Stretch‐Induced Increase in Ca²⁺‐Spark Rate in Rabbit Atrial Cardiomyocytes Requires TRPA1 and Intact Microtubule Network

Abstract

Background

Mechanical stretch of the myocardium is proarrhythmic and alters cellular Ca²⁺ handling, potentially involving cation nonselective mechano‐sensitive ion channels. This study aimed to assess the presence and mechanisms of stretch‐induced increase in Ca²⁺‐spark rate (SiS) in isolated atrial cardiomyocytes.

Methods

Freshly isolated rabbit, pig, and human left atrial cardiomyocytes were stretched axially using glass microrods. Free cytosolic Ca²⁺ concentration was monitored using confocal microscopy at resting sarcomere length (≈1.79 μm) and during severe (≈12%) increase in sarcomere length.

Results

Diastolic stretch provoked SiS, which was prevented by disrupting microtubules with colchicine, but unaffected by inhibition of NADPH oxidase 2 or scavenging of reactive oxygen species. SiS was absent in Na⁺‐ and Ca²⁺‐free external solution, suggesting that it requires transsarcolemmal influx of Na⁺ or Ca²⁺. Activation of Piezo1 increased baseline spark rate, which was further increased by stretch. TRPA1 (transient receptor potential ankyrin 1) activation also increased baseline spark rate, with no further change upon stretch. SiS was not detectable in the presence of streptomycin (a blocker of nonselective mechano‐sensitive ion channels), and HC‐030031 and A‐967079 (selective blockers of TRPA1), even when Piezo1 was activated. SiS was also observed in pig and human atrial cardiomyocytes.

Conclusions

In atrial cardiomyocytes, diastolic stretch enhances Ca²⁺‐spark rate through a mechanism that requires microtubular integrity and TRPA1 but that is independent of redox signaling. TRPA1 emerges as a key regulator of stretch‐induced Ca²⁺ handling in atrial cells, with potential implications for arrhythmogenesis.

Nonstandard Abbreviations and Acronyms

CaT

Ca²⁺ transient(s)

GsMTx‐4

Grammostola spatulata mechanotoxin 4

MSC

mechanosensitive channel(s)

MSC_NS

cation nonselective mechanosensitive ion channel(s)

NOX2

NADPH oxidase 2

ROS

reactive oxygen species

RyR

ryanodine receptor(s)

SiS

stretch‐induced increase in Ca²⁺‐spark rate

SL

sarcomere length(s)

SR

sarcoplasmic reticulum

TRPA1

transient receptor potential ankyrin 1

Research Perspective.

What Is New?

• Stretch is known to be proarrhythmic, but the underlying molecular mechanisms are poorly understood; this study focuses on stretch‐induced increase in Ca²⁺‐spark rate in atrial cardiomyocytes, revealing that Ca²⁺‐spark rate depends on an intact microtubule network, extracellular Na⁺ and Ca²⁺ influx, and functional TRPA1 (transient receptor potential ankyrin 1) channels but is independent of reactive oxygen species signaling.

What Question Should Be Addressed Next?

• Future research should delve deeper into TRPA1‐activated Ca²⁺ signaling pathways and the interaction between microtubules and TRPA1 under mechanical stretch in atrial cardiomyocytes.

• In vivo and clinical studies are needed to clarify the role of TRPA1 in both physiology and atrial fibrillation; these studies could further explore whether TRPA1 activity correlates with the risk or progression of atrial fibrillation, potentially identifying it as a predictive biomarker or therapeutic target.

Atrial and ventricular myocardium respond to stretch through an instantaneous increase in contractility (Frank – Starling response), followed by a slower rise in force production linked to a gain in cellular calcium (Anrepp effect), as well as through changes in electrophysiology, for example, in pacemaking rate (Bainbridge effect), action potential duration, refractory period, and conduction velocity (mechanoelectric coupling). ¹ , ² , ³ , ⁴ Although these mechanically induced changes contribute to normal pump function, they can also promote arrhythmogenesis. ⁵ Stretch, for example, has been linked to the triggering of premature beats and increased susceptibility to atrial and ventricular fibrillation. ⁶ , ⁷

Mechanical modulation of Ca²⁺ handling is of central importance to cardiac mechanosensitivity. ⁸ A number of mechanisms have been proposed to alter myocardial Ca²⁺ handling in response to stretch. In ventricular cardiomyocytes, the open probability of RyR (ryanodine receptors) on the sarcoplasmic reticulum (SR) is enhanced by diastolic stretch, leading to more frequent occurrence of small, nonsynchronized, local Ca²⁺ release events commonly referred to as ‘Ca²⁺ sparks’. ⁹ , ¹⁰ This stretch‐induced increase in Ca²⁺‐spark rate (SiS) has been suggested to serve as a feedback mechanism during cardiac contraction, whereby cell shortening will decrease the open probability of RyR and thus contribute to termination of Ca²⁺‐induced Ca²⁺ release. ¹¹

In ventricular cardiomyocytes, the effect of stretch on Ca²⁺‐spark rate can be eliminated by disrupting microtubule integrity. ⁹ Microtubules have been proposed to mediate the interplay between stretch and RyR activity by transmitting mechanical signals directly from the cell surface to the SR membrane, modulating RyR activity. ⁹ Cardiomyocytes also respond to axial stretch with an increased reactive oxygen species (ROS) production, which can enhance the open probability of RyR and give rise to a sudden rise (within 10 seconds) in Ca²⁺‐spark rate. ¹⁰ Activation of sarcolemmal cation nonselective mechano‐sensitive ion channels (MSC_NS) constitutes another plausible mechanism that may link cardiomyocyte stretch and Ca²⁺ signaling. MSC_NS may modulate RyR activity either directly via Ca²⁺ influx or indirectly via Na⁺ influx increasing local Ca²⁺ concentration through effects on NCX (Na⁺/Ca²⁺ exchanger) activity. This has not been observed in rabbit ventricular cardiomyocytes, ¹⁰ whereas in mouse ventricular cardiomyocytes the MSC_NS Piezo1 has been proposed to be critical for the occurrence of SiS. ¹²

Although SiS has been investigated in some detail in ventricular cardiomyocytes, ⁹ , ¹³ , ¹⁴ its presence in atrial cardiomyocytes is unknown. Here, we applied sustained axial stretch to freshly isolated rabbit atrial cardiomyocytes and observed SiS. We investigated potential roles of microtubules, ROS, and MSC_NS in SiS, with the aim to identify the potential candidate of MSC_NS that may link axial stretch to changes in Ca²⁺‐spark rate in atrial cardiomyocytes. Furthermore, we studied atrial cardiomyocytes from pig and human to assess the potential translational relevance of SiS.

METHODS

Data Availability

The data that support the findings of this study are available from the corresponding author upon reasonable request.

Ethics

All animal experiments were approved by the national animal welfare bodies responsible for participating laboratories (The Netherlands: DEC2014‐112; Germany: Regierungspräsidium Freiburg, X‐16/10R; Spain: Comunidad de Madrid, Ref#PROEX078.8/21). All animal procedures conformed to the guidelines stated in Directive 2010/63/EU of the European Parliament on the protection of animals used for scientific purposes. All procedures on human tissues conformed to the principles outlined in the Declaration of Helsinki. The human heart tissue, together with clinical patient data, were collected by the CardioVascular BioBank of the University Heart Center Freiburg. The use of these samples and data were approved by the ethics committee of the University of Freiburg (ethics vote number 393/16 for the CardioVascular BioBank, and 216/20 for this study). The informed written consent from patients or their legal guardians was given before the inclusion of human heart tissue in the study. Next we describe key aspects of the materials and methods used in this study; further detail can be found in Data S1.

Animal Model

Experiments at Cardiovascular Research Institute Maastricht used New Zealand white rabbits (age ≈ 12 weeks, both sexes, weight 2–3 kg, N=27). Rabbits were euthanized by a concussive blow to the head (approved by the Dutch animal welfare body [DEC2014‐112]). Experiments at the Institute for Experimental Cardiovascular Medicine in Freiburg used New Zealand white rabbits (age ≈ 10 weeks, both sexes, weight 2–3 kg, N=22). As previously described, ¹⁵ rabbits were anesthetized via intramuscular injection of esketamine hydrochloride (Ketanest S 25 mg/mL, Pfizer Pharma PFE GmbH, Berlin, Germany; 0.5 mL/kg body weight) and xylazine (Rompun 2%, Bayer Vital GmbH, Leverkusen, Germany; 0.2 mL/kg body weight). During anesthesia, 5 mg of esketamine hydrochloride was given intravenously. Euthanasia was performed by intravenous injection of thiopental (Thiopental Inresa 0.5 g, Inresa Arzneimittel GmbH, Freiburg, Germany; 12.5 mg/mL) until apnea.

The left atrial tissues from pigs were obtained from Filgueiras‐Rama's laboratory. The slow growing cross breeding Yucatan‐Large white pigs (age ≈ 1.7 years, both sexes, weight ≈ 102 kg, N=3) with sinus rhythm were used. The anesthesia and euthanasia employed were conducted in accordance with established protocols as described previously. ¹⁶ In brief, intramuscular ketamine (20 mg/kg) and midazolam (0.5 mg/kg) were used for premedication and a bolus of fentanyl (0.010 mg/kg) was given intravenously for anesthesia induction. Ventilation was performed using a volumetric ventilator and administration of 2% to 2.5% sevoflurane. A heparin bolus was administered (300 International Units/kg) and an intravenous bolus of fentanyl (50 mg/mL) was given for analgesic control during sternotomy. Deep anesthesia and mechanical ventilation were necessary to preserve the heart in optimal conditions before excision. Thereafter, euthanasia was performed by exsanguination. After excision, the pig left atrial tissue was flushed with cold (4 °C) cardioplegic solution (Dr Franz Köhler Chemie GmbH, Bensheim, Germany) for blood removal and preservation during tissue transport lasting of up to 15 hours.

Human right atrial tissue was obtained from patients undergoing transplantation for terminal heart failure or open heart surgery for complex congenital heart defects (Table S1).

Cardiomyocyte Isolation

Rabbit atrial cardiomyocytes were isolated using 2 methods: Langendorff perfusion‐ or tissue slice‐based enzymatic digestion (Data S1, Table S2). ¹⁵ Pig and human atrial cardiomyocytes were isolated by the latter method.

Cell Stretch

Two glass rods (35 μm in diameter) were attached to cardiomyocytes using the bioadhesive MyoTak glue (IonOptix LLC, Westwood, MA, USA). Cells were stretched at a rate of 0.5 μm/ms from diastolic resting length, using movement in the cell‐axial direction of one rod, controlled by a piezo‐electric actuator. To measure passive and active tension, the other glass rod was attached to a force transducer. Rod displacement was prescribed based on the distance between the 2 glass rods after attachment to the resting cell. The relative increase in rod distance was greater than the actual change in sarcomere length (SL). In this study, a 10% and 20% increase in distance between rods, referred to as moderate and severe stretch, caused an increase in SL of ≈ 6% and 12%, respectively (Table S3). In pig and human atrial cardiomyocytes, only severe stretch was applied. Larger strains could not be achieved consistently, due to glass rod sliding or detachment of cells.

Selection of cells for analysis was performed according to the following inclusion criteria: (1) rod‐shaped morphology with clear homogenous striations and without membrane blebbing; (2) ability to follow electrical pacing at a frequency of 2 Hz without extra or missing beats; (3) ability to withstand 20% stretch with limited glass rod sliding (Table S3); and (4) lack of stretch‐induced self‐sustained contractions or cell death.

Recording Conditions

Atrial cardiomyocytes were loaded with Fluo‐4 AM (5 μmol·L⁻¹; Thermo Fisher, Waltham, MA, USA) dissolved in DMSO (containing [w/v] 20%—Pluronic F127) for 10 minutes, followed by 20 minutes in physiological saline solution 1 containing (in mmol·L⁻¹): 135 NaCl, 5 KCl, 1 MgCl₂, 10 glucose, 10 HEPES at pH 7.4 with NaOH, supplemented with 1.8 mmol·L⁻¹ Ca²⁺ for deesterification of the dye at 37 °C. Fluorescence intensity was recorded using line‐scanning confocal microscopy (Data S1). After electrical pacing at 2 Hz for 1 minute to achieve steady‐state SR loading, baseline spark activity was recorded for 20 seconds, followed by 20 seconds of stretch; ¹⁷ time‐matched control activity was recorded without stretch in a subset of cells.

Isometric stretch (5% increments, 20 seconds per step) was applied to rabbit atrial cardiomyocytes during 1 Hz field stimulation, with simultaneous recording of SL and force. Isometric peak force and Ca²⁺ transients (CaT) were recorded during 2 Hz field stimulation at different levels of preload (baseline, moderate, and severe stretch). ¹⁷ For isometric force and SR Ca²⁺ content measurements, physiological saline solution 1 was used. However, atrial cardiomyocytes isolated by Langendorff‐based enzymatic digestion had very low Ca²⁺‐spark rates, and 4 of 6 isolations resulted in cells with no detectable Ca²⁺ sparks in the presence of 1.8 mmol·L⁻¹ Ca²⁺. Ca²⁺ sparks were therefore studied in physiological saline solution 1 containing 3.6 mmol·L⁻¹ Ca²⁺, after preconditioning of cells by electrical pacing (in these conditions, all cell isolations allowed for Ca²⁺ spark observation).

In order to assess the robustness of SiS, further experiments were conducted in an independent laboratory, using a slice‐based cell isolation technique ¹⁵ at room temperature (22 °C, external Ca²⁺ concentration 1.8 mmol·L⁻¹; Data S1). Ca²⁺ sparks were observed after preconditioning of the cells by electrical pacing (as described previously) in physiological saline solution 2 (in mmol·L⁻¹): 137 NaCl, 4 KCl, 1 MgCl₂, 10 HEPES, 10 glucose, 1.8 CaCl₂, 20 taurine, 10 creatine, 5 adenosine, and 2 L‐carnitine at pH 7.3 with NaOH.

The roles of ROS, microtubules, and MSC_NS were studied using pharmacological interventions, and the contribution of transsarcolemmal Na⁺ or Ca²⁺ flux was assessed using Na⁺‐ and Ca²⁺‐free solution (0NC; Data S1).

Expression and Function of Transient Receptor Potential Ankyrin 1 Channels

The Western blot and patch clamp (cell‐attached mode) techniques were used to detect protein expression and channel activity of TRPA1 channels, respectively, in rabbit atrial cardiomyocytes (Data S1, Table S4).

Statistical Analysis

Ca²⁺ sparks were analyzed automatically, using the ImageJ Plugin Spark Master, ¹⁸ and their frequency was expressed as events per second, normalized to 100 μm scan line length (sparks·100 μm⁻¹·s⁻¹). Data are shown as mean±SEM. Statistical comparisons were made using Wilcoxon matched‐pairs signed rank test, Mann–Whitney U test, mixed‐effects analysis followed by Šídák's multiple comparisons test ¹⁹ or Friedman test for multiple comparisons, as appropriate (stated in figure legends); a P value <0.05 was considered indicative of statistically significant differences between means. Data were plotted with PRISM 10 (GraphPad Prism Software Inc, Boston, MA, USA). All experiments were conducted on at least 3 independent animals or patients, with the study size aided by power calculations. Throughout the article, “n” indicates the number of cells included in the analysis, and “N” indicates the number of animals.

RESULTS

Acute Effects of Axial Stretch on Force and CaT in Atrial Cardiomyocytes

Atrial cardiomyocytes were stretched axially using 2 glass rods while being paced at 2 Hz of field stimulation (Figure 1A). Stretch increased diastolic SL (Figure 1B), average SL rose from 1.79±0.01 μm in nonstretched cells to 1.91±0.02 μm with moderate stretch (≈6% SL increase), and to 2.01±0.02 μm with severe stretch (≈12%, Table S3). The largest SL observed in individual cells during stretch was ≈ 2.2 μm (Figure 1C), which is within the physiological range during diastolic filling. ²⁰ Active force production increased with SL (Figure 1D). The presence of this stretch‐induced increase in active force development (Frank–Starling response) confirms that isolated cells showed physiological functional behavior.

Figure 1. Technical approach to single atrial myocyte stretch.

A, Rabbit atrial cardiomyocyte, attached to glass microrods coated with a biological adhesive. Top: nonstretched; Bottom: during severe stretch. Yellow box: area used for calculation of SL. B, Example of force recordings at different levels of isometric stretch, increasing resting SL. Force traces shown were recorded at the end of each 20‐second stretch step. C, Relation between amount of stretch (calculated as the relative increase in distance between glass rods) and change in SL. The right y axis shows relative changes in mean SL (ΔSL, expressed as percentage difference relative to mean SL at 0% stretch). Friedman test with Dunn's post hoc test was performed to test for statistical significance. n/N=5/3. D, Isometric peak force as a function of SL, linearly fitted, showing 95% CIs. n/N=8/3. All experiments were performed with physiological saline solution 1 containing 1.8 mmol·L⁻¹ CaCl₂ at 37 °C. SL, sarcomere length.

The Frank–Starling response occurs in the absence of noticeable changes in systolic CaT, ⁴ and this was confirmed here. To characterize the relationship between Ca²⁺, active force, and diastolic stretch in rabbit atrial cardiomyocytes, we simultaneously measured force of contraction (Figure 2A and 2C) and CaT (Figure 2B and 2D) during 3 consecutive beats immediately before and during 20 seconds after the onset of the different levels of stretch (2 Hz pacing). The superimposed traces (Figure 2A) illustrate a stretch‐dependent increase in peak force, as well as increases in maximum velocity of force generation and relaxation (Figure 2C). Time‐to‐peak force development and time from peak force to 50% relaxation were not significantly altered (Table S5). The superimposed CaT illustrated that stretch did not significantly affect the amplitude, the time to 50% peak fluorescence, or the time constant of CaT decay (Figure 2D). Ca²⁺‐transient parameters are summarized in Table S5.

Figure 2. Changes in systolic Ca²⁺ and force parameters in rabbit atrial cardiomyocytes during moderate and severe stretch.

Representative example of (A) force of contraction and (B) CaT, recorded simultaneously at 20 seconds after the onset of stretch in an atrial cardiomyocyte paced at 2 Hz. Superimposed black, blue, and red traces are representing baseline, moderate, and severe stretch, respectively. C, Summary data illustrating change in force (mN·mm⁻²) and maximum velocity of contraction (+dF/dt_max) or relaxation (−dF/dt_max) as a function of diastolic stretch. n/N=7/3. D, Left, CaT amplitude (F/F ₀); Middle, time to 50% peak fluorescence (TF₅₀, in ms); Right, time constant of decay of Ca²⁺ transient (tau, in ms). n/N=9/4. Friedman test with Dunn's post hoc test used for (C) and (D). All experiments shown in this figure were performed with physiological saline solution 1 containing 1.8 mmol·L⁻¹ CaCl₂ at 37 °C. CaT indicates Ca²⁺ transients; +dF/dt_max, maximum velocity of force generation; −dF/dt_max, maximum velocity of relaxation; Tau, time constant of decay of Ca²⁺ transient; and TF₅₀, time from peak force to 50% relaxation.

Atrial Cardiomyocyte Stretch Increases Diastolic Ca²⁺‐Spark Rate

Next, we examined the effects of stretch on Ca²⁺ sparks in atrial cardiomyocytes (here in the presence of high extracellular Ca²⁺: 3.6 mmol·L⁻¹). The general protocol is illustrated in Figure 3A, where the duration of stretch is indicated by the red bar below the fluorescence surface plot. SiS amplitude was dependent on the magnitude of stretch. It remained indistinguishable from control at moderate stretch, yet increased significantly upon severe stretch (from 1.18±0.14·[100 μm·s]⁻¹ to 1.84±0.19·[100 μm·s]⁻¹, P<0.0001; spark rate measured during 20 seconds of baseline or stretch, Figure 3B). Although data in Figure 3B are paired, data points are not linked for sake of visual clarity. Notably, 49 of 65 cells exhibited higher Ca²⁺‐spark rates during severe stretch and 11 of 18 cells exhibited higher Ca²⁺‐spark rates during moderate stretch.

Figure 3. Stretch‐activated diastolic Ca²⁺‐release events in rabbit atrial cardiomyocytes.

A, Representative example of a fluorescence surface plot of a confocal line scan through the center of a rabbit atrial cardiomyocyte, showing diastolic Ca²⁺‐release events (Ca²⁺ sparks) before and during severe stretch in cells preconditions by 60‐second pacing at 2 Hz. B, Effect of stretch on Ca²⁺‐spark rate during rest and moderate or severe stretch. n/N=15/4 and 65/12, respectively. C, The histogram (1 second bin size) shows the count of Ca²⁺ sparks illustrating a stretch‐induced increase in Ca²⁺‐spark rate upon severe stretch. n/N=65/12. D, Enlarged XT‐plot of fluorescence intensity of a single Ca²⁺ spark, illustrating the analysis of spark properties, including (E) amplitude (ΔF/F ₀), duration (FDHM), and width (FWHM), at baseline and during severe stretch. n/N=11/3. Wilcoxon matched pairs signed rank test was for (B) and (E). All experiments shown in this figure were performed with physiological saline solution 1 containing 3.6 mmol·L⁻¹ CaCl₂ at 37 °C. FDHM indicates full duration at half maximum fluorescence; FWHM, full width at half maximum fluorescence.

Application of severe stretch was accompanied by a robust increase in Ca²⁺‐spark rate. SiS was rapid in onset (i.e., detected within the first 1‐second analysis bin after application of stretch) and sustained over the 20‐second period of stretch, as illustrated in Figure 3C. As reported before in ventricular cells, whereas Ca²⁺‐spark rate was increased by severe stretch, individual Ca²⁺‐spark characteristics, such as spark amplitude, full duration at half maximum fluorescence, or full width at half maximum fluorescence were not significantly affected by cell stretch (Figure 3D and 3E). The caffeine‐induced SR Ca²⁺ release following Ca²⁺ spark recordings in atrial cardiomyocytes was not significantly different between baseline and severe stretch conditions (F _caff/F ₀: 7.75±0.65 at baseline versus 6.83±0.39 in severe stretch condition, P=0.2199, n/N=10/5 and 16/5 respectively, Figure S1), suggesting that the SR Ca²⁺ content was not significantly altered within the first 20 seconds of stretch in atrial cardiomyocytes.

SiS Requires Microtubular Integrity But Not ROS Production

As the results demonstrate that SiS is most pronounced when severe stretch is applied, underlying molecular mechanisms were investigated using this protocol. Previous work in rat ventricular cardiomyocytes suggested that SiS may require microtubular integrity, ⁹ and work on mouse ventricular cardiomyocytes suggested that SiS relies on mechanosensitive ROS production. ¹⁰ In the present study, disruption of microtubules with colchicine (10 μmol·L⁻¹ for 2 hours) abrogated SiS in rabbit atrial cardiomyocytes (Figure 4A and 4B).

Figure 4. Role of microtubules and ROS in stretch‐induced increase in Ca²⁺‐spark rate in rabbit atrial cardiomyocytes.

A, Fluorescence surface plots of atrial cardiomyocytes before and during severe stretch, in control cells and following administration of colchicine (an inhibitor of microtubule polymerization, 10 μmol·L⁻¹), gp91ds (an inhibitor of NOX2, 3 μmol·L⁻¹) and scrambled (nontargeted) peptides (3 μmol·L⁻¹), and N‐acetylcysteine (a ROS scavenger, 10 mmol·L⁻¹). B through D, Ca²⁺‐spark rate at baseline and during stretch in atrial cardiomyocytes exposed to the aforementioned pharmacological agents. n/N=24/5, 14/4, 12/5, 9/4, and 24/5, respectively. E, Ca²⁺‐spark rate at baseline and during stretch in control or Na⁺‐ and Ca²⁺‐free solution, n/N=42/9 and 15/4, respectively. Wilcoxon matched‐pairs signed rank test was used for (B), (C), and (D) to assess statistical significance. Mixed‐effects analysis followed by Šídák's multiple comparisons test was used for control, Na⁺/Ca²⁺‐free (E) and GsMTx‐4 (Figure S5) to assess statistical significance. All experiments were carried out in physiological saline solution 1 containing 3.6 mmol·L⁻¹ CaCl₂ at 37 °C. NAC indicates N‐acetylcysteine; NOX2, NADPH oxidase 2; and ROS, reactive oxygen species.

NOX2 (NADPH oxidase 2) inhibition by gp91ds‐tat failed to prevent SiS: Ca²⁺‐spark rate still increased from 1.17±0.20·(100 μm·s)⁻¹ to 3.06±0.94·(100 μm·s)⁻¹ upon stretch (P=0.0391, Figure 4A and 4C). To test whether ROS production by sources other than NOX2 influences Ca²⁺‐spark rate during stretch, we examined the effect of the ROS scavenger N‐acetylcysteine. We found that severe stretch still increased Ca²⁺‐spark rate from 0.97±0.14·(100 μm·s)⁻¹ to 1.44±0.22·(100 μm·s)⁻¹ (P=0.0150, Figure 4A and 4D). Neither gp91ds‐tat nor N‐acetylcysteine significantly altered Ca²⁺‐spark frequency in nonstretched cells. Taken together, these results suggest that in rabbit atrial cardiomyocytes, microtubules, but not ROS, are required for SiS.

SiS Relies on Trans‐Sarcolemmal Ca²⁺ Influx in Atrial Cardiomyocytes

We next examined whether transsarcolemmal Ca²⁺ entry is required for SiS in rabbit atrial cardiomyocytes. Cells were electrically stimulated as before (Figure 3A) and then superfused with the 0NC solution using a local perfusion system, followed immediately by Ca²⁺ fluorescence imaging. Solution exchange time was ≈ 0.5 second. In 0NC solution, baseline Ca²⁺‐spark rate was not significantly different compared with control conditions (Figure 4E), whereas SiS was abolished, suggesting that atrial cardiomyocytes require transsarcolemmal influx of Ca²⁺ or Na⁺ for SiS.

Robustness of SiS in Atrial Cardiomyocytes

To test the robustness of SiS in atrial cardiomyocytes, experiments were conducted in another laboratory with a modified cell isolation protocol (slice‐based method) and recording conditions (1.8 mmol·L⁻¹ Ca²⁺, room temperature). SiS was reproducible in these conditions. Upon severe stretch, Ca²⁺‐spark rate increased from 0.27±0.07·(100 μm·s)⁻¹ to 0.41±0.09·(100 μm·s)⁻¹ (P=0.0048, Figure 5A and 5B; no significant change in Ca²⁺‐spark rate was detected in a time‐matched control group without stretch, Figure S2). SiS was observed only in the stretched part of the cell (i.e., between 2 glass rods; spark rate increased from 0.34±0.12·(100 μm·s)⁻¹ to 0.68±0.13·(100 μm·s)⁻¹, P=0.0005), whereas the free cell ends showed no significant difference Ca²⁺‐spark rate (from 0.36±0.10·(100 μm·s)^‐1 to 0.38±0.12·(100 μm·s)^‐1, P>0.9999, Figure S3). Consistent with previous findings (Figure 3E), no stretch‐induced alterations of individual Ca²⁺ spark characteristics, such as amplitude, full duration at half maximum fluorescence, or full width at half maximum fluorescence were detected (Figure S4).

Figure 5. Role of MSC_NS in the stretch‐induced increase in Ca²⁺‐spark rate in rabbit atrial cardiomyocytes.

A, Fluorescence surface plots of atrial cardiomyocytes before and during application of severe stretch in control conditions, and following treatment with streptomycin (STP, a nonspecific blocker of MSC_NS, 40 μmol·L⁻¹), Yoda1 (a chemical activator of Piezo1, 20 μmol·L⁻¹), allylisothiocyanate (AITC, a selective agonist of TRPA1, 50 μmol·L⁻¹), HC‐030031 (HC, a selective blocker of TRPA1, 10 μmol·L⁻¹), A‐967079 (A96, a selective blocker of TRPA1, 10 μmol·L⁻¹); HC‐030031 plus Yoda1 (HC + Yoda1, 10 μmol·L⁻¹ and 20 μmol·L⁻¹, respectively), and A‐967079 plus Yoda1 (A96 + Yoda1, 10 μmol·L⁻¹ and 20 μmol·L⁻¹, respectively). B, Summary data on Ca²⁺‐spark rate at baseline and during stretch in atrial cardiomyocytes exposed to the aforementioned pharmacological agents. n/N=20/9, 10/4, 10/3, 10/4, 10/4, 11/4, 11/3 and 10/4, respectively. C, Role of MSC_NS and potential candidates in baseline Ca²⁺‐spark rate. D, Role of MSC_NS and potential candidates in Ca²⁺‐spark rate during stretch. Mixed‐effects analysis followed by Šídák's multiple comparisons test for (B), (C), and (D) was used. Pairwise comparisons were performed among the groups in (B). All experiments were carried out in physiological saline solution 2 containing 1.8 mmol·L⁻¹ CaCl₂ at room temperature. MSC_NS, cation nonselective mechanosensitive ion channel(s); and TRPA1, transient receptor potential ankyrin 1.

Involvement of MSC_NS

To test the potential involvement of MSC_NS in SiS, we applied streptomycin (STP, 40 μmol·L⁻¹), ²¹ a nonselective blocker of MSC_NS, and observed that SiS was abolished, indicating an involvement of MSC_NS (Figure 5A and 5B).

Various MSC_NS are expressed in cardiac tissue. ²² We assessed TRPA1, as it is present in cardiomyocytes, microtubule dependent, and known to be Ca²⁺ permeable. ²³ , ²⁴ To test the contribution of TRPA1 to SiS in atrial cardiomyocytes, we used 2 selective TRPA1 antagonists: HC‐030031 (10 μmol·L⁻¹) ²⁵ , ²⁶ and A‐967079 (10 μmol·L⁻¹). ²⁷ , ²⁸ We also applied the Piezo1 agonist Yoda1 (20 μmol·L⁻¹), ¹² as Piezo1 has recently been linked to Ca²⁺ sparks in mouse ventricular cardiomyocytes. ¹²

SiS was abolished in the presence of HC‐030031 and A‐967079, even if combined with Piezo1 activation by Yoda1, suggesting a critical role for TRPA1 in SiS (Figure 5A and 5B).

In the presence of allylisothiocyanate (a selective activator of TRPA1, 50 μmol·L⁻¹), ²⁸ , ²⁹ we observed a higher baseline Ca²⁺‐spark rate (0.91±0.08·(100 μm·s)⁻¹ compared with 0.25±0.06·(100 μm·s)⁻¹ in control, P<0.0001, Figure 5C). After pre‐activation of TRPA1, stretch did not induce a significant further increase in spark rate (Figure 5A and 5B). The Ca²⁺‐spark rate during stretch in allylisothiocyanate condition was significantly higher than in control condition (stretch with no drug, P<0.0001, Figure 5D).

We also tested Grammostola spatulata mechanotoxin 4 (GsMTx‐4, 2 μmol·L⁻¹; 10 minute preincubation), ⁹ a well‐known blocker of MSC_NS that, intriguingly, has been reported to also activate TRPA1. ³⁰ Cells exposed to GsMTx‐4 showed a higher Ca²⁺‐spark rate at resting length, compared with cells in control solution (Figure S5). This elevated baseline Ca²⁺‐spark rate was not significantly different from the Ca²⁺‐spark rate seen in control cells during stretch. In cells pretreated with GsMTx‐4, stretch did not significantly alter Ca²⁺‐spark rate (Figure S5).

Although preactivation of Piezo1 by Yoda1 induced a higher Ca²⁺‐spark rate in cells at resting length (0.66±0.10·(100 μm·s)⁻¹ compared with 0.25±0.06·(100 μm·s)⁻¹ in control, P=0.0010, Figure 5C), stretch still caused SiS (Ca²⁺‐spark rate increased from 0.51±0.10·(100 μm·s)⁻¹ to 0.74±0.08·(100 μm·s)⁻¹, P=0.0024, Figure 5A and 5B). Additionally, we observed that the Ca²⁺‐spark rates at baseline and during stretch in Yoda1 were both significantly higher than those in A‐967079 + Yoda1 condition (Figure 5C and 5D). Of note, in nonstretched cells, neither streptomycin nor the blockers of TRPA1 (whether combined with Yoda1 or not), significantly altered basal Ca²⁺‐spark rate (Figure 5C).

The 95% CI of differences between groups in Figure 5B through 5D are provided in Tables S6 and S8, respectively.

Protein Expression and Function of TRPA1 in Rabbit Atrial Cardiomyocytes

We combined biochemical and electrophysiological techniques to test further the presence and activity of TRPA1 in atrial cardiomyocytes. Western blot results indicate that TRPA1 protein was expressed in rabbit left atrial cardiomyocytes (Figure 6A).

Figure 6. Expression and function of TRPA1 channel in rabbit atrial cardiomyocytes.

A, TRPA1 and GAPDH protein expression in rabbit left atrial cardiomyocytes using the Western blot technique. N=3. B, Representative recording of control and allylisothiocyanate (50 μmol·L⁻¹) at rest (0 mm Hg) and during stretch (−80 mm Hg) in a cell‐attached patch at holding potential of −80 mV. C and D, Quantification of channel open probability and average currents in control and allylisothiocyanate at rest and during stretch. n/N=12/3 in control and 19/3 in allylisothiocyanate. Wilcoxon matched‐pairs signed rank test and Mann–Whitney U test for (C) and (D) were used for assessment of statistical significance. AITC indicates allylisothiocyanate; and TRPA1, transient receptor potential ankyrin 1.

The patch clamp technique was then employed to conduct voltage clamp experiments in the cell‐attached configuration. Higher channel opening probability was induced by stretch (suction at −80 mm Hg) applied to the membrane in control conditions (from 0% to 1.81±0.65%, P=0.0156) and in presence of allylisothiocyanate (from 7.57±2.91% to 25.77±5.64%, P<0.0001, Figure 6B and 6C). Additionally, a significantly higher channel opening probability was found in allylisothiocyanate‐treated cells at baseline (0 mm Hg; P=0.0046) and during stretch (−80 mm Hg; P<0.0001, Figure 6B and 6C) when comparing to control.

Although no significant differences of average currents were observed in control (P=0.79), stretch markedly enhanced the average currents in allylisothiocyanate‐treated cells (from −0.89±0.20 pA to −1.52±0.29 pA, P=0.0046, Figure 6B and 6D). Moreover, the average current was strongly enhanced in allylisothiocyanate‐treated cells at baseline (P=0.0180) and during stretch (P=0.0007, Figure 6B and 6D) when compared with the control.

These findings confirm the expression of functional TRPA1 in rabbit atrial cardiomyocytes, further supporting its function in modulating Ca²⁺ sparks in response to stretch in this study.

SiS in Atrial Cardiomyocytes of Other Species

To assess the potential translational relevance of atrial SiS, we analyzed Ca²⁺‐spark rate in atrial cardiomyocytes, freshly isolated from the left atrium of pigs with sinus rhythm and right atrium of patients with terminal heart failure undergoing transplantation and complex congenital heart defect. SiS was present in both species (Figure 7A through 7C) and was detected only in the stretched part of the cells, that is, between glass rods (Figure S6). As in rabbit cardiomyocytes, stretch did not significantly alter spark amplitude, full duration at half maximum fluorescence, and full width at half maximum fluorescence (Figure 7D and 7E).

Figure 7. SiS in atrial cardiomyocytes from pig and human.

A, Fluorescence surface plots of atrial cardiomyocytes from pig (left) and human (right) before and during application of severe stretch. B and C, Summary data on Ca²⁺‐spark rate at baseline and during stretch in atrial cardiomyocytes from pig and human, respectively. D and E, Amplitude (ΔF/F ₀), duration (FDHM), and width (FWHM) at baseline and during stretch in atrial cardiomyocytes from pig (n/N=14/3) and human (11/3), respectively. Wilcoxon matched‐pairs signed rank test was used for (B through E) to assess statistical significance. All experiments were carried out in physiological saline solution 2 containing 1.8 mmol·L⁻¹ CaCl₂ at room temperature. FDHM indicates full duration at half maximum fluorescence; FWHM, full width at half maximum fluorescence; and SiS, stretch‐induced increase in Ca²⁺‐spark rate.

Western blot results indicate that TRPA1 protein is expressed in human right atrial cardiomyocytes (Figure S7A), and TRPA1 activation by allylisothiocyanate leads to a significant increase in baseline Ca²⁺‐spark rate in human right atrial cardiomyocytes (Figure S7B).

DISCUSSION

We investigated the effect of axial stretch on Ca²⁺ sparks in atrial cardiomyocytes. Our main findings are as follows: (1) axial stretch increases rabbit atrial cardiomyocyte force production without significantly affecting systolic CaT amplitude and dynamics; (2) severe stretch, leading to ≈ 12% SL increase in rabbit atrial cardiomyocytes, raises Ca²⁺‐spark rate without significantly altering individual Ca²⁺‐spark properties; (3) SiS requires external Na⁺ or/and Ca²⁺, an intact microtubular network, and functional TRPA1, but not ROS; and (4) SiS is present also in pig and human atrial cardiomyocytes.

Stretch Effects on Ca²⁺ Transients and Sparks in Rabbit Atrial Cardiomyocytes

In rabbit atrial cardiomyocytes, increasing stretch enhanced pacing‐induced contraction amplitudes but was not associated with significant changes in CaT, in keeping with the Frank–Starling response in atrial cardiomyocytes. ³¹ Like in ventricular cells, ¹³ baseline Ca²⁺‐spark rate was highly variable between individual cardiomyocytes. The fluorescence surface plots also demonstrate a high variability in amplitude, duration, and width of the individual Ca²⁺ sparks that may be due to a variable number of RyR that constitute a single Ca²⁺‐release unit. ³² Ca²⁺‐spark rate increased upon acute stretch ⁹ , ¹⁰ and remained elevated for at least 20 seconds. SiS in rabbit atrial cardiomyocytes was not associated with significant changes in spark amplitude, duration, or width, confirming previous data in rat ventricular cells. ⁸ We also confirm that SiS is a local phenomenon as it appears only in the stretched region of a cell and not in unstretched parts. ⁹

Possible Mechanisms of SiS in Rabbit Atrial Cardiomyocytes

Direct evidence that stretch causes diastolic Ca²⁺ leak via RyR located in the SR was provided in studies of mouse and rat ventricular cardiomyocytes. ⁹ , ³³

Several mechanisms have been proposed to link mechanical stretch and RyR open probability. ⁹ , ¹⁰ , ³⁴ , ³⁵ , ³⁶ Our data demonstrate that interfering with microtubule integrity using colchicine ³⁷ prevents SiS in rabbit atrial cardiomyocytes. Being intertwined between the transverse‐tubular sarcolemma and the SR membrane, ⁸ microtubules could transmit mechanical cues from the sarcolemma to the membrane of the SR and thus induce a stretch‐dependent enhancement of RyR open probability.

Additionally, we observed that neither NOX2 nor ROS were critically required in SiS, data are in keeping with previous observations in rat ventricular cardiomyocytes (where SiS was observed in presence of L‐NAME) ⁹ but different from observations in mouse ventricular cardiomyocytes., ¹⁰ which found that the inhibition of NOX2 prevented SiS. Further investigations are necessary to clarify the role of NOX2 and ROS in SiS across species and cardiac chambers.

The RyR are highly sensitive to Ca²⁺. Increased Ca²⁺ concentration both on the intraluminal sarcoplasmic reticular side of the RyR and on the cytosolic side will enhance the open probability of the RyR. ³⁸ The former mechanism of increased RyR Ca²⁺ sensitivity is one of the rationales for loading the SR toward a steady state of Ca²⁺ content, the latter mechanism constitutes the basis for Ca²⁺‐induced Ca²⁺ release, such as in excitation contraction coupling. ³⁹

Whether SiS depends on Na⁺ or Ca²⁺ influx from the extracellular space was tested by acute removal of Na⁺ and Ca²⁺ from the perfusate. In this condition, the baseline Ca²⁺‐spark rate was not significantly different from control conditions (with Ca²⁺ in the bath), yet stretch no longer induced SiS. We conclude that, in contrast to ventricular cardiomyocytes, SiS requires transsarcolemmal influx of Na⁺ or Ca²⁺ in rabbit atrial cardiomyocytes. Whether stretch‐induced RyR activation depends directly on Ca²⁺ influx or indirectly on Na⁺ influx is not known. ⁴⁰ Enhanced Na⁺ influx is expected to reduce the driving force for NCX‐mediated Ca²⁺ efflux, thereby potentially increasing the Ca²⁺ concentration in the restricted space between the transverse tubular system and the SR.

Role of MSC_NS in SiS in Rabbit Atrial Cardiomyocytes

The dependence of SiS on extracellular Na⁺ or Ca²⁺ suggests that MSCs in the sarcolemma are involved. ²² Here, a nonselective blocker of MSC_NS, streptomycin, prevented SiS, confirming the involvement of MSC_NS as a mechanism of SiS in rabbit atrial cardiomyocytes. Counterintuitively, GsMTx‐4, a blocker of several MSC_NS (such as Piezo1, ⁴¹ TRPC1, ⁴² and TRPC6 ⁴³ ) increased baseline Ca²⁺‐spark rate to levels indistinguishable from stretched cells (and subsequent application of stretch did not give rise to a further significant change in spark rate). A possible explanation for this finding could be the activation of TRPA1 by GsMTx‐4. ³⁰ This channel is mechanosensitive and Ca²⁺ permeable and is also widely expressed in the cardiovascular system. ²⁴ It is involved in regulating intracellular Ca²⁺ concentrations ²⁵ and its activation might trigger Ca²⁺ sparks in rabbit atrial cardiomyocytes. This interpretation is in keeping with the effects of allylisothiocyanate, a selective activator of TRPA1, which increases baseline Ca²⁺‐spark rate (without further increase upon stretch), and with our observations upon selective block of TRPA1, using HC‐030031 and A‐967079, which prevented SiS.

Finally, Piezo1 opening by Yoda1 induced an increase in Ca²⁺‐spark rate. This indicates that Piezo1 could, if activated, contribute to the adjustment of Ca²⁺‐spark rate in atrial cardiomyocytes. Remarkably, the presence of SiS, even during Piezo1 activation by Yoda1, suggests an additional, Piezo1‐independent mechanism. This is in line with the absence of SiS when TRPA1 is blocked (Piezo1 being present), indicating that Piezo1 cannot compensate the role of TRPA1 in the conditions used for this study. A selective blocker for Piezo1 would be required to exclude a possible role of Piezo1 in SiS.

Interestingly, when TRPA1 is blocked by HC‐030031 or A‐967079, Yoda1 does not change significantly the Ca²⁺‐spark rate compared with control, with or without stretch. This suggests that Piezo1 cannot generate significantly more sparks without TRPA1 in these conditions, indicating a functional coupling between TRPA1 and Piezo1. This functional coupling could be ensured by Ca²⁺, similar to what previously described interactions between Piezo1 and TRPM4 ⁴⁴ or Ca²⁺‐activated, large conductance potassium‐channel BK, ⁴⁵ or KCa3.1 Gardos channel. ⁴⁶ A likely chain of events would be as follows: Piezo1 channels open, letting Ca²⁺ enter the cells, leading to activation of Ca²⁺‐sensitive channels. This would be in line with the Ca²⁺‐sensitive property of TRPA1. ²⁸ , ⁴⁷ Although unlikely, we cannot exclude a possible block of Piezo1 by HC‐030031 or A‐967079, or even an effect of Yoda1 on TRPA1 directly; such experiments will be the focus of future investigations.

Taken together, our findings suggest that the major MSC involved in SiS in rabbit atrial cardiomyocytes is TRPA1.

Protein Expression and Function of TRPA1 in Rabbit Atrial Cardiomyocytes

The TRPA1 protein expression was observed in rabbit left atrial cardiomyocytes by using the Western blot technique. However, due to the current technical limitations of anti‐TRPA1 antibody in fluorescence imaging, ⁴⁸ it remains challenging to determine the precise localization of TRPA1 in the sarcolemma (surface cell membrane or transverse‐axial tubular system). To address functional aspects, we conducted patch‐clamp experiments, demonstrating higher single channel activity in response to allylisothiocyanate and stretch.

The significantly enhanced channel opening probability during stretch was present in both control and TRPA1‐activated conditions compared with baseline. Additionally, the channel opening probability and current amplitude were notably higher in allylisothiocyanate‐activated conditions to control. This heightened response suggests a synergistic effect between chemical activation (via allylisothiocyanate) and stretch, amplifying Ca²⁺ influx and suggesting TRPA1 as a functional component in atrial mechanotransduction.

In the context of SiS in rabbit atrial cardiomyocytes, the augmented TRPA1‐like activity in response to stretch may play a direct role in modulating Ca²⁺ sparks, which could be a result of the localized Ca²⁺ entry through TRPA1.

The robust expression of TRPA1 protein found here in rabbit atria is also detected in rabbit ventricular myocardium, ⁴⁹ suggesting that even though molecular mechanisms underlying SiS appear to differ between cardiac chambers, expression of TRPA1 is unlikely to be the cause.

SiS in Atrial Cardiomyocytes of Other Species

SiS is present also in pig and human atrial cardiomyocytes, as in rabbit, severe stretch raises Ca²⁺‐spark rate, without affecting spark properties. Furthermore, functional TRPA1 protein expression was also detected in human right atrial cardiomyocytes. Activation of TRPA1 by allylisothiocyanate significantly increased the Ca²⁺‐spark rate in these cells, similar to rabbit atrial cardiomyocytes. Identification of SiS and confirmation of TRPA1 as mediating Ca²⁺ signaling in human cardiomyocytes highlight the translational potential of our observations.

Cross‐Laboratory Validation and Methodological Observations

This study was conducted in 2 different laboratories (in Maastricht and in Freiburg). We used 2 sets of experimental conditions, including differences in cell isolation protocols, Ca²⁺ concentrations, and temperature. The fundamental effects of stretch on rabbit atrial cardiomyocytes were qualitatively similar, increasing the confidence in observed behaviors.

Overall, the use of a live tissue slice‐based enzymatic digestion protocol, and of saline supplemented with amino acids (including taurine, creatine, adenosine, and L‐carnitine) would be the preferred model for further studies, as this yielded a larger number of usable cells with suitable Ca²⁺‐spark rates at rest (in 1.8 mmol·L⁻¹ Ca²⁺). A systematic analysis of the effects of recording temperature (22 °C versus 37 °C) would be relevant due to the reported temperature sensitivity of TRPA1 ⁵⁰ and deserves a separate investigation.

Limitations

Ca²⁺ sparks were recorded in line scan mode. The limited number of sparks that could be observed per cell may increase variability of the data.

We used MyoTak to achieve long (20 seconds) and robust stretches. Although MyoTak is widely used, we cannot entirely exclude the possibility of side effects associated with its application.

Evidence suggests that TRPA1 modulates intracellular Ca²⁺ via a CaMKII‐dependent pathway in rat cardiomyocytes. ²⁵ However, our study did not investigate the TRPA1‐activated Ca²⁺ signaling pathway in depth within rabbit atrial cardiomyocytes, which could represent an avenue for further research.

MSC_NS in Stretch‐Induced Atrial Arrhythmias

Acute atrial stretch can occur under various circumstances. Physiologically, it may result from increased venous return, ⁵¹ such as during coughing, intense exercise, or cold water immersion. Pathologically, stretch can occur as a consequence of acute atrioventricular valve regurgitation, for example upon rupture of chordae tendineae, or therapeutically such as during rapid intravenous fluid administration, or catheter displacement of the atrial wall leading to focal stretch. ⁵¹ , ⁵²

MSC_NS have been regarded as main candidates involved in stretch‐induced electrophysiological responses (such as alterations in action potential shape, including early and delayed after‐depolarizations), which may generate triggers and substrates for atrial fibrillation. ⁷ Blocking MSC_NS has been shown to reduce inducibility of atrial fibrillation during stretch, ⁵³ , ⁵⁴ making them a potential target for antiarrhythmic drug development. ⁵⁵ Our findings suggest that one of the major MSC_NS, TRPA1, enhances diastolic Ca²⁺‐spark rate during stretch in rabbit atrial cardiomyocytes, potentially contributing to a proarrhythmic environment. Future studies are warranted to investigate the specific roles of TRPA1‐activated Ca²⁺ signaling pathways in atrial physiology and during pressure‐ or volume‐load related atrial arrhythmias.

CONCLUSIONS

In atrial cardiomyocytes, stretch produces an increase in diastolic Ca²⁺‐spark rate via a mechanism that requires an intact microtubule network, external Na⁺ or Ca²⁺, and TRPA1 but not ROS. SiS may be translationally relevant as it is present also in pig and human atrial cardiomyocytes. Further research is needed to characterize the link between SiS, TRPA1, and mechanically induced atrial arrhythmias.

Sources of Funding

This work was supported by an EU Innovative Training Network (RADical reduction of oxidative stress in cardiovascular disease, No. 316738), by a grant from the Austrian Science Fund (I2013‐B27), and by the German Research Foundation (DFG) via a project grant (390939984) and the German Research Foundation (DFG) Excellence Strategy cluster CIBSS (EXC‐2189—2100249960). Jiaying Fu was supported by the China Scholarship Council (No. 202106380082). Teresa Schiatti was supported by the European Union’s Horizon 2020 research and innovation programme (under the Marie Skłodowska‐Curie grant agreement 860 974). Jiaying Fu, Breanne A. Cameron, Teresa Schiatti, Josef Madl, Ursula Ravens, Peter Kohl, Eva A. Rog‐Zielinska, and Rémi Peyronnet are members of the German Research Foundation (DFG) Collaborative Research Centre SFB1425 (DFG #422681845). Ursula Ravens is member of the board of German Atrial Fibrillation NETwork (AFNET). This work was further supported by the Netherlands Heart Foundation (Grant number 01‐002‐2022‐0118, Embrace: Electro‐Molecular Basis and Therapeutic Management of Atrial Cardiomyopathy, Fibrillation and Associated Outcomes) and the European Union (MAESTRIA: Machine Learning Artificial Intelligence Early Detection Stroke Atrial Fibrillation, No. 965286).

Disclosures

None.

Supporting information

Data S1–S2

Tables S1–S8

Figures S1–S7

UNEDITED GELS