Depletion of m6A reader protein YTHDC1 induces dilated cardiomyopathy by abnormal splicing of Titin

Abstract N 6‐methyladenosine (m6A) is the most prevalent modification in mRNA and engages in multiple biological processes. Previous studies indicated that m6A methyltransferase METTL3 (‘writer’) and demethylase FTO (‘eraser’) play critical roles in heart‐related disease. However, in the heart, the function of m6A ‘reader’, such as YTH (YT521‐B homology) domain‐containing proteins remains unclear. Here, we report that the defect in YTHDC1 but not other YTH family members contributes to dilated cardiomyopathy (DCM) in mice. Cardiac‐specific conditional Ythdc1 knockout led to obvious left ventricular chamber enlargement and severe systolic dysfunction. YTHDC1 deficiency also resulted in the decrease of cardiomyocyte contractility and disordered sarcomere arrangement. By means of integrating multiple high‐throughput sequence technologies, including m6A‐MeRIP, RIP‐seq and mRNA‐seq, we identified 42 transcripts as potential downstream targets of YTHDC1. Amongst them, we found that Titin mRNA was decorated with m6A modification and depletion of YTHDC1 resulted in aberrant splicing of Titin. Our study suggests that Ythdc1 plays crucial role in regulating the normal contractile function and the development of DCM. These findings clarify the essential role of m6A reader in cardiac biofunction and provide a novel potential target for the treatment of DCM.


| INTRODUC TI ON
N 6 -methyladenosine (m 6 A) is the most abundant internal epigenetic modification on eukaryotic mRNAs. 1-3 Over 12,000 m 6 A sites characterised by a typical consensus in the transcripts of more than 7000 human genes have been identified. 4 Built on the understanding of its widespread prevalence on mRNA, m 6 A has been shown to play crucial roles in mRNA fate determination and participates in multiple biological processes. Recent studies uncovered that m 6 A modification serves to facilitate critical steps in mRNA splicing, translation initiation and decay. 5,6 Similar to epigenetic marks on DNA and histone, m 6 A on RNA is also dynamic and reversible. The m 6 A modification is installed by the METTL3-METTL14-WTAP methyltransferase 'writer' complex 7 and can be removed by m 6 A demethylases FTO 8 and ALKBH5 ('eraser'). 9 Both the 'writer' and 'eraser' are essential in pathological remodelling of the heart. METTL3 deficiency in mice induces maladaptive remodelling and ultimately causes heart failure, 10 whilst reduced FTO expression is observed in failing mammalian hearts, which may attenuate the ischaemia-induced cardiac remodelling. 11 These results documented the importance of m 6 A modification in the aetiology of cardiac disease.
Whilst methyltransferases and demethylases act as the 'writer' and 'eraser' of m 6 A on mRNA, respectively, the downstream effects of RNA m 6 A modification are specifically deciphered by the 'reader' YTH (YT521-B homology) domain-containing proteins, including cytoplasmic YTHDF1-YTHDF3, 12-14 YTHDC2 15 and nuclear YTHDC1. 16,17 In the cytosol, YTHDF1 and YTHDF3 work in concert to enhance the translation of target mRNAs, 12,14 whilst YTHDF2 has been shown to affect the stability of m 6 A-modified RNAs by localising them to mRNA decay machinery. 13 YTHDC2 either promotes the translation efficiency of its targets or decreases RNA stability by interacting with different binding partners. 18 Intriguingly, as the sole nuclear m 6 A reader, YTHDC1 modulates nuclear processing of its targets including alternative splicing (AS), 16 export of m 6 A decorated mRNA 19 and turnover of chromatin-associated RNA. 20,21 Functional deficit of YTH family proteins can induce multiple diseases. However, the exact roles of m 6 A readers in heart are poorly understood.
Dilated cardiomyopathy (DCM) is one of the most common causes of heart failure and indication for heart transplantation worldwide. 22 An estimated 35%-40% of genetic DCMs may result from sarcomere gene mutations. 23,24 Of the known genetic mutations that cause DCM, Titin (TTN) mutations account for 20%-25% of cases. 25 There are two major Titin mRNA isoforms, N2BA and N2B. The increased ratio of N2BA:N2B due to the abnormal Titin pre-mRNA splicing can directly lead to DCM. 26 However, the regulatory molecular mechanisms of Titin splicing are still largely unknown.
In this work, we reported that amongst the four major YTH domain-containing proteins, only cardiac-specific ablation of Ythdc1 contributes to a typical DCM phenotype in mice. The loss of YTHDC1 leads to aberrant splicing of Titin, inducing an increased ratio of N2BA:N2B isoform, which finally leads to DCM.

| MATERIAL AND ME THODS
The data that support the findings of this study are available from the corresponding authors upon reasonable request.

| Mouse models and anaesthesia
This study conformed to the rules of the Guide for the Care and Use of Laboratory Animals made by the U.S. National Institutes of Health. All the animal experiments were approved by the Animal Care and Use Committee of Tongji University School of Medicine.
All the generation construction strategies of the YTH domaincontaining proteins were shown in Figure 1A and Figure S1A-S1C. Importantly, the conditional Ythdc1 targeted mice were generated by inserting the loxP sites covering the exon 5 to exon 7. To disrupt the Ythdc1 gene in cardiomyocytes, Ythdc1 knockout (KO) mouse embryonic stem cell clone was purchased from CAM-SU GRC, microinjected into mouse blastocysts and implanted into pseudo-pregnant mice. The resulting chimeric mice were crossed with FLPeR mice to excise the FRT flanked selection cassette to obtain Ythdc1 flox mice which were then crossed with α-MHC-Cre transgenic mice to generate Ythdc1 conditional knockout (cKO) mice. All mice used in this study were maintained in C57BL/6J genetic background. Primer sequences for genotyping are listed in Table S1.

| Quantitative real-time PCR (qRT-PCR)
Total RNA was extracted from mouse left ventricles using qRT-PCR was performed by using SYBR Green Supermix (636600, Toyobo) in a Thermo QuantStudio 6 Flex Real-Time PCR system.
Gene expression was determined relative to Gapdh using the log 2 fold change method. All qRT-PCR primers covered exon-exon junctions when possible. Primer sequences for qRT-PCR are listed in Table S1.

| Western blot
Western blot was performed as described previously. 27 Briefly, tissue homogenates were lysed in RIPA buffer (50 mM Tris, 10 mM EDTA, 150 mM NaCl, 0.25% deoxycholic acid, 0.1% SDS, 1% NP-40 substitute) with protease inhibitors (Roche). After centrifugation, the supernatant was collected and protein was separated on 10% SDS-polyacrylamide gel electrophoresis and transferred to PVDF membrane. The membranes were blocked with 3% bovine serum albumin and incubated with the following primary antibodies: YTHDC1 (ab220159,Abcam), GAPDH (60004,Proteintech) overnight at 4°C. The secondary antibodies conjugated to infrared dyes (LI-COR Biosciences) were applied, and the blots were visualised with an Odyssey imager and quantified by ImageJ software.

| Histology
Hearts were dissected from age-and sex-matched littermates, washed in PBS and fixed overnight in 4% paraformaldehyde.
Samples were subsequently dehydrated in 70% ethanol, embedded in paraffin and coronally sectioned (8μm thick). Sections were stained with haematoxylin and eosin according to previously published methods. 27

| Transmission electron microscopy (TEM)
Fresh left ventricular tissues were carefully kept in a relaxed and slightly stretched state during sample preparation and sectioned into 1 mm tissue blocks, then immediately immersed in ice-cold 2.5% glutaraldehyde to fix 1 h, at 4°C. Blocks were repeatedly washed in phosphate buffer and postfixed in 1% osmium tetroxide for 1 h. Samples were dehydrated in increasing concentrations of ethanol, starting with 10%, to 50%, 70% and 90% of ethanol, then embedded in 100% epoxy resin and left to polymerize at 55°C in 5% CO2 for 36 h. The resin blocks were then sectioned with an ultramicrotome. The ultrathin sections were placed on the grids, stained with uranyl acetate and lead citrate solution for TEM observation (JOEL TEM1230).

| Adult cardiomyocyte isolation
Cardiomyocytes were isolated as previously described, 27

| Cardiomyocyte contractility measurements
Cardiomyocyte contractility was measured as previously described. 29 Isolated cardiomyocytes were placed into a thermostatically controlled chamber with oxygenated Krebs balanced salt solution (pH 7.4) containing dextrose and Ca 2+ (1 mmol/L) at 37°C and imaged with an inverted microscope. Individual cardiomyocytes were selected for analysis on the basis of a characteristic rod-shaped morphology with no membrane blebbing and quiescence in the presence of extracellular Ca2 + (l mmol/L). The changes in light intensity at the cell edges were used to track cell motion. All parameters were calculated for each contraction, and the results were shown as average of all observed contractions.

| Isolation and transfection of neonatal rat ventricular myocytes.
Ventricles from neonatal rats were separated from the atria, cut into small pieces and then dissociated in Ca 2+ -free HBSS containing 0.125 mg/ml trypsin (Gibco), 10 mg, ml DNase II (Sigma) and 0.1 mg/ml collagenase type IV (Sigma). Digestion was performed at 37°C by stirring the digestion solution containing the heart sections throughout the repeated 5-min period of digestion for 8-10 times. The supernatant was collected with FBS (Gibco) after each digestion period to terminate the digestion. The cell pellets were resuspended in DMEM (Gibco) supplemented with 10% FBS and with 100 mM 5-bromo-20-deoxyuridine (Sigma) and seeded onto 100-mm plastic dishes for 2 h at 37°C in a 5% CO 2 and humidified atmosphere. The supernatant was the plated onto 1% gelatin (Sigma)-coated dishes. Twenty-four hours after the seeding, the medium was changed to DMEM (Gibco) containing 2% FBS (Gibco), 1% insulin-transferrin-selenium (ITS; Gibco),

| Sequencing data analysis
General pre-processing of reads: all samples were sequenced by Illumina HiseqX10 or Hiseq4000 with paired-end 150bp reads generated for analysis. Reads were quality controlled using fastqc and aligned to GRCm38 using HISAT2 31 with default parameters. All big-Wig files were obtained by using bamCoverage in deepTools. 32 For RNA-Seq, read counts were quantified using feature counts. 33 Differential expression testing was performed with R package DESeq2. 34 Genes with adjusted P values less than 0.05 and Fold Change >1.5 were marked as differentially expressed genes (DE genes). AS analysis was performed by using rMATS. 35  Metagene plot of m 6 A was performed by using Guitar. 40 Venn plots were conducted with TBTools 41 or R package eulerr. Other plots were obtained by using R package eulerr. All data generated or analysed during this study are included either in this article or in the supplemental information files.

| Statistics
All the data were presented as mean ±SEM. All statistical analyses were carried out using GraphPad Prism 8. Two-tailed Student's t test or nonparametric Mann-Whitney test was used for comparisons between two groups, and two-way ANOVA was used to compare the difference amongst multiple groups. Statistical significance was defined as a p value less than 0.05.

| Generation of cardiac-specific Ythdc1 conditional knockout mice
To understand the functional involvement of Ythdc1 in postnatal heart and potential contribution to heart diseases, we generated a conditional deletion allele of the mouse Ythdc1 gene. Two loxP sites were engineered to flank a region of Ythdc1 genomic DNA encoding exons 5-7. Mice carrying conditional Ythdc1 allele were crossed with mice carrying the Cre recombinase gene driven by the α-Mhc promoter ( Figure 1A), which expresses Cre recombinase in heart after birth. Cre-mediated recombination deleted Ythdc1 exon 5-7 between the loxP sites. This recombination was confirmed in the DNA of the Ythdc1-cKO mice. As shown in Figure 1B, homozygous Ythdc1-cKO mice were given the expected 391 bp mutant fragment compared with a 217 bp wild-type (WT) fragment. The RT-qPCR and Western blot analysis were used to evaluate the gene knockout efficiency of Ythdc1 at mRNA and protein levels, respectively.
The results showed a significant reduction of Ythdc1 expression in cKO mice both in mRNA level ( Figure 1C) and in protein level ( Figure 1D). Despite the newborn homozygous Ythdc1-cKO mice were viable and exhibited the expected mendelian ratios ( Figure 1E), all of them dead in around 10 weeks ( Figure 1F).
Likewise, to know the rest m 6 A 'reader' proteins' function in heart, we also generated the other three genetic knockout mice, including Ythdf1-KO, Ythdf2-cKO and Ythdf3-KO (Figure S1A-S1C), to allow assessment the role of each m 6 A 'reader' protein family member in cardiac function. We did not include Ythdc2, as its expression is barely detected in heart. As shown in Figure S2, the reduction level of Ythdf1  Figure S2E, S2F) was also assured by Western blot. Amongst all the four mice models, we found that only Ythdc1-cKO mice were extremely susceptible to premature death, whilst littermates of the rest YTH family members did not display abnormal phenotype after knockout ( Figure 1F). These results implicated that YTHDC1 plays an essential role in postnatal heart.

| Loss of cardiac Ythdc1 in the heart leads to DCM
To interrogate the exact cause of premature death, we evaluated the hearts by cardiac histopathology and echocardiography on Ythdc1-

| YTHDC1 deficiency attenuates the contraction of cardiomyocytes.
Considering that inability to produce sufficient contraction is the hallmark of DCM and heart failure, we measured the contractility of single cardiomyocytes isolated from the ventricle of 8-week-old Ctrl and cKO mice ( Figure 3A). We observed that YTHDC1 deficiency significantly prolonged the relaxing time of cardiomyocytes ( Figure 3B) and reduced the amplitude of length shortening ( Figure 3C).
To further confirm the finding in Ythdc1 deficiency mice and circumvent the off-target effects of the CRISPR gene knockout strategy, we subsequently measured the conduction of cultured cardiomyocytes through the multiple microelectrode array (MEA) recording. We first determined the efficiency of gene knockdown by Western blots, confirming that more than 50% reduction of YTHDC1 expression at protein level ( Figure S4A, S4B). And then we found that knockdown of YTHDC1 significantly decreased the contraction amplitude of the cardiomyocyte monolayers ( Figure 3D, 3E), implying that YTHDC1 is critical for regulating the contraction of cardiomyocytes.

| Multidimensional sequencing identifies Titin as a direct downstream target of YTHDC1
We next sought to investigate the molecular mechanism by which  Figure S5A). Then, we performed YTHDC1 RIPseq with mice at 2 weeks of age to identify YTHDC1-binding targets in heart and found about 860 hits overlapped with transcripts identified by m 6 A-seq which were defined as direct downstream targets of YTHDC1 ( Figure 4C). GO analysis showed that these genes were mainly enriched in chromatin organisation, actin filament-based process, heart development and heart morphogenesis ( Figure 4C).
Given that AS is an important function of YTHDC1, 16 we then performed mRNA-seq to identify the AS difference of the heart tissues between Ctrl and cKO mice. To avoid the altered transcriptional regulation caused by end-stage heart failure, we carried out the sequencing with mice at 2 and 8 weeks of age and analysed the overlapping dysregulated genes. By using rMATS, aberrant splicing targets of Ythdc1-cKO were obtained for further functional characterizations ( Figure 4D).
Taking these analyses together, we narrowed down targets to 42 high-confidence transcripts, which might be potential key genes in the development of YTHDC1-dependent DCM ( Figure 4E). GO analysis revealed enrichment of these genes involved in striated muscle cell development. More importantly, we found TTN, whose

| YTHDC1 deficiency increases the expression ratio of N2BA to N2B isoform of Titin
TITIN is a sarcomeric protein that determines the structure and biomechanical properties of striated muscle, and its defect is directly associated with DCM. 43 In the myocardium, its differential splicing leads to the expression of N2B and N2BA isoforms that differ in size. 26 Smaller mammals express predominantly N2B, whereas both N2B and N2BA are readily detectable in larger mammals, including humans. 44 Because of its shorter extensible I-band region, dominant expression of N2B results in higher passive myocardial stiffness than that of N2BA. Therefore, from RNA-Seq data, we further performed extensive analysis of alternative exon regulation of Titin with the Percent Spliced In (PSI) plot, showing the mean exon inclusion and exclusion.
Our data illustrated that Titin in cKO mice was predominantly spliced in the elastic region extending from the N2B to the PEVK segments, between which the N2A signalling domain is located ( Figure 5A). The alternatively spliced exons reside in the elastic PEVK and the immunoglobulin-rich region within the I-band, which may help to explain the increased distensibility of sarcomeres in YTHDC1 deficiency cardiomyocytes. 26 These findings implicated that loss of Ythdc1 induces aberrant splicing of Titin, leading to an increased ratio of N2BA:N2B isoform, which can directly result in DCM. We then examined the expression of the N2BA and N2B RNA isoforms in heart tissues from YTHDC1 deficiency mice and control by qRT-PCR. As illustrated in Figure 5B, 5C in agreement with the deep sequencing results, the N2BA isoform was highly expressed in YTHDC1 deficiency hearts, whereas no difference was observed in N2B isoform. In addition, high-resolution SDS-agarose gels also revealed that the ratio of N2BA:N2B protein in YTHDC1 deficiency hearts was dramatically higher than that in controls due to an increase in the more compliant N2BA isoform ( Figure 5D-5E

| DISCUSS ION
We demonstrated an essential role of YTHDC1 in cardiomyocytes for the maintenance of normal heart function, with loss of YTHDC1 leading to DCM. First, the Ythdc1-cKO mice exhibit early DCM, which ultimately proceeds to heart failure and postnatal lethality. Second, the Ythdc1-cKO cardiomyocytes manifested disrupted myofilament and abnormal contraction function. Ythdc1-induced alternative splicing events from mouse hearts (red, two weeks of age; blue, eight weeks of age). Right, GO analysis showed enrichment for these genes encoding proteins involved in striated muscle cell development expressed and the N2BA has less passive stiffness compared to N2B due to its larger size and increased elasticity. 45 It has been reported that an increased ratio of N2BA:N2B isoform of Titin can directly result in DCM. 26 However, the regulator that determines Titin splicing remains largely unknown. Recent study has identified RNA binding protein 20 as a splicing factor of TTN both in human and rat, 46,47 which again emphasises the key role of post-transcriptional regulation in cardiac function. Our study suggested a totally novel role of YTHDC1 in Titin splicing regulation, which supplemented the mode of TTN regulation.
Thus far, the importance of reversible m 6 A modifications on regulation of gene expression has been unveiled in different systems. 48 YTHDF2 modulates neural development by promoting m 6 A-dependent degradation of neural development-related mRNA targets. 49 Paris et al. found that inhibition of YTHDF2 specifically compromises the propagation of leukemic stem cells. 50  (C) Pooled data from B demonstrating that only the Titin N2BA isoform-specific exon was highly expressed in the homozygous mutant. N2B isoform observed no significant difference between Ctrl and Ythdc1-cKO group. The data are shown as means ±SEM. n = 3. * p < 0.01 compared with ctrl group by unpaired t test. (D) SDS-agarose gel electrophoresis of protein lysates from control and Ythdc1-cKO left ventricular myocardium, then gel was stained with Coomassie blue. N2BA bands are broader in Ythdc1-cKO samples than in controls. T2 is a minor degradation product. (E) Pooled data from D showing the ratio of N2BA to N2B, T2 to TT and TT to MHC from left ventricular homogenates (p < 0.001). Data are means ±SEM. *p < 0.01 compared with ctrl by unpaired t test reported that YTHDF3 overexpression is associated with brain metastasis of breast tumour, resulting in poor survival. However, as an important member in the m 6 A readers, the role YTHDC1 in heart is not clear. Our studies reported YTHDC1 participates in the onset and progression of DCM, providing new evidence that RNA modification involves in the processing of heart disease.
DCM is one of the most common causes of heart failure. 52 For patients with established DCM, treatment is directed at alleviating the major clinical manifestations of heart failure and arrhythmias. However, management of heart failure remains an unmet need. 23,53 Our study demonstrated that YTHDC1-depended Titin splicing is crucial for the postnatal heart development and normal cardiac function, which probably provides a potential target for treating DCM through tuning m 6 A modification of Titin mRNA.
Unfortunately, due to the limitation of single-base resolution m 6 A sequencing technique, we did not identify the precise m 6 A site of Titin. Meanwhile, we did not completely exclude m 6 A-independent pathway, such as chromatin organisation, which might also conduct the Ythdc1 deficiency-induced DCM. It will be of great interest to determine whether the dysfunctional Ythdc1 gene is correlated with cardiac malfunction in human, which may facilitate the development of new therapeutic strategies for reversing DCM.
In summary, this study highlights the close relationship between the homeostasis imbalance of mRNA m 6 A modification and the occurrence and development of DCM. YTHDC1-dependent Titin splicing is a potential new therapeutic target for DCM.