Cardiac overexpression of perilipin 2 induces atrial steatosis, connexin 43 remodeling, and atrial fibrillation in aged mice

Abstract

Atrial fibrillation (AF) is prevalent in patients with obesity and diabetes, and such patients often exhibit cardiac steatosis. Since the role of cardiac steatosis per se in the induction of AF has not been elucidated, the present study was designed to explore the relation between cardiac steatosis and AF. Transgenic (Tg) mice with cardiac-specific overexpression of perilipin 2 (PLIN2) were housed in the laboratory for more than 12 mo before the study. Electron microscopy of the atria of PLIN2-Tg mice showed accumulation of small lipid droplets around mitochondrial chains, and five- to ninefold greater atrial triacylglycerol (TAG) content compared with wild-type (WT) mice. Electrocardiography showed significantly longer RR intervals in PLIN2-Tg mice than in WT mice. Transesophageal electrical burst pacing resulted in significantly higher prevalence of sustained (>5 min) AF (69%) in PLIN2-Tg mice than in WT mice (24%), although it was comparable in younger (4-mo-old) mice. Connexin 43 (Cx43), a gap junction protein, was localized at the intercalated disks in WT atria but was heterogeneously distributed on the lateral side of cardiomyocytes in PLIN2-Tg atria. Langendorff-perfused hearts using the optical mapping technique showed slower and heterogeneous impulse propagation in PLIN2-Tg atria compared with WT atria. Cardiac overexpression of hormone-sensitive lipase in PLIN2-Tg mice resulted in atrial TAG depletion and amelioration of AF susceptibility. The results suggest that PLIN2-induced steatosis is associated with Cx43 remodeling, impaired conduction propagation, and higher incidence of AF in aged mice. Therapies targeting cardiac steatosis could be potentially beneficial against AF in patients with obesity or diabetes.

INTRODUCTION

Cardiac intracellular accumulation of lipid droplets (LDs), also known as “cardiac steatosis,” correlates significantly with obesity and diabetes. In 1977, Regan et al. (37) reported from autopsy investigation that myocardial triacylglycerol (TAG) content is threefold higher in diabetic patients than in nondiabetic individuals. Advancement in medical biotechnology and the use of magnetic resonance imaging have confirmed the presence of significantly higher levels of myocardial TAG in patients with obesity and type 2 diabetes mellitus (T2DM) compared with control groups (30) and that high TAG contents correlate with impaired systolic and diastolic functions (38, 48). These early studies laid the foundations for the link between cardiac steatosis and cardiovascular diseases in obese and diabetic patients and allowed the creation of the concept of lipotoxic cardiomyopathy (40, 56).

The pathophysiological roles of cardiac steatosis have commonly been investigated in rodent models of diabetes. LDs accumulate in cardiomyocytes of Zucker diabetic fatty rats, together with accumulation of TAG and ceramide, leading to cardiomyocyte apoptosis via induction of inducible nitric oxide synthase (55). The db/db mice exhibit cardiac steatosis, high levels of fatty acid (FA) oxidation, and generation of mitochondrial reactive oxygen species (ROS), associated with impaired cardiac function (6, 8). High myocardial levels of TAG and diacylglycerol (DAG), augmented FA oxidation, reduced glucose oxidation, and impaired cardiac function have been described in the mouse model of type 1 diabetes, streptozotocin-induced diabetic mice (5, 42).

Chronic hyperglycemia, which is often associated with insulin resistance enhancers, can provoke the complex pathogenic process of cardiomyopathy in diabetes. The latter condition is precipitated by 1) FA overload and increased ROS production, 2) LD accumulation and intracellular lipotoxicity, 3) accumulation of advanced glycation end products, 4) endoplasmic reticulum stress, 5) impairment of mitochondrial Ca²⁺ handling, and 6) activation of the renin-angiotensin-aldosterone system (RAS) (17, 19). In addition, recent studies identified other cellular dysfunctions, including:7) autophagy dysfunction (21), 8) epigenetic modification (43), and 9) dysregulation of micro-RNA in diabetic hearts (4). These factors aggregate to initiate the development of a complex pathophysiology that leads to cardiomyocyte apoptosis, inflammation, interstitial fibrosis, and eventually cardiac dysfunction.

Atrial fibrillation (AF), which is the most common arrhythmia in humans, is also prevalent in obese and diabetic patients. Previous studies reported a close association between body mass index and total fat mass and the risk of AF in elderly individuals (3). Recent epidemiological studies have identified obesity, T2DM, and aging as major risk factors for AF (33). Furthermore, a meta-analysis of cohort and case-control studies found that T2DM increases the relative risk of AF by 39% (13). Rodent models of Goto-Kakizaki diabetic rats and high-fat diet-induced obese mice exhibit high chance of atrial arrythmogenicity (20, 49). Chronic inflammation and the resultant interstitial fibrosis are known to be major factors for AF in obesity and diabetes (10), although the precise mechanism has not yet been clarified. In particular, the role of cardiac steatosis per se in the induction of AF has not been investigated to date.

On the basis of the evidence from human studies and animal models of diabetes, we hypothesized in the present study that cardiac steatosis plays a critical role in the pathogenesis of AF in obesity and diabetes. To study these effects, we utilized our recently described mice with cardiac-specific overexpression of perilipin 2 (PLIN2) (51). PLIN2 induces dynamic steatosis without adverse effects on ventricular function and without changes in plasma glucose and lipid levels. The present study was conducted in aged PLIN2-overexpressing mice to address the effects of atrial accumulation of LDs on the induction of AF. The results link atrial steatosis to remodeling of the gap junction protein connexin 43 (Cx43), a key molecule in AF.

MATERIALS AND METHODS

Cardiac-specific PLIN2-overexpressing mice.

Transgenic (Tg) mice with cardiac-specific PLIN2 overexpression (lines AD17 and AD19) were generated as described in detail previously (51). Both Tg lines were backcrossed to C57BL/6 mice for 10 generations, maintained by intercross, and experiments were performed using both Tg lines. The results were confirmed in both Tg lines, and representative data are presented unless otherwise noted. Heterozygous Tg mice with cardiac-specific overexpression of hormone-sensitive lipase (HSL) (52) were crossed with AD17 Tg mice to generate double-Tg mice, which overexpress both PLIN2 and HSL specifically in cardiomyocytes (Tg+HSL). All procedures were conducted in accordance with the “Regulations for Animal Research” at the University of Fukui and approved by the Animal Research Committee, University of Fukui.

Animal studies.

Heterozygous Tg mice and wild-type (WT) littermates were used in all experiments. Both male and female mice were used unless otherwise noted. The mice were maintained on a regular chow diet (MF, Oriental Yeast), kept under a 12:12-h dark-light cycle until the age of 12–15 mo, and housed in individual cages 2 wk before the experiments. The mice were fed ad libitum and had free access to water.

Western blot analysis.

Immunoblot analysis was performed using specific antibodies against PLIN2 (no. ab52356, Abcam), Cx43 (no. 3512, Cell Signaling Technology), phosphorylated Cx43 (no. 3511, Cell Signaling Technology), α-tubulin (no. ab18251, Abcam), and GAPDH (no. MAB374, Chemicon), as reported previously, and analyzed with a FluorChem IS-8000 (Alpha Innotech) (51). Phosphatase inhibitor cocktail (Nacalai Tesque, Kyoto, Japan) was added in the tissue homogenization step for Cx43 analysis.

Tissue lipid content.

The hearts were perfused with 3 ml of PBS from the left ventricles and excised. Both the right and left atria were removed on a chilled stage using a stereomicroscope and frozen in liquid nitrogen. The left and right atria of five mice were pooled together, weighed, and homogenized in 20 volumes of PBS, and lipids were extracted with 20 volumes of chloroform-methanol (2:1). TAG content was measured as described in detail previously (51).

Liquid chromatography-mass spectrometry.

Twelve male WT and AD17-Tg mice were used for this analysis. They were divided into two groups (n = 6). Both atria from the six mice were excised and pooled together, and lipids were extracted as described above. The extracted lipids were dissolved and subjected to liquid chromatography-mass spectrometry (LC-MS) analysis, as described in detail previously (51). References for each annotated compound were searched for in KNApSAcK and Kyoto Encyclopedia of Genes and Genomes (KEGG) databases (1), and then lipid compounds (quality grades A and B) were extracted using LipidSearch software (Mitsui Knowledge Industry). Comparison analyses were performed between WT and Tg mice and species that showed greater than twofold changes in the two independent comparison analyses were extracted.

Echocardiography.

Cardiac function was analyzed by echocardiography in awake mice with ultrasonography equipped with a 13-MHz linear transducer (ALOKA), as described in detail previously (51).

Microscopic examination.

The excised left atrium was spread and pinned on the silicone gel, fixed with 4% paraformaldehyde-PBS for 30 min, and embedded in O.C.T. compound (Sakura Finetek Japan). The sections were stained with specific antibodies against PLIN2 (no. ab52356, Abcam) or Cx43 (no. 3512, Cell Signaling Technology) before examination under the microscope (AX80, Olympus). For confocal microscopy, the sections were stained with specific antibodies against Cx43, N-cadherin (no. ab98952, Abcam), and fluorescent secondary antibodies (nos. A11011 and A21050, Invitrogen) and analyzed with a confocal microscope system (TCS Sp2, Leica). Electron microscopy was performed as described in detail previously, using a transmission electron microscope (H-7650, Hitachi) (51).

mRNA expression analysis.

Total RNA was extracted from the cardiac atrium using TRIzol reagent (Invitrogen), and reverse-transcribed using a Quantitect reverse transcription kit (Qiagen). The target genes were amplified and analyzed as described previously (51).

Microarray analysis.

Cardiac ventricular RNA from four mice per group (WT or AD17) was pooled and used for subsequent double-strand cDNA and biotin-labeled cRNA synthesis (47). The labeled cRNA was applied to an Affymetrix oligonucleotide array GeneChip Mouse Genome 430 v. 2.0 following the Affymetrix GeneChip expression analysis protocol. The data were analyzed by the Gene Chip Operating Software v. 1.4.

Induction of AF.

AF was induced as described previously (28). Briefly, two platinum bipolar electrodes set at 1-mm interelectrode distance were inserted into the esophagus to stimulate the epicardial surface of the left atrium and to record the atrial electrogram, respectively. A surface electrocardiogram (ECG) lead II was also recorded. The bipolar electrode was connected to an electrical stimulator (SEN7013, Nihon Kohden). AF was induced by electrical burst pacing from the esophagus at the left atrium (S₁S₁ = 20 ms) for 30 s. Sustained AF was defined as irregular atrial rhythm (atrial cycle lengths <80 ms) lasting longer than 5 min. When AF spontaneously terminated within 5 min, burst pacing was repeated up to five times, or the experiment ended when AF persisted longer than 5 min.

Langendorff-perfused mouse heart.

Mice were anesthetized and treated with sodium heparin (500 USP units/kg iv). The hearts were rapidly excised and connected to a modified Langendorff apparatus (12). A polytetrafluoroethylene-coated silver bipolar electrode was used to stimulate the epicardial surface of the atria, which included the pulmonary veins. Each preparation was perfused under constant-flow conditions with oxygenated (95% oxygen, 5% CO₂) Tyrode solution containing (in mM): 141.0 NaCl, 5.0 KCl, 1.8 CaCl₂, 25.0 NaHCO₃, 1.0 MgSO₄, 1.2 NaH₂PO₄, 5 HEPES, and 5.0 dextrose (pH 7.4, at 36 ± 1°C). Perfusion pressure was measured with a pressure transducer (Nihon Kohden) and maintained within the pressure range of 50–60 mmHg by adjusting the flow. Cardiac rhythm was monitored using three-pin electrodes fixed to the chamber in positions corresponding to ECG limb leads II.

Optical mapping.

Experiments were basically performed as described previously (12). The isolated heart in the Langendorff apparatus was allowed to stabilize for ~20 min before being stained with 8 µM potentiometric fluorophore (RH237, Promo Kine). To prevent motion artifact, cardiac contractions were arrested by treating the heart with the electromechanical uncoupler 15 µM blebbistatin (Toronto Research Chemicals). The isolated hearts were illuminated with two 530-nm LEDs (LEX2-G, Brain Vision). Fluorescence was acquired for 2 s through a 715-nm long-pass filter using a complementary metal oxide semiconductor (CMOS) camera (MiCAMA02-CMOS, Brain Vision) at a sampling rate of 1 kHz and 9 × 4.5 mm fields of view. The isolated hearts were subjected to electrical pacing from the epicardial surface of the atrium at a basic cycle length of 150 ms. In all experiments, automated algorithms were used to determine depolarization time relative to a single fiducial point (i.e., the stimulus). Depolarization time was defined as the point of maximum positive derivative in the action potential upstroke (dV/dt_max). Depolarization contour maps were computed for the entire mapping field. The impulse conduction time (ICT) on both atria was defined as the difference between the earliest and latest depolarization times in the mapping field.

Statistical analysis.

All values are expressed as means ± SE. Differences between groups were examined using Student’s t test unless otherwise noted. A P value <0.05 denoted the presence of significant difference.

RESULTS

Atrial steatosis in PLIN2-Tg mice.

Atrial PLIN2 expression and lipid accumulation were analyzed in AD17- and AD19-Tg mice. As shown in Fig. 1A, robust expression of enhanced green fluorescent protein (EGFP)-PLIN2 fusion protein was present in the atria of AD17 and AD19 mice, the expression of the fusion protein was higher in AD19 than in AD17 mice. Immunostaining showed strong expression of PLIN2 in atrial cardiomyocytes in Tg mice (Fig. 1B). Electron microscopy showed the presence of many LDs associated with mitochondrial chains in the left atria of both Tg lines (Fig. 1C). In contrast to Tg mice, few LDs were detected in the atria of WT mice. In parallel with the EGFP-PLIN2 protein expression, atrial TAG content was five- to sixfold (AD17) and seven- to ninefold (AD19) higher than that of WT mice (Fig. 1D). Despite severe steatosis, atrial weight/body ratio was comparable between WT and Tg mice in both young (4- to 5-mo-old) and aged (13- to 16-mo-old) mice and also between young and aged mice of the two genotypes (Table 1). Histological examination using light microscopy showed no infiltration of inflammatory cells or interstitial fibrosis in the Tg atria (Fig. 1, B and E). Electron microscopy showed no morphological changes, such as apoptotic changes (i.e., compaction of the nuclear chromatin or nuclear fragmentation) or mitochondrial degeneration, in the atrial cardiomyocytes of both Tg lines.

Fig. 1.

Atrial perilipin 2 (PLIN2) expression and steatosis in PLIN2-transgenic (Tg) mice. A: Western blot analysis for PLIN2 in atria of wild-type (WT), AD17, and AD19 mice. Tissue homogenates of atria of WT, AD17, and AD19 Tg mice were subjected to SDS-PAGE, blotted, and probed with anti-PLIN2 and anti-α-tubulin antibodies. EGFP, enhanced green fluorescent protein. B: representative light micrographs of atrial tissues of WT and AD17-Tg mice immunostained for PLIN2. Scale bars, 30 μm. C: representative electron micrographs showing accumulation of small lipid droplets (arrowheads) in AD17 and AD19 atria. Scale bars, 1 μm. D: atrial triacylglycerol (TAG) content in WT, AD17, and AD19 mice. Hearts were perfused with 3 ml of PBS from the left ventricle, and both atria were excised. Both atria from 5 mice (3 males, 2 females) were pooled and homogenized, and TAG content was measured. Each group consisted of 10 mice, and values of 5 mice are presented. E: atrial collagen deposition. Representative light micrographs of atrial tissues of WT and AD17-Tg mice stained with Masson’s trichrome. Scale bars, 50 μm.

Table 1.

Atrial weight/body weight in male WT and PLIN2-Tg AD17 mice

Genotype	Young (n = 4, 4–5 mo old)	Aged (n = 18, 13–16 mo old)
WT	0.26 ± 0.01	0.29 ± 0.03
Tg	0.25 ± 0.01	0.26 ± 0.02

Values are means ± SE. Both atria were excised and weighed. There is no significant difference between wild-type (WT) and perilipin-2-transgenic (PLIN2-Tg) mice in both young and aged mice or between young and aged mice of both genotypes.

Atrial gene expressions.

To determine whether PLIN2 overexpression would affect atrial gene expression, we analyzed the mRNA expression by RT-qPCR. As shown in Table 2, the expression of genes involved in FA uptake or oxidation and LD metabolism were comparable between WT and Tg mice. In addition, the expression of genes related to apoptosis, fibrosis, or cardiac stress were not altered in either Tg strain compared with WT mice. These results indicate that PLIN2-induced atrial steatosis is relatively benign and mimics the physiological steatosis reported previously in the ventricles of PLIN2-Tg mice (51).

Table 2.

Expression of various genes in atria of WT and PLIN2-Tg mice

Gene	WT	Tg
FA uptake/oxidation
Ppara	1.0 ± 0.08	0.88 ± 0.03
Cd36	1.0 ± 0.04	1.00 ± 0.03
Cpt1	1.0 ± 0.06	1.09 ± 0.08
Acadl	1.0 ± 0.06	1.13 ± 0.08
Acadm	1.0 ± 0.03	1.02 ± 0.04
Acox1	1.0 ± 0.05	0.87 ± 0.04
Ucp2	1.0 ± 0.07	0.99 ± 0.05
Ucp3	1.0 ± 0.24	0.72 ± 0.15
Lipid droplet metabolism
Pnpla2	1.0 ± 0.09	1.09 ± 0.02
Lipe	1.0 ± 0.18	1.13 ± 0.04
Dgat1	1.0 ± 0.08	1.21 ± 0.04
Dgat2	1.0 ± 0.05	1.11 ± 0.10
Apoptosis, fibrosis, cardiac stress
Casp9	1.0 ± 0.09	0.91 ± 0.06
Bcl2	1.0 ± 0.18	1.12 ± 0.30
Cola1a	1.0 ± 0.10	0.85 ± 0.05
Nppa	1.0 ± 0.08	1.01 ± 0.05
Nppb	1.0 ± 0.14	0.73 ± 0.07

Data are means ± SE of 4 or 5 male mice (AD17). WT, wild type; PLIN2-Tg, perilipin-2 transgenic; FA, fatty acid; Acadl, acyl-CoA dehydrogenase long chain; Acadm, acyl-CoA dehydrogenase medium chain; Acox1, acyl-CoA oxidase-1. Pnpla2, adipose triglyceride lipase; Lipe, hormone-sensitive lipase; Nppa, natriuretic peptide type A; Nppb, natriuretic peptide type B. There is no significant difference between WT and Tg mice.

Electrophysiological examination.

To determine whether cardiac steatosis affected cardiac electrophysiological properties, we analyzed the surface ECG in WT and Tg mice. Table 3 shows the ECG parameters in aged WT and AD17-Tg mice. The durations of the P wave and PR, QRS, and QT intervals were comparable between the two strains. However, the RR interval was significantly longer in Tg mice than in WT mice (148 ± 5.7 vs. 126 ± 3.3 ms). As shown in Table 4, ultrasonography showed comparable cardiac function between the WT and Tg mice, as reported previously in younger mice (51). These studies showed slower heart rate in Tg mice than in WT mice and suggested the potential effects of steatosis on cardiac electrophysiology.

Table 3.

Electrocardiographic parameters in male WT and PLIN2-Tg AD17 mice

Genotype	P	PR	QRS	QT	RR
WT (n = 8)	16 ± 1.4	46 ± 1.3	14 ± 0.5	24 ± 1.0	126 ± 3.3
Tg (n = 7)	15 ± 0.6	47 ± 1.9	14 ± 0.4	24 ± 0.8	148 ± 5.7*

Values are means ± SE in ms. WT, wild type; PLIN2-Tg, perilipin-2-transgenic.

^*P < 0.05 vs WT mice.

Table 4.

Echocardiographic parameters of WT and PLIN2-Tg mice

Genotype	HR, beats/min	LVIDd, mm	LVIDs, mm	IVS, mm	LVPW, mm	FS, %
WT	668 ± 6	3.05 ± 0.06	1.44 ± 0.03	0.78 ± 0.02	0.75 ± 0.03	52.4 ± 0.6
Tg	641 ± 16	2.90 ± 0.08	1.38 ± 0.05	0.81 ± 0.03	0.75 ± 0.02	53.0 ± 0.8

Data are means ± SE of 8 male mice aged 15–17 mo (AD17). WT, wild type; PLIN2-Tg, perilipin-2-transgenic; HR, heart rate; LVIDd, left ventricular internal dimension in diastole; LVIDs, left ventricular internal dimension in systole; IVS, interventricular septum; LVPW, left ventricular posterior wall; FS, fractional shortening. There is no significant difference between WT and Tg mice.

Prevalence of persistent AF.

To study whether cardiac steatosis affected AF susceptibility, we induced AF using transesophageal burst pacing in aged WT and Tg mice. As shown in Fig. 2A, Tg mice were prone to AF, and AF tended to persist much longer than in WT mice. As shown in Fig. 2B, 69% of the Tg mice (AD17 + AD19 total) developed sustained (>5 min) AF. The prevalence of sustained AF in each Tg line was similar (71% in AD17 vs. 67% in AD19). In contrast, aged WT mice were relatively resistant to AF, and only 24% of WT mice developed sustained AF. We also tried to induce AF in younger (4-mo-old) mice; however, both WT and AD17-Tg young mice were resistant to AF (WT 12.5% vs. Tg 0%, n = 8 WT and 10 Tg).

Fig. 2.

A: representative electrocardiogram (ECG) by transesophageal electrical burst pacing in male wild-type (WT) and perilipin 2-transgenic (PLIN2-Tg) mice (AD17). Two platinum bipolar electrodes at 1-mm interelectrode distance were advanced into the esophagus to stimulate the epicardial surface of the left atrium and to record the atrial electrogram, respectively. Surface ECG lead II was also recorded. In Tg mice, atrial fibrillation (AF) was easily induced by transesophageal electrical burst pacing, whereas WT mice were resistant to AF. B: prevalence of sustained AF, defined as an irregular atrial rhythm (atrial cycle length <80 ms) persisting longer than 5 min. When AF spontaneously terminated within 5 min, burst pacing was repeated up to 5 times, or experiment ended when AF persisted longer than 5 min; n = 17 (WT), 7 (AD17), 9 (AD19), and 16 (Tg total). *P < 0.05 vs. WT mice by Fisher's exact test.

Cx43 expression.

To explore the mechanism by which PLIN2 overexpression predisposed to AF, cardiac gene expression in ventricular tissues was analyzed by microarray in both WT and Tg mice. The expression levels of ~900 genes were altered by greater than twofold in Tg hearts compared with WT hearts. Pathway analysis using the KEGG database identified significant changes in the gap junction pathway, which includes gga1 (coding Cx43), in PLIN2-Tg hearts.

A literature search of the PubMed database using the keywords AF and gap junction also identified a close relation between gap junction protein Cx43 and AF (7, 14, 39). On the basis of these findings, we focused on Cx43 and analyzed its expression and localization in the following experiments. As shown in Fig. 3A, the mRNA expressions of atrial Cx43 were comparable between WT and Tg mice. The protein expression of Cx43 showed significant variation among each animal in both genotypes and was comparable between WT and Tg mice (Fig. 3B).

Fig. 3.

Atrial connexin 43 (Cx43) expression and localization in male wild-type (WT) and perilipin 2-transgenic (PLIN2-Tg) mice. A: Cx43 mRNA expression levels in atria of WT and Tg mice. Total RNA was extracted from cardiac atria, and mRNA expression levels were determined by RT-qPCR normalized to GAPDH. Values are means ± SE of 5 mice per group (AD17). B: Cx43 protein expression in atria of WT and Tg mice analyzed by immunoblotting. Graph shows densitometric analysis of protein expression normalized to GAPDH. Values are means ± SE of 10 mice per group (AD17). Left: representative immunoblot for Cx43 and GAPDH. C: representative images of atrial sections stained with specific antibody against Cx43. Arrows show Cx43 located in intercalated disks (ICD) in WT atria; arrowheads show lateral localization of Cx43 in Tg atria (AD19). Scale bars, 50 μm. D: representative images of atrial sections examined by confocal microscopy showing Cx43 (green) and N-cadherin (red). N-cadherin staining shows localization in ICD. E: quantification of Cx43 localized in ICD assessed on merged images of confocal microscopy (AD19, n = 4; 10 images per mouse). Values are means ± SE. ***P < 0.001 vs. WT mice. F: representative images of atrial sections stained with specific antibody against Cx43 in atria of young WT and Tg mice (4- to 5 mo old). Arrows show Cx43 located in ICD; arrowheads show lateral localization of Cx43. Scale bars, 20 μm. G. representative image of phosphorylation status of Cx43 analyzed by immunoblotting. Graph shows densitometric analysis of each phosphorylated Cx43 normalized to total Cx43. Values are means ± SE of 4 mice per group (AD17).

Localization of Cx43 expression.

Cx43 is normally located at intercalated disks (ICD) and forms gap junction channels, which play a role in intercellular communications by allowing passage of small molecules or ions (45). The localization of Cx43 is dysregulated in the presence of oxidized stress and in pathological conditions such as diabetes (24, 44). Given this background, we mapped atrial cardiomyocytes for Cx43 by immunostaining. As shown in Fig. 3C, Cx43 was localized in the ICD of WT atria, whereas it appeared heterogeneously distributed on the lateral side of cardiomyocytes in the atria of Tg mice. Quantification of Cx43 localized at ICD by fluorescent immunostaining indicated it was 50% less in Tg atria than in WT atria (Fig. 3, D and E). These results suggest that Cx43 undergoes remodeling in the atria of PLIN2-Tg mice. We also mapped myocardial Cx43 expression in young mice. The immunostaining of Cx43 revealed that, in addition to ICD, a significant amount of Cx43 was detected in the cytosol and the lateral side of cardiomyocytes in both genotypes, although Tg mice appeared to have an increase in lateral localization of Cx43 in atrial cardiomyocytes (Fig. 3F). Since localization of Cx43 is regulated by phosphorylation of Cx43 (15), we analyzed phosphorylated Cx43 by Western blotting. However, the results showed similar levels of phosphorylation of Cx43 in both Tg and WT atria (Fig. 3G).

Assessment of atrial conduction time and optical mapping.

Does Cx43 remodeling affect conduction propagation in PLIN2-Tg atrium? To answer this question, we analyzed atrial conduction velocity in Langendorff-perfused heart preparations by using an optical mapping system. Figure 4, A and B, shows representative examples of activation isochrone maps recorded from the epicardial surface of the left and right atria during steady-state pacing. Activation isochrone mapping demonstrated smooth propagation in the WT atria. In contrast, Tg atria displayed relative crowding of isochrone lines, indicating conduction slowing compared with the WT atria (Fig. 4B). In addition, the Tg atria displayed mosaic-patterned isochrone lines in the left atria. Measurement of the mean ICT on both atria showed significant prolongation in Tg atria compared with WT atria (Fig. 4C).

Fig. 4.

Atrial conduction propagation assessed by optical mapping system. A: image of mapping area (red box) showing epicardial surface of both atria including area of pulmonary veins and location of pacing site. B: representative examples of activation isochrone maps from the epicardial surface of both atria. Activation maps are shown with 1-ms isochrones. Note the mosaic-patterned isochrone lines in left atria of perilipin 2-transgenic (Tg) mouse (arrow, AD17). C: impulse conduction time recorded from the epicardial surface of both atria in male wild-type (WT) and Tg mice (AD17, n = 6). Values are means ± SE. *P < 0.05 vs. WT mice.

Effect of TAG depletion on AF susceptibility.

Is the increase in myocardial TAG the main cause of AF susceptibility? First, we induced AF in double-Tg mice (which overexpress both PLIN2 and HSL specifically in cardiomyocytes, Tg+HSL; Fig. 5A) (51). As shown in Fig. 5B, LDs could not be identified in the atrial cardiomyocytes of Tg+HSL mice despite diffuse expression and dense deposition of PLIN2. Atrial TAG content was lower in Tg+HSL mice than in WT mice (Fig. 5C). In parallel with the lack of LDs and low TAG content, the prevalence of sustained AF was decreased by 61% (although this decrease was not significant) (Fig. 5D). These results suggest that increased lipids rather than PLIN2 protein contribute to AF susceptibility in PLIN2-Tg mice.

Fig. 5.

Effect of triacylglycerol (TAG) depletion on susceptibility to afrial fibrillation (AF). Double-transgenic (Tg) mice with cardiac-specific overexpression of perilipin 2 (PLIN2) and hormone-sensitive lipase (HSL) (Tg+HSL) were created and maintained until 12–15 mo of age, and AF was induced by burst pacing, as described in Fig. 2. A: atrial enhanced green fluorescent protein (EGFP)-PLIN2 and HSL protein expression in wild-type (WT), Tg (AD17), and Tg+HSL mice assessed by immunoblotting. B: lipid droplet (LD) accumulation and EGFP-PLIN2 expression in left atria of WT, Tg, and Tg+HSL mice. Tissue sections were stained with LipidTOX and analyzed by confocal microscopy. Bottom: merged images of those focused on LDs (red) and EGFP (green). Representative micrographs of 4 mice/group (2 males, 2 females) are presented. Scale bars, 10 μm. C: atrial TAG content in WT, Tg, and Tg+HSL mice. Both atria from 6 mice (3 males, 3 females) were pooled and homogenized, and TAG content was measured. D: prevalence of sustained AF, defined as irregular atrial rhythm (atrial cycle lengths <80 ms) persisting longer than 5 min. When AF spontaneously terminated within 5 min, burst pacing was repeated up to 5 times, or experiment ended when AF persisted longer than 5 min; n = 8 (WT), 8 (Tg), and 7 (Tg+HSL). *P < 0.05 vs. WT mice by Fisher's exact test.

Atrial lipid profile.

To explore the lipid species that contribute to the induction of AF, we performed lipidomic analysis by LC-MS using atrial lipid extracts from WT and Tg mice. The analysis identified 413 lipid species in 14 classes. Comparison analysis showed increases greater than twofold in 100 of 113 species of TAG, 9 of 37 species of DAG, and 1 of 37 species of ceramide in Tg atria compared with WT atria (Table 5). Among these lipid species, the 10 most abundant TAGs consisted of long-chain FAs (C16 and C18). Interestingly, most of the top 10 TAGs in fold changes and TAG detected only in Tg atria contained docosahexaenoic acid (DHA, C22:6). Thus, Tg atria showed exclusive accumulation of acylglycerides (TAG and DAG), which have unique FA compositions.

Table 5.

Atrial lipidomic profile of PLIN2-Tg mice

Lipid Class	Lipid Search ID	Fatty Acids	Mass in Tg	Fold Change (Tg/WT)
Ceramide
	87	(m22:1_22:2)	378754	8.9
Diacylglycerol
	294	(18:2_22:6)	561843	16.7
	312	(14:0_18:2)	100362	6.6
	344	(18:2_18:2)	257453	5.1
	384	(18:1_22:6)	478352	3.3
	394	(18:1_14:0)	210063	3.2
	473	(18:1_18:1)	161948	2.9
	330	(22:5_18:2)	60217	∞
	229	(22:6_22:6)	44449	∞
	432	(18:1_20:3)	21229	∞
Triacylglycerol (top 10 in fold change)
	571	(18:1_22:6_22:6)	853037	14.5
	565	(18:2_18:2_22:6)	3415962	12.2
	631	(19:1_18:2_18:2)	265130	11.7
	573	(16:0_22:6_22:6)	1458927	9.2
	592	(18:1_18:2_22:6)	5281189	9.0
	583	(18:1_22:5_22:6)	297827	8.4
	563	(16:0_18:3_22:6)	924937	7.8
	594	(16:0_18:2_22:6)	5212056	7.8
	582	(16:1_18:2_22:5)	365472	6.5
	624	(18:0_18:2_22:6)	3478271	6.3
Triacylglycerol (top 10 in mass)
	671	(16:0_18:1_18:1)	29819776	2.6
	648	(18:1_18:1_18:2)	28770649	3.2
	643	(16:0_16:1_18:1)	27181828	2.7
	616	(16:1_18:1_18:2)	26439285	3.2
	673	(18:0_18:1_18:2)	23154034	2.8
	618	(18:1_18:2_18:2)	21878717	4.0
	677	(16:0_16:0_18:1)	16677374	2.4
	612	(16:1_16:1_18:1)	13162425	3.0
	591	(18:2_18:2_18:2)	8046123	5.9
	699	(18:0_18:1_18:1)	7317438	3.4
Triacylglycerol (∞, top 10 in mass)
	557	(22:5_18:2_22:6)	346342	∞
	548	(20:5_18:2_22:6)	159133	∞
	669	(16:0e_18:1_22:6)	140318	∞
	545	(22:6_22:6_22:6)	133380	∞
	561	(16:0_20:5_22:6)	120380	∞
	291	(20:2e_10:3_10:3)	108048	∞
	547	(14:0_22:6_22:6)	101767	∞
	690	(19:1_18:2_20:1)	92513	∞
	642	(14:1e_20:1_22:6)	90475	∞
	696	(16:0e_20:1_22:6)	82404	∞

∞Detected only in transgenic (Tg) mice. WT, wild type. One hundred triacylglycerol (TAG) species were extracted, and the top 10 in each category are presented.

DISCUSSION

The present study examined the impact of intracellular LD accumulation on AF by using a mouse model of cardiac steatosis, the PLIN2-Tg mouse. To date, only a few studies have addressed the effects of intracellular lipotoxicity on arrhythmia. For example, Son (46) and Morrow et al. (31) reported the presence of cardiac steatosis, Cx43 downregulation, slow conduction velocity, and ventricular tachycardia in transgenic mice with cardiac peroxisome proliferator-activated receptor-γ (PPARγ) overexpression. Although these studies highlighted the crucial role of cardiac steatosis in ventricular arrhythmia, the specific effect of cardiac steatosis on arrhythmia remains obscure, since PPARγ modulate several genes, including PLIN2 (9) and Cx43 (41), which could affect arrhythmogenicity in cardiomyocytes. In contrast, the use of PLIN2-Tg mice in the present study allowed us to analyze the specific effect of intracellular accumulation of LDs on arrhythmogenicity, since PLIN2 induces LD formation by limiting access of cytosolic lipases (25), resulting in benign and physiological steatosis (51). In addition, other systemic arrhythmogenic factors, including hyperglycemia, hyperlipidemia, and obesity, are absent in PLIN2-Tg mice. Thus, the PLIN2-Tg mouse appears to be a suitable mouse model for the study of the link between cardiac steatosis and AF.

It is important to note that the young (4-mo-old) PLIN2-Tg mice did not show significant AF susceptibility relative to WT mice. This finding indicates that steatosis is not sufficient to induce AF but that it does so with some additional factors related to aging. Indeed, several clinical studies, including the Framingham Heart Study (26), reported the high prevalence of AF in old age. Furthermore, experimental studies using rodents have shown that atrial fibrosis is a key risk factor for aging-related susceptibility to AF (53). In the present study, we did not detect significant interstitial fibrosis in the atria of the mice studied. However, potential aging-related structural remodeling or molecular alterations might underlie the atria of the mice. Aging is also reported to correlate with downregulation of Cx43 in atria (18). This decline may play a role in the enhanced susceptibility of aged PLIN2-Tg mice to AF observed in the present study. Other unknown aging-related factors for AF are of great interest and are targets for further investigation.

Structural and electrical remodeling are key factors in the pathogenesis of AF. Remodeling is induced by atrial inflammation, activation of RAS, calcium overload, ROS production, and transforming growth factor-β signaling (16, 35). On the basis of this background, we investigated the features of structural remodeling, including interstitial fibrosis, inflammation, and apoptosis of cardiomyocytes by use of light- and electron microscopy. However, there was no evidence for any of these processes in the atria of both PLIN2-Tg and WT mice. These results are consistent with those of a previous study (51), which demonstrated the presence of severe PLIN2-induced steatosis in the ventricles, although such finding appeared to be benign and represented physiological LD accumulation.

Despite the absence of structural remodeling, electrical remodeling was evident in our study, characterized by lateral localization of Cx43, slower electrical current (CV), and heterogeneous conduction propagation in the atria of aged PLIN2-Tg mice (Fig. 4). It has been reported that downregulation of Cx43 results in altered ion current and slow CV (22, 32) and that gap junction remodeling and lateral distribution of Cx43 are crucial for the pathogenesis of AF (50). On the basis of these findings and the reports showing that Cx43 gene transfer preserves CV and prevents AF (7, 14), we suggest that the observed electrical remodeling is the major cause of AF in PLIN2-Tg mice.

Localization of Cx43 is regulated by phosphorylation of Cx43, which is coordinately regulated by protein kinases PKA and PKCs (mainly PKCε). The activities of these protein kinases are affected by various hormones, cytokines, and conditions (15). For instance, epinephrine increases intracellular cAMP level, and the activated PKA phosphorylates Cx43, facilitating Cx43 at gap junctions. Conversely, angiotensin II or chronic hyperglycemia and resultant DAG accumulation activates PKCs, which phosphorylates distinct sites of Cx43 (e.g., Ser³⁶⁸ in rats), directing it to the lateral side of cardiomyocytes (29, 45). To assess the phosphorylation status of Cx43, we also performed Western blot analysis using specific antibodies against phosphorylated Cx43. However, the results did not show a significant difference in phosphorylated Cx43 between WT and Tg atria, probably due to the large variation in total Cx43 expression among the mice studied (data not shown). Nevertheless, since several DAG species were increased in the atria of PLIN2-Tg mice (Table 5), activation of PKC and resultant Cx43 phosphorylation might have occurred in the atria of Tg mice. This possibility remains to be clarified.

End-binding 1 (EB1), a microtubule-binding protein, plays an important role in Cx43 trafficking to ICD through the microtubules (44). EB1 interacts with desmoplakin, a desmosome protein, and regulates Cx43 localization (34). The function of this EB1 is reported to be inhibited by oxidative stress (44). To test whether EB1 is affected by PLIN2-induced steatosis, we analyzed EB1 expression and localization by immunostaining. However, we could not detect any difference between the atria of Tg and WT mice (data not shown). Thus, the impact of LD accumulation on EB1 function remains to be determined.

LDs were reported recently to provide lipid signals that regulate many aspects of cell physiology. For instance, Haemmerle et al. (11) reported that FAs released from intracellular LDs function as endogenous ligands for PPARα. Our lipidomic analysis showed exclusive increases in TAG, DAG, and ceramide in Tg atria. The results of the PLIN2-HSL double-Tg mouse confirmed the contribution of LDs on the induction AF. Considered together, the above findings suggest that the lipid signals released from LDs during aging could play a role in the increased susceptibility to AF. In particular, the lipidomic analysis showed exclusively an increase of DHA-containing TAG in the atria of Tg mice (Table 5). Considering the reports showing that ω3 polyunsaturated FA restore Cx43 abnormality in diabetic rat hearts (2, 36), the packaging of DHA in LDs might have affected Cx43 localization in the PLIN2-Tg atria. The precise mechanism remains to be clarified in our future investigation.

The lipidomic analysis also showed an ~30% increase in phosphatidylcholine in Tg atria (data not shown). Such an increase might cause significant changes in membrane phospholipid (PL) composition that could affect Cx43 localization in the Tg heart (27). Since LDs are coated with a PL monolayer, excess LD formation might alter PL composition of the plasma membranes, where signaling lipids, ion channels, and gap junctions are mounted.

Other potential mechanisms for Cx43 remodeling might be physical interference of LDs with microtubules in Cx43 trafficking, or interference in energy supply for Cx43 trafficking, because PLIN2 possesses a barrier function against cytosolic lipases (25). The potential effect of overexpression of PLIN2 protein on Cx43 remodeling should be considered also, because both PLIN2 and Cx43 have been shown to be degraded by the ubiquitin-proteasome pathway (23, 54). Overload of the proteasome pathway might have affected the quality maintenance of Cx43. These possibilities remain the subject of future research.

In conclusion, the present study has demonstrated that PLIN2-induced atrial steatosis is associated with Cx43 remodeling, aberrant conduction propagation, and increased susceptibility to AF. The results highlight a new role for LDs in the pathogenesis of AF and the potential therapeutic benefits of targeting atrial steatosis in the prevention of AF in elderly individuals with metabolic disorders.

GRANTS

This work was supported in part by a research grant from the Ministry of Education, Culture, Sports, Science and Technology in Japan 18K08471 (to J. Suzuki), Merit Review Award from the Department of Veterans Affairs Biomedical Laboratory Research and Development Service I01BX000398 (to F. B. Kraemer), and National Institute of Diabetes and Digestive and Kidney Diseases Grant P30 DK-116074 (to F. B. Kraemer).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

J.S. and M.H. conceived and designed research; S.S., J.S., M.H., and S.I. performed experiments; S.S., J.S., M.H., M.Y., Y.Z., S.T., S.I., and T.K. analyzed data; S.S., J.S., M.H., M.Y., Y.Z., T.N., M. Ichikawa, M. Imagawa, S.T., S.I., T.K., F.B.K., and T.I. interpreted results of experiments; J.S. and M.H. prepared figures; S.S., J.S., M.H., and F.B.K. drafted manuscript; S.S., J.S., M.H., and F.B.K. edited and revised manuscript; S.S., J.S., M.H., M.Y., Y.Z., T.N., M. Ichikawa, M. Imagawa, S.T., S.I., T.K., F.B.K., and T.I. approved final version of manuscript.