BMAL1 regulates mitochondrial fission and mitophagy through mitochondrial protein BNIP3 and is critical in the development of dilated cardiomyopathy

To see whether the circadian clock gene Bmal1 plays a role in cardiac function, we generated mice with global deletion of the Bmal1 gene (Fig. S1) and examined their heart morphology and function over time. Mice with Bmal1 KO showed measurable weight loss and heart rate reduction during their growth (Fig. S2A and S2B). At 32 weeks of age, histological analyses of cross sections of the heart at the level of papillary muscles showed that Bmal1 deficient mice displayed a larger left ventricular cavity plus thinning of left ventricular posterior wall and interventricular septum (Fig. 1A and 1B). The myocyte cross-sectional area was also increased in Bmal1 KO mice when compared to that of the wild type mice (Fig. 1C and 1D), indicating dilation of the myocardium and the ventricle. In addition, transmission electron microscopy showed that Bmal1 KO mice exhibited disorganization and compaction of sarcomeres. The sarcomeric Z-lines were also become condensed and wider (Fig. 1E).

From 12-weeks of age, we assessed heart functions of wild type and Bmal1 KO mice by echocardiography at four-weeks intervals (Fig. 1F). Bmal1 KO mice showed a significantly increased left ventricular end systolic internal diameter (LVIDs) at 32 weeks compared to that in wild type mice (Fig. 1G). The interventricular septum thickness at end-systole (IVSs) and left ventricular posterior wall thickness at end-systole (LVPWs) also gradually became thinner in Bmal1 KO mice than those in wild type mice (Fig. 1H and 1I). Increased left ventricular internal dimension at end-diastole (LVIDd), decreased interventricular septum thickness at end-diastole (IVSd), and left ventricular posterior wall thickness at end-diastole (LVPWd) were also observed in Bmal1 KO mice (Fig. 1J–L). Further, Bmal1 KO mice showed impaired cardiac contractility with a decreased fraction of shortening (FS) (Fig. 1M) and ejection fraction (EF) (Fig. 1N). Overall, Bmal1 KO led to a dramatic ventricular dilation and a ~50% reduction in cardiac contractility in mice at later age. These results are consistent with previous reports showing Bmal1 deficiency induced age-associated DCM in mice (Lefta et al., 2012).

To further assess the role of BMAL1 in human cardiomyocytes, we knocked out BMAL1 expression in H7 hESCs by CRISPR/Cas9 mediated genome editing (Fig. 2A). Indel mutations were introduced within exon 10 of BMAL1 by CRISPR/Cas9 nickase, which led to frameshifts and creation of a premature stop codon in BMAL1 locus (Fig. S3A–C). Western blot results indicated that BMAL1 protein expression was knocked out in the selected positive clones (Fig. 2B). In order to confirm whether both alleles of the BMAL1 gene were disrupted, we performed TA cloning of the amplified PCR products of the target cutting region and it showed that those positive clones were all homozygous. The two alleles of the clone 18 had the same genetic modification (Fig. S3D), while the others had two different genetic mutations (data not shown). The established BMAL1 KO hESC lines exhibited normal ES-cell-like morphologies and pluripotency nature (Fig. S4). Although BMAL1 KO hESCs spontaneously differentiated into the three germ layers in teratoma assays, an upregulation of the mesodermal, cardiac progenitor, and endodermal genes were observed (Fig. S5).

We further differentiated these BMAL1 KO hESCs into beating cardiomyocytes (Supplementary Movie 1 and 2) (Lian et al., 2012). The differentiation efficiency regularly reached greater than 90% (Fig. 2C). To evaluate the impact of BMAL1 KO on cardiomyocyte morphology and myofilament structure, immunofluorescence stainings were performed and it showed that BMAL1 KO hESC-cardiomyocytes exhibited abnormal sarcomeric cardiac troponin T (cTnT) and α-Actinin distribution (Fig. 2E). Assessing ultra-structures using transmission electron microscopy further showed that wildtype hESC-cardiomyocytes exhibited well organized sarcomeres and regular Z-lines, while BMAL1 KO hESC-cardiomyocytes displayed severe myofilament disorganization and irregular sarcomeric Z-lines (Fig. 2D). We also assessed the cell size of both widetype mice and BMAL1 KO hESC-cardiomyocytes. The average cell area is significantly larger in BMAL1 KO hESC-cardiomyocytes (wildtype hESC-cardiomyocytes, 3,370 ± 133.6 μm²; BMAL1^−/− hESC-cardiomyocytes, 6,965 ± 396.3 μm², n = 85 per group, P < 0.001) (Fig. 2F). Compared with wild type hESC-cardiomyocytes, the percentage of BMAL1 deficient hESC-cardiomyocytes with disorganized sarcomeres increased significantly (Fig. 2G). In addition, BMAL1 deficiency led to a significant increase in both early apoptosis and late apoptosis according to flowcytometry analysis (Fig. 2H–K). These data indicated that BMAL1 KO had a great impact on myofilament organization and morphology of human cardiomyocytes.

Our results showed that Bmal1 knockout mice exhibited reduced heart contractility and cardiomyocyte sarcomeric disruption. To assess whether BMAL1 KO has an impact on contractility of human cardiomyocytes, we next measured the contraction force of single BMAL1 KO hESC-derived cardiomyocytes. We found that single BMAL1 KO hESC-cardiomyocytes exhibited weakened contractile force generation per cell movement and irregular beating rhythm when compared with wildtype control hESC-cardiomyocytes (Fig. 3A–C).

Previous studies on Ca²⁺ homeostasis showed that calcium handling plays an important role in excitation-contraction coupling and subsequent force generation of cardiomyocytes (Cheng et al., 1993; Bers, 2002; Eisner et al., 2017). Myocardium from failing hearts also showed defective excitation-contraction coupling which contributes to impaired contractility (Gomez et al., 1997; Yano et al., 2000; Piacentino et al., 2003). We then explored the Ca²⁺ handling property of BMAL1 KO hESC-cardiomyocytes by calcium imaging. The results showed that BMAL1 KO hESC-cardiomyocytes exhibited more irregular Ca²⁺ transients compared with those of wild type control hESC-cardiomyocytes (Fig. 3D and 3G). Further, BMAL1 KO hESC-cardiomyocytes showed a reduction in Ca²⁺ transient amplitude, indicating that BMAL1 deficiency weakened myofilament Ca²⁺ sensitivity (Fig. 3E). BMAL1 KO hESC-cardiomyocytes also showed a slower peak to peak time, indicating a significantly elongated beating activity (Fig. 3F). There was no significant difference in the relative time to peak and Ca²⁺ decay kinetics between wildtype and BMAL1 KO hESC-cardiomyocytes (Fig. 3H–J). These data indicated that BMAL1 KO impaired contractility and calcium handling of hESC-cardiomyocytes.

Previous studies revealed that sympathetic nervous system plays a key role in regulating cardiac physiology(Hoover et al., 2004) by modulating the heart rate, contractility (Li et al., 2000; Shan et al., 2010), and structure of myocytes (Kanevskij et al., 2002; O’Connell et al., 2003; Kreipke and Birren, 2015). To evaluate the effect of BMAL1 deficiency on cardiomyocytes to neuro-hormonal regulation, non-invasive microelectrode arrays (MEAs) was utilized to analyze pharmacologic effects of heart-related drugs on hESC-cardiomyocytes. The drugs that can influence sympathetic nervous system, for instance adrenergic catecholamine norepinephrine (NE) and beta-adrenergic receptor agonist metoprolol (Mtl), were used to measure cardiac electrophysiological characteristics. Wildtype or BMAL1 KO hESC-cardiomyocytes day 30 post differentiation were seeded on MEA probe and the monolayer cardiomyocytes on the MEA surface began to beat spontaneously after 24 h in culture. Synchronous and stable traces for electrophysiological parameters were recorded on day 35 (Fig. 4A). Parameters of MEAs used for the evaluation were characterized by beating rate (Fig. 4B). Irregular beating activities were observed in the onset of BMAL1 KO hESC-cardiomyocytes (Fig. 4C). Moreover, baseline electrophysiological parameters showed that the average beating frequency of BMAL1 KO hESC-cardiomyocytes was much lower than that of the wild type hESC-cardiomyocytes (Fig. 4D).

Mitochondrial dysfunction is considered to be a critical mechanism underlying cardiovascular diseases (Wallace, 2013; Murphy et al., 2016; Friederich et al., 2018). Abnormal mitochondrial morphology was observed in some mitochondrial diseases and idiopathic cardiomyopathies (McManus et al., 2019) . To see whether BMAL1 ablation causes abnormality in mitochondrial function, we analyzed the ultrastructure of day 35 BMAL1 KO hESC-cardiomyocytes by transmission electron microscopy. Mitochondria of BMAL1 KO hESC-cardiomyocytes were extensively elongated, with abnormal cristae structures and an increased mitochondrial area/perimeter ratio (Fig. 5A–C). The most common intra-mitochondrial cristae malformations were circle cristae, partitioning cristae, swollen cristae, and cristolysis (Fig. S6). These ultrastructural mitochondria morphologies were consistent with previous findings (Jacobi et al., 2015; McManus et al., 2019). Immunostaining of BMAL1 KO hESC-cardiomyocytes with adCox8a-RFP showed that mitochondria were highly hyper-fusion compared to those of wildtype hESC-cardiomyocytes (Fig. 5D, and Supplementary Movie 3 and Movie 4). Likewise, expression level of MFN2, the marker of mitochondrial fusion, was significantly increased in BMAL1 KO hESC-cardiomyocytes (Fig. 5E and 5F). Similar findings were also present in BMAL1 KO mice (Fig. 6A–E). Taken together, these results indicated that Bmal1 deficiency led to hyper-fused mitochondria with abnormal cristae in cardiomyocytes.

A previous study demonstrated that the heart is one of the sturdiest mitophagic organs (Kanevskij et al., 2002). Mitophagy plays a crucial role in mitochondrial quality control, by removing dysfunctional, damaged, or aging mitochondria (Song et al., 2015; Song et al., 2017). Transmission electron microscopy imaging showed decreased mitophagy events in BMAL1 KO hESC-cardiomyocytes (Fig. 5G and 5H). We thus further examined mitophagy intensity in wildtype and BMAL1 KO hESC-cardiomyocytes using Mtphagy Dye (Iwashita et al., 2017). Consistent with transmission electron microscopy imaging, knocking out BMAL1 resulted in a decline in mitophagy intensity (Fig. 5I and 5J). Mitochondrial localization of the autophagy chaperone protein SQSTM1 can mediate autophagy of damaged mitochondria by binding to autophagosomal microtubule associated protein1 light chain 3 (LC3) (Narendra et al., 2010). We found that mitochondrial SQSTM1 and LC3 II protein level were also decreased in BMAL1 KO hESC-cardiomyocytes (Fig. 5K and 5L). Similar results were found in Bmal1 KO mice (Fig. 6F–I,). In addition, mitochondrial autophagy-associated protein BNIP3 expression was reduced in myocardium of Bmal1-deficient mice (Fig. 6H–L). These data indicated that BMAL1 ablation caused mitochondria abnormalities in and reduced mitophagy of cardiomyocytes. We further found that mitochondrial oxidative metabolisms of BMAL1 KO hESC-cardiomyocytes were significantly declined (Fig. 5M and 5N). Overall, our data showed that BMAL1 loss-of-function induced dysfunction and diminished respiration in mitochondria, suggesting that BMAL1 deletion impacted the dynamic remodeling of the mitochondrial network.

According to genome-wide transcript profiling, BMAL1 is estimated to target more than 150 sites in the human genome, including all known clock genes and genes regulating metabolism (Hatanaka et al., 2010) . BMAL1 is a core circadian transcription factor and regulates downstream genes by binding to the E-box elements in their promoters (Ueda et al., 2005). The transcription of Bnip3 gene exhibits a diurnal rhythm in C57 BL/6J livers (Jacobi et al., 2015). We also identified several canonical E-box sequences (CANNTG), and one noncanonical E-box sequence (CAGCTT) in the promoter region of Bnip3 gene (Table. 1). In addition, BMAL1 is able to bind to the promoter region of Bnip3 in osteosarcoma according to ChIP-seq analysis (Rey et al., 2016) (Fig. 7A). Since BNIP3 is a critical player in mitophagy by mediating autophagy of mitochondria and is located in myocardium (Azad et al., 2008; Chaanine et al., 2013; Xiao et al., 2018), we further examined BNIP3 protein expression in wild type and BMAL1 KO hESC-cardiomyocytes at day 35. Western blot confirmed that BNIP3 protein level was reduced in BMAL1 KO hESC-cardiomyocytes (Fig. 7B and 7C). The mRNA expression of BNIP3 in BMAL1 KO day 35 hESC-cardiomyocytes was also decreased (Fig. 7D). We therefore hypothesized that BMAL1 is responsible for the transcriptional activation of BNIP3 by binding to the E-box elements within its promoter region in cardiomyocytes. ChIP-qPCR assays showed that there was a BMAL1 binding-site from 0 to 85 bp in the promoter region of BNIP3 gene (Fig. 7E). We further used dual-luciferase reporter assays to evaluate the transcriptional activation of BMAL1 on BNIP3. As shown in Fig. 7F, wild type BNIP3 promoter facilitated highest transcriptional activity to luciferase gene along with the co-transfection of BMAL1 and CLOCK. However, E-box mutant BNIP3 promoter was not able to sufficiently trigger the transcription of reporter gene with the exogenous overexpression of BMAL1 and CLOCK. These results indicated that BMAL1 regulates the transcription of BNIP3 gene via binding with the E-Box element within its promoter region (Fig. 8).

The human embryonic stem cell line H7 was cultured in mTeSR (STEMCELL TECHNOLOGIES, #85851 and #85852). HESC-H7 cells were isolated from matrigel by accutase digestion and reseeded on the new matrigel at 2–6 × 10⁵ cells per 12 well in mTeSR. We utilized the protocols originally developed by Lian et al (Lian et al., 2012) to achieve high-purity cardiomyocytes in vitro. Briefly, on the day of cardiac induction (day 0), 6 μmol/L CHIR99021 (http://www.selleckchem.com↗) was added for 2 days in 1 mL of insulin-minus RPMI + B27 medium. At day 2, cells were recovered in insulin-minus RPMI + B27 for 24 h. At day 3, cells were treated with 5 μmol/L IWR-1 (Sigma) for 2 days to activate Wnt signaling. At day 5, cells were finally switched to RPMI + B27 plus insulin medium. Beating cells were observed at day 7–8 after differentiation. The mature CMs were cultured in in DMEM containing 10% fetal bovine-serum. All these cells needed to be cultured on Matrigel-coated plates (corning #354248) and maintained at 37 °C with 5% CO₂.

BMAL1 gene mutation was corrected back to human embryonic stem cell line 7 (hESC-H7), using low off-target CRISPR dual nickase plasmid (PX462). Guide-RNA sequences (gRNA-A: 5′:AATCCATCTGCTGCCTAATCAGG; gRNA-B: 5′:TTGTCGTAGGATGTGACCGAGGG) were designed at http://crispr.mit.edu/↗ and cloned as described at http://www.genome-engineering.org/crispr/↗. The editing efficiencies of CRISPR plasmids were validated with 293T cells using Sanger sequencing. 4.6 × 10⁵ cells of hESC-H7 were electroporated using Neon Transfection System(Thermo Fisher Scientific) with 3 μg of the most efficient plasmids. Successful electroporation cells were puromycin (0.22 μg/mL) selected after 1 day. Single selected positive clones were picked and verified with Sanger sequencing using TA cloning technology to measure allelic frequency. We derived BMAL1 gene mutation cell line during the CRISPR-editing process.

Human EB were generated as described previously(Ang et al., 2016). In ultra-low attachment plates (Corning), embryoid bodies can be formed in specific medium (DMEM, 20% FBS, 1% non-essential amino acid, 100 μmol/L BME, 1mmol/L L-glutamine) for 10 days followed by quantitative real-time PCR for evaluating gene expression of endoderm, mesoderm, ectoderm, cardiac progenitor, and mature cardiomyocytes.

All animal procedures conformed to the National Institutes of Health Guidelines for the Use of Laboratory Animals were approved by Fudan University’s Animal Care and Use Committee guidelines. The germLine Bmal1^−/− mice were imported from Jackson Lab (Bunger et al., 2000). Mice were kept on a 14-h:10-h light-dark schedule and had access to food and water. Mice were anaesthetized using 1%–2% isoflurane and euthanized by cervical dislocation.

The beating cardiomyocytes were digested with 1 mg/mL collagenase I (Sigma) from petri dishes, and followed by 0.25% Trypsin/EDTA (Gibco) to further isolate the cells. Then the isolated cells were fixed and infiltrated by permeabilization solution (BD Biosciences). The cells were stained for cardiac troponin T (cTnT, Thermo Scientific) for 2 h at 4 °C. The cells were incubated with the secondary antibody for 1 h before washing cells with PBS. Finally, the fluorescence-activated cell detection was performed on the FACSCalibur (BD Biosciences). The antibody information could be found in Table 2. Data was analyzed using FlowJo software.

Cardiomyocyte mitophagy was detected by Mitophagy Detection Kit (Dojindo). Cardiomyocytes were cultured in 37 °C with 100 nmol/L Mtphagy dyes for 30 min. Cells were wished 2 times with PBS, and then treated with 10 µmol/L CCCP for 8 h prior to induce mitophagy. Cells were suspended into single cells and subsequently subjected to FCM to determine the flourecence intensity at 700 nm.

The mitochondria of cardiomyocytes were visualized using adenovirus carrying Cox8a-RFP,which has sequences of mitochondrial matrix localization derived from Cox8a. The differentiated 30-day CMs were digested into single cells and were inoculated in a laser confocal dish at a suitable density for four days with DMEM + 10% FBS. The CMs were then cultured in DMEM + 10% FBS containing Cox8a-RFP for 24 h. After 24 h, the original liquid was discarded and replaced with DMEM + 10% FBS. The mitochondria of cardiomyocytes were photographed within 24 h.

The mitochondrial respiratory capacity of cardiomyocytes was determined by Seahorse XFe96 Analyzer (Agilent Technologies). The differentiated 34-day CMs were plated in 96-well plates at 2 × 10⁴ cells/well at 37 °C for 24 h. The cells were incubated in DMEM containing 0.584 g/L L-glutamine and 5.5 mmol/L D-glucose for 1 h at 37 °C before analysis. 0.5 mg/mL oligomycin, 1 mmol/L FCCP, 1 mmol/L rotenone and 2 mmol/L antimycin A were sequentially added into each assigned well at indicated time point.

The calcium flux was determined by laser confocal Ca²⁺ imaging technique to evaluate intracellular calcium transient. The autonomous beating cardiomyocytes labeled by 5 μmol/L Cal-520AM and 0.02% Pluronic F-127 (AAT Bioquest) were detected by LSM-710 laser scanning confocal microscope (Carl Zeiss) with a detection time of 10 milliseconds under the line scanning mode in Zen software. Zen software was used to record calcium transient and MATLAB software (MathWorks) was performed to analysis calcium imaging signals.

The cardiomyocytes were digested at 32 days after induction and then inoculated to the surface of the coated MEAs probe. The myocardial cell field potential waveforms were recorded using a MEA2100 system at 35 days post induction. A bright-field signal video was recorded using a heated MC-rack to ensure that the cells were maintained at 37 °C. The data of field potential waveforms were quantitatively analyzed by using the Spike2 7.19 software (CED, UK).

2–3 days before the measurement, the cardiomyocytes were digested and planted on a confocal dish coated with Matrigel. The contractility of cardiomyocytes was detected by using video-based motion edge detection system. Through a Zeiss CFM-500 inversion fluorescence microscope and Video Sarcomere Length software (900B: VSL, Aurora Scientific), beating cardiomyocytes were clearly projected onto the screen. The spontaneous contraction traces of cardiomyocytes were recorded by using data analysis software (FelixGX, PTI). During the testing process, these cells were kept in 37 °C using heating equipment, and only the cardiomyocytes of the digestive single were analyzed.

The pGL3 Basic plasmid was digested with two restriction endonucleases Hind III and Nhe I. Subsequently, the 2,000 bp fragment, the 363 bp fragment and the mutated 362 bp fragment of the 5′ upstream sequence of the Bnip3 gene were cloned into the pGL3 basic vector (Promega, Madison, WI). The open reading frame (ORF) of the Clock gene was cloned to pcDNA 3.1 by NotI and Apa 1 endonuclease. The ORF of the BMAL1 gene was cloned to pCMV using BamHI and EcoRI. The plasmid constructed above was added to the DH5α to be transformed and cultured on ampicillin agar plate. The positive clones were selected for amplification, and the plasmid was extracted by using MN large pumping kit, and the concentration was measured. The existence of the CLOCK, BMAL1 and BNIP3 genes were confirmed by sequencing following plasmid extraction.

The following plasmids were instantly transfected into the HEK293T cells in the 6-well plates by means of LipofectamineTM 2000 regent (Invitrogen,USA) according different distribution combinations. These plasmids included 500 ng of pGL3-BNIP3 plasmid, 100 ng of pRL-TK plasmid, 500 ng pcDNA 3.1, 500 ng pCMV, 500 ng BMAL1 plasmid, and 500 ng CLOCK plasmid. The cells were harvested after 24 h of transfection and the activity of firefly and renilla luciferase was detected by the Dual Luciferase Reporter Gene Assay Kit (Beyotime, China). The firefly luciferase activity was corrected by the renilla luciferase activity.

ChIP was performed as described previously (Lan et al., 2007). The whole cell extracts were sonicated and immunoprecipitated with antibody of BMAL1 (CST). Real-time quantitative PCR was used to analyze the precipitated DNA samples.

For statistical analysis, all data were expressed using mean ± SD. Student’s t test was used for comparison between the two groups. The statistical significance of differences between the experimental groups and several timepoints was determined using 2-way ANOVA with post-hoc test. P-values less than 0.05 was considered to be statistically different. All experimental results were analyzed using SPASS 17.0. GraphPad Prism 5 and Adobe illustrator software were used for analysis and graphing.

Below is the link to the electronic supplementary material.

BMAL1, brain and musle ARNT-like 1; Bnip3, BCL2/adenovirus E1B interacting protein 3; CMs, cardiomyocytes; DCM, cTnT, cardiac muscle Troponin T;dilated cardiomyopathy; EB,embryoid body; hESCs, human embryonic stem cells; IVSs, interventricular septal thickness at end systole; LVIDs, left ventricular internal dimension at end systole; LVPWs, left ventricular posterior wall thickness at end systole; MEAs, multi-electrode arrays; TEM, transmission electron microscopy; WGA, wheat germ agglutinin

Ermin Li, Xiuya Li, Jie Huang, Chen Xu, Qianqian Liang, Kehan Ren, Aobing Bai, Chao Lu, Ruizhe Qian, and Ning Sun declare that they have no conflict of interest.

All procedures followed were in accordance with the ethical standards of the responsible committee on human experimentation (Fudan University, China) and with the Helsinki Declaration of 1975, as revised in 2000 (5). Informed consent was obtained from all patients for being included in the study.

All institutional and national guidelines for the care and use of laboratory animals were followed.