labelled SCN5A^+/− + 1×S10s). This 80.4% increase was chosen according to the average increase in peak I_Na density observed in our voltage clamp experiments of isolated murine cardiomyocytes in Figure 1.

Figure 7

S10s gene therapy improves conduction and increases excitability in a simulated strand of human left ventricular cardiomyocytes. (A) Longitudinal conduction velocity in a simulated linear strand of human left ventricular cardiomyocytes as a function of gap junctional conductance (g_j). Complete loss-of-function mutation in SCN5A simulated by a 50% decrease in Na_V1.5 conductance (‘SCN5A^+/−’) and the effect of S10s, provided 100% successful transduction, as an 80.4% increase in remaining Na_v1.5 conductance to 90.2% of the wild type Na_v1.5 conductance (‘SCN5A^+/− + 1×S10s’). Data for 0.9×, 0.5×, and 0.1× S10s overexpression levels obtained by scaling the 80.4% increase accordingly. Individual cells of the 400-cell strand (inset) described according to the human left ventricular cell model by Ten Tusscher et al.,²⁰ as updated by Ten Tusscher and Panfilov.²¹ Myoplasmic resistivity set to 150 Ω·cm.³³ (B) Stimulus current threshold of the 400-cell strand as a function of g_j. (C) Structure of the 400-cell strand with branches at cells #200 and #201 of the strand used to simulate an increased electrical load. g_j set to 10 µS in the associated simulations. (D) Action potential conduction in the branched strand of (C) at stimulus rates of 50 (top), 100 (middle), and 150 beats/min (bottom). Action potentials of cells #100 (left) and #300 (right) of the strand, as indicated by ‘AP’ in (C), under control conditions (wild type) and in case of SCN5A^+/− and SCN5A^+/− + 1×S10s. Action potentials in case of 0.9×, 0.5×, and 0.1× S10s overexpression levels are also shown

As illustrated in Figure 7A (solid and dashed lines), modelling heterozygosity of the complete loss-of-function mutation in SCN5A resulted in a 17–22% decrease in CV over a wide range of g_j values, which was reduced to 3–4% upon the simulated application of S10s (with a transduction efficiency of 100%). This decrease in CV amounted to 4%–5%, 9%–11%, and 15%–20% when the S10s-induced increase in peak I_Na density was lowered by 10%, 50%, and 90%, respectively (mimicking 0.9×, 0.5×, and 0.1× S10s overexpression levels) (Figure 7A, dotted lines). At the same time, the mutation-induced decrease in excitability, expressed as increase in stimulus current threshold, amounted to ≈15% in absence of S10s, which was reduced to only ≈1% in presence of S10s at 1× overexpression level (Figure 7B, solid and dashed lines), further indicating that excitability could be largely restored by S10s. This increase in stimulus current threshold amounted to ≈2%, 4%–5%, and 7%–8% in presence of S10s at 0.9×, 0.5×, and 0.1× overexpression levels, respectively (Figure 7B, dotted lines).

We also constructed a linear strand with multiple branches and g_j set to 10 µS³³ (Figure 7C) that was stimulated at rates of 50–150 b.p.m. In this more demanding setting, the mutant I_Na elicited regular APs in the unbranched part of the strand at 50 and 100 b.p.m. (Figure 7D, left, top and middle panels) that were successfully conducted to the branched part of the strand (Figure 7D, right, top and middle panels). However, the mutant I_Na appeared unable to do so at 150 b.p.m. (Figure 7D, bottom panels). At this higher rate, every other AP in the unbranched part of the strand was not completely full-blown, resulting in alternating long-short APs (Figure 7D, left, bottom panel). The short AP was not successfully conducted to the branched part of the strand and this partial block thus resulted in a 2:1 conduction pattern in the strand (Figure 7D, right, bottom panel). Of note, such irregularities were not observed in the unbranched strand (data not shown) and in the case of antidromic propagation (see Supplementary data online, Figure S8). Application of S10s completely restored successful conduction, not only at 1× S10s overexpression level (Figure 7D, dashed lines) but also at significantly lower S10s overexpression levels (Figure 7D, dotted lines).

Although even at 0.1× overexpression level S10s appeared sufficient to rescue the 2:1 block that was observed at 150 b.p.m. (Figure 7D), it should be recognized that this result was obtained at a simulated transduction efficiency of 100%. In a more realistic setting, with a non-uniform mix of transduced and non-transduced cells, the result depends on the status of the cells (transduced vs. non-transduced) close to the splits in the strand, as shown in Figure 8. In the configuration of Figure 8A (top), only two nearby cells, at positions 200 and 204, are transduced (each showing the average increase in their Na_V1.5 conductance to 90.2% of the WT level, as observed in our voltage clamp experiments of Figure 1). Yet, the 2:1 block at 150 b.p.m. is still successfully rescued (Figure 8A, bottom). Conversely, in the configuration of Figure 8B (top), where many more cells are transduced throughout the strand but not near the splits, the rescue fails (Figure 8B, bottom). For the simulations of Figure 8A and B, the status of the cells 1–190 and 211–400 (transduced vs. non-transduced) was assigned in a similar pattern (compared to that shown in the configurations of Figure 8A and B), so that the overall transduction efficiency was 10% in panel A and 70% in panel B. However, the exact pattern in the remaining parts of the strand does not seriously affect the results of the simulations (data not shown).

Figure 8

Evaluation of S10s gene therapy in a setting of non-uniform transduction. (A,B) Configuration of transduced vs. non-transduced cells in the strand (top) and associated action potential conduction at 150 b.p.m. (bottom). (A) Successful rescue of the 2:1 block upon transduction of the cells at positions 200 and 204. (B) Rescue fails when none of the cells near the splits are transduced. Transduction simulated by increasing the Na_V1.5 conductance of the transduced cell to 90.2% of its WT level (i.e. the average level observed in the voltage clamp experiments of Figure 1)

Altogether, these results illustrate that S10s can prevent unidirectional conduction block at high stimulation frequencies. Such effect may be achieved already with a modest transduction efficiency, provided that the critical cells are effectively transduced. Therefore, it appears desirable to aim for a high local transduction efficiency to assure a high probability for correction of unidirectional block (as shown in Figure 7).

S10s overexpression reduces VT inducibility in a computational whole-heart model with SCN5A loss-of-function mutations

Following our demonstration of S10s’s capability to restore Na_V1.5 and increase CV and to prevent conduction block in computer-simulated 1D strands of human ventricular tissue, we advanced our investigation to a more complex and clinically relevant scenario. We hypothesize that S10s gene therapy has the potential to mitigate arrhythmia induction in a human heart affected by Na_V1.5 dysfunction. To test this hypothesis, we employed a 3D image-based computational whole-heart model and compared VT inducibility before and after the digital application of S10s gene therapy.

Firstly, we developed a novel computational cell model that incorporated the EP properties of SCN5A loss-of-function mutation (SCN5A^+/−). To model the effects of S10s gene therapy, we augmented the remaining I_Na by 80.4% in this SCN5A^+/− cell model as previously described, resulting in a post-treatment cell model referred to as SCN5A^+/− + S10s. Compared to the SCN5A^+/− model, the post-treatment SCN5A^+/− + S10s model has a higher peak current density (SCN5A^+/− + S10s 432 vs. SCN5A^+/− 242 pA/pF) and a faster upstroke velocity (SCN5A^+/− + S10s 329 vs. SCN5A^+/− 229 V/s), as shown in Figure 9A and B.

Figure 9

S10s suppresses arrhythmia induction in a 3D human whole-heart model. (A) Current–voltage relationships of I_Na. (B) The dV/dt_max of the steady-state action potentials at 1 Hz pacing. (C) Reconstructed ventricular model with three different tissue types: non-fibrotic myocardium (pink), fibrotic remodelling (grey) and dense scar (dark blue). Scar region is non-conductive; the non-fibrotic myocardium exhibits altered electrophysiological (EP) properties due to the SCN5A mutation, while the fibrotic regions include additional fibrosis-induced changes in electrophysiological properties. (D) Activation patterns of the ventricular tachycardia re-entrant circuit induced by validated in silico rapid pacing protocol at the base of the SCN5A^+/− heart model. (E) Schematics describing digital S10s treatment delivery method. The magenta patch indicates the region affected by gene therapy, with the spatial distribution of simulated injection sites visualized in cyan colour. In the right close-up figure, the semi-transparent circle represents a 1-cm radius region affected by each simulated injection. In total, 16 injections were applied to ensure full transmural coverage of the region associated with the ventricular tachycardia circuit. (F) Comparison of wavefront propagation following rapid pacing in the heart model before (top) and after (bottom) S10s gene therapy is applied. Each row shows a series of frames that depict the continuous wave propagation in a portion of the 3D whole-heart model. The images in each column show the same instant of the simulation. Yellow stars stand for the pacing location; red lines with yellow contours mark the conduction block. Electrical wave propagation trajectories were marked in white arrows. The dotted region marks the location of excitable gap that occurred at t1. For the SCN5A^+/− model, a sustained re-entry was induced. For the SCN5A^+/− + S10s model, no re-entry was induced

We then reconstructed the whole-heart model with personalized diffuse fibrosis and dense scar distributions derived from late-gadolinium enhanced magnetic resonance imaging scan, as described in previous publications^22,34 and depicted in Figure 9C. Simulations were conducted to evaluate VT inducibility in the SCN5A^+/− heart using a validated in silico rapid pacing protocol, mimicking settings in the clinical electrophysiological studies.^35,36 From the simulations, a re-entrant circuit was induced in the basal posterolateral wall of RV, as shown in Figure 9D. We initially applied S10s gene therapy (with a transduction efficiency of 100%) globally to the entire heart and found that this treatment successfully terminated the previously induced VT (data not shown). Next, we investigated whether locally applied S10s gene therapy could achieve a similar anti-arrhythmic effect. S10s gene therapy was applied with 16 injections to fully cover the region associated with the arrhythmia circuit (Figure 9E). The simulation was repeated, and results revealed that after locally applying the S10s gene therapy, the previously induced VT got effectively terminated.

To gain a deeper insight into the role of S10s in preventing VT induction, we conducted a detailed analysis of the re-entry simulated under SCN5A^+/− and SCN5A^+/− + S10s conditions, as depicted in Figure 9F. For both simulations at t0, the wavefronts elicited at the basal-inferior pacing sites encounter preceding wavetails that have been slowed by a region of diffuse fibrosis. This transient conduction block causes the wavefronts to curl clockwise. In the SCN5A^+/− case (Figure 9F, top row), the slower CV due to the I_Na alteration allows the formation of an excitable gap that provides enough time for the tissue to recover and become stimulated again by the curling wave. As a result, a re-entrant circuit evolves and anchors around an adjacent region of dense scar. In contrast, the SCN5A^+/− + S10 case (Figure 9F, bottom row) has a recovered I_Na and CV, allowing the wavefront to curl fast enough to encounter still refractory tissue. This functional block of the curling wavefront prevents the formation of a re-entrant circuit.

Discussion

Our present work evaluated the potential of S10s overexpression as a gene therapy for cardiac arrhythmias that stem from reduction in I_Na. Our data from ex vivo, in vivo, in vitro, and in silico experiments provide support for the concept that introduction of S10s in ventricular cardiomyocytes increases I_Na, thereby improving cardiac conduction and importantly compensating for I_Na insufficiency and mitigate the associated risk for arrhythmias.

Reductions in cardiac I_Na are implicated in various arrhythmia syndromes such as BrS, PCCD, SSS, AF, and scar or cardiomyopathy-related VT/VF.^2–5 Sodium channel gene transfer could provide an effective therapy to improve cardiac conduction,³⁷ yet the development of such an intervention is difficult because the coding sequences of sodium channels exceed the packaging capacity of AAV. Although dual vector systems have been developed to expand AAV packaging capacity,³⁸ and SCN5A has been successfully implemented in such a system,⁹ translation remains complicated due to the lower intrinsic efficiency (in comparison to a single vector), much higher costs and potential additional safety issues due to the need of a higher dose. Notably, Doisne et al. used a high AAV-vector dose in neonatal mice (comparable to our high dose in adult mice),⁹ while cardiac transduction is more easily realized in neonatal mice as compared to adult mice.^11,39,40 It is therefore reasonable to anticipate that a higher dose is required to achieve a similar effect in adult recipients. As a result, we and others have continued to search for smaller transgenes that can correct sodium channel dysfunction.

The S10s transcript was identified in the hearts of mouse and human as a result of investigations into genetic variants associated with conduction abnormalities and BrS, and was shown to represent an endogenous mechanism to modulate Na_V1.5-mediated I_Na  in vitro and in vivo.¹⁹ Compared to previous approaches, S10s has several advantages as a gene therapy target. First, it has a coding sequence of 2 kb, which easily fits into a single AAV vector, avoiding the use of a dual AAV-vector system. Second, it increases I_Na density without altering the gating kinetics or APD in cardiomyocytes. This assures APD homogeneity even when it is expressed at varying levels caused by inhomogeneous transduction. Third, S10s is only effective in the presence of Na_V1.5 and does not generate any I_Na when expressed alone.¹⁹ This lowers its potential side-effect in case off-target expression occurs. Indeed, we observed hepatic expression of S10s transcripts when applied at high dose but found no elevated liver damage parameters in blood chemistry. Finally, as an endogenously expressed gene product in mammals including humans, S10s is expected to be well tolerated by its recipients. Nevertheless, the mechanism of action of S10s is not fully understood. Because S10s does not affect Scn5a expression, and the S10s gene product co-localizes with Na_V1.5 on the cell membrane,¹⁹ it is likely that it increases I_Na density by physically interacting with Na_V1.5 proteins. Preliminary in vitro data from our lab showed no increase of Na_V1.5 membrane expression by S10s overexpression, suggesting a mechanism other than facilitated trafficking, although further studies are required to gain additional mechanistic insight into the function of S10s.

In this study, we found that S10s gene therapy significantly increased CV_L but not CV_T in mouse heart. This might be due to that Na_V1.5 locates more abundantly in the intercalated disc than in the lateral membrane,^41–43 since S10s is only effective in the presence of Na_V1.5.¹⁹ In fact, this different response to sodium channel abundancy between CV_L and CV_T is not new. Overexpression of Na_V1.4 in canine heart has been reported to only increase CV_L due to its presence in intercalated disks.⁴⁴ Also in two different Scn5a heterozygous mouse models, only CV_L was significantly decreased but not CV_T.^23,45 Furthermore, this increase in CV_L in mouse heart was anti-arrhythmic as demonstrated in Figure 5.

Loss-of-function mutations in SCN5A lead to reduction of I_Na in cardiomyocytes. This group of mutations can be divided into several categories based on their mechanism of effect. The first category is represented by mutations that generate a premature stop codon by nonsense mutations or frameshifts (e.g. L1393X, C280S*fs61).^46,47 Such mutations lead to loss of protein expression and often cause more severe phenotypes than others.^48,49 Our Scn5a^+/Δ7bp mouse model and the simulated human tissue models both represent this category, where the conduction slowing phenotype was largely rescued, demonstrating the usefulness of S10s gene therapy for these conditions. The second category is represented by mutations that generate non-functional channels due to gating or permeation defects (e.g. R878C, G1408R)^50,51 and the third category is represented by mutations that impair the cell surface localization of Na_V1.5 (e.g. G1743R, D1275N).^52,53 For mutations belonging to these two categories, we anticipate that S10s gene therapy could compensate the reduced I_Na by interacting with the WT Na_V1.5, as almost all human patients are heterozygous for their SCN5A mutation(s). Since these mutations do not generate functional channels, it is unlikely that S10s would interfere other than just augment the WT channel. The fourth category is represented by loss-of-function based on altered gating kinetics (e.g. G514C, I890T),^54,55 involving activation at more depolarized membrane potentials or inactivation at less depolarized potentials. Most uncertainty remains with regard to potential interaction of S10s with these mutated channels. S10s gene therapy might be beneficial for treating these mutations but further investigation of the interaction between S10s and these mutant channels is needed to better understand the potential therapeutic value in this context.

In addition to the loss-of-function mutations in SCN5A, cardiac I_Na can be reduced due to other causes as well. Genetic variants in non-coding regions have been reported to reduce SCN5A expression by altering transcription factor binding sites,^23,56,57 or by creating microRNA binding sites.⁵⁸ Since these regulations only affect the expression level but not the biophysical properties of Na_V1.5, we anticipate S10s gene therapy could be beneficial for patients carrying such variants and mutations.

Besides the genetic causes, I_Na reductions can also be caused by remodelling after cardiac injury such as myocardial infarction, contributing to the associated VT/VF.⁵⁹ Locally overexpressed sodium channels and associated conduction restoration could prevent re-entrant arrhythmias in animal models.^4,28 Our 3D whole-heart in silico experiment suggests that application of S10s gene therapy eliminates such arrhythmias in an SCN5A-haploinsufficient human heart with fibrotic scar tissues. This result further underlines the anticipated applicability of S10s gene therapy to restore conduction and prevent malignant arrhythmias. In this context, delivery via direct intramyocardial injections seems particularly appealing, because of the high transduction efficiency that can be achieved with AAV6 gene transfer (as reported in the literature and confirmed by our own observations; AAV6 appears to be the most potent WT serotype, while more potent variants are under development).^60–62 The anticipated dose needed for injection in a region with slow/abnormal conduction is 1 × 10¹² VG/injection,^61,63 which is significantly lower than what is currently applied in cardiac gene therapies that target the heart via the systemic circulation (e.g. 6.7 × 10¹³–1.1 × 10¹⁴ VG/kg in a recent clinical trial).⁶⁴ Moreover, areas of abnormal conduction are relatively easily identified using electroanatomical mapping, which is common practice in clinical cardiac EP procedures. Functional validation of our proposed approach of direct intramyocardial injections with AAV-S10s gene therapy in porcine models of malignant arrhythmias therefore represents a critical next step before application to human subjects can be considered.^65,66

The following limitations of our study should be recognized. First, the mechanism of S10s function remains incompletely understood. Further in depth biochemical and cell biological studies are needed to reveal details of this mechanism. Second, to test the anti-arrhythmic effect, our S10s gene therapy was applied prior to injury in the mouse I/R model, which is different from the anticipated clinical scenario where it will be applied after the disease onset and in a different setting. Third, the relatively small group size in the mouse VT experiment did not have adequate power to detect anything but huge differences. Therefore, the analysis could not rule out the presence of potentially clinically meaningful differences. Fourth, our 3D whole-heart studies could not model cellular heterogeneity in transduction. Yet the finding that S10s was still capable in preventing unidirectional block at relatively low transduction efficiencies was reassuring. Moreover, the transduction efficiency with intramyocardial injection of AAV6 is typically very high, i.e. a transduction efficiency of >80%. Last, here we performed in vivo testing only in murine models. Further translational testing in large animal models of cardiac arrhythmias represents a high priority to enable clinical trials.

In conclusion, we developed an AAV-based gene therapy utilizing S10s, a naturally occurring transcript, for the treatment of sodium current insufficiencies. In mouse and human models, we showed that S10s gene therapy increases cardiac CV and prevents cardiac arrhythmias by increasing I_Na in cardiomyocytes. Our results present S10s as a promising gene therapy target for treating cardiac arrhythmias based on conduction abnormalities, and we propose local AAV-mediated gene transfer as the most viable route of administration to treat human subjects.

Supplementary data

Supplementary data are available at European Heart Journal online.

Declarations

Disclosure of Interest

O.F.K., H.L.T., and G.J.J.B. report ownership interest in PacingCure B.V. J.W., A.R.B., K.N., O.F.K. and H.L.T are employees of PacingCure B.V. J.W., V.M.C., P.B., and G.J.J.B. filed patent applications concerning S10s gene therapy. Other authors declare no conflicts of interest.

Data Availability

The data supporting this study's findings are available from the corresponding authors upon reasonable request.

Funding

This work was supported by European Research Council (starting grant 714866 and associated proof-of-concept grant 899422 to G.J.J.B.), Health Holland (LentiPace II to G.J.J.B. and H.L.T.), Horizon 2020 (Eurostars E114245 and E115484 to G.J.J.B. and V.M.C.), Dutch Research Council (Open Technology Program 18485 to H.L.T. and G.J.J.B.), European Innovation Council (Pathfinder Project 101115295 - Nav1.5-CARED and TRANSITION - Project 101099608 – TRACTION to V.M.C. and G.J.J.B.), Dutch Research Council (OCENW.GROOT.2019.029 to V.M.C.), the Netherlands CardioVascular Research Initiative wtih support from Dutch Heart Foundation and Stichting Hartekind (CVON2019-2 OUTREACH to V.M.C.), the Netherlands CardioVascular Research Initiative (CVON-PREDICT2 2018-30 to C.A.R. and C.R.B.), Fondation Leducq (17CVD02 to C.A.R. and C.R.B., and 23CVD04 to N.A.T.), Stichting De Merel (to E.M.J.M. and C.R.B.); Dutch Heart Foundation (Dekker Fellowship T2020-23 to H.D.D., and PPP-Grant PREVENT, to C.A.R and G.J.J.B.), National Institutes of Health (R01HL166759 and R01HL174440 to N.A.T.), and National Science Foundation (DMS-2436738 to N.A.T.).

Ethical Approval

Animal care and experiments conform to Directive 2010/63/EU. All animal work was approved by the Animal Ethics Committee of the Academic Medical Center, Amsterdam and was performed in compliance with the Dutch government guidelines.

Pre-registered Clinical Trial Number

None supplied.

Author notes

Jianan Wang and Arie O Verkerk have contributed equally to this work.

Phil Barnett and Gerard J J Boink have contributed equally to this work.