Article
`
`pubs.acs.org/jmc
`
`Discovery of (1R,2S)‑2-{[(2,4-Dimethylpyrimidin-5-yl)oxy]methyl}-2-
`(3-fluorophenyl)‑N‑(5-fluoropyridin-2-yl)cyclopropanecarboxamide
`(E2006): A Potent and Efficacious Oral Orexin Receptor Antagonist
`†
`†
`†
`†
`Yu Yoshida,*,†
`Yoshimitsu Naoe,
`Taro Terauchi,
`Fumihiro Ozaki,
`Takashi Doko,
`Ayumi Takemura,
`†
`†
`‡
`‡
`§
`Toshiaki Tanaka,
`Keiichi Sorimachi,
`Carsten T. Beuckmann,
`Michiyuki Suzuki,
`Takashi Ueno,
`†
`∥
`Shunsuke Ozaki,
`and Masahiro Yonaga
`†
`‡
`Biopharmacology, §Physical Chemistry, and
`Drug Metabolism and Pharmacokinetics, Eisai Product Creation
`Medicinal Chemistry,
`Systems, Eisai Co., Ltd., 5-1-3 Tokodai, Tsukuba-shi, Ibaraki 300-2635, Japan
`*S Supporting Information
`
`†
`
`∥
`
`ABSTRACT: The orexin/hypocretin receptors are a family of
`G protein-coupled receptors and consist of orexin-1 (OX1)
`and orexin-2 (OX2) receptor subtypes. Orexin receptors are
`expressed throughout the central nervous system and are
`involved in the regulation of the sleep/wake cycle. Because
`modulation of these receptors constitutes a promising target
`for novel treatments of disorders associated with the control of
`sleep and wakefulness, such as insomnia, the development of
`orexin receptor antagonists has emerged as an important focus
`in drug discovery research. Here, we report the design, synthesis, characterization, and structure−activity relationships (SARs) of
`novel orexin receptor antagonists. Various modifications made to the core structure of a previously developed compound (-)-5,
`the lead molecule, resulted in compounds with improved chemical and pharmacological profiles. The investigation afforded a
`potential therapeutic agent, (1R,2S)-2-{[(2,4-dimethylpyrimidin-5-yl)oxy]methyl}-2-(3-fluorophenyl)-N-(5-fluoropyridin-2-yl)-
`cyclopropanecarboxamide (E2006), an orally active, potent orexin antagonist. The efficacy was demonstrated in mice in an in
`vivo study by using sleep parameter measurements.
`
`■ INTRODUCTION
`
`Insomnia, a disorder in which the patient suffers from an
`inability to sleep or stay asleep for as long as desired,
`constitutes a widespread issue in today’s
`society, as
`approximately 30% of adults in the United States experience
`some symptoms of insomnia and approximately 10% describe
`their insomnia as chronic and/or severe. Despite the prevalence
`of this disorder, research regarding medical treatments for
`insomnia is relatively limited.1
`leads to a
`The effects of
`insomnia are dramatic, as it
`reduction in the quality of life, poor productivity, and a high
`risk of traffic and work-related safety incidents. Additionally,
`insomnia is a costly disorder, as it is estimated to cost over $63
`billion per year in the United States.2 Commonly prescribed
`medications for the treatment of insomnia include agents that
`positively modulate GABAA receptors, with zolpidem, a
`nonbenzodiazepine sleep drug, being the current market leader.
`Nonbenzodiazepine sleep aids are structurally distinct from
`benzodiazepines, as reflected in their naming, and bind to
`benzodiazepine sites with high specificity. Additionally, their
`pharmacological profiles are believed to be better
`in
`comparison with those of benzodiazepines owing to the less
`severe side effects. Despite the availability of various sleep-
`modifying drugs, the prevalence of insomnia has not decreased
`substantially because of remaining concerns about the overall
`
`safety and efficacy of treatments that target the GABA signaling
`pathway.3,4 Notably, two new non-GABA-related sleep drugs
`have recently been approved for the treatment of insomnia.
`Ramelteon, a melatonin (MT) receptor MT1/MT2 agonist, was
`approved in 2005, and doxepin, a histamine H1 receptor
`antagonist, received regulatory approval in 2010. In contrast to
`benzodiazepine and nonbenzodiazepine agents,
`these two
`sleep-aid drugs are nonscheduled drugs in the U.S.. However,
`questions remain about the effectiveness of ramelteon and
`doxepin, owing to the limited number of reports demonstrating
`their superiority to other sleep drugs,
`including those that
`positively modulate GABAA.5,6 The medical needs of patients
`with insomnia warrant the development of sleep medications
`with novel mechanisms of action.
`Toward that end, antagonism of the orexin receptor may
`serve as a promising route. Orexin A and orexin B, also known
`as hypocretin 1 and hypocretin 2, are endogenous neuropeptide
`ligands for orexin/hypocretin receptors (OXRs), which are G
`protein-coupled receptors (GPCRs). Orexin-1 receptor
`(OX1R) and orexin-2 receptor (OX2R) mediate the action of
`these ligands by postsynaptic neuronal signal transduction.
`OX1R and OX2R are expressed throughout the central nervous
`
`Received: February 6, 2015
`Published: May 8, 2015
`
`© 2015 American Chemical Society
`
`4648
`
`DOI: 10.1021/acs.jmedchem.5b00217
`J. Med. Chem. 2015, 58, 4648−4664
`
`Downloaded via Javii Austin on July 6, 2020 at 14:15:03 (UTC).
`
`See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.
`
`Page 1 of 17
`
`EISAI EXHIBIT 1020
`
`

`

`Journal of Medicinal Chemistry
`
`Article
`
`Figure 1. Orexin receptor dual antagonists that have advanced to clinical trials.
`
`system and are involved in the regulation of the sleep/wake
`cycle.7−9 To date, the specific contribution of OXR to the
`regulation of
`the sleep/wake cycle has recently starting
`emerged. A number of genetic and pharmacological studies
`have suggested that the orexin pathway plays an important role
`in regulating the sleep/wake cycle.10−19 Studies performed in
`orexin/ataxin-3 transgenic mice and rats, which lack orexin
`peptide-producing neurons, and OXR-deficient mice have
`shown that both receptors are involved in the regulation of
`the sleep/wake cycle because the lack of activation of both
`receptors results in a narcolepsy−cataplexy phenotype, while
`the lack of activation of either
`receptor alone elicits an
`attenuated sleep phenotype.20−23 Furthermore, very low levels
`of orexin in cerebrospinal fluid (CSF),
`indicative of nearly
`complete or complete loss of orexinergic neurons, have been
`observed in humans suffering from narcolepsy−cataplexy
`syndrome.24−26
`A dual OXR antagonist, almorexant, evaluated by Actelion
`Pharmaceuticals/GlaxoSmithKline was found to significantly
`decrease the behavioral indices of wakefulness in animals and
`has shown effectiveness in clinical trials.11 Recently, a number
`of orally active selective OX2R antagonists have been evaluated
`in animal studies.27,28 However, whether a dual
`receptor
`antagonist29 or an OX2R-selective antagonist30 will be better
`suited for use as a sleep aid is a topic of ongoing debate. We
`think that it is feasible to speculate that a sleep drug which
`promotes both non-REM and REM sleep would provide
`patients with a more natural sleep architecture, and we are
`providing animal data in this manuscript which suggest that a
`dual orexin antagonist is a potential candidate to fulfill this
`requirement.
`To date, a number of dual OXR antagonists have been
`described. Almorexant (1) and was tested in the treatment of
`insomnia,31 but a phase III clinical trial was discontinued. SB-
`649868 (2) from GlaxoSmithKline proceeded to a phase II
`clinical trial.32 Suvorexant (3), developed by Merck, received
`approval from the Pharmaceutical and Medical Devices Agency
`in Japan and the US Food and Drug Administration (FDA) in
`2014 and is marketed for the treatment of insomnia.33,34 Merck
`
`has entered another dual OXR antagonist filorexant (4) into a
`phase II clinical trial for multiple indications,
`including the
`treatment of insomnia (Figure 1).35
`In this paper, we report the synthesis, structure−activity
`relationships (SARs), optimization of drug-likeness parameters,
`and in vivo efficacy of a series of cyclopropane compounds
`reported previously.36 Our efforts in this area led to the
`discovery of 34 (E2006) as a promising dual OXR antagonist
`that has advanced into clinical trials.
`A novel series of compounds containing a cyclopropane core
`structure were identified as promising orally active orexin
`receptor antagonists, as exemplified by (−)-536 (Table 1).
`(−)-5 exhibits low-nanomolar affinity toward human OX2R, as
`measured by radio ligand-displacement binding assays.
`However, this compound exhibits a number of drawbacks in
`its drug-likeness parameters that need to be improved, such as
`Table 1. Properties of Compound (−)-5
`
`in vitro
`binding
`affinity
`(Ki, nM)
`
`compd OX2R OX1R
`(−)-5
`5
`106
`
`solubilitya
`(μM)
`pH
`pH
`7.4b
`1.2c
`15
`25
`
`TDI of CYP3Ad
`
`(% of control at
`10 μM)
`49
`
`metabolic
`stabilitye
`
`(residual
`ratio %)
`80
`
`aDMSO solution precipitation method.36 bDulbecco’s phosphate-
`buffered saline (PBS). cJapanese Pharmacopoeia 1st fluid (aqueous
`HCl solution containing 34 mM NaCl). dTime-dependent inhibition
`evaluated using a cocktail of probe substrates with human liver
`microsomes. eMetabolism after incubation of compounds at 0.3 μM in
`0.1% DMSO with human liver microsomes for 15 min.
`
`4649
`
`DOI: 10.1021/acs.jmedchem.5b00217
`J. Med. Chem. 2015, 58, 4648−4664
`
`Page 2 of 17
`
`

`

`Journal of Medicinal Chemistry
`Scheme 1a
`
`Article
`
`aReagents and conditions: (a) (1) NaHMDS, THF, 0 °C, 3 h, (2) KOH, EtOH, reflux, 8 h, (3) HCl, 0 °C−rt, 3 h; (b) NaBH4, MeOH−THF, 0
`°C−rt; (c) TBDPS-Cl, imidazole, DMF, −10 °C−rt, or vinyl acetate, lipase acrylic resin from Candida antarctica, rt; (d) Ar1-OH, DIAD, PPh3, 0
`°C−rt, overnight or (1) MsCl, TEA, DCM, (2) Ar1-OH, Cs2CO3, MeCN, 70 °C; (e) TBAF, THF, rt, 1 h or NaOH, EtOH−H2O, rt, 1 h (2 steps);
`(f) (COCl)2, DMSO, TEA, DCM, −78 °C−rt, 1 h; (g) 2-methyl-2-butene, NaClO2, NaH2PO4, acetone−H2O, rt, 2 h; (h) amine, HATU, DIPEA,
`DMF, 60 °C, overnight or (COCl)2, cat. DMF, DCM, rt, 1 h, then amine, DIPEA, THF, 60 °C.
`afforded carboxylic acid intermediates 18a−i. Subsequent
`amidation of 18a−i with various amines yielded the desired
`products 19−38.
`Components of the heteroaromatic A ring were synthesized
`as outlined in Schemes 2 and 3. Pyrazole derivatives 40a and
`
`its time-dependent inhibition (TDI) of cytochrome P450 3A
`(CYP3A) and low aqueous solubility under both acidic and
`neutral conditions (pH 1.2 and pH 7.4). (−)-5 also
`demonstrated moderate reversible inhibition of CYPs (IC50
`values of 8 and 8.6 μM for CYP2C8 and CYP2C19,
`respectively, and >10 μM for CYP1A2, CYP2C9, and
`CYP2D6). In an effort to find a clinical candidate within the
`chemical series, the A, B, and C rings in the molecule were
`optimized through chemical modifications.
`
`Scheme 2a
`
`■ RESULTS AND DISCUSSION
`Chemistry. The general synthetic route used to produce the
`cyclopropane compounds is presented in Scheme 1. Chiral
`cyclopropane ring formation was carried out by the reaction of
`corresponding aryl acetonitriles 6−12 with (R)-epichlorohydrin
`to yield the desired lactones 13a−f.37 The reported racemic
`compounds36 can be obtained using racemic epichlorohydrin
`instead of (R)-epichlorohydrin by following a procedure similar
`to the one described in Scheme 1. Reduction of the lactone
`with sodium borohydride afforded diol products 14a−f
`in
`excellent yields. Protection of the hydroxy group of 14a with
`TBDPS-Cl gave the desired monoprotected compound 15a. In
`the selective monoacylation of diols 14b−f, enzymatic acylation
`with lipase acrylic resin from Candida antarctica was
`successfully applied in the presence of vinyl acetate to obtain
`compounds 15b−f in good yields. The Mitsunobu reaction of
`the corresponding alcohols with various phenols led to the
`production of 16a−i. Deprotection of 16a and 16g−i using
`TBAF, or hydrolysis of the acyl group of 16b−f with sodium
`hydroxide, followed by Swern oxidation and Pinnick oxidation,
`
`aReagents and conditions: (i) m-CPBA, CHCl3, rt, overnight.
`
`40b were synthesized from commercially available aldehydes
`39a and 39b via Baeyer−Villiger oxidation (Scheme 2). The
`Pd-catalyzed coupling of commercially available 2,4-dichloro-5-
`methoxy pyrimidine 41 with trimethylaluminum afforded 42 in
`a good yield. An analogous transformation yielded 45 from
`chloro intermediate 44, which was accessed by iron-catalyzed
`ethylation of 41 with ethyl magnesium chloride (Scheme 3).
`Deprotection of the methoxy group of 42 and 45 in the
`presence of BBr3 produced hydroxypyrimidines 43 and 46.
`Compound 49 was synthesized from 43. Protection of the
`hydroxyl group with benzyl bromide followed by selective
`bromination of the methyl group in the 4 position yielded the
`crude bromomethyl-containing compound. Subsequent sub-
`stitution of
`the alkyl bromide in the presence of sodium
`
`4650
`
`DOI: 10.1021/acs.jmedchem.5b00217
`J. Med. Chem. 2015, 58, 4648−4664
`
`Page 3 of 17
`
`

`

`Journal of Medicinal Chemistry
`Scheme 3a
`
`Article
`
`aReagents and conditions: (a) Me3Al, Pd(PPh3)4, THF, 75 °C, overnight; (b) BBr3, DCM, rt, 4 d; (c) EtMgCl, Fe(acac)3, THF, rt, overnight; (d)
`Me3Al, Pd(PPh3)4, THF, 70 °C, 2 d; (e) BBr3, DCM, rt, 4 d; (f) BnBr, NaH, THF, 0 °C to rt, overnight; (g) Br2, CHCl3, 0 °C to rt, overnight, then
`NaOMe, MeOH, 90 °C, 12 h; (h) Pd-C, H2 gas, EtOAc, rt, 1 h.
`
`methoxide afforded methoxymethyl product 48. Deprotection
`of the benzyl group gave desired product 49.
`Pharmacology. Our efforts were primarily aimed at
`reducing the TDI effect on CYP3A and improving the aqueous
`solubility. The TDI effect was believed to be related to the
`demethylation of the 4-OCH3 or 3-OCH3 groups on the A-
`ring, where a second oxidative metabolism step may produce
`quinone intermediates that can react with nucleophiles.38
`Therefore, to resolve the TDI and solubility issues, we tried to
`introduce a substituted hetero Ar into the A-ring position
`instead of 3,4-di-OMe-Ph. According to our previous work, the
`methoxy groups are critical pharmacophores; the orientations
`of the small
`lipophilic group and lone pair were shown to
`strongly influence in vitro binding affinity.36 On the basis of the
`information regarding 3,4-di-OMe-Ph in the Cambridge
`Structural Database (CSD),
`the Me groups face opposite
`directions (Figure 2). Therefore, two types of dimethylpyr-
`idines (50 and 51) were designed with the hope of imitating
`the pharmacophores of 3,4-di-OMe-Ph, the lone pair and small
`lipophilic group (Table 2).
`in vitro binding
`Experimentally, dramatically different
`affinities were observed between 50 and 51, as 3-pyridine-
`containing 50 exhibited a moderate affinity, whereas
`the 4-pyridine-containing 51 resulted in
`introduction of
`reduced in vitro binding affinity toward both OX1R and
`OX2R. From these results, computational simulations of
`superposition using Molecular Operating Environment
`(MOE, Ryoka System, Inc., Tokyo, Japan) were conducted
`with (−)-5 and 50 or 51, with a particular focus on the position
`of the three aromatic rings and the di-OMe substituents (Figure
`2). Accordingly, we considered the superposition of
`the
`aromatic rings of 50 and 51 with (−)-5, together with the
`overall molecular shape including directionality of the ether
`linker, the angle of the upper-left aromatic ring (C-ring), and
`the direction of the NH of amide bond. On the basis of these
`observations, it is reasonable to exhibit better in vitro binding
`affinity for 50 than 51. We successfully converted 3,4-di-OMe-
`Ph to a hetero Ar group with reasonable activity, and 50
`showed improved solubility. This finding encouraged us to
`
`Figure 2. Overlay of (−)-5 and 50 and 51.
`
`further investigate hetero Ar rings other than dimethylpyridine
`to improve the parameters, although the compound did not
`show an improvement in the TDI.
`Dimethyl pyrazole 52, which was expected to meet the
`structural requirements, was synthesized and it showed an
`affinity similar to that shown by 50 as well as a significant
`improvement in the TDI. To further increase the affinity,
`various substituents were installed at the 1-position of pyrazole
`52, owing to the results of the computational simulations.
`Ethyl-substituted 53 was thought
`to have more favorable
`interactions with the small lipophilic site, and it exhibited an in
`vitro binding affinity comparable to that of (−)-5. This racemic
`mixture could be resolved by chiral high-performance liquid
`chromatography (HPLC) to provide (+)-53 and (−)-53. The
`
`4651
`
`DOI: 10.1021/acs.jmedchem.5b00217
`J. Med. Chem. 2015, 58, 4648−4664
`
`Page 4 of 17
`
`

`

`Journal of Medicinal Chemistry
`Table 2. A-Ring Modifications: Compounds 50−53 and 19−22
`
`Article
`
`aKi values are calculated from single experiments run in triplicate. bMeasured in Dulbecco’s PBS by DMSO solution precipitation method.36 cTime-
`dependent inhibition using a cocktail of probe substrates with human liver microsomes. dP-Glycoprotein (P-gp) transport assay.36 eDerived from
`rac-epichlorohydrin. fDerived from (R)-epichlorohydrin.
`
`two stereoisomers exhibited dramatically different receptor
`affinities; the (−)-53 isomer exhibited potent affinity toward
`both OX2R and OX1R, while the (+)-53 isomer did not.
`it was found that (−)-53 could be synthesized
`Notably,
`stereoselectively from (R)-epichlorohydrin using the procedure
`shown in Scheme 1,36 suggesting that (−)-53 had the chiral
`configuration of (1R,2S), which is the preferred configuration
`for OXR antagonist activity.
`Another approach for enhancing the in vitro receptor binding
`affinity was attempted in diazine derivatives, as exemplified by
`pyrimidine- or pyrazine-containing compounds 19 and 21.
`Specifically, this approach was carried out to modulate the pKa
`of the ring because the di-OMe phenyl moiety of (−)-5 is
`nonbasic. As a result, a 5-fold increase in the in vitro binding
`affinity was achieved with compound 19. Compound 20, which
`contained an Et group instead of the Me group in 19, was
`synthesized with the same expectations as we had for Et
`substituted pyrazole (−)-53;
`it exhibited increased in vitro
`binding affinity compared to that shown by 19 and exhibited an
`affinity comparable to that of (−)-5. Moreover, the pyrazole
`and pyrimidine derivatives, especially 19,
`showed some
`improved properties compared to those of (−)-5, namely a
`
`reduced TDI (96% for 19 vs 49% for (−)-5) and aqueous
`solubility: 80 μM for 19 vs 15 μM for (−)-5. Pyrazine 21
`exhibited a similar affinity to those of the pyrimidine derivatives
`but exhibited a concomitant deterioration in certain aspects of
`its physicochemical and drug likeness such as its aqueous
`solubility and reversible inhibition of CYPs (data not shown).
`Further reduction of the lipophilicity by the introduction of an
`OMe group in the 4-position of the dimethyl pyrimidine, as
`exemplified in 22, resulted in a stronger binding affinity than
`that of 19. However, 22 served as a substrate for P-glycoprotein
`(P-gp) with a corrected flux ratio of 4.0. Evaluation of the SARs
`of the A ring revealed that dimethyl pyrimidine-containing 19
`possessed the optimal profile in terms of overall balance
`because it exhibited a low lipophilicity (ClogP of 3.3 for 19 vs
`3.9 for (−)-5) with acceptable affinity toward OX2R, improved
`TDI and solubility, and moderate liability in a P-gp transport
`assay. Therefore, compound 19 was selected for optimization of
`the B-ring moiety.
`First, 2-, 3-, and 4-pyridine were introduced to confirm the
`SAR regarding the position of nitrogen, as a comparison with
`our previous work (Table 3).36 As a consequence, increased P-
`gp susceptibility was observed in 3- and 4-pyridine-containing
`
`4652
`
`DOI: 10.1021/acs.jmedchem.5b00217
`J. Med. Chem. 2015, 58, 4648−4664
`
`Page 5 of 17
`
`

`

`Journal of Medicinal Chemistry
`Table 3. Results of B-Ring Exploration: Compounds 23−32
`
`Article
`
`aKi values are calculated from single experiments run in triplicate. bTime-dependent inhibition using a cocktail of probe substrates with human liver
`microsomes. cP-gp transport assay.36 dSleep measurements conducted using an established method.36
`
`Figure 3. Wakefulness time measured after oral administration of 26, 28, and 31. The vehicle (10% Cremophor EL, 5% DMSO in saline; n = 4 for
`26, n = 7 for 28 and 31), 26 (n = 4), 28 (n = 5), and 31 (n = 6) at 30 mg/kg doses were administered immediately prior to the beginning of the dark
`cycle; cumulative wakefulness times were measured during the first 3 h after administration. Values are mean ± SEM.* P < 0.05, ** P < 0.01, and
`*** P < 0.001 versus vehicle control (unpaired parametric two-tailed t test).
`
`derivatives (24 and 25, respectively), with a close resemblance
`to previously identified SARs involving the derivatization of the
`di-OMe phenyl series.36 Subsequently, we evaluated the SARs
`of 2-pyridine-containing moieties in the B-ring, with the aim of
`enhancing the in vitro binding affinity and in vivo efficacy
`without adversely affecting other parameters. Introduction of
`substituents in the 5-position, such as in compounds 26 and 28,
`resulted in a greater in vitro binding affinity than that of 23. In
`contrast, the introduction of an Me substituent in the 4- or 5-
`position (29, 30) also resulted in increased in vitro binding
`
`affinity relative to that of 23 although the magnitude of the
`observed increase was smaller with an Me group than with a
`halogen. These results suggested that the basicity of the B-ring
`may influence the in vitro binding affinity. Accordingly,
`incorporation of fluorine in the 5-position and an Me
`substituent in the 4-position (31) resulted in increased in
`vitro binding affinity, which was more potent than that of
`compound 30. However, 31 exhibited decreased solubility
`compared to that of 26 (31, 51 μM, vs 26, 92 μM, at pH 7.4).
`
`4653
`
`DOI: 10.1021/acs.jmedchem.5b00217
`J. Med. Chem. 2015, 58, 4648−4664
`
`Page 6 of 17
`
`

`

`Journal of Medicinal Chemistry
`Table 4. Exploration of C-Ring with Different Combinations of B and C-Rings in Compounds 33−38
`
`Article
`
`aKi values are calculated from single experiments run in triplicate. bTime-dependent inhibition using a cocktail of probe substrates with human liver
`microsomes. cP-gp transport assay.36 dExploratory screening of inhibition of the human ether-a-go-go related gene (hERG) potassium channel.39
`
`Introduction of an OMe group in the 4-position (32) also
`elicited a strong increase in the in vitro binding affinity.
`With highly potent compounds 26, 28, and 31 in hand, in
`vivo characterizations were performed. The effects of these
`three compounds on wakefulness were evaluated in sleep
`experiments with mice via electroencephalogram/electromyo-
`gram (EEG/EMG) recordings during the first 3 h of the dark
`phase, as shown in Figure 3.36 All compounds elicited a
`significant reduction in the time of wakefulness following an
`oral dose of 30 mg/kg (Figure 3). Because of its well-balanced
`in vitro profiles taking into account the in vitro binding affinity,
`solubility, and other parameters, compound 26 was selected for
`optimization of the C ring.
`Previous synthetic efforts regarding the cyclopropane series
`were focused on modifying the A- and B-rings, while
`modifications of the C-ring were not explored. To investigate
`the SARs owing to changes in this component, a fluorine scan
`was conducted to explore favorable positions for substitutions
`(Table 4). Substitution at the 3-position (34) slightly increased
`the in vitro binding affinity relative to that of 26 without
`altering other parameters, while 2- or 4-F derivatives (33, 35)
`exhibited 2−3-fold decreases in the in vitro binding affinity or
`increased hERG inhibition. 3,5-Difluoro-containing derivative
`36 and 3,4-difluoro-containing derivative 37 exhibited strong in
`vitro binding affinities, but 37 also exhibited increased hERG
`inhibitory activity. Compound 38 was derived from 31, which
`exhibited strong in vivo efficacy and relatively weak hERG
`inhibitory activity (31: hERG IC50 > 10 μM). Compound 38
`exhibited an enhanced in vitro binding affinity compared to that
`of 26, with reduced hERG inhibition, as expected. However, 38
`exhibited decreased solubility (34, 74 μM, vs 38, 46 μM at pH
`7.4).
`On the basis of the SAR studies, compound 34 showed the
`most promising balance of in vitro properties and was selected
`
`for further profiling (Table 5). Compound 34 exhibited very
`weak reversible inhibition of CYP with IC50 values of >20 μM
`
`Table 5. In Vitro Properties of Selected Compound 34
`
`assay
`
`34
`
`3
`0.44
`6
`5.7
`1.6
`76
`
`hOX2R RBA Ki (nM)
`hOX2R cell-based functional assaya Ki (nM)b
`hOXIR RBA Ki (nM)
`hOXIR cell-based functional assaya Ki (nM)b
`corrected P-gp FRc (MDR1 flux ratio/PK1 flux ratio)
`TDId (CYP3A) (%)
`solubility (pM)e
`pH 1.2f
`pH 7.4g
`metabolic stability (residual ratio %)h
`90
`human
`aSee cell-based functional assay section in Experimental section. bKi
`values are calculated from single experiments run; cP-gp transport
`assay;36 dTime-dependent inhibition using cocktail of probe substrates
`eDMSO solution precipitation
`with human liver microsomes;
`method;36 fJapanese Pharmacopoeia 1st fluid; gDulbecco’s PBS;
`incubation at 0.1 μM in 0.1% DMSO with
`hMetabolism after
`human liver microsomes for 15 min.
`
`>100
`74
`
`for CYP1A2, CYP2C8, CYP2C9, CYP2C19, CYP2D6, and
`CYP3A. The aqueous solubility of 34 at room temperature was
`74 μM at pH 7.4 and >100 μM at pH 1.2. Compound 34
`showed weak TDI effects on CYP3A, measured as 76% of the
`control at 10 μM. Further evaluations of the possible risk of
`drug−drug interactions caused by CYP3A TDI are warranted,
`but the potential risk is expected to be marginal when the in
`vitro binding affinity and TDI screening data are considered.
`The pharmacokinetic properties of 34 were evaluated in male
`mice after intravenous and oral administration at 3 mg/kg iv
`
`4654
`
`DOI: 10.1021/acs.jmedchem.5b00217
`J. Med. Chem. 2015, 58, 4648−4664
`
`Page 7 of 17
`
`

`

`Journal of Medicinal Chemistry
`
`Article
`
`Table 6. Pharmacokinetic Parameters for 34 (iv and po) in Mice
`
`iv (1 mg/kg)
`
`po (10 mg/kg)
`
`mean
`1060
`2710
`1030
`2.16
`
`SD
`375
`1380
`368
`0.289
`
`PK parametera
`CL (mL/h/kg)
`Vss (mL/kg)
`AUC(0−inf) (ng·h/mL)
`1910
`2.20
`t1/2 (h)
`0.25
`tmax (h)
`488
`Cmax (ng/mL)
`F (%)b
`18.5
`aSee the Experimental section bF (%) = ((mean AUC(0−inf), po/dosepo)/(mean AUC(0−inf), iv/doseiv)) × 100.
`
`mean
`
`SD
`
`403
`1.00
`(0.25−0.50)
`25.2
`
`po (30 mg/kg)
`
`mean
`
`SD
`
`4130
`1.94
`1.00
`961
`13.4
`
`555
`0.585
`(0.50−1.00)
`127
`
`and 10 mg/kg po doses (Table 6). The results at 10 mg/kg po
`dose exhibited a plasma clearance of 1060 mL/h/kg, a half-life
`of 2.16 h for iv and 2.20 h for po administration, a Tmax of 0.25
`h, and an oral bioavailability of 18.5%. A Cmax exhibited 488 ng/
`mL for 10 mg/kg and 961 ng/mL for 30 mg/kg po
`In addition, brain penetration of 34 was
`administration.
`evaluated in mice after 1 and 3 h following a single oral
`administration of 10 mg/kg. After 1 h postdose,
`the
`concentration of 34 reached 32.8 nmol/L in CSF and 633.3
`nmol/L in plasma and reached 7.6 nmol/L in CSF and 157.6
`nmol/L in plasma at 3 h, with 85.2% of the compound bound
`indicating that 34 had sufficient brain
`to plasma protein,
`penetrability and exposure was clearly above Ki on the target
`(Tables 5 and 7). Therefore, 34 was subjected to efficacy
`evaluations through measurements of its effects on sleep in
`mice.
`
`Table 7. Mouse Brain Penetration Data of 34
`
`34 (10 mg/kg)
`CSF/plasmaa (nmol/L)
`
`1 h
`
`3 h
`
`633.3 ± 67.8
`32.8 ± 22.1
`157.6 ± 22.3
`7.6 ± 2.3
`PPB (%)b
`85.2
`C57BL/6N mice
`aSee the Experimental section. bSee the Experimental section.
`
`plasma
`CSF
`plasma
`CSF
`
`Compound 34 was administered orally at 10 and 30 mg/kg
`in mice, and the effect on sleep was evaluated (Figure 4).36
`Accordingly, 34 resulted in a significant decrease in time of
`wakefulness during the first 3 h following administration of 10
`and 30 mg/kg doses (77.1 and 75.3 min, respectively, for 34 vs
`116.7 min for vehicle, P < 0.001, Figure 4D). Both non-REM
`and REM sleep were increased following oral administration of
`10 and 30 mg/kg doses (99.3 min at 10 mg/kg and 95.5 min at
`30 mg/kg for 34 vs 62.6 min for vehicle, P < 0.01, Figure 4E;
`3.6 min at 10 mg/kg and 9.3 min at 30 mg/kg for 34 vs 0.6 min
`for vehicle, not significant and P < 0.001, respectively, Figure
`4F). Non-REM sleep-promoting effect in this sleep experiment
`was thought to be saturated at 10 and 30 mg/kg dose by taking
`of the PK results into consideration (Table 6). These data
`the compound 34 possessed robust sleep-
`suggested that
`promoting effects36 and potential as a drug to treat sleep
`disorders such as insomnia.
`Interestingly, the effects on sleep architecture were different
`between lead compound (−)-536 and lead optimized
`compound 34 (Figure 4). Especially, 34 caused a dose-
`dependent increase of REM sleep time, while (−)-5 did not
`provoke REM-sleep even up to highest dose (100 mg/kg po).36
`The difference of
`the effects on REM sleep by these
`
`compounds probably can be explained by the difference of
`OXRs selectivity profiles, (−)-5 is more OX2R selective
`antagonist than 34, which is nearly dual antagonist (about 3-
`fold selective for OX2R). Our results strongly indicate that
`REM sleep increase is caused via OX1R antagonism. This result
`is in alignment with orexin receptor KO mouse studies,40,41
`which demonstrate that both receptors are involved in sleep/
`wake regulation, with OX2R more regulating non-REM sleep
`and OX1R more regulating REM sleep. In addition, another
`sleep study, where orexin receptors were functionally inabled by
`intervention by applying OX2R selective or
`pharmacological
`dual orexin receptor antagonists, has indicated similar results
`regarding the roles of OX1R and OX2R for sleep architecture.42
`Thus, in our humble opinion, the selectivity ratio for OX1R and
`OX2R can clearly explain the sleep architecture induced by both
`compounds. When clinical data using a SORA will become
`available in the future, it will be important for deepening our
`understanding of the predictability of rodent’s sleep architec-
`ture data for humans. Nonetheless, we believe that both
`receptors should be targeted to promote physiological sleep by
`increasing both non-REM and REM sleep. Therefore,
`compound 34 is thought to induce a more favorable sleep
`architecture than (−)-5.
`■ CONCLUSION
`In conclusion, we optimized the structure of cyclopropane
`compounds based on (−)-5 as a lead OXR antagonist
`compound. Modifications of
`the A, B, and C-ring, and
`appropriate combinations of these alterations, afforded several
`potentially valuable compounds. Compared to (−)-5, signifi-
`cant improvements in multiple aspects were achieved by our
`efforts. Specifically, 34 exhibited an overall
`improvement in
`other properties such as the TDI of CYP3A and aqueous
`solubility, which were previously recognized as the major
`drawbacks of (−)-5. Additionally, 34 exhibited high OXRs
`antagonist activity and sufficient brain penetration. Further-
`more, 34 demonstrated high efficacy in preclinical sleep
`experiments. On the basis of
`these findings,
`it has been
`selected as a candidate compound, E2006, for further clinical
`evaluations toward the treatment of sleep disorders. Additional
`preclinical and clinical studies regarding this series will be
`reported in due course.
`
`■ EXPERIMENTAL SECTION
`
`Chemistry. 1H NMR spectra were recorded on a Bruker Avance
`spectrometer (operating at 600 MHz) or Varian Mercury 400
`13C NMR spectra were
`spectrometer (operating at 400 MHz).
`recorded on a Bruker Avance spectrometer (operating at 150 MHz).
`Chemical shifts were calculated in ppm (δ) from the residual CHCl3
`signal at (δH) 7.26 ppm or DMSO signal at (δH) 2.50 and (δC) 77.0
`
`4655
`
`DOI: 10.1021/acs.jmedchem.5b00217
`J. Med. Chem. 2015, 58, 4648−4664
`
`Page 8 of 17
`
`

`

`Journal of Medicinal Chemistry
`
`Article
`
`Figure 4. Wake and sleep time measurements after administration of 34. (A−C) Vehicle (10% Cremophor EL, 5% DMSO in saline; n = 7) or 34
`(10 or 30 mg/kg; n = 6) were orally administered immediately prior to the dark cycle (lights off), and (A) wakefulness, (B) non-REM sleep, and (C)
`REM sleep were measured hourly. (D−F) Cumulative amounts of (D) wakefulness, (E) non-REM sleep, and (F) REM sleep times measured during
`the first 3 h following administration of vehicle or 34 (derived from the time courses shown in A−C). Values are mean ± SEM * P < 0.05, ** P <
`0.01, *** P < 0.001 versus vehicle control (one-way ANOVA followed by Dunnett’s multiple comparison test).
`
`ppm in CDCl3. Then 600 MHz 1H or 13C NMR data were processed
`using ACD Spectrus processor
`from ACD Laboratories. High-
`resolution mass spectra (HRMS) were recorded on a ThermoFisher-
`Scientific LTQ-Orbitrap XL spectrometer (using electrospray
`ionization).
`The purity of the tested biological compounds was determined by
`an analytical LC−MS or UPLC method and was found to be greater
`than or equal to 95%. LC−MS analyses were performed using a
`Shimadzu LCMS-2010 EV and UPLC analyses were performed using
`a ACQUITY UPLC. The tested biological compounds were not
`PAINS compound. Optical rotation(±) was measured by Shimadzu
`HPLC SCL-10A system with a corresponding chiral column and
`JASCO OR-2090 optical rotation detector. Column chromatography
`was carried out using a Hi-Flash column (40 μm, silica gel and NH-
`silica gel, Yamazen Corporation). Chemicals and solvents were
`purchased from commercial sources.
`(1S,5R)-1-Phenyl-3-oxabicyclo[3.1.0]hexan-2-one (13a).
`NaHMDS (85 mL, 2.0 M) was added dropwi