Kmt2d ^flox/flox mice (Lee et al., 2013) were obtained from Dr. Kai Ge (National Institute of Diabetes and Digestive and Kidney Diseases, NIH, United States). Tnnt2-rtTA/tetO-Cre (cardiomyocyte-specific Cre) mice (Wu et al., 2010) were kindly donated by Dr. Bin Zhou (Shanghai Institutes for Biological Sciences of the Chinese Academy of Sciences, China). Female Kmt2d ^flox/flox mice were crossed with Tnnt2-Cre male mice and bred in the approved Experimental Animal Center at West China Hospital of Sichuan University (Chengdu, China) to generate the Kmt2d ^flox/flox ; Tnnt2-rtTA/tetO-Cre mice. The mice were housed in specific pathogen-free conditions (3-5 mice/cage, temperature at 20–26°C, relative humidity at 40–70% and 12 h dark/light cycle) and given free access to sterile feed and water. Genotyping procedure was performed by polymerase chain reaction (PCR) using genomic DNA. The sequences of relative primers are listed in Supplementary Table S1.

Eight-week-old Kmt2d ^flox/flox ; Tnnt2-rtTA/tetO-Cre male mice were fed with doxycycline (Dox) at a concentration of 2 mg/ml (Chen et al., 2018) in drinking water for 14 days to knockout Kmt2d in cardiomyocytes (KMT2D-cKO). One week after Dox withdrawal, experiments were carried out. Their Kmt2d ^flox/flox without Tnnt2-Cre littermates were used as control (KMT2D-Ctl).

All animal experimental protocols were reviewed and approved by the Institutional Animal Care and Use Committees of West China Hospital of Sichuan University (No. 2020267A).

KMT2D-Ctl and KMT2D-cKO mice were anaesthetized with isoflurane, and fixed in a supine position. The neck and chest of mice were cleaned and disinfected with 75% absolute ethyl alcohol. The jugular anterior fascicles were bluntly dissected along the midline to expose the trachea through a central incision in the neck. The mice were intubated in trachea and the ventilator was used to assist breathing. An incision was created on left side of the sternum and the thorax was opened between the third and fourth costal cartilages. The heart was exposed, then the left anterior descending coronary artery was visualized using stereoscopic microscope and ligated with 7–0 silk suture. Schematic diagram visualized ligation of left anterior descending coronary artery for artificial myocardial ischemia (MI) mice model (Supplementary Figure S1A). Successful myocardial ischemia modeling was confirmed by epicardial cyanosis and raised ST-segment of the electrocardiogram (Supplementary Figures S1B,C). The chest and skin were then closed with 6–0 silk suture.

Mice were divided into four groups: control mice sham operation (KMT2D-Ctl-sham), control mice myocardial ischemia (KMT2D-Ctl-MI), KMT2D-cKO mice sham operation (KMT2D-cKO-sham), and KMT2D-cKO mice myocardial ischemia (KMT2D-cKO-MI). In this study, the 24-h time point of myocardial ischemia was selected. Myocardial tissues were harvested in the end of ischemia and the ischemic size was evaluated by TTC staining.

The excised hearts were immediately transferred to pre-cooled (4°C) phosphate buffered saline (PBS) solution to rinse blood, and frozen at −20°C for 30 min. Then each frozen heart was cut transversely into five 2-mm-thick slices. The slices were immersed in 1% TTC (Sigma-Aldrich, United States) in PBS solution, and incubated at 37°C for 15 min followed by treatment of 4% (v/v) paraformaldehyde at 4°C for 12 h. The infarcted tissues were unstained and appeared pale, while viable tissues showed as red (Guo et al., 2021). The infarcted size (IS) and total left ventricular (LV) area were measured by Image-Pro Plus 6.0 software. The infarct size percentage was quantified for each slide by using the following formulas: IS (%) = IS/total LV area × 100.

The hearts were harvested, washed with PBS, and fixed in 4% paraformaldehyde (Biosharp, Cat No. BL539A) at 4°C for 24 h, followed by incubation in 30% sucrose and then embedded in OCT at −80°C. Frozen sections (8-μm thickness) were processed and permeabilized using 0.3% Triton X-100 (Solarbio, Cat No. T8200) for 30 min, and then blocked with 1% bovine serum albumin (BioFroxx, Cat No. 4240GR) in PBS at 37°C for 1 h. The sections were incubated with primary antibodies: anti-TNNT2 (1:100, Abcam, Cat No. ab8295), anti-KMT2D (1:100, Merckmillipore, Cat No. ABE 1867), secondary antibody (1:500, Alexa Fluor 488, Proteintech, Cat No. SA00013-2; Alexa Fluor 568, Invitrogen, Cat No. A11031). The nuclear staining was performed with DAPI (Beyotime, Cat No. C1005). Immunofluorescences images were visualized using an OLYMPUS microscope BX43.

Heart tissue was ground into powder in liquid nitrogen and total RNA was isolated using TRIzol™ Reagent (Invitrogen™, Cat No. 15596026) according to the instructions. RNA purity and concentrations were confirmed by spectrophotometry (absorbance at 260/280nm; Thermo Scientific™ NanoDrop™ One).

Reverse transcription of total RNA was performed using the HiScript II Q RT SuperMix (Vazyme, Cat No. R223-01). The qPCR was performed by CFX96™ Real-Time System (Bio-Rad Laboratories, United States), according to the instructions of TB Green^® Premix (Takara, Cat No. RR820A). The quantification of target gene in expression was calculated using the 2^−ΔΔCt method, and beta-Actin was used as a normalized reference. The primers were shown in Supplementary Table S2.

Heart function was determined 7 days after MI using a small animal high-resolution ultrasound (VIVID7, GE) equipped with a i13L imaging transducer. Briefly, mice were anesthetized with 1–2% isoflurane and chest hairs were removed. The left ventricular end-diastolic diameter (LVEDD) and left ventricular end-systolic diameter (LVESD) were recorded and measured in M-mode. The left ventricular ejection fraction (EF) and fractional shortening (FS) were calculated according to the formula [(LVEDD)³ − (LVESD)³]/(LVEDD)³ × 100% and (LVEDD − LVESD)/LVEDD × 100%, respectively.

The total RNA was extracted from the heart. After quality quantification, RNA sequencing was conducted by Shanghai Liebing Biomedical Technology Co., Ltd. The FPKM (fragments per kilobase of exon model per million mapped fragments) data with a large deviation from the mean and the FPKM with zero values were discarded. The remaining FPKM data were used for describing advanced volcano plot, venn diagram and heatmap of gene expression.

The genes at genome-wide significant CAD risk loci were obtained from previous study (Erdmann et al., 2018). The orthologs in mice were identified by Gene ID Conversion Tool from DAVID (Huang et al., 2009a; Huang et al., 2009b).

The cardiac samples were ground in liquid nitrogen and lysed with RIPA buffer (Beyotime, Cat No. P0013) with protease inhibitor cocktail (Roche, Cat No. 04693116001). Equivalent amounts of protein were separated by sodium dodecyl sulphate-polyacrylamide gel electrophoresis (SDS-PAGE), and transferred onto PVDF membranes (Merck Millipore, Cat No. ISEQ00010). Nonspecific reactivity was blocked in 5% skim milk in PBST (0.01M phosphate buffer solution, pH 7.2, 0.6% Tween-20) for 1 h at room temperature. The membranes were incubated with the specific primary antibodies against KMT2D (1:1000, Merck Millipore, Cat No. ABE 1867), RASD1 (1:800, Abcam, Cat No. ab251924) and β-Actin (1:1000, ZSGB-BIO, Cat No. TA-09) overnight at 4°C, then with the horseradish peroxidase (HRP)–conjugated secondary antibodies (1:3000, Invitrogen, Cat No. #31430 and #31460) for 3 h at room temperature. Signals were detected by the FUSION FX system (VILBER, France) using SuperSignal™ West Pico PLUS (Thermo, Cat No. 34580). The β-Actin was used as a normalized reference to assess protein levels.

The 24 h post-sham/MI-operated tissues of hearts were harvested and ground into powder in liquid nitrogen. The powdered tissues were lysed and chromatin was extracted using the SimpleChIP^® Plus Sonication Chromatin IP Kit (Cell Signaling Technology, Cat No. #56383). The chromatin was sheared by sonication and genomic DNA was immunoprecipitated with a ChIP-grade anti-H3K4me1 antibody (Abcam, Cat No. ab8895) following the SimpleChIP^® Plus Sonication Chromatin IP Kit. The qPCR was performed by CFX96™ Real-Time System, according to the instructions of TB Green^® Premix (Takara, Cat No. RR820A). The quantification was calculated using the 2^−ΔΔCt method, and input DNA was used as a normalized reference. The primers were shown in Supplementary Table S3.

H9C2 cells were obtained from Procell Life Science Technology Co., Ltd. (Wuhan, China) and cultured at 37 °C and 5% CO₂ in DMEM with L-glutamine medium (Gibco, Cat No. C11995500BT), 1% penicillin–streptomycin (Gibco, Cat No. 15140-122), with 10% fetal bovine serum (FBS, Gibco, Cat No. 16140-071), unless noted otherwise, in 60 mm dishes (Thermo Scientific, Cat No. 150462), as previously described (Pesant et al., 2006) with some modifications. When the cells grew to 70% confluent, fresh culture medium was replaced absolutely. Twenty-4 hours later, cells were treated with 1% O₂ 8 h for hypoxic studies or serum-free medium 2–3 h for serum starvation experiments. The control cells were concomitantly cultured in normal conditions.

The oligodeoxynucleotides of gRNAs were annealed and cloned into PX458 plasmids (Addgene, Cat No. #48138), named PX458-Kmt2d-gRNA1 and PX458-Kmt2d-gRNA2, respectively, and confirmed by sequencing. H9C2 cells were detached by 0.25% trypsin-EDTA solution (Gibco, Cat No. 25200-056) and transfected with CRISPR/Cas9 PX458-Kmt2d gRNA1/2 or empty control plasmids by Nucleofector™ X Kit (Lonza, Cat No. V4XC-2024) according to the manufacture’s protocol, then plated in DMEM with 10% FBS without pen/strep. Cells were cultured for 48 h at 37°C. The PX458 plasmid contained a green fluorescent protein (GFP) marker that can be used to screen GFP positive cells for H9C2-KO monoclonal H9C2 cell line by aseptic flow cytometry.

GraphPad Prism 6.01 was used to analyze the data. The measurement data were presented as mean ± standard deviation. The student’s t-test was used for two-group comparison. One-way analysis of variance (ANOVA) was used for comparison between more than two groups, and Tukey’s post hoc test for pairwise comparison. Predictors were kept if they were significant at a p value <0.05.

To evaluate the knockout efficiency of Kmt2d, adult myocardial tissues at the apex of the left ventricle were used to extract mRNAs and proteins. The qPCR and western blotting results showed that the products of Kmt2d gene were significantly reduced in the myocardium of KMT2D-cKO mice, to about 50% of those in the KMT2D-Ctl mice (Figures 1A–C). Immunofluorescence staining was used to confirm the specificity of KMT2D knockout in the cardiomyocytes. In the KMT2D-Ctl mice, KMT2D was co-expressed with TNNT2 in the heart tissues (Supplementary Figure S2A). However, In the KMT2D-cKO mice, KMT2D was almost undetectable in the TNNT2 positive cardiomyocytes (Supplementary Figure S2B). The mortality of myocardial ischemia within 24 h was calculated, and the result shows that the death rate in KMT2D-cKO mice was significantly higher (26.5%) than that in the KMT2D-Ctl group (12.5%), as shown in Table1.

Twenty-four hours after MI, adult hearts were harvested and the ischemic areas were analyzed by TTC staining. The ratio of infarct size to left ventricular (IS/LV) was calculated to evaluate the myocardial injury. The results show that the proportion of ischemic area in left ventricle in KMT2D-cKO mice was significantly higher than that in the KMT2D-Ctl group (Figures 1D,E). The left ventricular systolic function was assessed by echocardiography. The left ventricular ejection fraction (EF) and fractional shortening (FS) were significantly decreased in the KMT2D-cKO group compared to the KMT2D-Ctl mice 7 days post-MI (Figures 1F–H). These data indicate that specific knockout of Kmt2d in cardiomyocytes significantly increased the mouse mortality after MI, while the proportion of ischemic area in the left ventricle was increased as well, and the left ventricular systolic function was weakened, suggesting that Kmt2d is indispensable for the improved tolerance to cardiac ischemia in adult mice.

To study the causes of aggravation of cell injury induced by KMT2D deletion after ischemia, we performed RNA-seq and determined transcriptional profiles in the sham and the 24-hours-post-MI hearts in the KMT2D-cKO and the KMT2D-Ctl mice. Transcriptomic analysis revealed 1,621 genes differentially expressed ( | log 2 ( F o l d C h a n g e ) | >1) in the MI group compared to the sham group of the KMT2D-Ctl mice, including 805 up-regulated and 816 down-regulated genes (Figure 2A; p < 0.05, FDR<0.1). The expression of 1,209 genes was significantly altered ( | log 2 ( F o l d C h a n g e ) | >1) by MI in the KMT2D-cKO mice, including 652 up-regulated and 557 down-regulated genes (Figure 2B; p < 0.05, FDR<0.1). Among the differential expressed genes (DEGs), 862 genes were specifically altered by MI in KMT2D-Ctl mice, while 450 genes were uniquely changed after MI in the KMT2D-cKO group (Figure 2C).

Then we focused on genes whose expression was altered by MI in the KMT2D-Ctl mice but not in the KMT2D-cKO mice, or vice versa. Top 115 of the 862 DEGs and top 60 of the 450 DEGs were identified, respectively (Figure 2D, Supplementary Table S4). To further screen out the truly important genes, we then examined if any of these genes was associated with coronary artery disease by analyzing the data from genome-wide associated studies (GWAS) (Erdmann et al., 2018). In the NCBI Homo sapiens database, 364 genes at the genomic significant CAD risk loci were identified with the gene IDs (Supplementary Table S5), and 359 of them have murine orthologs (Supplementary Table S6). Among these genes, venn analysis showed that four DEGs were in our list, namely Rasd1 (Ras Related Dexamethasone Induced Protein 1), Thsd7a (Thrombospondin, type I, domain containing 7A), Ednra (Endothelin Receptor Type A), and Tns1 (Tensin 1) (Figure 2E). The expression of the Rasd1 gene was significantly decreased by MI or the loss of KMT2D (Figure 2F).

To further verify our findings, we collected ischemic heart tissues, including the infarct area and the border zone, from the 24-hours-MI and sham-operated group of the KMT2D-cKO and the KMT2D-Ctl mice. Down-regulated expression of Rasd1 was detected in 24-hours-MI hearts compared with sham-operated ones in the KMT2D-Ctl mice at both mRNA (Figure 3A) and protein (Figures 3B,D) levels by qRT-PCR and western blot, respectively. Interestingly, the loss of KMT2D also led to a significant decrease in Rasd1 expression in the heart (Figures 3A–D), consisting with the RNA-seq result. However, the transcription of Rasd1 was not changed in the 24-hours-MI compared to the sham-operated hearts in the KMT2D-cKO mice (Figures 3A,D). The results suggested that KMT2D might be one of the upstream regulators of Rasd1 gene.

KMT2D is one of the prominent H3K4 mono-methyltransferases shaping enhancers. Therefore, we speculated that the effect of KMT2D on Rasd1 expression might be through H3K4 methylation regulation on the enhancer of Rasd1. Previous study identified a glucocorticoid response element (GRE) located about 2,300bp at the 3′-end of the Rasd1 gene, which is a critical enhancer of Rasd1 in human AtT-20 cells (Kemppainen et al., 2003). Interestingly, a GRE consensus element, GGTACATACTGTTCC was also located approximately 2,790bp downstream of the murine Rasd1 (Figure 3E). Therefore, it was speculated that this sequence might be the core GRE fragment of Rasd1 enhancer in mice. ChIP-qPCR results illustrated that, compared with sham-operated KMT2D-Ctl mice, H3K4me1 on the Rasd1 GRE was significantly decreased in MI-operated KMT2D-Ctl mice, sham-operated KMT2D-cKO mice and MI-operated KMT2D-cKO mice (Figure 3F), suggesting that either Kmt2d deletion or MI in cardiomyocytes could lead to the reduction of H3K4me1 in GRE of the murine Rasd1 enhancer, and further reduced Rasd1 transcription in the heart tissue.

To verify our findings and specifically investigate the effect of KMT2D on Rasd1 expression in cell culture, Kmt2d gene was knocked out in the H9C2 cell line using CRISPR/Cas9 system in vitro. The sequences, spanning 16th-19th exons, were the knockout fragment of Kmt2d in our transgenic mouse model. Therefore, CRISPR guide RNAs (gRNAs) were designed to target the 16th exon (gRNA-1: 5′-GCT​TAA​GGG​CTG​GCG​TTG​TGt​gg-3′, gRNA-2: 5′-CTT​GCA​CTT​CCA​GCC​ACC​CTt​gg-3′) of Kmt2d in H9C2 (Figure 4A). After monoclonal selection, H9C2-KO cells were successfully established and confirmed by genomic sequencing (Figure 4B). The mRNA and protein of Kmt2d were significantly reduced in the H9C2-KO cells, compared with control H9C2 cells (Figures 4C–F). In line with our finding in vivo, expression of Rasd1 was also weakened in H9C2-KO cells (Figures 4D,E,G). These results further confirmed that KMT2D regulates the expression of Rasd1.

To further investigate whether the protective effect of KMT2D on MI was associated with RASD1, H9C2 cells were treated with 1% oxygen to simulate the hypoxic injury during MI. The Kmt2d expression in the control H9C2 was slightly increased in hypoxia (1% O₂, 8 h) compared to the normoxia (Figures 5A,C,D). However, in the control H9C2, there was no significant differences of Rasd1 expression between hypoxia and normoxia conditions (Figures 5B,C,E). In the H9C2-KO cell line, both mRNA and protein levels of Kmt2d and Rasd1 were significantly decreased, compared with the control H9C2, but were also not affected by hypoxia (Figures 5B,C,E). These results suggested that down-regulation of Rasd1 expression in ischemia might not be caused by hypoxia.

During ischemia, adult cells indeed were not only deprived of oxygen, but also deficient in nutrition from serum. Therefore, we then treated H9C2 cells with serum-free medium to mimic nutritional deficiency in myocardial ischemia. After 2 h of serum starvation, the mRNA level of Kmt2d was significantly reduced compared with the normal culture conditions in control H9C2 cells (Figure 6A), and a similar trend of Rasd1 was also observed (Figure 6B). Three hours later of serum deprivation, the protein levels of KMT2D and RASD1 were both downregulated in the serum free condition from the H9C2-Ctl cells (Figures 6C–E). In the H9C2-KO cells, expressions of Kmt2d and Rasd1 maintained at a low level within serum or not, and were significantly lower than those in the H9C2-Ctl group (Figures 6C–E). These data suggest that serum depletion downregulated both Kmt2d and Rasd1 expression in adult cardiomyocytes, and that Rasd1 was only sensitive to serum free but not to hypoxia.