The animal study was reviewed and approved by the ethical committee of Bordeaux, Caen and Toulouse, France. All experiments were conducted in accordance with the European Union directive EU/2010/63.

6- to 10-week-old EPAC WT (C57Bl/6), EPAC1^−/−, and EPAC2^−/− male mice were used in this study. Global knock-out for EPAC1 isoform has been described previously (Laurent et al., 2015).

RAPGEF4 gene was floxed with two LoxP sites around Ex 17 by Institut Clinique de la Souris (IGBMC, Strasbourg) using homologous recombination in C57/Bl6N ES cells. The Myh6-Cre mice with a C57/Bl6J genetic background expressing Cre recombinase in cardiomyocytes were purchased from Janvier labs (France). Bl6N have a functional, wild-type Nicotinamide Nucleotide Transhydrogenase (NNT) while Bl6J have a defective allele and loss of enzymatic activity has been associated with mitigated heart failure upon pressure overload (Nickel et al., 2015). Heterozygous EPAC2 flox and Myh6-Cre mice were first crossed and their progeny were selected to obtain these genetic modifications along with an homozygous wild-type allele at NNT locus. Then, Myh6-Cre (+/−) EPAC2 (flox/wt) were crossed with Myh6-Cre (−/−) EPAC2 (flox/wt) and progeny was used for breeding to generate Myh6-Cre(+/−) EPAC2 (flox/flox) and Myh6-Cre (+/−) EPAC2 (wt/wt) which were used as control.

The following primers are used for genotype analysis. For Nnt locus, a set of three primers: common (5′-GTA​GGG​CCA​ACT​GTT​TCT​GCA​TGA-3′),

Wt (5′-GGG​CAT​AGG​AAG​CAA​ATA​CCA​AGT​TG-3’)

Mutant (5′- GTG​GAA​TTC​CGC​TGA​GAG​AAC​TCT​T-3′).

For Cre recombinase:

(5′-CCT​GGA​AAA​TGC​TTC​TGT​CCG-3′) and (5′-CAG​GGT​GTT​ATC​CGC​AAT​CCC-3′).

For EPAC2flox:

(5′-CCT​GTG​TGC​GTA​GGT​GTC​TTC​GTC-3‘) and (5'-GGC​ATT​TGT​AAC​CTT​CAA​AAC​ACA​TG-3')

Supplementary Figure S1 shows quantitative RT-PCR results.

Mice were anesthetized with isoflurane (3% induction and 1%–1.5% maintenance), placed on a heating pad and examined by echocardiography (Vevo2100, VisualSonics). M-mode in parasternal short axis view at the level of papillary muscle was used for wall thickness (LVAW, LVPW), cavity diameter (LVID) and fractional shortening (FS) measurements. 2D-mode long axis view was used for ejection fraction (EF). Images were analysed off-line with VevoLab software (VisulaSonics). Acquisition and analysis were performed blind.

Supplementary Table S1 shows that EPAC2^−/− are not significantly different from WT for heart size and cardiac function except for a small decrease in heart rate and left ventricle end-diastolic posterior wall thickness.

All solutions were prepared using ultrapure water supplied by a Milli-Q system (Millipore, USA). All of the reagents used in this study were purchased from Sigma Aldrich (France) unless otherwise stated. Liberase was purchased form Roche (France). The acetoxymethyl ester form of 8-CPT (8-CPT-AM) and the EPAC2 inhibitor were purchased from Biolog Life Science Institute (Germany). AM-001 was purchased from Abinter (France). To activate EPAC pathway, 8-CPT-AM was used at 10 μM. AM-001 was used at 20 μM and ESI-05 at 25 μM to selectively inhibit EPAC1 or EPAC2 isoform, respectively.

Atria from seven mice were dissected and immediately placed into All protect tissue solution (Qiagen, Germany). Atria were then stored at −80°C until RNA extraction. Total RNA was extracted from atria and purified using the QIAzol reagent (Qiagen) and the Qiagen RNeasy Kit (Qiagen). RNA quantity was then measured by spectrophotometry using the NanoDrop (Thermofisher, USA). 100 ng of RNA was reversed transcribed using cDNA Reverse Transcription kit (BIO-RAD). Primers were synthesized and obtained from BIO-RAD. RT-qPCR was performed in a 10 μL reaction volume composed by cDNA, SYBR Green mix (BIO-RAD, USA), ddH2O and primers on the BIO-RAD C100 Touch Thermal Cycler/CFX96 Real-time System. Expression levels for RAPGEF3 (EPAC1) and RAPGEF4 (EPAC2) were normalized using CFX Manager software (BIO-RAD) and housekeeping genes, GAPDH and HPRT1.

AF inducibility and probability were determined by transesophageal atrial stimulation (TS). Mice atria were stimulated using an octopolar catheter (WPI) introduced in the oesophagus under 2% isoflurane (Supplementary Figure S2A). Right atrial stimulation consisted of a burst pacing protocol of 40 stimuli at each pacing cycle with decreasing cycle length from 60 to 10 ms by 2 ms steps (Supplementary Figure S2B). Two forelimb electrodes were used to record ECG. Analysis was performed with Spike 2 software (Cambridge Electronic Design). AF was defined as an irregular and rapid rhythm lasting for more than 2 s and associated with loss of P wave (Verheule et al., 2004) (Supplementary Figure S2C). A mouse is considered as inducible when an AF episode was identified after burst pacing. AF probability (%) is defined by the ratio of the number of AF episodes divided by the number of burst pacing. The treatment of mice with the pharmacological activator and inhibitors of EPAC pathway or vehicle (0.1% dimethyl sulfoxide [DMSO]) was administered 30 min before the induction of AF via intraperitoneal injection.

Mice were anesthetized by inhalation of isoflurane (2% in air) and then humanly killed by cervical dislocation. Heart was quickly removed and atria were isolated in cardioplegic solution (in mmol/L: 110 NaCl, 1.2 CaCl₂, 16 KCl, 16 MgCl₂, 10 NaHCO₃, 9 Glucose). Then atria were pinned down on a Sylgar gel in the superfusion chamber before being perfused with Tyrode solution (in mmol/L: 130 NaCl, 24 NaHCO₃, 1.2 NaH₂PO₄, 1 MgCl₂, 5.6 Glucose, 4 KCl, 1.8 CaCl₂) gassed with 5% CO2 and 95% O2 (pH 7.4) at 37°C ± 0.5°C (Supplementary Figure S3A). Optical mapping experiments were performed using a high resolution CMOS camera (MICAM Ultima, SciMedia, USA) recording from a 100 x 100 field of view, at an 0.11 cm isotropic spatial resolution and a temporal resolution of 1 kHz. Right atria were paced with two unipolar electrodes at 2 ± 0.5 mA at frequencies ranging from 7 Hz to 12 Hz. Voltage-sensitive optical signals were obtained using di-4-ANEPPS (10 µM, VWR International) excited at 530 nm with LEDs (Cairn Research Ltd, Kent—United Kingdom). The fluorescence light was filtered with a band pass filter at 650 nm. Optical AP were analyzed using the PVwave software to quantify AP duration (APD) and calculate conduction velocities (CV). The effects of activation of the pathways for EPAC1 (8-CPT+ESI-05) and EPAC2 (8-CPT+AM-001) were evaluated by superfusion.

Atrial Myocytes were dissociated as previously described (Jansen and Rose, 2019). Briefly, mice were anesthetized by inhalation of isoflurane (2% in air) then heparinized by intraperitoneal injection of Heparin (200 UI). Mice anesthesia was checked by absence of the paw withdrawal reflex. Mice were subsequently killed by cervical dislocation and atrial appendages were rapidly excised. After the excision, all digestion steps were realized at 37°C. Atria were quickly washed and minced in modified Tyrode solution (in mmol/L: 140 NaCl, 5.4 KCl, 1.2 KH₂PO₄, 5 HEPES, 5.55 Glucose, 1 MgCl₂, 1.8 CaCl₂, 5 U/mL Heparin; pH 7.4 with NaOH) and transferred in a pre-digestion buffer solution (in mmol/L: 140 NaCl, 5.4 KCl, 1.2 KH₂PO₄, 5 HEPES, 18.5 Glucose, 50 Taurine, 0.066 CaCl₂, 1 mg/mL Bovine Serum Albumin; pH 6.9 with NaOH). After 5 min of pre-digestion, tissues were transferred in a digestion solution corresponding to the pre-digestion buffer supplemented by 0.11 mg/mL (equivalent to 0.34 Wünsch unit/mL and 36.7 units/mL Dispase) of Liberase (Medium Thermolysine, Roche, France). The digestion step lasted 20–23 min. After digestion was completed, atrial stripes were washed in a modified Kraft-Brühe solution (in mmol/L: 100 K-Glutamate, 10 K-Aspartate, 25 KCl, 10 KH₂PO₄, 2 MgSO4, 20 Taurine, 5 Creatine, 0.5 EGTA, 20 Glucose, 5 HEPES, 0,1% Bovine Serum Albumin; pH 7.2 with KOH), and mechanically triturated in this solution to allow cell isolation. Once the dissociation ended, cells were gradually reintroduced to 1 mmol/L calcium concentration by addition of calcium in the Kraft-Brühe solution (in mmol/L of free calcium: 0.125, 0.25, 0.375, 0.5, 0.625, 0.75, 0.875, and 1). Cells were used for patch clamp experiments during the 8 h following the dissociation. Only rod shaped and striated cells were used for experiments.

Isolated myocytes were studied in Petri dishes on the stage of an inverted microscope (Nikon TE200-S, Japan). AP were recorded at room temperature using the whole cell configuration of the patch-clamp technique in its current-clamp mode. For data acquisition, an Axopatch 200B (Molecular Devices, United State) amplifier connected to a Digidata 1322 A/D (Molecular Devices, United State) were used. Data were recorded and analyzed using pClamp software 9 (Molecular Devices, United State). Signals were digitized at a frequency of 10 KHz and filtered at 2 KHz using a 8-pole Bessel low pass filter. Patch pipettes resistance was usually comprised between 1.2 and 2.5 MΩ when filled with the intrapipette solution described below.

AP were elicited by 1 ms supra-threshold current steps at a frequency of 0.1 Hz. Bath solution was composed by (in mmol/L): 130 NaCl, 5.4 KCl, 1.4 MgCl₂, 0.4 NaH₂PO₄, 4.2 HEPES, 10 Glucose, 20 Taurine, 10 Creatine, 1 CaCl2; pH 7.4 with NaOH. Pipette solution was composed by (in mmol/L): 10 NaCl, 130 K-Glutamate, 9 KCl, 5 ATPMg, 0.5 MgCl₂, 10 HEPES, 0.4 GTP-Tris, 0.5 EGTA, 0.12 CaCl₂; pH 7.2 with KOH.

AP amplitude was measured as the difference between the peak of overshoot and the resting membrane potential. The maximum rate of rise of the AP (dV/dtmax) was calculated by differentiation of the AP upstroke using Clampfit software. Action potential duration (APD) was measured as the duration from the trigger of AP to 20%, 50% and 90% of repolarization (APD20, APD50 and APD90, respectively).

AP parameters under 8-CPT-AM superfusion (10 μmol/L) have been assessed at the steady state effect of the compound (∼5 min). To evaluate the EPAC1 selective inhibition by AM-001, cells were first treated by 8-CPT-AM (10 μmol/L) alone, then co-treated for at least 15 min by superfusion of both 8-CPT-AM (10 μmol/L) and AM-001 (20 μmol/L). The impact of the co-treatment has been evaluated at the steady state of the effect after this time lapse.

Data are reported as mean ± SEM. Statistical analysis was performed using GraphPad Prism 8 software. Data were tested for normal distribution using Shapiro-Wilk test. Fisher’s exact tests were used to evaluate AF inducibility. For optical mapping data, Student’s paired and unpaired t-test and non-parametric test (Mann-Whitney) were used to compare two populations, and ANOVA or Kruskal Wallis tests with appropriates post hoc tests (Tukey’s and Dunn’s test) were used to evaluate differences between more than two groups. For patch clamp data, two-way repeated measure ANOVA with Holm-Sidak post hoc test for normality distributed, otherwise Friedman’s ANOVA on ranks with Dunn’s multiple comparisons post hoc were used. p < 0.05 was considered statistically significant.

We first assessed the expression of EPAC1 and EPAC2 mRNA in mouse atria. Consistent with a previous study (Zhang et al., 2019), both isoforms are present (Supplementary Figure S4). Next, we determined the role of EPAC1 and EPAC2 in AF susceptibility in vivo using a pharmacological approach. Activation of EPAC1 and EPAC2 by injection of 8-CPT-AM slightly increased the proportion of animals showing AF episodes (Figure 1A). Indeed, AF was recorded only in 40% of WT mice (4/10) and after injection of vehicle (4/10) while it was detected in 64% of mice after 8-CPT injection (7/11). Inhibition of EPAC1 by intraperitoneal injection of AM-001 completely prevented AF initiation (0% vs. 37% in control). In contrast, blocking EPAC2 with ESI-05 did not change AF inducibility compared to control and vehicle (40% of mice in each group). Figure 1B showed that there is trend in an increase of AF probability after injection of 8-CPT injection (7.18% ± 4.15% vs. 7.88% ± 4.58% vs. 11.66% ± 4.17%, control vehicle, 8-CPT respectively). AF probability was completely suppressed following AM-001 injection (EPAC1 inhibition) whereas inhibition of EPAC2 with ESI-05 did not alter AF probability (7.18% ± 4.15% vs. 7.88% ± 4.58% vs. 6.79% ± 4.17%, control vehicle, ESI-05 respectively). Interestingly, we found that there was a tendency for longer AF episodes under EPAC1 and EPAC2 activation by 8-CPT-AM treatment (10.46 ± 6.65 vs. 16.45 ± 6.8 vs. 28.55 ± 5.95 s, control, vehicle, 8CPT respectively) and with EPAC2 inhibition (10.46 ± 6.65 vs. 16.45 ± 6.8 vs. 34.37 ± 5.38 s, control, vehicle, ESI-05 respectively) (Figure 1C). These data show that EPAC1 and not EPAC2 contribute to AF genesis.

To investigate further the role of EPAC1 in AF inducibility, we next used wild type (WT), EPAC1 KO and EPAC2 KO mice. Figure 2A shows that in EPAC1 KO mice occurrence of AF was slightly lower (3/12) than WT littermate (4/10). In contrast, in EPAC2 KO mice occurrence of AF was similar (4/10) than WT littermate (5/10). Furthermore, genetic deletion of EPAC1 decreased AF probability (5.34% ± 3.31% vs. 0.85% ± 0.45%) although it did not reach statistical significance (Figure 2B and see 4.5 Study limitations). No difference in AF probability was observed in EPAC2 KO mice compared to WT mice (7.18% ± 4.54% vs. 6.41% ± 3.49%). Lastly, there was tendency for a decrease in AF duration in EPAC1 KO mice (18.98 ± 8.27 s vs. 12.44 ± 1.21 s, WT and EPAC1 KO respectively) (Figure 2C).

Thus, it appears that EPAC1, but not EPAC2, is involved in atrial arrhythmia genesis. Subsequent experiments were then performed to characterize the role of EPAC1 and EPAC2 in atrial electrophysiology at tissue and cellular levels.

Next, we used high-resolution optical mapping to record optical AP in mice. Genetic deletion of EPAC1 significantly shortened APD20, APD50 and APD80 compared to WT mice (Figure 3A). Indeed, APD20 was decreased by 14% in EPAC1 KO mice (14.28 ± 0.35 m vs. 12.3 ± 0.30 m), APD50 was decreased by 16% (22.39 ± 0.39 vs. 18.86 ± 0.67 m), and APD80 by 10% (33.77 ± 0.43 vs. 30.30 ± 1.41 m). Conversely, genetic inhibition of EPAC2 significantly prolonged APDs (Figure 3B) since APD20, APD50, and APD80 were increased by 11% (14.28 ± 0.35 m vs. 15.86 ± 0.57 m), 10% (22.39 ± 0.39 vs. 24.60 ± 0.91 m), or 12% (33.77 ± 0.43 vs. 37.81 ± 1.34 m) in EPAC2 KO mice, respectively.

To further characterize the involvement of EPAC1 and EPAC2 in atrial electrophysiology, we used EPAC pharmacological probes approach. As reported in Figure 4A, 8-CPT-AM induced an increase of 27% of the APD20 (14.28 ± 0.35 m vs. 18.19 ± 0.72 m), an increase of 24% of the APD50 (22.39 ± 0.39 m vs. 27.86 ± 1.37 m), and of 26% of the APD80 (33.77 ± 0.43 vs. 42.58 ± 2.87 m). Next, we investigated which EPAC isoform was involved in this prolongation of the APD. Inhibition of EPAC1 with AM-001 did not change APD (Figure 4B). Adding 8-CPT-AM to activate EPAC2 in AM-001 treated mice induced significant prolongation by 37% of APD20 (13.95 ± 0.73 vs. 19.08 ± 1.54 m), by 32% of APD50 (21.40 ± 1.00 vs. 28.30 ± 2.06 m), and by 27% of APD80 (32.47 ± 0.97 vs. 41.40 ± 2.35 m). We did not observe any significant change in APD in ESI-05 treated mice whereas the use of 8-CPT-AM in the presence of ESI-05 to preferentially activate EPAC1 protein, significantly prolonged APD80 by 20% (32.83 ± 1.23 vs. 39.47 ± 2.17 m) (Figure 4C). Together, these results suggest that both EPAC1 and EPAC2 are involved in atrial electrophysiology modulation.

Optical mapping also facilitated study of AP propagation on across whole atrial tissue surface, and thus calculation of CV. Perfusion of 8-CPT-AM slowed CV by 27% (25.43 ± 3.7 vs. 18.43 ± 4.32 cm/s), and thus we observed an increase in total activation time (Figure 5A). On the contrary, blocking EPAC1 with AM-001 did not modify CV (32.67 ± 3.17 vs. 33.17 ± 3.7 cm/s, Figure 5B). However, AM-001 prevented the effect of 8-CPT-AM on CV, as no significant slowing was observed (32.67 ± 3.17 vs. 28.83 ± 3.80 cm/s). Of particular interest, we did not reveal any effect of EPAC2 pathway in CV since ESI-05 treatment did not significantly change this parameter (30.00 ± 2.24 vs. 29.88 ± 1.95 cm/s) (Figure 5C). However, the addition of 8-CPT-AM to activate EPAC1 in the presence of ESI-05 induced a significant slowing of CV by 30% (30.00 ± 2.24 vs. 21.00 ± 2.34 cm/s). Altogether, these data demonstrate that EPAC1 and not EPAC2 is involved in the modulation of CV.

To examine in more detail the role of EPAC1 in atrial electrophysiology, we next measured action potentials using the patch-clamp technique in atrial myocytes isolated from WT and EPAC1^−/− mice, before and after superfusion of the EPAC activator 8-CPT-AM (10 μmol/L). In WT atrial myocytes (n = 10), as in EPAC1−/− myocytes (n = 12), the main effect of EPAC activation was a significant lengthening of APD (Figure 6A) (+92.36 ± 25.22%, p < 0.001 and +39.05 ± 14.03%, p < 0.01 at 90% of repolarization vs. control, respectively, Figure 6B). However, this lengthening was significantly longer in WT than in EPAC1^−/− myocytes (p < 0.05). Interestingly, a similar lengthening profile was found at other APDs (at 20% and 50% of repolarization, Figure 6B). 8-CPT-AM had no other effects on AP parameters, except on AP amplitude (Supplementary Table S2). To evaluate whether this lengthening was linked to EPAC2 isoform, atrial myocytes from WT or EPAC1^−/− mice were concomitantly superfused with 8-CPT-AM and AM-001. Selective EPAC1 inhibition resulted in a partial but significant correction of the 8-CPT-AM dependent APD lengthening (+58.77 ± 17.49% vs. control at 90% of repolarization, p < 0.001 vs. control and p < 0.05 vs. 8-CPT-AM condition) in WT atrial myocytes (Figure 6B). In EPAC1^−/− atrial myocytes, co-treatment with 8-CPT-AM and AM-001 failed to influence the effect on APD compared to 8-CPT-AM alone (+45.76 ± 9.56% vs. control at 90% of repolarization, p < 0.001 vs. control and p > 0.1 vs. 8-CPT-AM condition). Interestingly, after EPAC1 selective inhibition with AM-001, 8-CPT-AM induced AP lengthening in WT atrial myocytes which was similar to EPAC1^−/− atrial myocytes (p = 0.85) (Figure 6B).

To determine whether EPAC1 played role in APD modulation at basal level, we superfused AM-001 (20 μmol/L) on WT or EPAC1^−/− atrial myocytes and recorded AP parameters (Figure 7A). Inhibition of EPAC1 in basal conditions had no effect on AP parameters in both WT and EPAC1^−/− atrial myocytes, except a time-dependent decrease of dV/dtmax (Supplementary Table S3). In addition, superfusion of 8-CPT-AM (10 μmol/L) after pretreatment with AM-001 resulted in APD prolongation both in WT and EPAC1^−/− atrial myocytes (p < 0.001, n = 9 for each background, Figure 7B), suggesting the involvement of EPAC2 in 8-CPT-AM dependent APD increase. Interestingly, after the pretreatment, 8-CPT-AM dependent APD prolongation was similar in both WT and EPAC1^−/− mice (+34.23 ± 27.87% and +31.23 ± 13.95% vs. control at 90% of repolarization, respectively, p = 0.73). Taken together, these results show that EPAC1 and EPAC2 isoforms are both involved in the effect of 8-CPT-AM in the APD prolongation in mouse atrial myocytes.

Prior work on EPAC signaling pathways in cardiac pathophysiology have focused on the ventricle. EPAC activates pro-hypertrophic pathways leading to cardiomyocyte hypertrophy (Morel et al., 2005). Other deleterious effects of EPAC are ventricular arrhythmias, first described in vitro, and confirmed in transgenic mice for EPAC1 and EPAC2 (Laudette et al., 2018). At the cellular level, EPAC regulates AP lengthening in rat ventricular cardiomyocytes (Brette et al., 2013), a process which is correlated with the genesis of arrhythmia by predisposing cardiomyocytes to early after depolarization and dispersion of repolarization (Nattel et al., 2007). In addition, EPAC activation increases Ca²⁺ spark frequency (Pereira et al., 2007). Chronic activation of EPAC influences the expression level of proarrhythmic channels including the slow delayed-rectifier potassium current (IKs) subunit potassium voltage-gated channel (KCN) and transient receptor potential canonical 3 and 4 channels (TRPC3/4) (Aflaki et al., 2014; Dominguez-Rodriguez et al., 2015). At the atrial level little is known. Acute EPAC activation inhibits the sodium current via ryanodine receptor type 2 activation. This decrease in Na⁺ current is associated with a reduction in maximum AP upstroke rate (Valli et al., 2018). In our study, we did not find any significant change in AP upstroke rate (Supplementary Table S2). The discrepancy can be due to the difference of genetic backgrounds of the mice and/or age of the animals. However, our data are consistent with recent studies showing an involvement of EPAC1 in the development of atrial and ventricular arrhythmias in mice (Prajapati et al., 2019; Zhang et al., 2019).

As described in ventricular cardiomyocytes, EPAC activation in atria lead to a lengthening of action potential at the tissue and cellular level. Our results demonstrated that both EPAC1 and EPAC2 isoforms are involved in the APD changes. In ventricular myocytes, APD lengthening is due to a decrease in sustained potassium current (Brette et al., 2013). Ion channels underlying AP are different in atria compared to ventricles (Hatem et al., 2010). The main difference is in potassium channels, albeit sustained potassium current is also present in atria. CV is an important parameter for atrial cardiac arrhythmia (Weiss et al., 2005). Our results from optical mapping experiments showed that CV is decreased upon EPAC activation. Ion channels regulating CV are connexins and sodium channels (Campbell et al., 2014). Given our patch-clamp data, it seems likely that the decrease in CV is due to changes in connexin properties. Further studies are needed to determine which ion channels are modulated by EPAC in atria, including connexin.

Our results showed that AF susceptibility was increased when EPAC was activated by 8-CPT. It seems likely that an AF episode is triggered by reduction in CV. To dissect which isoform was important in AF susceptibility, we used KO mice and our data showed that AF susceptibility was decreased in EPAC1 KO mice. These results are consistent with a study focusing on EPAC1-dependent phospholamban phosphorylation which also show that EPAC1 deficiency resulted in reduced susceptibility to AF (Okumura et al., 2014). In contrast, in EPAC2 KO mice AF susceptibility was similar to WT animals. The implication of EPAC1, and not EPAC2 is confirmed by our pharmacological approach, where AM-001, a specific EPAC1 inhibitor prevented AF. These results are consistent with a study showing that acute effects of CE3F4 treatment, an EPAC1 selective inhibitor (Courilleau et al., 2012; Courilleau et al., 2013), shorten the duration of AF in mice (Prajapati et al., 2019).

Until recently, the only EPAC1 selective inhibitor identified was CE3F4, a tetrahydroquinoline analogue, but its low bioavailability prevented its biological action in vivo, thereby excluding its use in large animals (Courilleau et al., 2012; Courilleau et al., 2013; Laudette et al., 2019). Another non-cyclic nucleotide small molecule designed as ESI-09 displays a favorable pharmacological/toxicological profile when administrated in vivo to mice but this compound does not discriminate between EPAC1 and EPAC2 isoforms (Almahariq et al., 2013). In the current study, we used AM-001, an EPAC1 non-competitive inhibitor, with in vivo application (Laudette et al., 2019). Our data demonstrate that AM-001 prevents AF in mice. Further studies are needed to confirm this finding in other mammal species.

There are several limitations to our study: First, mouse as an animal model to study AF is still controversial (Fu et al., 2022). However, numerous studies used mice to investigate the mechanisms involved in the initiation of AF, as reviewed in (Riley et al., 2012; Clauss et al., 2019; Blackwell et al., 2022). In addition, mouse models of AF may be useful to uncover novel mechanisms of AF. This is well illustrated with the studies demonstrating that abnormal ryanodine receptor Ca²⁺ release via CaMKII overactivation is directly responsible for the progression of paroxysmal AF to more persistent forms (Li et al., 2014; Mesubi and Anderson, 2016; Landstrom et al., 2017). A second limitation of our study concerns the data generated from transesophageal stimulation experiments. In the majority of results, we observed a trend which did not reach statistically significance. This was probably due to the large variability in this approach. Indeed, even if transesophageal stimulation protocol for the induction of AF in mice was carefully evaluated more than 20 years ago (Schrickel et al., 2002), data showed that AF inducibility in control mice could range from 0% to more than 70% indicating that optimization of this technique is still required (Murphy et al., 2022). Power analysis from our experiments indicates than more 30 mice in each condition should have been used, but this conflicts with our 3R ambitions. Another limitation would be the off-target effects of the drugs used but the specificity of 8-CPT-AM towards EPAC had been previously demonstrated in double KO mice for EPAC1 and EPAC2 (Pereira et al., 2013). Finally, the mechanistic link between EPAC activation, APD prolongation and the decreased CV in atria is unknown and requires extensive work in subsequent studies.