Male C57BL/6J mice (aged 4–6 weeks) were purchased from Xi’an Jiaotong University (Xi’an, China) and bred under standard laboratory conditions. After 1 week of acclimatization, a total of forty mice were randomly divided into 4 groups (n = 10 per group): 1) Standard diet group (STD; 20% fat, 56% carbohydrate, 24% protein; Research Diets Inc., New Brunswick, NJ); 2) STD + LCA group; 3) HFD group (HFD; 60% fat, 20% carbohydrate, 20% protein; Research Diets Inc.); 4) HFD + LCA group. After feeding HFD for 8 weeks, the mice became obese and showed greater propensity for AF. Subsequently, FAO activator, LCA (150 mg/kg⋅BW/d; TargetMol, Boston, United States) (Bakermans et al., 2013), was administrated via drinking water for another 4 weeks. Based on our preliminary experiment, the experimental dose of LCA was set to be 150 mg/kg⋅BW/d, which had no significant impact on mice body weight (BW) but suppressed obesity-induced AF susceptibility. At the end of the experiment, AF induction, echocardiography, intraperitoneal glucose tolerance test (IPGTT) and insulin tolerance test (ITT) were performed in each group before tissue sampling. After overnight fasting, atrial tissues and blood samples were collected from euthanized mice for further analysis. All the procedures of this study were approved by the Institutional Animal Care and Use Committee of Xi’an Jiaotong University.

Primary atrial cardiomyocytes were isolated from the atria of neonatal Sprague–Dawley rat (1∼3-day-old; Xi’an Jiaotong University, Xi’an, China). Briefly, atria tissue were surgically removed, trypsinized (0.08% trypsin; Solarbio, Beijing, China), digested with 0.1% collagenase Ⅱ (Solarbio), and the primary atrial cardiomyocytes were isolated by differential detachment and verified under the microscope. Isolated cells were cultured in 6-well plates at 37°C in 5% CO₂, with DMEM supplemented with 10% FBS and 1% penicillin/streptomycin solution. Final solutions of 200 μM PA (dissolved in 20% BSA; Sigma-Aldrich, St. Louis, United States) was added to replicate the effects of lipid overloading, while LCA (5 mM; TargetMol) was added to enhance FAO, and Compound C (CC, 0.5 μM; Sigma-Aldrich) was added to inhibit AMPK activation. After being treated for 24 h, cells were collected, washed and lysed for the following evaluations.

Programmed trans-esophageal stimulation was performed to assess AF inducibility as described previously (Fu et al., 2021). Briefly, mice were anesthetized with pentobarbital intraperitoneally (50 mg/kg⋅BW) and then electrical stimulated by an external simulator (SCOPE, Kaifeng, China), and surface electrocardiogram (ECG) was recorded by a physiologic signal-acquisition system (RM6240; Chengdu instrument factory, Chengdu, China). At first, baseline ECG was analyzed by 10 consecutive beats recorded in the initial stabilization period with heart rates between 300 and 500 bpm at first. Later, AF was induced with burst pacing (pulse width 1 ms; 1.5× capture threshold; 30, 35, and 40 Hz), and was considered sustained as persisted rapid irregular f-waves with irregular R-R intervals lasting for more than 1 s. In addition, atrial effective refractory period (ERP) was assessed by continuous stimulation applied with decreasing R-R intervals from 140 to 40 ms at 1 ms decrements. Sinoatrial node recovery time (SNRT_max) was measured as the longest duration between the last stimulus and the first sinus P-wave, and corrected by the R-R interval (_cSNRT_max).

2D echocardiography was employed to discern cardiac structural and functional differences among groups. Echocardiography (Vevo 2,100; VisualSonics Inc., Toronto, Ontario, Canada) was performed by an animal cardiologist blind to the experimental design in mice anaesthetized with inhalational isoflurane. Dimension of superoinferior (SI), anteroposterior (AP) and mediolateral (ML) were obtained in a long-axis view and a short-axis view, respectively. LA filling volume was calculated using the formula: LA Volume = (4π×SI×AP×ML)/(3 × 2×2 × 2) (Ujino et al., 2006). Cardiac function was calculated according to standard formulae, and the results were averaged of three cardiac cycles.

Glucose and insulin homeostasis were evaluated in vivo by IPGTT and ITT. For the IPGTT, D-glucose was injected intraperitoneally (2 g/kg⋅BW) into over-night fasted mice, and, subsequently, blood was taken from the tail vein at 5, 15, 30, 60, 90 and 120 min after glucose loading and blood glucose levels were measured by the Accu-Chek glucometer (Roche Diagnostics, Indianapolis, United States). Similarly, for the ITT, mice were fasted for 2 h before insulin administration intraperitoneally (1 U/kg⋅BW; Wanbang Biopharma, Xuzhou, China), following the glucose determination at 0, 15, 30, 45 and 60 min. Glucose tolerance and insulin sensitivity were calculated as the area under the curve (AUC) by GraphPad Prism 8.0 (GraphPad Software, San Diego, America).

After being isolated from heart tissue, atrial appendages were fixed with 4% paraformaldehyde overnight, embedded in paraffin, and cut into 5 µm slides longitudinally. Paraffin-embedded specimens were finally stained with hematoxylin-eosin (H&E), Wheat Germ Agglutinin (WGA), Masson, Periodic acid-Schiffs (PAS), selected antibodies (connexin-43; 1:200; Invitrogen, California, United States, NRF2; 1:500; Proteintech, Chicago, United States, and 8-Hydroxy-2′-deoxyguanosine (8-OH-dG), 1: 100; JalCA, Shizuoka, Japan), and TUNEL (in situ cell death detection kit, TMR red; Roche Diagnostics). Respectively, frozen sections (10 µm thickness) of atria samples were prepared and stained with oil red O, MitoSOX and dihydroethidium (DHE). Slides were visualized and photographed by an optical microscope (Nikon, Melville, NY, United States) at a magnification of 100× for histology, or by a confocal microscope (Leica, Bensheim, Germany) at a magnification of 20× for immunofluorescence. Images were further analyzed by ImageJ software (version 1.46r; National Institutes of Health, Bethesda, United States). Data were expressed as the percentage of the positive-stained region to the total area of cardiomyocyte area, and averaged from six random fields in each slide.

Biomarkers were assayed using commercially available kits in accordance with manufacturer’s guidelines, including fasted non-esterified FAs (NEFA) (Solarbio), lactate dehydrogenase (LDH) (Nanjing Jiancheng Bioengineering Institute, Nanjing, China), creatine kinase-MB (CK-MB) (Nanjing Jiancheng Bioengineering Institute), malondialdehyde (MDA) (Beyotime Institute of Biotechnology, Shanghai, China) and superoxide dismutase (SOD) (Beyotime Institute of Biotechnology).

Cellular FAO was measured with a FAO Assay Kit (ab222944; Abcam, Cambridge, United States) according to the manufacturer’s instruction. Fluorescence was detected using FLUOstar Omega (BMG LABTECH, Aylesbury, United Kingdom) and the results were normalized by protein concentration in each well through a micro-bicinchoninic acid (BCA; Pierce Chemical Company, Rockford, United States) kit.

Protein levels were measured by WB with β-actin as the loading control. Protein was extracted from atria samples or cells and quantitated by a BCA assay. Equal amount (10–30 μg) of protein was loaded and separated by SDS-PAGE using 10 or 12% acrylamide gradients, and later transferred to nitrocellulose membranes. The membranes were blocked with 5% skim milk and incubated with antibodies against total-AMPK (1:1,000; Cell Signaling Technology/CST, Massachusetts, United States), CD36 (1:1,000; Abcam), collagen Ⅰ (1:1,000; Abcam), collagen Ⅲ (1:1,000; Abcam), connexin-43 (1:200; Invitrogen), CPT1B (1:1,000; Proteintech), GLUT4 (1:500; Proteintech), NFκB (1:1,000; Proteintech), NRF2 (1:1,000; Proteintech), phoso-AMPK (Thr172; 1:1,000; CST), phoso-Akt (Ser473; 1:1,000; CST), phoso-NFκB (1:1,000; CST), PGC1α (1:1,000; Proteintech), SOD2 (1:1,000; Proteintech), TGF-β (1:1,000; Abcam), α-SMA (1:1,000; CST), β-actin (1:5,000; Proteintech). Protein levels were quantified as the intensity of bands using ImageJ software (NIH systems) and standardized to β-actin. Abbreviations are fully illustrated in corresponding figure legends.

RT-qPCR was carried out to quantify gene expressions among the groups. Briefly, RNA from atria tissues was extracted using TRIzol reagent (Accurate Biology, Changsha, China). Then, the high-sensitivity RT-qPCR reaction was measured using the SYBR green chimeric fluorescence method (Accurate Biology) and detected by CFX96™ Real-Time PCR System (Bio-Rad, Hercules, CA, United States). Results were quantified using the 2^-△△Ct comparative method and normalized by β-actin. The primer sequences and full names of genes are listed in Supplementary Table S1.

All data of animal and cell studies were analyzed by SPSS 23.0 (IBM SPSS software, New York, United States), visualized by GraphPad Prism 8.0 (GraphPad Software), and shown as mean ± SEM. Replicates are indicated in figure legends and table legends. Data normality was evaluated by Kolmogorov-Smirnov test, and comparisons between groups were performed with One-way ANOVA (in mice experiments) or Two-way ANOVA (in cell experiments) with Bonferroni post-hoc test. For all tests, p < 0.05 was considered significant.

In this study, we established a mouse model of obesity-related AF through 13-week HFD feeding, and LCA was administrated via drinking water (150 mg/kg⋅BW/d) during the last 4-week period to investigate whether enhancing FAO could attenuate obesity-mediated AF. The schematic of the experimental design is shown in Figure 1A, and basal characteristics of mice in each group are summarized in Table 1 .

As expected, at the end of the experiment, HFD mice gained 20% more BW than STD mice (Supplementary Figure S1A), together with increased heart weight (HW) and serum NEFA (Supplementary Figures S1B–D). Obesity-mediated AF was successfully established, as supported by increased AF frequency and prolonged duration in HFD mice (Figures 1B–D). LCA decreased the ratio of HW/NAL (nasal-anal length) in HFD mice, yet no significant change of BW gain was observed with or without LCA supplementation (Supplementary Figures S1A–C). In addition, LCA suppressed HFD-induced elevation of serum NEFA (Supplementary Figure S1D). Of note, LCA supplementation alleviated pacing-induced AF susceptibility in the obese mice. Surface ECG (Lead II) at baseline showed that LCA supplementation significantly decreased P-wave duration (P_max) and P-wave area (P_area), two of the independent predictors for AF, in obese mice (Figures 1E–G). In addition, electrophysiological abnormalities, including prolonged SNRT_max and shortened ERP, were regarded to increase the AF propensity. Supportively, LCA supplementation shortened SNRT_max and extended ERP in obese mice (Figures 1H–L).

Atrial structural remodeling, including atrial dilatation, cellular hypertrophy, gap junction disturbance, and interstitial fibrosis, offers substrates for AF. Thus, we later assessed atrial structural alterations among the groups.

Obese mice displayed excessive left atrial enlargement, as shown by marked increase of left atrium (LA) diameter and LA filling volume (Figures 2A–C). In atrial cardiomyocytes, obesity developed substantial cellular hypertrophy and disarray in the atria, as evidenced by H&E staining, WGA staining and β-MHC transcription (Figures 2D–G). Moreover, the main gap junction protein, connexin-43, was apparently upregulated in expression and heterogeneous distributed in the atrial cardiomyocyte of obese mice (Figures 2H–J). In addition, intra-myocardial fibrosis, a hallmark of AF, was observed to increase in the atria of obese mice, as evidenced by aggravated collagen deposition, upregulated pro-fibrosis signaling (TGF-β, α-SMA, Smad3, collagen Ⅰ, and collagen Ⅲ), and downregulated transcription of anti-fibrotic Smad7 (Figures 2K–O).

Coincided with reduced AF inducibility, LCA supplementation showed significant capacity to shrink LA, attenuate cellular hypertrophy, gap junction disturbance, and fibrosis while challenged by a long-term HFD (Figure 2), indicating FAO promotion via LCA supplementation constrained the AF substrates in obese mice.

Epigenetic studies demonstrated that AF is highly associated with lipid metabolic abnormalities. In line with this, apparent lipid deposition was observed in obese mice. What’s more, in accordance with reduced AF susceptibility and atrial remodeling, promoting FAO with LCA decreased the atrial lipid deposition effectively in obese mice (Figures 3A,B).

Promoted FAs influx and defective FAO can predispose to lipid deposition (Figure 3C). Thus, firstly, we assessed factors involved in lipid uptake and transportation. CD36, the main trans-membrane translocase of FAs, was decreased in gene level in obesity, and increased after LCA treatment (Figure 3D). However, CD36 was unchanged in total protein level among the groups (Figure 3E), but elevated in atrial membrane translocation in obese mice, and slightly reduced after LCA supplementation (Figures 3F,G). Besides, the transcription of other transportation-related genes, FABP3 and FABP-pm, was decreased in the atria of obesity and restored after LCA supplementation (Figure 3D).

Next, we evaluated the expression of the key enzymes of FAO, including CPT1B, AMPK and PGC1α, by RT-qPCR and Western blot. In contrast to increased FAs uptake, mitochondrial FAO was downregulated in obesity, as evidenced by the remarkable decrease of AMPK phosphorylation, and PGC1α expression in the atria of HFD mice compared to STD mice (Figures 3H–L). Notably, LCA supplementation upregulated the phosphorylation of AMPK as well as the expression of PGC1α and CPT1B in HFD mice (Figures 3D–L).

Lipid over-deposition undermines the structure of cardiomyocytes, namely, cardiac lipotoxicity (Sletten et al., 2018), thereby provoking a set of pathological processes, including oxidative inflammation, DNA damage and insulin resistance, which have been implicated as possible mechanisms for AF. Thus, firstly, we evaluated oxidative stress. Obesity increased mitochondrial superoxide production (measured by MitoSOX staining) and increased cellular oxidative stress (measured by DHE staining) in the atria (Figures 4A–D). Besides, obesity increased MDA contents in the serum and atria, and decreased atrial SOD, but unaffected serum SOD level. LCA supplementation thwarted oxidative damages provoked by HFD, as supported by the decreased MitoSOX Red and DHE fluorescence signal, the reduced MDA levels and the rise of SOD levels in the serum and atria of obese mice (Figures 4A–H).

Nuclear factor erythroid 2-related factor (NRF2) is an accepted master regulator participating in the cellular adaptive response to redox or energy stress, which accumulates in nuclear to induce a battery of defensive genes encoding detoxifying enzymes and antioxidant proteins (such as SOD2 and NFκB) (Figure 4E) (Tonelli et al., 2018). Next, we evaluated the NRF2-mediated cardio-protective pathways. Nuclear levels of NRF2 were comparably decreased in the atria of obese mice, and increased after LCA supplementation (Figures 4J,K). In addition, total protein level of NRF2 was elevated after LCA supplementation. In parallel with the distinct NRF2 activation, SOD2, one of the downstream of NRF2 for antioxidant defense, was noticeably increased in the atria of obese mice after LCA treatment (Figures 4L,M).

Consistent with obesity-induced oxidative stress, oxidative DNA damage (as shown by 8-OH-dG staining) and DNA segment (as shown by TUNEL staining) in the atira was aggravated after HFD, and were alleviated by LCA (Figures 4N–Q). Moreover, we evaluated myocardial cellular damage by measuring serum LDH and CK-MB levels. Obesity increased LDH levels apparently, yet LCA treatment unaffected its levels in obese mice. CK-MB is a more specific marker of myocardial cellular damage, and results showed that obesity unaffected serum CK-MB levels, yet LCA treatment reduced its levels (Figures 4R,S) in obese mice.

Next, we determined the transcription of pro-inflammation cytokines in the atria, including IL-1β, IL-6, IL-18, TNF-α, and MCP-1. Results demonstrated that HFD increased the transcription of IL-6 compared to STD, while LCA suppressed the transcription of most pro-inflammation cytokines, except for MCP-1, in both STD and HFD mice (Figure 4T). In addition, NFκB, a well-known downstream factor of NRF2, is a central activator of the immune response. Our results demonstrated that NFκB was activated in the atria of HFD mice, and was inhibited after LCA supplementation (Figure 4U), which is in line with our results showing LCA increased the expression and activation of NRF2 (Figures 4I–M).

Disordered glucose metabolism and insulin resistance are also novel risk factors for AF and usually occur in obesity (Nakamura and Sadoshima, 2020). Therefore, apart from FAs metabolism, the impacts of LCA on glycometabolism, including insulin signaling and insulin resistance, glucose and insulin tolerance, are also investigated in this study.

Both IPGTT (Figures 5A–C) and ITT (Figures 5D–F) confirmed glucose metabolic imbalance in obesity, including increased fasting blood glucose (FBG), impaired glucose tolerance, and decreased insulin sensitivity. Intriguingly, LCA had no effect on glucose tolerance and FBG, yet increased insulin tolerance in obese mice, indicating improved insulin sensitivity in peripheral tissues. Mechanistic investigation using WB analysis showed that LCA restored the serine/threonine protein kinase PKB (Akt) phosphorylation in the atria of obese mice (Figure 5G). Moreover, although LCA treatment had no effect on GLUT4 expression, its activation, as reflected by its translocation to the intracellular membrane, was promoted in obese mice (Figures 5H–J). Next, we assessed the transcription of glucose metabolism-related genes, including GLUT1, GLUT4, HK2, PFKM, PKM2, and PDK4. Most of these genes, except for PFKM, were inhibited in the atria of obese mice. LCA supplementation increased the transcription of GLUT4, PKM2, and PDK4, whereas decreased PFKM transcription in obese mice (Figure 5K). In addition, PAS staining showed aggravated glycogen accumulation in the atria of obese mice. LCA further increased glycogen accumulation in atrial cardiomyocytes (Figures 5L,M), probably due to the fact that LCA upregulated glucose uptake more than glycolysis (Figure 5N).

To investigate whether AMPK is crucial in LCA-conferred cardio-protection, we used PA, a saturated fatty acid, to mimic obesity in vitro, and CC, a pharmacological inhibitor of AMPK, to block AMPK signaling in primary atrial cardiomyocytes (Figure 6A). LCA enhanced FAO significantly in PA-treated cells, which might be ascribe to the activation AMPK and upregulation of PGC1α and CPT1B; however, inhibition of AMPK by CC reversed these effects (Figures 6C,D).

MDA examination was performed to assess the levels of oxidative stress. In line with the animal studies, LCA supplementation induced a decrement of PA-induced oxidative stress (Decreased cellular MDA) and enhanced corresponding anti-oxidative system (Restored NRF2/SOD2 signaling) in primary atrial cardiomyocytes; however, pretreatment with CC abolished LCA-conferred antioxidant effects. Similarly, pretreatment with CC also attenuated LCA-conferred anti-inflammation effects, as judged by NFκB activation (Figures 6E–G).

Altogether, these results supported that activation of AMPK signaling pathway may be the relevant molecular basis of LCA-mediated cardio-protection.