Binding of LncRNA-DACH1 to dystrophin impairs the membrane tracking of Nav1.5 protein and increases ventricular arrhythmia susceptibility

Dystrophin is a critical interacting protein of Nav1.5 that determines its membrane anchoring in cardiomyocytes. The study aims to explore whether lncRNA-DACH1(lncDACH1) can regulate the distribution of Nav1.5 by binding to dystrophin and participate in ventricular arrhythmogenesis. LncDACH1 was conrmed to bind to dystrophin. Cardiomyocyte-specic transgenic overexpression of lncDACH1(lncDACH1-TG) reduced the membrane distribution of dystrophin and Nav1.5 in cardiomyocytes. The opposite data were collected from lncDACH1 cardiomyocyte conditional knockout (lncDACH1-CKO) mice. Moreover, increased ventricular arrhythmia susceptibility was observed in lncDACH1-TG mice in vivo and ex vivo. The conservative fragment of lncDACH1 inhibited membrane distribution of dystrophin and Nav1.5 and promoted the inducibility of ventricular arrhythmia. Upregulation of dystrophin in lncDACH1-TG mice rescued the impaired membrane distribution of dystrophin and Nav1.5. The human homologue of lncDACH1 inhibited the membrane distribution of Nav1.5 in human iPS-differentiated cardiomyocytes. Collectively, lncDACH1 regulates Nav1.5 membrane distribution by binding to dystrophin and participates in ventricular arrhythmogenesis.


Introduction
The voltage gated sodium channel mediates the 0 phase depolarizing inward sodium currents of cardiomyocytes 1 . The expression and function of sodium channel is regulated at multiple levels encompassing gene mutation, post-transcriptional modi cation, post-translational modi cation and protein tra cking etc 1,2 . The disruption of either process is arrhythmogenic and occasionally causes sudden death 2, 3 .
The membrane targeting and localization of pore-forming subunit of sodium channel Nav1.5 was regulated by several interacting proteins such as ankyrin-G, MOG1, syntrophin and dystrophin etc 2,4 .
Dystrophin is a cytoplasmic protein that is encoded by duchenne muscular dystrophy (DMD) gene 5 . It distributes mainly on the cellular membrane of skeletal muscle cells and cardiomyocytes, and acts as a scaffold for Nav1.5 6 . In cardiomyocytes, dystrophin controls the expression and membrane anchoring of Nav1.5. Gavillet et al. showed that knockout of dystrophin in cardiomyocytes reduced peak sodium current, Nav1.5 protein expression and conduction velocity in mice 6 . Subsequently, they con rmed that knockout of dystrophin inhibits membrane distribution of Nav1.5 due to the disruption of dystrophinsyntrophin complex 7 .
Long noncoding RNAs (LncRNAs) are a new class of RNAs that are more than 200 nts long and possess little protein coding property. 8 LncRNAs have been shown to regulate multiple biological processes and participate in the pathogenesis of various diseases including cardiac diseases 9 . LncRNAs were shown to regulate cardiac electrophysiological property by altering the function of different ion channels. For example, the increased expression of lncRNA-Kcna2as in heart failure reduced I ks and prolonged action potential duration (APD) 10 . LncRNA-MALAT1 enhanced arrhythmia susceptibility by suppressing I to and prolonging APD 11 . In a previous study, we found that lncRNA-CCRR (cardiac conduction regulatory) interacts with connexin-43 interacting protein 85(CIP85) to promote connexin-43 membrane distribution and improve the impaired cardiac conduction of failing hearts 12 . However, to date, it remains unknown whether and how lncRNA regulates sodium channel.
LncDACH1 is an intronic lncRNA located on the rst intron of DACH1 gene 13 . We previously showed that lncDACH1 impairs cardiac function by promoting the degradation of sarco-endoplasmic reticulum ATPase 2a (SERCA2a), and exacerbates cardiac ischemia injury by inhibiting Yes-associated protein 1(YAP1) mediated proliferation of neonatal cardiomyocytes 13,14 . During analyzing the interacting proteins of lncDACH1 identi ed by mass spectrometry, we found that dystrophin is one of them, which drove us to hypothesize that lncDACH1 may be a critical regulator of sodium channel Nav1.5.
Therefore, in this study we explored whether lncDACH1 regulates Nav1.5 by interacting with dystrophin. We found that lncDACH1 inhibited the membrane tra cking of Nav1.5 by binding to dystrophin, which led to reduced sodium current and increased ventricular arrhythmia susceptibility. The study highlights a novel mechanism for the regulation of sodium channel tra cking, and reveals a potential therapeutic target for sodium channel dysfunction related cardiac arrhythmias.

Results
LncDACH1 binds to dystrophin and reduces Nav1.5 membrane distribution in cardiomyocytes of lncDACH1 transgenic mice We rstly validated the binding between lncDACH1 and dystrophin. The RNA pulldown plus immunoblot assay con rmed that lncDACH1 can successfully pulldown dystrophin ( Figure 1A). Conversely, the immunoprecipitation study showed that the antibody for dystrophin precipitated lncDACH1, while the negative control IgG did not ( Figure 1B). We next explored the in uence of lncDACH1 on cellular distribution of dystrophin. The western blot data showed that the total protein of dystrophin did not change, while the membrane fraction was reduced, and the cytoplasmic fraction increased in the hearts of lncDACH1-TG mice than wild-type (WT) controls ( Figure 1C). Consistently, the membrane and total protein levels of Nav1.5 were reduced, while cytoplasmic Nav1.5 increased in the hearts of lncDACH1-TG mice than WT controls ( Figure 1D). The reduced membrane distribution of dystrophin and Nav1.5 in the cardiomyocytes of lncDACH1-Tg mice was further con rmed by immuno uorescent staining ( Figure 1E). The mRNA levels of dystrophin and SCN5A did not change ( Figure 1F). We then evaluated the functional change of sodium channel.
Consistent with reduction of membrane Nav1.5, the peak I Na was signi cantly decreased in the ventricular myocytes of lncDACH1-TG mice than WT controls, while the kinetics of I Na (activation, inactivation, and recovery) did not change ( Figure 1G). Meanwhile, the conduction velocity was slower in the hearts of lncDACH1-TG than WT mice ( Figure 1H).
We then applied lncDACH1 adenovirus to cultured neonatal cardiomyocytes to con rm the regulation of lncDACH1 on sodium channel in vitro. Infection of adenovirus carrying lncDACH1 signi cantly upregulated the level of lncDACH1 ( Figure 2A) and remarkably inhibited peak I Na with no change in kinetics ( Figure 2B, C). The membrane distribution of dystrophin and Nav1.5 was remarkably reduced by overexpression of lncDACH1 as indicated by immuno uorescent staining ( Figure 2D). The mRNA levels of dystrophin and SCN5A were not altered by lncDACH1 overexpression ( Figure 2E).
Inhibition of lncDACH1 in cardiomyocytes increased membrane Nav1.5 distribution We then employed lncDACH1 cardiomyocyte conditional knockout(lncDACH1-CKO) mice to examine the regulation of lncDACH1 on dystrophin and Nav1.5. The western blot data showed that membrane distribution of dystrophin was increased in the hearts of lncDACH1-CKO mice than WT controls, while the total dystrophin protein and dystrophin mRNA did not change ( Figure 3A). Consistently, the membrane and total level of Nav1.5 was increased in the hearts of lncDACH1-CKO mice than WT controls, with no change on SCN5A mRNA ( Figure 3B). The change on dystrophin and Nav1.5 membrane distribution was further validated by immuno uorescent staining ( Figure 3C). Meanwhile, the peak I Na was larger in cardiomyocytes of lncDACH1-CKO mice than WT controls, while the kinetics of I Na (activation, inactivation, and recovery) did not change ( Figure 3D). Consistent with the increase of peak I Na , the conduction velocity in the hearts of lncDACH1-CKO mice was faster than WT controls ( Figure 3E). We further con rmed effects of lncDACH1 knockdown with its siRNAs on sodium channel in cultured neonatal cardiomyocytes in vitro. Infection of adenovirus carrying siRNAs for lncDACH1 signi cantly reduced the level of lncDACH1 ( Figure 4A). The patch-clamp recordings showed that knockdown of lncDACH1 signi cantly increased the current density of peak I Na with no change on channel kinetics ( Figure 4B, C). Membrane distribution of dystrophin and Nav1.5 were both increased after knockdown of lncDACH1 as indicated by immuno uorescent staining ( Figure 4D). The mRNA levels of dystrophin and SCN5A were not altered by lncDACH1 knockdown ( Figure 4E).

Transgenic overexpression of lncDACH1 is arrhythmogenic in mice
We next evaluated whether the inhibition of Nav1.5 by lncDACH1 is arrhythmogenic in lncDACH1-TG mice. Electrical pacing technique was employed to evaluate the arrhythmia susceptibility of intact hearts in vivo and isolated hearts ex vivo. The in vivo study showed that programmed pacing induced more ventricular arrhythmia in lncDACH1-TG mice that WT controls. Both induction rate and episodes of ventricular arrhythmia were higher in lncDACH1-TG mice ( Figure 5A). In consistent, the ex vivo electrical pacing study demonstrated that ventricular arrhythmia was more frequently occurred in lncDACH1-Tg mice ( Figure 5B). The optimal mapping study revealed that there are more breaking points in the perfused heart of lncDACH1-TG mice than WT controls ( Figure 5C-E). Conversely, no ventricular arrhythmia was induced in the hearts of lncDACH1-CKO mice ( Figure 5F).
The conservative fragment of lncDACH1 reduced peak sodium current and promoted ventricular arrhythmia The sequence blasting data showed that the fragment of lncDACH1 from 835 to 2085 nts is conservative between human and mouse. We then cut lncDACH1 into different fragments ( Figure 6A) to evaluate the sequence that is responsible for the binding with dystrophin. The data showed that only fragments containing the conserved sequence, fragments a and b, can pulldown dystrophin, indicating that the conserved sequence may be the functional region of lncDACH1 ( Figure 6A).
We then examined the in uence of the conserved sequence from 835 to 2085 nts (conserved fragment of lncDACH1, cF-lncDACH1) on cardiac sodium channel. The adenovirus carrying cF-lncDACH1 was constructed and administered to mice. The successful overexpression of cF-lncDACH1 was validated by qRT-PCR ( Figure 6B). Administration of cF-lncDACH1 reduced the membrane distribution, and increased cytoplasmic expression of both dystrophin and Nav1.5 as indicated by western blot and immuno uorescent data ( Figure 6C, D). The mRNA levels of dystrophin and Nav1.5 were not affected by cF-lncDACH1 (Supplemental Figure 1A). Overexpression of cF-lncDACH1 reduced peak I Na currents ( Figure 6E), and produced no in uence on channel kinetics (Supplemental Figure 1B). The optical mapping data showed that administration of cF-lncDACH1 reduced conduction velocity and increased break points of ventricular arrhythmias ( Figure 6F, G). The induction rate and episodes of ventricular tachycardia (VT) were higher in cF-lncDACH1 group than controls ( Figure 6H).
In cultured neonatal cardiomyocytes, overexpression of cF-lncDACH1 reduced peak I Na with no change in kinetics, inhibited membrane distribution of dystrophin and Nav1.5, and produced no in uence on the mRNA levels of dystrophin and SCN5A (Supplemental Figure 2). Activation of dystrophin transcription by dCas9-SAM system prevented the reduction of sodium current in lncDACH1 transgenic mice As lncDACH1 reduced Nav1.5 membrane targeting by interacting with dystrophin, we reasoned that overexpression of dystrophin may rescue the inhibition of Nav1.5 by lncDACH1. To test this notion, we constructed the AAV9 virus carrying dCas9-SAM system that can activate dystrophin transcription (AAV9-Dys-Act) to perform rescuing experiments on lncDACH1-TG mice (Supplemental Figure 3). Tail vein injection of AAV9-Dys-Act virus signi cantly increased the mRNA level of dystrophin in the hearts of both WT and lncDACH1-TG mice ( Figure 7A). The western blot data showed that overexpression of dystrophin with AAV9-Dys-Act virus increased both total and membrane protein expression of dystrophin, and rescued the reduction of dystrophin expression in lnDACH1-Tg mice ( Figure 7A). The mRNA level of SCN5A was not in uenced by AAV9-Dys-Act virus ( Figure 7B). Strikingly, AAV9-Dys-Act virus administration restored total and membrane expression of Nav1.5 in lnDACH1-TG mice ( Figure 7B). In addition, AAV9-Dys-Act virus injection rescued the reduction of peak I Na current in lnDACH1-TG mice ( Figure 7C). The kinetics of I Na current did not change among groups ( Figure 7C). Activation of dystrophin transcription with AAV9-Dys-Act virus restored the conduction velocity in lncDACH1-TG mice ( Figure 7D).
Both in vivo and ex vivo data indicated that activation of dystrophin transcription reduced the susceptibility to ventricular arrhythmia of lncDACH1-TG mice ( Figure 7E, F).
Hadhb binds to lncDACH1 and promotes its decay Reduced Nav1.5 expression and reduction of peak I Na in heart failure have been reported by a bunch of studies 15,16,17,18 . We therefore evaluated the contribution of lncDACH1 on sodium channel remodeling in transaortic constriction (TAC) induced heart failure model in mice. We found that lncDACH1 was increased in failing hearts than sham controls ( Figure 8A). Although lncDACH1 was upregulated in failing hearts, the mRNA of its host gene DACH1 did not change ( Figure 8B). This nding excluded the transcription related mechanism on lncDACH1 upregulation during heart failure. By analyzing the RNA Pulldown plus Mass Spectrometry data, we identi ed three potential interacting proteins of lncDACH1 that have been shown to regulate RNA stability. They are ANP32a (acidic leucine-rich nuclear phosphoprotein 32A), eIF4A1(eukaryotic initiation factor 4A1) and hydroxyacyl-CoA dehydrogenase subunit beta (hadhb). We therefore speculated that it may be the change of RNA stability that renders to the expression change of lncDACH1. We then tested whether these proteins can affect lncDACH1 level by knocking down their expression with siRNA. The data showed that knockdown of hadhb increased the expression of lncDACH1, while knockdown of ANP32a and eIF4A1 produced no in uence ( Figure 8C). The in uence of hadhb on lncDACH1 stability was further validated by the fact that knockdown of hadhb increased the decaying half-life of lncDACH1( Figure 8D). Furthermore, the sense sequence of lncDACH1 successfully pulled down hadhb, and the antibody of hadhb precipitated lncDACH1( Figure 8E). Additionally, the protein level of hadhb was reduced in mouse failing hearts ( Figure 8F), which is inversely correlated to the upregulation of lncDACH1. The siRNA for hadhb reduced the expression of Nav1.5 ( Figure 8G). These data indicated that hadhb is an upstream regulator of lncDACH1 which determines the stability of lncDACH1.
We lastly evaluated the human conserved sequence of lncDACH1(hcF-lncDACH1) on Nav1.5 distribution of human iPS induced cardiomyocytes. We found that overexpression of hcF-lncDACH1 reduced the membrane distribution of dystrophin and Nav1.5 ( Figure 8H).

Discussion
In this study, we discovered that lncDACH1 is critical regulator of sodium channel in the heart. LncDACH1 binds to dystrophin and thus inhibits membrane tra cking of Nav1.5, which leads to the reduction of peak sodium current and impairment of cardiac conduction. Therefore, upregulation of lncDACH1 increased the susceptibility to ventricular arrhythmia ( Fig 8I).
LncRNAs have been established to be critical regulators of various biological processes 19 . The action modes of lncRNAs are complex. One major mechanism for them to exert their biological function in the cytoplasm is to interact with the molecules such as proteins, miRNAs, mRNAs to alter protein translation, enzyme activity, protein degradation, etc 20 . For instance, lncRNA-CCRR was shown to inhibit the endocytic tra cking of connexin-43 by binding to CIP85 12 . LncDACH1 mainly distributes in the cytoplasm, and can bind to SERCA2a to promote its ubiquitination and degradation 13 . LncDACH1 can also bind to protein phosphatase 1 catalytic subunit alpha (PP1A) to inhibit its dephosphorylation activity on yes-associated protein 1 (YAP1), leading to the cytoplasmic sequestration of YAP1 14 . The unraveling of the molecular mechanism of Nav1.5 expression and function is critical for the insightful understanding of sodium channel function under physiological and pathological conditions. Several interacting proteins have been demonstrated to determine membrane distribution of Nav1.5 and sodium channel function 4 . Dystrophin is a well-characterized Nav1.5 partner protein. It indirectly interacts with Nav1.5 via syntrophin, which binds with the C-terminus of dystrophin and with the SIV motif on the Cterminus of Nav1.5 6,21 . In this study, we found that lncDACH1 binds to dystrophin and leads to the impairment of Nav1.5 tra cking and reduced membrane distribution. This is so far the rst evidence verifying the regulation of lncRNA on cardiac sodium channel.
Although the membrane distribution of both dystrophin and Nav1.5 was inhibited by lncDACH1, the total protein level of dystrophin was not affected, while Nav1.5 was reduced. The mechanism for the differential in uence of lncDACH1 on total protein levels of dystrophin and Nav1.5 is unclear. One explanation may be that Nav1.5 is a membrane channel protein. If they failed to target on the plasma membrane, they may be regarded as unnecessary protein and undergo the process of protein degradation. The E1-E3 enzymes in the ubiquitination systems have been shown to regulate the degradation of Nav1.5, which includes E1 enzyme UBE1(Ubiquitin-activating Enzyme1), UBA6(Ubiquitinlike modi er-activating enzyme 6), E2 enzyme, UBC9 (Ubiquitin-Conjugating Enzyme 9), and E3 ligase Nedd4-2 (neuronal precursor cell expressed developmentally downregulated 4-2) 22, 23, 24 . LITAF (lipopolysaccharide-induced tumor necrosis factor-alpha factor), a protein encoding a regulator of endosomal tra cking, was shown to reduce surface Nav1.5 by promoting degradation of NEDD4-2 25 . Therefore, cytoplasmic Nav1.5 that failed to target on plasma membrane may be quickly distinguished and then degraded by these ubiquitination enzymes.
The dysfunction of sodium channel is associated with various arrhythmias. Reduced peak I Na due to SCN5A loss-of-function mutation can cause a series of arrhythmias such as atrial brillation, Brugada syndrome, long QT syndrome, sudden cardiac death, and ventricular tachycardia etc 26,27 . Consistently, we found that accompanied with the reduction of peak I Na , overexpression of lncDACH1 reduced ventricular conduction velocity and increased the susceptibility to ventricular arrhythmia in mice. The increased peak I Na due to SCN5A gain-of-function mutation is associated arrhythmias such as atrial brillation, long QT syndrome; polymorphic ventricular complexes and ventricular arrhythmia 26,27 . In this study, although the peak I Na increased in lncDACH1-CKO mice, the susceptibility to arrhythmia did not increase. One difference of our data with SCN5A gain-of-function mutation is that the kinetics of peak I Na often change during mutation, while it is not the case of lncDACH1 knockout. One possible explanation may be that the increase of peak I Na has not reach the limit to break the electrophysiological balance of the heart. Sodium channel remodeling is commonly occurred in cardiac diseases especially heart failure. Despite of some discrepancy, the main observations are that the peak I Na , SCN5A mRNA and Nav1.5 protein are reduced during heart failure of both human patients and animal models 15,16,17,18 . In this study, we found that lncDACH1 was increased during heart failure, indicating that it may contribute to sodium channel remodeling and arrhythmogenesis during heart failure by interfering the action of dystrophin.
Interestingly, we found that activation of dystrophin with dCas9-SAM system restored the membrane distribution of Nav1.5 in cardiomyocytes of lncDACH1-TG mice, which implies its potential in counteracting sodium channel remodeling of patients with heart failure.
The low sequence conservation of lncRNAs among species is a critical issue that limits the extrapolation of data from animal studies to human beings 28,29 . In this study, we found that the conserved fragment of lncDACH1 exhibits the same effect as lncDACH1 on Nav1.5 tra cking and arrhythmogenesis. Moreover, the human conservative homologous fragment of lncDACH1 can inhibit the membrane distribution of Nav1.5 in cardiomyocytes derived from induced pluripotent stem cells (iPS-CMs). These ndings hint the clinical relevance of lncDACH1, which holds the potential to become a therapeutic target for treating sodium channel remodeling in clinic.
In conclusion, lncDACH1 is a novel regulator of sodium channel, which suppresses the membrane tra cking of Nav1.5 by disturbing the function of dystrophin. The current work enriched our understanding of the biology of sodium channel tra cking and function, and indicated that lncRNAs possess the potential to become therapeutic targets for ventricular arrhythmias.

Neonatal cardiomyocytes preparation
Neonatal cardiomyocytes were isolated from 3-day-old mice in accordance with the following procedures. Brie y, after dissection, hearts were washed and minced in 0.25% trypsin. Pooled cell suspensions were centrifuged and resuspended in Dulbecco's modi ed Eagle's medium (DMEM Hyclone, USA) supplemented with 10% fetal bovine serum, 100 U/ml penicillin and 100 μg/ml streptomycin. The suspension was incubated in culture asks for 90 min, which makes broblasts preferentially adhere to the bottom of the culture asks. Neonatal cardiomyocytes were removed from the culture asks and the medium was changed. Cell cultures were incubated for 48 h at 37 °C in a humidi ed atmosphere of 95% oxygen and 5% carbon dioxide before any experimentation.
Generation of cardiac myocyte-speci c lncDACH1 overexpressing mice Cardiomyocyte-speci c lncDACH1 overexpressing mice driven by murine αMHC promoter on a C57BL/6 background was generated by Biocytogen Co., Ltd (Beijing, China) as demonstrated in previous study 13 .
Construction of adeno-associated virus 9 (AAV9) carrying deactivated clustered regularly interspaced short palindromic repeats associated protein 9 nuclease-synergistic activation mediator(dCas9-SAM) system to activate the transcription of dystrophin Adeno-associated virus 9(AAV9) carrying dCas9-SAM system to activate the transcription of dystrophin was constructed as reported previously with brief modi cations 30 . The sgRNA targeting on the promoter region of dystrophin was designed and cloned into the multiple cloning site of plasmid GV639 (EFS-NLS-dSaCas9-NLS-VP64-bGHpA-U6). The constructed plasmid was packaged into AAV9 virus. The sequence of sgRNA is: 5'-CGCTTCCGCGGCCCGTTCAA -3'; The mock-sgRNA target sequence (5'-CGCTTCCGCGGCCCGTTCAA -3') was used as negative control. The obtained AAV9 virus volume was administered into C57BL/6 mice via tail vein injection at 1×10 11 genome containing particles (GC)/animal in 100µl.

Construction of adenovirus carrying cF-lncDACH1 and in vivo gene delivery
Adenovirus vectors carrying cF-lncDACH1(OE-cF-lncDACH1) and a negative control (NC) and a CAG promoter conjugated with green uorescent protein (GFP) were constructed by Genechem Co., Ltd.
(Shanghai, China). OE-cF-lncDACH1, control constructs at 1×10 9 genome containing particles (GC)/animal in 100µl volume was administered into C57BL/6 mice with body weights ranging from 18~22g via tail vein injection. Seventy-two hours after injection, the mice were subjected to further study.
Construction of adenovirus carrying lncDACH1, lncDACH1 siRNA, conserved fragment of human lncDACH1 and infection Adenovirus vectors carrying lncDACH1(OE-lncDACH1), a short RNA fragment for silencing lncDACH1 (Si-lncDACH1) or conserved fragment of human lncDACH1(hcF-lncDACH1) and a CAG promoter conjugated with green uorescent protein (GFP) were constructed by Genechem Co., Ltd. (Shanghai, China). Neonatal cardiomyocytes were infected with adenovirus for 48 hours, and then subjected to subsequent study.
Induction of ventricular arrhythmia C57BL/6 mice were anesthetized with 2,2,2-tribromoethanol (200 mg/kg, i.p.). An octapolar electrophysiological catheter (1.1F, SciSense Inc., Canada) was inserted into the right ventricle via the jugular vein. Intracardiac pacing was performed using an automated stimulator interfaced with the data acquisition system (GY6000; HeNan HuaNan Medical Science & Technology Ltd., Zhengzhou, China). The surface recording electrode was xed on LV epicardium to record pseudo-ECG. Inducibility of ventricular tachycardia (VT) was determined by applying a train of ten consecutive electrical pulses with a coupling interval of 80 ms (S1), followed by two extra stimuli (S2 and S3) at coupling intervals of 2 ms, respectively. Successful induction of VT was de ned as the appearance of rapid nonsinus rhythm ventricular activations lasting for three beats or more.

Construction of truncated LncDACH1 fragment plasmids
The sequence of lncDACH1 was divided into ve fragments. The cDNA of each fragment was inserted into the pCDNA3.1, respectively. The rst 417 nts of the entire sequences was cut off and constructed as fragment-a (418-2085 nts). Another 417 nts was cut off to generate fragment-b (835-2085 nts). The third 417 nts was cut off to generate fragment-c (1251-2085 nts). Fragment-d is from 835-1668 nts, and fragment-e is from 835-1251 nts.

Isolation of cardiac myocytes
Hearts were rapidly excised, cannulated, and perfused with Ca 2+ -free Tyrode solution (in mM): NaCl 137, KCl 5.4, NaH 2 PO 4 0.16, glucose 10, CaCl 2 1.8, MgCl 2 0.5, HEPES 5.0, and NaHCO 3 3.0 (pH 7.4 adjusted with NaOH) for 5 min. The heart was then perfused with a solution containing collagenases B and D (Roche) and protease XIV (Sigma) until digestion was complete. Tissue was dissociated using forceps, and dissociated left ventricular cardiomyocytes were gradually exposed to Ca 2+ (from 50 to 500 µM over 40 min) and plated in culture chambers for further studies.
Immunocytochemistry of isolated mouse ventricular myocytes.

Fluorescent in situ hybridization (FISH)
In situ hybridization was performed with a Fluorescent in Situ Hybridization (FISH) Kit (RiboBio, Guangzhou, China). Brie y, isolated cardiomyocytes were xed in 4% formaldehyde at 4°C for 10 min and dried out on the slides at room temperature (RT). The slides were rinsed and permeabilized with 0.5% Triton-100 in PBS at RT for 30 min, washed with PBS solution, and prehybridized at 37°C for 30 min before hybridization. The prehybridized slides were then incubated with lncRNA-probe in hybridization solution at 37°C for 16 h. After hybridization, the slides were washed six times with prewarmed wash buffer and PBS solution. Finally, the slides were counterstained with DAPI and visualized using a confocal laser-scanning microscope (Zeiss 800, Germany). The total, membrane and cytoplasmic protein samples were extracted from cardiac tissues of C57BL/6 mice for immunoblotting analysis. Total protein was collected with the treatment of RIPA lysis buffer (Beyotime, Beijing, China) and a protease inhibitor cocktail (Roche, Basel, Switzerland) at 4°C followed by centrifugation. Extraction of surface and cytoplasmic proteins was conducted using the Surface and Cytoplasmic Protein Reagent Kit (Cat#P0033; Beyotime, Shanghai, China) according to the manufacturer's instructions. Protein samples were fractionated by SDS-PAGE and then transferred to PVDF membrane. The membranes were blocked in Tris-buffered saline containing 5% milk and then incubated with primary antibodies at 4°C overnight. The primary antibodies include anti-Nav1.5 (ASC005, Alomone, 1:200), anti-dystrophin (MANDYS8, SIGMA, 1:500). The anti-β-actin (1:20000 dilution, 66009-1-Ig, Proteintech) and anti-N-cadherin antibody (Cat#ab76011, 1:5000; Abcam, Cambridge, UK) were used as internal controls. Western blot bands were captured on the Odyssey Infrared Imaging System (LI-COR Biosciences, USA) and quanti ed with Odyssey v1.2 software by measuring the band intensity (area × OD) in each group. The band intensity was normalized to the internal control. All antibodies were diluted in PBS buffer.

RNA pulldown and immunoblotting
The RNA pull-down was performed as described in the previous study 13 . Brie y, Biotin-labeled, full length lncDACH1 RNA and antisense RNA were prepared with the Biotin RNA Labeling Mix (Roche) and T7 RNA polymerase (Roche). Biotinylated RNAs were treated with RNase-free DNase I (Invitrogen) and puri ed on G-50 Sephadex Quick Spin columns (Roche). Biotinylated RNA (17μg) was heated to 65°C for 10 min and slowly cooled to 4°C. Then the RNA was mixed with tissue extracts in pulldown buffer supplemented with tRNA (0.1 μg/μl) and incubated at 4°C for 2 h. Washed Streptavidin agarose beads (60 μl, Invitrogen) were added to each binding reaction and further incubated at 4°C for 1 h. Beads were washed brie y ve times in pulldown buffer and boiled in SDS buffer, and the retrieved protein was visualized by immunoblotting.

RNA immunoprecipitation (RIP)
RNA immunoprecipitation (RIP) experiments were performed by using a Magna RIPTM RNA-Binding Protein Immunoprecipitation Kit (Millipore, USA) as previously reported 13 . Brie y, heart tissue was pieced and lysed in 220 μl of lysis buffer containing protease inhibitors and RNase Inhibitor and centrifuged at 14,000 × g for 10 min. The supernatants were incubated with anti-dystrophin, anti-hadhb and anti-rabbit IgG antibody for overnight at 4°C with gentle rotation. Protein G magnetic beads (50 μl) were added and incubated at RT with gentle rotation for 3 h. RNA was extracted with 400 μl phenol:chloroform:isoamyl alcohol (125:24:1, pH = 4.3) according to the manufacturer's instructions before quantitation by RT-qPCR.
Mouse models of heart failure (HF) by transaortic constriction (TAC) and by coronary artery ligation Mice were randomly divided into sham and TAC groups. In each group, mice were anesthetized with 2,2,2tribromoethanol (200 mg/kg, i.p.) for TAC model. The animal was orally intubated with 20-gauge tube, and ventilated (mouse ventilator, UGO BASILE, Biological Research Apparatus, Italy) at the respiratory rate of 100 breaths/min with a tidal volume of 0.3 ml. The transverse aorta was constricted by a 6-0 silk suture ligature tied rmly against a 27-gauge needle between the carotid arteries. Then, the needle was promptly removed to yield a constriction of 0.4 mm in diameter. For sham group mice, the animals received the same procedures without aorta constriction.

Statistics
Data are expressed as mean ± SEM. Statistical analysis was performed using unpaired Student's t test or One-Way Analysis of Variance (ANOVA) followed by Tukey's post-hoc analysis. A P< 0.05 was considered statistically different.v