Canagliflozin Inhibits Electrogenic Na⁺ Transport in Mouse Cortical Collecting Duct Cells

Abstract

Sodium-glucose cotransporter 2 inhibitors (SGLT2i) exhibit cardiorenal protective effects that likely involve mechanisms aside from SGLT2 inhibition. Still, many details surrounding these clinically important pleiotropic effects remain unclear. We previously showed that several SGLT2-independent proximal tubular transport functions are inhibited by canagliflozin, but not empagliflozin. Here, we demonstrate a canagliflozin-specific reduction in Sgk1 abundance in both opossum kidney proximal tubule and mouse cortical collecting duct (mCCD_cl1) cells, pointing to a possible underlying mechanism. Given the role of Sgk1 in the distal nephron, we hypothesized that canagliflozin would also inhibit epithelial Na⁺ channel (ENaC)-dependent Na⁺ transport. Canagliflozin inhibited ENaC-dependent Na⁺ transport (amiloride-sensitive short circuit current; I_SC) in mCCD_cl1 cells while empagliflozin had no effect. Selective membrane permeabilization revealed canagliflozin-induced inhibition of both apical conductance through ENaC and basolateral transport via the Na⁺/K⁺ ATPase. These effects were mimicked by the selective Sgk1 inhibitor, GSK650394. Surface labeling studies demonstrated reduced membrane localization of ENaC, but not Na⁺/K⁺ ATPase subunits, consistent with a mechanism involving Sgk1. Canagliflozin reduced I_SC in the presence and absence of rotenone, suggesting inhibition occurs independently of effects on mitochondrial complex I, another known target of canagliflozin. ENaC activity in mouse distal colon was also inhibited by canagliflozin, confirming these effects in native tissue. We identify Na⁺ transport through ENaC and the Na⁺/K⁺ ATPase as novel targets of inhibition by canagliflozin, with Sgk1 as a likely upstream mechanistic component. Canagliflozin-specific effects on transport mediated via this mechanism may contribute to non-class effects of this drug observed clinically.

Graphical Abstract

Graphical Abstract.

Introduction

Na⁺-glucose cotransporter 2 inhibitors (SGLT2i; “gliflozins”) help manage hyperglycemia in patients with type II diabetes,¹ and also benefit renal and cardiovascular health by reducing blood pressure, slowing the decline in glomerular filtration rate, reducing arterial stiffness, and lowering the incidence of adverse cardiovascular events.^2-6 Many of these cardiorenal protective effects manifest in non-diabetic individuals, particularly those with chronic kidney disease or heart failure, implying the involved mechanisms extend beyond glycemic control.^7-9

Some gliflozins exhibit SGLT2-independent effects including inhibition of Na⁺/H⁺ exchanger NHE1 and NHE3 activity,^10-12 inhibition of mitochondrial Complex I,^13,14 and activation of AMP kinase.¹⁵ While the underlying mechanisms and physiological implications of these off-target effects are not well-understood, they may contribute to some of the clinical outcomes not easily explained by SGLT2 inhibition alone. Early reports and interim analyses raised clinical concerns for increased risk of hyperkalemia with canagliflozin,^16-19 although large-scale trials and subsequent meta-analyses have shown no increased risk.^2,4,20-22 Other gliflozins have not raised concerns for hyperkalemia.^23-26 Meta-analyses point to greater reductions in HbA1c and bodyweight,²⁷ and higher risk of volume-related adverse outcomes^28,29 on high dose canagliflozin compared to other gliflozins. These findings suggest that drug-specific, noncanonical mechanisms may be clinically important.

We recently showed that canagliflozin, but not empagliflozin, inhibited NHE3 activity, fluid transport, albumin uptake, and mitochondrial Complex I function in opossum kidney (OK) cells,¹² a cell-culture model of the proximal tubule.³⁰ We therefore predicted that canagliflozin may alter the upstream signaling pathways involved in regulating epithelial transport more broadly. Here, further investigation revealed an acute canagliflozin-specific reduction of serum and glucocorticoid-regulated kinase 1 (Sgk1), which was previously identified as a regulator of NHE3 activity.³¹ Given the role of Sgk1 in regulating distal nephron ion transport,³² we hypothesized that canagliflozin would also inhibit Na⁺ transport in the cortical collecting duct (CCD). To test this, we used mCCD_cl1 cells (a mouse principal cell line)³³ as a model system to study the functional and biochemical effects of canagliflozin in this context. We then used mouse distal colon to study the effect of canagliflozin on Na⁺ transport in a native ENaC-expressing tissue that also lacks SGLTs.

Methods

Study Approval

Protocols for mice conformed with the National Institutes of Health Guide for the Care and Use of Laboratory Animals and were approved by the Institutional Animal Care and Use Committees of the University of Pittsburgh. Mice were housed in a temperature-controlled facility at the University of Pittsburgh on a standard 12-h light/dark cycle. All mice had free access to water and food.

Materials

Canagliflozin (AdooQ), empagliflozin (Cayman Chemical), rotenone (APExBIO), GSK650394 (MedChemExpress), and ouabain (Sigma) were commercially acquired. Adult male (10-12 wk old) C57BL/6 J mice were purchased from The Jackson Laboratory.

Culture of mCCD_cl1 Cells and OK Cells

mCCD_cl1 (passages 28-36) were maintained at 37°C and 5% CO₂ in complete culture media: DMEM/F12 supplemented with 2% decomplemented fetal bovine serum, 100 µg/mL penicillin/streptomycin, 50 nm dexamethasone, 5 µg/mL insulin, 5 µg/mL human apotransferrin, 10 ng/mL mouse epidermal growth factor (EGF), 1 nm triiodothyronine (T3), and 0.06 nM Na⁺ selenite. mCCD_cl1 cells were subcultured onto transwell plates (Corning) and media were changed every other day. Experiments were performed on cell monolayers after 7-8 d in culture and once trans-epithelial resistance reached at least 1 kΩ·cm².

OK cells (OK-P subclone) were cultured in 10 cm dishes in DMEM/F12 (Sigma; RNBL4456) supplemented with 5% FBS and 5 mm Glutamax (GIBCO, No. 35050061) at 37°C in 5% CO₂-95% air. For experiments, cells were dissociated from plates using Accutase (BD Biosciences, No. 561527) and seeded at 4 × 105 on 12 mm Transwell inserts in 12-well dishes (Corning, 3401), with 0.5 mL and 1.5 mL cell culture media in the apical and basolateral chambers, respectively. The next day, dishes were transferred to an orbital platform shaker set to 146 rotations/min in a 37°C CO₂ incubator for 72 h with daily media changes as previously described.¹²

Immunoblotting and Surface Biotinylation Using mCCD_cl1 Cells

For immunoblotting experiments, mCCD_cl1 cells were treated with serum-free media containing either 0.1% DMSO (vehicle control) or various drugs dissolved in DMSO, as described in the figure legends. Cells were lysed in CelLytic (Sigma C3228) buffer containing protease inhibitors (Thermo #A32963). For all experiments except phosphatase treatment, protein phosphatase inhibitors (Sigma #524625) were also added to the lysis buffer. Lysates were centrifuged at 12 000 × g to remove cell debris. Protein concentration was determined via BCA assay and equal amounts of protein (∼20 µg) were prepared for electrophoresis using Laemmli sample buffer containing 5% β mercaptoethanol. Proteins were resolved using Stain-free 4-15% polyacrylamide gels (BioRad), transferred to PVDF membranes, and blocked with 5% milk in tris-buffered saline containing 0.1% Tween-20 (TBST) for 2 h at room temperature. Membranes were then incubated in blocking solution containing one of the following primary antibodies: rabbit anti-Sgk1 (Cell Signaling #12103; 1:1000), rabbit anti-ENaCγ (Stressmarq #SPC-405; 1:1000), or mouse anti-NaKα (Developmental Studies Hybridoma Bank (DSHB) #a5; 1:500). The following day, membranes were washed with TBST and incubated in the appropriate horseradish peroxidase (HRP)-conjugated secondary antibody for 1 h at room temperature. Chemiluminescent images were captured using a ChemiDoc imager (BioRad) and densitometric analysis was performed with FIJI software.

For surface biotinylation experiments, drug/vehicle-treated mCCD_cl1 cells were first washed with ice cold phosphate-buffered saline (PBS), then incubated on both apical and basolateral surfaces with a 137 mm NaCl, 15 mm sodium borate solution (pH 9.0) containing 2 mg/mL EZ-αink Sulfo-NHS-SS-Biotin (Thermo #21331) for 30 min at 4°C with gentle rocking. Excess biotin was then quenched using ice cold FBS-supplemented DMEM for 30 min before final washing with ice cold PBS. Cells were lysed using Goldstein lysis buffer containing protease inhibitors. Five percent of the total lysate was saved as the input control, while the remaining lysates was incubated with constant rotation, overnight at 4°C, in Neutravidin agarose beads (Thermo #29200). The following day, recovered proteins were isolated by washing the beads several times in PBS, followed by elution in 2× Laemmli buffer at 42°C for 15 min. Proteins were resolved via electrophoresis as described above.

Immunoblotting Using OK Cells

Cells were washed once with serum free media prior to treatments as indicated. Cells were rinsed and solubilized in 250 µL detergent solution with protease and phosphatase inhibitors (5 µg/mL leupeptin, 7 µg/mL pepstatin A, 40 µm PMSF, 50 mm NaF, 15 mm sodium pyrophosphate, and Protease Inhibitor Cocktail (Roche, 04693159001). Protein concentration was determined by Lowry assay. Equal amounts of total protein (10-15 µg/sample) were separated on 4-15% SDS-polyacrylamide gels, transferred to PVDF membranes, blocked with 10% milk in Tris-buffered saline with 0.05% Tween-20 (TBST) and then incubated at 4°C overnight with the following primary antibodies: anti-Sgk1 (Cell Signaling #12103; 1:1000) and anti-β-tubulin (Cell Signaling #86298; 1:1000). The membranes were washed 3 times with TBST prior to incubation for 1 h with horseradish peroxidase-conjugated goat anti-mouse IgG (1:10000, Jackson ImmunoResearch, 115-035-166) or peroxidase-conjugated goat anti-Rabbit IgG (1:10000, Jackson ImmunoResearch, 111-035-144) and detected with chemiluminescence. All blots were imaged using the Bio-Rad ChemiDoc Touch Imaging System, and bands were quantitated using FIJI or Bio-Rad Image Lab software.

Measurement of Short-Circuit Currents

mCCD_cl1 cells were seeded onto Snap-well polyester filters (Corning) and grown as monolayers in complete media for 7-10 d prior to experiments and a trans-epithelial resistance of >1 kΩ·cm² was verified. Ten nanomolar aldosterone was added to the culture media 24 h prior to experiments. Cells were then mounted in an Ussing-style chamber (Physiologic Instruments) and bathed in normal Ringer’s solution containing the following, in mm: 135 NaCl, 25 NaHCO₃, 2.4 K₂HPO₄, 0.4 KH₂PO₄, 1.2 CaCl₂, 1.2 MgCl₂, and 10 glucose. The bath solutions were maintained at 37°C and continually gassed with 5% CO₂ balanced with oxygen to maintain a pH of 7.4. After a ∼10 min equilibration period, cell monolayers were voltage clamped to 0 mV and short-circuit current (I_SC) was recorded using a multichannel voltage clamp/amplifier (MCC6, Physiologic Instruments) connected through a multichannel analog/digital interface (Digidata 1440; Molecular Devices) to a computer equipped with pClamp 10.0 software.

For concentration-response analysis, canagliflozin-induced I_SC changes for concentrations of 0.3, 1, 3, 10, 30, and 100 µm were normalized to time-matched controls treated with vehicle only (0.1% DMSO) to account for rundown. Relative inhibition was calculated as the residual amiloride-sensitive I_SC remaining at each drug concentration [I_amil(cana)/I_amil(DMSO)] at 12 min post application. IC₅₀ was determined by fitting the data via 3-parameter nonlinear regression, where “top” and “bottom” parameters were constrained to 100 and >0, respectively.

For experiments with selective membrane permeabilization, an apical-to-basolateral Na⁺ gradient was generated by filling the basolateral chamber with a bath solution containing 25 mm NaHCO₃, with 115 mm Na⁺ replaced with equimolar N-methyl-d-glucamine and titrated to pH 7.4 with HCl. Apical or basolateral membranes were permeabilized by 100 µm nystatin added to one side of the chamber.

For experiments using mouse distal colons, tissues were harvested from mice maintained on a 5% KCl diet for 8 d to stimulate ENaC activity.³⁴ Colons were removed from mice under isoflurane anesthesia (3% balanced with air), flushed with ice cold, oxygenated Ringer’s solution and transferred to a dissecting dish under a stereomicroscope. Mice were then euthanized via bilateral thoracotomy. Mucosa/submucosa preparations were isolated via blunt dissection and mounted on Ussing chamber sliders (0.1 cm²; Physiologic Instruments). After a ∼10 min equilibration period, I_SC recordings were performed as described above for mCCD_cl1 cell experiments and total amiloride-sensitive currents were measured in the presence of 0.1% DMSO (vehicle control) or 25 µm canagliflozin in the apical and basolateral chamber bath solutions.

Results

Canagliflozin Reduces Sgk1 Abundance in Cell Culture Models of Both the Proximal Tubule and the CCD

Following our recent report showing canagliflozin-specific inhibition of NHE3,¹² we investigated effects on Sgk1, a known NHE3 regulator.³¹ OK cells treated with 25 µm canagliflozin for 45 min displayed a marked reduction in Sgk1 protein abundance, in contrast to cells treated with 25 µm empagliflozin (Figure 1A). In this timeframe, canagliflozin caused a preferential loss of the slower migrating (∼50 kDa) band in the Sgk1 doublet, previously shown to correspond to the phosphorylated, active form,^35,36 which we confirmed using λ protein phosphatase (Figure 1C). Similar to other reports, we were unable to detect this band using commercial antibodies directed against phosphorylated Sgk1 (pS422-Sgk1).^36,37

Figure 1.

Canagliflozin reduces Sgk1 abundance in kidney epithelial cells. (A) OK cells were treated for 45 min (bottom) with minimal culture media containing either 0.01% DMSO, 25 µm canagliflozin (cana), or 25 µm empagliflozin (empa). Cell lysates were then immunoblotted for Sgk1 (upper panel) and β-actin (loading control, lower panel). (B) mCCD_cl1 cells were treated for 45 min with minimal culture media containing 0.01% DMSO, 25 µm canagliflozin, or 25 µm empagliflozin, lysed and immunoblotted for Sgk1. The stain-free (s.f.) gel image in the lower panel confirms equal protein loading. (C) OK (left) and mCCD_cl1 (right) cell lysates were treated with λ protein phosphatase (λPP) prior to immunoblotting for Sgk1. λPP treatment caused selective loss of the upper ∼50 kD band. (D-G) Quantification of total Sgk1 and phospho-Sgk1 (50 kD band) abundance in OK and mCCD_cl1 cells treated for 45 min with DMSO, canagliflozin or empagliflozin, as shown in the representative immunoblots. Molecular weight markers are shown to the left. Line and error bars represent mean ± SD (n = 4-6). Data were compared by one-way ANOVA with Tukey’s post-hoc test.

Given the role of Sgk1 in distal tubule transport,^38-40 we tested the effect of canagliflozin and empagliflozin in mCCD_cl1 cells, a cell culture model of mouse principal cells.³³ Similar to our results in OK cells, 25 µm canagliflozin reduced both phosphorylated and total Sgk1 abundance after 45 min, while empagliflozin did not (Figure 1B, F-G), although the effect was more modest than in OK cells. Phosphatase treatment also reduced the ∼50 kDa band in mCCD_cl1 cell lysates, confirming a similar banding pattern between these cell types (Figure 1C; right), although the selective depletion of the phosphorylated Sgk1 band was also not as pronounced as in OK cells.

Canagliflozin Inhibits Both ENaC and Na⁺/K⁺ ATPase Activity in Cultured CCD Cells

We next assessed the functional consequences on transepithelial Na⁺ transport in mCCD_cl1 cells by measuring amiloride-sensitive short-circuit current (I_SC) in an Ussing-style chamber. When we simultaneously added 25 µm canagliflozin to the apical and basolateral chambers, we found that canagliflozin inhibited amiloride-sensitive I_SC. In contrast, the effects of bilateral empagliflozin (25 µm) addition were similar to the vehicle control (Figure 2A, B). The IC₅₀ for canagliflozin on amiloride-sensitive I_SC was 16 ± 2.5 µm under these conditions (Figure 2C).

Figure 2.

Canagliflozin, but not empagliflozin, inhibits short I_SC in mCCD_cl1 cells. (A) I_SC was recorded from mCCD_cl1 cells treated with either 0.1% DMSO (vehicle control), 25 µm canagliflozin (cana), or 25 µm empagliflozin (empa) in an Ussing chamber under voltage clamped conditions. One hundred micromolar amiloride was administered at the end of each recording to inhibit residual ENaC-dependent I_SC. (B) Summary data showing residual amiloride-sensitive I_SC for cells treated with canagliflozin or empagliflozin, normalized to DMSO controls for each experiment. Line and error bars represent mean ± SD (n = 6-9). Data were compared by one-way ANOVA with Tukey’s post-hoc test. (c) Concentration-response data for cells treated with canagliflozin at concentrations ranging from 0.3 to 100 µm. Residual (amiloride-sensitive) current for each concentration was calculated as a percentage of a time-matched vehicle control (∼12 min post treatment). Symbols and error bars represent mean ± SD (n = 5). Data were fitted using a 4-parameter nonlinear regression.

Inhibition of electrogenic Na⁺ transport in mCCD_cl1 cells could result from decreased apical membrane Na⁺ conductance via ENaC, decreased Na⁺/K⁺ ATPase activity in the basolateral membrane, or both. To identify the site of action, we selectively permeabilized the apical or basolateral membranes with 100 µm nystatin and measured I_SC under an apical-to-basolateral Na⁺ gradient, as described previously.⁴¹ Because both ENaC- and Na⁺/K⁺ ATPase-dependent transport are electrogenic, their activities can be measured directly as amiloride- or ouabain-sensitive I_SC, respectively. Amiloride-sensitive I_SC was smaller after basolateral nystatin treatment in canagliflozin-treated cells as compared to vehicle treated cells (Figure 3A). Likewise, ouabain-sensitive I_SC was smaller after apical nystatin treatment in canagliflozin treated cells as compared to vehicle treated cells (Figure 3B), with an overall effect that was more robust than for ENaC-mediated currents. Rates of canagliflozin-induced inhibition in both experiments were similar compared to intact cells (Figure 1) and faster than in basolateral-permeabilized cells (Figure 3E). These data support inhibition of both ENaC and Na⁺/K⁺ ATPase activity by canagliflozin, with the latter possibly being rate-limiting for canagliflozin’s effect on Na⁺ transport in intact cells.

Figure 3.

Canagliflozin inhibits electrogenic Na⁺ transport at both the apical and basolateral membranes in mCCD_cl1 cells. (A) I_SC was recorded from mCCD_cl1 cells treated first with 100 µm basolateral nystatin, followed by either DMSO or 25 µm canagliflozin (cana) on both sides. One hundred micromolar apical amiloride (amil) was administered at the end of each recording to inhibit residual ENaC-dependent I_SC. (B) I_SC recorded using mCCD_cl1 cells treated first with 100 µm apical nystatin, followed by either DMSO or 25 µm canagliflozin on both sides. Two hundred fifty micromolar basolateral ouabain was administered at the end of each recording to inhibit the residual Na⁺/K⁺ ATPase-dependent I_SC. Experiments in A and B were performed under an apical-to-basolateral Na⁺ gradient (140 to 25 mm). (C) Summary data showing residual amiloride-sensitive I_SC for cells treated with DMSO or canagliflozin under basolateral-permeabilized conditions. (D) Summary data showing residual ouabain-sensitive I_SC for cells treated with DMSO or canagliflozin under apical-permeabilized conditions. (E) Rates of 25 µm canagliflozin-induced I_SC inhibition under basolateral or apical permeabilization, as well as with intact mCCD_cl1 cells. (F, H) mCCD_cl1 cells were treated for 45 min with 0.1% DMSO or 25 µm canagliflozin. Western blots were then performed on biotinylated surface proteins, as well as 5% of the total cell lysate inputs, to quantify the relative surface expression of γ-ENaC (G) or NaKα (I). Molecular weight markers are shown to the left. Line and error bars represent mean ± SD (n = 6-7). Data were compared by unpaired t-test.

Canagliflozin Selectively Reduces Surface ENaC Abundance

Sgk1 is a major regulator of Na⁺ transport across both the apical and basolateral membranes of the principal cell, promoting functional expression of ENaC and Na⁺/K⁺ ATPase protein assemblies.⁴² We therefore assessed the effect of canagliflozin on membrane localization using surface biotinylation. Canagliflozin reduced surface expression of the ENaC γ subunit by ∼40% (Figure 3F, G), similar to the magnitude of inhibition of amiloride-sensitive currents when the apical membrane was isolated by nystatin permeabilization (Figure 3A). By contrast, canagliflozin had no effect on surface expression of NaKα, the catalytic subunit of the Na⁺/K⁺ ATPase (Figure 3H, I) despite near complete inhibition of ouabain-sensitive current in the basolateral membrane (Figure 3B). This suggests that canagliflozin inhibition of ENaC current likely depends on trafficking, whereas effects on the Na⁺/K⁺ ATPase may involve changes in its activity. These findings are consistent with known Sgk1-mediated effects on both transporters.^43-46

Inhibition of Sgk1 But Not Complex I Precludes Canagliflozin Inhibition of Amiloride-Sensitive I_SC

We theorized that canagliflozin-induced inhibition of Na⁺/K⁺ ATPase current was mediated at least in part through effects on Sgk1. If true, treating mCCD_cl1 cells with an Sgk1 inhibitor would be expected to mimic canagliflozin-induced inhibition under similar conditions. Although Sgk1 has been shown to regulate Na⁺/K⁺ ATPase activity in heterologous systems and in Xenopus-derived A6 renal epithelial cells,^45-47 inhibition of endogenous Sgk1 has not been shown to alter Na⁺/K⁺ ATPase activity in mammalian principal cells. We therefore assessed whether the selective Sgk1 inhibitor, GSK650394, directly reduces Na⁺/K⁺ ATPase-dependent I_SC in mCCD_cl1 cells.

As expected, bilateral 20 µm GSK650394 addition rapidly inhibited I_SC (Figure 4A). Subsequent inhibition by canagliflozin was accordingly reduced by pretreatment with GSK650394 versus vehicle control (Figure 4B), consistent with Sgk1 having a role in canagliflozin-induced I_SC inhibition. This is further supported by the similarity in the observed rates of I_SC inhibition (t_1/2 = 1.4 ± 0.1 min for GSK650394 vs. 1.9 ± 0.2 min for canagliflozin; Figure 2A, D), suggesting a common rate-limiting step. To directly test the effect of Sgk1 inhibition on Na⁺/K⁺ ATPase activity, we again utilized an apical membrane-permeabilized I_SC recording configuration to measure ouabain-sensitive currents in isolation. Ouabain-sensitive I_SC was substantially reduced in cells treated with GSK650394 prior to permeabilization (Figure 4C, D). On the other hand, ouabain-sensitive current was largely preserved after amiloride pretreatment, and greater than in GSK650394-treated cells. Together, these data demonstrate that Sgk1 acutely regulates Na⁺/K⁺ ATPase activity in CCD cells.

Figure 4.

The Sgk1 inhibitor, GSK650394, inhibits Na⁺/K⁺ ATPase-dependent Na⁺ transport mCCDcl1 cells. (A) I_SC recordings from mCCD_cl1 cells treated first with either DMSO or 20 µm GSK650394 (GSK), followed by 25 µm canagliflozin (cana). 100 µm apical amiloride (amil) was added at the end of each recording. B) Group data showing residual canagliflozin-sensitive I_SC for cells treated with DMSO or GSK650394. Line and error bars represent mean ± SD (n = 3). Data were compared by unpaired t-test and P values are shown in the figure. (C) I_SC recordings from mCCD_cl1 cells treated first with either DMSO, 20 µm GSK650934 or 100 µm amiloride, followed by 100 µm apical nystatin. Two hundred fifty micromolar basolateral ouabain was then administered to inhibit the residual Na⁺/K⁺ ATPase-dependent I_SC. (D) Group data showing residual ouabain-sensitive I_SC for cells treated with DMSO, GSK650934 or amiloride under apical-permeabilized conditions. Line and error bars represent mean ± SD (n = 4-5). Data were compared by one-way ANOVA with Tukey’s post-hoc.

We also tested the idea that canagliflozin inhibition of Na⁺/K⁺ ATPase activity was downstream of effects on Complex I activity, as we and others have reported.^12-14,48 The Complex I inhibitor, rotenone (10 µm) inhibited I_SC (Figure 5A). However, inhibition was partial, even with concentrations >5-fold higher than the drug’s IC₅₀ (<2 µm).⁴⁹  I_SC recovered and stabilized after ∼5 min following rotenone treatment, and adding canagliflozin afterward resulted in further inhibition (Figure 5B). In OK cells, we observed that acute (15 min) rotenone treatment did not reduce phosphorylated Sgk1 abundance. In contrast, 15 min canagliflozin and GSK650394 treatments selectively abolished the slower migrating Sgk1 band (Figure 5C-D). Canagliflozin and GSK650394 therefore elicit similar rapid responses in terms of Sgk1 protein reduction in OK cells and kinetics of I_SC inhibition in mCCD_cl1 cells. This response profile is quite different from that observed with rotenone treatment. These data suggest that (1) the global inhibitory effects of canagliflozin cannot be explained by Complex I inhibition alone, and (2) these effects are consistent with Sgk1 as a mediating effector of inhibition.

Figure 5.

Inhibition of mitochondrial complex 1 does not recapitulate the inhibitory effect of canagliflozin on I_SC in mCCDcl1 cells. (A) Representative I_SC recordings of mCCDcl1 cells treated first with 10 mm rotenone, then with or without 25 µm canagliflozin. One hundred micromolar apical amiloride was administered at the end of each recording to inhibit residual I_SC. (B) Group data showing residual I_SC following treatment with either 10 µm rotenone or 100 µm canagliflozin treatment. Line and error bars represent mean ± SD (n = 3). Data were compared by unpaired t-test and P values are shown in the figure. (C) Representative Sgk1 blot of duplicate filters of OK cells treated with vehicle (DMSO), 10 µm GSK650394, 25 µm canagliflozin and empagliflozin, and 10 µm rotenone for 15 min. (D) Blot quantifications.

Finally, we tested whether canagliflozin inhibited ENaC-dependent transport in native tissue by measuring amiloride-sensitive I_SC in mouse distal colon. We chose this because of the similar ion transport pathways shared between the distal nephron and distal colon, with the distal colon being anatomically compatible for Ussing chamber studies. To stimulate ENaC activation in the distal colon, we K⁺ loaded mice by feeding a 5% KCl diet for 1 wk prior to experiments.³⁴ Similar to our mCCD_cl1 experiments, 25 µm canagliflozin caused a clear downward deflection in I_SC, resulting in a ∼50% reduction of amiloride-sensitive I_SC, compared to DMSO controls (Figure 6). These data indicate that canagliflozin-dependent Na⁺ transport inhibition occurs in native tissue as well as in cultured cells.

Figure 6.

Canagliflozin inhibits ENaC activity in mouse distal colon. (A) Representative I_SC recordings from high K⁺-fed mouse distal colon segments treated with vehicle (0.1% DMSO) or 25 µm canagliflozin in an Ussing chamber system. One hundred micromolar amiloride was added at the end of the recording to determine residual ENaC-mediated Na⁺ transport. (B) Quantification of residual amiloride-sensitive I_SC in DMSO versus canagliflozin-treated tissues. Line and error bars represent mean ± SD (n = 4). Data were compared by unpaired t-test.

Discussion

Our results provide evidence for novel pleiotropic canagliflozin effects that impact the function of kidney transporters. Inhibition of transepithelial Na⁺ transport in mCCD_cl1 cells occurred at both membranes through both ENaC and the Na⁺/K⁺ ATPase. Effects on Sgk1 strongly suggest a shared upstream molecular target. Accordingly, the Sgk1 inhibitor GSK650394 recapitulated the transport effects of canagliflozin and precluded inhibition by canagliflozin. Canagliflozin effects on ENaC were slower than those on the Na⁺/K⁺ ATPase, and coincided with reductions in the ENaC surface pool, consistent with an Sgk1-mediated mechanism, although we did not directly measure Nedd4-2 activity as a readout of canonical Sgk1 signaling.⁴³ In contrast, Na⁺/K⁺ ATPase inhibition did not coincide with reduced surface expression, suggesting a reduction in catalytic activity as previously reported for Sgk1.⁴⁵

Importantly, we saw no inhibition of I_SC in mCCD_cl1 cells by empagliflozin, suggesting this is not a class effect. These results echo our recent report that canagliflozin, but not empagliflozin, inhibited NHE3 activity in both OK cells that endogenously express NHE3, and in NHE3-transfected, NHE1-null Chinese hamster ovary cells.¹² It is important to note that other groups have reported NHE3 inhibition by empagliflozin in mice.^11,50 As Sgk1 regulates NHE3,³¹ canagliflozin’s effects on Sgk1 may partly account for the NHE3 inhibition we previously reported, though this remains to be tested. Whether or not canagliflozin alters the activity of other Sgk1-regulated transporters (eg, NCC or NKCC2)⁵¹ is also unknown.

We and others reported that canagliflozin inhibited mitochondrial Complex I, but that empagliflozin did not.^12,14 Although this could have accounted for inhibition of amiloride-sensitive I_SC, pretreatment with the Complex I inhibitor rotenone did not preclude inhibition by canagliflozin. Furthermore, rotenone did not recapitulate canagliflozin’s effects on Sgk1 in OK cells, failing to specifically reduce the λPP-sensitive slower migrating Sgk1 band after 15 min. These data are inconsistent with Complex I as the sole upstream target for canagliflozin’s specific effects on the Sgk1-ENaC-Na⁺/K⁺-ATPase axis.

Early investigations of Sgk1 and Na⁺/K⁺ ATPase interaction have suggested that cell-surface expression is alternately affected or unaffected, depending on the model system.^42,45,46 We found that canagliflozin inhibited the Na⁺/K⁺ ATPase without affecting surface expression of its catalytic α subunit, and with kinetics matching that of the Sgk1 inhibitor, GSK6503954. Here we also report that Sgk1 inhibition reduced Na⁺/K⁺ ATPase activity, while previous studies showed that Sgk1 overexpression increased Na⁺/K⁺ ATPase activity.⁴⁵ Together, our data are consistent with a model where canagliflozin inhibits the Na⁺/K⁺ ATPase via a mechanism involving Sgk1. Previous reports have hinted at gliflozin effects on Sgk1 in vivo. Dapagliflozin reduced Sgk1 levels in T helper 17 cells and myocardial tissue isolated from diabetic rabbits.^52,53 Empagliflozin was also reported to reduce Sgk1 levels in myocardial tissue isolated from db/db mice (10 mg/kg/day).⁵² Empagliflozin (10 mg/kg/day for 2 wk) had no effect on ENaC abundance in kidney tissue from control or hypertensive rats.⁵⁴ In contrast, dapagliflozin reduced ENaC γ subunit levels in myocardial tissue isolated from diabetic rabbits.⁵² Here we found that acute canagliflozin treatment reduced Sgk1 levels and inhibited Sgk1-dependent Na⁺ transport.

An important consideration in our study is the clinical relevance of drug concentrations used. Pharmacokinetic studies in human studies show that 300 mg canagliflozin doses result in plasma concentrations that peak at ∼10 µm.⁵⁵ In mice, 30 mg/kg canagliflozin doses resulted in peak plasma concentrations of 8 µm in males and 30 µm in females.^56,57 More modest sex-dependent differences were observed in humans.⁵⁸ In humans, canagliflozin has a 119-L volume of distribution,¹⁷ indicating extensive distribution into tissues. Accordingly, concentrations of the drug in mice were 3-fold⁵⁶ to 9-fold⁵⁷ higher in renal tissues than in plasma. In this context, an IC₅₀ of 16 µm for canagliflozin’s effects in tubular cells is likely relevant in humans.

Our in vitro data raise the possibility that canagliflozin treatment may alter or impair Sgk1 function in the kidney. Future studies will be required to determine if these effects can be observed in vivo. Under normal dietary conditions, Sgk1-deficient mice have no overt abnormalities regarding salt or fluid homeostasis, aside from slightly elevated plasma K⁺ and aldosterone levels^59,60 Phenotypic differences were only revealed during Na⁺ restriction or K⁺ loading conditions, during which ENaC-dependent transport was reduced despite significantly elevated aldosterone levels.^59-61 Surprisingly, Sgk1-deficient mice had even more colonic ENaC activity under Na⁺ restriction than wild type mice.⁶² Together, these data suggest that other mechanisms contribute substantially to ENaC regulation in vivo when Sgk1 activity is abolished.

Off-target effects of gliflozins have long been speculated.⁶³ For instance, the blood pressure reducing effects of gliflozins exceed what would be expected from glycosuria and osmotic diuresis alone.⁶⁴ Among SGLT2i’s, 300 mg canagliflozin demonstrated the greatest reduction in systolic blood pressure.⁶⁵ Volume-related adverse events were also dose-dependent at 100 and 300 mg canagliflozin, despite SGLT2 inhibition likely being saturated at lower doses.^28,29 SGLT1 inhibition in the kidney at this dose is poorly supported due to extensive binding to plasma proteins.⁶⁶ These findings suggest drug-specific, noncanonical mechanisms may contribute to canagliflozin’s distinct clinical profile. Indeed, meta-analyses have consistently ranked 300 mg canagliflozin highest for reducing HbA_1c and body weight,^27,65 further supporting canagliflozin-specific effects. Our findings identify Sgk1 inhibition and the downstream reductions in ENaC and Na⁺/K⁺-ATPase activity as a novel mechanism that may contribute to canagliflozin-specific outcomes.

Author Contributions

Andrew J. Nickerson (Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Project administration, Supervision, Writing - original draft, Writing - review & editing), Wafaa N. Albalawy (Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Writing - original draft, Writing - review & editing), Elynna B. Youm (Data curation, Formal analysis, Investigation, Methodology, Writing - review & editing), Nicole A. Joseph (Data curation, Formal analysis, Methodology, Writing - review & editing), Kennedy G. Szekely (Data curation, Formal analysis, Methodology, Writing - review & editing), Thomas R. Kleyman (Conceptualization, Funding acquisition, Investigation, Project administration, Resources, Supervision, Writing - review & editing), Ora A. Weisz (Conceptualization, Funding acquisition, Investigation, Methodology, Project administration, Resources, Supervision, Writing - review & editing), and Ossama B. Kashlan (Conceptualization, Formal analysis, Funding acquisition, Investigation, Project administration, Supervision, Writing - original draft, Writing - review & editing)

Funding

This work was supported by the National Institutes of Health (DK136356 and DK140634 to A.J.N., DK137167 and HL147818 to T.R.K., DK125439 to O.B.K., DK125049 to O.A.W., DK137329, AG021885, DK119180), an American Society of Nephrology predoctoral fellowship to W.N.A., and a Saudi Ministry of Education Scholarship to W.N.A.

Conflict of Interest Statement

O.A.W. is an Advisor for Judo Biosciences, and a consultant for Maze Therapeutics and Merck. O.A.W. holds the position of Executive Editor for Function and is blinded from reviewing or making decisions for the manuscript. The authors declare no other conflicts of interest.

Data Availability

The data underlying this article are available in the article.