Empagliflozin reduces renal lipotoxicity in experimental Alport syndrome

Sodium-glucose cotransporter-2 inhibitors (SGLT2i) are anti-hyperglycemic agents that prevent glucose reabsorption in proximal tubular cells. SGLT2i improves renal outcomes in both diabetic and non-diabetic patients, indicating it may have beneficial effects beyond glycemic control. Here, we demonstrate that SGLT2i affects energy metabolism and renal lipotoxicity in experimental Alport syndrome (AS). In vitro, we found that SGLT2 protein expression levels in human and mouse podocytes were similar to tubular cells. Newly established immortalized podocytes from Col4a3 knockout mice (AS podocytes) accumulate lipid droplets along with increased apoptosis when compared to wildtype podocytes. Treatment with SGLT2i empagliflozin reduces lipid droplet accumulation and apoptosis in AS podocytes. Empagliflozin inhibits the utilization of glucose/pyruvate as a metabolic substrate in AS podocytes. In vivo, we demonstrate that empagliflozin reduces albuminuria and prolongs the survival of AS mice. Empagliflozin-treated AS mice show decreased serum blood urea nitrogen and creatinine levels in association with reduced triglyceride and cholesterol ester content in kidney cortices when compared to AS mice. Lipid accumulation in kidney cortices correlates with the decline in renal function. In summary, empagliflozin reduces renal lipotoxicity and improves kidney function in experimental AS in association with the energy substrates switch from glucose to fatty acids in podocytes.

SGLT2i. Here, we aimed at investigating if podocytes and tubular cells change their preferences 8 with regard to their metabolic fuel in response to SGLT2i. In the kidney, podocytes highly rely on 9 glucose as the substrate for ATP production (Abe et al., 2010), while tubular cells use free fatty 10 acid as the preferred energy source (Kang et al., 2015). Therefore, a reduction of glucose 11 availability by SGLT2i may trigger the utilization of alternative energy substrates, such as lipids 12 (Osataphan et al., 2019). To study the effect of SGLT2i on energy substrate switch, we first 13 investigated whether SGLT2i exercises a protective effect on immortalized podocytes and tubular 14 cells derived from AS mice. We demonstrate that AS podocytes have increased cytotoxicity and 15 apoptosis when compared to WT podocytes ( Figure 2). Empagliflozin treatment protected AS 16 podocytes from apoptosis but not cytotoxicity. On the other hand, AS tubular cells did not show 17 increased apoptosis but were found to exhibit a tendency to increased cytotoxicity when 18 compared to WT tubular cells. Tubular cytotoxicity was significantly reduced by empagliflozin 19 treatment. The apparent discrepancy between cytotoxicity and apoptosis could be explained by 20 the fact that apoptosis is a coordinated and energy-reliant process that involves the activation of 21 caspases, while cell death characterized by loss of cell membrane integrity (which was measured 22 in our cytotoxicity assay) is energy-independent (Cummings & Schnellmann, 2004), therefore the 23 two processes can take place independently, sequentially, as well as concurrently (Elmore, 2007). 24 Interestingly, we observed a similar trend with regard to LD accumulation. We found a significantly 25 increased number of LDs in AS podocytes compared to WT podocytes, which was reduced by 26 empagliflozin treatment. However, no difference in LD accumulation was observed in tubular cells 1 (Figure 2), suggesting that the mechanisms leading to LD accumulation in podocytes in AS are 2 cell-specific. As podocytes but not tubular cells in AS are in contact with an abnormal glomerular 3 basement membrane, the possibility that the LD accumulation is the result of a cross talk between 4 matrix and lipid metabolism is possible, as it was recently suggested by others (Romani et al.,  To test the preference of energy substrate in association with AS and empagliflozin treatment, we appears to promote a shift in substrate utilization for ATP production. The accumulation of LD in 20 AS podocytes could lead to an increase in the availability of fatty acids and contribute to the 21 elevated FAO-linked respiration observed in these cells. Given that podocytes rely more on 22 glucose oxidation and are therefore more vulnerable to glucose deprivation, it is feasible to 23 speculate that empagliflozin only affects podocyte respiratory metabolism and not that of tubular 24

cells. 25
While the major mechanisms by which SGLT2i reduces albuminuria is thought to be linked to a 26 modulation of the tubulo-glomerular feedback resulting in improved glomerular hyperfiltration 1 was also included in our study in order to determine whether a combination of empagliflozin and 11 ramipril would have a superior effect to ramipril. We show that treatment of AS mice with both 12 ramipril and empagliflozin is not superior in preserving renal function when compared to treatment 13 with ramipril alone. While this is not consistent with the findings reported in patients with DKD, the 14 very sizable effect of ramipril in experimental AS may account for the inability to report additional 15 renoprotective effects of empagliflozin. Interestingly, we found that empagliflozin or the combined 16 treatment of empagliflozin and ramipril reduces triglyceride content in the kidney of AS mice, while 17 ramipril did not have any effect on the renal triglyceride content but affected esterified cholesterol  Our study has several limitations. First, we did not study empagliflozin's off-target effect on other 7 transporters such as sodium-hydrogen exchanger (NHE) 1 in the heart or NHE3 in the kidney 8 In summary, our study demonstrates that empagliflozin reduces renal lipotoxicity and improves 17 kidney function in experimental AS. This beneficial effect is associated with a shift in the use of 18 energy substrates from glucose to fatty acids in podocytes. Lipid accumulation in kidney cortices 19 correlates with kidney disease progression, therefore reducing renal lipid content by empagliflozin 20 and other agents may represent a novel therapeutic strategy for the treatment of patients with AS. 21 Results obtained from this study may allow us to better define the mechanisms leading to SGLT2i-22 mediated renoprotection in non-diabetic kidney disease. 23 24

Methods 25
Animal studies 26

Phenotypic analysis of mice 1
Col4a3 -/mice (a model of AS) are in a 129X1/SvJ background and were purchased from Jackson 2 Laboratory (129-Col4a3 tm1Dec /J, stock #002908). Mice were fed empagliflozin-supplemented chow 3 (70 mg/kg) versus a regular diet starting at 4 weeks of age. Ramipril was added to the drinking 4 water at a concentration that would lead to a daily uptake of 10 mg/kg body weight (Kim et al., 5 2021). Five groups of mice were examined: WT + placebo, AS + placebo, AS + empagliflozin, AS 6 + ramipril and AS + empagliflozin + ramipril. Both male and female mice were used. Mice were 7 sacrificed at 8 weeks and analyzed as described below. 8 Urinary albumin-to-creatinine ratios 9 Morning spot urine samples were collected bi-weekly. Urinary albumin-to-creatinine ratios were To measure podocyte number per glomerulus, glomerular sections embedded in OCT were 1 stained with a WT1 antibody (Santa Cruz Biotechnology, Dallas, TX, sc-192, 1:300), followed by 2 a secondary antibody (Invitrogen, Waltham, MA, A-11008, 1:500) and Mounting Medium with 3 DAPI (Vectorlabs, Newark, CA, H-1200). Images were acquired using Olympus IX81 confocal 4 microscope (Tokyo, Japan) coupled with a 60x oil immersion objective lens and images were 5 processed using Fiji/Image J. 5-10 glomeruli per mouse were quantified.

Triglyceride (TG) assay 3
The TG content was determined using Triglyceride Colorimetric Assay Kit (Cayman, Ann Arbor, 4 MI) following the manufacturer's protocol. TG standards and lipid samples from above-mentioned 5 extraction were added into a 96 well plate. The reaction was initiated by adding 150 µl enzyme 6 buffer to each well. Absorbance at 540 nm was measured using a SpectraMax M5 plate reader 7 (Molecular Devices, San Jose, CA). 8

Cytotoxicity and apoptosis assay 6
Cytotoxicity and apoptosis assays were performed using the ApoTox-Glo Triplex assay 7 (Promega, Madison, WI) according to the manufacturer's protocol. Fluorescence and 8 luminescence were measured on a SpectraMax i3x multi-mode microplate reader (Molecular 9 Devices, San Jose, CA). 10

Lipid droplet quantification 11
Cultured cells were fixed with 4% paraformaldehyde (PFA) and 2% sucrose and then stained with 12 Nile red (Sigma-Aldrich, St. Louis, MO) and HCS Cell Mask Blue (Invitrogen, Waltham, MA) 13 according to the manufacturer's protocols. Images were acquired using the Opera high content

Cellular respiration measurements 17
Oxygen consumption rate (OCR) was measured using a high-resolution respirometer (O2k-Fluo- endogenous respiration was measured in intact cells. For substrate-driven respiration, cells were 26 permeabilized with 2.5 ug/ml digitonin (optimal digitonin concentration for podocytes and tubular 1 cells was established prior following the SUIT-010 protocol) and supplemented with 2.5 mM ADP. 2 FAO-linked substrates (0.5 mM octanoylcarnitine plus 0.1 mM malate) were then added to the 3 chamber using a Hamilton microsyringe, and the coupled FA-driven OCR was measured. Finally, 4 2 mM malate, 5 mM pyruvate and 10 mM glutamate were added to initiate coupled NADH-linked 5 respiration, and the additive effect of NADH-driven OCR was measured. Mitochondrial outer 6 membrane integrity was tested by addition of 10 uM cytochrome c. Respiration was inhibited by 7 the addition of 100 mM sodium azide, which is a specific mitochondrial complex IV (CIV) inhibitor. 8 Cell respiration was recorded as pmol O2 consumed for 1 s and normalized to cell numbers. 9 Pyruvate dehydrogenase activity assay 10 Pyruvate dehydrogenase (PDH) activity in cells was determined using PDH Colorimetric Assay 11 Kit (BioVision, Milpitas, CA) according to manufacturer's protocol. 1x10 6 cells were used. 12 Absorbance at 450 nm was measured using a SpectraMax M5 plate reader (Molecular Devices, 13 San Jose, CA). 14

Statistics 15
For each statistical test, biological sample size (n), and p-value are indicated in the corresponding 16 figure legends. All values are presented as mean ± SD. Statistical analysis was performed using 17 Prism GraphPad 7 software. Significant outliers were determined by GraphPad outlier calculator 18 and excluded from further statistical analysis. Animals were grouped according to genotypes then 19 randomized, and investigators were blinded for the analyses. When comparing two groups, a two-20 tailed Student's t-test was performed. Otherwise, results were analyzed using One-way ANOVA 21 followed by Holm-Sidak's multiple comparison. A p-value less than 0.05 was considered 22 statistically significant. Only data from independent experiments were analyzed. 23

Study Approval 24
All studies involving mice were approved by the Institutional Animal Care and Use Committee 25   to the Katz family for supporting this study. 10 11

Data availability 12
All data generated or analyzed during this study are included in the manuscript and supporting 13 files.