New Approach to Drug Discovery of a Safe Mitochondrial Uncoupler: OPC-163493.

We serendipitously found a mitochondrial uncoupler (mUncoupler), compound 1, in the process of screening for inhibitors of a gene product related to calorie restriction (CR) and longevity. Compound 1 has a unique 4-cyano-1,2,3-triazole structure which is different from any known mUncoupler and ameliorated HbA1c in Zucker diabetic fatty (ZDF) rats. However, its administration at high doses was not tolerated in an acute toxicity test in rats. We therefore tried to optimize cyanotriazole compound 1 and convert it into an agent that could be safely administered to patients with diabetes mellitus (DM) or metabolic disorders. Considering pharmacokinetic (PK) profiles, especially organ distribution targeting the liver and avoiding the brain, as well as acute toxicities and pharmacological effects of the derivatives, various conversions and substitutions at the 5-position on the cyanotriazole ring were carried out. These optimizing processes improved PK profiles and effectiveness, and acute toxicities became negligible even at high doses. We finally succeeded in developing an optimized compound, OPC-163493, as a liver-localized/targeted mUncoupler.

Introduction

Obesity, nonalcoholic fatty liver disease (NAFLD), and insulin resistance (IR) associated with visceral, hepatic, or ectopic fat are major risk factors for a number of chronic diseases including diabetes mellitus (DM), cardiovascular diseases, and cancer.[1−3] In particular, the global pandemic of DM is a serious public health emergency.[4] These metabolic disorders are intrinsically involved in an energy imbalance between energy expenditure and calorie intake. An appropriate degree of calorie restriction (CR) ameliorates these disorders, and moreover, it is the only proven way to extend lifespan in mammals like rodents.[5] It has already been shown in both primates and rodents that CR improves health, decreases age-related mortality, and extends lifespan.[6]

Rogina et al. found a single gene mutation that doubled the lifespan of fruit flies and named it Indy (for “I’m not dead yet”). Since Indy is closely related to a sodium dicarboxylate cotransporter (NaDC) which transports TCA cycle intermediates into the cell, the authors speculated that fruit flies with the Indy mutation could not absorb adequate nutrition, and consequently CR-like effects were observed.[7] Inoue et al. cloned a new gene similar to NaDC, a sodium-dependent citrate transporter (NaCT/SLC13A5), as a candidate for the human Indy homologue,[8] which differed from two already known genes, NaDC1 and NaDC3. Since we also found it independently and named it HepNaDC,[9] we established constitutively NaCT-expressing CHO cell clones to obtain CR mimetics. We found 4-cyano-1,2,3-triazole derivative 1 in the hit compounds, which reduced the intracellular radioactive count; however, it did not work as a NaCT inhibitor. Later we discovered that compound 1 had mitochondrial uncoupling (mUncoupling) activity that induced TCA cycle activation. In short, the compound vigorously converted [14C]-citrate into [14C]-carbon dioxide in the TCA cycle and volatilized it. Consequently, the intracellular radioactive count was reduced, as if it inhibited radioactive citrate uptake via NaCT. This activity reliably disappeared in the presence of antimycin A, a mitochondrial respiratory chain complex III inhibitor.

The mUncoupler, 2,4-dinitrophenol (DNP), was widely used as a weight-loss agent in the 1930s; however, its use was accompanied by many severe adverse effects including hyperthermia, cataracts, agranulocytosis, and even death. These effects were ascribed to the narrow therapeutic window, and finally the FDA banned its use in 1938.[10,11] Since then, the use of chemical mUncouplers has been confined to their use as reagents for basic research. Nevertheless, there has been a revival in interest in their therapeutic applications,[12,13] and attempts to discover safe chemical uncouplers have been made[14−19] due to their energy-consuming benefit. In particular, liver-targeted mUncoupling proposed by Shulman and colleagues[20,21] is a promising means of an efficacious and safe treatment for DM and hepatic steatosis.

Cyanotriazole derivative 1 has a unique chemical structure different from any known mUncouplers. We attempted to optimize it as a safe therapeutic option for metabolic disorders such as DM. In this optimization, we regarded safety as the most important factor; therefore, we primarily assessed organ distribution and the acute toxicity of the compounds as well as antidiabetic efficacy. We describe here the optimization process from hit compound 1 to a liver-localized mUncoupler OPC-163493, which recently demonstrated its potent antidiabetic and cardiovascular beneficial effects with acceptable safety.[22]

Results

Hit compound 1 having a cyanotriazole structure was identified from high-throughput screening using NaCT-expressing CHO cells. Its EC50 in the TCA activation assay was 0.34 μM, and the activity was completely abolished by treatment with antimycin A (Figure a). Moreover, compound 1 increased the oxygen consumption rate (OCR) in the presence of oligomycin, an ATP synthase inhibitor, in cultured human liver carcinoma Hep G2 cells, indicating that 1 had mUncoupling activity (Figure b). Compound 1 significantly increased the OCR in the concentration range from 1.25 to 10 μM. Carbonyl cyanide p-trifluoromethoxyphenylhydrazone (FCCP) and DNP also significantly increased the OCR at 0.25 μM and 40 μM, respectively (Figure c). Subsequently, the TCA cycle activation assay was used for in vitro evaluation of other cyanotriazole derivatives in the process of optimization.

Figure 1

Biological profile of compound 1. (a) TCA cycle activation activity of compound 1. The activity of the compound was determined as the reduction of radioactivity accompanied by metabolism of [14C]-labeled citric acid which is taken up into the NaCT-expressing CHO cells. Open and filled circles indicate % of residual intracellular radioactivity in the presence and absence of antimycin A, a mitochondrial complex III inhibitor. (b) Oxygen consumption rate (OCR) in cultured human liver carcinoma HepG2 cells as measured by an extracellular flux analyzer. ΔOCR evoked by compound 1 was defined as its mUncoupling activity. (c) mUncoupling activity of 1 in HepG2 cells. ΔOCR is plotted against compound 1 concentrations (filled circle). Reference compounds, FCCP (filled triangle) and DNP (open triangle), were measured at 0.25 μM and 40 μM, respectively. Data represent mean ± SE (n = 8) **P < 0.01, OPC, FCCP, or DNP vs DMSO-treated group (open circle). The evaluations of compound 1 and reference compounds were independently performed by one-way ANOVA followed by a two-tailed Dunnett’s test. (d) HbA1c-lowering effects of compound 1 in ZDF rats. Effect of compound 1 on HbA1c changes from baseline was determined in a one-week oral dosing study (BID). Data represent mean ± SE (n = 8). Significant efficacy was found in the group receiving 200 mg/kg/day (*P < 0.05, compared with the vehicle group using Dunnett’s test with one-way ANOVA).

A PK study of compound 1 showed that it was well absorbed orally; however, distribution to the liver, the supposed target organ, was lower than the plasma concentration (Kp value of 0.5 and 0.6 at 2 and 8 h after dosing, respectively) (Table ). To evaluate the antidiabetic efficacy of compound 1 in a model animal of type 2 DM, we conducted an oral dosing study in male Zucker diabetic fatty (ZDF) rats.[23−25] One-week dosing of compound 1 significantly suppressed HbA1c elevation by 0.73% at 200 mg/kg/day (Figure d). To monitor whether hypoglycemia occurred, we measured blood glucose in a single-dose toxicity study; however, two out of three animals in the 1000 mg/kg dosing group of normal rats had unexpectedly elevated blood glucose levels exceeding 150 mg/dL (Figure b), and both animals with hyperglycemia died after blood glucose measurement at 4 and 8 h. Accordingly, we aimed at finding cyanotriazole derivatives that exhibit high antidiabetic efficacy without any adverse effects.

Table 1

Chemical and Biological Properties of Cyanotriazole Derivatives

log P and pKa were measured by PION Sirius T3.[35]

Each value represents the mean of 2 animals.

Detection limit: 0.01 μg/mL (plasma), 0.04 μg/g (liver and brain).

Liver/plasma concentration ratio for each compound.

Fatal cases after exhibiting hyperglycemia were observed within 8 h of administration.

Results from 6-week dosing study.

Figure 3

Impact of high doses of cyanotriazole compounds on rectal temperature (a) and blood glucose levels (b) in SD rats at 8 h after dosing. (a) Rectal temperatures at 8 h after dosing of compound 3a and 3b are shown as individual value plots (n = 2–4). (b) Blood glucose levels at 8 h after dosing are presented as individual value plots. The blood glucose levels of animals that died within 8 h after dosing were plotted by the last measured values before death and are represented by gray-filled rectangles (2 h after dosing) and gray-filled triangles (4 h after dosing), respectively (n = 2–4).

Several key cyanotriazole derivatives (3a–d) were synthesized as shown in Scheme . OPC-163493 was prepared from the corresponding aldehyde in the same way.[22,26] The aldehydes (2a–c) were prepared by alkylation or Suzuki coupling of the corresponding benzaldehydes. Aldehyde 2d was prepared with reference to the method reported by Abhale et al.[27] The cyanotriazole derivatives (3a–d) were synthesized by 1,3-dipolar cycloaddition of azide to the condensation intermediate of the aldehydes (2a–d) and (phenylsulfonyl)acetonitrile and obtained in good yields.[28,29] Compounds 3a–d and OPC-163493 all showed TCA cycle-enhancing activities and had the characteristic chemical structure of mUcouplers, a hydrophobic part and weak acid cyanotriazole part that could disperse the charge of their conjugate base[30] (Table ). To confirm the necessity of having an acidic proton on the cyanotriazole ring, we synthesized the methylated OPC-163493 (4). Activation of the TCA cycle did not occur in the presence of 2-methyl-4-cyanotriazole 4 (data not shown).

Scheme 1

Synthesis of Cyanotriazole Derivatives

First, we examined the characteristics of compound 3a, which has a methylene group inserted into the ether moiety and trifluoromethyl groups added onto the benzene ring of compound 1, having an EC50 of 0.30 μM in the TCA cycle activation. The PK study showed that 3a was well absorbed orally and distributed into the liver at a higher concentration than compound 1 (Kp value of 2.9 and 3.3 at 2 and 8 h after dosing, respectively) (Table ). Compound 3a significantly reduced HbA1c at doses of 6 and 20 mg/kg/day in ZDF rats after two-week repeated dosing (Figure ). Improvement in the liver distribution of compound 3a was thought to enhance its in vivo efficacy, although the in vitro activity was similar to that of compound 1. We next examined the effects of high doses of compound 3a on blood glucose levels in rats. Compound 3a also elevated the blood glucose level and rectal temperature of SD rats at doses of 100 and 300 mg/kg, and one of three animals receiving 300 mg/kg of the compound died after developing hyperthermia and hyperglycemia (Figure ).

Figure 2

HbA1c-lowering effects of a series of compounds in ZDF rats. Effect of repeated oral dosing with each compound (BID) on HbA1c changes from baseline in male ZDF rats. Test compound treatments were assessed in rats aged between 11 and 13 weeks. Data represent mean ± SE (n = 6). Single- and double-labeled symbols show p < 0.05 and p < 0.01, respectively, determined by a Williams test (vs each vehicle group (represented as (v)).

Next, we characterized compound 3b which was converted from phenoxyphenyl to biphenyl with trifluoromethyl and fluoro groups as substituents at the 5-position on the cyanotriazole ring of compound 1. Compound 3b showed similar activity in the TCA cycle activation assay (EC50 = 0.35 μM) as compounds 1 and 3a. The PK study of compound 3b revealed that it was also well absorbed and had a similar Kp value in the liver as compound 3a (Table ). Compound 3b consequently demonstrated a remarkable antidiabetic effect in ZDF rats (Figure ). However, compound 3b still raised blood glucose levels at doses of 30 mg/kg or higher, while no elevation of rectal temperature was observed (Figure ). As far as acute toxicity of mUncoupling was concerned, we focused on the translocation of the compounds into the brain and investigated the exposure of the brain to mUncouplers. Although both compounds 3a and 3b were distributed to the brain at a detectable level (Table ), only compound 3a elevated rectal temperature despite both compounds raising blood glucose levels. Therefore, the blood glucose level was thought to be a more sensitive toxicity marker than rectal temperature.

Compound 3c could not be detected in the brain (<0.04 μg/g) after oral administration; it was the first cyanotriazole derivative which did not elevate blood glucose in an acute toxicity study (Figure ). The foregoing results indicated that mere exposure to compounds with mUncoupling activity in the brain was critical for hyperglycemia, and thus blood glucose elevation was deemed to be a sensitive marker of toxicity. Although its in vitro activity was one-third of compound 1 (EC50 of the TCA activation, 1.07 vs 0.34 μM), a significant HbA1c lowering effect was seen in the ZDF rats (Figure ). However, a metabolite implying the possibility of future adverse effects was found in a PK study of compound 3c, and further screening was carried out.

Since compound 3d, having 2-(4-chlorophenyl)-4-thiazoyl as a substituent at the 5-position on the cyanotriazole ring, was also indetectable in the brain, no blood glucose elevation was observed. On the other hand, its antidiabetic effects were prominent in ZDF rats (Table and Figures and 3). We further refined compound 3d and finally obtained OPC-163493.[22] OPC-163493 exhibited a better PK profile than compound 3d, although the EC50 value is similar to that of compound 3d (Table ). A six-week oral dosing study in male ZDF rats revealed that OPC-163493 suppressed HbA1c elevation, and the HbA1c-lowering effect was 0.45, 0.53, and 1.3% points at 2, 4, and 10 mg/kg/day, respectively, which was calculated as 13.2%, 14.5%, and 38.1% reduction from the value of the vehicle (3.41% points; Figure a and Table S2). OPC-163493 also reduced the fasting blood glucose level without any changes in body weight, food intake, and insulin level.[22] Furthermore, to investigate the combined effects of OPC-163493 used in conjunction with metformin or sitagliptin, we conducted a 6-week repeated oral dose study in ZDF rats. Compared with metformin monotherapy, a combination of both low (4 mg/kg/day) and high dose (10 mg/kg/day) of OPC-163493 with metformin (100 mg/kg/day) significantly suppressed HbA1c elevation. Treatment using higher dose OPC-163493 in combination with sitagliptin (10 mg/kg/day) significantly reduced elevated HbA1c compared with sitagliptin monotherapy. Since metformin alone had no effect on this animal model, the add-on effect of OPC-163493 at 10 mg/kg/day was similar, 35.0% reduction from the value of vehicle (2.97% points) to that of OPC-163493 monotherapy. On the other hand, the HbA1c-lowering effect of sitagliptin monotherapy was 11.4% reduction from the vehicle value, and the combination therapy with OPC-163494 showed an add-on effect, 50.5% reduction (Figure b and Table S3). Therefore, OPC-163493 demonstrated add-on effects on both metformin and sitagliptin in this combination study. Our preclinical toxicity studies demonstrated the tolerability of OPC-163493.[22]

Figure 4

HbA1c-lowering effect of OPC-163493 monotherapy (a) and in combination with metformin (100 mg/kg/day) or sitagliptin (10 mg/kg/day) (b). HbA1c changes from baseline were evaluated in 6-week repeated oral dose (BID) studies in male ZDF rats. Data represent mean ± SE (n = 16 in (a), n = 8 in (b)). *P < 0.05, **P < 0.01 compared with vehicle group (v). ##p < 0.01, $p < 0.05 compared with the metformin and sitagliptin monotherapy group, respectively. Monotherapy data are taken from a previous report.[22]

Discussion

In this report, we described the process of optimizing cyanotriazole derivatives with mUncoupling activity to create a potential antidiabetic agent, OPC-163493.

Initially, we assessed the possibility of using NaCT inhibitors to treat diseases caused by excessive calorie intake such as DM since NaCT is thought to be a candidate homologue of Indy, a single mutation of which has been shown to extend the lifespan of fruit flies via CR-like effects.[7,8] We serendipitously found an mUncoupler, compound 1, while using a cell-based screening system aimed at obtaining NaCT inhibitors. Although compound 1 exhibited antidiabetic effects to some degree, there was still room for improvement. Therefore, we endeavored to optimize compound 1 to create a safe and effective mUncoupler that can be administered to patients with DM.

In this study, we showed that manipulation of a compound’s tissue distribution, specifically avoiding the brain with most of the compound distributed to the liver, was of paramount importance to preclude a toxic response to an mUncoupling agent while maintaining its antidiabetic efficacy. At that phase of the study, monitoring blood glucose levels was a more helpful and sensitive marker to ascertain toxic responses than body temperature in the acute toxicity tests. We also confirmed that DNP obviously elevated blood glucose level at a dose of 10 mg/kg and 30 mg/kg in a similar acute toxicity test (Figure S3). Trace amounts of compounds 3a and 3b were detected in the brain after oral administration (10 mg/kg), and both compounds elevated blood glucose levels when administered at 3 or 10 times the dose used in the PK study (Figure b). We speculated that even slight exposure to the mUncoupler in the brain could cause hyperglycemia. It is well-known that the glucose-excited (GE) neurons and glucose-inhibited (GI) neurons in the ventromedial hypothalamus (VMH) play a coordinated role in regulating blood glucose levels. The GI neurons are particularly important for the counter-regulatory response to hypoglycemia. As a possible mechanism, AMP-activated protein kinase (AMPK), a key sensor of cellular energy status, is thought to be involved in activation of GI neurons. Hypoglycemia induces a decrease in intracellular ATP: an increase in the AMP/ATP ratio, which activates AMPK. Activated AMPK phosphorylates neuronal nitric oxide (NO) synthase (nNOS), leading to NO production. NO then binds to its receptor, soluble guanylyl cyclase (sGC), which increases cyclic guanosine monophosphate (cGMP) production. cGMP ultimately leads to closure of the cystic fibrosis transmembrane regulator (CFTR) Cl– conductance and depolarization in VMH GI neurons, which results in activation of the sympathetic nerves innervating the liver and stimulation of the release of counter-regulatory hormones such as glucagon and catecholamines.[31−34] Since mUncoupling reduces the efficiency of ATP production in oxidative phosphorylation, it leads to an increase in the AMP/ATP ratio and activation of AMPK.[22] Thus, once an mUncoupler is distributed in the brain, this would activate GI neurons and cause blood glucose levels to rise.

On the other hand, in the optimization process, it was difficult to predict the pattern of tissue distribution and toxicities on the basis of the physicochemical properties such as the octanol/water partition coefficient (log P) or the acid dissociation constant (pKa) of each compound (Table ). Clarifying the distribution in the liver and brain by PK studies was the only way to predict the efficacy and acute toxicities of each compound.

We also demonstrated that OPC-163493 was highly effective not only as monotherapy but also when used in combination with metformin or sitagliptin. OPC-163493 also ameliorated HbA1c in a wide variety of DM model animals, including Akita mice (type 1 DM model) and aged ZDF rats (extreme insulin resistance model).[22] These data suggest that OPC-163493 may be a possible therapeutic option in the treatment of DM. It is hoped that future clinical studies will prove its efficacy and safety in humans.

Experimental Section

Chemistry

All reagents and solvents were of reagent grade and purchased from commercial suppliers (Sigma-Aldrich Japan, FUJIFILM Wako Pure Chemical Corporation, and Tokyo Chemical Industry Co., Ltd.) and were used directly without further purification. Compound purification by column chromatography was carried out using commercial prepacked columns from Biotage or Yamazen. Thin-layer chromatography (TLC) analyses were conducted on silica gel plates (60 F254, 0.2 mm thick, Merck). 1H-nuclear magnetic resonance spectroscopy (NMR) and 13C NMR spectra were recorded on a Bruker Avance III HD 500 MHz or Bruker Avance III 400 MHz NMR spectrometer using TMS as an internal reference. When peak multiplicities are reported, the following abbreviations are used: s = singlet, d = doublet, t = triplet, m = multiplet, br = broadened, dd = doublet of doublets, dt = doublet of triplets. Coupling constants, when given, are reported in hertz. Infrared (IR) spectra were recorded using a HORIBA FT-720 spectrophotometer. High-resolution mass spectra (HRMS) were obtained using a WATERS H-class/Xevo G2-XS for electrospray ionization (ESI). Elemental analysis was performed on a CHN corder MT-6, from J-SCIENCE LAB. The purity was determined by high-performance liquid chromatography (HPLC). The purity of all final compounds was 95% or higher. The instrument was a Shimadzu LC6A and LC20, from SHIMADZU. The column was a TSK-GEL ODS-80Ts (150 mm × 4.6 mm), from TOSHO BIOSCIENCE.

3-[2,5-Bis(trifluoromethyl)benzyloxy]benzaldehyde (2a)

To a suspension of m-hydroxybenzaldehyde (9.89 g, 81.1 mmol) in NMP (180 mL) was added 2,5-bis(trifluoromethyl)benzylbromide (24.9 g, 81.1 mmol) and K2CO3 (16.8 g, 122 mmol). The reaction mixture was stirred at 80 °C for 2 h. Water was slowly added to the reaction mixture. The precipitate was filtered and washed with water, giving the title compound (28.3 g, quant.) as a brown solid. 1H NMR (CDCl3) δ: 10.01 (1H, s), 8.08 (1H, s), 7.86 (1H, d, J = 8.1 Hz), 7.73 (1H, d, J = 8.1 Hz), 7.56–7.49 (3H, m), 7.31–7.28 (1H, m), 5.35 (2H, s). 13C NMR (CDCl3) (125 MHz) δ: 191.8, 158.6, 138.0, 136.5 (d, J = 1.3 Hz), 134.4 (q, J = 33.1 Hz), 130.6 (q, J = 31.7 Hz), 130.4, 126.8 (q, J = 5.6 Hz), 125.7 (q, J = 3.7 Hz), 125.0 (q, J = 3.7 Hz), 124.4, 123.6 (q, J = 274.3 Hz), 123.3 (q, J = 273.0 Hz), 121.9, 113.4, 65.7 (q, J = 1.3 Hz). IR (ATR): 1704, 1594, 1309, 1259, 1089 cm–1.

3′-Fluoro-5′-(trifluoromethyl)-[1,1′-biphenyl]-3-carbaldehyde (2b)

To a solution of 1-bromo-3-fluoro-5-(trifluoromethyl)benzene (2.94 g, 12.1 mmol) in DME (40 mL) were added 3-formylphenylboronic acid (2.0 g, 13.3 mmol), 2 M Na2CO3 aqueous solution (7.3 mL), and 1,1′-bis(diphenylphosphino)ferrocene-palladium dichloride (297 mg, 0.364 mmol). The reaction mixture was stirred at 80 °C under a N2 atmosphere for 2 h. The reaction mixture was diluted with water and extracted twice with ethyl acetate. The combined organic layers were washed with brine, dried over MgSO4, and concentrated in vacuo. The residue was purified by flash column chromatography (hexane/ethyl acetate = 1:0 to 9:1) to give the desired product, the title compound (2.90 g, 89%), as a white solid. 1H NMR (CDCl3) δ: 10.12 (1H, s), 8.10 (1H, s), 7.96–7.94 (1H, m), 7.87–7.85 (1H, m), 7.69–7.66 (2H, m), 7.52 (1H, d, J = 8.2 Hz), 7.37 (1H, d, J = 8.2 Hz). 13C NMR (CDCl3) (125 MHz) δ: 191.8, 162.9 (d, J = 249.5 Hz), 143.1 (d, J = 7.7 Hz), 139.5 (d, J = 1.8 Hz), 137.2, 133.3 (qd. J = 33.1, 8.3 Hz), 132.9, 130.2, 130.0, 127.9, 123.3 (qd, J = 272.8, 3.0 Hz), 119.8 (dq, J = 3.6, 3.6 Hz), 117.6 (d, J = 22.5 Hz), 112.3 (dq, J = 24.5, 3.7 Hz). IR (ATR): 1702, 1604, 1450, 1355, 1268, 1168, 1126 cm–1.

(E)-3-Chloro-5-[3-(trifluoromethyl)styryl]benzaldehyde (2c)

To a suspension of 3-bromo-5-chlorobenzaldehyde (3.12 g, 14,3 mmol) in DME (45 mL) and water (15 mL) were added (E)-[3-(trifluoromethyl)styryl]boronic acid pinacol ester (4.25 g, 14.3 mmol), sodium phosphate tribasic dodecahydrate (16.8 g, 122 mmol), and the 1,1′-bis(diphenylphosphino)ferrocene–palladium(II) dichloride CH2Cl2 complex (0.35 g, 0.428 mmol). The reaction mixture was stirred at 80 °C under a N2 atmosphere for 2 h. The reaction was quenched by addition of water, and then ethyl acetate was added. After separation, the aqueous phase was extracted with ethyl acetate. The combined organic layers were washed with brine, dried over MgSO4, and concentrated in vacuo. The residue was purified by flash column chromatography (hexane/ethyl acetate = 1:0 to 9:1) to give the desired product, the title compound (3.15 g, 71%), as a pale yellow solid. H NMR (CDCl3) (400 MHz) δ: 10.01 (1H, s), 7.91 (1H, s), 7.78–7.69 (4H, m), 7.58–7.49 (2H, m), 7.23 (1H, d, J = 16.4 Hz), 7.15 (1H, d, J = 16.4 Hz). 13C NMR (CDCl3) (125 MHz) δ: 190.7, 139.5, 138.1, 137.0, 135.9, 131,9, 131.5 (q, J = 32.1 Hz), 130.3, 129.9, 129.4, 128.6, 127.6, 125.7, 125.0 (q, J = 3.6 Hz), 124.0 (q, J = 272.4 Hz), 123.4 (q, J = 3.8 Hz). IR (ATR): 1697, 1590, 1571, 1450, 1336, 1322, 1110, 1068 cm–1.

5-{3-[2,5-Bis(trifluoromethyl)benzyloxy]phenyl}-3H-[1,2,3]triazole-4-carbonitrile (3a)

To a solution of compound 2a (32.26 g, 92.7 mmol) in toluene (300 mL) was added (phenylsulfonyl)acetonitrile (16.8 g, 92.7 mmol) and potassium acetate (9.09 g, 92.7 mmol) in one portion, and the mixture was stirred at 80 °C for 1.5 h. The solvent was removed under reduced pressure. To a solution of the obtained residue in NMP (300 mL) was added sodium azide (9.04 g, 139 mmol), and the mixture was stirred at 110 °C for 2 h. After the reaction mixture cooled to room temperature, NH4Cl aqueous saturated solution was added. The mixture was extracted twice with ethyl acetate. The combined organic layer was washed twice with water and brine, dried over anhydrous Na2SO4, and concentrated under reduced pressure. The obtained residue was purified by crystallization from acetonitrile to produce the title compound as a white powder (16.71 g, 44%). 1H NMR (DMSO-d6) (500 MHz) δ: 8.21 (1H, s), 8.09 (1H, d, J = 8.2 Hz), 8.02 (1H, d, J = 8.2 Hz), 7.61–7.52 (3H, m), 7.28 (1H, dd, J = 7.2, 1.6 Hz), 5.41 (2H, s). 13C NMR (DMSO-d6) (125 MHz) δ: 158.7, 146.5, 136.8, 133.4 (q, J = 32.6 Hz), 131.4, 131.1 (q, J = 31.5 Hz), 128.2 (q, J = 5.5 Hz), 127.8, 127.5 (q, J = 3.6 Hz), 126.3 (q, J = 3.6 Hz), 123.9 (q, J = 274.5 Hz), 123.8 (q, J = 272.9 Hz), 120.1, 117.2, 116.6, 113.8, 113.4, 66.3. IR (ATR): 2242, 1598, 1496, 1427, 1313, 1257, 1166, 1110, 1087 cm–1. HRMS (ESI) m/z: [M + Na]+ calcd for C18H10F6N4ONa, 435.0656; found, 435.0653. Elemental analysis calculated (%) for C18H10F6N4O: C 52.44, H 2.44, N 13.59. Found: C 52.20, H 2.51, N 13.56. HPLC analysis: retention time = 8.154 min; peak area, 99.9%; eluent, CH3CN:H2O:AcOH = 65:35:1; 30 min with a flow rate of 1 mL min–1 and detection at 254 nm; column temperature, rt.

5-(3′-Fluoro-5′-trifluoromethyl-[1,1-biphenyl]-3-yl)-3H-[1,2,3]triazole-4-carbonitrile (3b)

Compound 3b was made in a manner similar to compound 3a, resulting in beige granules (CH3CN). Yield: 59%. 1H NMR (DMSO-d6) (500 MHz) δ: 8.27 (1H, s), 8.03–7.95 (4H, m), 7.78–7.75 (2H, m). 13C NMR (DMSO-d6) (125 MHz) δ: 162.5 (d, J = 247.0 Hz), 145.7 (br), 142.8 (d, J = 8.3 Hz), 138.3, 131.8 (qd, J = 32.7, 8.3 Hz), 130.3, 129.2, 126.8, 126.5 (br), 125.4, 123.3 (qd, J = 272.8, 3.1 Hz), 119.5 (qd, J = 6.2, 3.1 Hz), 117.8 (d, J = 22.4 Hz), 116.5, 113.3, 112.2 (dq, J = 24.9, 3.5 Hz). IR (ATR): 2244, 1596, 1359, 1265, 1162, 1126, 1095 cm–1. HRMS (ESI) m/z: [M + Na]+ calcd. For C16H8F4N4Na, 355.0583; found, 355.0576. Elemental analysis calculated (%) for C16H8F4N4: C 57.84, H 2.43, N 16.89. Found: C 57.75, H 2.71, N 16.80. HPLC analysis: retention time = 5.547 min; peak area, 99.8%; eluent, CH3CN:H2O:AcOH = 70:30:1; 30 min with a flow rate of 1 mL min–1 and detection at 254 nm; column temperature, rt.

(E)-5-(3-Chloro-5-(3-(trifluoromethyl)styryl)phenyl)-3H-1,2,3-triazole-4-carbonitrile (3c)

Compound 3c was made in a manner similar to compound 3a, resulting in white powder (hexane-ethyl acetate). Yield: 73%. 1H NMR (DMSO-d6) (500 MHz) δ: 8.05 (1H, s), 8.02 (1H, s), 7.96–7.94 (2H, m), 7.82 (1H, s), 7.68–7.64 (2H, m), 7.57–7.53 (2H, m). 13C NMR (DMSO-d6) (125 MHz) δ: 146.2, 140.4, 137.9, 134.9, 130.9, 130.2, 130.1 (q, J = 31.6 Hz), 129.1, 128.5, 127.9, 125.7, 124.9 (q, J = 3.6 Hz), 124.6 (q, J = 272.5 Hz), 124.1, 123.6 (q, J = 3.7 Hz), 117.1, 113.5. IR (ATR): 3224, 2252, 1577, 1332, 1110 cm–1. HRMS (ESI) m/z: [M + Na]+ calcd. for C18H10N4ClF3Na, 397.0444; found, 397.0438. Elemental analysis calculated (%) for C18H10ClF3N4: C 57.69, H 2.69, N 14.95. Found: C 57.38, H 2.78, N 14.87. HPLC analysis: retention time = 11.442 min; peak area, 99.9%; eluent, CH3CN:H2O:AcOH = 70:30:1; 30 min with a flow rate of 1 mL min–1 and detection at 254 nm; column temperature, rt.

5-[2-(4-Chlorophenyl)-thiazol-4-yl]-3H-[1,2,3]triazole-4-carbonitrile (3d)

Compound 3d was made in a manner similar to compound 3a, resulting in a pale brown solid (ethyl acetate–hexane). Yield: 66%. 1H NMR (DMSO-d6) (300 MHz) δ: 8.37 (s), 8.07 (2H, d, J = 8.5 Hz), 7.66 (2H, d, J = 8.6 Hz). 13C NMR (DMSO-d6) (125 MHz) δ: 167.0, 142,3 (br.), 141.9 (br.), 135.4, 131.0, 129.3, 127.8, 120.4, 116.9, 112.7. IR (ATR): 2240, 1594, 1496, 1450, 1402, 1091, 1070 cm–1. HRMS (ESI) m/z: [M + Na]+ calcd. for C12H6ClN5SNa, 309.9930; found, 309.9925. Elemental analysis calculated (%) for C12H6ClN5S: C 50.09, H 2.10, N 24.34. Found: C 49.89, H 2.13, N 24.33. HPLC analysis: retention time = 4.416 min; peak area, 99.9%; eluent, CH3CN:H2O:AcOH = 70:30:1; 30 min with a flow rate of 1 mL min–1 and detection at 254 nm; column temperature, rt.

2-Methyl-5-{5-methyl-2-[4-(trifluoromethyl)phenyl]thiazol-4-yl}-2H-1,2,3-triazole-4-carbonitrile (4)

To a suspension of OPC-163493 (500 mg, 1.491 mmol) in DMF (5 mL) were added potassium carbonate (412 mg, 2.98 mmol) and iodomethane (0.112 mL, 1.789 mmol) at room temperature. The reaction mixture was stirred at room temperature for 3 days. To the reaction mixture was added water and ethyl acetate, and the precipitate was filtered. Water was added to the precipitate solution of DMSO, and the precipitate was filtered and washed with water, giving the desired product, the title compound (213 mg, 41% yield), as a pale yellow powder. The powder was recrystallized in DMSO, and a colorless column was obtained. Using X-ray crystal structure analysis, the methyl position on the triazole ring of the compound was determined (see Supporting Information). 1H NMR (DMSO-d6) (500 MHz) δ 8.16 (2H, d, J = 8.3 Hz), 7.92 (2H, d, J = 8.3 Hz), 4.37 (3H, s), 2.80 (3H, s). 13C NMR (DMSO-d6) (125 MHz): δ 162.34, 146.00, 139.37, 136.14, 135.82, 130.12 (q, J = 31.1 Hz), 126.43, 126.31 (q, J = 3.7 Hz), 123.84 (q, J = 271.7 Hz), 118.40, 112.25, 43.07, 12.35. HRMS (ESI) m/z: [M + H]+ calcd. for C15H11F3N5S, 350.0687; found, 350.0692. Elemental analysis calculated (%) for C15H10F3N5S: C 51.57, H 2.89, N 20.05. Found: C 51.45, H 2.69, N 20.10. HPLC analysis: retention time = 12.125 min; peak area, 99.5%; eluent, CH3CN:H2O:AcOH = 70:30:1; 30 min with a flow rate of 1 mL min–1 and detection at 254 nm; column temperature, rt.

Measurement of log P, pKa

log P and pKa of 1, 3a–d, and OPC-163493 were measured by PION Sirius T3[35] according to the manufacturer’s manual.

Reagents, Cells, and in Vitro Studies

Compound 1 was obtained from Namiki Shoji Co., Ltd. FCCP was synthesized at Medicinal Chemistry Research Laboratories, Otsuka Pharmaceutical Co., Ltd. DNP was obtained from Sigma-Aldrich Co., LLC. [1,5-14C] citric acid was purchased from PerkinElmer, Inc.

CHO-K1 (#EC85051005) and HepG2 (#EC85011430) cells were purchased from DS Pharma Biomedical Co., Ltd. Control CHO and NaCT-CHO were established by transfection, initially with the pME18S (GenBank accession No. AB009864) and subsequently with pcDNA3.1 (+) vectors, using this approach to insert the NaCT gene into CHO-K1 cells, while control cells were transfected without insertion of the NaCT gene. The methods of the TCA cycle activation assay, mUncoupling activity, and cellular fluxes using an extracellular flux analyzer are described elsewhere.[22]

Animal Studies

1,1-Dimethyl biguanide (metformin) hydrochloride was purchased from FUJIFILM Wako Pure Chemical Corporation. Sitagliptin phosphate hydrate was extracted from JANUVIA (Merck & Co., Inc.). Sprague–Dawley (SD) rats and ZDF rats were purchased from Charles River Laboratories Japan, Inc. Animals were housed under specific pathogen-free (SPF) conditions using a 12 h light and 12 h dark cycle with access to food and water ad libitum. MF and CRF-1 (Oriental Yeast Co., Ltd.) were fed to ZDF and SD rats, respectively. All experiments were carried out in accordance with “Guidelines for Animal Care and Use at Otsuka Pharmaceutical Co., Ltd.”

Pharmacokinetic studies were performed using male SD rats (aged 6–7 weeks). Compound suspensions were administered to two animals on the basis of their body weights, and blood samples were collected via the tail vein. To determine the tissue distribution of each compound, the collection of blood samples was conducted via the inferior vena cava under isoflurane anesthesia at 2 and 8 h after dosing, and the liver and brain were isolated from animals euthanized by exsanguination.

An acute toxicological study was carried out using male SD rats (aged 6–7 weeks). Taking animal welfare into account, we determined that the minimum number of animals needed for evaluating the toxicities of cyanotriazole derivatives was 3 or 4, so no statistical analysis was performed on the toxicological indexes. After dosing, the time-dependent changes in blood glucose levels were measured up to 8 h, and rectal temperatures were monitored as necessary.

An antidiabetic study was conducted using male 11-week-old ZDF rats. The animals were allocated on the basis of HbA1c, fasting blood glucose level, and body weight. Body weights and food intake were periodically monitored. Compounds were suspended in 5% gum Arabic solution and administered orally twice a day based on the latest body weight.

Measurement of Compound Concentrations in the Plasma, Liver, and Brain

The plasma and tissue concentrations of each compound were determined by high-performance liquid chromatographic–electrospray ionization tandem mass spectrometry (LC-ESI-MS/MS). The plasma or tissue homogenates and 10-fold acetonitrile–methanol (1:1, v/v) were mixed and centrifuged. The supernatant was used for the LC-MS/MS analysis. High-performance liquid chromatography (HPLC) analysis was carried out with the Prominence UFLCXR system (Shimadzu Corporation). Tandem mass spectrometry (MS/MS) analysis was carried out with the AB SCIEX API4000 (AB SCIEX) equipped with an ESI source. HPLC analysis was conducted using the binary gradient method. The analytical column was a CAPCELLPAK C18 MGII (3 μm, 2.1 mm ID × 50 mm, Shiseido Company, Limited). Mobile phase: 10 M ammonium acetate aqueous solution and acetonitrile with a flow rate of 0.55 mL/min. Selected reaction monitoring MS/MS analysis was conducted in negative ionization mode for each analyte.

Measurement of Blood Glucose Level, HbA1c Values, and Rectal Temperature Levels

Blood for blood glucose measurement was taken from the tail vein without anesthesia. Blood glucose levels in plasma were determined by Glutest Mint (Sanwa Kagaku Kenkyusho Co., Ltd.). Heparinized blood samples for baseline HbA1c and efficacy measurements were taken from the tail vein without anesthesia and the inferior vena cava under isoflurane anesthesia, respectively. HbA1c values were determined using a DCA Vantage Analyzer (Siemens). Rectal temperature levels were measured with BWT-100, a small animal thermometer (Bio Research Center Co., Ltd., while the animals were awake.

Statistical Analysis

All statistical analyses were performed using SAS Software (release 9.3, SAS Institute Japan Ltd.). The level of significance was two-tailed with P < 0.05. Each statistical test was determined after consultation with the statistician in charge of statistics at Otsuka. Methods used for analysis are described in the figure and Supporting Information figure legends.