Emerging Gene Therapies for Inherited Arrhythmia Syndromes

Authors, Journal, Affiliations, Type, DOI

• Authors: Luca Grisorio — Department of Molecular Medicine, University of Pavia, Italy; Jason D. Roberts — Population Health Research Institute, Hamilton Health Sciences and McMaster University, Ontario (Marianne Barrie/PHRI Chair in ARVC Research)
• Journal: Canadian Journal of Cardiology. 2026;42:982–1003
• Type: Invited narrative review; no systematic search or evidence grading
• DOI: https://doi.org/10.1016/j.cjca.2026.01.003
• Disclosures: none declared

Overview

A field survey of nucleic-acid therapeutics for inherited arrhythmia syndromes, organised by the three modalities — gene replacement, small-nucleic-acid silencing, and gene editing — and then by disease. The translational position it documents is that three conditions now have human trials running: PKP2-ARVC (three Phase I programmes), RYR2-CPVT (AAV8-CASQ2, Phase Ib, including a paediatric cohort), and PLN p.R14del cardiomyopathy (subcutaneous siRNA, Phase I).

Its most valuable original contribution is not the target list but the quantitative treatment of heterogeneous transduction. Because AAV transduction is never uniform, the review assembles computational and animal evidence that only 30–50% of cardiomyocytes need correcting to abolish triggered arrhythmias — far below the >70% required for structural repair — while warning that the same non-uniformity can create the dispersion of repolarisation that causes reentry. That tension, rather than vector efficiency, is framed as the central unsolved problem in the field.

Keywords

Gene therapy · cardiac arrhythmias · channelopathies · cardiomyopathy · AAV · precision medicine · sudden cardiac death

Key Takeaways

Gene therapy modalities

• AAV cargo ceiling is ~4.5 kb, which excludes full-length RYR2 (~15,000 nt) and SCN5A (~6,048 bp) and forces pathway-targeted or split-vector workarounds.
• Suppression-replacement (Sup-Rep) is a single AAV carrying an shRNA that knocks down the endogenous gene plus an interference-immune (shIMM) cDNA — one construct can serve every pathogenic variant in a gene.
• Base and prime editors convert C·G→T·A or A·T→G·C and make small substitutions/indels without double-strand breaks; their size usually mandates dual-AAV split-intein delivery.
• Safety has become the binding constraint. In May 2025 Rocket Pharmaceuticals reported a death in the pivotal phase 2 RP-A501 trial in Danon disease, potentially related to a newly added C3 inhibitor in the pre-treatment immunosuppressive regimen. In the same period the FDA disclosed three deaths from acute liver failure after Sarepta's AAVrh74 therapies for Duchenne and limb-girdle muscular dystrophy. The authors read these as dose-selection, capsid-engineering and immunomodulation problems rather than as a verdict on the platform.

CPVT

• Allele-specific silencing works but does not scale. siRNA against RYR2 p.R4496C reduced mutant transcript 50% and total RYR2 protein 15%, and abolished arrhythmias in neonatal and adult mice; modelling predicted fully mutant homotetramers falling from 6.25% to 1.23%. CRISPR-SaCas9 against a linked silent restriction site eliminated the phenotype with only ~47% of cardiomyocytes transduced. But each mutation needs its own construct, and CPVT1 has a large variant repertoire.
• The CRDS problem is explicit. RYR2 loss-of-function causes calcium release deficiency syndrome, so any silencing strategy risks converting a gain-of-function disease into a loss-of-function one. The authors note that knockdown models showed no CRDS-like electrophysiological remodelling and argue for a "permissible therapeutic window", but recommend organ-level CRDS susceptibility testing before clinical use.
• Mutation-agnostic alternatives work across genotypes. AAV9 delivering the autocamide inhibitory peptide 2 (AIP) under a cardiac troponin T promoter confined CaMKII inhibition to cardiomyocytes and eliminated arrhythmias in p.R176Q mice, and reverted calcium handling in hiPSC-CMs carrying five different RYR2 variants.
• CASQ2 overexpression is the clinical candidate. It protects RYR2 p.R4496C mice, and in silico work suggests only 1.5–2-fold overexpression is needed — far below the 10–20-fold that causes hypertrophy in transgenic mice.
• CPVT2 needs replacement, not silencing. AAV9-CASQ2 restored junctional SR ultrastructure and gave protection to 12 months in CASQ2 p.R33Q mice; a transduction rate above 33% prevented arrhythmia entirely while lower rates still prevented sustained VT.
• Autosomal dominant CASQ2 variants have now been described — they preserve dimerisation but block polymerisation, and are proposed to act as a "poison peptide". Whether overexpression rescues them is untested, and a Sup-Rep approach is the fallback.

Long QT and short QT syndromes

• Sup-Rep is the scalable answer. A single lentiviral construct silences KCNQ1 and expresses shIMM KCNQ1, shortening action potentials in hiPSC-CMs carrying four different LQT1 mutations, and has been extended to LQT2 and SQT1 cells. In LQT1 transgenic rabbits, AAV9-Sup-Rep normalised QT and QT dispersion at baseline and after isoproterenol with only ~30% reduction in mutant KCNQ1 mRNA. In SQT1 rabbits, 60% silencing of the mutant transcript recovered QTc and apicobasal APD₉₀ heterogeneity, with I_Kr normalised in 47% of cells.
• Two caveats the authors raise against their own optimism: the LQT1 rabbit study used a ubiquitous U6 promoter, so silencing was not confined to cardiomyocytes; and QT normalisation was demonstrated without comprehensive testing of arrhythmia susceptibility.
• Base editing offers a steeper dose–response than expected. In SCN5A p.T1307M mice, mRNA editing efficiency up to 99% normalised QTc at >60%, yet homozygotes edited at only 20% — without QT normalisation — were still completely protected from carbachol-induced VT.

Calmodulinopathies

• Three genes, one identical protein — so silencing one CALM gene is tolerable, and the redundancy is what makes the disease treatable.
• CRISPR interference against the CALM2 promoter normalised action potentials and Ca²⁺/CaM-dependent L-type channel inactivation in patient hiPSC-CMs, sparing CALM1 and CALM3.
• A single universal Sup-Rep construct knocked down all three endogenous CALM genes with three shRNAs while expressing shIMM CALM1, and normalised action potential duration in all three CALM-LQTS subtypes.
• ASO knockdown is the most tunable option. A CALM1-selective ASO conjugated to palmitic acid normalised action potentials in CALM1 p.F142L hiPSC-CMs and abolished caffeine/adrenaline-induced bidirectional VT in p.Asn98Ser knock-in mice — but required >85% silencing of the mutant transcript, with compensatory up-regulation of CALM2 and CALM3. The effect lasted 4 weeks.

Brugada syndrome

• Gene replacement is impossible — SCN5A exceeds AAV capacity — so the field targets Nav1.5 trafficking or total sodium current instead.
• MOG1 (a Nav1.5 chaperone) delivered by AAV9 normalised the phenotype in SCN5A p.G1746R knock-in mice (J waves eliminated, sinus node behaviour normalised, late sodium current changes) and improved the cardiomyopathy, sinus node dysfunction and conduction disease of SCN5A p.D1275N mice.
• SCN10A-short — a 7-exon protein generated from an intronic SCN10A enhancer — increases I_Na and restored upstroke velocity, peak I_Na density and conduction velocity in SCN5A haploinsufficient mice. The authors flag that the same construct reduced reentrant VT after ischaemia-reperfusion, so it may serve both inherited and acquired disease.

Arrhythmogenic cardiomyopathies

• PKP2-ARVC is the most advanced. Truncating variants predominate and act through haploinsufficiency, which is precisely the mechanism replacement can fix. AAVrh74, AAV9 and AAVrh10 programmes all rescued Pkp2 knockout and knock-in mice; treatment after disease onset (2.5 weeks post-deletion) prevented further functional decline and prolonged survival, but did not eliminate vulnerability to ventricular arrhythmia — a possible sign that an established substrate cannot be removed.
• TMEM43 p.S358L illustrates the opposite logic. Truncating TMEM43 variants occur in the general population without disease, implying haploinsufficiency is tolerated, while the p.S358L missense is malignant. So the field is pursuing WT TMEM43 overexpression (which extended median survival from 23.9 to 34.6 weeks in double-transgenic mice, reduced fibrosis, and shortened P-wave and QRS duration) and allele-specific knockdown of p.S358L as alternatives.
• PLN p.R14del is not straightforwardly gain-of-function. The longstanding model — constitutive SERCA2a inhibition — has been revised: inhibitory capacity may actually be impaired, with disease arising from aberrant SR/ER cluster formation. Silencing ≥50% of phospholamban is tolerated, which is what makes knockdown viable: AAV CRISPR reduced mutant expression 15–20% with functional benefit, an ASO cut PLN protein 50% and tripled median survival (7.5 → 24 weeks), and DWORF overexpression — which counters phospholamban and activates SERCA2a — extended survival to ~17–20 weeks through reducing aberrant SR clusters rather than through SERCA2a activation.

Human clinical trials

• All three PKP2 trials require an existing ICD, exclude patients with elevated anti-AAV neutralising antibodies, and give a single dose with up to 5 years of follow-up. Efficacy endpoints are ventricular arrhythmia burden, ICD discharges and myocardial PKP2 expression on biopsy.
• The CPVT1 trial is notable for enrolling children aged 7–17 in its third cohort and for keeping participants on optimal medical therapy, with the first two patients in each cohort carrying an ICD.
• PULSE is the only non-viral, re-dosable programme in the table and the only one delivering by subcutaneous injection.

The therapeutic threshold to prevent arrhythmias

This is the review's conceptual core and is treated on its own page: concepts/Gene-Therapy-Transduction-Threshold.

• In a 1-D strand, 70–80 contiguous cardiomyocytes must simultaneously reach a suprathreshold DAD or EAD to initiate a propagating action potential; in 2-D and 3-D the requirement rises to ~7,000–8,000 and ~700,000–800,000 cells. This source–sink buffering is why random triggered activity rarely propagates in healthy tissue.
• In silico, suppressing ectopic action potentials in a DAD-prone cable required 30–50% of cells to be "stabilisers", falling to 30–40% in 2-D and 3-D — and rising again when stabilisers were clustered rather than dispersed, and when gap junction conductance was reduced or fibrosis simulated.
• Animal data agree: CPVT models achieved complete suppression of adrenergic arrhythmia at 40–66% transduction, and CASQ2 p.D307H mice prevented sustained VT below 33%.
• But the same heterogeneity can be proarrhythmic. Where the therapy alters action potential duration, non-uniform correction increases APD dispersion and therefore reentry vulnerability — seen in silico when stabiliser cells were clustered or when an untreated central region was modelled.
• Consequence for structural disease: fibrosis and reduced gap junction coupling raise the threshold for suppressing triggers and independently favour reentry, so gene therapy for cardiomyopathy is likely to work best early, before remodelling is established.

Future directions

• Post-MI VT: strategies either prolong refractoriness locally (dominant-negative KCNH2 p.G628S) or improve conduction in the border zone (Nav1.4, Cx32, SCN10A-short, SERCA2a). None has reached clinical trials, chiefly because of the difficulty of achieving precise temporal and spatial transgene control; short-acting mRNA is proposed as the solution.
• Atrial fibrillation: epicardial adenoviral dominant-negative KCNH2, Cx43 and Cx40 overexpression, and AAV9 siRNA knockdown of the atrial-specific TASK-1 channel have all reduced AF burden in pigs. The obstacles are atrial-specific delivery and the impossibility of repeat dosing with AAV.
• The authors' closing judgement is that inherited arrhythmias are the ideal test case — monogenic, severe, and poorly served by lifelong drugs and devices — but that extending the approach to acquired disease depends on tunable, re-dosable, non-viral platforms.

Limitations of the document

• Invited narrative review with no systematic search, no formal quality appraisal, and no evidence grading. Treatment claims are the authors' synthesis; there is no PRISMA process behind them.
• Almost the entire evidence base is murine. The exceptions are two transgenic rabbit studies (LQT1, SQT1) and hiPSC-CM work; human anatomical scale, cardiac tropism, and anti-AAV seroprevalence are unaddressed, and the authors themselves call for large-animal models.
• Transduction efficiency is frequently inferred rather than measured. Several studies (the LQT1 rabbit Sup-Rep work in particular) report a therapeutic effect without quantifying the proportion of transduced cardiomyocytes, which undermines the quantitative threshold argument the review is built on.
• The clinical trial table is a moving target. Phase I, ascending dose, safety-primary; with 9–18 patients per programme, no efficacy conclusion can be drawn for years, and the review does not discuss what a positive result would look like.
• Safety is under-weighted relative to efficacy. The 2025 AAV deaths are reported in the modality section but do not inform the per-disease optimism, and no risk–benefit framework is offered for a first-in-human therapy in a population that already has an ICD.
• Paediatric durability is not addressed — the CPVT1 trial will treat children aged 7–17 with a non-integrating vector whose episomal copies dilute with cardiomyocyte growth, and re-dosing is blocked by the immunity the first dose creates.
• No cost, manufacturing, or access analysis, despite all five trials being single-dose systemic AAV or siRNA programmes at doses near or above the immunogenic threshold.

Key Concepts Mentioned

concepts/Gene-Therapy-Transduction-Threshold · concepts/SupRep-Therapy · concepts/Gene-Silencing-Therapy · concepts/AAV-Gene-Delivery · concepts/CRISPR-Cas9-in-Channelopathies · concepts/Gene-Editing-Risk-Benefit-Framework · concepts/Calmodulinopathy · concepts/iPSC-Derived-Cardiomyocytes · concepts/Tachycardia-Induced-Cardiomyopathy

Key Entities Mentioned

entities/CPVT · entities/RYR2 · entities/CASQ2 · entities/Long-QT-Syndrome · entities/Short-QT-Syndrome · entities/Brugada-Syndrome · entities/SCN5A · entities/PKP2 · entities/TMEM43 · entities/PLN · entities/ARVC · entities/KCNQ1 · entities/KCNH2 · entities/Atrial-Fibrillation

Named but unlinked: DSC2 · DSP · FLNC · LMNA · RBM20 · GJA1 · GJA5 · SCN10A · MOG1 · DWORF · TECRL · TRDN · HEROIC-PKP2 · RIDGE-1 · RP-A601 · PULSE

Wiki Pages Updated

• concepts/Gene-Therapy-Transduction-Threshold — created
• concepts/SupRep-Therapy — universal CALM construct, LQT1 rabbit caveats, new contradictions
• concepts/AAV-Gene-Delivery — 2025 AAV fatalities, cargo/modality table, transduction-threshold cross-link
• concepts/Gene-Silencing-Therapy — palmitic-acid conjugation, the CALM ASO, tunability
• concepts/Calmodulinopathy — CRISPRi, universal Sup-Rep, ASO
• entities/PKP2 — three named Phase I trials with designs
• entities/PLN — PULSE trial, mechanism revision, ASO and DWORF
• entities/TMEM43 — gene replacement data and the haploinsufficiency-tolerance argument
• entities/CPVT — CASQ2 Phase Ib trial, CRDS safety concern, CaMKII AIP
• entities/CASQ2 — autosomal dominant variants and the transduction threshold
• entities/Brugada-Syndrome · entities/SCN5A — SCN10A-short as a second sodium-current route
• Genetics-index · wiki/wikiindex · wiki/sourceindex