Dapagliflozin attenuates diabetes-induced diastolic dysfunction and cardiac fibrosis by regulating SGK1 signaling

The study protocol was approved by the local Institutional Animal Care and Use Committee (IACUC) of Yonsei University Health System (YUHS-IACUC: 2016-0157) and complies with the ARRIVE reporting guidelines. Healthy New Zealand white rabbits were purchased from Dooyeol Biotech (Dooyeol Biotech, Seoul, Korea) and were maintained under the same standard laboratory conditions, housed at room temperature with a 12-h light cycle with free access to diet and water in each cage. All animals were submitted to daily health status monitoring including weight, food intake, and general activity. All protocols followed the guidelines for the care and use of laboratory animals (National Research Council, USA). The main outcome variable was diastolic dysfunction assessed by echocardiography. Based on a previous study [18], a case number estimation had yielded a group size of n = 10 (8 animals + 2 reserve animals). A total of 30 male rabbits (3.0–3.5 kg, 22–24 weeks) were randomly allocated to three groups: control (n = 10), diabetes (n = 10), and diabetes + dapa (dapagliflozin, 1 mg/kg/day/P.O. for 8 weeks) (n = 10). Diabetic condition was induced by intravenous injection of Alloxan monohydrate (ALX, Sigma-Aldrich, St. Louis, MO, USA) at a dose of 150 mg/kg. Rabbits exhibiting a fasting blood glucose level above 200 mg/dl were diagnosed as diabetic. All rabbits were fed for a 1% cholesterol diet (Dooyeol Biotech) for 6 weeks and a normal diet for 2 weeks thereafter. After follow-up, echocardiography was performed in the prone position after anesthesia with an intramuscular injection of an appropriate mixture of Zoletil and Rompun. After echocardiography, the animals were euthanized to minimize the discomfort experienced by animals. The experimental protocol is shown in Additional file 1: Fig. S3 in detail.

Glucose and cholesterol levels were measured in blood samples at baseline, diabetes modeling, and follow-up after 8 weeks using a Blood Glucose Monitoring System (Osang healthcare, Anyang, Korea) and DRI-CHEM 4000i (Fujifilm, Tokyo, Japan). Blood samples were drawn from the ear veins of the rabbits after they had been fasted overnight.

All images were obtained using a commercial ultrasound machine (Vivid 7 Dimension; GE Vingmed Ultrasound AS, Horten, Norway) with an S10 probe (2.5 megahertz). Images were acquired from apical three-chamber, four-chamber, and two-chamber views; and short-axis views of the mitral valves, papillary muscles, and apex [19].

The left atrial end-diastolic diameter (LVEDD), left atrial end-systolic diameter (LVESD), septal, LV posterior wall thicknesses, and left atrial diameter (LAD) were measured from standard planes. The LV ejection fraction (EF) was calculated using the Teicholz formula [19]. Pulsed Doppler echocardiography of the transmitral flow was performed. The sample volume was positioned at the level of the mitral tips in the apical four-chamber view. From the transmitral recording, the peak early (E) and late diastolic filling velocities were obtained. An apical four-chamber view was also used to obtain Doppler tissue imaging of the mitral annulus. Sample volumes were placed on the septal and lateral sides of the mitral annulus. Values for systolic (S′), early (e′), and late (a′) diastolic annular velocities were obtained. Echocardiography and analysis were performed in blind conditions.

The samples were cut into 1 mm squares and immediately placed in primary TEM fixation. After pretreatment, specimens were embedded with a Poly/Bed 812 kit (Polysciences, Warrington, USA) and then placed in resin and polymerized in an electron microscope oven (TD-700, DOSAKA, Kyoto, Japan) at 65 °C for 12 h. Ultrathin sections (80 nm) were placed on copper grids and double stained with 3% uranyl acetate and 3% lead citrate for 30 min and 7 min, respectively. The stained sections were then imaged using a transmission electron microscope (JEM-1011, JEOL, Tokyo, Japan) equipped with a Mega-View III CCD camera (Soft imaging system, Münster, Germany).

Heart tissue was fixed in 10% normal buffered formalin, embedded in paraffin, sectioned at 4μm thickness, cut on a microtome RM2235 (Leica, Wetzlar, Germany), then deparaffinized through the dewatering process. Masson’s Trichrome and Sirius Red were used to stain for collagens. Immunohistochemistry (IHC) and immunofluorescence (IF) were used to evaluate fibrosis, macrophage, or inflammation expression. Tissue sections were immunostained at 4°C overnight with antibody. IHC was used to detect α-SMA (Abcam, Cambridge, UK, ab-7817), Fibronectin (Abcam, ab-6328), TGF-β1 (Abbkine, Wuhan, China, ABP52598), 3-nitrotyrosine (Abcam, ab-61392), Receptor for advanced glycation end products (RAGE) (LifeSpan Biosciences, Seattle, USA, LS-C122375), RAM11 (DAKO, CA, USA, M0633), tumor necrosis factor-α (TNF-α) (Abcam, ab6671), NHE1 (Santa Cruz Biotechnologies, CA, USA, sc-136239), SGLT1 (Millipore, Overijse, Belgium, 07-1417), SGLT2 (Abcam, ab85626), Fis1 (Santa Cruz Biotechnologies, CA, USA, sc-376447), and Mfn1/Mitofusin1 (Santa Cruz Biotechnologies, CA, USA, sc-166644). The primary antibody was detected using a peroxidase-based kit (DAKO, Glostrup, Denmark) and visualized using DAB substrate with enhancer (DAKO). The sections were subsequently counterstained with hematoxylin (DAKO). The IHC staining was performed as previously described [20]. Digital images of the heart tissue were scanned using a SCN 400 scanner (Leica, Wetzlar, Germany), and histomorphometry was performed using LAS 4.2 software (Leica). Ten random images from 10 heart tissues per group were analyzed in a blinded procedure.

IF staining of the heart tissue to detect serum and glucocorticoid-regulated kinase 1 (SGK1) (ABCAM, ab43606) and epithelial sodium channel (ENaC) (Biorbyt, Cambridge, UK, orb100662) was performed following a published protocol [21]. The sections were washed for 10 min in 1% PBS and then incubated with FITC-conjugated secondary antibodies (Santa Cruz Biotechnologies) for 1 h in the dark at room temperature. The sections were washed in PBS for 10 min, mounted with Fluoroshield containing DAPI (ImmunoBioscience, Mukilteo, WA, USA), and stored in the dark at 4°C. Confocal microscopy was performed with an LSM 700 system (Carl Zeiss, Oberkochen, Germany).

Cells of the rat cardiomyoblast cell line H9C2 were cultured in DMEM containing 10% fetal bovine serum (both from Biowest, MO, USA) supplemented with 10% non-essential amino acids, 1% 2-mercaptoethanol, and 10% penicillin (all from Gibco, Carlsbad, CA, USA). Cells were maintained at 37°C in humidified air with 5% carbon dioxide. Before treatment, the cells were washed twice with pH 7.4 phosphate-buffered saline (PBS, Gibco). The cells were incubated in 500 μM palmitate (diluted in 5% bovine serum albumin [BSA]) with or without 35mM high glucose (HG) for 24 h and then treated with 0.4 μM dapagliflozin with 10 μg/ml lipopolysaccharide (LPS) for 24 h (all from Sigma-Aldrich).

siRNA targeting rat siSGK1 (5′- AGGAGAACAUCGAGCACAATT -3′) and siControl (5′-UUCUCCGAACGUGUCACGUTT-3′) were synthesized (Bioneer, Daejeon, Korea). H9C2 cells were then transfected with the siRNAs using Lipofectamine^TM RNAiMAX (Invitrogen, Carlsbad, CA, USA) according to a previously described method [22].

The heart tissues and H9C2 cells of total RNA were isolated using a published procedure [23]. cDNA was synthesized using Quantitect Reverse Transcription Kit (QIAGEN, Hilden, Germany), then the cDNA was amplified using AccuPower PCR Premix (Bioneer, Daejeon, Korea) and the SYBR Green kit of the 2X Fast Q-PCR Master Mix (SMOBIO, Hsinchu City, Taiwan). Relative mRNA levels were determined by comparison with GAPDH or β-actin. The rabbit and H9C2 primers used for the target genes are shown in Additional file 1: Table S1 and S2.

The heart tissues and H9C2 cells were lysed with RIPA buffer (Biosesang, Seongnam, Korea) containing Complete Mini and EDTA-free protease inhibitor cocktail (Roche, Basel, Switzerland). The protein samples were resolved by SDS-PAGE and then electrotransferred to an Immuno-Blot PVDF membrane (Bio-Rad, Hercules, CA, USA). Membranes were blocked with 5% skim milk (Noble Bio, Hwaseong, Korea) in 10% TBS-T for 1 h at room temperature. Membranes were incubated with primary antibodies against Fibronectin (Abcam, ab-6328), TGF-β1 (Abbkine, Wuhan, China, ABP52598), SGK1 (Abcam, ab43606), ENaC Gamma (Biorbyt, Cambridge, UK, orb100662), NHE1 (Santa Cruz Biotechnologies, CA, USA, sc-136239), TNF-α (Abcam, ab6671), IL-6 (Santa Cruz Biotechnologies), NF-kB (p65, Enzo life sciences, Farmingdale, NY, USA), and pNF-kB (p65, Santa Cruz Biotechnologies) at 4°C overnight and washed with TBS-T. They were incubated with horseradish peroxidase-conjugated secondary antibody for 1 h room temperature and then subjected to ECL (GE Healthcare, Chicago, USA) detection. GAPDH was detected on the same membrane to serve as a loading control. Densitometry analysis was performed using Image J software (National Institutes of Health, Bethesda, MD, USA).

Baseline characteristics were comparable among the three groups; however, the fasting blood glucose level was significantly higher in the diabetes groups compared with the control group (P < 0.001, Table 1). At 8 weeks follow-up, body weights did not differ among the three groups, but total cholesterol, triglyceride, low-density lipoprotein, and fasting blood glucose were significantly higher in the diabetes group compared to those in the control and diabetes+dapa groups (P < 0.001, P < 0.01, P < 0.05, Table 1).

Baseline echocardiographic parameters were similar among the three groups. Table 2 shows comparisons of echocardiographic parameters at 8 weeks follow-up among the three groups. In the diabetes+dapa group, septal e′ and lateral e′ velocities were significantly higher than that of the diabetes group (P < 0.05). The E/septal e′ ratio was significantly lower in the diabetes+dapa group compared with that in the diabetes groups (P < 0.05).

A qualitative assessment by Masson’s trichrome stain revealed prominent fibrosis in the myocardium of the diabetes group (Fig. 1A). Sirius red staining revealed that the Diabetes group exhibited significantly increased myocardial fibrosis compared with the control and diabetes+dapa groups (P < 0.001, Fig. 1A, B), reflecting increased collagen deposition. Extracellular remodeling was assessed by immunostaining for α-SMA, fibronectin, and TGF-β1 protein levels (Fig. 1C–E). The diabetes+dapa group showed significantly decreased fibronectin and TGF-β1 compared with the diabetes group (P < 0.01, P < 0.001). Dapagliflozin significantly reduced expression of fibronectin and TGF-b1 based on RT-PCR and Western blot findings (P < 0.001, P < 0.01, P < 0.05, Fig. 1I, J). This finding provides visual evidence that dapagliflozin can influence the extracellular remodeling of rabbit myocardium.

To further investigate the mechanism of dapagliflozin in the attenuation of cardiac fibrosis and extracellular remodeling, we compared the expression of myocardial SGK1 protein and its downstream proteins, ENaC and NHE1.

Myocardial SGK1 and ENaC proteins were significantly decreased in the diabetes+dapa group compared with the diabetes group, and comparable to the control group, when assessed by immunofluorescence techniques (P < 0.001, P < 0.01, Fig. 2A, B). Similarly, NHE1 was significantly decreased in the Diabetes+Dapa group compared with the Diabetes group (P < 0.01, Fig. 2C). The RT-PCR showed a significant reduction in SGK1, ENaC, and NHE1 levels in the diabetes + dapa group compared to that in the diabetes group (P < 0.01, P < 0.05, Fig. 2D) and similarly tend to decreased expression level in Western blot (Fig. 2E).

To confirm the effect of dapagliflozin at the cellular level, hyperglycemic fibrosis was induced in cardiomyoblast H9C2 cells using RT-PCR and Western blot (Fig. 3A, B). Fibronection and TGF-β1 mRNA and protein levels were significantly lower in the H9C2 cells in the high glucose group treated with dapagliflozin than in those in the high glucose group without dapagliflozin treatment (P < 0.001, P < 0.01). SGK1, ENaC, and NHE1 mRNA and protein levels were significantly lower in the dapagliflozin-treated group compared with those in the high glucose group (P < 0.001, P < 0.01, P < 0.05).

To test whether SGK1 was directly inhibited by dapagliflozin, we assessed SGK1 expression in dapagliflozin-treated H9C2 cells. SGK1 siRNA (20 nM) effectively suppressed the transcriptional and translational level of SGK1 (P < 0.001, P < 0.01, P < 0.05, Fig. 4A, B). Dapagliflozin suppressed the SGK1 level and subsequently suppressed ENaC and NHE1, similar to SGK1 downregulated cells (P < 0.001, P < 0.01, P < 0.05, Fig 4C, D).

To explore other possible mechanisms for the cardiovascular benefit of dapagliflozin, 3-nitrotyrosine (an oxidative stress marker) expression was compared between the three groups. 3-nitrotyrosine expression was significantly lower in the diabetes+dapa group than in the Diabetes group, and comparable to the control group (P < 0.05, Fig. 1F). In the expression of macrophage and inflammation protein markers, including RAM11, RAGE, and TNF-α, using immunohistochemistry analysis. RAM11, RAGE, and TNF-α expression were significantly lower in the Diabetes+Dapa group compared with the Diabetes group, and comparable to the Control group (Additional file 1: Fig. S1 A-C). Decreased expression of inflammatory markers, TNF-α, IL-6, and NF-kB, was observed in dapagliflozin-treated H9C2 cells compared with the high glucose group. (Additional file 1: Fig. S2 A-B).

To address possible mechanisms for decreased 3-nitrotyrosine expression in the diabetes+dapa group, mitochondria structures were analyzed. In the diabetes group, mitochondria showed a disrupted cristae structure with diminution of matrix electron density, loss and fusion of cristae, and mitochondrial fragmentation compared with the control group (Fig. 5A). In the diabetes+dapa group, the cristae structure in the mitochondria was preserved and a lesser degree of the diabetes-induced ultrastructural anomalies of the mitochondria was noted. Additionally, to further investigate the impact of dapagliflozin on mitochondrial dysfunction, we analyzed the expression levels of mitofission or mitofusion proteins, Fis1 and Mfn1. Levels of Fis-1 land Mfn-1 were significantly reduced in the Diabetes+Dapa group compared with the diabetes group (P < 0.05, Fig. 5B). RT-PCR and Western blot analysis also showed similar results (P < 0.05, Fig. 5C, D).

In the expression levels of SGLT1 and SGLT2 in heart tissue, both SGLT1 and SGLT2 were significantly higher expressions in the diabetes group when compared to those in the control group (P < 0.001, P < 0.01, P < 0.05, Fig. 1G–H and K).