In this study, we used male mice at the age of 12–16 weeks. The generation of mice with a podocyte-specific (Podo GC-A KO) or global deletion (GC-A KO) of the natriuretic peptide receptor guanylylcyclase-A (GC-A) has been described elsewhere (Lopez et al., 1995; Staffel et al., 2017). C57BL/6J mice were purchased from Charles River (Sulzfeld, Germany). Offspring of homozygous breeder pairs were used throughout the study. Animals had free access to food and water. All experimental procedures were conducted in accordance with the German Animal Welfare Act and approved by the local authorities (government of Lower Franconia, Germany, file number RUF 55.2.2-2532.2-896-13).

For all models (see overview Table 1), UNx was used to induce non-diabetic hyperfiltration in the remaining kidney. UNx was performed as described in a previous study by Staffel et al. (2017). No increased mortality after UNx intervention was observed.

Experimental group 1, UNx in wildtype mice (Tables 1, 2): As a proof of principle, the effect of EMPA on glomerular non-diabetic hyperfiltration in kidneys of uninephrectomized (UNx) wildtype (WT) animals was tested. WT animals received only UNx or sham surgery without any further intervention and were sacrificed 2 weeks post UNx. WT animals with UNx do not develop albuminuria and renal damage within several weeks post UNx (Staffel et al., 2017), why this model was not further monitored for progression of kidney disease. Experimental group 2, UNx/DOCA/salt in Podo GC-A KO (Tables 1, 2): The cardiac natriuretic peptides ANP and BNP activate the same receptor, the membrane-bound guanylyl cyclase-A (GC-A). GC-A is expressed in various cell types in the body and is critically involved in the regulation of the blood pressure and the extracellular volume. We have recently shown that podocytes have a strikingly high expression of GC-A (Staffel et al., 2017). Podocyte-specific deletion of GC-A (Podo GC-A KO) does not alter blood pressure, plasma volume and plasma renin concentration under control conditions (Tables 1, 2). Treatment of Podo GC-A KO with UNx/DOCA/salt not only resulted in arterial hypertension, hypervolemia and massive suppression of plasma renin concentration (Table 2, data from Staffel et al., 2017), but also in massive glomerular damage and albuminuria (Staffel et al., 2017). Since glomerular damage induced by UNx/DOCA/salt was markedly aggravated in Podo GC-A KO compared with Podo GC-A WT despite of similar effects on blood pressure, plasma volume and plasma renin concentration in both genotypes, podocytes of Podo GC-A KO mice appear to be more susceptible to these stressors. Therefore, this genetic model was used to induce severe glomerular damage. Podo GC-A KO mice received UNx and a 60 days-release pellet (150 mg) of the mineralocorticoid deoxycorticosterone acetate (DOCA) was implanted subcutaneously (Innovative Research of America, Sarasota, FL, United States). From day seven post UNx (regeneration from surgery), the mice were fed a high-salt diet (4% NaCl, ssniff-Spezialdiäten GmbH, Soest, Germany) until the end of the experiment. Due to severe kidney damage with marked albuminuria at 4 weeks post UNx the experiment was terminated prematurely. Experimental group 3, UNx/DOCA/salt in WT mice (Tables 1, 2): In order to generate a less aggressive kidney injury model with moderate albuminuria the described UNx/DOCA/salt model (group 2) was applied to WT C57BL/6J animals with only minor changes. As shown in Table 2 UNx/DOCA/salt in wildtype mice induces arterial hypertension, hypervolemia and a massive suppression in plasma renin concentration. A 21 days-release DOCA-pellet (50 mg) was implanted in WT animals at day seven post UNx. The high-salt diet started on the same day and mice were sacrificed after 4 weeks.

Experimental group 4, UNx/high salt in Podo GC-A KO (Tables 1, 2): As a model for mild glomerular damage/albuminuria uninephrectomized Podo GC-A KO mice were challenged with a high-salt diet (4% NaCl, ssniff-Spezialdiäten GmbH, Soest, Germany) for 6 weeks, but without concomitant DOCA treatment. As shown in Table 2, data of previous experiments showed that this model does not develop arterial hypertension or an increase in plasma volume, while plasma renin concentration was significantly suppressed.

Experimental group 5, UNx in GC-A KO (Tables 1, 2): In order to avoid the use of high-salt diet and DOCA, mice with global deletion of the receptor for ANP and BNP (GC-A KO) were chosen. Even under control conditions, GC-A KO mice are hypertensive (+17 mm Hg), hypervolemic (plasma volume + 0.3% of bodyweight) and have reduced plasma renin concentration (Lopez et al., 1995; Demerath et al., 2014). The effects of UNx on these parameters have not been determined in GC-A KO yet. To induce glomerular hyperfiltration, GC-A KO mice received UNx only and renal function was monitored for 6 weeks.

EMPA treatment in all groups started 3 days before UNx at a dose of 30 mg/kg/day (Carbosynth Limited, Compton, United Kingdom) administered via drinking water. The EMPA concentration was adjusted to the drinking behavior of each experimental group, in particular for animals with increased water uptake during DOCA and high-salt diet. Fresh drinking water was prepared every 3–4 days. As proof of efficacy, urinary glucose levels were monitored on a regular base using a colorimetric glucose quantification kit (Cayman Chemical, Ann Arbor, MI, United States).

GFR was assessed in conscious mice at prespecified time points post UNx using the transdermal GFR technology (MediBeacon Inc., Mannheim, Germany). Briefly, the back of the mice was shaved under 3% isoflurane anaesthesia and the GFR monitor was installed on the shaved skin. Background fluorescence signal was measured for 5 min before Fluorescein isothiocyanate (FITC)-labeled sinistrin (15 mg/100 g bodyweight) was injected intravenously. After injection, animals were placed in individual cages and the fluorescence signal in the skin was measured for 90 min in conscious and freely moving mice. The recorded clearance of FITC-sinistrin was used to calculate the excretion half-life t_1/2 of FITC-sinistrin according to the manufacturer’s instructions (MPD Studio2 software; remove artifact filter; 3 compartment model with linear correction). FITC-sinstrin t_1/2 was converted to GFR using a mouse-specific conversion factor (Schreiber et al., 2012): GFR [µl/min/100 g b. w.] = 14616, 8 [µl/100 g b. w.]/t_1/2 [min]. Since we compare single kidney GFR in our analysis, GFR values of mice without UNx were divided by two.

Throughout the experiment, spot urine was collected each week to determine the albumin/creatinine ratio in the urine as a marker of progressing kidney damage. Albumin was quantified using a specific mouse albumin ELISA according to the manufacturer’s instructions (Dunn Labortechnik, Asbach, Germany). A colorimetric assay (improved Jaffe method) was used to determine creatinine (BioAssay Systems, Hayward, CA, United States).

For long-term assessment of drinking behavior and urine excretion, mice were kept in metabolic cages with free access to food and water for a period of 72 h. Urine was collected and drinking water was replaced every 24 h. The health status of animals was closely monitored by visual inspection and daily body weight measurements.

At the end of the experiment, blood samples were taken from the facial vein of conscious mice and mice were killed by cervical dislocation thereafter. Determination of plasma renin concentration (PRC) in these plasma samples was based on the generation of angiotensin I after the addition of plasma from bilaterally nephrectomized male rats as excess renin substrate. The generated angiotensin I [ng/ml*h⁻¹] was determined by ELISA [Angiotensin I (PRA) ELISA; IBL International, Germany].

At the end of the experiment, mice were killed by cervical dislocation and prepared for perfusion via the abdominal aorta. Kidneys were rinsed with a 0.9% NaCl solution, perfusion-fixed with 4% paraformaldehyde (3 min; 100 mmHg constant pressure) and stored in 70% methanol for paraffin-embedding. After sectioning (5 µm), kidney sections were stained for collagenous, fibrotic tissue using a standard Masson-Goldner-Trichrome staining protocol (Sigma-Aldrich, HT15-1KT, Taufkirchen, Germany). Stained kidney sections were examined by light microscopy (Axio Observer 7, Carl Zeiss, Jena, Germany). For unbiased quantification of the fibrotic tissue areas (blue staining) we used the Zeiss ZEN Intellesis Image Segmentation software (Carl Zeiss, Jena, Germany). In brief, the software uses a machine-learning algorithm for automated identification of stained areas based on an individual training for each segmentation class. We defined three different classes recognizing background, healthy kidney or fibrotic tissue areas and used the fibrosis/kidney ratio as read out parameter for our analysis.

Blood was collected from WT animals after a 2 weeks EMPA treatment (30 mg/kg/day) via submandibular puncture of the facial vein. Total and free EMPA concentration in plasma were determined by HPLC-MS/MS. Further details regarding sample preparation and HPLC-MS/MS measurement are given in the supplementary data.

Plasma volume was measured in conscious mice by single intravenous injection of 50 µl FITC-labeled bovine serum albumin (BSA; Sigma-Aldrich, Taufkirchen, Germany) under mild isoflurane anesthesia. A 10 µl blood sample was collected from the tail vein before and 20 min post FITC-BSA injection. Plasma fluorescence was measured on a Nanodrop 3,000 (Peqlab Biotechnologie GmbH, Erlangen, Germany) and FITC-BSA concentration was calculated according to an appropriate standard curve. We used the equation c (FITC-BSA stock)* v (FITC-BSA stock) = c (FITC-BSA plasma)* v (plasma) to determine the plasma volume of mice.

Renal mRNA was extracted from 4 paraffin-embedded kidney sections (15 µm) per sample using the Quick-RNA FFPE Miniprep-Kit (Zymo Research Europe GmbH, Freiburg, Germany). Cardiac mRNA was extracted from left ventricles of excised hearts using the RNeasy Mini Kit (Qiagen, Hilden, Germany) according to the manufacturer’s recommendation. cDNA was transcribed from 1 µg RNA using random primers, PCR nucleotide mix, RNasin ribonuclease inhibitor, reverse transcriptase and reverse transcriptase 5x reaction buffer (Promega GmbH, Walldorf, Germany) for 1 h at 37°C. mRNA abundance of renal (fibronectin, α-smooth muscle actin, collagen 1a1) and cardiac target genes (BNP, α-smooth muscle actin, TGF-β, collagen 1a1, collagen 3a1) were measured using the SYBR Green PCR (Roche Diagnostics Deutschland GmbH, Mannheim, Germany) or TaqMan Gene Expression (Thermo Fisher Scientific GmbH, Dreieich, Germany) detection method. For relative mRNA expression analysis according to the comparative threshold cycle (Ct) relative quantification analysis method (Livak and Schmittgen, 2001) Rpl-32 (kidney) and Gapdh (heart) were used as housekeeper genes. Relative target gene mRNA expression are shown as percentage of the expression level in H₂O animals of the respective genotype (100%). Primer sequences and the corresponding detection methods can be found in Supplementary Tables S1, S2.

Data are shown as mean ± SEM. For single-group comparisons, an unpaired Student’s t test was used to calculate the level of significance. Accordingly, for multi-group comparisons at different time points, a two-way ANOVA followed by a Bonferroni post hoc test was used. All statistical analyses were performed using the GraphPad Prism software. Differences between groups were considered significant at a p < 0.05.

To verify that EMPA administration via the drinking water results in therapeutic plasma levels, EMPA plasma concentration of WT animals was determined by HPLC-MS/MS. After 2 weeks of EMPA treatment, total plasma EMPA concentration was 350.8 nM, which corresponds to the plasma concentration determined in clinical trials (Scheen, 2014) (Figure 1A). Protein precipitation revealed that 87% of EMPA was bound to plasma proteins and the concentration of unbound EMPA was 46 nM. In line with the expected inhibition of SGLT2, urinary glucose/creatinine ratio was markedly increased in EMPA treated WT animals (714.31 ± 126.15 mg/mg; Figure 1B) compared with H₂O control animals (0.1 ± 0.011 mg/mg; Figure 1B).

Here, we investigated whether EMPA affects GFR of control (sham-operated) and hyperfiltrating (UNx) kidneys of non-diabetic WT animals. To induce hyperfiltration, the left kidney of WT animals was excised and single kidney GFR was determined 2 weeks later. EMPA had no effect on GFR in sham-operated animals (Figure 2A). UNx induced a marked increase in single kidney GFR compared to sham-operated control animals. This effect was blunted in EMPA treated animals, in which hyperfiltration in response to UNx was significantly reduced compared with H₂O treated UNx animals (65 vs. 41% GFR increase, Figure 2A). UNx not only results in an increase in GFR but also in kidney hypertrophy. This parallelism of GFR and kidney weight was abolished in EMPA-treated mice, because EMPA reduced GFR in UNx mice, but kidney weight tended to be even slightly higher in EMPA than in H₂0 mice (p = 0.06; Figure 2B). Plasma renin concentration was not altered by EMPA in sham or UNx mice (Figure 2C). Histological examination of renal slices did not reveal any signs of enhanced fibrosis in response to UNx and EMPA did not have any effect on kidney histology and fibrosis (Figure 2D).

GFR and progression of renal damage were not further monitored in these mice, as previous studies proved that WT animals with UNx but without any further intervention do not even develop albuminuria and renal damage within several weeks post UNx (Staffel et al., 2017). Therefore, this model is unsuitable for investigations of nephroprotective mechanisms of EMPA and UNx was combined with additional treatments and genetic alterations to enhance glomerular damage.

Podo GC-A KO mice with UNx/DOCA/salt treatment were used to study the nephroprotective potential of EMPA in a model system for severe glomerular damage and progressive renal failure. After 2 weeks of UNx/DOCA/salt treatment, single kidney GFR increased in Podo GC-A KO mice to +172% in the H₂O group and +158% in the EMPA group compared to baseline levels (Figure 3A). The slight trend toward reduced hyperfiltration in EMPA treated animals did not reach statistical significance and did not persist until week 4. The induced damage of the glomerular filter resulted in a massive increase in albuminuria (∼150x compared to baseline) to the same extent in H₂O and EMPA treated animals (Figure 3B). Due to incipient symptoms of nephrotic syndrome in individual animals of both groups, the experiment was terminated prematurely after 4 weeks. EMPA induced glucosuria (Figure 3C) to a similar extent as in untreated wildtype mice (Figure 1B). Preexisting diuresis and increased water intake, caused by DOCA/salt treatment, were numerically higher in EMPA than in H₂O mice (+5.4 and +5.7 ml respectively). However, this numerical difference did not reach statistical significance (Figure 3D). Plasma renin concentration was massively suppressed compared with untreated wildtype mice (shown in Figure 2) and EMPA did not significantly alter plasma renin concentration (Figure 3E). Histological examination of kidney sections revealed no differences in glomerular or tubular morphology between H₂O and EMPA treated animals. Both groups showed the characteristic model-dependent collagen casts and slight glomerular hypertrophy (Figure 3F) (Staffel et al., 2017). Increased collagenous fibrotic structures, visualized by Masson-Goldner staining, were found in glomeruli of all animals (compared to healthy kidneys; for comparison see Figure 2), and automated quantification of fibrotic tissue in whole kidney overview images confirmed no significant effect of EMPA treatment on renal fibrosis (Figure 3G). In line with the histological examination, qPCR did not reveal significant differences in renal mRNA expression of the fibrosis genes fibronectin, alpha-SMA and collagen1a1 between the H₂O and EMPA groups (Figure 3H). Likewise, marker genes for cardiac fibrosis were not altered by EMPA treatment (Supplementary Figure S1). In response to the removal of the left kidney and subsequent hyperfiltration, all animals developed pronounced hypertrophy of the remaining right kidney irrespective of the treatment (Figure 3I). Body weight did not change in the H₂0 group but decreased significantly in the EMPA group during 4 weeks of DOCA/salt treatment (−9.8% vs. baseline, Figure 3J). Overall, the progression of renal damage caused by a combination of a genetic model with increased glomerular vulnerability (Podo GC-A KO) and the UNx/DOCA/salt model was not ameliorated by EMPA treatment.

Since a potential nephroprotective effect of EMPA might be masked in an aggressive kidney injury model, we repeated the UNx/DOCA/salt model in a cohort of WT animals. Within the first 2 weeks post UNx, single kidney GFR rose to 135% in control vs. 149% in EMPA treated animals compared to baseline GFR levels (Figure 4A). At 4 weeks, hyperfiltration had further increased to 157 and 161% (Figure 4A), respectively, with no significant differences between groups. Animals of both groups developed a persistent moderate albuminuria throughout the course of the experiment (Figure 4B). Again, inhibition of SGLT2 by EMPA caused a massive loss of glucose in the urine (Figure 4C). DOCA-high salt diet induced marked diuresis and water uptake (Figure 4D). There was a slight trend toward increased water intake and urine volume in the EMPA group, without reaching statistical significance (Figure 4D). Plasma renin concentration was markedly reduced compared to untreated WT mice (shown in Figure 2) and was not affected by EMPA treatment (Figure 4E). In line with glomerular damage, both groups developed renal fibrosis to the same extent, visualized by histological staining of collagen deposits in glomeruli and tubulointerstitial areas (Figures 4F,G). Similarly, renal mRNA abundance of fibronectin, alpha-SMA and collagen1a1 was not different between H₂O and EMPA treated mice (Figure 4H) and cardiac fibrosis genes were unaffected by EMPA treatment (Supplementary Figure S1). While the extent of hyperfiltration was similar in both groups (Figure 4A), the increase in kidney weight was significantly greater in EMPA treated animals (kidney weight was related to body weight at baseline, Figure 4I). Body weight remained stable in the H₂O group, while it decreased significantly in the EMPA group (Figure 4J). In conclusion, no protective effect of EMPA on hyperfiltration, albuminuria or renal fibrosis was detected in this model.

In a next approach, we examined the effect of EMPA on a rather mild kidney damage model. Therefore, Podo GCA KO mice received UNx in combination with a high salt diet, but without concomitant DOCA treatment. Expecting a slow progression of renal damage, the experimental protocol was extended from 4 to 6 weeks. At 6 weeks post UNx, single kidney GFR was more than doubled in both groups (212 vs. 210% compared to basal levels) with no effect of EMPA on either healthy (basal; Figure 5A) or hyperfiltrating kidneys (6 weeks; Figure 5A). Glomerular damage was rather mild, as indicated by low albumin levels in the urine of control animals (Figure 5B). Urinary albumin excretion was numerically slightly increased in EMPA-treated animals, but the difference did not reach statistical significance. EMPA induced glucosuria to a similar extent as in the other models (Figure 5C). Water intake and diuresis were markedly lower compared with the DOCA/salt treated mice shown in Figures 3, 4 and EMPA induced significant increases in both parameters (Figure 5D). Plasma renin concentration was not altered by EMPA (Figure 5E). Histological (Figure 5F) and quantitative (Figure 5G) examination of fibrotic collagen deposits in kidney sections revealed no difference between groups and results were similar to those obtained for healthy kidneys (for comparison see Figure 2). Fibronectin, alpha-SMA and collagen1a1 mRNA levels tended to be higher in EMPA treated mice, without statistically significant difference (Figure 5H). Matching the strong increase in single kidney GFR, all animals, independent of treatment, showed a massive renal hypertrophy (>2x increase in kidney weight compared to control kidneys) after UNx. This effect was pronounced in EMPA treated mice, since kidney weight (related to body weight at baseline) was significantly higher than in H₂0 treated mice (Figure 5I). As in the models before (Figures 3–5) the body weights of EMPA treated mice decreased in the course of the treatment phase while it remained constant in the H₂O group (Figure 5J).

As a final approach, GC-A knockout mice, which are hypervolemic and hypertensive without DOCA/salt treatment (Tables 1, 2) were subjected to UNx. UNx in GC-A KO mice induced a constant hyperfiltration in the remaining kidney that was not affected by EMPA treatment (H₂O: 170–180% vs. EMPA: 165–170% compared to basal levels) (Figure 6A). Albuminuria increased in both groups within the first 4 weeks post UNx until it reached a stable level (Figure 6B). A strong trend towards higher urinary albumin/creatinine ratio in EMPA treated mice compared with H₂0 control animals was detected, but the observed differences did not reach statistical significance (Figure 6B). EMPA induced urinary glucose loss (Figure 6C), marked diuresis (+4.21 ml/24 h; Figure 6D) and increased water uptake (Figure 6D). EMPA did not affect the plasma volume (H₂O: 1.20 ± 0.22 ml; EMPA: 1.19 ± 0.26 ml, data not shown), which has been shown to be elevated in GC-A KO animals under control conditions previously (Sabrane et al., 2005; Demerath et al., 2014). As in the other models, EMPA did not affect plasma renin concentration (Figure 6E). Visual inspection of kidney sections with Masson-Goldner staining revealed no clear signs for renal fibrosis in H₂O and EMPA-treated GC-A KO animals (Figure 6F). These results were confirmed by an unbiased automated quantification analysis (Figure 6G) and by qPCR (Figure 6H). Moreover, EMPA did not affect cardiac fibrosis markers (Supplementary Figure S1). Again, we observed renal hypertrophy after UNx, but in this model, EMPA had no significant effects on elevated kidney weights (Figure 6I) or body weight (Figure 6J).

In summary, SGLT2 inhibition by EMPA reduced single kidney GFR in hyperfiltrating kidneys of WT mice independent of the glycemic state of the animal. Moreover, EMPA treatment increased glucosuria in all models with varying degrees of renal dysfunction. However, prominent markers of kidney damage such as albuminuria and renal fibrosis were not ameliorated by EMPA treatment, probably due to the lack of an effect of EMPA on single kidney GFR under challenging conditions like high salt intake, hypervolemia or arterial hypertension.