Empagliflozin reduces the senescence of cardiac stromal cells and improves cardiac function in a murine model of diabetes

Abstract The sodium‐glucose cotransporter 2 (SGLT2) inhibitor empagliflozin reduces heart failure in diabetes, but underlying mechanisms remain elusive. We hypothesized that empagliflozin could counteract the senescence of cardiac stromal cells (CSC), the action of which limits cardiac damage and cardiac fibrosis in diabetic‐like conditions in vitro and in vivo. CSC were isolated from murine heart biopsies (n = 5) through cardiosphere (CSp) formation and incubated for 3 or 48 hours with 5.5 mmol/L normal glucose (NG), high glucose (12‐5 and 30.5 mmol/L, HG) or a hyperosmolar control of mannitol (HM) in the presence or absence of empagliflozin 100 nmol/L. The senescent CSC status was verified by β‐gal staining and expression of the pro‐survival marker Akt (pAkt) and the pro‐inflammatory marker p38 (p‐P38). The cardiac effects of empagliflozin were also studied in vivo by echocardiography and by histology in a murine model of streptozotocin (STZ)‐induced diabetes. Compared to NG, incubations with HG and HM significantly reduced the number of CSps, increased the β‐gal‐positive CSC and P‐p38, while decreasing pAkt, all reversed by empagliflozin (P < .01). Empagliflozin also reversed cardiac dysfunction, cardiac fibrosis and cell senescence in mice with (STZ)‐induced diabetes (P < .01). Empagliflozin counteracts the pro‐senescence effect of HG and of hyperosmolar stress on CSC, and improves cardiac function via decreasing cardiac fibrosis and senescence in diabetic mice, possibly through SGLT2 off‐target effects. These effects may explain empagliflozin unexpected benefits on cardiac function in diabetic patients.


| INTRODUC TI ON
Heart failure (HF), with or without ischaemic heart disease, remains the leading causes of death worldwide. 1 This can be attributed at least in part to the inability of the heart to regenerate damaged tissue. 2 As a result, the lost myocardium is replaced with fibrosis, initiating a series of subsequent events that lead to adverse ventricular remodelling and organ failure. 3 Cardiac stromal cells (CSC) are a heterogeneous population of non-cardiomyocyte CD45-, CD34-, CD31-and CD105+ cells including fibroblasts, resident in the heart with supportive and paracrine functions. The crosstalk between cardiomyocytes and CSC plays a fundamental role for the repair processes after cardiac damage, through the release of growth factors, pro-angiogenic factors and the regulation of cardiac metabolism. [4][5][6] Over and above normal cellular ageing, exposure to cardiovascular risk factors accelerates the senescence of cardiomyocytes and CSC, hampering their cardioprotective and repairing functions, and ultimately leading to HF. 7 One important condition that accelerates CSC senescence is diabetes. 7 Diabetes features a state of accelerated ageing, as several fundamental age-related changes occur at the cellular level in various tissues and organs (especially the blood vessels and the heart). 8,9 The crosstalk between CSC and cardiomyocytes can be hindered by the senescence of CSC due to metabolic and biophysical insults, such as in the presence of high glucose (HG) and related hyperosmolar stress, which can compromise cardiac function, promoting HF. 10 In contrast, new drugs, such as gliflozines, these ones so far tested in a diabetic setting, exert a positive impact on repair processes, limiting HF and its progression. 11,12 However, cellular and molecular mechanisms that confer protection from the development of HF in diabetes with gliflozines are largely unknown. In this context, we hypothesized that a specific gliflozine, empagliflozin, can specifically tackle the processes of CSC and cardiac senescence that ultimately favour HF. We then achieved this by comparing a condition of accelerated senescence-diabetes-vs 'normal' controls and investigated the mechanism of action of empagliflozin compared to untreated diabetic conditions in a murine model. We subsequently measured senescence markers in vitro and in vivo in the murine model of streptozotocin-induced diabetes and aimed at interfering with such pathways by empagliflozin.

| Materials
d-glucose and d-mannitol (this latter devoid of metabolic activities and used as a purely hyperosmolar controls) were purchased from Sigma. Empagliflozin was purchased from Boehringer Ingelheim.

| Isolation of cardiospheres and cardiac stromal cells
The murine tissue was derived from hearts (6 months-old adult mice) of male C57BL/6 mice (Charles River Italia). The isolated myocardial tissue was cut into 1 mm pieces, washed with Ca-Mg-free phosphate-buffered solution (PBS) (Invitrogen) and digested for 5 minutes at 37°C with 0.2% trypsin (Invitrogen). The obtained cells were discarded, and the remaining tissue fragments were washed with complete explant medium (Iscove's Modified Dulbecco's Medium [IMDM] supplemented with 10% foetal calf serum, 100 U/mL penicillin G, 100 μg/mL streptomycin, 2 mmol/L l-glutamine, and 0.1 mmol/L 2-mercaptoethanol) and cultured as explants in fibronectin-coated plates in complete explant medium at 37°C and 5% CO 2 . After 3 weeks, a layer of fibroblast-like cells was generated from adherent explants on which small, phase-bright cells migrated.

| Cardiospheres and cardiac stromal cell treatments
CSps at passage 1 and subconfluent CSC were synchronized by starvation in IMDM supplemented with 2% foetal calf serum, 100 U/ mL penicillin G, 100 μg/mL streptomycin, 2 mmol/L l-glutamine for 24 hours, then incubated for 3 or 48 hours with 5.5 mmol/L glucose

| Animal care and experimental procedures
Male C57BL/6 mice (body weight: 30 ± 4 g, 6 months old) were purchased from Charles River Italia. The mice were housed under a 12 hour light/dark cycle (7 am-7 pm) in temperature and humidity controlled rooms and were provided with ad libitum rodent chow (Teklad 7001, 4.4%; Harlan Teklad Global Diets) and water. Type 1 diabetes mellitus (T1DM) was induced by one single intraperitoneal injection of streptozotocin (STZ, Sigma-Aldrich) at 150 mg/kg dissolved in 0.01 mol/L citrate buffer (pH 4.5). 13 Control mice received citrate buffer injections (pH 4.5). Blood glucose was tested using a glucometer (Accuchek Nano, Roche) one week after STZ through tail vein blood collection. Diabetes was defined by blood glucose equal to 280.9 ± 35.8 mg/dL after a six-hour daytime fast. Mice with lower levels were excluded from the study. Empagliflozin was dissolved in water and administered to mice in the experimental group (30 mg/kg) by oral gavage for 1 month, whereas the vehicle group was given the same volume of water. For echocardiograms and minor procedures, the mice were anaesthetized by intraperitoneal injection of ketamine (100 mg/kg, Clorketam; Vétoquinol).
The animals were then killed under 2% isoflurane anaesthesia by cardiac puncture (Online Supplement, Online Figure 1

| Echocardiography
Using a portable ultrasound apparatus (Esaote) equipped with a 21-MHz linear probe, we performed trans-thoracic echocardiography

| Tissue preparation and histological evaluations
After induction of anaesthesia, the hearts of diabetic and age/sexmatched C57/BL6 control mice were removed and embedded in OCT without fixation or in paraffin after formalin fixation, respectively. The blocks were transversely cut to obtain 15 sections (each 5 μm thick): 5 from the base of the ventricles, 5 from the middle of the ventricles and 5 from the apex. All histological sections were analysed in a blinded fashion. Fibrotic areas were analysed on the paraffin-embedded sections by picro-sirius staining. Colour digital images were obtained, and the extent of LV fibrosis was evaluated with the ImageJ software (NIH, http://rsb.info.nih.gov/ij) in five randomly chosen cardiac areas per each heart section and expressed as % of the total LV area.
SA-b-gal activity was detected on frozen myocardial samples by the SA-b-gal Activity Senescence-Associated-β-Galactosidase Staining (Cell Biolabs, Inc), as previously described. 14 The frozen sections were post-fixed in 4% paraformaldehyde at room temperature for 15 minutes, rinsed with sterile PBS and incubated overnight with fresh senescence-associated β-galactosidase staining solution (1 mg/mL X-gal in 40 mmol/L citric acid/sodium phosphate, pH 6.0, 5 mmol/L potassium ferricyanide, 5 mmol/L potassium ferrocyanide, 150 mmol/L sodium chloride, 2 mmol/L magnesium chloride) at 37°C. Then, the staining solution was removed, the sections were immersed in 70% glycerol, and the development of the blue colour was assessed under standard light microscope. The extent of the blue-stained area was evaluated with the ImageJ software (NIH, http://rsb.info.nih.gov/ij) in five randomly chosen cardiac area per each heart section and expressed as % of total LV area.

| Immunoblotting
Total protein extracts of the whole left cardiac ventricle from the mice or CSC in each experimental group were isolated in an ice-cold RadioImmuno Precipitation Assay. The proteins were separated under reducing conditions and electroblotted onto polyvinylidene fluoride membrane (Immobilon-P, Millipore). After blocking, the membranes with proteins harvested from CSC were incubated overnight at 4°C with the following primary antibodies: (a) insulin receptor-α subunit (IRα, Santa Cruz Biotechnology), (b) insulin receptor substrate type 1 (IRS-1, Santa Cruz), (c) phosphorylated isoform of Akt (Cell Signaling), (d) phosphorylated isoform of p38 (Santa Cruz), as previously described. 15 The membranes with proteins harvested from the hearts were incubated overnight at 4°C with primary antibody against type III collagen (Sigma-Aldrich, 1:10 000 dilution).
Equal loading/equal protein transfer was verified by stripping and reprobing the blot with anti-beta actin (Sigma).  Figure 1C,D). These results indicate that HG interferes with CSps formation and exerts a pro-senescent effect on CSC in part through a hyperosmolar mechanism. These effects were significantly attenuated by 100 nmol/L empagliflozin targeting SGLT-2 ( Figure 1A-D). Under basal conditions in the absence of HG and HM, empagliflozin 100 nmol/L did not show any effect on the cardiosphere number ( Figure 1A,B). In contrast, 500 nmol/L empagliflozin exerted cytotoxic effects on CSps by decreasing their number at 3 and 48 hours after treatment (data not shown).

| Empagliflozin increases the expression of phosphorylated Akt, while decreases the expression of phosphorylated p38 in cardiac stromal cells under high glucose culture conditions
PI3K/Akt/eNOS is one of two parallel pathways activated by insulin after binding to its receptor. 16 In particular, Akt is an important signalling molecule involved in the regulation of cell survival. 17 We F I G U R E 3 Effects of streptozotocin and empagliflozin on cardiac function in mice. Panels A, Ejection Fraction (EF) and B, Fractional Shortening (FS) measured by echocardiography in the different treatment groups such as control (vehicle), streptozotocin-treated mice and streptozotocin + empagliflozintreated mice. Panels C, Representative M-mode and B-mode images recorded by echocardiography in the different treatment groups. Data are expressed as means ± standard deviations (n = 5 mice per treatment group). **P < .01 vs vehicle-treated mice; °°P < .01 vs without empagliflozin. E, empagliflozin; STZ, streptozotocin have recently shown that HG impairs the PI3K/Akt/eNOS pathway and shifts insulin signalling towards the activation of mitogenic and pro-inflammatory effectors, such as p38 and ERK1/2. 18 Here, we investigated the effects of HG and HM in phosphorylation of the pro-survival marker Akt and the pro-inflammatory marker p38 in CSC. In HG and HM, the levels of Ser473-phosphorylated Akt in CSC were reduced (Figure 2A). This was accompanied by an increase in the levels of the pro-inflammatory marker phosphorylated p38 (Figure 2B), indicating a selective down-regulation of the PI3/Akt pathway caused by HG through a hyperosmolar mechanism, which spares the mitogenic insulin signalling pathway. These effects were significantly attenuated by 100 nmol/L empagliflozin (Figure 2A,B). Under basal conditions in the absence of HG and HM, empagliflozin 100 nmol/ L did not show any effect on the expression of AKT nor P-p38 (Figure 2A,B). To examine whether the PI3K/Akt down-regulation upon incubation with HG and HM is due to insulin resistance through insulin receptor (IR)-α degradation, we semi-quantified the total amount of IR-α by immunoblotting. The expression of IR-α was reduced by HG but was not influenced by HM ( Figure 2C). 100 nmol/L empaglifozin was able to restore IR expression in CSC under HG culture conditions and induced it under in normoglycaemia, suggesting an insulin-sensitizing effect both in normoglycaemic and hyperglycaemic conditions ( Figure 2C). We also tested the possibility that incubation with HG and HM could affect the protein levels of phosphorylated (pIRS)-1 and total insulin receptor substrate (IRS)-1. Unlike IR-α, incubation with full range concentrations of HM and HG reduced pIRS-1 and IRS-1 ( Figure 2D), indicating that a reduction in insulin signalling by HGinduced hyperosmolar stress should depend by a disturbance in the signalling pathway downstream of the IR. 100 nmol/L empaglifozin was able to restore the expression of IRS-1 in the CSC under HG and HM culture conditions, suggesting an insulin-sensitizing effect ( Figure 2D).

| Empagliflozin reduced high glucose-induced cardiac dysfunction in streptozotocin-treated mice
To study the impact of empagliflozin on systolic function of the left ventricle, trans-thoracic echocardiography was performed in mice treated with STZ, or STZ + emapgliflozin or vehicle (saline). Figure 3 shows the representative images of the echocardiographic analyses. Empagliflozin significantly attenuated these changes (Figure 3).

| Empagliflozin reduced high glucose-induced cardiac fibrosis and senescence
To study the effect of empagliflozin on HG-induced cardiac fibrosis, we performed histopathological staining and western blotting analysis.
The Sirius red staining showed that the fibrotic area was increased in STZ-treated mice compared to the vehicle ( Figure 4A,B). The collagen area was smaller in the STZ + empagliflozin treatment group than in the STZ treatment group ( Figure 4C). Western blotting showed that STZ treatment increased the expression of the fibrosis marker collagen III ( Figure 4D). Notably, the expression of collagen type III was significantly reduced upon empagliflozin treatment ( Figure 4E,F). STZ induced a significantly higher percentage of SA-β-gal-positive cardiac area, which was reduced by empagliflozin treatment (Figure 5).

| D ISCUSS I ON
In the present study, we demonstrated that HG, mimicking the in vivo conditions of type 1 and type 2 diabetes, induces cellular senescence in CSC. This effect is accompanied by the attenuation of the anti-inflammatory and pro-survival insulin signalling in CSC through a down-regulation of the PI3K/Akt pathway. The effects of HG on the PI3K/Akt pathway depend in part on the hyperosmolar component of HG, since they are mimicked by CSC exposure to the metabolically inactive monosaccharide mannitol.
Since these effects are observed only in the PI3K/Akt pathway, while in the same conditions CSC do not lose the ability to respond to insulin with the activation of P38, these data support the hypothesis that HG promotes an imbalance in CSC insulin signalling between pro-survival and pro-inflammatory arms, with insulin preserving its signalling capacity through the mitogenic and pro-inflammatory arm. Attenuated insulin signalling can also result from changes in insulin resistance and/or post-receptor signalling. 18 We show here that exposure to HG affects the total amount of IR-α and IRS-1 levels, with the hyperosmolar component starting to act on the signal further downstream of the IRS-1. Overall, this indicates that the altered CSC responsiveness to insulin is due to a disturbance in the signalling pathway starting from IR-α ( Figure 6).
Hyperglycaemia has been reported to contribute to tissue injury, particularly in heart disease. 19,20 Several signalling pathways have been analysed and targeted in cardiac fibrosis, senescence and ischaemia, on a dysmetabolic or cardiotoxic basis, including activation of the Stat3-mediated pathway. [21][22][23] In non-diabetic mice F I G U R E 5 Effects of streptozotocin and empagliflozin on cardiac senescence-associated β-galactosidase activity in mice. Panels A-C, Senescence β-galactosidase (SA β-Gal) staining (magnification 5× and 40×) and quantitative evaluations (Panel D) in cardiac tissue. Data are expressed as means ± standard deviations (n = 5 mice per treatment group). **P < .01 vs vehicle-treated mice; °°P < .01 vs without empagliflozin. E, empagliflozin; STZ, streptozotocin undergoing ischaemia-reperfusion injury, empagliflozin reduced myocardial infart size through STAT-3. 24 Guo et al 20  reduces myocardial interstitial fibrosis and remodelling in hypertensive rats. [33][34][35] Although the SGLT1 isoform is the only form expressed in the capillaries and the myocardium of human and rodent hearts, 36 and we did not find any consistent data supporting the expression of SGLT2 receptor in murine CSC and in the heart, it is possible that empagliflozin has off-target effects outside of SGLT2mediated pathways in the heart, regulating CSC activity. CSC are a mixed population of non-cardiomyocyte cells including fibroblasts with paracrine activity, and can therefore represent the cellular target of empagliflozin for its antifibrotic and anti-senescence effects.
Indeed, one of the relevant component of CSC, the cardiac fibroblasts, is known to express sodium-hydrogen exchanger 1 (NHE-1), 37,38 which is an established molecular target of SGLT2 inhibitors, including empagliflozin. 39 We have previously shown that NHE-1 is one of the molecular targets of hyperosmolar stress, capable of mediating the down-regulation of the PI3/Akt/eNOS signalling pathway induced by hyperosmolar stress in endothelial cells exposed to HG and HM, and that this effect is reverted by NHE-1 blockade through cariporide. [40][41][42] In the present work, we have demonstrated a similar down-regulation of the PI3/Akt signalling pathway exerted by hyperosmolar stress in CSCs exposed to HG and HM, and that this effect is reversed by empagliflozin. Inhibition of NHE-1 attenuated cardiac fibrosis and HF in rat models of high blood pressure, 43,44 and cariporide treatment caused a significant increase in the lifespan of both hypertensive and normotensive rats, attenuating cardiac fibrosis typically associated with ageing. 45 We recognize limitations of this study. A first one is that we here used murine cardiac samples to explain effects of empagliflozin in humans. Indeed, there are intrinsic constraints when using murine specimens, including diverse pharmacodynamic and pharmacokinetic profiles of the drug. Another limitation is that the different mechanisms by which Akt can exert a pro-survival action have not been studied in our cell system of CSC exposed to HG, and this needs to be studied in the future. Finally, this study does not provide direct data to support the mechanistic hypothesis of the antifibrotic and anti-senescence effects of empagliflozin on the heart.
In conclusion, our study shows that HG-induced senescence and cardiac fibrosis are reduced by empagliflozin possibly through SGLT2 off-target effects. These effects may help explain unexpected benefits of empagliflozin on cardiac function in patients with diabetes.

CO N FLI C T O F I NTE R E S T
The authors declare no conflict of interest. F I G U R E 6 A scheme of the putative signalling pathway through which empagliflozin target cardiac stromal cell senescence induced by hyperglycaemia and hyperosmotic, based on findings from the in vitro part of the present study. In the plasma membrane, hyperglycaemia induces hyperosmotic stress, and together, these metabolic and biophysical stressors impair the ability of cardiac stromal cells to respond to insulin. This causes the perturbation of the pro-survival PI3K/Akt and pro-inflammatory p38 pathways with insulin preserving its signalling capacity through the mitogenic and pro-inflammatory arm, leading to the development and progression of a pro-senescent phenotype. Empagliflozin is able to revert CSC senescence by targeting the insulin signalling pathway

DATA AVA I L A B I L I T Y S TAT E M E N T
Data available on request from the authors.