Angiotensin II‐induced redox‐sensitive SGLT1 and 2 expression promotes high glucose‐induced endothelial cell senescence

Abstract High glucose (HG)‐induced endothelial senescence and dysfunction contribute to the increased cardiovascular risk in diabetes. Empagliflozin, a selective sodium glucose co‐transporter2 (SGLT2) inhibitor, reduced the risk of cardiovascular mortality in type 2 diabetic patients but the protective mechanism remains unclear. This study examines the role of SGLT2 in HG‐induced endothelial senescence and dysfunction. Porcine coronary artery cultured endothelial cells (ECs) or segments were exposed to HG (25 mmol/L) before determination of senescence‐associated beta‐galactosidase activity, protein level by Western blot and immunofluorescence staining, mRNA by RT‐PCR, nitric oxide (NO) by electron paramagnetic resonance, oxidative stress using dihydroethidium and glucose uptake using 2‐NBD‐glucose. HG increased ECs senescence markers and oxidative stress, down‐regulated eNOS expression and NO formation, and induced the expression of VCAM‐1, tissue factor, and the local angiotensin system, all these effects were prevented by empagliflozin. Empagliflozin and LX‐4211 (dual SGLT1/2 inhibitor) reduced glucose uptake stimulated by HG and H2O2 in ECs. HG increased SGLT1 and 2 protein levels in cultured ECs and native endothelium. Inhibition of the angiotensin system prevented HG‐induced ECs senescence and SGLT1 and 2 expression. Thus, HG‐induced ECs ageing is driven by the local angiotensin system via the redox‐sensitive up‐regulation of SGLT1 and 2, and, in turn, enhanced glucotoxicity.

premature death from cardiovascular disease, the leading cause of death, in diabetes mellitus. 2 Endothelial dysfunction is an early key event contributing to the initiation and development of diabetic vascular complications. 3,4 Indeed, reduced endothelium-dependent nitric oxide (NO)-mediated vasodilatation is observed in patients with type 1 and 2 diabetes, [5][6][7] and blunted NO-mediated vasorelaxation in experimental models of diabetes. [8][9][10] Since exposure of isolated arteries and cultured endothelial cells (ECs) to high glucose (HG) reduced the formation of NO, [11][12][13] ECs appear to be particularly sensitive to glucotoxicity. The mechanisms underlying the hyperglycemia-induced endothelial dysfunction have been shown to involve NADPH oxidase-and mitochondria-derived oxidative stress, 10,14 uncoupled endothelial NO synthase, 10 advanced glycation end products (AGEs) and receptors for AGEs signalling, 15 and increased levels of asymmetric dimethylarginine. 16 More recently, endothelial senescence has been identified as a potential contributor to the premature alteration of the endothelial function in diabetes. Indeed, senescent ECs were observed in the aorta of diabetic rats, [17][18][19] and the exposure of ECs to a high concentration of glucose (HG) increased the number of senescence associated-beta-galactosidase (SA-β-gal) positive cells. 18,19 Endothelial senescence, characterized by an irreversible form of growth arrest, is associated with functional and structural changes and altered gene expression including increased oxidative stress and the down-regulation of the eNOS-derived formation of NO, and, as a consequence, an up-regulation of pro-atherothrombotic responses. 18,19 Gliflozins are a new class of anti-diabetic drugs targeting the sodium-glucose co-transporter 2 (SGLT2), located in the S1-S2 segments of the proximal tubule in the kidney and responsible for nearly 90% of renal glucose reabsorption. In addition to SGLT2, SGLT1 in the distal segment S3 accounts for the reabsorption of the remaining 10%. 20 Thus, inhibition of SGLT2 leads to a significant reduction in the blood glucose level by preventing glucose reabsorption by the kidney. Empagliflozin, a selective SGLT2 inhibitor, reduced cardiovascular mortality in type 2 diabetic patients by 38%, in parallel to a moderate reduction in HbA1c, 21 and decreased the progression of kidney disease and lowered the rate of clinically relevant renal events. 22 The role of SGLT2 in diabetesrelated endothelial dysfunction and vascular complications has been poorly studied. Recently, SGLT inhibitors have been shown to attenuate the endothelial dysfunction in streptozotocin-treated mice and rats 23,24 and in Zucker diabetic fatty rats. 20 However, it is unclear whether the vasoprotective effect of SGLT inhibitors is related to the decreased renal sodium and glucose reabsorption and/or to a direct effect at the arterial wall. Indeed, SGLT1 expression has been observed in cultured ECs and in the endothelium of the mouse thoracic aorta, and its expression is increased in capillaries and small vessels after brain ischemia and reperfusion. 25,26 In contrast to SGLT1, studies regarding the expression of SGLT2 in ECs have indicated that SGLT2 mRNA levels were undetectable in control cultured human pulmonary and coronary artery EC lines. 27 Since cultured ECs often undergo pronounced phenotypic changes following numerous passages, SGLT2 expression remains to be determined in primary and low passages of ECs under both physiological and pathological conditions. Therefore, the aim of this study was to investigate the possibility that SGLT2 may play a role in endothelial senescence and dysfunction following chronic exposure to HG raising the interest and the understanding of the use of SGLT2 inhibitors in type 2 diabetes.

| Determination of SA-β-gal activity by flow cytometry
SA-β-galactosidase activity was determined by flow cytometry using the fluorogenic substrate C 12 FDG (5-dodecanoylaminofluorescein Di-β-D-galactopyranoside, Invitrogen) as described previously. 29 ECs were pretreated with 300 μmol/L chloroquine for 1 hour to induce lysosomal alkalinization. C 12 FDG (33 μmol/L) was then added to the incubation medium for 1 hour. At the end of the incubation period, ECs were washed with ice-cold PBS, resuspended following trypsinization and analysed using a flow cytometer (FACScan, BD Bioscience, Le Pont de Claix, France). Data were acquired and analysed using the Cellquest software (BD Bioscience). Light scatter parameters were used to eliminate dead cells and subcellular debris.
The C 12 -fluorescein signal was measured on the FL1 detector, and the proportion of ECs with SA-β-gal activity was estimated using the median fluorescence intensity of the population. Autofluorescence assessed in parallel in ECs not exposed to C 12 FDG was negligible.

| Analysis of mRNA expression by RT-PCR
Total RNAs were isolated from ECs or tissues using the mirVANA ® Isolation kit (Invitrogen). In some experiments, mRNAs were purified from the total RNAs using a poly (dt) primer, in accordance with the manufacturer's instructions (Macherey-Nagel, Hoerdt, France). Results were normalized with GAPDH and expressed as fold change over control.

| Determination of NO formation by electron paramagnetic resonance
Nitric oxide formation was assessed in ECs cultured on Cytodex-3 beads by electron paramagnetic resonance after formation of Fe(II) NO(DETC) 2 , a paramagnetic diethyldithiocarbamate iron complex with NO, at 77 K in a Dewar flask using a MS100 spectrometer (Magnettech Ltd., Berlin, Germany). The ESR methodology was used as reported previously. 30 The values are expressed in signal amplitude (arbitrary units).

| Determination of platelet aggregation
Washed human platelet suspensions were kindly provided by the Etablissement Français du Sang -Alsace (Strasbourg), and were prepared as previously described. 28

| Determination of oxidative stress
ECs were seeded into Lab-Tek ® chamber slide for 24 h, then exposed to serum-free MCDB 131 (Invitrogen) for 6 hours. The redox-sensitive fluorescent dye dihydroethidium (DHE) was used to evaluate the formation of reactive oxygen species (ROS). ECs were incubated with DHE (5 μmol/L) for 30 minutes at 37°C in a light protected manner. ECs were then washed and mounted in DAKO medium (fluorescence medium, DAKO, Les Ulis, France) and examined under confocal microscope (Leica SP2 UV DM Irbe). Images were analysed using Image J software.

| Determination of glucose uptake analysis by flow cytometry
ECs were seeded at a density of 2 × 10 5 cells per well in a 6 well plate and incubated overnight. Then, ECs were incubated in serum-free medium for 6 hours before being washed once with PBS and incubated with 100 μmol/L of 2-(N- for 1 h at 37°C. Thereafter, ECs were trypsinized before being centrifuged at 300 g for 5 minutes at room temperature and washed once with PBS. Cell pellets were resuspended in 300 μl PBS and the 2-NBD-glucose fluorescence was determined in the FITC channel (FL-1) using a flow cytometer (FACScan). Mean fluorescence intensity of 2-NBD-glucose was used to measure glucose uptake by ECs.
Unstained control was used to optimize FACS settings.

| Immunofluorescence studies
For immunofluorescence histochemistry, 0.14 μm cryomicrotome sections of porcine coronary artery segments were fixed with para-

| Statistical Analysis
Data are presented as mean ± SEM of n different experiments. Mean values were compared using Student's paired t test or an analysis of variance followed by the post-hoc Bonferroni test to identify significant differences between treatments using GraphPad Prism (v5.0).
The difference was considered to be significant when the P value was less than 0.05.

| High glucose-induced endothelial senescence is prevented by empagliflozin
The role of SGLT2 in HG-induced endothelial senescence, an early event promoting endothelial dysfunction, 31  gliflozin amounted to about 63% at 100 nmol/L ( Figure 1A). In contrast, empagliflozin did not affect the low level of senescence in ECs at P1 and replicative senescence assessed in ECs at P3 ( Figure 1B).
Since oxidative stress has been involved in HG-induced ECs senescence, 19 experiments were performed to determine the possibility that empagliflozin affects oxidative stress-induced premature ECs induced increased SA-β-gal activity was markedly prevented by the antioxidant N-acetyl-cysteine but not by empagliflozin ( Figure 1C) indicating that the SGLT2 inhibitor does not affect the signal transduction pathway leading to senescence in response to oxidative stress. Consistent with an increased SA-β-gal activity, HG up-regulated the expression level of the senescence markers p21 and p16 but not p53, and this effect was prevented by empagliflozin ( Figure 1D).
Empagliflozin alone did not affect the low basal expression level of p53, p21 and p16 ( Figure 1D).

| Empagliflozin prevents glucotoxicity in ECs
Since oxidative stress plays a key role in HG-induced endothelial senescence and dysfunction, 19  but not COX-1, all of these effects were prevented by empagliflozin ( Figure 2C).

| High glucose-induced endothelial dysfunction is prevented by empagliflozin
Since HG and associated oxidative stress are known to promote an impaired eNOS-derived NO formation and availability, and to activate the expression of pro-atherothrombotic responses in ECs, 18,19 experiments have evaluated whether empagliflozin is able to pre- prevented the platelet anti-aggregatory effect, 32 it can be attributed to eNOS-derived NO. In contrast, HG-treated ECs had a reduced ability to inhibit platelet aggregation whereas empagliflozin partially but significantly restored the anti-aggregatory effect of the HG-treated ECs ( Figure 3C). The empagliflozin treatment alone did not affect the anti-aggregatory effect of ECs ( Figure 3C).

| Role of the local angiotensin system in the HGinduced ECs senescence
Since previous studies have indicated that both AT1 receptor antagonists and inhibitors of angiotensin-converting enzyme F I G U R E 1 The selective SGLT2 inhibitor empagliflozin prevents the high glucose-induced ECs senescence but not replicative senescence and H 2 O 2 -induced senescence. A, ECs at P1 are exposed either to normal (5 mmol/L, NG) or high glucose (25 mmol/L, HG) in the absence or presence of increasing concentrations of empagliflozin for 96 h. B, ECs at P1 or P3 are exposed to increasing concentrations of empagliflozin for 48 h. C, ECs at P1 are exposed to N-acetyl cysteine (NAC, an antioxidant, 1 mmol/L) or empagliflozin (EMPA, 100 nmol/L) for 30 min before being exposed to 100 μmol/L of H 2 O 2 for 1 h. Thereafter, the medium is replaced and ECs are incubated for 48 h. SA-β-gal activity is determined by flow cytometry. D, ECs are exposed either to normal or high glucose for 96 h in the presence or absence of empagliflozin (100 nmol/L) before Western blot analysis of p53, p21, and p16. Results are shown as representative immunoblots (upper panels) and corresponding cumulative data (lower panels). Data are expressed as mean ± SEM of n = 3-4. *P < 0.05 vs. control NG and # P < 0.05 vs. control HG improve the endothelial function in type 2 diabetic patients and in experimental models of diabetes [33][34][35] and that Ang II is a strong inducer of endothelial senescence, 36 experiments were performed to determine the role of the local angiotensin system in HG-induced endothelial senescence and its modulation by empagliflozin. Both the ACE inhibitor perindoprilat and the AT1R antagonist losartan prevented the HG-induced increase in SA-β-gal activity in ECs ( Figure 4A). In addition, HG-induced endothelial senescence was associated with an increased protein expression level of both ACE and AT1 receptors, both of these effects were abolished by empagliflozin ( Figure 4B). Empagliflozin alone affected neither the expression level of ACE nor that of AT1R in ECs (Figure 4 B).

| HG promotes SGLT1 and 2-mediated glucose uptake and expression of SGLT1 and 2 in ECs
Since previous studies have indicated that native and cultured ECs express SGLT1 mRNA and protein, whereas SGLT2 mRNA was not detectable, 26 Figure 5A,B). Thus, these findings suggest that about 40% of glucose uptake involves Na + -and glucose-dependent transport mechanisms. Although empagliflozin did F I G U R E 2 High glucose causes a redox-sensitive induction of ECs senescence and promotes oxidative stress in ECs involving NADPH oxidase and cyclooxygenases, which is inhibited by empagliflozin. A, ECs are exposed to either normal or high glucose for 24 h in the presence or absence of empagliflozin (100 nmol/L), before dihydroethidium staining. Ethidium fluorescence is determined by confocal microscope. B, ECs are exposed either with N-acetyl cysteine (NAC, 1 mmol/L), VAS-2870 (VAS, a NADPH oxidase inhibitor, 5 μmol/L), or Indomethacin (INDO, a COX inhibitor, 30 μmol/L) for 30 min before the addition of high glucose for 96 h, and the subsequent determination of SA-β-gal activity by flow cytometry. C, ECs are exposed to normal or a high glucose for 96 h in the presence or absence of empagliflozin (100 nmol/L) before Western blot analysis of the NADPH oxidase subunits p22 phox , p47 phox , COX-1 and COX-2. Results are shown as representative immunoblots (upper panels) and corresponding cumulative data (lower panels). Data are expressed as mean ± SEM of n = 4-5. *P < 0.05 vs. control normal glucose and # P < 0.05 vs. control high glucose not affect glucose uptake into ECs, LX-4211 significantly reduced the Na + -and glucose-dependent uptake by about 52.88% indicating a major role of SGLT1 in ECs. In addition, H 2 O 2 increased basal 2-NBD-glucose uptake into ECs by about 27% or about 46% when expressed relative to the Na + -and glucose-dependent transport mechanism, and high glucose by about 18% or 30% ( Figure 5C,D). Both the stimulatory effect

| HG increases SGLT1 and 2 protein expression levels in native endothelium of coronary artery segments
Since all investigations were performed with cultured ECs, the possibility that HG increases the protein expression level of SGLT1 and 2 in native endothelium, and its potential consequences on responses promoting endothelial dysfunction were assessed using freshly harvested porcine coronary artery segments with endothelium and in the presence of SGLTs inhibitors. A low SGLT1 immunofluorescence signal was detectable predominantly at the luminal surface of the coronary artery whereas that for SGLT2 was barely detectable ( Figure 7A

| D ISCUSS I ON
The major findings of this study indicate that the selective SGLT2 inhibitor empagliflozin very effectively prevented the HG-induced premature endothelial senescence and dysfunction as characterized by the down-regulation of the eNOS-derived NO formation, the expression of pro-atherosclerotic molecules including tissue factor and VCAM-1, and a reduced platelet anti-aggregatory activity. The protective effect of empagliflozin involves its ability to prevent the F I G U R E 4 Up-regulation of the local angiotensin system mediates the high glucose-induced ECs senescence, an effect prevented by empagliflozin. A, ECs are exposed to either an angiotensin-converting enzyme inhibitor (Perindoprilat, 0.1 μmol/L) or an AT1R antagonist (Losartan, 30 μmol/L) for 30 min before the addition of high glucose for 96 h, and the subsequent determination of SA-β-gal activity by flow cytometry. B, ECs are exposed to normal or a high glucose for 96 h in the presence or absence of empagliflozin (100 nmol/L) before Western blot analysis of angiotensin-conversion enzyme (ACE), and AT1R. Results are shown as representative immunoblots (upper panels) and corresponding cumulative data (lower panels). Data are expressed as mean ± SEM of n = 3-4. *P < 0.05 vs. control normal glucose and # P < 0.05 vs. control high glucose HG-induced NADPH oxidase-and COXs-mediated oxidative stress triggering the subsequent activation of the pro-senescent local angiotensin system and is explained, at least in part, by the up-regulation of SGLT2 contributing to promote glucose entry into ECs.
Importantly, the effect of empagliflozin has been observed within the range of 1-100 nmol/L, and is in good agreement with the IC 50 value and drug exposure in clinical practice. Altogether, these observations indicate that SGLT2 appears to be an interesting target to protect the vascular system in diabetes, and possibly also to retard ageing-and cardiovascular risk factor-related endothelial dysfunction.
The recent EMPA-REG OUTCOME trial has indicated that empagliflozin reduced the risk of major adverse cardiovascular events in type 2 diabetic patients with established cardiovascular disease with a remarkable 38% relative risk reduction in death from cardiovascular causes and a 35% relative risk reduction in F I G U R E 5 LX-4211 but not empagliflozin inhibits basal glucose entry into ECs whereas both SGLT inhibitors prevent the increased glucose entry into H 2 O 2 -and high glucose-treated ECs. After a 6-h incubation period in serum-free medium without glucose, ECs are incubated with either empagliflozin, LX-4211 (a dual SGLT1 and 2 inhibitor) or high glucose (25 mmol/L) for 30 min in the presence (A) or absence of sodium replaced by choline chloride (B) before the addition of 2-NBD-glucose for 1 h. C,D, ECs are either untreated or exposed to H 2 O 2 for 24 h, and high glucose for 48 h. After a 6-h incubation period in serum-free medium without glucose, ECs are exposed to either empagliflozin or LX-4211 for 30 min before the addition of 2-NBD-glucose for 1 h. Thereafter, the ECs-associated 2-NBD-glucose signal was determined by flow cytometry. Results are shown as cumulative data (left panels) and representative flow cytometry overlay histograms (right panels). Data are expressed as mean ± SEM of n = 3-5. *P < 0.05 vs. respective control and # P < 0.05 vs control H 2 O 2 or high glucose F I G U R E 6 Effect of H 2 O 2 and high glucose on SGLT1 and SGLT2 mRNA and protein expression in ECs, and role of AT1R, NADPH oxidase and COXs. A,C) ECs are exposed to H 2 O 2 (100 μmol/L) and, thereafter, the expression level of SGLT1 and 2 mRNA was determined after a 1-h incubation period by RT-PCR (A), and SGLT1 and 2 protein after a 24-h period by Western blot analysis (C). B,D) ECs are exposed to high glucose and, thereafter, the expression level of SGLT1 and 2 mRNA was determined after a 4-h incubation period by RT-PCR (B), and SGLT1 and 2 protein after a 96-h period by Western blot analysis (D). E,F) ECs are exposed to either an AT1R antagonist (Losartan, 1 μmol/L) VAS-2870 (VAS, a NADPH oxidase inhibitor, 1 μmol/L), or indomethacin (INDO, a COX inhibitor, 30 μmol/L) for 30 min before the addition of high glucose for 96 h and, thereafter, the expression level of SGLT1 and SGLT2 was assessed by Western blot analysis. Results are shown as representative immunoblots (upper panels) and corresponding cumulative data (lower panels). Data are expressed as mean ± SEM of n = 3-6. *P < 0.05 vs respective control and # P < 0.05 vs high glucose hospitalization for heart failure. 22 Although the mechanisms behind the cardiovascular protective benefits of empagliflozin still remain unclear, several potential mechanisms have been involved such as decreases in arterial stiffness and systolic blood pressure, natriuresis, diuresis and reduced albuminuria, changes in the lipid profile with reduced triglycerides level, reduced sympathetic nervous system activity, improved myocardial energetics by augmenting ketone body oxidation, and small reductions of hyperglycemia, body weight and visceral adiposity. 37,38 In addition, experimental studies have revealed that empagliflozin improves endothelial dysfunction in streptozotocin-induced type 1 diabetic rats and in Zucker diabetic fatty rats (type 2 diabetes experimental model 23,39 ), and ipragliflozin in streptozotocin-induced type 1 diabetic mouse. 24   ECs is consistent with previous observations. Indeed, the BTBR ob/ ob type 2 diabetic mice is characterized by about a 50% reduction in the SGLT1 mRNA level associated with a 2-fold higher SGLT1 protein level in the renal cortex. 45 Moreover, the injection of a glucose solution directly into the intestinal lumen of ruminant sheep for 4 days resulted in an increased SGLT1 mRNA level by about 2-fold in the proximal intestine whereas SGLT1 protein and activity levels were increased by about 60-90-fold. 46 Thus, these findings indicate that the principal level of SGLT1 regulation by luminal sugar is translational or post-translational. In addition, a 2.5-fold higher SGLT2 protein level has been observed in the renal cortex of adult female compared to male rats despite a similar expression level of SGLT2 mRNA. 47 Thus, these previous findings in conjunction with the present ones indicate that changes in SGLT1 and SGLT2 mRNA levels are not associated with corresponding changes in the respective protein levels demonstrating the involvement of complex regulatory mechanisms that still remain to be clarified.
Moreover, the present findings provide also evidence that HG is able to up-regulate markedly SGLT1 and, also to some extent, SGLT2, almost exclusively in the native endothelium of porcine coronary arteries, and that this effect is associated with a down-regulation of eNOS and an up-regulation of VCAM-1 in the endothelium.
Since both empagliflozin and LX-4211 very effectively prevented all stimulatory effects of HG on the endothelium, both SGLTs appear to play a key role.
In conclusion, the major novel findings of this study indicate that SGLT2, as well as SGLT1, is expressed in cultured and native ECs under pathological conditions including hyperglycemic conditions and oxidative stress. They further indicate that SGLT2 appears to contribute to excessive glucose entry promoting endothelial senescence and dysfunction, in part, via the local angiotensin system acting in a feed forward mechanism to sustain oxidative stress. Since SGLT2 expression is up-regulated by oxidative stress in ECs, SGLT2 appears to be an interesting target to prevent the induction of endothelial senescence and dysfunction and, hence, to retard premature vascular ageing.

ACK N OWLED G EM ENTS
We thank Romain Vauchelles and Sophie Martin (UMR CNRS 7213, Faculty of pharmacy, Strasbourg University) for technical assistance in microscopy and image analysis, and RT-PCR analysis, respectively, and Claudine Ebel (Flow cytometry core facility, IGBMC, Strasbourg). wrote the paper.