The mechanical effects of CRT promoting autophagy via mitochondrial calcium uniporter down‐regulation and mitochondrial dynamics alteration

Abstract The mechanism of cardiac resynchronization therapy (CRT) remains unclear. In this study, mitochondria calcium uniporter (MCU), dynamin‐related protein‐1 (DNM1L/Drp1) and their relationship with autophagy in heart failure (HF) and CRT are investigated. Thirteen male beagle's dogs were divided into three groups (sham, HF, CRT). Animals received left bundle branch (LBB) ablation followed by either 8‐week rapid atrial pacing or 4‐week rapid atrial pacing and 4‐week biventricular pacing. Cardiac function was evaluated by echocardiography. Differentially expressed genes (DEGs) were detected by microarray analysis. General morphological changes, mitochondrial ultrastructure, autophagosomes and mitophagosomes were investigated. The cardiomyocyte stretching was adopted to imitate the mechanical effect of CRT. Cells were divided into three groups (control, angiotensin‐II and angiotensin‐II + stretching). MCU, DNM1L/Drp1 and autophagy markers were detected by western blots or immunofluorescence. In the present study, CRT could correct cardiac dysfunction, decrease cardiomyocyte's size, alleviate cardiac fibrosis, promote the formation of autophagosome and mitigate mitochondrial injury. CRT significantly influenced gene expression profile, especially down‐regulating MCU and up‐regulating DNM1L/Drp1. Cell stretching reversed the angiotensin‐II induced changes of MCU and DNM1L/Drp1 and partly restored autophagy. CRT's mechanical effects down‐regulated MCU, up‐regulated DNM1L/Drp1 and subsequently enhanced autophagy. Besides, the mechanical stretching prevented the angiotensin‐II‐induced cellular enlargement.


| INTRODUC TI ON
Heart failure (HF) is associated with high risk of mortality and morbidity. The onset and progress of HF are closely related to mitochondria injury. 1 The intact mitochondrial double-layer membrane possesses the ionic selectivity, particularly, for calcium.
Calcium-overload in mitochondrial matrix leads to increased membrane permeability and dissipated membrane potential, and subsequent cell death is inevitable. 2 The mitochondrial calcium uniporter (MCU) localized at the inner membrane of mitochondria confers the high selectivity of calcium. From the current perspective, MCU is the most important calcium channel mediating inward calcium flux into the mitochondrial matrix. 3 Consequently, MCU determined mitochondrial calcium concentration and even buffered cytosolic calcium. 2,4,5 The cytosolic calcium can influence the mitochondrial fission by the dynamin-related protein 1 (DNM1L/ Drp1), 5 and such alteration of mitochondrial dynamics could further change the autophagy and mitophagy. 6 It is reported that the level of autophagy is changed in HF, and the elevated autophagy is believed to be an important mechanism to degrade unnecessary cellular component. This process permits the sequential degradation and recycling of cellular components. 6,7 Autophagy can be influenced on condition of MCU function or expression anomaly. 8,9 Besides, impairment of autophagy in company with the suppressed DNM1L/Drp1 and mitochondrial fission in HF indicates that alteration of mitochondrial dynamics and autophagy should take part in the compensatory mechanism in HF. 6 Additionally, variation of mitochondrial dynamics could further influence the mitophagy which is important to maintain cardiomyocyte integrity. [10][11][12] In terms of HF treatment, recent years have witnessed great progression in cardiac devices development, especially the cardiac resynchronization therapy (CRT), an established biventricular-pacing method to effectively improve cardiac function. 13 CRT activates myocardium based on electromechanical coupling theory, 14 and its mechanical effects are believed to correlate with cardiac function improvement. 15 Although CRT shows impressive performance in clinical practice, the molecular mechanism of CRT still remains unknown. Considering a number of patients do not respond to CRT, 13,16 further study on the mechanism of CRT is helpful to deeply understand its principle and offer solution for improving the response rate of CRT. Consequently, distinct illumination of the mechanism of CRT on molecular level is important.
In the present study, a canine model is introduced to investigate the CRT's effects on HF and its relationship with MCU and autophagy in vivo. To some extent, the ultimate effect of CRT on heart is the mechanical stretch of the ventricular wall. Therefore, a cell stretching culture method is applied to imitate CRT's mechanical effect on cardiomyocytes in vitro. In general, this study aims to disclose the important role of the alteration of MCU and DNM1L/Drp1 and their influences on autophagy in CRT. We suggest that decreased MCU, elevated DNM1L/Drp1 and subsequent enhanced autophagy after CRT provide cardioprotection. To our knowledge, the connection among MCU, DNM1L/Drp1 and autophagy to elucidate the molecular mechanism of CRT has not been reported before.

| Establishment of canine model with experimental heart failure
Thirteen male beagle's dogs with ages ranging from 1 year to 1.5 year and weights around 15-20 kg were originally purchased from the College of Agriculture and Biology, Shanghai Jiaotong University for this study. Animals were randomly divided into three groups: sham group (n = 4), HF group (n = 5), CRT group (n = 4). However, as one dog in HF group died from serious post-operation infection, 12 dogs were finally included. The asynchronous HF was established through left bundle branch (LBB) ablation followed by 8 weeks of rapid atrial pacing with 200 bpm. The LBB potential was mapped and characterized as biphasic or triphasic waves between His bundle potential and ventricular potential. LBB block was verified by intracavitary electrocardiograph after frequency ablation. CRT group was treated with 4 weeks of rapid atrial pacing followed by 4 weeks of biventricular pacing at the same heart rate ( Figure S1). At terminal study, dogs were anaesthetized and samples were collected as pre-

| Echocardiography
Echocardiography was performed at three time points (baseline,

| Electrocardiography
Electrocardiographs (ECGs) were performed to validate the LBBB and biventricular pacing. All ECGs were depicted at a paper speed of 25 mm/s. QRS duration and QRS morphology were measured by two independent cardiologists. When their opinions went against each other, another physician was brought in to give an ultimate decision.

| Microarray analysis of differentially expressed genes
Tissue samples of lateral LV wall in the distribution of the left circumflex artery were acquired and mixed from four dogs in each group, and the total RNA was isolated with TRIzol reagent (Invitrogen Life Technologies, Carlsbad, CA) after homogenization ( Figure S4A). Four pieces of mixed RNA were quantified by the spectrophotometers (NanoDrop ND-1000, Thermo Scientific Inc, Wilmington, DE) respectively. Additionally, RNA integrity was assessed by standard denaturing agarose gel electrophoresis.
Afterwards, total RNA was amplified and labelled with the dye Cy3.
Then the labelled RNA was hybridized onto the canine specific Value") were chosen for further data analysis. Genes with more than 4-fold changes were identified as differentially expressed ones, and hierarchical clustering analysis was performed to show the distinguishable gene expression pattern among three independent mixed samples. Finally, gene ontology (GO) and proteinprotein interaction (PPI) analysis were applied through the STRING online tool, 18 Cytoscope software 19 and DAVID. 20 Fisher's exact test is used to find if there is more overlap between the DEGs and the GO terms that would be expected by chance, and P < 0.05 was considered statistically significant.

| Morphological examination
The hearts of the murine models were promptly obtained, and then fixed in formalin, embedded in paraffin, sliced into 5-μm thick sec-

| Cell stretching culture and validation of the effect of MCU inhibitor
To date, it was hardly seen the simulation of CRT on the cellular level. In the current viewpoints, CRT improved cardiac function and asynchrony through promoting the mechanical function of delayed myocardium. 21 Even baseline myocardial mechanical stretching could be regarded as a predictor of CRT response and good prognosis. 22 It's reasonable to deduce that CRT ultimately influenced the heart by stretching the cardiomyocyte. Therefore, the cardiomyocyte stretching culture was introduced to imitate the mechanical effect of CRT ( Figure S7A). Cells were divided into three groups: control group, angiotensin-II group and angiotensin-II+stretching group. The murine ventricles were harvested from neonatal Sprague-Dawley rats within 48 hours of birth. Then they were minced into pieces and digested in the tyrisin (0.1%). After the digestion, the cardiomyocyte suspension was centrifuged at 1000 × g for 5 minutes at 4°C, and the pellet was re-suspended in low-glucose DMEM culture (Gibco) containing 10% foetal bovine serum (Gemini). Afterwards, cardiomyocytes were purified by the differential adhesion method. The suspended cardiomyocytes were

| Immunofluorescence
After cell stretching, the silicone membrane loaded with cardiomyocytes was fixed in 4% paraformaldehyde, cut into tiny pieces

| Statistical analysis
Continuous variables were presented as mean ± standard error of mean (SEM) with at least three repeated experiments, and inter-group difference was assessed by Student's t test or ANOVA via SPSS software 19.0. A P < 0.05 was regarded as significant.

| Assessment of canine model and the effects of CRT in vivo
Body weight, echocardiographic parameters and QRS duration at baseline showed insignificant difference among three groups ( Figure S2). The LBB potential was mapped and characterized as biphasic or triphasic waves between His bundle potential and ventricular potential. LBBB was verified by intracavitary electrocardiograph after frequency ablation ( Figure 1E). Overdrive pacing-induced heart failure was confirmed by decrease in LVEF and myocardial asynchrony. Compared with sham group, LVEF outstandingly decreased in HF group 65.78 ± 3.26 vs 40.47 ± 1.34, P < 0.05) after 4 weeks of overdrive pacing (Supplemental table).

| CRT decreasing cardiomyocyte's size, alleviating cardiac fibrosis, promoting the formation of autophagosome and mitophagosome and mitigating mitochondrial injury
In HF group, the CSA and the FAP increased dramatically compared with sham group, while CRT could alleviate the cell enlargement and myocardial fibrosis ( Figure 3A,B). TEM study revealed an eminent increase of autophagosome, mitophagosome and mitochondrial fission in CRT group ( Figure 3C).
Moreover, TEM study indicated severe swelling and disruption of myocardial fibre in HF group ( Figure S3B), while CRT could relieve these damages ( Figure S3C). Besides, morphological disorder of mitochondria (disarrangement and cristae loss) indicated mitochondrial damage in HF dogs. However, after CRT treatment, the mitochondrial injury was significantly relieved ( Figure 3C and Figure S3).

| CRT significantly influencing gene expression profile
GO analysis of DEGs indicated kinds of biological processes involved (Figure 2A), especially various mitochondria-associated processes (P < 0.05) and autophagy (P < 0.001). Meanwhile, gene cluster analysis validated the DEGs abounded with plenty of genes related to metabolism and autophagy ( Figure 2B and Figure S6).
Further analysis revealed the proportion of mitochondria-and autophagy-related genes in the total DEGs was 12.7% ( Figure 2C).
The PPI analysis showed dense nodes with close PPI relationships among mitochondrial proteins and autophagic proteins ( Figure 2D In accordance with the TEM study, Pink1 was down-regulated in HF group, while both Parkin and Pink1 were up-regulated in HF+CRT group, indicating enhanced mitophagy ( Figure 3E,F).

| Effects of cardiomyocyte stretching culture on angiotensin-II-induced pathological stress
Consistent with the results of animal model, MCU was elevated in angiotensin-II treated cells compared with control cells, however, cell stretching could reduce the expression of MCU and increase the expression of DNM1L/Drp1 under angiotensin-II-induced pathological stress ( Figure 4A,B). LC3B II/I ratio was reduced, while SQSTM1/p62 was increased after angiotensin-II treatment, indicating impaired autophagy. However, cell stretching could up-regulate LC3B II/I, Parkin and Pink1 and down-regulate SQSTM1/p62, referring to enhancement of autophagy and mitophagy ( Figure 4A,B).
Changes of intracellular fluorescence intensity of MCU and DNM1L/ Drp1 were consistent with the results of western blots ( Figure 5).
Immunofluorescence of the cardiomyocyte membrane-specific GJA1/ while MCU knockout could partly restore them. Therefore, intervention on MCU could protect from pathological injury by autophagy/ mitophagy enhancement ( Figure S9).

F I G U R E 4 Analysis of the protein expression changes after cell stretching or mitochondria calcium uniporter inhibition in vitro
F I G U R E 5 In vitro analysis of the effects of cellular mechanical stretching on angiotensin-II treated cardiomyocytes

| D ISCUSS I ON
CRT brings great benefits to HF patients, especially to someone with refractory HF which hardly response to medication therapy. CRT was proved to improve cardiac function, 14 CRT could re-balance the sympathetic-parasympathetic system through inhibiting the cholinergic pathway and improve cardiac function. 27 It was reported that CRT could partly correct the heterogeneously expressed ryanodine receptor and sodium channel in HF to mitigate cardiac dysfunction and myocardial asynchrony. 25,26 CRT had effect on the oxidative modification of cysteine residue in ATP synthase, and the function of ATP synthase was regulated subsequently to increase the ATP production. 33 Recently, a translational study reported microRNA-30d could effectively predict CRT non-response in patients, and their animal experiment suggested elevated circulating microRNA-30d and cardiac in situ expressed microRNA-30d might be the key factor against apoptosis. 28 Abovementioned studies suggested CRT should influence cardiomyocyte on energetic metabolism. Therefore, roles of autophagy and mitophagy in CRT were investigated herein. To date, it was hardly seen the simulation of CRT on the cellular level. CRT improved cardiac function and asynchrony through promoting the mechanical function of the heart. 14,21 Moreover, baseline myocardial mechanical stretching could be regarded as a prognostic predictor of CRT response. 22 It is reasonable to deduce that CRT ultimately influenced the heart by stretching the cardiomyocyte.
Mechanical cell stretching promoted the metabolic activity of ventricular cells and optimized the utilization of ATP production. 34 Our findings indicated increased MCU, decreased DNM1L/Drp1 and impaired autophagy and mitophagy in HF. MCU was believed to take an active part in mitochondrial pathophysiology. 2

| Limitation
The Beagle's dogs in this study were of wild type, and MCU gene loss-of-function study could not be carried out as the transgenic dog model was unfeasible. Besides, the animal sample size was relatively small, and this study should be treated as a pilot study which needs more research in future. The cell stretching model was designed to imitate the mechanical effects of the CRT, however, the CRT-related electrical effects could not be elucidated by the present study. Therefore, it is worthy of further investigation of the mechanism of CRT.

| CON CLUS ION
The

CO N FLI C T S O F I NTE R E S T
None.