Animal model

Six- to eight-week-old male Wistar rats were used for this study (n = 8). All procedures on animals were performed according to the institutional animal ethical committee guidelines (IAEC) of NCCS. Streptozotocin (STZ, Sigma-Aldrich) dissolved in citrate buffer (pH 4.5) was given to the animals at a dose of 55 mg/kg body weight to induce diabetes. Control animals received an equivalent volume of citrate buffer. After confirmation of diabetes, control and diabetic animals were randomized into four groups with or without AKG treatment. The treatment groups received a daily oral dose of 250 mg/kg body weight AKG (Sigma-Aldrich) until the end of the experiment. The total duration of the experiment was three months.

Echocardiography and hemodynamic parameters

Rats were anesthetized using urethane (1 g/kg BW, i.p.) at the end of the experiment. ECG was recorded using ECG surface electrodes (AD Instruments, Colorado Springs, CO). A microtip pressure transducer catheter (SPR-320, Millar Instruments, Houston, TX, United States) was advanced into the left ventricles to record the hemodynamic parameters. The catheter was connected to an 8-channel Powerlab instrument via a bridge amplifier (AD Instrument). Data were recorded and analyzed by Lab Chart 7 software (AD Instrument). Left ventricular apical tissue was either snap frozen, embedded in OCT compound for IHC, or immersed in buffered formalin. Blood was collected from the abdominal aorta, and serum was isolated and stored at − 80 °C until further analysis.

Serum analysis

Insulin levels were measured by an ELISA kit (Mercodia AB, Sweden). Total cholesterol, triglycerides, and cardiac tissue damage markers creatine phosphokinase (CPK) and creatine kinase-myocardial band (CK-MB) were analyzed.

Histological staining

Formalin-fixed heart tissues were embedded in paraffin wax. Sections of 3–4 µm thickness were taken with a rotary microtome (Leica, Germany) and stained with hematoxylin & eosin (H & E) reagent. Prepared slides were examined under a microscope (Olympus, Japan), and images were taken at 40X magnification to note histopathological lesions. The severity of the lesions was determined by grading system; as 0 = within normal limits, 1 = Minimal, 2 = Mild/Slight, 3 = Moderate, 4 = Severe, and distribution of lymphocytic infiltration was recorded as focal, multifocal, and diffuse at ventricular myocytes as described previously [45].

Picrosirius red staining

Interstitial tissue collagen deposition and fibrosis were analyzed by picrosirius red (PSR) staining. Sections were deparaffinized and serially rehydrated and stained with 0.1% PSR stain. Then, sections were mounted with DPX mounting medium and observed under a light microscope (Nikon, Tokyo, Japan) and images were taken at 40 × magnification to note collagen deposition.

Cell culture

Embryonic cardiomyoblast H9c2 cells were maintained in Dulbecco’s Modified Eagle Medium (DMEM) supplemented with 10% serum. Upon reaching 70% confluency, cells were seeded into 100 mM dishes provided with DMEM containing 1% serum. These cells were treated with normal glucose (NG, 5.5 mM glucose) or high glucose (HG, 33 mM glucose) for 72 h with or without AKG (3.5 mM) treatment.

DNMT3B knockdown

A short hairpin RNA (shRNA) sequence targeting DNMT3B was cloned into the pLKO.1—TRC vector. Lentiviral particles were generated by transfecting HEK293FT cells with the DNMT3B shRNA construct along with packaging vectors (pMD2. G and psPAX2). A stable H9c2 cell line expressing shRNA was obtained by lentiviral infection followed by puromycin (1.5 µg) selection. The efficiency of DNMT3B knockdown was confirmed by western blot and qPCR analysis. Scrambled shRNA was used as a negative control. Oligonucleotide sequences of shRNA targeting DNMT3B are given in Additional file 1: Table S5.

Immunohistochemical (IHC) and immunofluorescence (IF) studies for 5mC

Frozen heart sections and H9c2 cells were stained with an anti-5mC antibody (28692S, Cell Signaling Technology, USA) according to the manufacturer’s protocol for IHC and IF. In brief, sections and cells were treated with 70% ethanol for 10 min followed by treatment with 2 N HCl for 1 h. After blocking nonspecific antibody binding using 3% bovine serum albumin, specimens were incubated overnight at 4 °C with primary antibody (1:60 dilution). Subsequently, specimens were stained with anti-rabbit Alexa Fluor® 488-conjugated secondary antibody (Invitrogen, USA). Nuclei were stained with 4′-6- diamidino-2-phenylindole (DAPI). Sections and cells were then mounted and imaged by a confocal laser scanning microscope (CFLSM) (Leica SP5II). Primary antibody was omitted for negative control experiments. For DNMT3B expression, the IF protocol was carried out as described previously [46]. Images were analyzed using ImageJ software (National Institutes of Health).

Dot Blot

Genomic DNA from left ventricular apical tissue and H9c2 cells was isolated using a DNeasyTissue kit (Qiagen). Dot blot analysis was performed to determine the levels of 5mC/5hmC in gDNA as described by Ito S., et al. with some modifications [47]. In brief, 1 µg gDNA was sonicated and subjected to alkali denaturation at 95 °C for 10 min. Denatured DNA was spotted manually onto the Hybond-N + nylon membrane (Amersham Biosciences). Then, DNA was crosslinked to the membrane by UV irradiation at 1200 mJ/cm2 dosage followed by blocking with 5% nonfat dry milk and probing with 5mC (rabbit)/5hmC (mouse) antibody (Cell Signaling Technology, USA) overnight at 4 °C, followed by respective HRP-conjugated secondary antibodies and detection with a chemiluminescence kit. To ensure equal spotting of total DNA on the membrane, the same blot was stained with 0.05% methylene blue (MB) in 0.3 M sodium acetate (pH 5.2). Densitometric quantification of the nucleic acid dot blot intensities was performed with ImageJ software (National Institutes of Health), and the values were normalized to methylene blue staining.

MeDIP/hMeDIP sequencing

Purified genomic DNA from tissue samples with two biological replicates from control and STZ groups was subjected to (h) MeDIP-seq library preparation. In brief, 5 μg of genomic DNA was fragmented by sonication to a mean size of approximately 200–400 bp, followed by end repair, A-base tailing, and adapter ligation. After adapter ligation, 10% of the DNA was secured, which served as input DNA. The remaining DNA was immunoprecipitated by using a MagMeDIP kit for 5mC and an hMEDIP kit for 5hmC (Diagenode). This captured DNA and input DNA were then purified with Zymo DNA Clean & Concentrator™-5 (Zymo Research) and subsequently amplified with 12 cycles using KAPA HiFi HotStart ReadyMix (Kapa Biosystems, Wilmington, MA, USA) and Illumina Multiplexing PCR Primers. An Agilent 2100 Bioanalyzer then analyzed the constructed libraries and finally sequenced on the Illumina platform (Illumina X Ten, paired-end, 150 bp).

Data processing

The quality of raw data from MEDIP, hMEDIP, and input samples was evaluated with FASTQC, the adapter sequence was trimmed with the help of Trimmomatic software (v0.36). Clean reads were mapped on rat genome Rn6 (Rnor_6.0) using bowtie2 (v2.4.4) followed by peak calling by MACS2 (v2.2.7.1) compared to input. Differentially methylated regions (DMRs) and differentially hydroxymethylated regions (DhMRs) between control and diabetic animals were detected using PePr (v1.1.24) software with default parameters. PePr uses a window-based approach and models read counts across intragroup and intergroup replicates with a negative binomial distribution [48]. Differential peaks were annotated using ChipSeeker (v1.8.6). The promoter region is identified as a − 1 kb region from the transcription start site (TSS). Kyoto Encyclopedia of Genes and Genomes (KEGG) and Gene Ontology (GO) enrichment analyses were performed for the DMRs/DhMRs using PathfindR (v1.6.2) and ShinyGO (v0.76.3) software packages, and peaks were visualized using Integrative Genomics Viewer (IGV) (v2.15.2) software.

MEDIP/hMEDIP-qPCR

Purified gDNA from ventricular tissue (1 µg for MEDIP and 2 µg for hMEDIP) was sonicated to obtain a fragment size of 300–500 bp by using Bioruptor pico (Diagenode). MEDIP was performed as described previously [49]. Methylated DNA was immunoprecipitated by using an anti-5mC antibody (286928, CST). Immunoprecipitation of 5hmC was performed with an hMEDIP kit (Diagenode) according to the manufacturer's protocol. After immunoenrichment, DNA was subjected to proteinase K digestion, resuspended in TE buffer, and used for qPCR analysis. The primers used for qPCR analysis are given in Additional file 1: Table S5. The following formula calculated percent enrichment relative to 10% input,

% 5mC/5hmC enrichment relative to input = 2[(Ct(10% input)−3.32)−Ct(IP)] × 100%

Where 3.32 is the compensatory factor for 10% input taken from the sample.

Real-time PCR

Ventricular tissue or H9c2 cells were used to analyze gene expression by real-time PCR (qPCR). Total RNA was isolated using TRIzol® reagent (Invitrogen, Carlsbad, CA, USA), and 1 µg of RNA was reverse transcribed using a Verso cDNA synthesis kit (Thermo Scientific, Waltham, MA, USA) according to the manufacturer’s protocol. qPCR was performed with the primers given in Additional file 1: Table S5 by using Power Up™ SYBR™ Green Master Mix (Applied Biosystems, Austin, TX, USA) on QuantStudio 3 (Applied Biosystems, Foster City, CA, USA). Gene expression was determined by normalization to β-actin by employing the ΔΔCt method.

Western blotting

Frozen tissue samples and H9c2 cells were lysed in RIPA buffer and processed as described previously [46] for western blot analysis using the following primary antibodies: DNMT1 (ab13537, Abcam), DNMT3A (3598, Cell Signaling Technologies), DNMT3B (ab2851, Abcam), MeCP2 (3456, CST), MBD2 (sc-514062, Santa Cruz Technologies), TET1 (ab157004, Abcam), CTGF (sc-14939, Santa Cruz Technologies), Troponin I (4002, Cell Signaling Technologies), and Vinculin (V9131, Sigma). Immune complexes were detected by SuperSignal West Pico Plus substrate (Thermo Fisher, Carlsbad, CA, USA) on a Chemidoc XRS + system (Bio-Rad). Densitometric quantification of the band intensities was performed with ImageJ software (National Institutes of Health), and the values were normalized to those of Vinculin.

Chromatin immunoprecipitation (ChIP)

To determine the binding activity of DNMT3B and TET1 to the (hydroxy)methylated regions, we performed ChIP-PCR as per the protocol described by Perna and Alberi [50] with some modifications. In brief, 35 mg of frozen tissue samples were minced and crosslinked with 1% formaldehyde solution for 15 min. The cross-linking reaction was stopped by adding 0.125 M glycine solution. Then, samples were ground in lysis buffer with protease inhibitor with the help of a Dounce homogenizer. Chromatin was sonicated to obtain a fragment size of 200–500 bp. One percent of the input DNA was stored at 4 °C. Protein concentration was determined by the Bradford method, and 30 µg of protein was used for immunoprecipitation for each sample. Immunoprecipitation of chromatin fragments was carried out using ChIP-grade antibodies against DNMT3B (ab2851, Abcam) or TET1 (ab157004, Abcam). Preimmune IgG was used as a negative control. Specific antibody–chromatin complexes were then precipitated by incubation with ChIP-grade protein-G-coated agarose beads (Invitrogen, USA) for 4 h at 4 °C. DNA samples were eluted, and supernatants were used for reverse cross-linking of the protein/DNA complexes to free DNA. The purified DNA was subjected to PCR amplification using TGFBR2 and TGFBR3 primer pairs (Additional file 1: Table S5) encompassing the sequence analysis. PCR products were visualized on 1.2% agarose gel.

Statistical analysis

Data are expressed as the mean ± standard deviation (SD). Unpaired t-tests were used for comparisons between two groups, and one-way ANOVA with Tukey’s multiple comparisons post-test was used for multiple groups. GraphPad Prism software (GraphPad, San Diego, CA, USA) was used to analyze statistical significance. A value of p < 0.05 was considered statistically significant.