Angiotensin II Upregulates Sodium-Glucose Co-Transporter2 (SGLT2) Expression and SGLT2 Inhibitor Attenuates Ang II-Induced Hypertensive Renal Injury in Mice

Clinical trials indicate that sodium-glucose co-transporter 2 inhibitors (SGLT2i) improve kidney 3 function, yet, the molecular regulation of SGLT2 expression is incompletely understood. Here, 4 we investigated the role of the intrarenal renin-angiotensin-system (RAS) on SGLT2 expression. 5 In adult non-diabetic participants in the Nephrotic Syndrome Study Network (NEPTUNE, 6 N=163), multivariable linear regression analysis showed SGLT2 mRNA was significantly 7 associated with angiotensinogen (AGT), renin, and angiotensin converting enzyme (ACE) 8 mRNA levels (p<0.001). In vitro , angiotensin II (Ang II) dose-dependently stimulated SGLT2 9 expression in HK-2, human immortalized renal proximal tubular cells (RPTCs); losartan and 10 antioxidants inhibited it. Sglt2 expression was increased in transgenic mice specifically 11 overexpressing Agt in their RPTCs, as well as in WT mice with a single subcutaneous injection 12 of Ang II (1.44 mg/kg). Moreover, Ang II (1000 ng/kg/min) infusion via osmotic mini-pump in 13 WT mice for 4 weeks increased systolic blood pressure (SBP), glomerulosclerosis, 14 tubulointerstitial fibrosis, and albuminuria; canaglifozin (Cana, 15 mg/kg/day) reversed these 15 changes, with the exception of SBP. Fractional glucose excretion was higher in Ang II+Cana 16 than WT+Cana, whereas Sglt2 expression was similar. Our data demonstrate a link between intrarenal RAS and SGLT2 expression and that SGLT2i ameliorates Ang II-induced renal injury 18 independent of SBP.

Recent large clinical trials have reported that sodium-glucose cotransporter 2 inhibitors 2 (SGLT2i) have renoprotective effects in a glucose-independent manner in patients with type 2 3 diabetes [1][2][3]. Despite the compelling evidence, the mechanisms underlying renoprotection are 4 incompletely understood. One proposed mechanism is that SGLT2i exerts renal hemodynamic 5 effects by reducing the intraglomerular pressure via the activation of tubulo-glomerular feedback 6 [4,5]. Other studies have suggested that SGLT2i may decrease the inflammatory and fibrotic 7 responses to hyperglycemia in renal proximal tubular cells (RPTCs) [6,7] and mesangial cells 8 [8]. Based on such findings, SGLT2i has been proposed as a potential renoprotective agent for 9 patients with non-diabetic chronic kidney disease (CKD) [9]. Indeed, results from recent clinical 10 trials have indicated that SGLT2i exerts a positive effect in non-diabetic CKD [10,11]. 11 In order to assess the efficacy of SGLT2i in various CKD, it is important to consider the 12 endogenous expression level of SGLT2. While some studies reported increased SGLT2 13 expression in type 1 and 2 diabetes in murine experimental models [12] and humans [13], others 14 found decreased SGLT2 mRNA expression in kidney biopsies of patients with diabetes 15 compared to healthy controls or patients with non-diabetic glomerular diseases in a cohort of 16 patients with samples in the European Renal cDNA Bank [14]. There is only limited information 17 available on SGLT2 expression in non-diabetic CKD. Furthermore, it is uncertain whether the 18 clinical efficiency of SGLT2i corresponds to endogenous SGLT2 expression levels. 19 We previously showed unchanged expression of renal angiotensinogen (Agt, the sole 20 precursor of all angiotensins) in non-diabetic [15] or diabetic [16] mice treated with SGLT2i. 21 Here, we investigated the impact of increased renal Agt and its down-stream angiotensin II (Ang 22 II) on SGLT2 expression. Moreover, we also asked whether altered SGLT2 expression by Ang II 23 (N=16) [23]. Some mice (Ang II-Cana, N=8) received canagliflozin in the drinking water (15 1 mg/kg/day) [24] starting from the second week. Mice without Ang II infusion received a sham 2 surgery, and were divided into 2 groups; Ctrl (N=8) with regular water and Ctrl-Cana (N=8) with 3 canagliflozin (15 mg/kg/day) for the same duration. 4 5

Physiological measurements in chronic Ang II infusion study 6
The mice were housed in a temperature-controlled room regulated on a 12-hour light/dark cycle 7 and had free access to standard mouse chow and water. The concentration of canagliflozin was 8 adjusted at least 3 times a week based on the amount of water intake. Two weeks before the 9 pump implantation surgery, mice were acclimated to SBP measurements daily for a week. 10 Starting from 1 week before to 4 weeks after surgery, SBP was measured at least 3 times per 11 week. BW was measured weekly. Non-fasting BG was measured weekly by Accu-Chek 12 Performa (Roche Diagnostics, Laval, QC, Canada). 13 Twenty-four hours before euthanasia, the mice were housed in individual metabolic cages, 14 food intake and water consumption were measured, and urine was collected. The GFR was 15 estimated from fluorescein isothiocyanate inulin plasma clearance in conscious mice as 16 recommended by the Animal Models of Diabetic Complications Consortium 17 (http://www.diacomp.org/) with slight modifications [25,26]. Immediately after the GFR 18 measurements, mice were euthanized with intraperitoneal administration of pentobarbital. The 19 kidneys and hearts were weighed. Kidneys were harvested for histology and RPT isolation by 20 Percoll gradient [15,19]. 21 22

Mouse serum and urine biochemical measurements 23
We measured urinary albumin and glucose concentration using ELISA kit Albuwell (Exocell,1 Inc., Philadelphia, PA) and colorimetric-enzymatic method Autokit Glucose (Wako Diagnostics, 2 Richmond, VA), respectively. Urine samples were extracted with C18 Sep-Pak columns (Waters,3 Mississauga, ON) and assayed for Ang II by specific ELISA (Bachem America, Torrence, CA) 4 according to the number III protocol [15]. Serum sodium, potassium, creatinine and BUN, as 5 well as urine sodium, potassium and creatinine were measured by the Comparative Medicine and 6 Animal Resources Centre, McGill University (Montreal, QC, Canada) (N=6/group). Hematocrit 7 was determined using glass microcapillaries at the first time point of inulin clearance. 8 Fractional sodium excretion (FeNa) and fractional glucose excretion (FeGlu) were 9 calculated using the 24-hour urine volume, glomerular filtration rate (GFR), and serum and urine 10 sodium or glucose concentration [27,28]. Electrolyte-free water clearance was also calculated 11 using 24-hour urine volume, serum and urine sodium, and urine potassium, as previously 12 described [28]. 13 14

Statistical analysis 2
Continuous data are reported as mean ± standard error if normally distributed and as median and 3 interquartile range (IQR) for non-normally distributed data. Categorical data were expressed as 4 frequencies and percentages. For between group comparisons, one-way analysis of variance 5 (ANOVA) was used, followed by the Bonferroni post-hoc test. Pearson's correlation coefficient 6 was used to analyze correlation between continuous variables. In the NEPTUNE database, 7 univariable and then multivariable linear regression analyses were performed to identify the 8 independent determinants of SGLT2 mRNA expression. We confirmed that the presence of 9 multicollinearity was unlikely based on the variance inflation factors (VIFs). Statistical analyses 10 were performed using the statistical software packages SAS University Edition (SAS Institute, 11 Cary, NC) for NEPTUNE data and GraphPad Prism 8 (GraphPad Software, San Diego, CA) for 12 animal and in vitro data. P values of <0.05 were considered statistically significant. 13 14

Ethics statement 15
All animals procedures were approved by the CRCHUM Animal Care Committee (approval# 16 M2R3 CM16016JCs) and performed at CRCHUM in accordance with the Principles of 17 We analyzed mRNA expression of SGLT2 and RAS components in kidney biopsies from 4 participants of the NEPTUNE cohort. SGLT2 mRNA expression from micro-dissected TI was 5 available from 183 adult participants; 163 non-diabetic patients with proteinuria and 20 control 6 subjects (15 nephrectomies due to cancer and 5 living donors). Baseline clinical characteristics 7 are shown in Table 1 (n=163) with the exception of controls, for which data were not available. 8 There was no difference in SGLT2 expression between the disease groups ( Figure 1A). SGLT2 9 mRNA positively correlated with AGT (r = 0.55, p<0.001), renin (r = 0.46, p<0.001) and 10 angiotensin converting enzyme (ACE) (r = 0.47, p<0.001) mRNA ( Figure 1B-D). There was a 11 weak but statistically significant negative correlation between SGLT1 and SGLT2 mRNA levels 12 (r = -0.28, p<0.001; Figure 1E). These variables were analyzed with multivariable linear 13 regression to identify the independent associations with SGLT2 expression ( Table 2). In 14 univariable analysis, SGLT2 mRNA positively correlated with eGFR and negatively correlated 15 with interstitial fibrosis ( Table 2), which was consistent with a previous report [14]. 16 Multivariable analysis revealed AGT, renin, ACE, and SGLT1 mRNA as independent 17 determinants of the SGLT2 mRNA expression ( Table 2). RAS blocker use was not associated 18 with altered SGLT2 expression ( Table 2). 19 20

Ang II up-regulates SGLT2 expression in HK-2 cells 21
Next, we investigated the impact of Ang II on SGLT2 expression in vitro, using the 22 human immortalized renal proximal tubular cell line (HK-2). Culture of HK-2 cells with Ang II 23 led to induction of SGLT2 mRNA in a concentration-dependent manner (Figure 2A). Incubation 1 of HK-2 cells with losartan (AT1R antagonist) abolished the effects of Ang II (Figure 2A), 2 indicating that Ang II stimulates SGLT2 expression via AT1R. Human SGLT2 promoter activity 3 ( Figure 2B) and SGLT2 immunostaining (Figure 2C, D) in response to Ang II with and without 4 losartan confirmed this observation. Furthermore, the stimulatory effect of Ang II on SGLT2 5 promoter activity was attenuated by catalase (H 2 O 2 decomposer enzyme), apocynin and DPI 6 (inhibitors of NADPH oxidase (NOX)) ( Figure 2E), indicating that activation of NADPH 7 oxidase followed by reactive oxygen species (ROS) generation is required for Ang II to stimulate 8

SGLT2. 9
Furthermore, SGLT2's glucose transport activity was evaluated by the entry of 10 fluorescence-labeled glucose 2-NBDG into the HK2 cells. Ang II significantly increased glucose 11 uptake to the HK2 cells, which was abolished by losartan treatment (Figure 2F). We have generated Tg mice overexpressing Agt with and without Cat specifically in their 16 RPTCs (Agt-Tg and Agt/Cat-Tg, respectively). Agt-Tg mice have a high risk of developing mild 17 hypertension, albuminuria, and renal injury [19,22,35,36], and these abnormalities are 18 attenuated in Agt/Cat-Tg mice [21]. Figure 3A shows enhanced Sglt2-immunopositive staining 19 in Agt-Tg compared with WT mice, and these increases were attenuated by Cat overexpression 20 and abolished by losartan. Since increased urinary Ang II levels in Agt-Tg mice were attenuated 21 by losartan treatment [22], we speculated that increased renal Ang II generated from up-22 regulated renal Agt ( Figure 3B) likely stimulates SGLT2 expression via oxidative stress. Semi-1 quantitative analysis of Sglt2 immunostaining was consistent with this notion ( Figure 3C). 2 3

Single-dose Ang II injection stimulates Sglt2 expression in mice 4
Next, we tested whether systemic Ang II injection could stimulate renal Sglt2 expression 5 in vivo. WT mice received a single bolus injection of Ang II (1.44 mg/kg) or its vehicle (0.9% 6 NaCl) subcutaneously (N=6/each group), and were euthanized 18 hours later ( Figure 4A). A 7 separate group of mice received losartan (15 mg/kg/day in the drinking water) for 7 days before 8 the Ang II injection (N=6). Seven-day treatment with losartan did not affect SBP (115.9±1.6 9 mmHg without losartan, n=12 vs 116.2±4.7 mmHg with losartan, n=6). Injection of Ang II 10 increased Sglt2 mRNA and protein expression ( Figure 4B-D) and these changes were attenuated 11 by losartan. Although up-regulated trend of Agt mRNA expression was observed, the change did 12 not reach statistical significance ( Figure 4B). 13

SGLT2 Inhibitor attenuates Ang II-induced hypertensive renal injury in mice 15
Having shown that Ang II up-regulates SGLT2 expression, we investigated whether 16 inhibition of SGLT2 could ameliorate Ang II-induced hypertensive renal injury in mice. We 17 administered high-dose Ang II (1,000 ng/kg/min, corresponding to 1.44 mg/kg/day) via 18 subcutaneously-implanted osmotic minipumps in WT mice for 4 weeks. One week after 19 initiating Ang II infusion, mice in Ang II-Cana group were started on an SGLT2i, canagliflozin 20 (15 mg/kg/day) in the drinking water ( Figure 5A). Control mice underwent sham surgery, and 21 canagliflozin (15 mg/kg/day) was added to the drinking water one week after the sham surgery 22 (Ctrl-Cana). Physiological and renal pathological changes in these 4 groups (N=8/group) were 1 compared at week 4. 2 SBP was significantly higher in both the Ang II and Ang II-Cana groups (p<0.001) 3 compared with controls; there was no detectable differences between SBP in the two groups 4 ( Figure 5B). Mice receiving Ang II infusion with or without canagliflozin had lower body 5 weight ( Figure 5C, p<0.001), consistent with the appetite suppressing action of high-dose Ang 6 II [37, 38]. However, food intake was significantly higher in the Ang II-Cana group compared 7 with the Ang II group, likely resulting from a compensation of the caloric loss from glycosuria 8 ( Table 3). Urine volume and water intake were significantly elevated in the Ang II-Cana group 9 (Table 3). Urinary glucose concentration was similarly increased in mice treated with 10 canagliflozin, regardless whether they received Ang II or not ( Figure 5D), but the 24-hour urine 11 glucose excretion was significantly increased in Ang II-Cana compared with Ctrl-Cana (Table 3). 12 Blood glucose (BG) levels were not significantly changed among the 4 groups ( Figure 5E). 13 FeGlu was considerably higher in the Ang II-Cana group than in any other groups studied 14 ( Figure 5F). Ang II with and without Cana evoked increases in hematocrit ( Table 3). As 15 anticipated, urinary potassium excretion was increased in the Ang II-Cana group without causing 16 detectable hypokalemia (Table 3). Ang II-Cana had significantly increased serum sodium level 17 and electrolyte free water clearance ( Table 3). 18 Ang II group had significantly lower GFR/BW ratio ( Figure 5G), as well as significantly 19 elevated serum creatinine and BUN levels ( Table 3). Kidney weight (KW)/tibia length (TL) ratio 20 was the lowest in the Ang II group. Canagliflozin did not reduce Ang II-induced increases in 21 heart weight (HW) ( Table 3). HW/BW strongly correlated with week 4 SBP regardless of the 22 treatment (Supplementary Figure 1, r=0.69, p<0.001). At week 1, just before canagliflozin 23 treatment was initiated, there was no difference in urinary albumin/creatinine ratio between 1 control and Ang II group, 30.8 ± 7.7 g/mg vs. 23.1 ± 7.8 g/mg, respectively (N=6 in each 2 group, NS). Urinary albumin excretion after 4 weeks was significantly increased in the Ang II 3 group compared to Ctrl, and was attenuated in the Ang II-Cana group ( Figure 5H). Ang II 4 infusion evoked mesangial expansion, glomerulosclerosis, and tubulointerstitial fibrosis, and 5 these changes were partially reversed by canagliflozin treatment (Figure 6A-D). 6 Unexpectedly, Sglt2 and Agt expressions were similar in the groups studied ( Figure 7A hypertensive kidney injury [19,36,40]. Our findings may potentially lead to a new treatment 18 option for hypertensive nephrosclerosis, which is the second most common cause of CKD and 19 end-stage kidney disease in the world [41]. 20 Here, we demonstrated that Ang II dose-dependently increases SGLT2 expression in 21 vitro, and that Agt-Tg mice as well as WT mice treated with a bolus injection of Ang II have up-22 regulated Sglt2 expression. These findings are consistent with previous findings of Bautista et al. 23 [42], who reported increased Sglt2 expression in rats with aortic coarctation, a model of 1 renovascular hypertension, and rats with low-dose Ang II infusion. Moreover, we demonstrated 2 the inhibitory role of antioxidants on the stimulation Sglt2 expression by Ang II in HK-2 cells 3 and in Agt/Cat-Tg mice, indicating increased oxidative stress or ROS generation may mediate 4 the stimulatory effect of Ang II on Sglt2 expression. 5 In contrast to the effects of bolus injection of Ang II, chronic Ang II infusion (1000 6 ng/kg/min subcutaneously for 4 weeks) resulted in no detectable increases in Sglt2 expression. 7 The reasons for these apparently contradictory observations are not clear at present. One 8 possibility is that sustained stimulation of AT1R by Ang II led to desensitization of AT1R, as has 9 been reported in many models for cardiovascular disease [43,44]. SGLT2's biphasic response to 10 different concentrations of Ang II was also observed in our in vitro study as Ang II at 10 -5 M 11 inhibited SGLT2 mRNA expression. Another possibility might be that Sglt2 expression is 12 affected by pressure natriuresis response, similar to Na + /H + exchanger 3 (NHE3), another sodium 13 transporter in the proximal tubules. It has been reported that a short-term low-dose Ang II (200 14 ng/kg/min for 3 days) in rats, which did not affect SBP, increased the expression of NHE3 Lower Sglt2 expression has also been reported following ischemia-reperfusion-induced tubular 2 cell injury in mice [48]. 3 The chronic high-dose Ang II infusion did not alter intrarenal Agt expression in this study, 4 which supports a previous finding by Gonzalez-Villalobos, et al. demonstrating that 1,000 5 ng/kg/min of Ang II infusion for 12 days did not increase intrarenal Agt expression levels while 6 400 ng/kg/min of Ang II infusion enhanced the Agt expression. Though we were aware of their 7 findings, we needed to use higher-dose of Ang II to detect renal pathological changes by Ang II 8 in C57BL/6 mice which are known to be relatively resistant to kidney injury [49]. 9 The present study showed that canagliflozin prevented Ang II-induced hypertensive renal 10 injury in mice. It has been proposed that the potential mechanisms of renoprotection with 11 SGLT2i include improved renal hemodynamics via tubule-glomerular feedback [50], lowered 12 systemic hypertension [51], and direct anti-fibrotic effects [52,53]. In our study, canagliflozin 13 did not lower GFR or systemic blood pressure in mice with Ang II infusion. However, it 14 decreased glomerular and tubulointerstitial fibrosis shown in pathology as well as profibrotic 15 gene expressions in isolated RPTCs. Therefore, it can be speculated that canagliflozin prevented 16 renal damage by Ang II via anti-fibrotic effects. 17 It is important to note that the Ang II-Cana group had significantly higher FeGlu than 18 Ctrl-Cana, indicating that the glycosuric effect of canagliflozin was enhanced by co-19 administration of Ang II compared with the effects of canagliflozin in euglycemice mice. These 20 findings would suggest that the effects of SGLT2i may not necessarily correlate with SGLT2 21 mRNA or protein expression. Indeed, Ang II may directly enhance the effect of SGLT2i. It has 22 been proposed that increased SGLT2 expression would render SGLT2 an attractive target for 23 inhibition, thus increasing the efficacy of the inhibitors [14,48,54]. To date, several 1 retrospective studies, which were conducted to determine the clinical parameters for monitoring 2 the therapeutic efficacy of SGLT2i in type 2 diabetes patients, consistently indicated that higher 3 initial HbA1c and eGFR are associated with a better glycemic response [55][56][57]. Identifying 4 intrarenal RAS as one of the determinants of the efficacy of SGLT2i is of clinical relevance for 5 glucose control in type 2 diabetes patients. 6 Another novel aspect of the present study is that the diuretic action of canagliflozin was 7 enhanced by Ang II. Indeed, Ang II-Cana mice had the highest urine volume, water intake, 8 serum sodium level, electrolyte free water clearance, and hematocrit among the groups studied. Another important finding is the analysis of the large human kidney gene database of the 22 NEPTUNE cohort. We detected significant positive correlations of AGT, renin, and ACE gene 23 expressions with SGLT2 mRNA expression. In multivariable analysis, those factors previously 1 reported to correlate with SGLT2 expression in human kidneys, such as interstitial fibrosis and 2 eGFR [14], were no longer associated with SGLT2 after adjustment for RAS gene expression. 3 Since AGT is the sole precursor of all angiotensin peptides and renin is the rate-limiting enzyme 4 to generate Ang II, we surmise that the downstream molecule Ang II is likely an important 5 regulator of SGLT2 expression in human kidneys. 6 Our study has several limitations. First, the molecular mechanism of SGLT2 regulation 7 by ROS remains to be investigated. Second, we could not directly analyze the relation between 8 renal Ang II level and SGLT2 expression in the human biopsy material. Though RAS blocker 9 use was not associated with SGLT2 expression in NEPTUNE participants, we don't know how 10 much of the generated RAS was successfully blocked by RAS blockers they were taking. Third, 11 our chronic Ang II infusion study did not include a group that was treated simultaneously with a 12

RAS blocker and SGLT2i. 13
In conclusion, the present study demonstrates a link between intrarenal RAS and SGLT2 14 expression both in humans and mice. Consistently, treatment with SGLT2i effectively attenuated 15 Ang II-induced hypertensive renal injury in mice independent of changes in SBP. Our data 16 provide a rational basis for future clinical studies of SGLT2i in non-diabetic kidney disease. 17 Further understanding of the mechanism of SGLT2 regulation has clinical relevance and 18 provides important insights into the reno-and cardio-protective actions of SGLT2i. 19 20 Clinical Perspective 21 1. Hypertensive nephrosclerosis, the second most common cause of chronic kidney disease and 1 end-stage kidney disease in the world, is associated with an inappropriately activated intrarenal 2

RAS. 3
2. Our present study demonstrates an association between intrarenal RAS and SGLT2 expression 4 both in humans and mice. 5 3. SGLT2 inhibitors might have a role for the prevention and/or treatment of hypertension-6 associated deterioration of kidney function. 7 8

Data Availability 9
The data used to support the findings of this study are available from the corresponding author 10 upon request.     RAS, renin-angiotensin system; BP, blood pressure; eGFR, estimated glomerular filtration rate; UPCR, urine protein/creatinine ratio. *Variables entered in the model: eGFR, interstitial fibrosis, log 2 SGLT1 mRNA, log 2 AGT mRNA, log 2 Renin mRNA, and log 2 ACE mRNA.