Interaction of the Sodium/Glucose Cotransporter (SGLT)

Home , Canagliflozin, SGLT2 inhibitor

Supplemental material to this article can be found at: http://jpet.aspetjournals.org/content/suppl/2016/04/27/jpet.116.232025.DC1

1521-0103/358/1/94–102$25.00 http://dx.doi.org/10.1124/jpet.116.232025 THE JOURNAL OF PHARMACOLOGY AND EXPERIMENTAL THERAPEUTICS J Pharmacol Exp Ther 358:94–102, July 2016 Copyright ª 2016 by The American Society for Pharmacology and Experimental Therapeutics

Interaction of the Sodium/Glucose Cotransporter (SGLT) 2 Inhibitor Canagliflozin with SGLT1 and SGLT2: Inhibition Kinetics, Sidedness of Action, and Transporter-Associated Incorporation Accounting for its Pharmacodynamic and Pharmacokinetic Features s

Ryuichi Ohgaki, Ling Wei, Kazunori Yamada, Taiki Hara, Chiaki Kuriyama, Suguru Okuda, Kiichiro Ueta, Masaharu Shiotani, Shushi Nagamori, and Yoshikatsu Kanai Department of Bio-system Pharmacology, Graduate School of Medicine, Osaka University, Osaka, Japan (R.O., L.W., S.O., S.N., Y.K.); and Research Division, Mitsubishi Tanabe Pharma Corporation, Saitama, Japan (K.Y., T.H., C.K., K.U., M.S.) Downloaded from Received January 12, 2016; accepted April 26, 2016

ABSTRACT Canagliflozin, a selective sodium/glucose cotransporter (SGLT) than that of a-methyl-D-glucopyranoside. Canagliflozin inhibited 2 inhibitor, suppresses the renal reabsorption of glucose a-methyl-D-glucopyranoside–induced SGLT1- and SGLT2- and decreases blood glucose level in patients with type mediated inward currents preferentially from the extracellu- jpet.aspetjournals.org 2 diabetes. A characteristic of canagliflozin is its modest lar side and not from the intracellular side. Based on the SGLT1 inhibitory action in the intestine at clinical dosage. To Ki value, canagliflozin is estimated to sufficiently inhibit reveal its mechanism of action, we investigated the interaction SGLT2 from the urinary side in renal proximal tubules. The Ki of canagliflozin with SGLT1 and SGLT2. Inhibition kinetics value for SGLT1 suggests that canagliflozin suppresses and transporter-mediated uptake were examined in human SGLT1inthesmallintestinefrom the luminal side, whereas SGLT1- or SGLT2-expressing cells. Whole-cell patch-clamp it does not affect SGLT1 in the heart and skeletal muscle,

recording was conducted to examine the sidedness of drug considering the maximal concentration of plasma-unbound at ASPET Journals on October 1, 2021 action. Canagliflozin competitively inhibited SGLT1 and canagliflozin. Similarly, SGLT1 in the kidney would not be SGLT2, with high potency and selectivity for SGLT2. Inhibition inhibited, thereby aiding in the prevention of hypoglycemia. constant (Ki) values for SGLT1 and SGLT2 were 770.5 and After binding to SGLT2, canagliflozin may be reabsorbed by 4.0 nM, respectively. 14C-canagliflozin was suggested to be SGLT2, which leads to the low urinary excretion and pro- transported by SGLT2; however, the transport rate was less longed drug action of canagliflozin.

Introduction (SGLTs) in the apical membrane and then leaves the epithelial cells through the facilitative glucose transporters in the Under euglycemic conditions, glucose in the glomerular basolateral membrane to the bloodstream (Wright et al., filtrate is completely reabsorbed in the proximal tubules of the 2011). Under physiological conditions, the bulk of filtered kidney so that it is not excreted into urine. In the process of glucose (∼90%) is reabsorbed by SGLT2 (SLC5A2), a low- glucose reabsorption, glucose in the luminal fluid is taken up affinity high-capacity transporter in the proximal convoluted by tubular epithelial cells via sodium/glucose cotransporters tubule (particularly in the S1 segment) (Kanai et al., 1994; You et al., 1995). The remaining filtered glucose (∼10%) is This work was supported by Grants-in-Aid for Scientific Research from the reabsorbed by SGLT1 (SLC5A1), a high-affinity low-capacity Japan Society for the Promotion of Science; the Ministry of Education, Science, transporter in the proximal straight tubule (particularly in Culture, Sports and Technology of Japan (MEXT); and the Mitsubishi Tanabe Pharma Corporation. L.W. was supported by a scholarship from MEXT for the S3 segment) (Hediger et al., 1987; Lee et al., 1994). In the foreign students. small intestine, SGLT1 is present in the apical membrane of This work was previously presented in part as follows: L.W., R.O., S.N., and epithelial cells and is responsible for the intestinal absorption Y.K. Interaction of a novel SGLT2 inhibitor Canagliflozin with human SGLT2. The 57th Annual Meeting of the Japan Diabetes Society, 2014 May 22–24; of glucose and galactose (Wright et al., 2011). To target SGLT2 Osaka, Japan; and R.O., L.W., S.N., K.Y., T.H., C.K., K.U., M.S., and Y.K. in renal proximal tubules, selective SGLT2 inhibitors have Study on the transport of SGLT2 inhibitor canagliflozin by human SGLT2 and been developed as antidiabetic agents. These compounds in- its inhibitory action from extracellular and intracellular sides. The 58th Annual Meeting of the Japan Diabetes Society, 2015 May 21–24; Fukuoka, hibit the renal reabsorption of glucose to increase its urinary Japan. excretion, thereby lowering plasma glucose levels in an insulin- R.O. and L.W. contributed equally to this work. dx.doi.org/10.1124/jpet.116.232025. independent manner (Liang et al., 2012; Boyle and Wilding, s This article has supplemental material available at jpet.aspetjournals.org. 2013; Fujita and Inagaki, 2014; Peene and Benhalima, 2014).

ABBREVIATIONS: AMG, a-methyl-D-glucopyranoside; 8-Br-cAMP, 8-bromo-cAMP; DMSO, dimethylsulfoxide; DOX, doxycycline; HBSS, Hank’s balanced salt solution; IAMG, a-methyl-D-glucopyranoside–induced current; IAMG/Cm, current density; Ki, inhibition constant; PCR, polymerase chain reaction; SGLT, sodium/glucose cotransporter.

94 Canagliflozin Actions on Sodium/Glucose Cotransporters 95

Canagliflozin, a C-glucoside with a thiophene ring (Nomura Japan). a-Methyl-D-glucopyranoside (AMG) was purchased from et al., 2010), is the first selective SGLT2 inhibitor approved by Sigma-Aldrich (St. Louis, MO). Phlorizin was obtained from TCI 14 the US Food and Drug Administration to improve glycemic (Tokyo, Japan). C-Canagliflozin ((1S)-1,5-anhydro-1-C-(3-{[5-(4- 14 control in individuals with type 2 diabetes mellitus (Elkinson fluorophenyl)thiophen-2-yl]-[ C]methyl}-4-methylphenyl)-D-glucitol hemihydrate) (2.15 GBq/mmol) was synthesized by Janssen Research and Scott, 2013). The European Medicines Agency also 14 & Development, LLC (Spring House, PA). C-AMG {methyl-a-D- approved canagliflozin for this use in 2013. The administra- [glucose-14C(U)]glucopyranoside} (10.69 GBq/mmol) was from tion of canagliflozin at a clinical dosage to patients with type 2 PerkinElmer (Boston, MA). Blasticidin S was purchased from ∼ diabetes mellitus induces urinary glucose excretion at 100 g/day, InvivoGen (San Diego, CA). 8-Bromo-cAMP (8-Br-cAMP) was which would not increase the risk of hypoglycemia (Devineni obtained from Sigma-Aldrich. Penicillin-streptomycin solution, et al., 2012, 2013; Iijima et al., 2015). In addition to anti- hygromycin B, G418 sulfate, and doxycycline (Dox) were from Wako hyperglycemic action, canagliflozin, as a selective SGLT2 Pure Chemical Industries (Osaka, Japan). inhibitor, exhibits other favorable effects for the treatment of Cell Lines. For the construction of the tetracycline-inducible type 2 diabetes mellitus, including loss in body weight and expression vector, a plasmid backbone of pJTI R4 DEST [except the l hypotensive effects (Sha et al., 2011; Lavalle-Gonzalez et al., region flanked by integrase attR1 and attR2 sites (nucleotide – 2013). Among the selective SGLT2 inhibitors currently avail- position 3842 5545)], and a partial fragment of pcDNA4/TO (nucleotide position 232–1319: composed of cytomegalovirus promoter, able for clinical use, a characteristic of canagliflozin is its mod- tetracycline operator, multiple cloning sites, and bovine growth est SGLT1 inhibitory action in the intestine at clinical dosage hormone polyadenylation site) were amplified by polymerase chain

(Grempler et al., 2012; Kuriyama et al., 2014). This is probably reaction (PCR) using Pwo Super Yield DNA Polymerase (Roche Downloaded from because of its relatively low SGLT2/SGLT1 selectivity as well as Diagnostics, Mannheim, Germany). The obtained two PCR products its relatively high clinical dosage, owing to its high plasma were then ligated together by homologous recombination using In- protein binding (Fujita and Inagaki, 2014; Devineni et al., 2015). Fusion HD Cloning Kit (Clontech, Palo Alto, CA). The nucleotide A major question regarding the mechanisms of action of sequences of the primer pairs were as follows: forward 59-GGCTT- SGLT2 inhibitors is the sidedness of action [i.e., whether GGCTCCGGTGcccgt-39 and reverse 59-GGCGTAATCATGGTCatagctg-39 these compounds inhibit SGLT2 from the extracellular side for pJTI R4 DEST; and forward 59- CACCGGAGCCAAGCCacgcgttga- 9 9 jpet.aspetjournals.org (luminal side) or from the cytoplasmic side]. A canonical cattgattattgactag-3 and reverse 5 -GACCATGATTACGCCtccccag- catgcctgctattgtc-39 for pcDNA4/TO, wherein the complementary nonselective SGLT inhibitor, phlorizin, the parent compound sequences between the primer sets are indicated by capital letters. of selective SGLT2 inhibitors, exhibits a ∼600-fold lower The coding sequences of human SGLT1 (GenBank accession inhibition constant (Ki) value on SGLT1 from the extracellular #AB463272) and SGLT2 (GenBank accession #NM_003041.3) were side than it does from the intracellular side (Eskandari et al., amplified by PCR from Flexi ORF SGLT1 Clone pF1KB8725 2005). Phlorizin is freely filtrated by the glomerulus and (Promega, Madison, WI) and from a full-length synthetic SGLT2 CDS mainly excreted into urine, which is in line with the idea that clone (GenScript USA, Piscataway, NJ), respectively. The PCR products the glomerular-filtrated phlorizin acts on SGLTs from the were integrated into the tetracycline-inducible expression vector at ASPET Journals on October 1, 2021 luminal side to increase urinary glucose excretion (Silverman, (HindIII/XbaI linearized) by homologous recombination using In- 1974). In contrast, canagliflozin exhibits high plasma protein Fusion HD Cloning Kit (Clontech). The nucleotide sequences of the 9 binding (99%), and its urinary excretion is 1% or less of a single primer pairs were as follows: forward 5 -ACTTAAGCTTggtaccaccatg- gacagtagcacctggagc-39 and reverse 59-TTAAACGGGCCCTCTAGAct oral dosage (Inagaki et al., 2014; Devineni et al., 2015; Kinoshita caggcaaaatatgcatggc-39 for SGLT1; and forward 59-GTTTAAACTT- and Kondo, 2015). These observations raised the possibility that AAGCTTccaccatggaggagcacacagaggcaggct-39 and reverse 59-AAAC- canagliflozin enters epithelial cells of the renal proximal tubule GGGCCCTCTAGAttaggcatagaagccccagaggaac-39 for SGLT2, wherein through the basolateral membrane to exert its inhibitory effect the primer sequences identical to that of the tetracycline-inducible on SGLT2 in the apical membrane from the cytoplasmic side. expression vector were indicated by capital letters. There is currently no experimental evidence for the sidedness of To introduce an R4 attP retargeting site and promoter-less the action of clinically used selective SGLT2 inhibitors. In the neomycin resistance gene into the genomic PhiC31 pseudo-attP site, study using TA-3404, a compound with a canagliflozin-related the TetR-expressing T-Rex-CHO and T-Rex-293 cells (Invitrogen, structure, it has been suggested that the SGLT2 inhibition Carlsbad, CA) were cotransfected with pJTI phiC31 Int and pJTI/Neo occurs from the extracellular side (Ghezzi et al., 2014). In ad- using X-tremeGENE 9 DNA Transfection Reagent (Roche Diagnostics). dition, we speculated that canagliflozin is not only an inhibitor After the selection for hygromycin B resistance and cloning, the established Jump-In TI platform cells (termed as JTREx-CHO and of SGLT2, but also a substrate for SGLT2, and that SGLT2- JTREx-293 cells) were further cotransfected with pJTI R4 Int and mediated active transport, at least in part, explains the low uri- the tetracycline-inducible expression vectors for human SGLT1 or nary excretion rate observed with the use of canagliflozin. SGLT2 constructed as described above. Successful retargeting of the In this study, to reveal the molecular mechanisms of action expression vectors at the R4 attP site confers neomycin resistance to of canagliflozin, we investigated its interaction with SGLTs in the Jump-In TI platform cells by cointegration of the EF1a promoter vitro using cultured cell lines stably expressing human SGLT1 upstream of the promoter-less neomycin resistance gene. Retargeted and SGLT2. We examined the concentration-dependent pro- cells were selected for G418 sulfate resistance, cloned, and used for the measurements. files of inhibition to determine the Ki values of canagliflozin for each SGLT, the uptake of canagliflozin by SGLTs using Cells were grown at 37°C in a humidified incubator supplied with ’ ’ ’ radioactive 14C-canagliflozin, and the sidedness of the drug 5% CO2. Dulbecco s modified Eagle s medium (DMEM)/Ham s F-12 and DMEM (Wako Pure Chemical Industries) supplemented with 10% action using whole-cell patch-clamp recording. fetal bovine serum (ThermoFisher Scientific, Grand Island, NY), 1% (v/v) penicillin-streptomycin solution, and 5 mg/ml blasticidin S Materials and Methods were used to maintain the T-Rex-CHO and T-Rex-293 cells, respectively. Where indicated, hygromycin B (100 mg/ml) and G418 sulfate Chemicals. Canagliflozin was synthesized by the Mitsubishi (500 mg/ml) were additionally supplied to the medium for selection and Tanabe Pharma Corporation, Medicinal Chemistry Laboratory (Saitama, maintenance of the constructed stable cell lines. 96 Ohgaki et al.

14C-AMG Uptake Measurement and Inhibition Experi- Clampfit 10.4 software (Molecular Devices). The whole-cell capacitance ments. Uptake of AMG, an SGLT-selective substrate, was measured (Cm) was measured at the beginning of each measurement by the in the cell lines. Cells were seeded on 24-well plates and cultured for square-pulse method with the Membrane Test tool of the Clampex 8.2

24 hours in the presence of 10 ng/ml DOX to induce the expression of software. The current density [AMG-induced current (IAMG)/Cm] was human SGLTs. Uptake measurements were performed as previously calculated to normalize the recorded IAMG. described (Wiriyasermkul et al., 2012). Briefly, cells were incubated at To investigate the sidedness of inhibition by canagliflozin on 37°C in glucose-free HBSS containing 14C-AMG at concentrations of SGLT1, the measurements were performed in the presence (n 5 10) 400 mM (25.4 MBq/mmol) and 1500 mM (50.8 MBq/mmol) for SGLT1 or absence (n 5 11) of intracellular canagliflozin (20 mM). In the case of and SGLT2, respectively. NaCl in glucose-free HBSS was replaced experiments on SGLT2, all the measurements were performed with with choline chloride for the sodium-free condition. For inhibition intracellular application of canagliflozin (200 nM), and the IAMG/Cm experiments, the indicated concentrations of phlorizin or canagliflozin values before and after the application of extracellular canagliflozin dissolved in dimethylsulfoxide (DMSO) were added to the assay (200 nM) were compared (n 5 7). The indicated concentrations of buffer. The assay buffer contained up to 0.1% DMSO (final concen- canagliflozin itself did not induce a significant current when applied tration), unless otherwise noted. Up to 0.2% of DMSO did not extracellularly (Supplemental Fig. 4). Replacement of the intracellu- significantly affect the uptake of 14C-AMG and 14C-canagliflozin (data lar fluid with high-sodium internal solution containing canagliflozin not shown). After cell lysis, the radioactivity was measured by liquid was confirmed under current-clamp mode (0 pA) after achieving scintillation counting and normalized according to the protein whole-cell configuration. ’ amount. The IC50 of phlorizin and canagliflozin was determined by Statistical Analyses. Unpaired, two-tailed Student s t tests were fitting the data to inhibition curves using nonlinear regression (four- performed with Excel 2013 software (Microsoft, Redmond, WA), where parameter Hill function; SigmaPlot 12.5; SYSTAT). To determine the p values ,0.05 were considered to be significant. Downloaded from kinetic properties of inhibition, uptake measurements were performed with varied concentrations of 14C-AMG in the presence or absence of the indicated concentrations of inhibitors. Uptake rates were plotted against 14C-AMG concentration and fitted to the Michaelis-Menten Results curve. Ki was calculated using the following equation when competitive Functional Expression of SGLT1 and SGLT2. To 5 1 1 inhibition was observed: V Vmax/{1 [Km/(S)] [1 (I)/Ki]}, where V is confirm the functional expression of SGLT1 and SGLT2, 14C-

14 14 jpet.aspetjournals.org the C-AMG uptake rate, Vmax is the maximal C-AMG uptake rate, AMG uptake was measured in doxycycline-treated JTREx- K is the Michaelis constant; [S] is the 14C-AMG concentration, and [I] m CHO-SGLT1 and JTREx-CHO-SGLT2 cells. Doxycycline is the inhibitor concentration. All analyses were performed using the 14 enzyme kinetics module of SigmaPlot 12.5 (SYSTAT, San Jose, CA). increased C-AMG uptake in the JTREx-CHO-SGLT1 and 14C-Canagliflozin Uptake Measurements. For 14C-canagliflozin JTREx-CHO-SGLT2 cells but not in the parental JTREx-CHO 14 uptake measurement, cells were incubated with the indicated concen- cell line (Fig. 1, A and B). The expressed C-AMG uptake was 1 1 trations of 14C-canagliflozin (2.15 GBq/mmol) in the presence or ab- Na dependent and was abolished in the absence of Na . sence of sodium. Phlorizin (50 mM) was added to the assay buffer where Phlorizin almost completely suppressed 14C-AMG uptake in indicated. To examine the temperature dependence of the uptake, the both cell lines (Fig. 1, A and B). The 14C-AMG uptakes by at ASPET Journals on October 1, 2021 measurements were performed using 14C-canagliflozin (5 and 10 nM) or SGLT1 and SGLT2 were linearly dependent on the incubation 14 C-AMG (500 mM, 25.4 MBq/mmol) for 20 minutes on ice or at 37°C. time for (at least) 30 and 90 minutes, respectively (Fig. 1, C Whole-Cell Patch-Clamp Recordings. Cells were seeded on and D). These results indicate that SGLT1 and SGLT2 were poly-D-lysine–coated glass coverslips and cultured in the presence functionally expressed in the cell lines constructed. Protein of DOX (10 ng/ml) for 12–20 hours. To increase the activity of SGLT2, expression of SGLT1 and SGLT2 was also confirmed by Western 8-Br-cAMP (0.1 mM) was added to the medium 1 hour before the measurements (Ghezzi and Wright, 2012). Whole-cell patch-clamp blot analysis using specific antibodies (Supplemental Fig. 1). Based recording was performed at room temperature for SGLT1 (JTREx- on the linear range observed over the time course, the uptakes 293-SGLT1 cells) and at 32°C for SGLT2 (JTREx-293-SGLT2 cells) in were measured for 5 and 30 minutes on SGLT1 and SGLT2 a stage-top chamber on an Eclipse TE300 inverted microscope (Nikon, cells, respectively, in the subsequent kinetic analyses. Tokyo, Japan) with gravitational perfusion of external solution Kinetic Properties of the Inhibition of 14C-AMG [100 mM Mannitol, 150 mM NaCl, 1 mM CaCl2, 1 mM MgCl2, and Uptake by Canagliflozin. The concentration dependence 10 mM HEPES-Tris (pH 7.4)]. To elicit the transporter-mediated of the effects of canagliflozin on 14C-AMG uptake was current, the indicated concentration of AMG was added to the external examined and compared with that of the effects of phlorizin. solution, at which no significant inward currents were induced by the Phlorizin inhibited 14C-AMG uptake mediated by SGLT1 and equivalent osmolarity of mannitol (Supplemental Fig. 4). Patch SGLT2 in a concentration-dependent manner (Fig. 2, A and pipettes were prepared from borosilicate glass capillary GD-1.5 (Narishige, Tokyo, Japan) pulled by a PD-7 pipette puller (Narishige), B). The IC50 values of phlorizin on SGLT1 and SGLT2 were giving a resistance of 7–10 MV when filled with internal solution. The 439 and 56 nM, respectively. Similarly, the inhibition of composition of the standard internal solution was 145 mM CsCl, 5 mM SGLT1 and SGLT2 by canagliflozin also exhibited clear NaCl, 11 mM EGTA, and 10 mM HEPES-Tris (pH 7.4). A high-sodium concentration dependence with an IC50 of 1550 nM for SGLT1 internal solution [150 mM NaCl, 11 mM EGTA, and 10 mM HEPES- and 6.8 nM for SGLT2 (Fig. 2, C and D). Tris (pH 7.4)] was used for the intracellular application of canagli- To further characterize the kinetic properties of the in- flozin. Canagliflozin was dissolved in DMSO, and was diluted into hibitory effects of canagliflozin, the Ki values and mode of internal and external solutions. External and internal solutions with inhibition were determined. Nonlinear regression analysis of and without canagliflozin contained 0.5% DMSO (final concentration), the inhibitory effects of the compounds (Supplemental Fig. 2, the concentration at which DMSO did not significantly affect the A, C, E, and G) revealed that canagliflozin and phlorizin AMG-induced inward current (Supplemental Fig. 4). The whole-cell 14 currents were recorded at a holding potential of 260 mV using an competitively inhibit SGLT1- and SGLT2-mediated C-AMG AXOPATCH 200B amplifier (Axon Instruments, Foster City, CA) with uptake. The competitive inhibition was confirmed by Eadie- a low-pass Bessel filter set at 5 kHz. Data were digitized at 10 kHz by a Hofstee plots (Supplemental Fig. 2, B, D, F, and H). The Ki Digidata 1320A AD/DA converter (Axon Instruments) with Clampex values were determined from three or five independently 8.2 software (Molecular Devices, Sunnyvale, CA) and analyzed using performed trials, and are listed in Table 1. Canagliflozin Actions on Sodium/Glucose Cotransporters 97 Downloaded from jpet.aspetjournals.org at ASPET Journals on October 1, 2021

Fig. 1. Properties of 14C-AMG uptake in the JTREx-CHO-SGLT1 and JTREx-CHO-SGLT2 cells. (A and B) Na+ dependence and phlorizin sensitivity of doxycycline-induced 14C-AMG uptake. Doxycycline-treated (Dox+) or untreated (Dox2) cells were used to determine 14C-AMG uptake in the presence or absence of Na+ and phlorizin (100 mM). For the JTREx-CHO-SGLT1 cells (SGLT1) and parental JTREx-CHO cells (Mock), the uptake of 14C-AMG (400 mM) was measured for 5 minutes (A). For the JTREx-CHO-SGLT2 cells (SGLT2) and parental JTREx-CHO cells (Mock), the uptake of 14C-AMG (1500 mM) was measured for 30 minutes (B). (C and D) Time course of 14C-AMG uptake examined in the presence of Na+ (glucose-free HBSS). 14C-AMG of 400 mMwas used for the Dox-treated JTREx-CHO-SGLT1 (SGLT1) and parental JTREx-CHO (Mock) cells (C). 14C-AMG of 1500 mM was used for the Dox-treated JTREx-CHO-SGLT2 cells (SGLT2) and parental JTREx-CHO cells (Mock) (D). Data shown depict the mean 6 S.E.M. n =3–4. ***p , 0.001.

SGLT2-Mediated Uptake of 14C-Canagliflozin. To ex- SGLT2-mediated transport. The uptake of 14C-canagliflozin amine whether canagliflozin is transported by SGLT1 and via SGLT2 increased during the first 20 minutes of the SGLT2, we measured the uptake of 14C-canagliflozin in incubation time (Fig. 3C). the JTREx-CHO-SGLT1 and JTREx-CHO-SGLT2 cells. To confirm that the SGLT2-associated increase of No statistically significant uptake of 14C-canagliflozin 14C-canagliflozin radioactivity is because of the SGLT2- was detected for SGLT1 at 10 and 20 nM, although the mediated transport and not SGLT2 binding, its tempera- JTREx-CHO-SGLT1 cells tended to exhibit a slightly higher ture dependency was examined. When similar measurements uptake compared with mock cells in the presence of Na1 were performed on ice, uptake of the 14C-AMG in JTREx-CHO- (Supplemental Fig. 3). In contrast, the JTREx-CHO-SGLT2 SGLT2 cells decreased to the background level (i.e., to that of cells, showed a significantly higher 14C-canagliflozin up- the parental JTREx-CHO cells), confirming the transporter- take compared with mock cells at 5–20 nM in the presence, mediated uptake to be highly sensitive to temperature (Fig. but not in the absence, of Na1 (Fig.3A).Theconsiderable 3D). Similarly, the JTREx-CHO-SGLT2 cells showed a 14C-canagliflozin uptake observed in the parental JTREx- higher 14C-canagliflozin uptake compared with the parental CHO cells most likely represents the incorporation via JTREx-CHO cells at 37°C, whereas this difference disap- passive diffusion through the plasma membrane. SGLT2- peared on ice (Fig. 3D). Such temperature dependence mediated 14C-canagliflozin uptake increased in a concentration- suggests that the observed 14C-canagliflozin uptake repre- dependent manner (Fig. 3A) and was completely inhibited sents SGLT2-mediated transport but not the binding of by phlorizin (Fig. 3B), consistent with the properties of 14C-canagliflozin to SGLT2. 98 Ohgaki et al. Downloaded from jpet.aspetjournals.org at ASPET Journals on October 1, 2021

Fig. 2. Concentration-dependent inhibition of 14C-AMG uptake by phlorizin and canagliflozin. 14C-AMG uptake mediated by SGLT1 or SGLT2 was plotted against inhibitor concentration. SGLT1- or SGLT2-mediated uptake was obtained by subtracting 14C-AMG uptake in the parental JTREx-CHO cells from that in the JTREx-CHO-SGLT1 or JTREx-CHO-SGLT2 cells, respectively. To measure SGLT1-mediated uptake, cells were incubated with 400 mM 14C-AMG for 5 minutes with varied concentrations of phlorizin or canagliflozin (A and C). To measure SGLT2-mediated uptake, cells were incubated with 1500 mM 14C-AMG for 30 minutes with varied concentrations of phlorizin or canagliflozin (B and D). The 14C-AMG uptake in the presence of inhibitors is shown as the percentage of the uptake without inhibitors. Data shown depict the mean 6 S.E.M. n =3–4.

Sidedness of Canagliflozin Action. The sidedness of mode (0 pA). After attaining the whole-cell configuration, the canagliflozin action on SGLTs was investigated using the original inside-negative membrane potential was promptly whole-cell configuration of the patch-clamp method by shifted toward 0 mV, as expected from the ionic composition applying canagliflozin to the outside or inside of the cells of the internal solution and reached a plateau at ∼150 seconds (Fig. 4A). Because AMG-induced electric currents in the (Supplemental Fig. 5), indicating complete replacement of JTREx-CHO-SGLT2 cells used for the uptake experiments intracellular fluid with the drug-containing pipette solution were too small for reliable electrophysiological recording, we within this time range. Therefore, all the measurements were prepared other stable cell lines (JTREx-293-SGLT1 and performed after 150 seconds. JTREx-293-SGLT2 cells) established from the JTREx-293 Neither the JTREx-293-SGLT1 nor the JTREx-293-SGLT2 cellsandusedthemforwhole-cellpatch-clamprecording. cells, without induction by doxycycline treatment, showed When the AMG uptakes were measured in the two SGLT2- any significant background inward current by the addition expressing cell lines, JTREx-293-SGLT2 cells showed a 2.32- of AMG into the external solution (Fig. 4B). After the fold higher uptake at 30 minutes than JTREx-CHO-SGLT2 cells (data not shown). Furthermore, to enhance the activity TABLE 1 of SGLT2, 8-Br-cAMP was added to the medium, and measure- Ki values of phlorizin and canagliflozin ments were conducted at 32°C (Hummel et al., 2011; Ghezzi and Wright, 2012). Inhibitor Ki for SGLT1 (nM) Ki for SGLT2 (nM) For the intracellular application of the drug, the time course Phlorizin 545.7 6 56.89 (n = 3) 102.3 6 27.41 (n =3) 6 6 for the replacement of intracellular fluid with a canagliflozin- Canagliflozin 770.5 85.19 (n =5) 4.0 0.77 (n =3) containing pipette solution was estimated under current-clamp Values are presented as the mean 6 S.E.M. Canagliflozin Actions on Sodium/Glucose Cotransporters 99 Downloaded from jpet.aspetjournals.org

Fig. 3. 14C-canagliflozin uptake mediated by SGLT2. (A and B) The measurement of 14C-canagliflozin uptake by SGLT2. The uptake of 14C-canagliflozin at ASPET Journals on October 1, 2021 (5, 10, and 20 nM) was measured for 20 minutes in the JTREx-CHO-SGLT2 (SGLT2) and parental JTREx-CHO (Mock) cells with or without Na+ (A) and phlorizin (50 mM) (B). (C) Time course of 14C-canagliflozin uptake by SGLT2. The uptake of 5 nM 14C-canagliflozin was measured in the JTREx-CHO- SGLT2 (SGLT2) and parental JTREx-CHO (Mock) cells in the presence of Na+. (D) Temperature dependence of 14C-canagliflozin uptake by SGLT2. The uptakes of 14C-canagliflozin (5 and 10 nM) (left) and 14C-AMG (500 mM) (right) were measured either on ice or at 37°C for 20 minutes in the presence of Na+ using the Dox-treated JTREx-CHO-SGLT2 (SGLT2) and parental JTREx-CHO (Mock) cells. Data shown depict the mean 6 S.E.M. n =3–6. *p , 0.05; **p , 0.01; ***p , 0.001. doxycycline treatment, a concentration of 10 mM AMG in the Discussion external solution induced an inward current in the JTREx- We performed a kinetic analysis of the inhibitory effect of 293-SGLT1 cells (Fig. 4C). This AMG-induced inward canagliflozin and confirmed its high potency and selectivity for current was reverted to the basal level when either AMG 1 SGLT2 (Fig. 2 and Supplemental Fig. 2). This is the first or Na was washed out from the external solution during report to describe the K values of canagliflozin, which are the measurement (Supplemental Fig. 4). The AMG-induced i useful to evaluate the in vivo action of canagliflozin on SGLT2 current was almost completely suppressed by extracellular application of 20 mM canagliflozin (Fig. 4C). In contrast, and SGLT1 in various organs. Canagliflozin competitively ∼ when the same concentration of canagliflozin was applied inhibited human SGLT1 and SGLT2 with 200-fold higher intracellularly, no significant inhibition of the SGLT1- selectivity for SGLT2 than for SGLT1, based on the ratio of Ki mediated AMG-induced current was observed (Fig. 4E). values (Table 1). Based on the Ki values, canagliflozin has ∼ Similarly, an external solution of 20 mM AMG induced an an 25-fold higher affinity for SGLT2 than for phlorizin, appreciable inward current in the JTREx-293-SGLT2 cells whereas both inhibitors exhibited similar affinity for SGLT1 in the presence of intracellular 200 nM canagliflozin. This (Table 1). These results are in line with those of previous inward current was almost completely reverted to the basal reports, wherein the potency and selectivity of canagliflozin level when the same concentration of canagliflozin was were discussed based on IC50 values (Nomura et al., 2010; extracellularly applied (Fig. 4D). Figure 4F shows the Grempler et al., 2012). comparison of the current densities of the AMG-induced The whole-cell patch-clamp method allowed us to control the current before and after the addition of extracellular composition of intracellular as well as extracellular fluid, canagliflozin, indicating that extracellular canagliflozin enabling the delivery of canagliflozin into the cells for remarkably reduces AMG-induced currents. Taken to- examining the sidedness of the inhibition (Fig. 4). We gether, these results demonstrate that canagliflozin demonstrated that canagliflozin preferentially inhibits preferentially inhibits SGLT1 and SGLT2 from the extra- SGLT2 from the extracellular side and not from the intracel- cellular side. lular side. This suggests that canagliflozin acts on SGLT2 100 Ohgaki et al. Downloaded from jpet.aspetjournals.org at ASPET Journals on October 1, 2021

Fig. 4. Sidedness of canagliflozin action on SGLTs. (A) Schematic representation of whole-cell recording of transporter-mediated AMG-induced electric currents. JTREx-293-SGLT1 and JTREx-293-SGLT2 cells were subjected to whole-cell patch-clamp recording at a holding potential of 260 mV. (B) The current trace of whole-cell recording demonstrating the absence of the background AMG-induced current in JTREx-293-SGLT1 (top) and JTREx-293- SGLT2 (bottom) cells without the induction by doxycycline. AMG (10 mM for SGLT1 and 20 mM for SGLT2) was added to the external solution. The traces of AMG-induced currents showing the inhibitory effect after extracellular application of canagliflozin on SGLT1 (C) and SGLT2 (D). AMG (10 mM for SGLT1 and 20 mM for SGLT2) and canagliflozin (20 mM for SGLT1 and 200 nM for SGLT2) were added to the external solution as indicated. Canagliflozin Actions on Sodium/Glucose Cotransporters 101 from the luminal side because SGLT2 is localized on the apical been reported that SGLT1 is involved in the pathophysiology of membrane of proximal tubules (Wright et al., 2011). Because a murine model of PRKAG2 cardiomyopathy (Ramratnam the plasma protein binding of canagliflozin is ∼99% (Devineni et al., 2014). Because the maximal plasma level of unbound et al., 2015) and the maximal plasma concentration after oral canagliflozin in clinical studies (20–60 nM, see above) is ∼1/30 administration of clinical doses is 2.1–6.14 mM (Kinoshita and to 1/10 of the Ki value for SGLT1 (770.5 nM), the oral Kondo, 2015; Sha et al., 2015), the maximal concentration of administration of clinical doses of canagliflozin would not unbound canagliflozin in plasma and in the glomerular filtrate inhibit SGLT1 in the heart or skeletal muscle. In the is ∼20–60 nM. SGLT2 should be exposed to this approximate concentration range of 20–60 nM, it is estimated that canagli- canagliflozin concentration from the luminal side. This is flozin inhibits SGLT1 by 0.2–0.7% under euglycemic conditions above the determined Ki value of canagliflozin for SGLT2 [100 mg/dl (5.6 mM) D-glucose] and by 0.1–0.2% when the blood (4.0 nM), suggesting that canagliflozin fully inhibits SGLT2 glucose level is elevated to 300 mg/dl (16.7 mM); this is based on from the luminal side. In the concentration range of 20–60 nM, the assumption that the Km for D-glucose is 0.5 mM in SGLT1 canagliflozin is estimated to inhibit SGLT2 by 68–86% (Lu et al., 2014). Similarly, oral administration of canagliflozin under euglycemic conditions [100 mg/dl (5.6 mM) D-glucose] at clinical dosage would not inhibit SGLT1 in the proximal and by 49–74% when the blood glucose level is elevated to straight tubules of the kidney, aiding in the prevention of 300 mg/dl (16.7 mM); this is based on the following equation: hypoglycemia in conjunction with the paradoxical increase in V 5 Vmax/{1 1 [Km/(S)] [1 1 (I)/Ki]}, where V is the uptake rate endogenous glucose production caused by SGLT2 inhibition 5 for D-glucose, Vmax is the maximal uptake rate, Km 4.0 mM (Abdul-Ghani et al., 2013; Cefalu, 2014). Downloaded from for D-glucose (Lu et al., 2014), [S] is the D-glucose concen- Pharmacokinetic studies have revealed that the excretion tration, [I] is the unbound canagliflozin concentration, and of canagliflozin in urine is ,1% of the administered dose Ki 5 4.0 nM. This supports the idea that canagliflozin (Inagaki et al., 2014; Devineni et al., 2015; Kinoshita and inhibits the renal tubular SGLT2 by acting from the luminal Kondo, 2015). We showed that canagliflozin is incorporated side but not from the basolateral side. into the SGLT2-expressing cells by SGLT2-mediated trans- Previous studies demonstrated that the oral administration port in addition to passive diffusion through the plasma jpet.aspetjournals.org of canagliflozin delays postprandial glucose absorption from membrane (Fig. 3). Because the exogenous expression of the intestine and enhances glucose-induced glucagon-like SGLT2 in conventional expression systems is, for an unknown peptide-1 secretion (Polidori et al., 2013; Kinoshita and reason, generally low in efficiency (Kanai et al., 1994), the Kondo, 2015; Sha et al., 2015). Because other SGLT2 inhib- uptake of 14C-canagliflozin by SGLT2 was low (Fig. 3). itors with higher selectivity to SGLT2 have not been reported Therefore, based on the data obtained, it is difficult to estimate to elicit such effects, transient inhibition of SGLT1 in the the contribution of the possible SGLT2-mediated reabsorption upper small intestine seems to be the reason underlying the of canagliflozin to total tubular canagliflozin reabsorption, effect on intestinal glucose absorption (Polidori et al., 2013; with the latter also involving the passive diffusion of the at ASPET Journals on October 1, 2021 Oguma et al., 2015a,b). Like SGLT2, we showed that SGLT1 is compound through the renal tubule because of its high preferentially inhibited by canagliflozin from the extracellular hydrophobicity. We propose that the tubular reabsorption of side, with a Ki value of 770.5 nM. Because of the high plasma canagliflozin involving SGLT2-mediated transport and the protein binding ratio (Devineni et al., 2015), the clinical low filtration rate because of its high plasma protein binding dosage of canagliflozin is higher (100–300 mg) than other contribute to the low urinary excretion of canagliflozin. More- SGLT2 inhibitors. Therefore, it is reasonable to suppose that over, a local circulation of canagliflozin through the apical the luminal concentration of canagliflozin in the upper small membrane of renal proximal tubules may be facilitated by the intestine is transiently elevated after oral administration to a SGLT2-mediated reabsorption of canagliflozin with the pos- level considerably higher than the Ki value for SGLT1, sible involvement of its efflux transporters. This would retain implying that canagliflozin could inhibit SGLT1 from the canagliflozin in the proximity of SGLT2, enabling the re- luminal side of the intestine. The intraluminal concentration current inhibition of SGLT2 from the luminal side, which of canagliflozin in the small intestine is substantially elevated explains the prolonged drug action beyond the decrease in after oral administration (Oguma et al., 2015b). blood canagliflozin levels (Inagaki et al., 2014; Kuriyama In a survey of the expression of SGLTs in human tissues by et al., 2014). The significance of SGLT2-mediated canagliflozin quantitative PCR, SGLT1 was expressed at high levels in the uptake in pharmacodynamics as well as pharmacokinetics in heart and skeletal muscle, in addition to the small intestine vivo remains to be investigated. (Chen et al., 2010). In the heart, SGLT1 mRNA was detected 14C-canagliflozin uptake was not detected for SGLT1 over in cardiomyocytes by in situ hybridization (Zhou et al., 2003), the concentration range tested (Supplemental Fig. 3). Higher and SGLT1 protein was localized to the cardiac myocyte concentrations of 14C-canagliflozin could not to be examined sarcolemma (Banerjee et al., 2009). SGLT1 expression levels because of a high background cellular incorporation. However, were increased in diabetic or ischemic cardiomyopathy considering its lower affinity to SGLT1, it is important to test (Banerjee et al., 2009). Although the functional roles of SGLT1 the uptake of canagliflozin by SGLT1 at a higher concentra- have not been determined in such nonepithelial tissues, it has tion using high-expression systems.

Canagliflozin (200 nM) was also contained in the internal solution for (D). (E) The IAMG/Cm in JTREx-293-SGLT1 cells in the presence or absence of intracellularly applied 20 mM canagliflozin [shown as (+) and (2), respectively]. Data shown depict the mean 6 S.E.M. [n = 11 for (2) and n = 10 for (+)]. (F) Comparison of the current densities for the currents induced by 20 mM AMG before (2) and after (+) the application of extracellular canagliflozin (200 nM) under the same condition as described in (D). Data shown are the mean 6 S.E.M. n =7.*p , 0.05. 102 Ohgaki et al.

In conclusion, canagliflozin inhibited SGLT2 and SGLT1 in patients with type 2 diabetes mellitus and moderate renal impairment. Clin Drug Investig 34:731–742. a competitive manner with high potency and selectivity for Kanai Y, Lee WS, You G, Brown D, and Hediger MA (1994) The human kidney low SGLT2. Canagliflozin acted from the extracellular side, sug- affinity Na1/glucose cotransporter SGLT2. Delineation of the major renal reab- sorptive mechanism for D-glucose. J Clin Invest 93:397–404. gesting that the inhibition occurs from the urinary side in renal Kinoshita S and Kondo K (2015) Evaluation of pharmacokinetic and pharmacody- proximal tubules. We also suggested that canagliflozin is, at namic interactions of canagliflozin and teneligliptin in Japanese healthy male volunteers. Expert Opin Drug Metab Toxicol 11:7–14. least in part, transported by SGLT2 after binding, which may Kuriyama C, Xu JZ, Lee SP, Qi J, Kimata H, Kakimoto T, Nakayama K, Watanabe Y, contribute to its pharmacodynamic and pharmacokinetic Taniuchi N, and Hikida K, et al.. (2014) Analysis of the effect of canagliflozin on renal glucose reabsorption and progression of hyperglycemia in zucker diabetic characteristics. Fatty rats. J Pharmacol Exp Ther 351:423–431. Lavalle-González FJ, Januszewicz A, Davidson J, Tong C, Qiu R, Canovatchel W, Acknowledgments and Meininger G (2013) Efficacy and safety of canagliflozin compared with placebo The authors thank Takanori Kobayashi and Miyuki Kurauchi for and sitagliptin in patients with type 2 diabetes on background metformin mono- therapy: a randomised trial. Diabetologia 56:2582–2592. their technical assistance. Canagliflozin was developed by Mitsubishi Lee WS, Kanai Y, Wells RG, and Hediger MA (1994) The high affinity Na1/glucose Tanabe Pharma Corporation in collaboration with Janssen Research cotransporter. Re-evaluation of function and distribution of expression. J Biol – & Development, LLC. Chem 269:12032 12039. Liang Y, Arakawa K, Ueta K, Matsushita Y, Kuriyama C, Martin T, Du F, Liu Y, Xu J, and Conway B, et al.. (2012) Effect of canagliflozin on renal threshold for glucose, Authorship Contributions glycemia, and body weight in normal and diabetic animal models. PLoS One 7: Participated in research design: Kanai, Ueta, Shiotani, and e30555. Lu Y, Griffen SC, Boulton DW, and Leil TA (2014) Use of systems pharmacology Nagamori modeling to elucidate the operating characteristics of SGLT1 and SGLT2 in renal glucose reabsorption in humans. Front Pharmacol 5:274.

Conducted experiments: Ohgaki, Wei, Yamada, Hara, Ueta, and Downloaded from Kuriyama Nomura S, Sakamaki S, Hongu M, Kawanishi E, Koga Y, Sakamoto T, Yamamoto Y, Ueta K, Kimata H, and Nakayama K, et al.. (2010) Discovery of canagliflozin, a Performed data analysis: Ohgaki and Wei novel C-glucoside with thiophene ring, as sodium-dependent glucose cotransporter Wrote or contributed to the writing of the manuscript: Ohgaki, Wei, 2 inhibitor for the treatment of type 2 diabetes mellitus. J Med Chem 53: – Yamada, Hara, Kanai, Okuda, and Kuriyama 6355 6360. Oguma T, Kuriyama C, Nakayama K, Matsushita Y, Yoshida K, Kiuchi S, Ikenaga Y, Nakamaru Y, Hikida K, and Saito A, et al.. (2015a) The effect of combined treat- References ment with canagliflozin and teneligliptin on glucose intolerance in Zucker diabetic Abdul-Ghani MA, DeFronzo RA, and Norton L (2013) Novel hypothesis to explain fatty rats. J Pharmacol Sci 127:456–461. why SGLT2 inhibitors inhibit only 30-50% of filtered glucose load in humans. Oguma T, Nakayama K, Kuriyama C, Matsushita Y, Yoshida K, Hikida K, Obokata jpet.aspetjournals.org Diabetes 62:3324–3328. N, Tsuda-Tsukimoto M, Saito A, and Arakawa K, et al.. (2015b) Intestinal sodium Banerjee SK, McGaffin KR, Pastor-Soler NM, and Ahmad F (2009) SGLT1 is a novel glucose cotransporter 1 inhibition enhances glucagon-like peptide-1 secretion in cardiac glucose transporter that is perturbed in disease states. Cardiovasc Res 84: normal and diabetic rodents. J Pharmacol Exp Ther 354:279–289. 111–118. Peene B and Benhalima K (2014) Sodium glucose transporter protein 2 inhibitors: Boyle LD and Wilding JP (2013) Emerging sodium/glucose co-transporter 2 inhibitors focusing on the kidney to treat type 2 diabetes. Ther Adv Endocrinol Metab 5: for type 2 diabetes. Expert Opin Emerg Drugs 18:375–391. 124–136. Cefalu WT (2014) Paradoxical insights into whole body metabolic adaptations fol- Polidori D, Sha S, Mudaliar S, Ciaraldi TP, Ghosh A, Vaccaro N, Farrell K, lowing SGLT2 inhibition. J Clin Invest 124:485–487. Rothenberg P, and Henry RR (2013) Canagliflozin lowers postprandial glucose and Chen J, Williams S, Ho S, Loraine H, Hagan D, Whaley JM, and Feder JN (2010) insulin by delaying intestinal glucose absorption in addition to increasing urinary

Quantitative PCR tissue expression profiling of the human SGLT2 gene and re- glucose excretion: results of a randomized, placebo-controlled study. Diabetes Care at ASPET Journals on October 1, 2021 lated family members. Diabetes Ther 1:57–92. 36:2154–2161. Devineni D, Morrow L, Hompesch M, Skee D, Vandebosch A, Murphy J, Ways K, Ramratnam M, Sharma RK, D’Auria S, Lee SJ, Wang D, Huang XYN, and Ahmad F and Schwartz S (2012) Canagliflozin improves glycaemic control over 28 days in (2014) Transgenic knockdown of cardiac sodium/glucose cotransporter 1 (SGLT1) subjects with type 2 diabetes not optimally controlled on insulin. Diabetes Obes attenuates PRKAG2 cardiomyopathy, whereas transgenic overexpression of car- Metab 14:539–545. diac SGLT1 causes pathologic hypertrophy and dysfunction in mice. J Am Heart Devineni D, Curtin CR, Polidori D, Gutierrez MJ, Murphy J, Rusch S, and Rothenberg Assoc 3:e000899. PL (2013) Pharmacokinetics and pharmacodynamics of canagliflozin, a sodium Sha S, Devineni D, Ghosh A, Polidori D, Chien S, Wexler D, Shalayda K, Demarest K, glucose co-transporter 2 inhibitor, in subjects with type 2 diabetes mellitus. J Clin and Rothenberg P (2011) Canagliflozin, a novel inhibitor of sodium glucose co- Pharmacol 53:601–610. transporter 2, dose dependently reduces calculated renal threshold for glucose Devineni D, Curtin CR, Marbury TC, Smith W, Vaccaro N, Wexler D, Vandebosch A, excretion and increases urinary glucose excretion in healthy subjects. Diabetes Rusch S, Stieltjes H, and Wajs E (2015) Effect of hepatic or renal impairment on Obes Metab 13:669–672. the pharmacokinetics of canagliflozin, a sodium glucose co-transporter 2 inhibitor. Sha S, Polidori D, Farrell K, Ghosh A, Natarajan J, Vaccaro N, Pinheiro J, Rothenberg Clin Ther 37:610–628. P, and Plum-Mörschel L (2015) Pharmacodynamic differences between canagliflozin Elkinson S and Scott LJ (2013) Canagliflozin: first global approval. Drugs 73:979–988. and dapagliflozin: results of a randomized, double-blind, crossover study. Diabetes Eskandari S, Wright EM, and Loo DD (2005) Kinetics of the reverse mode of the Obes Metab 17:188–197. 1 Na /glucose cotransporter. JMembrBiol204:23–32. Silverman M (1974) The in vivo localization of high-affinity phlorizin receptors to the Fujita Y and Inagaki N (2014) Renal sodium glucose cotransporter 2 inhibitors as a brush border surface of the proximal tubule in dog kidney. Biochim Biophys Acta novel therapeutic approach to treatment of type 2 diabetes: clinical data and 339:92–102. mechanism of action. J Diabetes Investig 5:265–275. Wiriyasermkul P, Nagamori S, Tominaga H, Oriuchi N, Kaira K, Nakao H, Kitashoji 1 Ghezzi C and Wright EM (2012) Regulation of the human Na -dependent glucose T, Ohgaki R, Tanaka H, and Endou H, et al.. (2012) Transport of 3-fluoro-L- cotransporter hSGLT2. Am J Physiol Cell Physiol 303:C348–C354. a-methyl-tyrosine by tumor-upregulated L-type amino acid transporter 1: a cause Ghezzi C, Hirayama BA, Gorraitz E, Loo DD, Liang Y, and Wright EM (2014) SGLT2 of the tumor uptake in PET. J Nucl Med 53:1253–1261. inhibitors act from the extracellular surface of the cell membrane. Physiol Rep 2: Wright EM, Loo DD, and Hirayama BA (2011) Biology of human sodium glucose e12058. transporters. Physiol Rev 91:733–794. Grempler R, Thomas L, Eckhardt M, Himmelsbach F, Sauer A, Sharp DE, Bakker You G, Lee WS, Barros EJG, Kanai Y, Huo TL, Khawaja S, Wells RG, Nigam RA, Mark M, Klein T, and Eickelmann P (2012) Empagliflozin, a novel selective SK, and Hediger MA (1995) Molecular characteristics of Na(1)-coupled glu- sodium glucose cotransporter-2 (SGLT-2) inhibitor: characterisation and compar- cose transporters in adult and embryonic rat kidney. JBiolChem270: ison with other SGLT-2 inhibitors. Diabetes Obes Metab 14:83–90. 29365–29371. Hediger MA, Coady MJ, Ikeda TS, and Wright EM (1987) Expression cloning and Zhou L, Cryan EV, D’Andrea MR, Belkowski S, Conway BR, and Demarest KT (2003) 1 cDNA sequencing of the Na /glucose co-transporter. Nature 330:379–381. Human cardiomyocytes express high level of Na1/glucose cotransporter 1 (SGLT1). Hummel CS, Lu C, Loo DD, Hirayama BA, Voss AA, and Wright EM (2011) Glucose J Cell Biochem 90:339–346. transport by human renal Na1/D-glucose cotransporters SGLT1 and SGLT2. Am J Physiol Cell Physiol 300:C14–C21. Iijima H, Kifuji T, Maruyama N, and Inagaki N (2015) Pharmacokinetics, pharma- Address correspondence to: Dr. Yoshikatsu Kanai, Department of Bio- codynamics, and safety of canagliflozin in Japanese patients with type 2 diabetes system Pharmacology, Graduate School of Medicine, Osaka University, 2-2 mellitus. Adv Ther 32:768–782. Yamadaoka, Suita, Osaka 565-0871, Japan. E-mail: [email protected]. Inagaki N, Kondo K, Yoshinari T, Ishii M, Sakai M, Kuki H, and Furihata K (2014) osaka-u.ac.jp Pharmacokinetic and pharmacodynamic profiles of canagliflozin in Japanese Interaction of the sodium/glucose cotransporter (SGLT) 2 inhibitor canagliflozin with SGLT1 and SGLT2: inhibition kinetics, sidedness of action, and transporter-associated incorporation accounting for its pharmacodynamic and pharmacokinetic features Ryuichi Ohgaki, Ling Wei, Kazunori Yamada, Taiki Hara, Chiaki Kuriyama, Suguru Okuda, Kiichiro Ueta, Masaharu Shiotani, Shushi Nagamori, Yoshikatsu Kanai Journal of Phrmacology and Experimental Therapeutics (JPET)

Supplemental Figure 1 Protein expression of SGLT1 and SGLT2 in the constructed stable cell lines. The parental JTREx-CHO (Mock) cells and JTREx-CHO-SGLT1 cells (A) or JTREx-CHO-SGLT2 cells (B) were incubated with (Dox+) or without (Dox-) doxycycline and subjected to Western blot analysis using specific antibodies. The bands corresponding to SGLT1 and SGLT2 detected specifically in Dox-treated stable cell lines were indicated with arrows.

Supplemental Figure 2 Kinetic properties of the inhibition of 14C-AMG uptake by phlorizin and canagliflozin. The concentration dependence of SGLT1-mediated 14C-AMG uptake was determined in the presence or absence of 500 nM phlorizin (A) or 1000 nM canagliflozin (E). Their Eadie–Hofstee plots are shown in (B) and (F). The concentration dependence of SGLT2-mediated 14C-AMG uptake was determined in the presence or absence of 80 nM phlorizin (C) or 8 nM canagliflozin (G). The corresponding Eadie–Hofstee plots are shown in (D) and (H). Eadie–Hofstee plots of 14C-AMG uptake in the presence and the absence of inhibitors intersect at the same y-intercept, confirming competitive inhibition. The SGLT1- or SGLT2-mediated uptake was obtained by subtracting 14C-AMG uptake in parental JTREx-CHO cells from that in JTREx-CHO-SGLT1 or JTREx-CHO-SGLT2 cells, respectively. To measure the SGLT1- or SGLT2-mediated uptake, the cells were incubated with the indicated concentration of 14C-AMG for 5 min or 30 min, respectively. For the experiments shown in (A) and (B), the assay buffer contained 0.2% DMSO (final concentration), otherwise DMSO concentration is up to 0.1%. Data shown depict means ± S.E.M. n = 3.

Supplemental Figure 3 The measurement of 14C-canagliflozin uptake by SGLT1. The uptake of 14C-canagliflozin (10 and 20 nM) was measured in the JTREx-CHO-SGLT1 (SGLT1) and parental JTREx-CHO (Mock) cells in the presence or absence of Na+ for 5 min. Data shown depict means ± S.E.M. n = 3–6.

Supplemental Figure 4 Characteristics of AMG-induced currents recorded with whole-cell patch clamp on JTREx-293-SGLT1 (A) and SGLT2 cells (B and C). Whole-cell patch clamp recordings were conducted as described in Figure 4. When indicated with bars above the current traces, cells were perfused with the external solutions with different compositions as follows: To elicit the transporter-mediated current, the indicated concentration of AMG was added to the external solution (10 or 20 mM AMG). Equivalent molar of mannitol instead of AMG was added to the external solution to test the effects of osmotic change (10 or 20 mM Mannitol). Na+ was washed out with Na+-free external solution in which NaCl was substituted with the equivalent concentration of N-methyl-D-glucamine (Na+-free). DMSO, the solvent of cangaliflozin, was added to the external solution at the concentration of 0.5% together with AMG (10 or 20 mM AMG+0.5% DMSO). Canagliflozin was added to the external solution (20 μM or 200 nM Canagliflozin). Reproducible results were obtained from at least three independent measurements. The representative traces were shown in the figure.

Supplemental Figure 5 Replacement of the intracellular fluid with high-sodium internal solution containing canagliflozin. Whole-cell patch clamp recordings in the JTREx-293-SGLT1 (A) and SGLT2 cells (B) were set up as described in Figure 4. Immediately after achieving the whole-cell configuration, the membrane potential was monitored under the current-clamp mode (0 pA). Because the ionic composition of the high sodium internal solution [150 mM NaCl, 11 mM EGTA, and 10 mM HEPES–Tris

(pH 7.4)] is similar to that of the external solution [150 mM NaCl, 1 mM CaCl2, 1 mM

MgCl2, and 10 mM HEPES–Tris (pH 7.4)], replacement of the intracellular fluid would result in loss of the original inside-negative membrane potential (usually ~−60 mV). As expected, the membrane potential promptly shifted toward 0 mV and reached a plateau ~150 s after attaining the whole-cell configuration. Supplemental Figure 1

A B Mock SGLT1 Mock SGLT2 Dox - + - + Dox - + - +

250 250 150 150

100 100 75 75 ➡

➡ 50 50

37 37

25 25

20 20 (kDa) (kDa) Supplemental Figure 2 A B 14 SGLT1 30 SGLT1 12 25

e Phlorizin (-)

k 10 Phlorizin (-) 20 Phlorizin (500 nM) 8 15 6 10 C-AMG upt a 4 14

(nmol/mg protein/min) 5 2 Phlorizin (500 nM) V (nmol/mg protein/min) 0 0 0 200 400 600 800 1000 0.000 0.004 0.008 0.012 0.016 0.020 [AMG] (µM) V (nmol/mg protein/min)/[AMG] (µM) C D 1.5 3.0 SGLT2 SGLT2 2.5

e Phlorizin (-) k 2.0 1.0 Phlorizin (-) Phlorizin (80 nM) 1.5

0.5 1.0 C-AMG upt a 14 0.5 (nmol/mg protein/min) Phlorizin (80 nM) V (nmol/mg protein/min) 0.0 0.0 0 500 1000 1500 2000 0.0000 0.0004 0.0008 0.0012 0.0016 [AMG] (µM) V (nmol/mg protein/min)/[AMG] (µM) E F 20 30 SGLT1 SGLT1 25 Canagliflozin (-) e 15

k Canagliflozin (-) 20 Canagliflozin (1000 nM)

10 15

10 C-AMG upt a

14 5 5 V (nmol/mg protein/min) (nmol/mg protein/min) Canagliflozin (1000 nM) 0 0 0 200 400 600 800 1000 0.00 0.01 0.02 0.03 0.04 0.05 0.06 [AMG] (µM) V (nmol/mg protein/min)/[AMG] (µM) G H 1.5 SGLT2 3.5 SGLT2 3.0 e

k 2.5 Canagliflozin (-) 1.0 Canagliflozin (-) Canagliflozin (8 nM) 2.0 1.5 0.5 C-AMG upt a 1.0 14 (nmol/mg protein/min) V (nmol/mg protein/min) 0.5 Canagliflozin (8 nM) 0.0 0.0 0 500 1000 1500 2000 0.0000 0.0004 0.0008 0.0012 0.0016 [AMG] (µM) V (nmol/mg protein/min)/[AMG] (µM) Supplemental Figure3

14C-canagliflozin uptake (pmol/mg protein) 10 15 20 25 30 0 5 Na SGLT1 M NS + o + ck 10 Canagliflozin (nM) Na NS + - Na NS + + 20 Na NS + - Supplemental Figure 4

A SGLT1

Na+-free 10 mM AMG 10 mM Mannitol 10 mM AMG + 0.5% DMSO 10 mM Mannitol 10 mM AMG 20 µM Canagliflozin

20 pA

1 min

B SGLT2

Na+-free

20 mM AMG 20 mM Mannitol

50 pA

1 min C SGLT2

20 mM Mannitol 20 mM AMG 20 mM AMG + 0.5% DMSO 200 nM Canagliflozin

50 pA

1 min Supplemental Figure 5 A B

0 SGLT1 0 SGLT2

-10 -10

-20 -20

-30 -30 Membrane potential (mV) Membrane potential (mV) -40 -40

-50 -50 0 50 100 150 200 250 300 0 50 100 150 200 250 300 Time (sec) Time (sec) Interaction of the sodium/glucose cotransporter (SGLT) 2 inhibitor canagliflozin with SGLT1 and SGLT2: inhibition kinetics, sidedness of action, and transporter-associated incorporation accounting for its pharmacodynamic and pharmacokinetic features Ryuichi Ohgaki, Ling Wei, Kazunori Yamada, Taiki Hara, Chiaki Kuriyama, Suguru Okuda, Kiichiro Ueta, Masaharu Shiotani, Shushi Nagamori, Yoshikatsu Kanai Journal of Phrmacology and Experimental Therapeutics (JPET)

Supplemental Methods Western blot. Western blot analysis was performed basically as previously described (Khunweeraphong et al., 2012). The parental JTREx-CHO cells, JTREx-CHO-SGLT1 cells and JTREx-CHO-SGLT2 cells were incubated with or without doxycycline for 24 h. Cells were collected in ice-cold phosphate buffered saline (PBS) and lysed by sonication in lysis buffer containing 20 mM Tris-HCl (pH7.5), 150 mM NaCl, 1 mM EDTA and protease inhibitor cocktail (Roche Diagnostics GmBH, Mannheim, Germany). Whole lysate was centrifuged at 1000 × g for 5 min and the supernatant was further centrifuged at 391,000 × g for 15 min to obtain crude membrane fraction. After solubilization with the lysis buffer containing 1% FC-12 and mixing with Laemmli buffer, 50 μg/lane of the crude membrane factions were separated on 10% SDS-PAGE gel. The proteins were transferred electrophoretically onto Hybond-P PVDF membrane (GE Healthcare; Buckinghamshire, UK). Membranes were blocked at room temperature for 1 h with Tris-buffered saline Tween-20 (TBS-T) containing 5 % skim milk and incubated overnight at 4°C with primary antibodies. The following antibodies were used: anti-SGLT1 antibody (Abcam, Cambridge, MA, ab14686), anti-SGLT2 antibody (Santa Cruz Biotechnology, Santa Cruz, CA, sc-47402). The membranes were then washed with TBS-T and incubated at room temperature for 1h with appropriate secondary antibodies conjugated to horseradish peroxidase. Signals were developed by ECL Plus Western Blotting Detection System (GE Healthcare, Piscataway, NJ) and visualized using LAS-4000 mini Ver. 2.0 (Fujifilm Corporation, Tokyo, Japan).

Reference for Supplemental Methods Khunweeraphong N, Nagamori S, Wiriyasermkul P, Nishinaka Y, Wongthai P, Ohgaki R, Tanaka H, Tominaga H, Sakurai H, and Kanai Y (2012) Establishment of stable cell lines with high expression of heterodimers of human 4F2hc and human amino acid transporter LAT1 or LAT2 and delineation of their differential interaction with α-alkyl moieties. J Phamacol Sci 119:368-380.