C57BL/6N mice (10–12 weeks of age, no gender restrictions) were provided by the Experimental Animal Center of Southwest Medical University and maintained in a specific pathogen-free facility with ad libitum access to food and water. All procedures were approved by the Institutional Animal Ethics Committee at Southwest Medical University, Luzhou, China (No: 20221108-010), which conforms to the Animal Research: Reporting of In Vivo Experiments (ARRIVE) guidelines.

CIT (Cat. No: 59729-32-7, MedChemExpress, New Jersey, United States) was administered intraperitoneally as a solution at a high therapeutic concentration of 40 mg/mL (dissolved in PBS), at a dosage of 9.6 mg/kg/day for 4 weeks. The dosage and treatment protocol were based on a published report and were confirmed to have a plasma concentration of over 40 μg/L (Honig et al., 2009).

Mice were first anesthetized with 3%–5% isoflurane, maintained on 1%–1.5% isoflurane, and then attached with four leads for baseline ECG monitoring (M150, Biopac, Goleta, California, United States) (heart rate was maintained at 400–500 bpm). ECG recordings were examined at a stable state for over 5 min. The QT interval, PR interval, and RR interval were measured and analyzed. Abnormal cardiac events, such as premature ventricular beats (PVCs), were also recorded. For echocardiogram measurement, mice were anesthetized with 1%–1.5% isoflurane, and then parasternal short-axis section (PSAX) B mode and M mode (Vevo 3100, FUJIFLIM VisualSonics Inc., Canada) were used to evaluate potential changes in hemodynamic and structural remodeling. Primary indicators of cardiac function, such as ejection fraction (EF), fractional shortening (FS), end-systole (s) and end-diastole (d) of the intraventricular septal thickness (IVSs and IVSd), end-systole and end-diastole of left ventricular posterior wall thickness (LVPWs and LVPWd), and end-diastolic and end-systolic left ventricular dimensions (LVIDs and LVIDd), were measured and quantified.

The detailed procedures of optical mapping have been thoroughly described previously (Li et al., 2023b). In brief, hearts were quickly extracted and mounted on the Langendorff perfusion system. Excitation–contraction uncoupler and optical dyes were perfused in order. Programmed electrical pacing was then performed to induce cardiac alternans and arrhythmias. After acquiring optical signals, basic electrophysiological parameters such as action potential duration (APD), calcium transient duration (CaTD) and their spatial heterogeneity, conduction velocity (CV), and cardiac alternans with reentrant circuits were quantified and analyzed.

When the heart reached a stable state, it was first paced at a sequence of 100 ms, and then the pacing cycle length was decreased by 10 ms until a pacing cycle length of 50 ms was reached. After testing frequency-dependent alternans and restitution properties, consecutive 50 Hz burst pacing was conducted to induce ventricular arrhythmias (Supplementary Figure S1). This procedure was repeated after ISO (final working concentration 1 µM) intervention.

The detailed protocol refers to our previous publication (Li et al., 2023b). After intraperitoneal heparin treatment and anesthesia with 3% isoflurane, the hearts were quickly isolated and perfused with Krebs solution in NaCl 119, NaHCO₃ 25, NaH₂PO₄ 1.0, KCl 4.7, MgCl₂ 1.05, CaCl₂ 1.8, and glucose 10 (mM); the pH was adjusted to 7.35–7.45 with NaOH, and the baths were oxygenated with 95% O₂ and 5% CO₂ to maintain the stability of pH. After the Langendorff-perfused hearts reached a stable state, as confirmed by real-time ECG monitoring, 50 µL of blebbistatin stock solution was diluted in 50 mL of circulating Krebs solution (final concentration 10 µM) to minimize contraction artifacts. Then, 15 µL stock solution of Rhod2-AM (final concentration 0.356 mM, 1 mg/mL, stock solution, Cat: ab142780, Abcam, Cambridge, United Kingdom) and 15 µL Pluronic F127 (20% w/v in DMSO, Cat: P3000MP, Invitrogen, California, United States) were diluted in 20 mL Krebs solution and circulated in a total volume of 50 mL for 15 min; next, 10 µL of the stock solution of RH237 (1 mg/mL, final concentration 0.603 mM; stock solution Cat: S1109, Thermo Fisher Scientific, Waltham, Massachusetts, United States) was perfused to obtain simultaneous membrane potential (Vm) and intracellular Ca²⁺ recordings. Once dye loading was accomplished, a custom-designed optical mapping system, including an electron-multiplying charge-coupled device (EMCCD) camera (Evolve 512, Photometrics, Tucson, Arizona, United States) and two light-emitting diodes (LEDs) of 530 nm, were used for excitation of the Ca²⁺ transient and transmembrane voltage probes. CaT fluorescence was acquired using a 585-/40-nm bandpass filter, while Vm fluorescence was collected using a 662-nm long-pass filter. The high-speed EMCCD camera enabled high temporal resolution with a sampling rate of up to 1,000 frames/s and a minimum spatial resolution at the sample of 32 × 32 µm per pixel. The programmed stimulation and ECG recordings were simultaneously controlled using Spik2 software (Cambridge Electronic Design limited, Cambridge, United Kingdom). For optical mapping signal analysis, data were semi-automatically processed using ElectroMap (University of Birmingham, Birmingham, United Kingdom) and OmapScope5 software (MappingLab, Manchester, United Kingdom).

The transcriptomics array was performed at Novogene Co. Ltd. (Beijing, China), in accordance with the manufacturer’s protocols. Total RNA was obtained, and its integrity was evaluated using the RNA Nano 6000 Assay Kit of the Bio-analyzer 2100 system (Agilent Technologies, Clara, California, United States). Subsequently, first-strand cDNA was synthesized, and a library was constructed for transcriptome sequencing. The index-coded samples underwent clustering, and the library preparations were sequenced on an Illumina NovaSeq platform, generating 150 base-pair paired-end reads. The resulting data were subjected to Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses.

Total RNA was extracted from cardiac tissue using the RNA Easy Fast Tissue/Cell Kit (cat. DP451; Tiangen Biotech, Shanghai, China). Following this, the total RNA was subjected to reverse transcription to synthesize cDNA using ReverTra® Ace qPCR RT Master Mix (cat. FSQ-201; Toyobo Life Science, Osaka, Japan). qPCR was conducted using the Taq Pro Universal SYBR qPCR Master Mix (Vazyme Biotech, Nanjing, China) under the following thermal cycling conditions: 95°C for 10 s, 60°C for 25 s, and 72°C for 30 s for 40 cycles. The mRNA expression levels of the target genes were normalized to GAPDH using the 2^−ΔΔCt method (Livak and Schmittgen, 2001). Primers used in this study are displayed in Table 1.

Mice were heparinized and anesthetized first; the hearts were then quickly excised and placed into precooled Ca²⁺-free Tyrode’s solution. After cannulation of the aorta, the heart was mounted on a Langendorff apparatus and perfused with Ca²⁺-free Tyrode’s solution [NaCl 137, KCl 5.4, MgCl₂ 1.0, NaH₂PO₄ 0.33, D-glucose 10, and HEPES 10, (mM); pH 7.4] at 37°C for approximately 2 min. Then, the hearts were perfused with the same solution containing collagenase (Type II, Worthington, 450 U/mL), BSA (BBI, 1 mg/mL), protease (Type XIV, Sigma, 0.21 U/mL), and Ca²⁺ (50 μM) for 8–10 min. All solutions were equilibrated with 5% CO₂/95% O₂ and maintained at 37°C. When the tissues were well digested, the atria were removed, and ventricles were cut into small pieces and gently triturated to release single myocytes in B buffer (Ca²⁺-free Tyrode’s solution with 5 mg/mL BSA and 50 μM Ca²⁺). After filtering and centrifugation at 100 g, cells were stored in KB solution [L-glutamic acid 120, KOH 120, KCl 10, KH₂PO₄ 10, MgSO₄ 1.8, EGTA 0.5, D-glucose 20, taurine 10, and HEPES 10, (mM); bovine serum albumin 0.2%; pH 7.2] at 4°C until Ca²⁺ (1.8 mM)-tolerant and ready for use. The obtained ventricular myocytes can be used simultaneously for patch clamp and intracellular calcium measurements.

Action potentials (APs) were recorded using the whole-cell patch-clamp technique under the current clamp mode (PatchMaster, EPC-10 amplifier, HEKA, Harvard Bioscience Inc, Reutlingen, Germany). The bath solution contained the following: NaCl 138, KCl 4, MgCl₂ 1.0, NaH₂PO₄ 0.33, D-glucose 10, HEPES 10, and CaCl₂ 2 (pH adjusted to 7.4 with NaOH) (mM). The pipette was filled with a solution of the following composition (mM): K-aspartate 120, KCl 20, MgCl2 1, HEPES 5, EGTA 5, and Mg-ATP 2 (pH adjusted to 7.2 with KOH). Action potentials were evoked by current pulses of 800 pA (4 ms in duration) at a frequency of 1 Hz. The time course of the action potentials was analyzed using pClamp software (Clampfit 11.3, Axon, Molecular Devices, Sunnyvale, California, United States). The action potential configuration was analyzed by measuring APD from the peak of the action potential to 10%, 30%, 50%, 70%, and 90% of repolarization.

Transient outward potassium current (I_to) was recorded using the whole-cell patch-clamp configuration. Cells were superfused with a bath solution of the following composition (in mM): NaCl 146, KCl 4, MgCl₂ 2.5, HEPES 5, dextrose 5.5, and CaCl₂ 2 (pH adjusted to 7.3 with NaOH). The pipette was filled with a solution of the following composition (mM): KCl 135, EGTA 10, HEPES 10, dextrose 5.5, and Mg-ATP 3 (pH adjusted to 7.2 with KOH). Then, 20 μM nifedipine and 20 μM TTX were added to the bath solution to block the activity of I_Ca and I_Na, respectively. After membrane rupture, the cell capacitance was measured, and the series resistance (Rs) was compensated electronically by 70%. I_to was evoked by 500-ms depolarizing pulses to potentials between −60 and +60 mV (10 mV steps) from a holding potential of −70 mV. The peak current density is calculated as I_to. The amplitude and kinetics of I_to were analyzed using Clampfit 11.3 software. The L-type calcium current (I_Ca-L) was recorded by using the whole-cell patch-clamp configuration. Ventricular myocytes were superfused with a bath solution of the following composition (in mM): NaCl 140, CaCl₂ 1.8, CsCl 10, MgCl₂ 1.0, glucose 10, and HEPES 10 (pH was adjusted to 7.4 with NaOH). The pipette was filled with a solution of the following composition (mM): CsCl 130, TEACl 20, HEPES 5.0, Na-GTP 0.1, EGTA 5.0, and Mg-ATP 5.0 (pH was adjusted to 7.2 with CsOH) (Zhang et al., 2024). The statistical analysis was carried out using GraphPad Prism 8.0.2. For AP, I_to, and I_Ca-L, patch electrode resistance was 3–5 MΩ. The recordings were conducted at room temperature (24 °C–26 °C).

Total protein was extracted from ventricular tissues, with the entire procedure conducted on ice. The collected tissues were lysed with RIPA lysis buffer for 20 min. Lysates were centrifuged at 12,000 rpm at 4 °C for 15 min. The supernatant was stored at −80 °C after determining its protein concentration using Bradford assay. A measure of 20 μg protein for each lane was separated using 10% SDS-PAGE and transferred to a PVDF membrane (Millipore, United States). The membrane was incubated in TBST containing 5% non-fat milk for 2 h at room temperature to block nonspecific binding and was incubated with the primary antibody (Kv 4.2, 1:1,000, ab192762, Abcam, United States) overnight at 4 °C. The membrane was incubated with the horseradish peroxidase (HRP)-conjugated goat anti-rabbit or mouse IgG (1:3,000, BBI, China). The secondary antibody was incubated for 1 h at room temperature. The membrane was incubated in chemiluminescent HRP substrate (Millipore, United States) at room temperature for 30 s and then imaged using the Universal Hood II System (Bio-Rad, United States) (Eslami and Lujan, 2010).

[Ca²⁺]_i was quantified using Fluo-4 AM (Molecular Probes, Thermo Fisher Scientific, United States). In brief, ventricular myocytes were placed in tubes containing Fluo-4 AM (5 μmol/L reconstituted in DMSO) along with 0.1% Pluronic F-127 (P2443, Thermo Fisher Scientific, United States) for 30 min in the dark and then washed for 20 min in Tyrode’s solution. The ventricular myocytes were then placed on coverslips and perfused with Tyrode’s solution (1.8 mM Ca²⁺) at room temperature for 10 min. Images were obtained by the line scan mode using a confocal microscope (LSM980, Carl Zeiss, German). The ventricular myocytes were field-stimulated at 0.5 Hz. A wavelength of 488 nm was applied to excite the samples and collect fluorescence emission spectra at 505 nm (40 nm band-pass filter) for single-color calcium imaging. To standardize the fluorescence intensity of intracellular Ca²⁺ indicators, the change in [Ca²⁺]_i (∆[Ca²⁺]_i) was expressed as ΔF/F0 = (F − F0)/F0. Here, F0 is a measure of the baseline fluorescence at the resting state, F is the fluorescence intensity at any given time, and ∆F is the moment-by-moment deviation from that baseline (Li et al., 2023a).

All data are expressed as mean ± standard deviation. Differences between group means were examined using two-tailed, unpaired Student’s t-test or two-way analysis of variance (ANOVA) with Sidak’s test for multiple comparisons. P < 0.05 was considered statistically significant.

As shown in Figure 1A, Echo and Masson staining showed that CIT did not significantly induce hemodynamic or structural remodeling of the hearts (Figures 1B,C; Table 2). Surface ECG recordings in vivo revealed that mice undergoing CIT treatment presented significant QT interval prolongation (day 29 SAL: 48.49 ± 1.94 ms vs. day 29 CIT: 51.31 ± 2.92 ms, P = 0.0138; day 0 CIT: 47.45 ± 2.98 ms C vs. day 29 CIT: 51.31 ± 2.92 ms, P = 0.0011) (Figures 1F,G) compared with those treated with SAL, while the PR interval and RR interval were not significantly altered (Figures 1D,E). In addition to the effects on QT interval prolongation, CIT-treated mice also exhibited multiple abnormalities in ECG recordings and higher incidence of spontaneous premature ventricular beats (PVCs), as shown in Figures 1H,I. All mice displayed a normal sinus rhythm at the baseline; however, following CIT intervention, electrical abnormalities such as PVCs, T wave alternans (TWA), and atrioventricular conduction abnormalities were recorded, which indicated the proarrhythmic effect of CIT. Then, the occurrence and number count of PVCs were quantified in both SAL and CIT groups during 15 min. Five of eleven mice showed spontaneous PVCs in the CIT group (day 0 CIT: 0/11 vs. day 29: 5/11, P = 0.0351), with an average of approximately 50 PVCs for each mouse (an average of 100 PVCs for the five mice with electrical abnormalities), while no PVCs were observed in the SAL group (Figure 1J).

Consistent with the in vivo surface ECG results, both the QT interval and the QTc interval were significantly prolonged in isolated hearts from mice treated with CIT in vitro, while the PR and RR intervals were not affected (Figures 2A,B). Under ISO challenge, the PR interval remained unchanged, while the RR interval was significantly reduced (SAL: 138.9 ± 8.98 ms vs. SAL + ISO: 124.2 ± 6.23 ms, P = 0.0366; CIT: 140.5 ± 12.15 ms vs. CIT + ISO: 126.5 ± 12.17 ms, P = 0.0490; n = 8 for each group). Meanwhile, the QT interval and QTc interval were also reduced by ISO treatment (QT interval: SAL: 51.66 ± 3.30 ms vs. SAL + ISO: 45.19 ± 2.07 ms, P = 0.0007; CIT: 56.21 ± 4.04 ms vs. CIT + ISO: 49.40 ± 1.63 ms, P = 0.0004; n = 8 for each group. QTc interval: SAL: 138.7 ± 8.35 ms vs. SAL + ISO: 125.8 ± 4.75 ms, P = 0.0176; CIT: 150.2 ± 10.93 ms vs. CIT + ISO: 138.8 ± 7.15 ms, P = 0.0417; n = 8 for each group). Moreover, hearts isolated from mice treated with CIT showed an increased incidence of ventricular arrhythmias, manifested as PVCs and VT (Figures 2E–G, SAL 0/8 vs. CIT 5/8, P = 0.0256). Representative ECG recordings are shown in Figure 2E; sustained polymorphic ventricular tachycardias (PVTs) were induced in CIT-treated hearts with significantly prolonged QT intervals (QT interval = 60.1 ms), while the control hearts presented normal QT intervals (QT interval = 49.3 ms) and maintained sinus rhythm before and after ISO intervention. Then, these abnormal cardiac events were quantified as shown in Figures 2F,H; CIT-treated hearts presented higher PVC counts (SAL 52.75 ± 29.05 vs. CIT 275.4 ± 256.4, P = 0.0286; n = 8 for each group) and longer VT duration (SAL 0 vs. CIT 101.1 ± 129.7, P = 0.0447; n = 8 for each group). These data suggested that CIT treatment would induce prolonged QT intervals and higher vulnerability of ventricular arrhythmias in the absence of cardiac structural remodeling.

Proarrhythmic electrophysiological mechanisms of CIT were explored using cardiac optical mapping (Figures 3–6). Representative action potential duration at 80% repolarization (APD₈₀) maps and corresponding optical traces before and after ISO application are illustrated in Figure 3A. The APD in hearts treated with CIT was significantly prolonged before and after ISO challenge compared with the control hearts (APD₈₀ baseline: 30.76 ± 2.78 ms vs. CIT: 37.44 ± 4.88 ms, P < 0.0001; APD₈₀ isoproterenol: 27.12 ± 1.85 ms vs. CIT: 34.51 ± 3.80 ms, P < 0.0001; n = 8 for each group) (Figures 3B,C). APD dispersion was increased in CIT-treated hearts. ISO reduced APD dispersion in SAL-treated hearts, but not in CIT-treated hearts (baseline: 0.490 ± 0.064 vs. CIT: 0.591 ± 0.084, P = 0.037; isoproterenol: 0.399 ± 0.051 vs. CIT: 0.532 ± 0.083, P = 0.0042; n = 8 for each group) (Figure 3D). Since studies have revealed that Nav 1.5 was influenced in CIT-related cardiac toxicity (Nakatani and Amano, 2021), the conduction of action potential was then assessed. ISO was effective in increasing the conduction velocity (CV) in both groups, while no difference between SAL- and CIT-treated hearts was found (Figures 3E–G).

Meanwhile, intracellular calcium-transient signals simultaneously recorded with membrane voltage (V_m) were investigated. The calcium transient duration at 80% repolarization (CaTD₈₀) maps and corresponding traces under programmed 10 Hz S1S1 pacing are shown in Figure 4A. The results suggested that CIT treatment significantly prolonged CaTD₈₀ (Figures 4B,C) without altering CaTD₈₀ heterogeneity (Figure 4D) when compared with SAL treatment (CaTD₈₀ baseline: 45.94 ± 1.09 ms vs. CIT: 50.92 ± 2.40 ms, P < 0.0001; ISO: 36.67 ± 1.35 ms vs. CIT + ISO: 42.42 ± 1.39 ms, P < 0.0001; n = 8 for each group). ISO decreased the time constant (tau) of decay in both groups, but no difference was observed in the tau of decay and time to peak (TTP) between SAL- and CIT-treated hearts (Figures 4E,F). We further investigated the physiological delay between AP and CaT (voltage–calcium latency) at three time points: the upstroke midpoint latency, the peak latency, and the decay latency, as shown in Figure 4H. Representative AP and CaT coupling traces at baseline and under ISO challenge are shown in Figure 4G; the peak and upstroke midpoint latencies (Figures 4I,J) were not altered with CIT treatment. However, the decay latency of CIT-treated hearts was significantly shortened compared with SAL hearts (Figure 4K), suggesting an alteration or turbulence of the relationship between AP and CaT (baseline: 23.69 ± 2.805 ms vs. CIT: 20.16 ± 2.310 ms, P = 0.045; ISO: 24.16 ± 3.411 ms vs. CIT + ISO: 20.59 ± 1.891 ms, P = 0.041; n = 8 for each group). The different degree of APD and CaTD prolongation resulted in impaired Vm–Ca coupling that may interrupt the physiological orchestrated relationship of membrane currents and intracellular calcium handling.

Since AP and CaT alternans are widely believed to be predictors of lethal arrhythmias (Kanaporis and Blatter, 2023), programmed S1S1 pacing from a cycle length of 100 to 50 ms was conducted to induce cardiac alternans, according to previous studies (Fukaya et al., 2019; Liao et al., 2021). Figure 5A presents representative APD alternans traces at 80–60 ms at baseline and after ISO challenge. CIT-treated hearts showed frequency-dependent APD alternans. Meanwhile, accompanied by APD alternans, CIT-treated hearts also showed frequency dependence and suffered larger and more severe CaT alternans in a broader region among all the pacing frequencies (Figure 5B). The CaT alternans algorithm is illustrated in Figure 5C. CaT_Large (L) and CaT_Small (S) are the amplitudes of the large and small CaTs of a pair of alternating CaTs, respectively. The CaT alternans ratio = 1 − S/L. By this definition, the CaT alternans ratio values fall between 0 and 1, where the CaT alternans ratio = 0 indicates no Ca²⁺ alternans and CaT alternans ratio = 1 indicates a situation where SR Ca²⁺ release is completely abolished on every other beat. CaTs were considered alternating when the beat-to-beat difference in CaT amplitude exceeded 10% (CaT alternans ratio > 0.1) (Kanaporis and Blatter, 2023). Quantification of CaT alternans showed that CIT-treated hearts possessed a higher CaT alternans ratio than the SAL-treated hearts at 80–60 ms (Figures 5D,E). Another CaT alternans algorithm quantified the duration of CaT(n)/CaT(n + 1), and the ratio reflects the degree of calcium alternation well as the ratio of 1 − S/L (Supplementary Figure S2). APD alternans accompanied by CaT alternans were deteriorated in CIT-treated hearts, which promote CIT-related arrhythmias.

Since CIT-treated hearts exhibited significant frequency-dependent AP and CaT alternans, we investigated the relationship between alternans and arrhythmias in CIT-treated hearts, especially CaT alternans. Representative ECG and corresponding AP and CaT traces under burst pacing in a SAL-treated heart are shown in Figure 6A; the normal sinus rhythm was maintained at baseline and after burst pacing without AP or CaT alternans. Phase maps indicated that AP conduction from the pacing region to the bottom was even and synchronous. However, as shown in Figure 6B, the CIT-treated heart exhibited spontaneous and sustained VT after two sequences of 50 Hz burst pacing. In further investigation of corresponding AP and CaT signals, severe CaT alternans with intermittent AP alternans were observed during fast pacing. Based on previous research studies on CaT alternans and arrhythmias, spatial discordant alternans (SDA) is believed to be firmly associated with arrhythmias (Aras et al., 2022; Chien et al., 2023); therefore, the spatial heterogeneity of CaT alternans with subsequent reentry activities were explored. Figure 6C shows six regions exhibiting large (regions 1, 2, and 3) and small (regions 4, 5, and 6) CaT alternans. By the combination analysis of phase maps and spatial properties of CaT alternans, the phase singularities (PSs) were found to be rotating and meandering in regions 1, 2, and 3 (Figure 6D), even in the absence of SDA. Regions with the most severe alternans overlapped with the PS trajectory in the reentrant circuits. These findings, to some extent, also indicated that the “spatial discordance” of CaT alternans was a potential electrophysiological substrate of CIT-related tachyarrhythmias. The dynamic process of VT initiation and maintenance of VT is shown in Figure 6D. In beat 1 of the upper panel, the ectopy was generated from the apex of the heart first, and the ectopic action potentials spread to the bottom rapidly. However, with impaired function caused by serious CaT alternans, the electrical pulses were terminated at the circle region border and shifted to the right side to escape from these regions with higher refractoriness, thus causing a wave-break and the formation of the phase singularity (Handa et al., 2020), which led to reentrant activities. The reentry circuits meandered around the PS and did not stop until the PS shifted and generated another ectopy in beat 11 shown in Figure 6D. A phase map (Figure 6D beat 11–20) showed the identical process of the maintenance of reentry activity; the trajectory of the PS existed in the dotted region for much of the duration in the subsequent VT episodes. Regional differences in CaT alternans initiated ectopic beats and led to reentry activities to maintain the sustained VT.

To further investigate the underlying mechanisms of CIT-induced arrhythmias, the transcriptomics array was carried out using cardiac ventricular tissue from mice undergoing different treatments (Figure 7). A heat map of gene clusters in each group is displayed in Figure 7A. The results revealed that CIT treatment induced alterations in gene clusters. KEGG and GO enrichment analyses of differentially expressed genes (DEGs; P < 0.05 and Log₂ fold of change > 1) demonstrated that DEGs were enriched in pathways including the calcium signaling pathway, cellular potassium ion transport, potassium channel complex, and regulation of ion transmembrane transport, which may contribute to the effects of CIT on cardiac electrophysiology and arrhythmias (Figures 7B,C; Supplementary Table S1).

Further analysis on the expression of main ion channels revealed that expressions of multiple potassium channels were downregulated (Figure 7D). Subsequently, the expressions of genes involved in K⁺ and Ca²⁺ channels were confirmed using RT-qPCR (Figure 7E). These data revealed that KCND2, KCNH2, KCNIP2, and KCNE1 were remarkably reduced following treatment with CIT (P < 0.05), while the expressions of CACNA1C and KCND3 were not significantly altered.

Action potentials were further measured using the patch-clamp technique at the single-cell level for deeper mechanism searching. Examples of APs recorded in isolated cells from CIT and SAL ventricles (at 1 Hz) are shown in Figure 8A. APD₃₀ of CIT-treated cardiomyocytes was significantly prolonged compared with SAL-treated cardiomyocytes, (6.481 ± 5.183 ms vs. 2.941 ± 1.241 ms, P < 0.05, Figure 8B). APD₉₀ of CIT-treated cardiomyocytes was also significantly prolonged compared with SAL-treated cardiomyocytes (P < 0.05, Figure 8C). In agreement with APD prolongation, the mean I _to density was significantly reduced in CIT-treated cardiomyocytes, and the I _to stimulation protocol and representative traces are shown in Figures 8D,E. Changes in the rates of activation and inactivation of I _to were quantified. CIT treatment caused a positive shift in the voltage-dependent activation and a negative shift in inactivation (Figure 8F). The half-maximal voltage of activation (V_{0.5, act}) estimated by fitting the activation curve with a Boltzmann function was shifted from 28.42 ± 2.336 to 17.00 ± 3.040 mV in cells treated with CIT (P < 0.05, Figure 8G). The voltages at half-maximum inactivation (V_{0.5, inact}) estimated by fitting the inactivation curve with a Boltzmann function were shifted from −35.43 ± 1.330 to −27.92 ± 1.814 mV in cells treated with CIT (P < 0.01, Figure 8H), illustrating that CIT treatment changed I _to gating kinetics and regulated the voltage dependence of I _to activation and inactivation, which led to a prolongation of APD in ventricular cardiomyocytes. To further clarify the protein expression after CIT treatment, we then conducted Western blot and found that the Kv4.2 expression level was significantly reduced (P < 0.05, Figures 8I,J), which was consistent with the decreased mRNA expression of KCND2.

The aforementioned data suggest that the abnormal Ca²⁺ handling is the effect of CIT on the prolongation of CaTD, which prompted us to further investigate whether CIT can directly affect I_Ca-L and intracellular calcium. The results showed that CIT reduced I_Ca-L current density (Figure 9A). In particular, CIT reduced the peak current density of I_Ca-L (at 0 mV) from −4.73 ± 0.32 pA/pF to −3.46 ± 0.31 pA/pF (P = 0.013) and (at 10 mV) from −4.48 ± 0.40 pA/pF to −2.80 ± 0.28 pA/pF (P = 0.002, Figure 9B). Next, Δ[Ca²⁺]_i was monitored using the Fluo-4 AM indicator. The results showed that ∆[Ca²⁺]i was significantly reduced in CIT-treated cardiomyocytes, with ΔF/F0 from 2.31 ± 0.20 to 1.1 ± 0.16 (Figures 9C,D).

Previous studies have found that citalopram and escitalopram cause prolongation of the QT interval on ECG and pathologic QTc prolongation in some patients (Akturk et al., 2023; Faraj et al., 2023; Chen et al., 2024). At the same time, CIT carries the risk of TdP (Krøll et al., 2022). The cardiac side effects of QT interval prolongation and higher risk of TdP by citalopram were reproduced and validated in our current work. Subsequent optical mapping studies suggested that APD was significantly prolonged in CIT-treated hearts, which is consistent with results of other studies on drug-induced long QT syndrome. Generally speaking, prolonged APD provides the electrical substrate for generating early after depolarization (EAD) or delayed after depolarization (DAD), which may deteriorate into life-threatening tachyarrhythmias (Bersell et al., 2023; Chien et al., 2023). These results, in some aspects, demonstrate the proarrhythmic potential of CIT treatment. Increased APD dispersion was observed in CIT-treated hearts, which may cause discordant repolarization and result in lethal cardiac events (Khwaounjoo et al., 2022; Yamazaki et al., 2022). CaTD was also found to be prolonged before and after ISO challenge. Despite the regional heterogeneity in APD between CIT-treated hearts and controls, we did not find spatial differences in CaTD, similar to a recent study (Wang et al., 2020). Accumulating evidence over the past few decades has revealed that normal cardiac function resulted from sophisticated interactions between ion channels on the membrane and intracellular calcium-handling proteins; disturbance of these interactions may lead to severe arrhythmias (Lei et al., 2018; Lei and Huang, 2020). As shown in Figure 4G, APD and CaTD of CIT-treated hearts were prolonged to different extents compared with SAL-treated hearts, which, in turn, resulted in shortened voltage–calcium decay latency. The shortened Vm–Ca²⁺ decay latency did not imply a more inseparable but aggravated coupling relationship because the physiological decay was not completed during the diastolic interval (Zhou et al., 2016). Therefore, the impaired voltage–calcium coupling may lead to discordant interactions of membrane ion channels and intracellular calcium-handling proteins, which may evolve into arrhythmias.

Prolonged APD and CaTD may lead to changed refractoriness in generating abnormal depolarization or intracellular calcium release, which are responsible for AP and CaT alternans (Liu et al., 2018; Qu and Weiss, 2023). Previous studies have indicated that SDA is inclined to generate lethal tachyarrhythmias (Cacheux et al., 2019; Liao et al., 2021). No SDA was found in all CIT-treated hearts nor in SAL hearts. However, after quantification of the CaT alternans ratio, we observed that hearts with the most severe arrhythmias would first express serious CaT alternans in a specific region, and the regions with electrical remodeling may provide the substrate for the initiation of arrhythmias under increased stress. Prolonged APD in spatially discordant regions caused increased APD dispersion; the AP repolarization in regions with short APD would finish at first and conduct to nearby regions, while the conduction would be blocked or deviated because of the not fully depolarized regions with prolonged APD, thus causing incomplete depolarization and formation of AP alternans. The AP alternans correlated well with CaT alternans. Moreover, being triggered by action potentials, the release of sarcoplasmic reticulum was well coordinated with APs, while the impaired Vm–Ca²⁺ coupling may interrupt the correlation and induce irregular calcium release and calcium alternans (Kanaporis et al., 2023). Combining the results of decreased I_Ca-L current and [Ca²⁺]_i after CIT treatment, the membrane potential may also be influenced, which may, in return, result in AP oscillation, AP alternans, and CaT alternans.

In our study, we found that APD was prolonged in spatially discordant regions of ventricles in CIT-treated hearts by optical mapping results, and the transcriptomics data revealed that KCND2, KCNH2, KCNIP2, and KCNE1 were reduced following treatment with CIT, which are related to potassium channels and APD prolongation. Although there have been studies suggesting that CIT affects Nav1.5 in HEK293 (Nakatani and Amano, 2021), our findings from ECG (QRS wave morphology), optical mapping (conduction velocity), and transcriptomics data did not demonstrate any alterations in Nav1.5 expression or function. This absence of change may be attributed to compensatory mechanisms that arise following prolonged exposure to CIT. More research may be needed in future to explore the effects of CIT on Nav1.5 in different cell types or species, especially in human cardiomyocytes. In rodent hearts, fast transient outward K⁺ currents (I_{to, f}) are generated by K_V4 α-units (i.e., K_V4.2 and K_V4.3) and the Ca²⁺-sensing auxiliary subunit (KCNIP2), with KCND2 being the main gene that codes for K_V4.2. Meanwhile, KCNE1 also plays a critical role in the regulation of K_V4.3. We found that APD was profoundly prolonged during the entire repolarization process of CIT-treated cardiomyocytes, as shown by the whole-cell patch clamp technique, and I_to current density was remarkably reduced. Even though the protocol was used for total potassium current recording and peak current density calculated as I_to, the variation tendency of I_to caused by CIT was confirmed. Prolonged APD₃₀ was consistent with weakened I_to, while APD₉₀ was also prolonged at the same time, suggesting that I_ks and I_kr may be influenced to some extent since the mRNA expressions of KCNE1 and KCNH2 were also downregulated (KCNE1 is involved in I_ks, and KCNH2 is the main coding gene of I_kr). Additionally, Chen et al. found that the polymorphisms of the KCNQ1, KCNE1, and KCNH2 genes have a complementary effect on escitalopram-induced QTc prolongation in elderly patients (Chen et al., 2024). Meanwhile, the Kv4.2 protein level was significantly reduced, which is the main component of I_to. Taken together, CIT had a detrimental effect in prolonging APD by downregulating the activities of multiple potassium channels. Admittedly, the electrophysiology of rodent hearts cannot be completely equivalent to that of human cardiomyocytes or iPSCs because of the action potential plateau. However, experiments based on murine hearts can serve as a good example for studying arrhythmogenesis in drug-induced arrhythmias based on the repolarization and potassium current. Moreover, in our I_CaL current tests using the patch clamp, we also recorded a reduction in I_Ca-L current after CIT treatment. This might also be one of the reasons for the prolongation of APD.

Fluctuation of the sarcoplasmic reticulum calcium content plays a key role in cardiac alternans (Fakuade et al., 2021). Meanwhile, Qu et al. also mentioned that alterations in the expression or function of various calcium handling proteins are responsible for most cardiac alternans (Qu et al., 2013). In our study, we failed to observe any change in the expression of calcium-handling proteins in CIT-treated hearts, even though cardiac alternans are proven to not rely on calcium fluctuation (Picht et al., 2006); however, we found reduced [Ca²⁺]_i and I_Ca-L in the patch-clamp technique, which was one of the factors that generate CaT alternans. Meanwhile, to our knowledge, CaT alternans are widely accepted to precede AP alternans and are a predictor of arrhythmias (Kanaporis and Blatter, 2015). Although a considerable number of researchers still hold different opinions, they believe that prolonged action potentials and AP restitution play key roles in generation of AP alternans (Watanabe et al., 2012; Bingen et al., 2013; Kanaporis et al., 2023). CaT alternans and AP alternans are synchronized in time and magnitude (Kanaporis et al., 2023), and shortened APD rescued calcium alternans (Kanaporis et al., 2019). In addition, short APs generated by K⁺ channel activity have been associated with Vm-driven alternans (Fish and Antzelevitch, 2008). A recent study also demonstrated that activation of SK channels suppressed AP and CaT alternans (Kanaporis and Blatter, 2023). These studies suggest that potassium channels are widely linked to AP and CaT alternans. In our study, downregulation of I_to and other potential potassium currents prolonged APD and evoked aggravated AP alternans, which induced severe accompanying calcium alternans and led to ventricular arrhythmias. In addition, potassium current enhancement may alleviate AP alternans by weakening APD–CaT coupling (Livshitz and Rudy, 2007). I_to downregulation (potassium current weakening) then, in turn, enhanced AP alternans. APD–CaT uncoupling can also lead to CaT alternans in the absence of AP alternans (Kanaporis et al., 2023). The chaotic relationship of Vm and CaT may thus play an auxiliary role in the genesis of calcium alternans, even though we cannot currently clarify the complex and sophisticated interactions between the membrane channels and calcium-handling proteins. In general, CaT alternans manifested as the arrhythmogenic substrate after CIT treatment, and it was mainly generated by downregulation of Ito, weakened I_CaL, and the impaired coupling of AP and CaT.