elts hssuyasse hraooia rpriso empagliflozin of properties pharmacological assessed study This mellitus. ∗ to: Correspondence in reduction a [5]. and kidney, levels urine glucose the the blood via in elimination glucose glucose to reabsorb to shown to leading been capacity has cortex, body’s SGLT-2 the kidney of block Inhibition the [3,4]. tubule segment) in proximal early (S1 the expressed of membrane apical exclusively the cotransporter in family. specifically glucose almost SGLT capacity is high the affinity, that of low member a is a is SGLT-2 [1,2], reabsorption SGLT-2 glucose by of facilitated proportion 2 significant type in A potential therapeutic diabetes. has the inhibition their of such, [1], as mediation kidney and the of through tubules proximal the glycaemia from the glucose of state in reuptake steady involved of are (SGLTs) control cotransporters glucose Sodium Introduction 2011 October 5 acceptance final of date 2011; June 26 decision first of date 2011; May 27 submitted Date Conclusions: Results: Methods: Aims: 4 3 2 1 Grempler inhibitors R. SGLT-2 other with comparison and characterisation inhibitor: (SGLT-2) glucose cotransporter-2 sodium selective novel a Empagliflozin, article original ftetse GT2ihbtr vrhGT1 malflznrpeet ninvtv hrpui praht ra diabetes. treat to approach therapeutic innovative an represents Empagliflozin hSGLT-1. over Keywords: inhibitors SGLT-2 tested the of ohigrIglemGb oK,Bredre tae6,D837Bbrc nder [email protected] an E-mail: Biberach D-88397 65, Straße Germany. Birkendorfer Riss, Co.KG, & GmbH Ingelheim Boehringer rats. eetvt eeietfid malflznpamckntc nZFrt eecaatrsdb oeaettlpam laac C)and (CL) clearance plasma SGLT-1 total in moderate differences Species by high. 6. was characterised and BA and were 5 low 4, rats was SGLT-1, CL ZDF over dogs beagle selectivity in in of pharmacokinetics while degree (BA), Empagliflozin high bioavailability a identified. has were empagliflozin inhibitors, selectivity SGLT-2 other with Compared GT1 n .Tonwcl ie vrepesn SL- n SL- eeetbihdad[ and established were hSGLT-6 and hSGLT-5 over-expressing lines cell new Two 4. and 2 SGLT-1, sasdvlpd idn ieiswr nlsduigardoiadbnigasywt [ with assay binding radioligand a using analysed were kinetics Binding developed. assays ebae.Acute membranes. oec n eetvt ihohrSL- inhibitors. SGLT-2 other with selectivity and potency .Mark M. ohatoscnrbtdequally. contributed authors Both adoeaoi iessRsac,Berne nehi hraGb oK,Bbrc,Germany Biberach, Co.KG, & GmbH Pharma Ingelheim Boehringer Research, Diseases CardioMetabolic rgDsoeySpot ohigrIglemPam mH&C.G ieah Germany Biberach, Co.KG, & GmbH Pharma Ingelheim Boehringer Support, Discovery Drug eiia hmsr,Berne nehi hraGb oK,Bbrc,Germany Biberach, Co.KG, & GmbH Pharma Ingelheim Boehringer Chemistry, Medicinal rgMtbls n hraoieis ohigrIglemPamcuiasIc,Rdeed T USA CT, Ridgefield, Inc., Pharmaceuticals Ingelheim Boehringer Pharmacokinetics, and Metabolism Drug malflzni eetv oimguoectasotr2(GT2 niio nciia eeomn o h ramn ftp diabetes 2 type of treatment the for development clinical in inhibitor (SGLT-2) cotransporter-2 glucose sodium selective a is Empagliflozin malflznhsa IC an has Empagliflozin 1 [ ibts malflzn hoii,SL,SL- niio,tp diabetes 2 type inhibitor, SGLT-2 SGLT, , empagliflozin, diabetes, 14 .Klein T. , ]apamty lcprnsd AG paeeprmnswr efre ihsal ellnsoe-xrsighmn(h) human over-expressing lines cell stable with performed were experiments uptake (AMG) glucopyranoside C]-alpha-methyl malflzni oetadcmeiieSL- niio iha xeln eetvt rfieadtehgetslciiywindow selectivity highest the and profile selectivity excellent an with inhibitor SGLT-2 competitive and potent a is Empagliflozin rRl rmlr eateto adoeaoi iessResearch, Diseases CardioMetabolic of Department Grempler, Rolf Dr 1 ∗ ,L.Thomas nvivo in 1 .Eickelmann P. & seseto hraoieiswspromdwt omgyamcbal osadZce ibtcfty(ZDF) fatty diabetic Zucker and dogs beagle normoglycaemic with performed was pharmacokinetics of assessment 50 1 ∗ f31n o SL-.Isbnigt GT2i opttv ihguoe(aflf prxmtl h). 1 approximately (half-life glucose with competitive is SGLT-2 to binding Its hSGLT-2. for nM 3.1 of ,M.Eckhardt 1 2 .Himmelsbach F. , n/rsdu ebopin GT4hsbe hw to shown been has monosaccharide the renal SGLT-4 in in reabsorption. expressed role sodium are a 6, play and/or and potentially monosaccharide 5 and SGLT-4, kidney of [10]. SGLTs, binding capable glucose other not upon Three sodium is in transports and but expressed transport, muscle, is junctions skeletal neuromuscular SGLT-3 in of and understood. intestine small the well of neurons less are control [8,9]. SGLT-1 the gene in mutations with individuals dehydration in seen severe diarrhoea and by characterised disease a ‘glucose-galactose malabsorption’, in result glucose can of intestine the inhibition from as reabsorption important, is SGLT-1 main over inhibitors SGLT-2 its small SGLT for of the with selectivity consequence, in a lung, absorption As [6,7]. and galactose intestine also and liver glucose is heart, being SGLT-1 intestine, function SGLT-2, the the to of in segment contrast found S3 In the in [5]. tubule located is proximal and similar [3] a glucose has for which affinity SGLT-1, member, family SGLT another by nvitro in h oe fteohrSL aiymmesi glycaemic in members family SGLT other the of roles The reabsorbed is SGLT-2 by up taken not is that glucose Residual 2 ,A.Sauer n hraoiei properties pharmacokinetic and 3 ]lble malflznadHK9-SL- cell HEK293-hSGLT-2 and empagliflozin H]-labelled 3 14 ,D.E.Sharp ]mnoead[ and C]-mannose ibts bst n Metabolism and Obesity Diabetes, 4 © .A Bakker A. R. , nvivo in 14 01BakelPbihn Ltd Publishing Blackwell 2011 ]moioio uptake C]-myo-inositol n oprdits compared and 4 39,2012. 83–90, 14: 1 ,

original article original article DIABETES, OBESITY AND METABOLISM transport mannose, 1,5-anhydro-D-glucitol and fructose in a sodium-dependent manner and is also highly expressed in 160 hSGLT-2 hSGLT-1 the small intestine [11]. SGLT-5 has a similar localisation to 140 hSGLT-4 SGLT-2, being exclusive to the kidney cortex, although its role 120 hSGLT-5 (mannose) 100 in monosaccharide transport is not established [12]. SGLT-6 is hSGLT-6 (myo-inositol) a high-affinity myo-inositol transporter also found in the brain 80 60 [% control] and intestine [13]. C]-AMG-uptake 40 14 The concept of SGLT-2 inhibition has been pursued by [ 20 several pharmaceutical companies, with the development of 0 -11 -10 -9 -8 -7 -6 -5 -4 T-1095 [14] (Tanabe Seiyaku, Osaka, Japan) and later aryl log [M] and heteroaryl O-glucosides, for example, sergliflozin [15] and remogliflozin [16] (Glaxo SmithKline, Brentwood, UK) Figure 1. Potency of empagliflozin for sodium glucose cotransporter and AVE2268 [17] (Sanofi-Aventis, Paris, France). Although (SGLT)-2 and selectivity over SGLT-1, 4, 5 and 6. Inset shows the chemical these compounds established the proof of concept in humans, structure of empagliflozin. Results show inhibition of monosaccharide their pharmacokinetic properties are probable to have pre- uptake by hSGLT-2, 1, 4, 5 and 6, respectively, at different concentrations vented their further clinical development. Today, only C- of empagliflozin. [14C]-AMG was used as substrate for hSGLT-1, 2 and 4, glucoside SGLT-2 inhibitors are in clinical development for the mannose was used as substrate for hSGLT-5 and myo-inositol for hSGLT- treatment of type 2 diabetes, including dapagliflozin [18–20] 6. Data are shown from representative experiments and are expressed as mean % of respective control ± standard error of the mean, where the (Bristol-Myers Squibb, NY, USA/AstraZeneca, London, UK), control value was measured at 10−11 M empagliflozin of each individual canagliflozin [21] (Johnson & Johnson, New Brunswick, NJ, assay. USA), empagliflozin (Boehringer Ingelheim, Ingelheim, Ger- many), ipragliflozin (Astellas, Tokyo, Japan), tofogliflozin (HEK293-hSGLT-2), hSGLT-4 (HEK293-hSGLT-4), hSGLT- (Roche, Basel, Switzerland) and TS-071 (Taisho, Tokyo, 6 (HEK293-hSGLT-2) and mouse (m) SGLT-1 (HEK293- Japan) [22]. mSGLT-1). CHO-K1 cells (ATCC) were stably transfected This manuscript presents the in vitro properties of with vectors for rat (r) SGLT-1 (CHO-rSGLT-1) and rSGLT- the potent and selective competitive SGLT-2 inhibitor 2 (CHO-rSGLT-2). T-REx 293 cells (Invitrogen) were stably empagliflozin in comparison with other SGLT inhibitors and transfected with a vector for hSGLT-5 (T-REx 293-hSGLT-5). its pharmacokinetic profile in rats and dogs. See the Supporting Information for more details.

[14C]-monosaccharide uptake inhibition experiments Materials and Methods In brief, 0.6 μCi [14C]-labelled monosaccharide was added to ◦ Chemicals stable cell lines pre-incubated at 37 C in 200 μluptakebuffer. Empagliflozin (BI 10773; 1-chloro-4-(β-D-glucopyranos-1- Cells were incubated for 60 min (hSGLT-5), 90 min (hSGLT-4) ◦ yl)-2-[4-((S)-tetrahydrofuran-3-yl-oxy)-benzyl]-benzene; or 4 h (hSGLT-2) at 37 C, then washed three times with figure 1) was synthesised at Boehringer Ingelheim Pharma phosphate-buffered saline (PBS) and lysed in 0.1 N NaOH. The GmbH & Co.KG, Biberach, Germany. Empagliflozin can be lysate was mixed with 200 μl MicroScint 40, shaken for 15 min synthesised as described in patent: WO 2005/092877 A1 or in and counted for radioactivity in the TopCount NXT (Canberra WO 2006/120208 A1. The crystalline form of empagliflozin Packard, Schwadorf, Austria). A dose-response curve was fitted is described in WO 2006/117359 A1. [3H]-empagliflozin (32 to an empirical four-parameter model using XL Fit (IDBS, Ci/mmol) was synthesised at Tritec (Teufen, Switzerland). Guildford, UK) to determine the inhibitor concentration at Dapagliflozin, canagliflozin, remogliflozin, sergliflozin and half-maximal response (IC50). See the Supporting Information T-1095A were synthesised at Boehringer Ingelheim Pharma for more details. GmbH & Co.KG, Biberach, Germany. Ipragliflozin was synthe- sised at Mercachem (Nijmegen, The Netherlands). Foetal calf HEK293-hSGLT-2 Membrane preparation serum was from Biological Industries (Kibbutz Beit-Haemek, At 90% confluency, cells were collected in cold PBS without ◦ Israel), Dulbecco’s Minimal Essential Medium from Cambrex Ca2+ and Mg2+ (4 C) and sonificated (Dr Hielscher GmbH, (East Rutherford, NJ, USA), Zeocin from Invitrogen (Carlsbad, UP 50H) for 3 × 30 s on ice; cell debris was removed by ◦ CA, USA), ethylene diamine tetraacetic acid, sodium chloride centrifugation at 33 × g for 5 min at 4 C and supernatants ◦ (NaCl) and sodium hydroxide (NaOH) from Merck (Darm- centrifuged at 18 000 × g for 60 min at 4 C. Supernatant was ◦ stadt, Germany) and alpha-D-glucose and polyethylenimine discarded and membrane pellet was stored at −80 C. Protein from Sigma (St. Louis, MO, USA). Packard Unifilter-96 GF/B content was determined using the BCA Protein Assay Reagent Filterplates, TopSeals, Microscint 20 and UltimaGold were Kit (Bio-Rad, CA, USA). obtained from Perkin Elmer (Waltham, MA, USA). Radioligand binding assays using [3H]-labelled Cell lines empagliflozin HEK293 cells (ATCC) were stably transfected with vec- Membranes (60 μg/well) were assayed in a 10 mM 4- tors for human (h) SGLT-1 (HEK293-hSGLT-1), hSGLT-2 (2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES)

84 Grempler et al. Volume 14 No. 1 January 2012 DIABETES, OBESITY AND METABOLISM original article buffer (pH 7.4) containing 137 mM NaCl in the presence Mean residence time was calculated as AUMC0−∞/AUC0−∞, or absence of 20 mM glucose and indicated concentrations of total plasma clearance (CL) as dose/AUC0−∞,and 3 [ H]-empagliflozin in 96-well plates at room temperature for steady state volume of distribution (Vss) as (Dose × 2 h. Incubations were stopped by rapid filtration through GF/B AUMC0−∞)/(AUC0−∞ × AUC0−∞). Apparent bioavailability Filterplates impregnated with polyethyleneimine 0.5% and pre- (BA) was calculated as (oral AUC0−∞/oral dose)/(intravenous wetted with 0.9% NaCl solution, and washed four times with AUC0−∞/intravenous dose) × 100. Maximum concentration ◦ 0.9% NaCl solution (4 C) using a Harvester Filtermate 96 (Cmax) and time of maximum concentration (tmax)arealso (Perkin Elmer, Waltham, MA, USA). Filterplates were dried reported. Individual and mean pharmacokinetic parameters for 2 h and 50 μl of Microscint 20 was added to each well. were calculated using Kinetica version 4.41 (ThermoScien- Radioactivity retained on the filters was measured using the tific Corp., Philadelphia, PA, USA) or ToxKin™ version 3 TopCount NXT. In parallel, the actual amount of activity used (LogicaCMG ITS AG, Basel, Switzerland). The means and in the assays was determined by adding the same amount of standard deviations (SDs) of the plasma concentrations were [3H]-empagliflozin that was added per well in the radioligand calculated using Microsoft Excel 2002 or ToxKin™. binding studies and 4 ml Ultima Gold Scintilator into 5 ml vials and measuring using a Tricarb 2900TR (Waltham, MA, USA). Statistical analysis Non-specific [3H]-empagliflozin-binding was determined in Results are given as mean ± SEM (standard error of the the presence of 30 μM dapagliflozin. mean) or mean ± SD, as indicated. IC values were calculated Kinetic binding parameters were determined in the presence 50 using regression analysis. For comparisons, unpaired Student’s or absence of 20 mM glucose (see the Supporting Information t-tests were used. Statistical and data analysis was performed for more details). Graphpad Prism 5.0 (Graphpad Software, using GraphPadPrism Software Version 5.02 (GraphPad Soft- Inc., CA, USA) was used for calculating the equilibrium dis- ware, Inc.). sociation constant (Kd) using a nonlinear regressions for a single binding site model, and for calculating the association rate constant (Kon) and the dissociation rate constant values by Results means of a global fitting procedure using the ‘association and Potency and selectivity of empagliflozin for SGLT-2 then dissociation’ nonlinear regression. The potency of empagliflozin to inhibit hSGLTs was analysed Animals in vitro by measuring the uptake of the non-metabolisable Animal procedures were approved by the local animal glucose analogue into cells stably over-expressing hSGLT-1, 14 ethics committee and complied with National Institutes of 2, 4, 5 and 6. Empagliflozin inhibited the uptake of [ C]- Health guidance (Guide for Care and Use of Laboratory alpha-methyl glucopyranoside (AMG) via hSGLT-2 in a Animals, Institute of Laboratory Animal Resources, 1996). dose-dependent manner with an IC50 of 3.1 nM, but was less Male beagle dogs were obtained from a breeding colony at BASF potent for other SGLTs (IC50 range: 1100–11000 nM; Table 1, (Ludwigshafen, Germany). Animals were housed in groups figure 1). [Zucker diabetic fatty (ZDF) rats] or pairs (male beagle dogs) Empagliflozin showed >2500-fold selectivity for hSGLT- in controlled temperature and humidity, with a 12-h light/dark 2 over hSGLT-1 (IC50 8300 nM) and >3500-fold selectivity cycle (lights out between 18:00 and 06:00 hours). Rats were fed over hSGLT-4 (IC50 11000 nM; Table 1). In two novel assays diet number 2437 (Provimi Kliba, Kaiseraugst, Switzerland). for hSGLT-5 and hSGLT-6 (measuring uptake of mannose for Beagle dogs were fed once daily between 11:00 hours and 13:00 hSGLT-5 and myo-inositol for hSGLT-6), empagliflozin exhib- hours with a standard pelleted diet (Diet number 3363, Provimi ited >350-fold selectivity over hSGLT-5 (IC50 1100 nM) and Kliba, Switzerland). Animals had access to water ad libitum. >600-fold selectivity over hSGLT-6 (IC50 2000 nM; Table 1). Animals were used for experiments between 12 and 15 weeks The ability of empagliflozin to inhibit the facilitative glucose of age (rats) and at 37 months (dogs). transporter-1 (GLUT1) [23] was also tested in HEK293 cells using [14C]-deoxyglucose as a substrate in uptake experiments Pharmacokinetics and pharmacodynamics (see the Supporting Information for more details). No relevant of empagliflozin in beagle dogs and ZDF rats inhibition of GLUT1 was observed up to 10 μM empagliflozin (data not shown). Animals were fasted overnight before and for 2 (rats) or 4 (dogs) h after dosing, and serial blood samples taken up to Comparison of potency and selectivity of C-glucoside 24 or 48 h after dosing from rats and dogs, respectively. See the Supporting Information for details of dosing. Pharmacokinetic and O-glucoside SGLT-2 inhibitors parameters were calculated by non-compartmental methods The potency of empagliflozin for SGLT-2 was compared against as follows. Area under the plasma concentration-time curve other SGLT-2 inhibitors. (AUC0−t) to the last quantifiable time point was calculated In general, all C-glucoside SGLT inhibitors (empagliflozin, using the linear trapezoidal method. AUC0−t was extrapo- dapagliflozin, canagliflozin, ipragliflozin and tofogliflozin) lated to infinity (AUC0−∞) using log-linear regression of the exhibited slightly higher potency to inhibit SGLT-2 than with terminal portion of the individual curves to estimate the termi- the O-glucosides (sergliflozin, remogliflozin and phlorizin), nal elimination half-life (t 1/2). Area under the moment curve with the exception of T-1095A. Empagliflozin had the highest (AUMC0−∞) was calculated in a manner similar to AUC0−∞. selectivity for SGLT-2 over SGLT-1 (>2500-fold), followed

Volume 14 No. 1 January 2012 doi:10.1111/j.1463-1326.2011.01517.x 85 original article DIABETES, OBESITY AND METABOLISM

Table 1. Overview of potency on SGLT-2 and selectivity over SGLT-1, 4, 5 and 6 of structurally known SGLT-2 inhibitors.

Compound SGLT-2 (AMG) SGLT-1 (AMG) SGLT-4 (AMG) SGLT-5 (mannose) SGLT-6 (myo-inositol)

C-glucosides IC50 pIC50 ± SEM Empagliflozin 3.1 8300 11000 1100 2000 (BI 10773) 8.50 ± 0.02 5.08 ± 0.03 4.94 ± 0.09 5.98 ± 0.15 5.70 ± 0.08 Dapagliflozin 1.2 1400 9100 820 1300 (BMS-512148) 8.94 ± 0.06 5.86 ± 0.07 5.04 ± 0.12 6.09 ± 0.22 5.88 ± 0.09 Canagliflozin 2.7 710 7900 1700 240 (JNJ-28431754;TA-7284) 8.56 ± 0.02 6.15 ± 0.06 5.10 ± 0.02 5.77 ± 0.12 6.61 ± 0.09 Ipragliflozin 5.3 3000 16000 740 7800 (ASP-1941) 8.27 ± 0.04 5.53 ± 0.02 4.80 ± 0.07 6.13 ± 0.11 5.11 ± 0.06 Tofogliflozin 6.4 12000 14000 3000 n.d. (CSG-452, RG-7201) 8.18 ± 0.12 4.92 ± 0.09 4.84 ± 0.12 5.53 ± 0.25

O-glucosides IC50 pIC50 ± SEM Sergliflozin 7.5 2100 6000 1100 14000 8.12 ± 0.01 5.69 ± 0.11 5.22 ± 0.23 5.95 (n = 1) 4.84 ± 0.08 Remogliflozin 12 6500 1500 190 6200 7.93 ± 0.13 5.19 ± 0.19 5.81 ± 0.24 6.72 ± 0.19 5.21 ± 0.11 T-1095A 4.4 260 2300 1100 3300 8.36 ± 0.08 6.58 ± 0.04 6.63 ± 0.22 5.97 ± 0.19 5.48 ± 0.15 Phlorizin 21 290 6100 1500 10000 7.67 ± 0.03 6.54 ± 0.05 5.22 ± 0.32 5.82 ± 0.18 4.99 ± 0.10

14 Results are shown as mean IC50 [nM] and pIC50 ± SEM for inhibition of hSGLT-2, 1, 4, 5 and 6. [ C]-AMG was used as substrate for hSGLT-1, 2 and 4, [14C]-mannose was used as substrate for hSGLT-5 and [14C]-myo-inositol for hSGLT-6. Data are derived from at least three independent experiments unless indicated otherwise. AMG, alpha-methyl glucopyranoside; n.d., not determined; SEM, standard error of the mean; SGLT, sodium glucose cotransporter. by tofogliflozin (>1875-fold), dapagliflozin (>1200-fold), Table 2. Kinetic binding data of [3H]-empagliflozin for hSGLT-2 in the ipragliflozin (>550-fold) and canagliflozin (>250-fold). All absence or presence of 20 mM glucose. compounds were highly selective for SGLT-2 over SGLT- 4, and all apart from remogliflozin showed >150–600-fold [3H]-empagliflozin selectivity over SGLT-5 (Table 1). Furthermore, all SGLT-2 No glucose 20 mM glucose inhibitors exhibited >500-fold selectivity over SGLT-6, apart from canagliflozin which has less than 90-fold selectivity. Mean SD Mean SD Kd (nM) 57.0 37.3 194 98.7 3 Binding characteristics of [ H]-empagliflozin t1/2 (min) 59 5 62 5 −1 to hSGLT-2 Kon (mol /min) 314747.7 265243 68302.7 29471.0 K (min−1) 0.01181 0.00098 0.01132 0.00101 [3H]-empagliflozin displayed a high affinity for SGLT-2 with off ± ameanKd of 57 ± 37 nM in the absence of glucose in kinetic Data shown are the mean SD of three independent experiments. binding experiments (Table 2, see the Supporting Information Every experiment was performed in triplicate and with three different 3 figure S1 for more detail). High physiological concentrations of concentrations of [ H]-empagliflozin. Kd, equilibrium dissociation glucose (20 mM) slightly lowered the affinity of empagliflozin constant; Kon, association rate constant; Koff , dissociation rate constant; ± SD, standard deviation; SGLT, sodium glucose cotransporter; t1/2, terminal to a mean Kd of 194 99 nM. Kinetic binding experiments elimination half-life. showed a half-life of [3H]-empagliflozin-binding to SGLT- ± 2of59 5 min in the absence of glucose, which was not in humans and mice, at >2500-fold and approximately significantly affected by the presence of 20 mM glucose. 2000-fold, respectively, with lower selectivity for rSGLT-2 However, a significantly lower Kon was observed in the over rSGLT-1 (approximately 60-fold lower; figure 2A). This −1 presence of 20 mM glucose (68302.7 mol /min) vs. no glucose reduced selectivity for SGLT-1 vs. SGLT-2 in rats has also −1 (314747.7 mol /min). These data show the competitive nature been observed with all other tested SGLT-2 inhibitors in of empagliflozin-binding to hSGLT-2 (Table 2). this study and has been previously observed for dapagliflozin and phlorizin [18]. This may be explained by differences in Selectivity of empagliflozin for SGLT-2 vs. SGLT-1 the primary amino acid sequence of human, mouse and across different species rat SGLT-1: alignment between amino acids 551 and 660 Results from [14C]-AMG uptake experiments found that revealed that rSGLT-1 differs from hSGLT-1 and mSGLT-1 empagliflozin was highly selective for SGLT-2 over SGLT-1 in several residues, of which three are responsible for glucose

86 Grempler et al. Volume 14 No. 1 January 2012 DIABETES, OBESITY AND METABOLISM original article

120 Table 3. Pharmacokinetic parameters of empagliflozin in ZDF rat and A Rat SGLT-1 dog. 100 Human SGLT-1 Mouse SGLT-1 80 Rat Dog 60 Dose (mg/kg), iv 1 0.5 CL (ml/min/kg), iv 43 ± 2.71.8 ± 0.2

[% control] [% 40 C]-AMG-uptake ± ± 14 Vss (l/kg), iv 1.4 0.10.6 0.1 [ 20 Dose (mg/kg), po 3 5 ± ± 0 Cmax (nM) 167 82.9 17254 1557 ∗ -10 -9 -8 -7 -6 -5 -4 -3 tmax (h) 2 (2–3) 1 (0.7–1) log empagliflozin [M] AUC0−∞ (nM*h) 858 ± 200 93560 ± 9304 t1/2 (h) 1.5 ± 0.56.3 ± 2.1 B 140 BA (%) 33.2 ± 7.789.0 ± 3.9 0.5 mM glucose 120 2 mM glucose The pharmacokinetic parameters after a single intravenous or oral 100 administration of empagliflozin to ZDF rats and beagle dogs are depicted 10 mM glucose 80 as mean ± SD (n = 4). AUC, area under the plasma concentration- time curve; BA, bioavailability; CL, total plasma clearance; C , 60 max maximum concentration; SD, standard deviation; tmax, time of maximum [% control] C]-AMG-uptake 40 concentration; Vss, steady state volume of distribution; ZDF, Zucker 14 [ 20 diabetic fatty; t1/2, terminal elimination half-life. ∗Median and range. 0 -9 -8 -7 -6 -5 -4 -3 log empagliflozin [M] measured 24 h after administration of 5 mg/kg empagliflozin (data not shown). Despite only moderate CL and BA of Figure 2. Inhibition of sodium glucose cotransporter (SGLT)-1 empagliflozin in rats, acceptable exposure was also achieved by empagliflozin. (A) Species differences in SGLT-1-selectivity of (Table 3). Thus, in beagle dogs and ZDF rats, plasma 14 empagliflozin. Results show inhibition of [ C]-AMG uptake of human, concentrations well above the SGLT-2 IC can be achieved rat and mouse SGLT-1 at different concentrations of empagliflozin. Data 50 are shown from one representative experiment and are expressed as mean with low oral doses of empagliflozin. % control ± standard error of the mean (SEM), where the control value was measured at 10−9 M empagliflozin of each individual assay. Selectivity overhumanandmouseSGLT-1ismorethan20-foldgreaterthanoverrat Discussion SGLT-1. (B) Inhibition of rat SGLT-1 by empagliflozin in the presence of Empagliflozin is a potent and highly selective inhibitor physiological glucose concentrations. Results show inhibition of rat SGLT- of SGLT-2 that is in development for the treatment of 1-mediated [14C]-AMG uptake by empagliflozin in the presence of glucose concentrations up to 10 mM. Data are shown from one representative type 2 diabetes [25–27]. The pharmacologic characteristics experiment and are expressed as mean % control values ± SEM (all values of this compound are described here in detail for the first have been normalised to the value at 10−8 M empagliflozin at 0.5 mM time. glucose). Radioligand binding experiments were employed to analyse the binding characteristics of SGLT-2 inhibitors to hSGLT-2. binding [24]. To evaluate whether this phenomenon could In these experiments, we showed that binding of empagliflozin impact on the interpretation of rat experiments using SGLT-2 to hSGLT-2 is competitive to glucose, although even very inhibitors, [14C]-AMG uptake experiments in the presence high physiological concentrations of glucose (20 mM) have of physiological glucose concentrations up to 10 mM were only relatively small effects on the affinity of empagliflozin for performed. These experiments revealed a right shift in the SGLT-2. The half-life of the empagliflozin-SGLT-2 complex AMG uptake inhibition curve with empagliflozin, suggesting was approximately 60 min, which may explain a prolonged inhibition of SGLT-2 even when plasma concentrations an increase in IC50 for rSGLT-1 inhibition, with increasing of empagliflozin are below the IC . Furthermore, the glucose concentrations (figure 2B). At 10 mM glucose the IC50 50 was shifted to approximately 15 μM and only partial inhibition kidney concentration of another SGLT-2 inhibitor (TS- was achieved by concentrations up to 100 μM empagliflozin 071) was reported to be approximately 10–30-fold above (figure 2B). This shows the competitive nature of SGLT-1 the actual plasma concentration up to 4 h after compound inhibition and indicates that inhibition of SGLT-1 by SGLT administration [28], suggesting that potent inhibition of SGLT- inhibitors is dependent upon the glucose concentration in the 2 in the kidney may be expected even when the SGLT-2 inhibitor proximal tubules or intestine. is already cleared from plasma. For the first time, in this study, the in vitro potency of several SGLT-2 inhibitors, and their selectivity against other family Pharmacokinetics of empagliflozin in ZDF rats and members, namely SGLT-1, 4, 5 and 6, were compared. SGLT-2 beagle dogs inhibitors fall into two classes: C-glucosides (empagliflozin, Pharmacokinetic parameters in rats and dogs are summarised dapagliflozin, canagliflozin, ipragliflozin and tofogliflozin) in Table 3. High exposure of empagliflozin was achieved and O-glucosides (sergliflozin, remogliflozin, T-1095A and in dogs, with plasma concentrations >100-fold above IC50 phlorizin).

Volume 14 No. 1 January 2012 doi:10.1111/j.1463-1326.2011.01517.x 87 original article DIABETES, OBESITY AND METABOLISM

Despite the fact that all SGLT inhibitors are structurally When compared between species, several of the SGLT-2 relatively similar, the compounds differ in their respective inhibitors evaluated showed reduced selectivity over rSGLT-1 selectivity profiles. Empagliflozin was shown to have the highest compared with human and mouse SGLT-1, which may selectivity of the inhibitors tested for hSGLT-2 over hSGLT-1. be explained by differences in conserved amino acids in This is highly desirable, as SGLT-1 has an important role in the C-terminus that are involved in phlorizin-binding of normal intestinal glucose absorption, and thus its inhibition SGLT-1 [23]. Inhibition of SGLT-1 was shown to be dependent may lead to diarrhoea and severe dehydration, symptoms on the glucose concentration at the site of action (e.g. observed in people with inherited mutations in the SGLT-1 proximal tubules or intestine) and the local concentration gene [8,9]. An inhibitor with high specificity for SGLT-2 over of the inhibitor. On the basis of the linear pharmacokinetics SGLT-1 may have better gastrointestinal tolerability properties of this compound (data not shown), the Cmax after 10 mg/kg than one with low specificity. Clinical studies will determine empagliflozin can be estimated to be approximately 500 nM; whether empagliflozin does indeed have fewer gastrointestinal thus even if a 10-fold higher concentration in the kidney is side effects than other less-selective SGLT-2 inhibitors. SGLT-1 assumed, this would be below the apparent Ki (inhibition is also highly expressed in the heart [29], thus its inhibition constant) of 8500 nM of empagliflozin for rSGLT-1. Therefore, by non-selective compounds may impact on cardiac glucose at physiological glucose concentrations, no relevant inhibition uptake. of human, mouse or rat SGLT-1 by pharmacological doses of All SGLT inhibitors investigated in this study were shown empagliflozin is expected. to be highly selective over SGLT-4, which, like SGLT-1, is In a parallel study [34], empagliflozin was shown to expressed in both kidney and intestine, and is involved in effectively increase urinary glucose excretion and reduce intestinal monosaccharide uptake [11]. blood glucose levels in diabetic rats. These findings have SGLT-5 is a new member of the SGLT family, which, also been shown with empagliflozin in patients with type 2 like SGLT-2, is expressed exclusively in the kidney [30]. Its diabetes [26,35]. While inhibition of glucose reabsorption is an function has not been previously identified. In this study, effective means of removing excess glucose from the body, this hSGLT-5 was identified as primarily a mannose transporter, mechanism may result in slightly increased urine volume and but was also shown to be able to transport fructose and, to pollakiuria, and may increase the risk of genital and urinary a lesser extent, AMG (data not shown). Together, these new tract infections [35–37]. findings indicate that SGLT-5 could be involved in the renal Overall, these experiments show that empagliflozin is a reabsorption of mannose and/or fructose as well as sodium potent, competitive SGLT-2 inhibitor with high selectivity over alongside SGLT-4. However, as the physiological relevance of SGLT-1, 4, 5 and 6, and has potential as a treatment for type 2 SGLT-5 in mannose and/or fructose homeostasis is not clear, a diabetes. Phase III clinical trials are ongoing to evaluate the high degree of selectivity is desirable for SGLT-2 inhibitors in safety and efficacy of empagliflozin in type 2 diabetes patients. clinical development. SGLT-6 is important for myo-inositol uptake from the intestine and also for the reabsorption of myo-inositol Acknowledgements from the primary filtrate in the kidney. Myo-inositol is a We thank Dr Ralf Kiesling, Dr Claudia Heine, Dr Bodo stereoisomer of inositol, which is involved in cell signalling via Betzemeier and Dr Thorsten Lehmann-Linz for their help in inositol phosphate pathways [31]. Preliminary data suggest synthesis of reference compounds or labelling of empagliflozin. that inositol/myo-inositol may play important roles in We acknowledge the excellent technical assistance of Verena psychiatric conditions, for example, depression, bulimia or Zell, Annette Halder, Nicola Zimmermann, Martin Steiner, panic disorder [32,33]. As the physiological function of SGLT- Stefanie Eisele, Petra Veit, Stefan Weigele, Monika Krauth, 6 is not fully understood, a high degree of selectivity seems to be Sabrina Hummel, Peter Kaptein and Rebecca Janek. We thank desirable; and empagliflozin has shown >600-fold selectivity Andrea Lorenz for the excellent performance of radioligand over SGLT-6. Other compounds were also selective over SGLT- binding experiments. We thank Tetsuo Seki and Seiichiro 6 with the exception of canagliflozin, which was a relatively Nishimura for providing the HEK293-hSGLT-1 and -hSGLT-2 potent inhibitor of SGLT-6 (IC50 240 nM). cell lines. We thank Dr Robert Augustin for critical reading of Selectivity of empagliflozin over SGLT-3 was not analysed the manuscript and excellent discussions about GLUT biology. as SGLT-3 does not transport monosaccharides, and an assay We acknowledge Dr Stefanie Frohner,¨ who performed the for SGLT-3-binding was not available. Structural similarities initial experiments on hSGLT-5. The authors acknowledge the between SGLT-1 and SGLT-3 [10] suggest that empagliflozin editorial assistance of Stephanie Lockett and Lindsay Napier, may also have a high degree of selectivity over SGLT-3; however, Fleishman-Hillard Group Limited, London, UK, whose services further investigation is required to support this speculation. In were funded by Boehringer Ingelheim. addition, empagliflozin was shown to be highly selective over the facilitative GLUT1 (data not shown), which is responsible for glucose uptake in almost all tissues [23]. Owing to the Conflict of Interest similarity of GLUT1 to other GLUT family members, no At the time of completing the studies, all authors were inhibition of GLUTs is anticipated for empagliflozin; however, employees of Boehringer Ingelheim Pharma GmbH & Co.KG further investigation is required to elucidate the drug effects on or Boehringer Ingelheim Pharmaceuticals Inc. as indicated. All other GLUTs. authors contributed substantially to the concept and design

88 Grempler et al. Volume 14 No. 1 January 2012 DIABETES, OBESITY AND METABOLISM original article and/or to acquisition, analysis and interpretation of data. R.G., 13. Aouameur R, Da CS, Bissonnette P, Coady MJ, Lapointe JY. SMIT2 mediates T.K. and L.T. drafted the article, and the article was revised and all myo-inositol uptake in apical membranes of rat small intestine. Am the final version approved by all authors. J Physiol Gastrointest Liver Physiol 2007; 293: G1300–G1307. Specifically, R.G. and L.T. were responsible for conception, 14. Oku A, Ueta K, Arakawa K et al. T-1095, an inhibitor of renal Na+-glucose design, analysis and interpretation of data regarding compar- cotransporters, may provide a novel approach to treating diabetes. ison of potency and selectivity of SGLT-2 inhibitors, species Diabetes 1999; 48: 1794–1800. selectivity of empagliflozin and for conception and design of 15. Fujimori Y, Katsuno K, Ojima K et al. Sergliflozin , a selective radioligand binding experiments. SGLT2 inhibitor, improves glycemic control in streptozotocin-induced diabetic rats and Zucker fatty rats. Eur J Pharmacol 2009; 609: 148–154. A.S. and D.S. were responsible for the design, acquisition of data, analysis and interpretation of the pharmacokinetic data 16. Fujimori Y, Katsuno K, Nakashima I, Ishikawa-Takemura Y, Fujikura H, of ZDF rats or beagle dogs, respectively. R.B. was responsible Isaji M. Remogliflozin etabonate, in a novel category of selective low-affinity sodium glucose cotransporter (SGLT2) inhibitors, exhibits for the acquisition, analysis and interpretation of radioligand antidiabetic efficacy in rodent models. J Pharmacol Exp Ther 2008; 327: binding data. 268–276. M.E. and F.H. designed and synthesised empagliflozin. 17. Bickel M, Brummerhop H, Frick W et al. Effects of AVE2268, a substituted M.M., T.K. and P.E. contributed to the concept and design glycopyranoside, on urinary glucose excretion and blood glucose in mice of the article. and rats. Arzneimittelforschung 2008; 58: 574–580. 18. Han S, Hagan DL, Taylor JR et al. Dapagliflozin, a selective SGLT2 inhibitor, Supporting Information improves glucose homeostasis in normal and diabetic rats. Diabetes 2008; 57: 1723–1729. Additional Supporting Information may be found in the online version of this article: 19. Meng W, Ellsworth BA, Nirschl AA et al. Discovery of dapagliflozin: a potent, selective renal sodium-dependent glucose cotransporter 2 (SGLT2) Figure S1. Kinetic binding of empagliflozin in the absence inhibitor for the treatment of type 2 diabetes. J Med Chem 2008; 51: (A) and presence (B) of 20 mM glucose. 1145–1149. Appendix S1. Materials and Methods. 20. Obermeier M, Yao M, Khanna A et al. In vitro characterization and Please note: Wiley-Blackwell is not responsible for the pharmacokinetics of dapagliflozin (BMS-512148), a potent sodium-glucose content or functionality of any supporting materials supplied cotransporter type II inhibitor, in animals and humans. Drug Metab Dispos by the authors. Any queries (other than missing material) 2010; 38: 405–414. should be directed to the corresponding author for the article. 21. Nomura S, Sakamaki S, Hongu M et al. Discovery of canagliflozin, a novel C-glucoside with thiophene ring, as sodium-dependent glucose References cotransporter 2 inhibitor for the treatment of type 2 diabetes mellitus. J Med Chem 2010; 53: 6355–6360. 1. Ghosh RK, Ghosh SM, Chawla S, Jasdanwala SA. SGLT2 inhibitors: a new emerging therapeutic class in the treatment of type 2 diabetes mellitus. 22. Nair S, Wilding JP. Sodium glucose cotransporter 2 inhibitors as a new J Clin Pharmacol 2011 [Epub ahead of print]. treatment for diabetes mellitus. J Clin Endocrinol Metab 2010; 95: 34–42. 2. Ferrannini E. Learning from glycosuria. Diabetes 2011; 60: 695–696. 23. Augustin R. The protein family of glucose transport facilitators: It’s not only about glucose after all. IUBMB Life 2010; 62: 315–333. 3. Hummel CS, Lu C, Loo DD, Hirayama BA, Voss AA, Wright EM. Glucose + transport by human renal Na+/D-glucose cotransporters SGLT1 and SGLT2. 24. Althoff T, Hentschel H, Luig J, Schutz H, Kasch M, Kinne RK. Na( )-D- Am J Physiol Cell Physiol 2011; 300: C14–C21. glucose cotransporter in the kidney of Squalus acanthias: molecular identification and intrarenal distribution. Am J Physiol Regul Integr Comp 4. Kanai Y, Lee WS, You G, Brown D, Hediger MA. The human kidney low + Physiol 2006; 290: R1094–R1104. affinity Na /glucose cotransporter SGLT2. Delineation of the major renal reabsorptive mechanism for D-glucose. J Clin Invest 1994; 93: 397–404. 25. Grempler R, Thomas L, Eckhardt M et al. In vitro properties and in vivo effect on urinary glucose excretion of BI 10773, a novel selective SGLT-2 5. Isaji M. Sodium-glucose cotransporter inhibitors for diabetes. Curr Opin inhibitor. Diabetes 2009; 58: A521–P. Investig Drugs 2007; 8: 285–292. 26. Port A, Macha S, Seman L et al. Safety, tolerability, pharmacokinetics and 6. Wright EM, Loo DD, Panayotova-Heiermann M et al. ’Active’ sugar pharmacodynamics of BI10773, a sodium-glucose co-transporter inhibitor transport in eukaryotes. J Exp Biol 1994; 196: 197–212. (SGLT-2), in healthy volunteers. Diabetes 2010; 59: A155 [569–P]. 7. Wright EM, Hirayama BA, Loo DF. Active sugar transport in health and disease. J Intern Med 2007; 261: 32–43. 27. Ferrannini E, Seman L, Seewaldt-Becker E, Hantel S, Pinnetti S, Woerle HJ. The potent and highly selective sodium glucose cotransporter-2 8. Meeuwisse GW. Glucose-galactose malabsorption. Studies on renal (SGLT-2) inhibitor BI 10773 is safe and efficacious as monotherapy glucosuria. Helv Paediatr Acta 1970; 25: 13–24. in patients with type 2 diabetes mellitus. Diabetologia 2010; 53: 9. Turk E, Zabel B, Mundlos S, Dyer J, Wright EM. Glucose/galactose malab- S531 [877]. + sorption caused by a defect in the Na /glucose cotransporter. Nature 28. Kakinuma H, Oi T, Hashimoto-Tsuchiya Y et al. (1S)-1,5-anhydro-1-[5-(4- 1991; 350: 354–356. ethoxybenzyl)-2-methoxy-4-methylphenyl]-1-thio-D-glucito l (TS-071) is 10. Bianchi L, Diez-Sampedro A. A single amino acid change converts the a potent, selective sodium-dependent glucose cotransporter 2 (SGLT2) sugar sensor SGLT3 into a sugar transporter. PLoS One 2010; 5: e10241. inhibitor for type 2 diabetes treatment. J Med Chem 2010; 53: 3247–3261. 11. Tazawa S, Yamato T, Fujikura H et al. SLC5A9/SGLT4, a new Na+- 29. Zhou L, Cryan EV, D’Andrea MR, Belkowski S, Conway BR, Demarest KT. dependent , is an essential transporter for mannose, Human cardiomyocytes express high level of Na+/glucose cotransporter 1,5-anhydro-D-glucitol, and fructose. Life Sci 2005; 76: 1039–1050. 1 (SGLT1). J Cell Biochem 2003; 90: 339–346. 12. Diez-Sampedro A, Hirayama BA, Osswald C et al. A glucose sensor hiding 30. Leicht S, Grohmann S, Page K, Streicher R, Mark M, Eickelmann P. hSGLT5 in a family of transporters. Proc Natl Acad Sci U S A 2003; 100: (SLC5A10) is a sodium dependent sugar transporter exclusively expressed 11753–11758. in the kidney. Diabetes 2006; 55: 1512–P.

Volume 14 No. 1 January 2012 doi:10.1111/j.1463-1326.2011.01517.x 89 original article DIABETES, OBESITY AND METABOLISM

31. Coady MJ, Wallendorff B, Gagnon DG, Lapointe JY. Identification of a novel 35. Rosenstock J, Jelaska A, Seman L, Pinnetti S, Hantel S, Woerle HJ. Efficacy Na+/myo-inositol cotransporter. J Biol Chem 2002; 277: 35219–35224. and safety of BI 10773, a new sodium glucose cotransporter (SGLT-2) 32. Belmaker RH, Bersudsky Y, Benjamin J, Agam G, Levine J, Kofman O. inhibitor, in type 2 diabetes inadequately controlled on . Manipulation of inositol-linked second messenger systems as a therapeutic Diabetes 2011; 60 A271 [989–P]. strategy in psychiatry. Adv Biochem Psychopharmacol 1995; 49:67–84. 36. Wilding JP, Norwood P, T’joen C, Bastien A, List JF, Fiedorek FT. A study 33. Palatnik A, Frolov K, Fux M, Benjamin J. Double-blind, controlled, crossover of dapagliflozin in patients with type 2 diabetes receiving high doses of trial of inositol versus fluvoxamine for the treatment of panic disorder. plus insulin sensitizers: applicability of a novel insulin-independent J Clin Psychopharmacol 2001; 21: 335–339. treatment. Diabetes Care 2009; 32: 1656–1662. 34. Thomas L, Grempler R, Eckhardt M et al. Long-term treatment with 37. Rosenstock J, Polidori D, Zhao Y et al. Canagliflozin, an inhibitor of sodium empagliflozin, a novel, potent and selective SGLT-2 inhibitor, improves glucose co-transporter 2, improves glycaemic control, lowers body weight, glycaemic control and features of metabolic syndrome in diabetic rats. and improves beta cell function in subjects with type 2 diabetes on Diabetes Obes Metab 2011 [Epub ahead of print]. background metformin. Diabetologia 2010; 53: S531 [873].

90 Grempler et al. Volume 14 No. 1 January 2012