Beneficial action potential duration–shortening effects, but deleterious negative inotropism of I_Ks-activator docosahexaenoyl glycine in long QT syndrome type 2

Abstract

Aims

Loss-of-function mutations in KCNQ1 and KCNH2 (α-subunits of the slow delayed I_Ks and rapid delayed I_Kr-conducting repolarising K⁺ channels) lead to long QT syndrome type 1 (LQT1) and 2 (LQT2), respectively. These channelopathies present with longer action potential duration (APD) and prolonged QT interval on electrocardiogram, which can ultimately lead to deadly arrhythmias. Here, we investigated the therapeutic potential of the polyunsaturated fatty acid docosahexaenoyl glycine (DHA-gly) in normalizing APD and QT interval in LQT2 by increasing I_Ks.

Methods and results

The effects of DHA-gly on electrical and mechanical parameters were assessed in Xenopus laevis oocytes, wild-type (WT), LQT1 (KCNQ1-Y315S), and LQT2 (KCNH2-G628S) transgenic rabbit models and human induced pluripotent stem cell-derived cardiomyocytes (hiPSC-CMs). DHA-gly increased I_Ks in oocytes and WT ventricular cardiomyocytes (VCMs) in a dose-dependent manner. Consequently, DHA-gly shortened APD in vitro and QT interval ex vivo in WT and LQT2 rabbits, but not in LQT1. However, DHA-gly was unable to reduce arrhythmia formation in LQT2. Beneficial APD/QT shortening effects were accompanied by a detrimental decrease in both cellular and ventricular contraction across all genotypes, including LQT1, which could be due to a shortening in Ca²⁺ transient duration observed in VCMs and hiPSC-CMs.

Conclusion

DHA-gly–induced I_Ks enhancement shows promising results in shortening APD/QT in LQT2 rabbits, while having no effect on LQT1 (impaired I_Ks). However, its adverse effect on cardiac contractility, even in LQT1, makes it unsuitable to treat LQTS patients. Our study highlights the importance of considering both electrical and mechanical effects of potential therapeutic compounds prior to clinical translation.

Graphical Abstract

Graphical Abstract.

Translational perspective.

Beta-blockers are used as first-line of treatment for LQTS to reduce arrhythmias but are not effective in each patient and do not target the underlying cause, i.e. the prolonged cardiac APD. In this study, we demonstrate that the polyunsaturated fatty acid (PUFA) docosahexaenoyl glycine (DHA-gly) increases I_Ks, thereby shortening APD and QT interval in LQT2. While we observe unwanted off-target effects with DHA-gly-induced decreased cardiac contractility, we confirm that PUFA derivatives with their APD/QT-shortening effect could be a promising therapeutic strategy to treat LQT2 in the future, if the absence of deleterious effects on mechanical function is confirmed.

What’s new?

• Docosahexaenoyl glycine (DHA-gly) increases rabbit I_Ks in Xenopus oocytes and isolated ventricular rabbit cardiomyocytes.

• This increase leads to a shortening of APD in vitro and QT interval ex vivo in LQT2 (with intact I_Ks) but not LQT1 (absent I_Ks) and a decrease in short-term variability of the APD.

• Deleterious effects of DHA-gly on cardiac contractility and intracellular Ca²⁺ dynamics might mask potential antiarrhythmic effects as DHA-gly does not reduce arrhythmia inducibility in LQT2 at the organ level.

• These data indicate that contractility screening should be a requisite for all promising PUFAs prior to clinical translation.

Introduction

Congenital long QT syndrome (LQTS) is an arrhythmogenic cardiac disease characterized by a prolonged QT interval on the electrocardiogram (ECG) due to impaired cardiac repolarization. LQTS patients carry a heightened risk for arrhythmogenic syncope and life-threatening arrhythmias, such as torsades de pointes, that can lead to sudden cardiac death.^1–3 More than 80% of LQTS cases are caused by loss-of-function variants in the KCNQ1 or KCNH2 genes, encoding for the α-subunits of the slow delayed I_Ks-conducting or rapid delayed I_Kr-conducting K⁺ channels, classified as LQT1 and LQT2, respectively.^1,3 Reduced or absent I_Ks or I_Kr leads to prolonged ventricular action potential duration (APD), which in turn leads to a prolonged QT interval.⁴ Beta-blockers are used as the first-line treatment for LQT1 and LQT2 to reduce arrhythmias.^1,5,6 While they exert their antiarrhythmic effect by preventing adrenergic-triggered early afterdepolarizations (EADs), they do not target the underlying cause of arrhythmias, i.e. the prolonged cardiac APD. Additionally, despite appropriate beta-blocker therapy, syncope and aborted cardiac arrest may occur, especially in patients who were symptomatic prior to pharmacological therapy.⁷ Hence, there is a persisting, unmet need for alternative pharmacological therapies in LQTS that can normalize the QT interval and abolish arrhythmias triggered by prolonged QT.

Polyunsaturated fatty acids (PUFAs) have long been considered as antiarrhythmic compounds,^8–11 to the extent that the American Heart Association recommends eating fish rich in oil (natural PUFAs) twice a week to reduce the risk of developing cardiovascular diseases.¹² As amphiphilic molecules, PUFAs can target and modulate several cardiac ion channels.¹³ Notably, it has been suggested that their antiarrhythmic effect is due to the inhibition of voltage-gated Na⁺ and L-type Ca²⁺ channels.^13–15 Interestingly, over the last decade, PUFAs have been shown to also increase the slow delayed rectifier K⁺ repolarizing current I_Ks by electrostatic interactions between the negative charge of the PUFAs head group and KCNQ1 channels.^16–22 While natural PUFAs, such as docosahexaenoic acid (DHA), already have an activating effect on I_Ks, modified PUFAs, in which the head group is replaced, such as N-arachidonoyl taurine (N-AT) and DHA-glycine (DHA-gly), show an even stronger I_Ks activating effect.¹⁸ Hence, PUFAs might represent a new class of compounds to treat LQTS. By increasing the I_Ks current, they can accelerate cardiac repolarization, which shortens APD and the corresponding QT interval.

Therapeutic use of PUFAs in LQTS has already been investigated in a drug-induced LQT setting in guinea pigs, in which DHA, DHA-gly, and N-AT all shortened the QT interval ex vivo with a greater effect of DHA-gly.²³ We also recently demonstrated that DHA normalizes the QT interval and APD in our transgenic LQT2 rabbit model (KCNH2-G628S, loss of I_Kr),²⁴ both ex vivo and in vivo, while having no effect in our transgenic LQT1 rabbit model (KCNQ1-Y315S, loss of I_Ks),²⁴ in which the I_Ks could not be enhanced due to the dominant negative loss-of-function variant.²⁵

In this study, we investigated the potential of the modified PUFA DHA-gly to normalize cardiac repolarization in our transgenic LQT1 and LQT2 rabbit models.²⁴ We show that DHA-gly shortens APD in isolated ventricular cardiomyocytes (VCMs) and the QT interval in ex vivo hearts from LQT2 rabbits, but not LQT1. Interestingly, while DHA-gly reduces short-term variability (STV) of the APD—a surrogate for antiarrhythmic effects—in LQT2 at the cellular level, its antiarrhythmic effects are not reflected at the whole-heart level. As a potential mechanism underlying the unexpected lack of antiarrhythmic effects at the whole-heart level, we demonstrate that DHA-gly decreases cardiac contractility and disrupts intracellular Ca²⁺ dynamics in healthy wild-type (WT), LQT1, and LQT2 rabbits, indicating unwanted off-target effects.

Methods

Ethical use of animal models

Adult transgenic LQT1 (KCNQ1-Y315S, loss of I_Ks),²⁴ LQT2 (KCNH2-G628S, loss of I_Kr),²⁴ and WT New Zealand White rabbits of both sexes, aged from 3 months to 1 year, were used in this study. All rabbit experiments were performed in compliance with EU legislation (directive 2010/63/EU) and the German (TierSchG and TierSchVersV) and Swiss Animal Welfare Ordinance, approved by the local Institutional Animal Care and Use Committees in Germany (Regierungspraesidium Freiburg, approval number G14/111) or the Cantonal Veterinary Office and the Animal Welfare Office (Kanton Bern, approval numbers BE132-20 and BE74-23). All experiments were performed in compliance with EU legislation (Directive 2010/63/EU), and animal housing and handling were in accordance with FELASA standards.

Two-electrode voltage-clamp experiments

Xenopus laevis oocytes were purchased from EcoCyte Bioscience (Dortmund, Germany) and maintained as described previously.²⁶ Oocytes were injected with 0.9 ng rabbit KCNQ1 and 0.55 ng rabbit KCNE1 RNA (KCNQ1: XM_008252197.2; KCNE1: NM_001109822) and incubated at 16°C for 1–2 days post-injection.

Two-electrode voltage-clamp recordings were performed at room temperature using an AxoClamp 900A amplifier (Molecular Devices, CA, USA). Docosahexaenoyl glycine (Cayman Chemicals, MI, USA) was stored at −20°C as per supplier instructions. The control solution contained (in mM) 88 NaCl, 1 KCl, 0.4 CaCl₂, 0.8 MgCl₂, and 15 HEPES (with pH set to 7.4 using NaOH). For DHA-gly experiments, the control solution was supplemented with indicated concentrations of DHA-gly. The holding voltage was set to −80 mV, with activation curves generated in steps between −80 and +60 mV in 10 mV increments (5 s duration) and a tail voltage of −20 mV. Control (CTRL) solution or DHA-gly supplemented solution was continuously perfused into the oocyte chamber at a speed of 1 mL/min using a Minipuls 3 peristaltic pump (Gilson, WI, USA). Docosahexaenoyl glycine was applied until a stable effect on current amplitude was observed, achieved after approximately 8 min, and monitored by running an application protocol stepping from a holding voltage of −80 mV to a test voltage of 0 mV every 10 s. Between cells, tubing and recording chamber were cleaned with 70% EtOH and distilled H₂O.

To quantify the voltage dependence of channel opening, tail currents were measured shortly after stepping to the tail voltage and plotted against the preceding activation voltage. A Boltzmann function was fitted to the data to generate the conductance vs. voltage [G(V)] curve:

$$G(V)=G_{min}+(G_{max}-G_{min})/\left\{1+\exp \left[{\displaystyle \frac{V_{50}-V}{s}}\right]\right\}$$ (1)

where G_min is the minimum conductance, G_max is the maximum conductance, V₅₀ is the midpoint (i.e. the voltage at which the conductance is half the maximal conductance determined from the fit), and s is the slope of the curve (which was shared between CTRL and DHA-gly in each oocyte). The difference in steady-state current amplitude induced by the compound in each oocyte (i.e. ΔI_amp) was calculated at the end of the activation pulse to 30 mV and normalized to the current amplitude in the CTRL solution. The effects on G_max or I_amp are shown as relative changes (1 means no change). The effect on V₅₀ is shown in mV (0 mV means no change).

The concentration dependence of the DHA-gly effect was fitted with the following concentration–response equation:

$$\mathrm{\Delta }\mathrm{Effect}=\mathrm{\Delta }\mathrm{effec}{\mathrm{t}}_{\max }/\left[1+{\left({\displaystyle \frac{E{C}_{50}}{C}}\right)}^H\right]$$ (2)

where Δeffect_max is the maximal shift in V₅₀, change in current amplitude or change in G_max, EC₅₀ is the concentration needed to induce 50% of the maximal effect, C is the concentration of DHA-Gly, and H is the Hill coefficient. When fitting the concentration dependence of the DHA-gly effect on the relative current or conductance, the bottom of the curve was set to 1 (i.e. no change).

Rabbit heart extraction

Transgenic LQT1, LQT2, and WT rabbits of both sexes were anaesthetized with an i.m. injection of ketamine S (Ketanest S®, Pfizer, 12.5 mg.kg⁻¹ body weight) and xylazine (Rompun®, Bayer, 3.75 mg.kg⁻¹ body weight), as this anaesthesia regimen does not affect cardiac repolarization.²⁷ After deep anaesthesia (no eye or pinch reflexes), heparin (liquemin®, 500 UI) was injected intravenously (i.v.) in the ear vein to prevent clotting upon heart extraction. Terminal euthanasia was carried out by i.v. injection of pentobarbital (Esconarkon®, 150 mg.kg⁻¹ body weight), and hearts were immediately excised and immersed in a cold (4°C) Tyrode solution (in mM: 135 NaCl, 0.4 NaH₂PO₄, 5 KCl, 10 HEPES, 1 MgCl₂, 1.8 CaCl₂, 10 Glucose, 10 creatine, and 20 taurine, adjusted to pH = 7.4 with NaOH) before ex vivo whole-heart experiments or VCM isolation.

Rabbit ventricular cardiomyocyte isolation

After extraction (see above), hearts were rapidly mounted on a Langendorff perfusion system via cannulation of the aorta. Hearts were then perfused with an oxygenated, body temperature (37°C) Tyrode solution to flush all the blood. Shortly after, hearts were perfused with a Tyrode solution supplemented with 0.1 mM EGTA to stop the contraction and loosen cell-to-cell adhesions, followed by 20–25 min of collagenase digestion (Worthington type 1, 200 mg) in 80 µM Ca²⁺ Tyrode solution.²⁸ Once digested, left ventricles were dissected and cut into small pieces that were further digested in sequential 3 min steps with collagenase (Worthington type 1, 50 mg) in 80 µM Ca²⁺ Tyrode solution supplemented with 15 mM bovine serum albumin (BSA). Isolated VCMs were then seeded in 0.2 mM Ca²⁺ Tyrode solution.

Patch-clamp recordings

Voltage- and current-clamp recordings were performed in isolated left VCMs using an Axopatch 200B amplifier (Molecular Devices, USA). Voltage control, data acquisition, and data analysis were performed with pClamp 11.1/Clampfit (Axon Instruments). All recordings were performed at body temperature (37°C) using the ThermoClamp-1 temperature controller (Digitimer, UK). Borosilicate glass pipettes were pulled using a DMZ-Universal-Electrode-Puller (Munich, Germany) to reach a resistance of 2–3 MΩ.

I_Ks recordings

I_Ks current was recorded in whole-cell condition. Internal solution contained (in mM) 125 KCl, 5 NaCl, 1 MgCl₂, 5 K₂ATP, 10 HEPES, 0.5 EGTA, 4.5 PIP₂, and 1 MgATP adjusted to pH = 7.2 with KOH. Ventricular cardiomyocytes were constantly perfused with Tyrode solution supplemented with 0.3 µM dofetilide (I_Kr inhibition), 1 µM nisoldipine (I_CaL inhibition), and 0.1 µM forskoline to elicit I_Ks. I_Ks current was recorded in two groups of VCMs incubated for 10 min with either our modified Tyrode solution supplemented with ethanol (vehicle, final concentration < 0.001%) (CTRL condition) or with modified Tyrode solution supplemented with 10 µM DHA-gly (DHA-gly condition). Holding voltage was set to −80 mV and I_Na was inactivated with a 100 ms prepulse at −40 mV. I_Ks was then elicited with 5 s depolarizing steps from −20 to +30 mV. The same protocol was subsequently applied in presence of the specific I_Ks-blocker HMR 1556 (0.5 µM). I_Ks was then defined as the HMR 1556-sensitive current by subtracting the two traces. Steady-state I_Ks current was analysed at the end of the depolarizing steps. Potentials were not corrected due to the small calculated liquid junction potential (−4 mV).

Action potential recordings

Action potentials (APs) were recorded in perforated patch-clamp condition. Internal solution contained (in mM) 80 K-aspartate, 45 KCl, 5 NaCl, 1 MgCl₂, 5 K₂ATP, 10 HEPES, and 0.5 EGTA, adjusted to pH = 7.2 with KOH and supplemented with 0.44 mM amphotericin-B. Ventricular cardiomyocytes were constantly perfused with Tyrode solution supplemented with ethanol (vehicle, final concentration < 0.001%) (CTRL condition) before switching to Tyrode solution supplemented with DHA-gly 10 µM (DHA-gly condition). Action potentials were elicited at 1 Hz by a 4 ms current pulse at ∼1.5 × AP threshold. Resting membrane potential (RMP), AP amplitude (APA), maximal AP upstroke velocity, AP duration at 90% repolarization (APD₉₀), and STV of the APD₉₀ were measured and averaged on 20 consecutive APs at steady-state control condition and after 2 min of DHA-gly perfusion. Potentials were corrected for the calculated liquid junction potential (14 mV). As we previously reported that AP parameters remain stable over the time frame of our experiments, we have not included temporal control data in this study.²⁹

Rabbit ex vivo whole-heart experiments

After extraction (see above), hearts were rapidly mounted on a Langendorff perfusion system (IH5, Hugo Sachs Electronic-Harvard Apparatus, Hugstetten, Germany) via cannulation of the aorta and perfused with an oxygenated, body temperature (37°C) Krebs-Henseleit (KH) solution consisting of (in mM) 118 NaCl, 4.7 KCl, 1.2 KH₂PO₄, 24.9 NaHCO₃, 1.8 CaCl₂, 0.8 MgSO₄, 5.6 Glucose, and 2 Na-pyruvate, adjusted to pH 7.4 using HCl. Perfusion was kept at a constant flow of 50 mL.min⁻¹ and a constant aortic pressure of 80 mmHg. Electrocardiograms were recorded via three electrodes positioned to the left, right, and apex of the heart to measure the three bipolar leads I, II, III, and the unipolar leads aVR, aVL, and aVF.^30,31 A latex balloon was inserted in the left ventricle and coupled to a pressure transducer (Hugo Sachs Electronic-Harvard Apparatus, Hugstetten, Germany) to measure the left ventricular pressure (LVP) with a fixed end diastolic pressure of 8 mmHg. Complete electrical dissociation of atria and ventricles was performed by electrocauterizing the atrioventricular (AV) node with an electrocauter to enable a slow (<2 Hz) ventricular escape rhythm. Hearts were then paced at 2 Hz in control condition and after 30 min of 10 µM DHA-gly perfusion to measure the QT interval without having to correct it for the underlying heart rate. As we previously reported that QT interval remains stable over the time frame of our experiments, we have not included temporal control data in this study.³² Arrhythmias were triggered in WT, LQT1, and LQT2 hearts by perfusion of 10 µM BaCl₂ to inhibit the inward rectifier K⁺ current I_K1 without any pacing—in the setting of bradycardia (slow intrinsic ventricular escape rhythm), which facilitates arrhythmia formation, especially in LQT2 as already described.³³ Hence, experiments were carried out as follows: arrhythmia induction (KH + 10 µM BaCl₂, no pacing, 5 min), washout BaCl₂ (KH, pacing 2 Hz, 5 min), baseline (CTRL condition, KH, pacing 2 Hz, 5 min), DHA-gly incubation (KH + 10 µM DHA-gly, pacing 2 Hz, 30 min), and arrhythmia induction in DHA-gly condition (KH + 10 µM DHA-gly + 10 µM BaCl2, no pacing, 5 min).

Ca²⁺ transient and sarcomere shortening recordings in rabbit cardiomyocytes

Intracellular Ca²⁺ levels and sarcomere shortening were measured in isolated VCMs using the IonOptix system (IonOptix LLC, USA). Data acquisition and analysis were conducted using the IonWizzard software (IonOprix LLC, USA). Cell contraction and sarcomere shortening were imaged with IonOptix Myocam-S. Ca²⁺ transients were recorded through epifluorescence microscopy after 15 min incubation of VCMs at room temperature with 1.5 µM Fura2-AM ratiometric dye (Abcam, UK). Ventricular cardiomyocytes were seeded onto laminin-coated dishes (10 mg.mL⁻¹, L2020-Sigma) and excited by LEDs at 340 nm (Ca²⁺-bound) and 380 nm (Ca²⁺-unbound) while Fura2-AM fluorescence at 510 nm was detected by a photomultiplier tube (PMT). Ventricular cardiomyocytes were paced at a stimulation frequency of 1 Hz using a MyoPacer field stimulator (IonOptix LLC, USA), and signals were integrated using a Fluorescence System Interface (FSI). Ca²⁺ transients and sarcomere shortening were recorded in body temperature (37°C) Tyrode solution or after 15 min perfusion of 10 µM DHA-gly. Ca²⁺ transient duration at 90% return to baseline (CaTD90) and peak shortening of sarcomere length were analysed taking into account the clustering of VCMs isolated from the same rabbit (inter-individual differences) using the hierarchical statistical method from Sikkel et al.³⁴

Ca²⁺ transient recordings in human induced pluripotent stem cell-derived cardiomyocytes

Human induced pluripotent stem cell-derived cardiomyocytes (hiPSC-CMs, Fujifilm Cellular Dynamics) were plated following the manufacture’s protocol. Briefly, hiPSC-CMs were suspended in plating medium, supplemented with 10 µM of Y27632 (Biogems) to improve cell survival. Then, cells were directly plated in silicone slides (SMI), at 170 000 cell/cm² density, previously coated with grooved 3D gel from a mixture of 2 mg.mL⁻¹ Matrigel® (Corning), 8% Gelatin (Sigma-Aldrich), and 4% of m-transglutaminase (Ajinomoto). After 4 h of being attached, the plating medium was replaced by a maintenance medium. On Day 4 (D4), three days of treatment (from D4 to D7) of Poli IC (0.95 µg.cm2⁻¹, Sigma-Aldrich) was applied. After removing the treatment, the maintenance medium was replaced every 2–3 days. Cells were cultured at 37°C and 5% CO₂ atmosphere.

Optical Ca²⁺ transients on hiPSC-CMs monolayers were recorded using a MacroFluo Leica Microscope coupled with a Leica APOZ6 zoom and a 5×/0.5 LWD objective. Data acquisition was performed using a photodiode connected to an Axopatch 200B amplifier and an Axon Digidata 1550B digitizer. Prior to recordings, hiPSC-CMs were incubated with Fluo-4F-AM (2 µM) for 30 min. The hiPSC-CM monolayers were paced at 1 Hz (12 volts) with a concentric bi-polar electrode and a Grass S 9 stimulator. Recordings were performed at 37°C in saline solution (control condition) containing (in mM) 150 NaCl, 5.4 KCl, 1.8 CaCl₂, 1 MgCl₂, 15 glucose, 15 HEPES, and 1 Na-pyruvate, adjusted to pH = 7.4 with HCl.³⁵ Recordings after application of 20 µM DHA-gly were taken every 10 min.

For each recording in each condition, 10 Ca²⁺ transients were averaged. The Ca²⁺ transient duration was measured at 80% of repolarization (CaTD80), reflecting the entire repolarization process (Phases 1, 2, and 3). The absolute change in CaT duration (ΔCaTD80) between DHA-gly and control condition was given as a percentage of change. CaTD80 was corrected (CaTD80c) in all measurements by the modified version of Fridericias’s formula³⁶:

$$CaTD80c={\displaystyle \frac{CaTD80}{(\sqrt[3]{RR})}}$$

where RR represents the interval between two Ca²⁺ transients.

Statistics

Statistics were performed using Prism 9.5.1 (GraphPad Software, USA). Data are expressed as mean ± standard error of the mean (SEM). Normality of the data was assessed with Shapiro–Wilk test and parametric or non-parametric tests were used accordingly, as specified in the figure legends. P < 0.05 was considered statistically significant.

Results

DHA-gly increases the I_Ks current in Xenopus oocytes and rabbit cardiomyocytes

We previously demonstrated that DHA-gly activates human I_Ks expressed in Xenopus laevis oocytes.^17,18 Here, we now also studied the potential of DHA-gly to increase rabbit I_Ks. Two-voltage clamp experiments on rabbit I_Ks expressed in Xenopus laevis oocytes—transduced with rabbit KCNQ1/KCNE1 mRNA—showed that DHA-gly increased rabbit I_Ks in a concentration-dependent manner (Figure 1A and B). This increase in rabbit I_Ks was mediated by both a leftward shift in the voltage dependence of activation towards more negative potentials and an increase in the maximal conductance (G_max) (see Supplementary material online, Figure S1A and B). Concentration–response fits for these two parameters predict a maximal effect of −26 mV shift for the half activation (V₅₀, baseline = + 28.6 ± 3.2 mV; DHA-gly 30 µM: + 6.5 ± 1.7 mV) and a 2.4-fold increase in G_max/G_max0 ratio (baseline = 1053.4 ± 489.4 nA; DHA-gly 30 µM: 2205.0 ± 780.2 nA) with 50% of the maximal effect achieved by 2–3 µM DHA-gly for both parameters (see Supplementary material online, Figure S1A and B). Combination of the V₅₀ shift and increase in G_max resulted in a concentration-dependent response of the current amplitude at +30 mV with a predicted 4.6-fold increase in I/I₀ ratio at maximal effect (see Supplementary material online, Figure S1C).

Figure 1.

DHA-gly increases I_Ks in Xenopus oocytes and rabbit VCMs. (A) Example trace recordings of rabbit I_Ks current in Xenopus oocytes in control conditions (CTRL) or during perfusion with 3 and 10 µM DHA-gly (coloured trace indicates the current sweep at 0 mV). (B) Corresponding current–voltage relationship of the steady-state I_Ks current in response to increased concentration of DHA-gly (0.3, 1, 3, 10, and 30 µM). (C) Example trace recordings of the I_Ks current in isolated left VCMs from WT rabbits in control conditions (CTRL) and after 10 min incubation with 10 µM DHA-gly. (D) Current–voltage relationship and current density at +30 mV (E) of the I_Ks steady-state current in control conditions (black circle, n = 7 cells, N = 2 rabbits) or after incubation with 10 µM DHA-gly (green square) (n = 7, N = 2). **P < 0.01 with unpaired t-test. Cartoons were created with BioRender.

We then tested whether DHA-gly could similarly activate I_Ks in mammalian rabbit cardiomyocytes. To ensure a sufficient activation of I_Ks of at least 50%, we used 10 µM DHA-gly, based on our results in Xenopus oocytes, and we performed patch-clamp experiments on freshly isolated left VCMs from WT rabbits. Since the measured current and the cell membrane capacitance did not show a direct proportionality (see Supplementary material online, Figure S1D),³⁷ steady-state I_Ks was not expressed in current density but as ‘raw’ I_Ks current in Figure 1. In VCMs, DHA-gly significantly increased I_Ks at positive potentials, with an average 65% increase at +30 mV (Figure 1C–E). Taken together, these data indicate that DHA-gly can be used to potentiate I_Ks in a rabbit model.

DHA-gly shortens the action potential duration in WT and LQT2 rabbit cardiomyocytes

To investigate whether the DHA-gly-mediated increase in I_Ks could reduce or even normalize APD in LQTS, we tested its effect on APD in isolated VCMs from WT, LQT1, and LQT2 rabbits. LQT1 and LQT2 VCMs exhibited longer APD at 90% repolarization (APD₉₀) than WT CMs (LQT1: 434.5 ± 24.7 ms; LQT2: 434.5 ± 32.4 ms; WT: 316.3 ± 16.5 ms) without any differences in RMP, maximal velocity of depolarization (upstroke velocity), and amplitude (APA) (see Supplementary material online, Figure  S2A-D). Moreover, LQT1 and LQT2 VCMs showed a tendency towards higher STV of the APD₉₀ than WT (see Supplementary material online, Figure S2E).

Perfusion of 10 µM DHA-gly had no impact on RMP, upstroke velocity, or APA across all genotypes (see Supplementary material online, Figure S3A–C). However, importantly, it significantly shortened APD₉₀ in WT (baseline: 316.3 ± 16.5 ms; DHA-gly: 281.0 ± 22.4 ms) and LQT2 VCMs (baseline: 434.5 ± 32.4 ms; DHA-gly: 364.1 ± 34.0 ms), while it had no effect in LQT1 VCMs (baseline: 434.5 ± 24.7 ms; DHA-gly: 417.6 ± 23.9 ms) (Figure 2A–C; Supplementary material online, Figure S3D). Of note, there were some heterogeneities in the response to DHA-gly (Figure 2A–C), as it has already been described for other drugs.^38,39 Interestingly, though not significant, DHA-gly-induced decrease in APD₉₀ showed a trend to be greater in LQT2 than in WT VCMs (delta APD₉₀, WT: −35.3 ± 12.1 ms; LQT2: −70.4 ± 30.4 ms) (see Supplementary material online, Figure S3E). Indeed, in VCMs in which DHA-gly shortened APD₉₀, we observed a positive correlation between the extent of shortening and the length of the APD at baseline, i.e. the longer the APD at baseline, the greater the shortening by DHA-gly (see Supplementary material online, Figure S3F). Additionally, DHA-gly tended to reduce the rather small STV of the APD₉₀ in WT (baseline: 5.7 ± 0.9 ms; DHA-gly: 4.0 ± 0.4 ms, P = 0.062) and significantly reduced STV of the APD₉₀ in LQT2 (baseline: 21.3 ± 11.8 ms; DHA-gly: 9.8 ± 3.7 ms, P < 0.05), but not in LQT1 (baseline: 13.6 ± 2.4 ms; DHA-gly: 11.0 ± 1.5 ms), suggesting a potential antiarrhythmic effect of DHA-gly in LQT2 (Figure 2A–C). In conclusion, DHA-gly is able to shorten APD in LQT2 VCMs towards a WT-like phenotype, but not in LQT1 VCMs.

Figure 2.

DHA-gly shortens action potential duration (APD₉₀) in WT and LQT2 rabbit VCMs. (A) Representative action potential in control condition (CTRL, grey) and after perfusion of 10 µM of DHA-gly (green), corresponding average APD₉₀, example of STV of the APD₉₀ in one cell with a Poincaré plot and average STV of the APD₉₀ in control condition (CTRL, grey) and after 10 µM DHA-gly perfusion (green) in isolated left VCMs from WT (top row, CTRL = grey, n = 7 cells, N = 2 rabbits), LQT1 (B) (middle row, CTRL = blue, n = 10, N = 3) and LQT2 (C) (bottom row, CTRL = red, n = 7, N = 4) rabbits. *P < 0.05 with paired t-test.

DHA-gly shortens the QT interval in WT and LQT2 rabbit hearts ex vivo

Given the promising shortening of APD and STV APD₉₀ in LQT2, we then tested whether DHA-gly could shorten the QT interval and prevent arrhythmias in LQT2 using ex vivo Langendorff-perfused whole hearts. These hearts were subjected to an atrioventricular (AV) node ablation, to induce a bradycardic ventricular escape rhythm, and paced at the right ventricle base at 2 Hz (Figure 3A–C). At baseline, LQT1 and LQT2 hearts showed significantly longer QT intervals than WT hearts at 2 Hz stimulation (WT: 214.6 ± 2.7 ms; LQT1: 236.1 ± 8.3 ms; LQT2: 273.6 ± 6.0 ms) (see Supplementary material online, Figure S4A). Similarly to the observations in isolated VCMs, perfusion of 10 µM DHA-gly decreased the QT interval in WT (baseline: 214.6 ± 2.7 ms; DHA-gly: 206.6 ± 2.8 ms, P < 0.05) and LQT2 (baseline: 273.6 ± 6.0 ms; DHA-gly: 246.9 ± 2.2 ms, P = 0.06), but not in LQT1 hearts (baseline: 236.1 ± 8.3 ms; DHA-gly: 233.2 ± 10.9 ms) (Figure 3A–C; Supplementary material online, Figure S4B). Besides, we once again observed a trend in a greater shortening of the QT interval by DHA-gly in LQT2 compared to WT hearts (delta QT, WT: −8.0 ± 3.1 ms; LQT2: −26.7 ± 14.5 ms) (see Supplementary material online, Figure S4C), and a positive correlation between the extent of shortening and the QT interval at baseline in hearts in which DHA-gly shortened the QT (see Supplementary material online, Figure S4D).

Figure 3.

DHA-gly shortens the QT interval in WT and LQT2 ex vivo rabbit hearts. Representative ECG traces of two consecutive beats in ventricular escape rhythm after AV node ablation and average QT interval in control condition and after 30 min of 10 µM DHA-gly perfusion in ex vivo Langendorff-perfused hearts from WT (A) (CTRL = grey, DHA-gly = green, N = 8), LQT1 (B) (CTRL = blue, DHA-gly = green, N = 5), and LQT2 (C) (CTRL = red, DHA-gly = green, N = 5) rabbits. (D–F) Example traces of LVP (top row) and simultaneous ECG recordings (bottom row) during episodes of bigeminy (D), VT (E), and VF (F). Proportion of arrhythmias elicited by perfusion of 10 µM barium chloride in control conditions (BaCl₂) or after 30 min of 10 µM DHA-gly perfusion (DHA-gly + BaCl₂) in individual hearts from WT (G), LQT1 (H), and LQT2 (I) (VER, ventricular escape rhythm, grey; bigeminy, orange; VT, pink; VF, dark red). * P  <  0.05 with paired t-test.

As we have previously demonstrated that the inhibition of the inward rectifier K⁺ current I_K1 with BaCl₂ prolonged APD/QT and triggered arrhythmia in bradycardic LQT2, but not in healthy WT rabbit hearts,^{33  ,40} we here investigated whether DHA-gly could prevent arrhythmia formation in our LQT2 rabbit model during perfusion with 10 µM BaCl₂ in bradycardic ventricular escape rhythm (Figure 3D–I). At baseline, I_K1-inhibition triggered episodes of bigeminy, ventricular tachycardia (VT), and ventricular fibrillation (VF) in LQT2 hearts (Figure 3D–F and I). However, except for episodes of bigeminy in only one WT and one LQT1 heart, I_K1-inhibition did not trigger any arrhythmia in WT and LQT1 (Figure 3G and H; Supplementary material online, Figure S4E). Unexpectedly, concomitant perfusion of 10 µM BaCl₂ and 10 µM DHA-gly did not prevent arrhythmia formation in LQT2 hearts (Figure 3I; Supplementary material online, Figure S4H); but also did not increase arrhythmias in WT and LQT1 hearts (Figure 3G and H; Supplementary material online, Figure S4F and G).

Taken together, these results indicate that DHA-gly is able to shorten the APD/QT interval in LQT2 hearts but fails to prevent cardiac arrhythmia formation ex vivo.

Deleterious effects of DHA-gly on cardiac contractility

While measuring the effect of DHA-gly on the QT interval, we also assessed cardiac contractility by recording the LVP in these Langendorff-perfused ex vivo whole hearts (Figure 4A–C). Perfusion of 10 µM DHA-gly significantly reduced the LVP in WT (baseline: 100.3 ± 4.3 mmHg; DHA-gly: 61.7 ± 4.7 mmHg), LQT1 (baseline: 86.5 ± 12.4 mmHg; DHA-gly: 56.2 ± 7.9 mmHg), and LQT2 hearts (baseline: 105.5 ± 7.0 mmHg; DHA-gly: 61.2 ± 6.8 mmHg) (Figure 4A–C). Surprisingly, while DHA-gly did not shorten APD/QT in LQT1 (Figures 2 and 3), it still decreased the contractility just as in WT and LQT2 hearts, suggesting a mechanism independent of its APD-shortening effect.

Figure 4.

DHA-gly decreases cardiac contractility in WT, LQT1, and LQT2 rabbit hearts. (A–C) Example trace and corresponding average LVP in control condition and after 30 min of 10 µM DHA-gly perfusion in ex vivo Langendorff-perfused hearts from WT (A) (CTRL = grey, DHA-gly = green, N = 8), LQT1 (B) (CTRL = blue, DHA-gly = green, N = 5), and LQT2 (C) (CTRL = red, DHA-gly = green, N = 5) rabbits. (D–F) Example trace of sarcomere length and corresponding average peak shortening in control condition and after incubation with 10 µM DHA-gly in isolated left VCMs from WT (D) (CTRL = grey, n = 53 cells, N = 4 rabbits; DHA-gly = green, n = 50, N = 4), LQT1 (E) (CTRL = blue, n = 68, N = 5; DHA-gly = green, n = 74, N = 5), and LQT2 (F) (CTRL = red, n = 65, N = 5; DHA-gly = green, n = 71, N = 5) rabbit hearts. *P < 0.05, ***P < 0.001 with paired t-test (A–C) or a hierarchical statistical method from Sikkel et al.³⁴ (D–F). Cartoons were created with BioRender.

We then went on to assess the effect of DHA-gly on cellular contractility by measuring the sarcomere length of isolated VCMs from WT, LQT1, and LQT2 rabbit hearts. Similarly as observed at the whole-heart level, 10 µM DHA-gly reduced the peak shortening of sarcomere length (i.e. contraction) in WT (baseline: 10.16 ± 0.43%; DHA-gly: 7.85 ± 0.36%), LQT1 (baseline: 9.06 ± 0.45%; DHA-gly: 6.98 ± 0.39%), and LQT2 VCMs (baseline: 9.16 ± 0.42%; DHA-gly: 8.11 ± 0.35%) (Figure 4D–F).

DHA-gly disrupts Ca²⁺ dynamics in WT, LQT1, and LQT2 rabbit cardiomyocytes and human iPSC-CMs

Given the deleterious effect of DHA-gly on cardiac contractility, we investigated whether DHA-gly affects Ca²⁺ dynamics in isolated cells by measuring Ca²⁺ fluorescence in Fura-2AM loaded left VCMs (Figure 5A–C). Perfusion of 10 µM DHA-gly significantly decreased Ca²⁺ transient duration at 25% (CaTD25) and 90% return to baseline (CaTD90) in WT (baseline: 364.0 ± 18.5 ms; DHA-gly: 297.0 ± 11.0 ms), LQT1 (baseline: 408.3 ± 12.5 ms; DHA-gly: 364.7 ± 11.1 ms), and LQT2 (baseline: 411.0 ± 18.8 ms; DHA-gly: 368.9 ± 15.6 ms) (Figure 5A–C; Supplementary material online, Figure S5A–C). Interestingly, some VCMs presented with an early secondary Ca²⁺ rise before the end of the Ca²⁺ decay (Figure 5D). These secondary Ca²⁺ rises can lead to early after depolarizations (EADs) and could therefore be seen as potential arrhythmic events.⁴¹ Hence, we calculated the effect of DHA-gly on the occurrence of these secondary Ca²⁺ rises in WT, LQT1, and LQT2 VCMs. Though not significant, DHA-gly tended to decrease the occurrence of secondary Ca²⁺ rises in WT (baseline: 7.9 ± 3.0%; DHA-gly: 0.0 ± 0.0%, P = 0.07), LQT1 (baseline: 26.5 ± 8.0%; DHA-gly: 7.8 ± 3.8%, P = 0.06), and to a lesser degree in LQT2 VCMs (baseline: 23.8 ± 7.5%; DHA-gly: 14.6 ± 8.3%, P = 0.37) (Figure 5A–C), suggesting a potential antiarrhythmic effect.

Figure 5.

DHA-gly shortens Ca²⁺ transient duration in rabbit VCMs and hiPSC-derived cardiomyocytes. Representative traces (top panels) and average Ca²⁺ transient duration at 90% return to baseline (CaTD₉₀) in isolated left VCMs from WT (A) (CTRL = grey, n = 39 cells, N = 4 rabbits; DHA-gly = green, n = 49, N = 4), LQT1 (B) (CTRL = blue, n = 56, N = 5; DHA-gly = green, n = 73, N = 5), and LQT2 (C) (CTRL = red, n = 57, N = 5; DHA-gly = green, n = 65, N = 5) rabbit hearts. Number of individual VCMs exhibiting secondary Ca²⁺ rise events (as a ratio) per hearts in WT (D) (three out of four hearts), LQT1 (E) (five out of five hearts), and LQT2 (F) (four out of five hearts). (G) Example traces depicting a secondary Ca²⁺ rise during the decay phase of a Ca²⁺ transient. (H) Representative traces of optically recorded Ca²⁺ transients from hiPSC-CM monolayers and average Ca²⁺ transient duration at 90% return to baseline (CaTD_90c) in control conditions (black) and after perfusion of 20 µM DHA-gly (green) (N = 4). *P < 0.05, **P < 0.01 with a hierarchical statistical method from Sikkel et al.³⁴ (A–C) or paired t-test (E). Cartoons were created with BioRender.

Interestingly, this DHA-gly-induced shortening of the Ca²⁺ transient duration was not only restricted to rabbit VCMs. Indeed, perfusion of 20 µM DHA-gly also significantly decreased CaTD25 and CaTD90 in monolayers of hiPSC-CMs (baseline: 522.1 ± 50.4 ms; DHA-gly: 424.4 ± 56.2 ms) (Figure 5E; Supplementary material online, Figure S5D).

Taken together, these data indicate that DHA-gly disrupts contractility and Ca²⁺ dynamics even in LQT1, suggesting a mechanism distinct from the APD shortening due to I_Ks activation.

Discussion

Our study shows, for the first time, the potential of the I_Ks-activator DHA-gly in shortening APD/QT in a medium-sized animal model of LQT2. Although promising, we show that this shortening effect does not prevent arrhythmia formation and is accompanied by a deleterious effect on cardiac contractility. Finally, we show that DHA-gly disrupts intracellular Ca²⁺ dynamics in VCMs.

DHA-gly acts as an activator of the I_Ks current

Over the last decade, PUFAs have been identified as potent activators of KCNQ1 channels, able to increase the I_Ks current.^17–22,42 While this increase in I_Ks was mostly studied in Xenopus oocytes expressing human KCNQ1/KCNE1, we previously showed that the PUFA DHA potentiates I_Ks in Xenopus oocytes expressing rabbit KCNQ1/KCNE1.²⁵ Furthermore, the PUFA N-AT has also been shown to increase I_Ks in mammalian cells, e.g. in isolated embryonic rat cardiomyocytes.¹⁷ Here, we demonstrated that another PUFA, DHA-gly, is also able to increase rabbit I_Ks in a concentration-dependent manner in Xenopus oocytes expressing rabbit KCNQ1/KCNE1 and can acutely increase I_Ks in isolated rabbit VCMs (Figure 1). This effect was mediated by both a shift in the voltage dependency towards more negative potentials and an increase in the maximal conductance G_max (see Supplementary material online, Figure S1). Interestingly, these two effects have been reported as a hallmark of the I_Ks activation by PUFAs.^20,21,42 Electrostatic interaction between the negatively charged PUFA head group and the arginine residue R228 (in human KCNQ1) in the voltage sensor domain induces the shift in the voltage dependency, whereas electrostatic interaction with the lysine residue K326 in the pore domain increases the maximal conductance G_max.^20,21,42 Therefore, as long as these two binding sites of PUFAs are preserved among species, PUFAs should be able to increase I_Ks. For instance, there is a 100% predicted amino acids similarity in PUFAs binding site between rabbit and human KCNQ1 (see Supplementary material online, Figure S6).⁴³ However, this raises a concern about DHA-gly efficacy to increase I_Ks in the presence of variants in KCNQ1 localized at PUFAs binding sites. To date, more than 600 KCNQ1 variants spanning the whole gene have been identified to cause LQT1 syndrome.^44–46 Interestingly, we recently demonstrated that the effect of the endocannabinoid N-arachidonoyl-L-serine (ARA-S) in increasing I_Ks varied between LQT1 variants, with reduced effect when the mutation was localized in one of the two putative binding sites of ARA-S.⁴⁷ Hence, some KCNQ1 mutations might prevent PUFA-induced activation of I_Ks, which can render the therapeutic strategy of using PUFAs to treat LQT1 variant- and patient-dependent.

In addition to their ability to increase I_Ks, PUFAs also interact with other cardiac ion channels/currents.¹³ Notably, PUFAs inhibit voltage-gated Na⁺  I_Na and L-type Ca²⁺  I_CaL currents.^{13–15,48,49} However, we previously showed that DHA-gly, at 10 µM, has minimal to no effect on I_Na and I_CaL, unlike more potent PUFAs activators of I_Ks.^18,35 Furthermore, 10 µM DHA-gly has already been shown to shorten the QT interval in a drug-induced LQT2 setting in ex vivo guinea pig hearts.^17,23 Hence, DHA-gly seemed a good candidate to study its therapeutic effect in our transgenic LQT1 and LQT2 rabbit models.

DHA-gly shortening effects on APD/QT

Polyunsaturated fatty acids’ propensity to enhance I_Ks, thus accelerating cardiac repolarization, makes them potential candidates to treat LQTS, in which repolarization is prolonged. In this study, we used rabbit models of LQT1 and LQT2 that express dominant-negative pore mutants of the human genes KCNQ1 (Y315S) and KCNH2 (G628S), respectively.²⁴ Rabbit cardiac electrophysiology resembles that of humans in terms of the shape of the AP and the underlying cardiac ion currents.^50,51 Notably, rabbit repolarization relies on I_Kr and I_Ks, as in humans, which makes them ideal models to study therapeutic effects in pathologies affecting the repolarization, such as LQTS.^25,28,52 Therefore, we tested the effect of DHA-gly in shortening the prolonged APD/QT interval in our LQT1 and LQT2 rabbit models.

In a drug-induced LQT2 guinea pig model, DHA, DHA-gly, linoleoyl glycine (Lin-gly), and N-AT have all been shown to shorten APD and QT interval.^17,23 Additionally, we previously demonstrated that DHA normalizes the QT interval in our LQT2 rabbit model to WT levels, but had no beneficial effects in our LQT1 model.^25,53 Here, we showed that DHA-gly shortens APD in isolated VCMs and QT interval in ex vivo hearts from WT and LQT2 rabbits, but not in LQT1 (Figures 2 and 3). Indeed, both WT and LQT2 rabbits have a functional I_Ks current, which can be increased by DHA-gly and lead to a shortening of APD and QT interval. However, our LQT1 rabbits lack I_Ks due to a dominant-negative mutation in the pore domain (Y315S).²⁴ Hence, it is possible that the mutation, lying only 11 residues away from the critical lysine K326,^20,21 disrupts DHA-gly’s binding site in the pore domain, preventing the increase in I_Ks conductance in these LQT1 CMs (Figure 1, Supplementary material online, Figure S1). In addition, since the pore mutation overexpressed in our LQT1 rabbit model has a dominant negative effect, which leads to non-conducting KCNQ1 channels,²⁴ even if DHA-gly was still able to interact with KCNQ1 via its voltage-sensor binding site, a shift in the voltage dependency would likely not restore an absent I_Ks current. While using PUFAs to treat LQTS appears to be variant-dependent in LQT1, it is nonetheless promising in LQT2, where KCNQ1 channels are functional. This suggests that PUFAs may offer new therapeutic strategies for LQT types with intact I_Ks.

Interestingly, we noticed a heterogeneity in the extent of APD change in response to DHA-gly in isolated VCMs, i.e. DHA-gly did not shorten APD in every individual VCM, and not to the same magnitude (Figure 2). Similarly, it has recently been shown that electrophysiological heterogeneity can be considerable in rabbit VCMs, even in healthy hearts, with highly variable changes in APD upon I_Kr or I_CaL block.^38,39 Nonetheless, in VCMs in which DHA-gly shortened the APD, we observed that the shortening effect of DHA-gly in WT and LQT2 rabbits was dependent on the length of the APD/QT at baseline, i.e. the longer the APD/QT, the greater the effect of DHA-gly (see Supplementary material online, Figures S3 and S4). Such a relationship between the extent of the APD/QT-altering effect of a drug and the initial length of APD/QT has already been well described for APD-prolonging or APD-shortening drugs.^54,55 While the APD differences between WT and LQT2 in this study are attributable to the different genotypes, this phenomenon has also been observed in relation to heart rate-induced changes in APD within a single model, known as rate-reverse dependency. In every species with heart rate-dependency of the APD/QT, such as rabbits and humans, where lower heart rates are associated with longer APD/QT, DHA-gly should also exhibit a greater APD/QT shortening effect at lower heart rates with accentuated DHA-gly-induced I_Ks activation mediated by its slow activating properties. Interestingly, lower heart rates have been associated with a higher incidence of life-threatening arrhythmias in LQT2.⁵⁶ Hence, it would be interesting to investigate whether DHA-gly can prevent arrhythmia formation in LQT2 by shortening the QT interval.

At the cellular level, we observed a reduction of the pro-arrhythmic marker STV of the APD₉₀ in both WT and LQT2 models, but not in LQT1 (Figure 2). Additionally, the occurrence of secondary Ca²⁺ rise tended to decrease (Figure 5). This suggests a potential antiarrhythmic effect of DHA-gly in isolated VCMs that was not reflected at the whole-heart level (Figure 3). While DHA-gly did not show any pro-arrhythmic effect in WT and LQT1, it also did not reduce the occurrence of arrhythmias (bigeminy) induced by I_K1-inhibition with BaCl₂ in 4/4 LQT2. This was rather unexpected as we had previously shown that progesterone, another APD/QT-shortening compound, decreased occurrence of mild (and more severe) arrhythmia such as bigeminy (and VT/VF) in LQT2 rabbits.⁵⁷ Harsher pro-arrhythmic conditions such as the combination of I_K1-inhibition and hypokalaemia can induce also longer episodes of VT and VF in LQT2,³³ but as we did not even see any antiarrhythmic effect with BaCl₂ alone, we doubt that we would be able to reveal any potential antiarrhythmic effect of DHA-gly in these even more pro-arrhythmic settings. Such a discrepancy between the antiarrhythmic effect at the cellular level and the absence of such effect at the whole heart is difficult to explain, especially since we previously observed an antiarrhythmic effect of DHA in LQT2 at the whole-heart level.²⁵ However, the deleterious effects of DHA-gly on contractility (Figure 4), not observed with DHA, might counteract any potential antiarrhythmic effect. Indeed, it has been shown that enhanced sarcoplasmic reticulum (SR) Ca²⁺ extrusion leads to decreased cardiac contractility, which goes along with spontaneous arrhythmia in the rabbit.^58,59 Moreover, growing evidence suggests that mechanical stimuli can trigger electrical responses and arrhythmias,^60–62 particularly in LQT2 patients, who often display altered mechanical function—a feature recapitulated in our LQT2 rabbit model.³⁰ Hence, further studies would be needed to clearly demonstrate this proposed mechanistic link between decreased contractility and pro-arrhythmic effects.

Deleterious DHA-gly effect on cardiac contractility

The promising shortening of APD/QT in our LQT2 rabbit model was counteracted by an unwanted, deleterious effect of DHA-gly on cardiac contractility, both at whole-heart and cellular levels (Figure 4). Interestingly, this effect was also observed in our LQT1 model, where DHA-gly was unable to shorten APD/QT. This suggests that the decrease in contractility is not a consequence of the increase in I_Ks and the consecutive shortening of cardiac repolarization, which could otherwise lead to decreased contraction due to a shortened plateau phase. The PUFA-induced negative inotropism has been previously reported in guinea pig and rat VCMs, likely due to a decrease in the L-type Ca²⁺ current (I_CaL) and an inhibition of SR-dependent Ca²⁺ release.^49,63 Although 10 µM DHA-gly has no effect on I_CaL,³⁵ we cannot rule out the possibility that DHA-gly affects the Ca²⁺ release from the SR. Indeed, we observed that DHA-gly shortened Ca²⁺ transient duration in VCMs in all genotypes (including LQT1) as well as in hiPSC-CMs (Figure 5), which could explain the decrease in contractility.

Hence, by increasing the I_Ks current, PUFAs are promising candidates for treating LQT2 and may also work in a variant-specific manner in LQT1—depending on where the KCNQ1 variant is located. However, we show here that their deleterious, negative inotropic effects on cardiac contractility and Ca²⁺ dynamics might limit their clinical use, as these effects could mask potential antiarrhythmic effects and/or even lead to heart failure. Nonetheless, several natural or modified PUFAs are capable of increasing I_Ks, and some of them may not affect cardiac contractility. As we observed a similar shortening in Ca²⁺ transient duration in both hiPSC-CMs and rabbit VCMs, hiPSC-CMs might be a good model for rapidly screening PUFAs for adverse effects on Ca²⁺ dynamics. Our study highlights the importance of considering both electrical and mechanical effects of potential new therapeutic compounds, underscoring the need for contractility screening of all promising candidates prior to their further development towards clinical translation.

Conclusion

This study demonstrates the ability of DHA-gly to shorten APD/QT by increasing I_Ks and confirms the potential of PUFAs as a therapeutic strategy in LQT2. Despite beneficial effects on APD/QT, DHA-gly failed to prevent arrhythmia formation in LQT2 and exerted deleterious effects on cardiac contractility, which could counteract a potential antiarrhythmic effect. Hence, screening PUFAs for APD-shortening effects in LQT2 should always be accompanied by a thorough assessment of their impact on cardiac contractility, something that could easily be done in hiPSC-CMs in a genotype- and variant-specific manner.

Supplementary material

Supplementary material is available at Europace online.

Funding

This work was supported by an NIH ROI (2R01HL131461-05 to H.P.L. and K.E.O.) and a Bern Center of Precision Medicine Lighthouse Grant ‘PACE’ (to K.E.O.). This work was also supported by the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation programme (grant agreement no. 850622 to S.I.L.) and by the Swedish Research Council (#2022-00844 to H.P.L.).

Data availability

Data are made available upon reasonable request.