The Specific Pathogen Free (SPF)-grade male wild type (WT) SD rats (200–250 g) and neonatal rats (1–3 days) were provided by Wuhan University Experiment Animal Center. The rats were hosted in an environment with 50–60% humidity and a 12 h:12 h light–dark cycle at 22–24°C, and with free access to food and water. The animal experiments were performed in strict accordance with the published NIH Guidelines for the Care and Use of Laboratory Animals and approved by the Animal Care and Use Committee of Wuhan University.

A CRISPR/Cas9 genome-editing technology was operated to generate KDM3A global knock-out rats (KDM3A-KO rat, SD background) as described in our previous study (25). Briefly, one single guide RNA (sgRNA) flanked exon 5 of the KDM3A gene in rat was designed and created. When purified Cas9 mRNA and sgRNA were mixed and microinjected into embryos, PCR products were TA cloned and sequenced to define the exacted indel mutations of generated founders. The primers were listed as follows: KDM3A F: 5′-TCCCTGGAGTTGCAGTAGTTT-3′; KDM3A R: 5′-AGGTCTATCACTGGCTAACTCA-3′. The founder rat was chosen as the breeding line and mated with the WT (kdm3a^+/+) SD rat strain to generate F1 heterozygote (kdm3a^±). The homozygous kdm3a^–/– strain was obtained through sibling mating of heterozygous F1 offspring and was identified by sequencing chromatograms of PCR products. The western blotting assay was also performed to validate KDM3A protein expression.

The rats received a high-fat diet for 4 weeks and were intraperitoneally injected with a low dose of streptozocin (STZ) (30 mg/kg) to establish a reliable type 2 diabetes mellitus (T2DM) model (fasting blood glucose >13.9 mmol/L). Subsequently, these T2DM rats were fed with a high-fat diet for another 16 weeks and received subcutaneous injections of saline daily from the time point of 8 weeks to construct an in vivo hyperglycemia myocardial damage model (DM group). Meanwhile, T2DM rats were fed with a high-fat diet for another 16 weeks and received subcutaneous injections of insulin (Novolin N, 3 U/d) from the time point of 8 weeks to establish an in vivo intensive GC model (DM–GC). The rats in the Sham group were fed with a normal diet for 20 consecutive weeks.

The primary cardiomyocytes were isolated from the heart of 1–3-day-old rats by differential adhesion method and identified by morphology and immunofluorescence as previously described (25). Subsequently, the cells in the normal glucose (NG) group were incubated with Dulbecco’s modified Eagle’s medium (DMEM)/F12 containing 5.5-mM glucose for 72 h. Cells in the HG group were treated with DMEM/F12 containing 30-mM glucose for 72 h to establish an in vitro hyperglycemia myocardial injury model. Cardiomyocytes in the HG–NG group were cultivated with a medium containing 30-mM glucose for 24 h followed by incubated with a medium containing 5.5-mM glucose for 48 h to establish a diabetic intensive GC model in vitro.

The short-hairpin RNA adenoviruses AdshKDM3A (KDM3A inhibition) and its negative control AdshRNA, as well as AdKDM3A (KDM3A over-expression) and its negative control AdGFP, were constructed by Shanghai GeneChem Co., Ltd. The primary cardiomyocytes were seeded into a 6-well plate at 10⁵/well and infected with various multiplicities of infection (MOI). The cells were transiently transfected for 2 h and continuously transfected for 48 h after the original medium was replaced by the complete medium. Finally, the number of GFP-positive cells and their activities were observed under a fluorescence microscope, and the cells with an optimal MOI value were selected for the subsequent experiments.

The siNF-κB/P65 plasmid was synthesized by RiboBio (Guangzhoug, China) and transfected with Lipofectamine 2000. The primary cardiomyocytes were transfected with siNF-κB/P65 plasmid for 6 h followed by incubated with normal medium for another 48 h.

At the cellular level, the primary cardiomyocytes were incubated at 37°C for 30 min with a dihydroethidium (DHE) fluorescent probe (Beyotime Institute of Biotechnology Co., Ltd., China), at a final concentration of 10 μmol/L. Under the fluorescence microscope, ROS-positive cells were dyed in red in the whole nuclear region. Later, at least five fields of view were randomly selected, photographed, and analyzed by the ImageJ software. Besides, a flow cytometry assay (FACS Calibur; BD Biosciences, United States) was also used to analyze the percentage of DHE positive cells. To further detect the intercellular superoxide generation and lipid peroxidation, the malondialdehyde (MDA) concentration and the superoxide dismutase (SOD) enzymatic activity in cardiomyocytes were also measured, in accordance with the descriptions of commercial kits (Nanjing Jiancheng Bioengineering Institute, China). To evaluate ROS generation in myocardial tissue, fresh frozen left ventricular samples (5-μm sections) were incubated with DHE (2 mmol/L) for 1 h at room temperature in the dark. The slides were visualized by fluorescence microscopy, and the mean fluorescence intensity (MFI) of each slide was quantified by Image-Pro Plus software version 6.0. The MDA concentration and SOD enzymatic activity in fresh myocardial tissue homogenate were also measured in the same manner as that in vitro.

Flow cytometry was used to estimate apoptotic cells as previous demonstrations (27). The cardiomyocytes in different groups were harvested and resuspended in 1 × binding puffer at a final concentration of 1 × 10⁶ cells/ml. Then, 5 μl of 7-aminoactinomycin D (7-AAD) was added into the cell suspension and then the cells underwent further staining with Annexin V-APC (AV-APC) dye, followed by the fluorescence-activated cell sorting on a flow cytometric assay (CytoFLEX; Beckman Coulter, Inc., Brea, CA, United States). The Annexin V-APC-positive cells were identified as early apoptosis, while both annexin V-APC- and 7-AAD-positive cells were identified as late apoptosis. The total apoptosis rate comprised the proportions of early apoptotic cells and late apoptotic cells.

The expression of inflammatory cytokine was detected by commercialized enzyme-linked immunosorbent assay (ELISA) kits (Nanjing Jiancheng Bioengineering Institute, Nanjing, China). The cardiomyocytes and fresh myocardial tissues were collected and crushed into homogenate to examine the levels of pro-inflammatory mediator tumor necrosis factor-alpha (TNF-α) and Interleukin-6 (IL-6) according to the manufacturer’s instructions.

The myocardial tissues were fixed with 4% paraformaldehyde for 72 h, dehydrated, transparentized, and embedded in paraffin to prepare 4-μm slices. The paraffin sections were stained with an in situ TUNEL detection kit (Roche, Basel, Switzerland) to detect apoptotic cells as previously described (28). Red fluorescent dye and 4′,6-Diamidino-2′-phenylindole (DAPI) were performed to specifically mark apoptotic nuclei and overall nuclei, respectively. The TUNEL-positive nuclei (red) and total cells (blue) were photographed with a fluorescent microscope (Olympus, Tokyo, Japan) in at least five randomly selected fields per section (200 × magnification). The apoptotic index was expressed as the ratio of apoptotic cells over total myocytes of DAPI-dyed nuclei. Hyperglycemia-induced myocardial remodeling after infarction was determined by Masson’s staining, areas stained in blue indicate the deposition of collagen fibrils. Hematoxylin and eosin (H&E) and wheat germ agglutinin (WGA) staining were also performed to detect cardiac hypertrophy as previously described (25).

Transthoracic echocardiography was carried out to determine cardiac function after animals were lightly anesthetized by inhaling 1.5–2.5% isoflurane. The transthoracic echocardiograms were recorded using MyLab 30CV ultrasound system (Biosound Easote, Inc.). Two-dimensional (2D) images were collected in the short axes with at least three representative cycles. Left ventricular (LV) M-mode tracing at mid-papillary levels was measured and averaged for the calculation of ejection fraction (EF) and factional shortening (FS).

The primary cardiomyocytes were transfected with AdKDM3A or AdshKDM3A. The pGL6-nuclear factor kappa-B (NF-κB) plasmid DNA, pRL-TK plasmid DNA (Beyotime, Shanghai), and Lipofect (Ribobio, Guangzhou) were prepared according to the manufacturer’s instructions and were used to transfect the cells. Subsequently, the cells were fully lysed and centrifuged to obtain the supernatant. Thereafter, 100 μl of Firefly luciferase detection reagents (Beyotime, Shanghai) was added to 50 μl of samples and the relative light unit (RLU) was determined on the machine. The reporter gene cell lysate was used as a blank control. Thereafter, 100 ml of Renilla luciferase working solution (Beyotime, Shanghai) was added to continuously measure the RLU and the ratio of Firefly luciferase/Renilla luciferase was calculated in each tube.

The proteins were extracted from the nucleus, and their concentration was determined by the bicinchoninic acid (BCA) assay (ASPEN). An electrophoresis gel was prepared according to the manufacturer’s instructions (Thermo Scientific), and the proteins were subjected to pre-electrophoresis and electrophoresis and transferred onto a membrane, followed by cross-linking for 10 min. After 15 min of blocking, the membrane was incubated with a diluted antibody for 15–20 min, washed with the eluent and equilibrated for 5 min using the equilibration solution. Finally, enhanced chemiluminescence (ECL) was used for image development, and the AlphaEaseFC software processing system was used to analyze the bands.

The Radio Immunoprecipitation Assay (RIPA) lysis buffer (Roche) was used to extract total proteins from the heart tissues or cells. After quantitative analysis by BCA, the samples were separated by 8–12% SDS–PAGE and then transferred onto PVDF membranes (Merck Millipore). After blocked by 5% skim milk, the membranes were incubated with corresponding primary (KDM3A ab106456, H3K9me2 ab1220, NF-κB/P65 ab194726) and secondary antibodies at the recommended dilution. The glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (ab37168) was chosen as an internal control. The protein bands were visualized using an ECL system (Thermos Fisher Scientific, Inc).

All data were analyzed using the SPSS 19.0 software package. The measured data were expressed as mean ± standard deviation. Student’s t-test was used to compare between two groups, and one-way analysis of variance (ANOVA) was applied for the comparison among multiple groups. The value, p < 0.05, represented a statistically significant difference. All results were independently reproduced at least 3 times with similar results.

According to the results of DHE staining, HG treatment results in evident accumulation of ROS in the cardiomyocyte. However, there was no significant difference in the MFI between the HG and the HG + NG groups (Figure 1A). Besides, as shown in Figure 1B, the HG obviously improved the level of MDA but decreased the level of SOD. Moreover, HG-induced oxidative stress continued to be activated even though the HG medium was replaced by an NG medium (Figures 1A,B). In addition, the results of flow cytometry and ELISA also showed that HG treatment remarkably facilitated the apoptotic ratio of cardiomyocytes as well as markedly increased expression of the IL-6 and TNF-α. However, the restoration of NG neither obviously reduce the apoptosis of cardiomyocytes nor distinctly suppress the expressions of inflammatory factors (Figures 1C,D). The above results suggest that HG treatment could induce continuous cardiomyocytes dysfunction, and the sustained activation of ROS, apoptosis and inflammatory responses might be the vital underlying pathophysiological mechanism.

Interestingly, as a key regulator of ROS, apoptosis, and inflammation, the expression of KDM3A was also gradually increased with the extension of HG treatment time and peaked at 6 h, which was maintained for 48 h. Meanwhile, the expression of H3K9me2 was gradually declined and reached the minimum value at 24 h, which was maintained for 48 h (Figure 1E). Moreover, even if the glucose concentration returned to normal, KDM3A was still highly expressed while H3K9me2 remain expressed at a relatively low level (Figure 1F). As a result, we proposed the hypothesis that KDM3A might be a crucial mediator of HG-induced long-lasting cardiomyocytes dysfunction.

To verify this hypothesis, recombinant adenovirus vectors with KDM3A over-expression (AdKDM3A) and knock-down (AdshKDM3A) were successfully constructed (Supplementary Figures 1A,B), which could meet the needs of subsequent experiments. According to the results of flow cytometry, cardiomyocytes in HG and HG–NG groups generated more ROS than that in the NG group. The KDM3A knock-down significantly reduced ROS generation in both HG group and HG–NG group, whereas KDM3A over-expression remarkedly increased ROS generation in both HG group and HG–NG group. Moreover, no differences were found in the number of DHE positive cells between the HG and HG–NG groups after up-regulating or down-regulating KDM3A expression (Figure 2A). Additionally, KDM3A knock-down or over-expression have no perceptible effect on the apoptosis and inflammatory responses of cardiomyocytes when incubated in the NG medium. However, down-regulation of KDM3A obviously decreased HG and HG–NG triggered cell apoptosis and reduced the expression of the pro-inflammatory mediators IL-6 and TNF-α. However, KDM3A up-regulation exerted the opposite effects. Noteworthy, there were no statistical differences in apoptosis rates and inflammatory cytokines expression between HG and HG–NG groups regardless of up-regulation or down-regulation of KDM3A (Figures 2B,C). Data above indicated that KDM3A inhibition could regulate high glucose-induced long-term cardiomyocytes injury by modulating ROS generation, apoptosis, and inflammation.

In our previous study, we demonstrated that KDM3A/NF-κB/P65 signaling pathway was closely associated with ROS generation, apoptosis, and inflammation in VSMCs, suggesting that KDM3A may be a key upstream regulator of NF-κB/P65 (27). In addition, the investigators also confirmed the continuous activation of transcriptional activity of NF-κB/P65 were involved in the process of hyperglycemia-induced persistent myocardial injury (29). The evidence discussed here indicates a potential mechanistic relationship between KDM3A and NF-κB/P65 activation in HG-induced long-lasting cardiomyocytes injury. Thus, western blots were performed, and the results showed that the expression of P65 protein was increased after HG and HG–NG treatments, which was accompanied by an obvious reduction in the H3K9me2 expression. In addition, KDM3A knock-down inhibited NF-κB/P65 transcription and further up-regulated H3K9me2 expression under HG and HG–NG conditions (Figure 3A). However, KDM3A over-expression exerted the opposite effects (Figure 3B). Furthermore, the results of the Electrophoretic Mobility Shift Assay (EMSA) (Figure 3C) and the dual-luciferase reporter gene assay (Figure 3D) also showed that KDM3A down-regulation weakened the DNA binding ability and transcriptional activity of NF-κB/P65 in cardiomyocytes under both HG and HG–NG conditions. Nevertheless, KDM3A over-expression exerted the opposite effects. These results demonstrate that KDM3A may participate in the progression of HG-induced cardiomyocytes injury by regulating the expression and transcriptional activity of NF-κB/P65.

In view of the results obtained in vitro, we would like to further validate these conclusions in vivo. The results of fasting blood glucose monitoring showed that blood glucose in the DM and DM + GC groups were significantly increased from week 4 to week 12 and maintained at a relatively high level (blood glucose >13.9 mmol/L). However, after administration with insulin from week 12, the blood glucose level in the DM + GC group returned to normal (Supplementary Figure 2A). The above results indicate that the DM rat model was successfully established, and insulin treatment obviously controls the blood glucose level.

Compared with rats in the control group, the LVEF and LVFS were markedly decreased in the DM group. However, intensive blood glucose control exerted no evident improvement on cardiac function as compared with the DM group (Figure 4A). Meanwhile, diabetes-induced marked myocardial hypertrophy and fibrosis; however, the intensive blood glucose control neither reduced the cross-sectional area of cardiomyocytes and HW/TL ratio (Figure 4B) nor alleviate the degree of myocardial fibrosis (Figure 4C). As a result, it can be concluded that hyperglycemia exerts sustained impairment on cardiac function and lead to extensive myocardial injury and cardiac remodeling even after the blood glucose level was tightly controlled.

Consistent with the results from in vitro, DHE fluorescent probes and MDA/SOD detection results revealed that even if the glucose level returned to normal, the generation of ROS induced by hyperglycemia was still maintained at a high level (Figures 4D,E). Besides, the results of TUNEL fluorescence staining indicated that intensive blood glucose control did not significantly reduce cell apoptosis (Figure 4F). Additionally, the TNF-α and IL-6 levels increased in varying degrees in the DM group compared with that in the control group, while the intensive glucose lowering treatment did not evidently inhibit expression of TNF-α and IL-6 (Figure 4G). Evidence again indicated that persistent activation of ROS, apoptosis and inflammatory responses might be the key pathophysiological basis for hyperglycemia-induced sustained myocardial injury and cardiac dysfunction. Importantly, KDM3A expression was significantly increased accompanied by an obvious decrease in H3K9me2 expression in DM rats (Figure 4H), however, no statistical difference was observed in KDM3A or H3K9me2 expression between the DM and DM + GC groups.

We have attempted to construct cardiomyocyte-specific KDM3A knock-out and over-expression rats, which unfortunately resulted in embryonic death. Only rats with the global KDM3A knock-out were successfully constructed and used in this study. According to the results of screening and sequencing identification, KDM3A global gene knock-out chimeric Founder rats were successfully constructed (Supplementary Figures 2B–E). The KDM3A-KO did not affect cardiac function in the basal state when compared with the WT group. However, after the DM model was established, rats in the DM–KO group exhibited higher LVEF and LVFS levels than that in the WT + DM group. Moreover, in the diabetic rats that received intensive GC therapy, the cardiac function of KDM3A-KO rats was also significantly improved compared to the WT rats (Figure 5A). Meanwhile, it was found from HE/WGA staining and Masson staining results that after DM and DM + GC treatment, the cross-sectional area and fibrosis degree of myocardial cells in the WT group and KO group were prominently increased compared with those in the sham group. Consistent with the echocardiography results, KDM3A-KO obviously relieved myocardial hypertrophy and fibrosis compared with the WT rats after DM and DM + GC treatments (Figure 5B).

Therefore, KDM3A knock-out alleviates long-term myocardial dysfunction and hypertrophy as well as mitigates fibrosis caused by hyperglycemia in the diabetic myocardium.

Consistent with the in vitro results, data of DHE staining demonstrated that after DM and DM + GC treatment, the accumulation of ROS was significantly increased in the WT and KDM3A-KO diabetic rats compared with that in the Sham group. However, ROS generation was significantly reduced in KDM3A-KO rats compared to WT rats in either the DM group or the DM-GC group (Figure 5C). In accordance with the ROS results, the TUNEL staining and IL-6/TNF-α results also showed that KDM3A knock-out evidently alleviated sustained apoptosis and inflammatory responses induced by hyperglycemia (Figures 5D,E).

The results of the western blot assay showed that the expression of P65 protein was significantly increased in both WT DM rats that didn’t receive intensive GC treatment and WT DM rats that received intensive GC treatment, accompanied by an obvious reduction in the H3K9me2 expression (Figure 6A). Moreover, the P65 protein level was significantly decreased and H3K9me2 protein expression was markedly increased in DM and DM + GC group after KDM3A knock-out (Figure 6A). Interestingly, there was no significant difference in P65 protein expression in both KDM3A-KO and WT rats between the DM and DM + GC groups. EMSA also demonstrated that the P65 DNA binding activity in the myocardium of DM rats continued to be increased even after being administrated with insulin, however, P65 DNA binding activity was remarkably declined after KDM3A knock-out in the DM and DM + GC groups (Figure 6B). The above in vivo results revealed the protective efficiency of KDM3A knock-out against hyperglycemia-induced persistent myocardial injury and cardiac dysfunction might be associated with the inhibition of the expression and transcriptional activity of NF-κB/P65.

To further investigate whether NF-κB/P65 is the specific target for KDM3A regulating HG-induced persistent cardiomyocytes injury. The rescue experiments were performed to investigate whether P65 inhibition could block the biological function of KDM3A. Primary cardiomyocytes were infected with AdKDM3A or/and transfected with a P65-siRNA plasmid (P65 expression was inhibited). As shown in Figures 7A,B, P65 knock-down evidently alleviated KDM3A overexpression-induced excessive ROS production and apoptosis both in HG and HG–NG groups. Besides, P65 knock-down significantly ameliorated KDM3A over-expression-induced inflammatory reaction both in HG and HG–NG groups as well (Figures 7C,D). Data above demonstrated that P65 inhibition could partially contract KDM3A over-expression-induced persistent cardiomyocytes injury in the condition of HG.