Male Sprague-Dawley rats (Koatech, Pyeongtaek, Korea), 6 weeks of age and weighing 280 to 300 g, were used in this study. A combination of a high-fat diet (HFD, 60% fat content, D12492; Research Diets, New Brunswick, NJ, USA) and a low dose of streptozotocin (STZ; Sigma, St. Louis, MO, USA) was used to induce a rat model of diabetic cardiomyopathy [2021]. The rats were kept in a specific pathogen-free facility at Hanyang University Medical School Animal Experiment Center under controlled temperature (23℃±2℃) and humidity (55%±5%) with a 12-hour artificial light and dark cycle. This research protocol was approved by the Hanyang University Institutional Animal Care and Use Committee (HY-IACUC-19-0016), and the experiments were performed in compliance with the Animal Research: Reporting of In Vivo Experiments (ARRIVE) guidelines on animal research [22].

Seven-week-old rats were separated into diabetic rats (n=15) and normal rats (n=8). The normal rats were fed a standard diet and the diabetic rats were fed an HFD for 7 weeks. After 6 weeks (at 13 weeks of age), the rats in the diabetic rat model group were intraperitoneally injected with 30 mg/kg STZ dissolved in 100 mM citrate buffer pH 4.5. We excluded diabetic rats with a fasting blood glucose (FBG) level below 200 mg/dL. The normal rats were intraperitoneally injected with only citrate buffer. At 1 week after STZ injection (at 14 weeks of age), the diabetic rats were randomly divided into two groups of treatment with saline (n=7) or G-CSF (n=8). The rats were injected intraperitoneally with either saline or 200 µg/kg/day recombinant human G-CSF (Leucostim; Dong-A Pharmacological, Seoul, Korea) for 5 days. At the time of initiation of G-CSF and saline administration, the diabetic rats were removed from HFD and fed with a standard diet (Fig. 1). All rats were euthanized at 18 weeks of age.

The H9c2 rat cardiomyocyte cell line was purchased from American Type Culture Collection (ATCC, Manassas, VA, USA). H9c2 cells were cultured in Dulbecco's modified Eagle's medium (DMEM; Gibco, Grand Island, NY, USA) supplemented with 5.5 mM glucose and 10% fetal bovine serum (FBS; Gibco), 100 U/mL penicillin and 100 mg/mL streptomycin (Gibco). The cells were maintained in a humidified 37℃ incubator with 5% CO₂. The medium was replaced every 2 to 3 days, and cells were subcultured once they reached 80% to 90% confluence. When the cell populations reached 60% to 70% confluence, the cultures were exposed to D-glucose (Sigma) at a final concentration of 33 mM high glucose (HG) treatment according to previous publications [23] or to 5.5 mM D-glucose as normal glucose (NG). The cells were pretreated with HG (33 mM) for 48 hours with or without G-CSF (500 ng/mL). Cells were plated at an appropriate density according to experimental design.

Echocardiography was performed using an ultrasound machine VIVID E9 (GE Healthcare, Milwaukee, WI, USA). The rats were anesthetized by intramuscular injection of a mixture of zoletil 50 (30 mg/kg; Virbac SA, Carros, France) and rompun (10 mg/kg; Bayer Korea, Seoul, Korea). All rats were placed in the left lateral decubitus position for serial echocardiographic examinations. The measurements included left ventricular ejection fraction (LVEF), peak velocity of the early diastolic filling wave (E), and early mitral annulus velocity (E'), as performed previously [1819]. All measurements represent the mean of five consecutive cardiac cycles and heart rate was calculated based on the strip chart of Doppler echocardiography.

Blood samples were collected from tail veins after 12 hours of fasting and plasma was separated by centrifugation at 3,500×g for 10 minutes. Body weight (BW), FBG, total cholesterol (TC), triglyceride (TG), and fasting insulin levels were measured. Serum glucose, TC, and TG were measured using an Olympus AU400 auto analyzer (Olympus GmbH, Hamburg, Germany). Fasting insulin was measured using an EZRMI 13K kit (Linco, St. Charles, MO, USA) according to the manufacturer's instructions. Insulin resistance was estimated by the homeostasis model assessment of insulin resistance (HOMA-IR) using the following formula: HOMA-IR=fasting insulin (µU/mL)×fasting plasma glucose (mmol/L)/22.5 [24].

Fluorescein isothiocyanate (FITC)-conjugated annexin V and propidium iodide (PI) were used to identify apoptotic cells using a FITC-annexin V apoptosis detection kit (BD Biosciences Pharmingen, San Diego, CA, USA) according to the manufacturer's instructions [25]. The H9c2 cells were harvested after the experimental procedures and washed twice with PBS. The cells were resuspended in binding buffer and FITC annexin V and PI were added. The mixture was incubated for 15 minutes in the dark at room temperature. The resulting fluorescence was measured by flow cytometry using a fluorescence-activated cell sorting flow cytometer (BD Biosciences).

The rats were anesthetized by intramuscular injection of a mixture of zoletil 50 (30 mg/kg) and rompun (10 mg/kg). Rats were weighed and their hearts were removed and divided into two halves along the anterior longitudinal middle line. One half of each heart was fixed in formalin, embedded in paraffin, and cut into 4 µm thick sections. The other half was frozen in liquid nitrogen and stored at −80℃ for real-time polymerase chain reaction (PCR) and Western blot analyses. The extent of myocardial fibrosis was determined by visualizing fibrotic tissue using Masson's trichrome (MT) staining. Apoptotic cardiomyocytes were evaluated using the terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) assay in paraffin sections with an In Situ Cell Death Detection kit (Roche, Mannheim, Germany). The stained sections were photographed using a light microscope (Leica DM 4000B; Leica, Wetzlar, Germany). Five regions from each digitized images were selected at random from the individual sections and quantified using the Leica image analysis system (Leica DM 4000B). All data were evaluated by an independent blinded investigator.

Total RNA was harvested from heart tissue and H9c2 cells using Qiazol reagent (Qiagen, Valencia, CA, USA), according to the manufacturer's instructions [26]. The concentration of each sample was measured using a Nanodrop ND-2000 spectrophotometer (Thermo Fisher Scientific Inc., Wilmington, DE, USA). For real-time PCR analysis, total RNA was reverse transcribed with stem-loop primers and the TaqMan MicroRNA Reverse Transcription kit (Applied BioSystems, Foster City, CA, USA), according to the manufacturer's instructions [24]. Real-time PCR was performed in duplicate using the TaqMan MicroRNA assay kit and TaqMan Universal PCR MasterMix (Applied Biosystems) for miR-34a, miR-92a, miR-21, miR-320, miR-23a, and miR-15b, according to the manufacturer's instructions. Real-time PCR was performed using the LightCycler480 program (Roche) for 40 cycles, (10 seconds each, at 95℃, 60℃, 72℃). Relative miRNA expression levels were normalized using the RNU6B (U6) small non-coding RNA as an endogenous control.

Transfection was carried out using Lipofectamine 2000 (Invitrogen, Carlsbad, CA, USA). For RNA interference, H9c2 cells were transfected with a miR-34a mimic (50 nM), miR-34a inhibitor (50 nM), or matched negative control (NC, 50 nM; GenePharma, Shanghai, China). All transfections were incubated for 24 hours, and then switched to NG (5.5 mM) media or HG (33 mM) media conditions. To examine whether miR-34a regulates the expression of B-cell lymphoma 2 (Bcl-2), a predicted target of miR-34a, H9c2 cells were co-transfected with luciferase vector (100 ng) containing the Bcl-2 3'UTR and miR-34a mimic or inhibitor using Lipofectamine 2000 (Invitrogen). Co-transfection with non-targeting NC RNA was performed as a control. The cells were harvested 24 hours after transfection, and luciferase activity was measured with a dual luciferase reporter assay kit (Promega, Madison, WI, USA) on a luminometer (Molecular Devices, Sunnyvale, CA, USA) following the manufactures instructions. Firefly luciferase activity was normalized to Renilla luciferase activity. All experiments were performed in triplicate.

The excised heart tissues were homogenized and total proteins were extracted using protein lysis buffer (Pro-preb; iNtRON, Seongnam, Korea). H9c2 cells were harvested and total proteins were extracted using RIPA cell lysis buffer (Genedepot, Hanam, Korea). Samples containing 60 µg of protein were transferred to sample buffer, separated by 10% sodium dodecyl sulfate polyacrylamide gel electrophoresis and transferred to an immobilon-P transfer membrane (PVDF, 0.45 µm pore size; Millipore, Billerica, MA, USA). After blocking in 5% skim milk solution for 60 minutes, the membranes were incubated with primary antibody for Bcl-2 (1:250; Santa Cruz Biotechnology, Santa Cruz, CA, USA), caspase-9 (1:250; Santa Cruz Biotechnology), or glyceraldehyde 3 phosphate dehydrogenase (GAPDH; 1:3,000; Cell Signaling Technology, Boston, MA, USA) overnight at 4℃. Membranes were then incubated with horseradish peroxidase (HRP)-conjugated anti-mouse antibody (1:1,000; Jackson Immunoresearch, West Grove, PA, USA) or anti-rabbit antibody (1:3,000; Jackson Immunoresearch) for 1 hour 30 minutes at room temperature. GAPDH was used as a protein loading control. Positive protein bands were visualized using an ECL kit (GenDEPOT, Barker, TX, USA), and the results were quantified with an image analyzer (Image lab 3.0; Bio-Rad, Hercules, CA, USA).

SPSS version 21.0 software (IBM Co., Armonk, NY, USA) was used for statistical analyses. All data are expressed as mean±standard deviation, except for histological data, which are expressed as mean±standard error. Histological data were analyzed using Kruskal-Wallis nonparametric analysis of variance (ANOVA, for multiple comparisons). Echocardiographic values between the groups were analyzed using two-way repeated measures ANOVA with Bonferroni's post hoc test for multiple comparisons. The remaining data were analyzed using oneway ANOVA (for multiple comparisons) with Tukey's post hoc test. Values of P<0.05 were considered statistically significant.

Supplementary Table 1 shows BW and results of biochemical analysis before and after treatment with G-CSF or saline. There were no differences in BWs between saline- and G-CSF-treated diabetic rats. FBG, TC, TG, insulin, and HOMA-IR were significantly increased in diabetic rats compared to the normal rats. G-CSF treatment significantly decreased TC and TG compared to saline-treated rats. However, G-CSF treatment did not significantly affect insulin or HOMA-IR in diabetic rats.

To investigate the effect of G-CSF treatment on heart function of diabetic rats, echocardiography was performed to measure cardiac parameters. Before G-CSF treatment (pre-treatment), E' velocity was significantly decreased and E/E' ratio was significantly increased in diabetic rats compared to normal rats. This result suggests that the diabetic rats had developed diastolic dysfunction. After treatment with G-CSF (post-treatment), LVEF was significantly decreased in diabetic rats compared to normal rats. E' velocity (3.8%±0.05% vs. 3.0%±0.12%, P<0.05) was significantly increased and E/E' ratio was significantly decreased (17.3%±0.64% vs. 22.7%±1.15%, P<0.05) in G-CSF-treated rats compared to saline-treated rats (Fig. 2). These results suggest that G-CSF improved diastolic dysfunction.

Cardiac fibrosis resulting from collagen deposition was investigated by Masson's trichrome staining. Interstitial collagen deposition increased significantly in saline-treated rats compared with G-CSF-treated rats (7.3%±3.14% vs. 2.3%±1.45%, P<0.05) (Fig. 3A and C). G-CSF treatment significantly decreased collagen deposition on cardiomyocytes.

H9c2 cells were transfected with miR-34a mimic or inhibitor to up-regulate or down-regulate the expression of miR-34a, respectively. Transfection with NC did not significantly change miR-34a level. Transfection with miR-34a mimic significantly increased miR-34a level and transfection with miR-34a inhibitor significantly decreased miR-34a level compared to transfection with NC (Supplementary Fig. 1).

miR-34a expression in the diabetic myocardium was significantly decreased in G-CSF-treated rats compared to saline-treated rats (Fig. 4A). In addition, miR-34a expression was significantly increased in H9c2 cells under HG condition compared to NG condition. Also, miR-34a expression were significantly decreased in G-CSF treated H9c2 cells under HG condition (Fig. 4B). However, miR-92a, miR-21, miR-320, miR-23a, and miR-15b were not significantly different in diabetic rats treated with G-CSF compared to diabetic rats treated with saline (Fig. 4C).

The TUNEL assay was performed to investigate the effects of G-CSF on cardiomyocyte apoptosis in the diabetic myocardium. G-CSF treatment significantly decreased the amount of TUNEL-positive cardiomyocytes compared to saline-treated rats (25.7%±3.22% vs. 40.9%±7.23%, P<0.05) (Fig. 3B and D). Based on annexin V staining, the number of apoptotic cells was significantly increased in H9c2 cells under the HG condition compared to the NG condition (37.5%±11.7% vs. 23.3%±4.3%, P<0.05). In addition, the number of apoptotic cells was significantly decreased in G-CSF-treated H9c2 cells under HG condition compared to control H9c2 cells under HG condition (20.8%±3.0% vs. 37.5%±11.7%, P<0.05) (Fig. 5A and B).

The apoptotic effect of miR-34a on H9c2 cells under HG condition was analyzed by annexin V staining. Transfection with miR-34a mimic significantly induced cell apoptosis compared to transfection with NC in H9c2 cells under HG condition (48.55%±3.62% vs. 42.82%±2.82%, P<0.05). Transfection with miR-34a inhibitor significantly reduced cell apoptosis compared to transfection with NC in H9c2 cells under HG condition (35.20%±1.34% vs. 42.82%±2.82%, P<0.05) (Fig. 5C). G-CSF treatment did not reduce apoptosis in H9c2 cells transfected with the miR-34a mimic under HG condition. In contrast, G-CSF treatment reduced apoptosis in H9c2 cells transfected with the miR-34a inhibitor under HG condition (Fig. 5D). These data indicate that miR-34a plays an important role in induction of cardiomyocyte apoptosis.

The bioinformatic target prediction algorithm revealed that Bcl-2 was a predicted target gene of miR-34a. Bcl-2 protein was measured by Western blotting (Fig. 6A and B). Bcl-2 expression in the diabetic myocardium was significantly decreased in saline-treated rats compared to G-CSF-treated (12.82%±7.4% vs. 37.62%±16.7%, P<0.05). In addition, Bcl-2 expression was significantly decreased in control H9c2 cells under HG condition compared to G-CSF-treated H9c2 cells under HG condition. (42.39%±6.77% vs. 67.37%±10.06%, P<0.05). Bcl-2 expression was also significantly decreased after transfection with miR-34 mimic compared to transfection with NC in H9c2 cells under HG condition (102.21%±6.2% vs. 137.16%±18.32%, P<0.05) (Fig. 6C).

The promoter activity of Bcl-2 was confirmed by a dual-luciferase assay (Fig. 6D). The luciferase activity of the Bcl-2 construct was significantly decreased in H9c2 cells transfected with miR-34a mimic compared to transfection with NC or miR-34a inhibitor (0.81%±0.11% vs. 1.00%±0.08% and 1.08%±0.1%, P<0.05). These results implied that miR-34a was able to bind to Bcl-2 directly and inhibit its expression.

Caspase-9 protein was measured by Western blotting (Supplementary Fig. 2). Caspase-9 expression was significantly decreased in diabetic rats treated with G-CSF compared with diabetic rats treated with saline (91.40%±21.91% vs. 139.23%±17.66%).