NRF2 regulates the glutamine transporter Slc38a3 (SNAT3) in kidney in response to metabolic acidosis

Expression of the glutamine transporter SNAT3 increases in kidney during metabolic acidosis, suggesting a role during ammoniagenesis. Microarray analysis of Nrf2 knock-out (KO) mouse kidney identified Snat3 as the most significantly down-regulated transcript compared to wild-type (WT). We hypothesized that in the absence of NRF2 the kidney would be unable to induce SNAT3 under conditions of metabolic acidosis and therefore reduce the availability of glutamine for ammoniagenesis. Metabolic acidosis was induced for 7 days in WT and Nrf2 KO mice. Nrf2 KO mice failed to induce Snat3 mRNA and protein expression during metabolic acidosis. However, there were no differences in blood pH, bicarbonate, pCO2, chloride and calcium or urinary pH, ammonium and phosphate levels. Normal induction of ammoniagenic enzymes was observed whereas several amino acid transporters showed differential regulation. Moreover, Nrf2 KO mice during acidosis showed increased expression of renal markers of oxidative stress and injury and NRF2 activity was increased during metabolic acidosis in WT kidney. We conclude that NRF2 is required to adapt the levels of SNAT3 in response to metabolic acidosis. In the absence of NRF2 and SNAT3, the kidney does not have any major acid handling defect; however, increased oxidative stress and renal injury may occur.


Results
NRF2 regulates SNAT3 in the kidney upon metabolic acidosis. In agreement with our previous work 22 , we confirmed that kidneys from Nrf2 KO mice express depleted levels of Snat3 mRNA and significantly lower protein compared to WT kidneys (Fig. 1a,b and Supplemental Fig. S1). However, deletion of Nrf2 did not affect Snat3 mRNA levels in liver and brain, suggesting that the Nrf2-dependent regulation was kidney-specific ( Fig. 1c,d). Interestingly, Nrf2 KO mice showed a trend to induce SNAT3 mRNA; however, they were unable to induce SNAT3 significantly in the kidney upon metabolic acidosis (Fig. 1a,b). Under baseline homeostatic conditions, SNAT3 is mostly expressed in the basolateral membrane of the S3 segment of the proximal tubule and extends to the S2 segment during metabolic acidosis [11][12][13] . To determine whether NRF2 is required for SNAT3 induction in the proximal tubule upon acid loading and if primary proximal convoluted tubular cells (PCT) could be utilized as a model to investigate the biological mechanisms of action of the SNAT3 transporter, we depleted Nrf2 in PCT with targeted siRNA molecules and exposed the cells to acidic media at pH 6.5 or control media at pH 7.4. siRNA depletion of Nrf2 caused a decrease in the mRNA and protein expression levels of Nrf2 and of mRNA levels of the NRF2 target genes Nqo1, Gstm1, Gsta3, Gclc and Gclm at both pH 7.4 and pH 6.5 (Supplemental Fig. S2). PCT transfected with scrambled control siRNA showed a significant increase in Snat3 mRNA expression following exposure to acidic media; whereas depletion of Nrf2 abrogated the induction of Snat3 mRNA at pH 6.5 (Fig. 1e). However, unlike the in vivo study, we did not detect any SNAT3 protein expression under any of the conditions tested and therefore cannot utilize this model for further mechanistic testing (Supplemental Fig. S2). Also, although no morphological changes were observed upon incubating cells at lower pH, we cannot fully exclude the contribution of toxicity at the non-physiological pH of 6.5.
Nrf2 KO mice do not exhibit any major acid handling defects. Since SNAT3 has been considered a critical transporter in the proximal tubule to supply glutamine for ammoniagenesis, we next tested blood and urine acid-base parameters in WT and Nrf2 KO mice under baseline conditions and after a dietary acid-loading. At baseline, Nrf2 KO mice exhibited no major defect in acid handling compared to WT mice. However, plasma pH and bicarbonate showed a trend to be lower while plasma chloride tended to be elevated in Nrf2 KO mice compared to WT (Table 1). Urinary pH was significantly more acidic in Nrf2 KO mice with higher urinary ammonium as well as phosphate, titratable acidity and net acid excretion compared to the control group (Table 2,  Supplementary Table S1). During the acute acid load, Nrf2 KO mice were able to decrease their urinary pH and increased their urinary ammonium and phosphate excretion as much as the control mice (Table 2). Finally, after 7 days of dietary HCl load, blood pH, bicarbonate and pCO 2 remained lower than under control diet in both groups of mice while blood chloride was increased (Table 1). Meanwhile, Nrf2 KO mice could still acidify their urinary pH and significantly increased their urinary ammonium and net acid excretions as much as WT mice (Table 2). Phosphate as well as titratable acid excretions were lower than at baseline after 7 days HCl load in both genotypes; however, Nrf2 KO mice excreted more phosphate and titratable acids than WT mice when normalized for creatinine. However, when normalized for 24 h urine volume, the difference for titratable acids and phosphate disappeared (Supplementary Table S1). Since urine volumes were small, sampling may not have been complete and quantitative in all animals. On the other hand, normalization to creatinine may also induce a bias as creatinine is SCIeNTIfIC RepoRts | (2018) 8:5629 | DOI:10.1038/s41598-018-24000-2 actively secreted into urine and the role of Nrf2 in this process unknown. Nevertheless, the reduction in SNAT3 induction during metabolic acidosis in Nrf2 KO mice was not associated with a failure to adapt to an acid-load. NRF2 does not affect the expression of ammoniagenic and gluconeogenic enzymes induced by metabolic acidosis. In addition to SNAT3, the mRNA of enzymes involved in ammoniagenesis and Figure 1. NRF2 regulates basal and metabolic acidosis induced SNAT3 levels in the kidney. WT and Nrf2 KO mice were fed a normal diet or a HCl containing diet for 7 days. (a) qPCR analysis of Snat3 mRNA expression in the kidney. Snat3 mRNA levels were normalized to Ppia. (b) Immunoblot detection of SNAT3 using 20 µg total kidney membrane preparation. β-TUBULIN was used as a house-keeping control. (c,d) qPCR analysis of Snat3 mRNA expression in liver and brain. (e) Primary proximal convoluted tubular cells (PCT) from WT kidneys were isolated, followed by depletion of Nrf2 with targeted siRNA molecules and exposure of the cells to normal or acidic media (pH 7.4 or 6.5) for 24 h. qPCR analysis of Snat3 mRNA expression in primary PCT. Data represent mean ± S.D. of n = 4-6 animals per group (a-d) or n = 4 independent PCT preparations (e). Statistical analysis for qPCR was performed with a one-way analysis of variance (with Tukey's post test); ***P ≤ 0.001; ****P ≤ 0.0001. Statistical analysis for immunoblotting was performed with Student's t-test; a = 0.05. . Data represent mean ± S.D. of n = 6 animals per group. Statistical analysis was performed with a one-way analysis of variance (with Tukey's post test); *P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.001 for differences between genotypes for the same treatment and # P ≤ 0.05; ## P ≤ 0.01; ### P ≤ 0.001 for differences between untreated and treated mice for the same genotype. subsequent gluconeogenesis, GLS and PEPCK, respectively, were also up-regulated in response to metabolic acidosis ( Fig. 2a-d). However, no differences were observed in mRNA and protein expression of Gls and Pepck in Nrf2 KO mice upon metabolic acidosis compared to WT control ( Fig. 2a-d).
Nrf2 KO mice exhibit altered amino acid profiles and amino acid transporter expression upon metabolic acidosis. Although we found that NRF2 was indispensable for SNAT3 induction upon metabolic acidosis both in vivo and in vitro, we did not observe any major defect in ammonium excretion in the kidney. Since SNAT3 has been postulated to be the major upregulated basolateral glutamine transporter in the kidney during chronic metabolic acidosis, we explored possible mechanisms underlying the ability of the kidney to compensate for its loss in Nrf2 KO mice. We first checked for differences in the amino acid profile in plasma and kidney of Nrf2 KO compared to WT upon metabolic acidosis. The profile in plasma and kidney of Nrf2 KO mice differed for several amino acids from WT mice (Supplementary Tables S2 and S3). From the 30 amino acids and derivatives measured in plasma glutamine, glutamate, histidine, phenylalanine, 3-methylhistidine and GABA were significantly increased in the Nrf2 KO mice when compared to WT mice. Glutamine and histidine are substrates of SNAT3, and glutamate and GABA are involved in glutamine metabolism, which could explain the altered profile observed in the plasma of Nrf2 KO mice. In kidney, from the 33 amino acids and derivatives measured histidine, glycine and beta-alanine were decreased and phenylalanine, valine, anserine and carnosine were significantly increased in the Nrf2 KO mice when compared to WT mice. No changes between genotypes could be observed in the glutamine, glutamate and GABA concentrations. Next, we examined a panel of genes encoding transporters which handle glutamine and other amino acids that can fuel ammoniagenesis (Snat 1,2,4, and 7 (Slc38a1,2,4, and 7), b 0,+ at (Slc7a9), y + Lat1 (Slc7a7), 4F2hc (Slc3a2), Tat1 (Slc16a10), Lat2 (Slc7a8), and B 0 at1 (Slc6a19)), transport ammonium (Rhcg, Nkcc2, Nhe4), protons (Nhe3) (Slc9a3 and 4), and calcium (Ncx1, Trpv5, Calb1 and Pmca4). Snat1 mRNA was significantly upregulated in Nrf2 KO kidney upon metabolic acidosis (Fig. 3a). None of the other members of the Snat (Slc38) family tested were upregulated at mRNA level by metabolic acidosis (Fig. 3b-d). The mRNA expression of the amino acid transporters b 0,+ at and y + lat1 were significantly suppressed upon metabolic acidosis in WT kidney, in agreement with our previous work 12 , whereas the expression remained unchanged in Nrf2 KO kidneys ( Supplementary Fig. S3). The same pattern was observed for 4F2hc ( Fig. 4a). No change in mRNA expression was observed for Lat2, Tat1, and B 0 at1. (Fig. 4c,e,g). At protein level, no difference was observed between WT and Nrf2 KO at baseline and in acid loaded conditions for 4F2hc, LAT2, TAT1 and B 0 AT1 (Fig. 4b,d,f,h). Additionally, no change in mRNA levels was observed for the ammonia transporter (Rhcg), the ion and ammonium transporter Nkcc2, and the Na + /H + exchanger Nhe3 ( Supplementary Fig. S3). However, the mRNA levels of the Na + /H + exchanger Nhe4 were significantly up-regulated in kidneys from Nrf2 KO mice under metabolic acidosis ( Supplementary Fig. S3). Furthermore, no change was observed in the mRNA expression of proteins involved in active renal calcium handling, namely Ncx1, Trpv5, Calb1, and Pmca4A ( Supplementary Fig. S4).
Metabolic acidosis induces NRF2 activity in WT kidney. In order to determine whether metabolic acidosis could induce NRF2 activity we measured a panel of its target transcripts. Accordingly, the mRNA levels of Nqo1, Glutathione S-Transferase Mu 1 (Gstm1), Glutathione S-Transferase Alpha 3 (Gsta3), Cytochrome P450 Family 2 Subfamily A Member 5 (Cyp2a5), Dihydropyrimidinase (Dpys), Glutamate-Cysteine Ligase Modifier Subunit (Gclm), and Gclc were significantly upregulated upon metabolic acidosis and glutathione (GSH) levels were maintained at levels similar to those in WT animals on a normal diet (Figs 5a-e and 6a-c). Nrf2 mRNA levels remained unchanged suggesting that regulation of NRF2 activity by metabolic acidosis occurs on a posttranscriptional level (Fig. 5f). In agreement with our previous study, we found a significant down regulation of the Nrf2 target genes Nqo1, Gstm1, Gsta3, Cyp2a5, Dpys, Gclm, and Gclc in Nrf2 KO kidneys (Figs 5a-e and 6a,b) and a subsequent reduction in the stimulation of their mRNA expression during metabolic acidosis 22 . Also, GSH levels were low in Nrf2 KO kidneys at baseline and remained almost unmeasurable during metabolic acidosis (Fig. 6c). The absence of induction of transcripts involved in the defense against oxidative stress during metabolic acidosis and the very low levels of GSH in Nrf2 KO kidneys were associated with a significant increase in the levels of the proximal tubule specific kidney injury and oxidative stress markers Kim1 (Havcr1) and carbonic anhydrase III(Car3), respectively (Fig. 6d,e) 23,24 .

Discussion
In this study, we have shown for the first time that Nrf2 regulates the induction of the renal glutamine transporter SNAT3 (Slc38a3) upon metabolic acidosis. Further, we confirmed our previous observation that NRF2 also controls basal SNAT3 expression in mouse kidney 22 . Importantly, we found that NRF2 does not control basal Snat3 mRNA expression in the liver and brain, and identified that both organs are insensitive to Snat3 induction upon metabolic acidosis, suggesting a kidney-specific mechanism of control.
With respect to the current knowledge of the transcriptional regulation of Snat3, work by Balkrishna et al. demonstrated that tissue-specific Snat3 expression is predominantly controlled by various epigenetic factors 25 . Importantly, they also identified two SP1 binding sites, one important for basal Snat3 promoter transactivation and a second pH sensitive SP1 binding site to upregulate Snat3 under metabolic acidosis 25 . Our data now indicate that NRF2 is also critical for the transcriptional control of Snat3 in mouse kidney. Evidence is increasing that NRF2 forms complexes with the transcription factor SP1, which directly affect the regulation of SP1 and NRF2 target genes 26,27 . Furthermore, it has been reported that NRF2 and the SP family proteins, including SP1, synergistically enhance the expression of the ion transporters, colonic H/K-ATPase and kNBC1, as shown in transfected HEK-293T and CV-1 kidney cells under potassium-depleted conditions 28 . This effect was abolished upon transfection with dominant negative NRF2 28 . Importantly, chromatin immunoprecipitation showed that NRF2 and SP1 bound to the promoter of kNBC1, yet only SP1 bound to the promoter of colonic H/K-ATPase, suggesting the formation of a NRF2-SP1 complex in certain situations 28 . The hypothesis that Snat3 transcription may be regulated by a NRF2-SP1 complex warrants further investigation.
The SLC38 family of amino acid transporters, to which SNAT3 (SLC38A3) belongs, is split into two major subgroups: system A and system N 14 . SNAT3 belongs to the system N transporters together with SNAT5 and the more recently characterized SNAT7 29 . Each transporter has specific activities for the amino acids glutamine, histidine and asparagine. During metabolic acidosis, both glutamine and histidine levels in blood were increased in Nrf2 KO mice as compared to WT mice, whereas in kidney tissue of Nrf2 KO mice glutamine remained the same and histidine levels were significantly reduced. The changes in plasma might be a direct effect of reduced SNAT3 expression but suggest that glutamine levels in the kidney are maintained at normal levels by other mechanisms in the absence of SNAT3. We observed no difference in the mRNA levels of the N-type transporters Snat5 and Snat7 in the kidneys of Nrf2 KO mice, therefore it is unlikely that these transporters are involved in a compensatory mechanism. The characterized system A transporters; SNAT1, SNAT2 and SNAT4 can carry glutamine but prefer small neutral amino acids 14 . Interestingly, our data show that Snat1 mRNA is significantly upregulated in Nrf2 KO mice upon metabolic acidosis, whilst in the whole animal Snat3 KO mouse, SNAT1 protein was profoundly elevated in the brain 7 . Although the fold induction of Snat1 was modest, it may contribute to elevating kidney glutamine levels during metabolic acidosis when SNAT3 is lost. The exact localization of SNAT1 in kidney, however,  is currently unknown. We also observed a significant induction of the sodium/proton exchanger Nhe4 (Slc9a4) in the Nrf2 KO mouse kidney following metabolic acidosis. NHE4 is critical for ammonia transport in the thick ascending limb of the loop of Henle, interstitial ammonia accumulation, and maintenance of systemic pH 30 . Its upregulation in Nrf2 KO mice may serve compensatory functions.
Metabolic acidosis results in the upregulation of enzymes which direct the metabolism of glutamine towards ammoniagensis (GLS and glutamate dehydrogenase 1) and gluconeogenesis (PEPCK). When working efficiently, both pathways combine to produce two NH 3 and two HCO 3 − ions per glutamine molecule 5 . NRF2 has also been shown to influence the direction of the metabolism of glutamine in favor of GSH biosynthesis and the TCA cycle 21 . In addition, NRF2 can inhibit gluconeogenesis, although the mechanism remains to be fully understood 19 . We show that GSH levels are significantly reduced in the Nrf2 KO kidney compared to WT, probably due to the downregulation of GCLC and GCLM, which code for the two subunits of the rate limiting enzyme in glutathione synthesis, GCL 21 . We show that the mRNA levels of Gclc and Gclm are significantly induced and that GSH levels are maintained upon metabolic acidosis in WT kidney, which suggests that glutamine is being utilized for all the pathways mentioned above during metabolic acidosis. This would imply that an active NRF2 system regulates pathways that compete for glutamine in the kidney and therefore reduces the efficiency by which the proximal tubular cells can utilize glutamine for ammoniagenesis and gluconeogenesis. Therefore, it is possible that in the absence of NRF2 in this setting, the demand for glutamine is reduced since it is no longer fluxed into GSH biosynthesis. If this is the case, in the absence of SNAT3, intracellular demands for glutamine may be met by the induction of the SNAT1 transporter or additional mechanisms yet to be identified. This hypothesis has to be tested in future experiments.
The primary role of NRF2 is to coordinate the upregulation of antioxidant and detoxification genes to defend against the detrimental effects of oxidative stress 31 . The importance of this pathway, with respect to the kidney have been demonstrated to protect against a number of nephrotoxic insults in Nrf2 KO animals 32 . Importantly, animal models of chronic kidney disease (CKD) showed reduced levels of NRF2 and its target genes [33][34][35][36] . Although this current investigation focused on SNAT3 and metabolic acidosis, it also gave us the opportunity to examine what happens to the NRF2 system in the kidney during metabolic acidosis. We show for the first time in vivo that the NRF2 target genes Gclm, Gclc, Nqo1, Gstm1, Gsta3, Cyp3a5 and Dyps are significantly induced upon metabolic acidosis and that Nrf2 is needed to maintain GSH levels in the WT kidney under such conditions 22 . Thus, our data suggest that metabolic acidosis in this setting is an oxidative insult and that the kidneys baseline defense needs to be stimulated in a NRF2-dependent manner in order to deal with the perturbation in homeostasis. Indeed there is in vitro evidence that NRF2 is activated in the prostate cancer cell AT-1 following exposure to acidic media; however, the opposite was observed in the MCF-7 breast cancer cell line 37,38 . Additionally, in the Nrf2 KO kidney mRNA levels of Gclm, Gclc, Nqo1, Gstm1, Gsta3, Cyp2a5 and Dyps were significantly depleted, as well as total GSH, in agreement with our previous study 22 . Importantly, upon metabolic acidosis we observed a significant increase in the proximal tubular injury and oxidative stress markers Kim1 24 and carbonic anhydrase type 3 (Car3) 23 specifically in the Nrf2 KO kidneys. Although the kidney does not have any major acid handling defects in this experimental setting after 7 days, this is an indicator that the loss of Nrf2 may be detrimental to the function of the proximal tubule following long term exposure and should be investigated further. The data may also indicate that the role of SNAT3 in ammoniagenesis is less critical than suggested by previous data and that SNAT3 may rather contribute to the anti-oxidative defense by supplying substrates for GSH synthesis. Clearly, this will require further detailed analyses developing cell-specific Snat3 deficient mouse models.
In summary, NRF2 regulates the levels of SNAT3 in response to metabolic acidosis. However, in the absence of SNAT3 induction, the kidney does not show any impairment in the ability to adapt acid-excretion to an acid-load. Compensatory adaption of other transporters in the kidney or metabolic redirection of glutamine away from glutathione synthesis may account for the loss of SNAT3 induction but this is associated with increased vulnerability to oxidative stress.

Methods and Materials
Animals. Male Nrf2 KO mice, age 10-14 weeks old (C57BL/6 background, generation, and genotyping have been described previously 39,40 ), were bred at the University of Liverpool Biomedical Services Unit and transferred to the Institute of Physiology, University of Zürich. Age and sex matched WT C57BL/6 mice were supplied by Janvier, France. All animal experiments were performed according to the Swiss Law of Animal Welfare and approved by the local authority (Veterinäramt Zurich).
For experiments, mice were housed in metabolic cages (Techniplast, Buguggiate, Italy). Mice were given water ad libitum and were fed with a standard powdered laboratory chow mixed with water (50/75 w/v) (Kliba, Augst, Switzerland) to adapt to metabolic cages for 2 days. Two sets of experiments were performed. The first set of 6 WT and 6 Nrf2 KO mice received a standard diet for 2 days, during which 24 h urine samples were collected under light mineral oil in the urine collector to determine daily urinary parameters. After 2 days, before euthanasia, retro-orbital blood samples were taken under light anesthesia with isoflurane. Plasma, kidneys, brains and livers were harvested after euthanasia. The second set of 6 WT and 6 Nrf2 KO mice received a standard diet during the first 24 h urine collection and were then switched to a dietary acid load for 7 days where mice were given a HCl-containing diet with 150 ml of 0.33 M HCl added to 100 g powdered standard food. Food, water intake, and urine excretion were monitored at day 0, 2 and 7 of the HCl diet. On the last day of experiments, before euthanasia, retro-orbital blood samples were taken under light anesthesia with isoflurane. Plasma, kidneys, brains and livers were harvested after euthanasia.
Analytic procedures. Blood pH, pCO 2 , and electrolytes were measured with a pH/blood-gas analyzer (ABL77 Radiometer). Urinary pH was measured with a pH-meter (Metrohm AG, Canada) and creatinine by a modified kinetic Jaffé colorimetric method 41 . Urinary NH 4 + was measured with the Berthelot protocol 42 . Urinary inorganic phosphate (Pi) concentration was determined by the phosphomolybdate method 43 . Urinary titratable acidity was measured according to Jorgensen and Siggaard-Andersen 44,45 . Briefly, CO 2 was eliminated by hydrochloric acid addition. Then, titratable acidity was measured by sodium hydroxide (1 N) titration to pH 7.40 using DL 55 Mettler Toledo ® titrator, ST20 Mettler ® sample changer and Inlab ® Semi micro pH electrode. Urinary pH was measured on fresh urine with a pH-meter. Citric acid was measured using a kit from Boehringer Mannheim/ R-Biopharm.
Amino acid measurements. Quantitative amino acid measurements were performed using targeted LC-MS/MS based on the method described by Harder et al. 46 . Briefly, plasma samples (10 µl) were dissolved in 500 µl ice-cold methanol containing an internal standard mixture of 15 deuterated amino acids. After centrifugation (10 min, 10 °C, x g), samples were dried. Kidney tissue samples were extracted in 30x methanol mass using TissueLyser. Amino acids in plasma and kidney samples were derivatized to their butyl esters as described by Gucciardi et al. 47 . Briefly, a mixture of 95% n-butanol and 5% acetylchloride (v/v) was added to the dried samples. Subsequently, the samples were incubated at 60 °C for 15 minutes at 600 rpm (Eppendorf Thermomixer Comfort; Eppendorf, Hamburg, Germany). The samples were dried and reconstituted in a 200 µl mixture of methanol/ water/formic acid (70/30/0.1% v/v).
The analysis was performed on a triple quadrupole QTRAP 5500 LC-MS/MS system operating in positive ESI mode (AB Sciex, Framingham, MA) equipped with a 1200 series binary pump (Agilent, Santa Clara, CA) and coupled to an HTC pal autosampler (CTC Analytics, Zwingen, Switzerland). Chromatographic separation was achieved using a Zorbax Eclipse XDB-C18 column (length 150 mm, internal diameter 3.0 mm, particle size 3.5 µm; Agilent). Analytes were measured in scheduled multiple reaction monitoring (MRM). For absolute quantification, a 10-point calibration was performed, using a mixture containing all amino acids in the measurement (A9906 amino acid standards, Sigma-Aldrich, Taufkirchen, Germany). Data analysis was done using Analyst 1.5.1 ® software (AB Sciex).
Analysis of mRNA expression. For isolation of total RNA kidney tissue (n = 6 per group), the Direct-Zol RNA MiniPrep kit (Zymo Research, Freiburg, Germany) including on-column DNAse treatment was applied. RNA concentration was assessed using a NanoDrop ND 1000 (Fisher Scientific, Reinach, Switzerland). cDNA was synthesized from 2 μg total RNA using Superscript III reverse transcriptase (Invitrogen, Carlsbad, CA, USA) following the manufacturers protocol. Relative mRNA quantification was performed by real-time quantitative RT-PCR using a rotor-gene 6000 (Corbett Research, Sydney, Australia). Briefly, the cDNA (10 ng) was mixed with gene-specific primers (Microsynth, Balgach, Switzerland) (200 nM) (supplementary Table S4) and KAPA SYBR FAST qPCR reagent (Kapasystems, Boston, MA, USA) (5 μl), in a final volume of 10 μl. Thermal cycler parameters were as follows: 15 min at 95 °C, followed by amplification of cDNA for 40 cycles with melting for 15 s at 94 °C, annealing for 30 s at 56 °C and extension for 30 s at 72 °C. For each sample, three technical replicates were analyzed. Expression was normalized to cyclophylin A (Ppia) control. Fold changes were quantified as 2 −(ΔCt sample-ΔCt control) , as described previously 48 . Immunoblotting. Total kidney homogenates were prepared by mechanical dispersion (Polytron Kinematica PT10/36, Switzerland) in 200 μl ice-cold resuspension buffer (200 mM Mannitol, 80 mM HEPES, 41 mM KOH, pH 7.5) containing EDTA-free proteinase inhibitor mix (Roche, Mannheim, Germany). The total membrane fraction was isolated from total homogenates by first centrifuging it at 2000 rpm (5415 R Centrifuge, Eppendorf, Hamburg, Germany) for 20 min at 4 °C. Next the supernatant was further centrifuged at 41000 rpm (Sorvall RC M120 EX Centrifuge, RP45A-0030 rotor) at 4 °C for 1 hour and the pellet was resuspended in resuspension buffer 7 .
Brush border membrane vesicles were prepared as described previously 49 . Briefly, mouse kidneys were homogenized in 2 ml of homogenization solution (300 mM Mannitol, 5 mM EGTA, 12 mM Tris, pH 7.1 containing EDTA-free proteinase inhibitor mix) followed by addition of 2.8 ml ice-cold distilled water and 58 μl of 1 M MgCl 2 . The mixture was kept for 15 min on ice and centrifuged at 4600 rpm for 15 min at 4 °C (5415 R Centrifuge). The supernatant was centrifuged at 16000 rpm for 30 min at 4 °C (Sorvall RC 5 C Plus centrifuge, SS34 rotor) and the pellet was resuspended in 2 ml membrane buffer (300 mM Mannitol, 20 mM HEPES, 12 mM Tris, pH 7.4, containing EDTA-free proteinase inhibitor mix). The mixture was centrifuged again at 16000 rpm for 30 min at 4 °C and the pellet resuspended in 200 µl membrane buffer.
Whole PCT cell lysates were obtained by lysing cells in 100 µl of ice cold RIPA buffer (Sigma, St. Louis, MO) supplemented with 1x complete mini protease inhibitors (Roche, Mannheim, Germany).