The Discovery, Preclinical, and Early Clinical Development of Potent

The Discovery, Preclinical, and Early Clinical Development of Potent...

0 downloads 276 Views 7MB Size

This is an open access article published under an ACS AuthorChoice License, which permits copying and redistribution of the article or any adaptations for non-commercial purposes.

Drug Annotation pubs.acs.org/jmc

The Discovery, Preclinical, and Early Clinical Development of Potent and Selective GPR40 Agonists for the Treatment of Type 2 Diabetes Mellitus (LY2881835, LY2922083, and LY2922470) Chaﬁq Hamdouchi,*,† Steven D. Kahl,† Anjana Patel Lewis,† Guemalli R. Cardona,† Richard W. Zink,† Keyue Chen,† Thomas E. Eessalu,† James V. Ficorilli,† Marialuisa C. Marcelo,† Keith A. Otto,† Kelly L. Wilbur,† Jayana P. Lineswala,† Jared L. Piper,† D. Scott Coﬀey,† Stephanie A. Sweetana,† Joseph V. Haas,† Dawn A. Brooks,† Edward J. Pratt,‡ Ruth M. Belin,† Mark A. Deeg,† Xiaosu Ma,† Ellen A. Cannady,† Jason T. Johnson,† Nathan P. Yumibe,† Qi Chen,† Pranab Maiti,§ Chahrzad Montrose-Raﬁzadeh,† Yanyun Chen,† and Anne Reifel Miller† †

Lilly Research Laboratories, A division of Eli Lilly and Company, Lilly Corporate Center, DC: 0540, Indianapolis, Indiana 46285, United States ‡ Eli Lilly and Company, 237994 Singapore § Jubilant Biosys Research Center, 560 022 Bangalore, India S Supporting Information *

ABSTRACT: The G protein-coupled receptor 40 (GPR40) also known as free fatty acid receptor 1 (FFAR1) is highly expressed in pancreatic, islet β-cells and responds to endogenous fatty acids, resulting in ampliﬁcation of insulin secretion only in the presence of elevated glucose levels. Hypothesis driven structural modiﬁcations to endogenous FFAs, focused on breaking planarity and reducing lipophilicity, led to the identiﬁcation of spiropiperidine and tetrahydroquinoline acid derivatives as GPR40 agonists with unique pharmacology, selectivity, and pharmacokinetic properties. Compounds 1 (LY2881835), 2 (LY2922083), and 3 (LY2922470) demonstrated potent, eﬃcacious, and durable dose-dependent reductions in glucose levels along with signiﬁcant increases in insulin and GLP-1 secretion during preclinical testing. A clinical study with 3 administered to subjects with T2DM provided proof of concept of 3 as a potential glucoselowering therapy. This manuscript summarizes the scientiﬁc rationale, medicinal chemistry, preclinical, and early development data of this new class of GPR40 agonists.

■

INTRODUCTION The incidence of type 2 diabetes mellitus (T2DM) has increased signiﬁcantly in recent years, becoming a serious health care problem, with an estimated 422 million adults living with diabetes worldwide in 2014 compared to 108 million in 1980.1 Although several oral treatments for T2DM are available, some of them have been associated with undesired adverse eﬀects such as hypoglycemia, liver damage, gastrointestinal symptoms, and weight gain. Therefore, new, safe treatment options based on novel mechanisms are desired to eﬀectively treat or prevent diabetes in more patients. The G protein-coupled receptor, GPR40, also known as FFA1 or FFAR1 (free fatty acid receptor 1) is highly expressed in pancreatic β-cells and responds to endogenous medium and long chain unsaturated fatty acids, resulting in ampliﬁcation of insulin secretion only in the presence of elevated glucose levels.2−6 The glucose dependency of insulin secretion makes this receptor an excellent target for developing eﬃcacious therapies with a desired safety proﬁle for use in the © XXXX American Chemical Society

treatment of T2DM. Although the mechanism of action of GPR40 is not completely understood, previous disclosures provided evidence that GPR40 is predominantly coupled with the G protein α-subunit of the Gq family (Gαq).7−9 Activation of Gαq protein-coupled receptors triggers an increased in phospholipase C (PLC) activity. The latter induces an inositol 1,4,5-triphosphate (IP3)-mediated intracellular calcium mobilization and protein kinase C (PKC) activation which are known to be linked to enhanced insulin secretion in pancreatic β-cells.10 In addition to pancreatic β-cells, GPR40 is also expressed in the enteroendocrine cells of the gastrointestinal tract, with activation resulting in the secretion of incretins such as glucagon like peptide 1 (GLP-1) and glucose dependent insulinotropic polypetide (GIP).11,12 Therefore, activation of GPR40 has the potential to modulate insulin secretion not only directly in Received: June 15, 2016 Published: October 17, 2016 A

DOI: 10.1021/acs.jmedchem.6b00892 J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry

Drug Annotation

Figure 1. Representative GPR40 Agonists. ClogP values were computed using Chemaxon software.

Figure 2. GPR40 membrane receptor model includes three characterized binding sites31 and multiple signaling pathways.

pancreatic β-cells but also indirectly through the regulation of incretin secretion.13 Because of the potential eﬃcacy and desirable safety proﬁle compared to existing therapies such as insulin and sulfonylureas, considerable eﬀort has been focused on the development of small molecule GPR40 agonists.7,9 A range of synthetic agonists based on scaﬀolds such as alkanoic acids and their bioisosteres have been reported to stimulate insulin secretion in a glucose dependent manner and correct impaired glucose tolerance in rodents, suggesting that GPR40 agonists would be novel and potential insulinotropic drugs (Figure 1).2,14−22 More recently, phase 2b data provided clinical validation of glucose-lowering in patients with T2DM. In early stage clinical trials, TAK-875 (4) showed potent glucose-lowering eﬀects. The oral administration of 4 for 12 weeks in patients with T2DM resulted in HbA1c reduction superior to historical data with the DPPIV class of inhibitors and comparable with that of sulfonylureas with a lower

incidence of hypoglycemia. No GLP-1 secretion was detected in a small, early stage study.14,23−25 However, in December 2013, 4 was discontinued in phase 3 due to liver-associated adverse events in patients.26 Despite that setback, there is still a high level of interest in this target and a number of pharmaceutical companies are believed to have active programs. The basic pharmacophore of GPR40 agonists is derived from medium to long chain fatty acids such as eicosapentaenoic acid (EPA), docosahexaenoic acid (DHA), and linoleic acid. Historically, synthetic GPR40 agonists were designed to mimic the fatty acid structure with an acidic headgroup and a hydrophobic tail. The β-position to the carboxylic acid group can be substituted by small residues or cyclized to an aromatic ring (Figure 1). Compounds are generally highly lipophilic and relatively planar, resulting from free fatty acid derivatives used as initial leads and from the inherent lipophilic nature of the GPR40 binding site.27−29 B

DOI: 10.1021/acs.jmedchem.6b00892 J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry

Drug Annotation

Figure 3. Hypothesis-driven identiﬁcation of spiropiperidine seed molecule 6 with an acceptable potency in G protein signaling assays.

Figure 4. Proposed metabolites following incubation of 6 in rat, dog, monkey, and human hepatocytes.

acceptable potency in G protein signaling assays, selectivity against PPARs, and ClogP of 3.3 (Figure 3). Because GPR40 is a Gαq-coupled GPCR, evaluation of the aﬃnity of compounds was initially performed with a calcium ﬂux primary assay using ﬂuorescence imaging plate reader (FLIPR) technology. Other groups have reported the use of GPR40 aequorin assays in addition to FLIPR to measure intracellular calcium changes.34 As the program advanced and we faced the lack of in vitro/in vivo connectivity, we developed competitive radioligand binding assays, a β-arrestin agonist assay to assess this non-G-protein mediated signaling response, and an IP-1 agonist assay to measure Gq-signaling using longer compound incubation times than calcium ﬂux. We also investigated receptor−ligand binding kinetics to better understand the structure−activity relationship (SAR) of these GPR40 ligands from a drug−target residence time perspective. The in vitro metabolism of these chemical starting points using liver microsomes and hepatocytes suggested multiple sites of oxidation, including β-oxidation at the headgroup, O-dealkylation of the benzyl ether, and oxidation followed by glucuronidation (Figure 4). We found that the hepatocyte metabolism was the rate-limiting step in the hepatic clearance of

Some of the historical challenges in the development of GPR40 agonists resulted from high plasma protein binding of highly lipophilic synthetic ligands and usually high intrinsic clearance predominantly through phase 2 hepatocyte conjugative metabolism that can result in low free drug concentration at the target.30 Furthermore, multiple binding sites, with diﬀerential signaling pathways and the lack of selectivity against peroxisome proliferator-activated receptors (PPARs), can complicate the in vitro/in vivo correlation needed to rationally design an optimal therapeutic agent. Three distinct binding sites (Figure 2) have been described for GPR40 receptor agonists: one which binds the endogenous ligands such as DHA, one which binds compound 4 at the allosteric 1 site, and one which binds the reported full agonist AM-1638 at the allosteric 2 site.13,31−33

■

COMPOUND DESIGN, LEAD IDENTIFICATION, AND OPTIMIZATION Hypothesis-driven structural modiﬁcations of the endogenous ligand free fatty acid focused on the search for sites where polar functionalities would be tolerated with a more conformationally constrained and functionalized analogue of DHA led to the identiﬁcation of spiropiperidine seed molecule 6 with an C

DOI: 10.1021/acs.jmedchem.6b00892 J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry

Drug Annotation

Table 1. In Vitro Activities of Spiropiperidine Phenyl Propanoic Acids (Acidic Headgroup)

Competitive inhibition using [3H]-4 with human GPR40 membranes and conﬁrmed with [3H]-1135 (see Figure 5C). bCalcium ﬂux with human GPR40. cβ-Arrestin recruitment with human GPR40. dβ-Arrestin recruitment with mouse GPR40. eβ-Arrestin recruitment with rat GPR40. fIP-1 measurement with human GPR40. gClogP and pKa values were computed using Chemaxon software a

subsequent reports on GPR40 agonists, we undertook a focused structure−activity relationship (SAR) around the acidic headgroup. Our initial eﬀorts focused on improving the in vitro potency and pharmacokinetic (PK) properties and eliminating the risk of PPAR activity. The carboxylic acid function is typically attached by a 2 carbon atom linker to an aromatic ring, and a 3aryl propionic acid is advantageous for activity. A broad range of substituted propionic acid analogues were made and examined for their in vitro and in vivo properties. We found that the substitution at the β-position of the carboxylic acid in these systems by a medium sized group such as the linear propargyl16,18 was optimal in terms of binding aﬃnity to the allosteric 1 site, Gprotein, and β-arrestin activities (Table 1).

this class of molecules. Therefore, hepatocyte clearance measurements were a paramount consideration during the optimization and selection of compounds. Compound 6 was investigated in rat, dog, monkey, and human hepatocyte incubations, and the relative abundance of the metabolites was compared based on the MS peak intensities. The results suggested greater abundance of the carboxylic acid E metabolite, formed via β-oxidation, compared to the less abundant Odealkylation (carboxylic acid B) and glucuronidation (metabolite A) products (Figure 4). Optimization of the initial hit series focused on a careful examination of the three key areas of 6: the acidic headgroup, the center linker, and the hydrophobic tail. Inspired by the D

DOI: 10.1021/acs.jmedchem.6b00892 J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry

Drug Annotation

Table 2. In Vitro Activities of Spiropiperidine Phenyl Propanoic Acids (Center Linker)

Competitive inhibition using [3H]-4 with human GPR40 membranes and conﬁrmed with [3H]-1135 (see Figure 5C). bCalcium ﬂux with human GPR40. cβ-Arrestin recruitment with human GPR40. dβ-Arrestin recruitment with mouse GPR40. eβ-Arrestin recruitment with rat GPR40. fIP-1 measurement with human GPR40. gClogP and pKa values were computed using Chemaxon software. a

β-Substitution was also found to be a key to addressing metabolism at this position. In vitro and in vivo metabolism

results showed no detectable metabolites that were likely derived from the β-oxidation pathway. However, the O-dealkylation E

DOI: 10.1021/acs.jmedchem.6b00892 J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry

Drug Annotation

Table 3. In Vitro Activities of Spiropiperidine Phenyl Propanoic Acids

Competitive inhibition using [3H]-4 with human GPR40 membranes and conﬁrmed with [3H]-1135 (see Figure 5C). bCalcium ﬂux with human GPR40. cβ-Arrestin recruitment with human GPR40. dβ-Arrestin recruitment with mouse GPR40. eβ-Arrestin recruitment with rat GPR40. fIP-1 measurement with human GPR40. gClogP and pKa values were computed using Chemaxon software. a

being much more potent than the corresponding R-isomers. Thus, we pursued our chemical optimization by employing only

product remained as a potential risk. Stereochemistry at this position is critical with S-isomers such as compounds 1 and 9 F

DOI: 10.1021/acs.jmedchem.6b00892 J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry

Drug Annotation

Table 4. Proﬁling for Key Representatives of Spiropiperidine Compounds and Comparison to Reference Standards (Compound 4 and Compound 5)

Competitive inhibition using [3H]-4 with human GPR40 membranes and conﬁrmed with [3H]-1135 (see Figure 5C). bCalcium ﬂux with human GPR40. cβ-Arrestin recruitment with human GPR40. dβ-Arrestin recruitment with mouse GPR40. eβ-Arrestin recruitment with rat GPR40. fIP-1 measurement with human GPR40. gKinetic binding aﬃnity (Kd) and residence time were determined from competitive association experiments.38 h ClogP values were computed using Chemaxon software; pKa values were measured using potentiometric methods. iGlucose-dependent insulin secretion in mouse MIN6 cells or rat islet cells. “+” indicates insulin secretion above glucose alone level and “−“ indicates no observed insulin secretion above glucose alone level. jIPGTT intraperitoneal glucose tolerance test measured in mice. a

the more active enantiomer. Substitution of the aromatic ring with medium and large sized group led to diminished potency. Incorporation of a ﬂuorine atom as in 9 was tolerated, although with a nearly a 4-fold drop in potency and no identiﬁed advantage related to PK properties. Of particular interest was the observation that modiﬁcation of the acidic headgroup and the improvement in potency observed with the substitution at the βposition was not associated with increased lipophilicity.

Compounds generally showed ClogP in the 3−4 range, largely driven by the presence of the spiropiperidine moiety. Having established the S-β-1-propynyl benzenepropanoic acid as an optimal head piece, we turned our attention to minimizing the risk of O-dealkylation at the benzylic positions of the center region and seeking additional means to introduce polarity with the goal of decreasing overall lipophilicity. We prepared a number of alternative central cores designed to maintain the orientation of the terminal groups. Table 2 illustrates some key G

DOI: 10.1021/acs.jmedchem.6b00892 J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry

Drug Annotation

binding (PPB) using such in vitro assays did not predict the in vivo eﬃcacy of our compounds, no SAR attempt was made to increase free drug fraction (f u). Instead, structural modiﬁcations focused on optimizing the key variables that inﬂuence free drug concentration after oral administration such as fraction of dose absorbed and intrinsic clearance,30 which could be incorporating polarity to the highly lipophilic carboxylic acid pharmacophore and engineering the metabolic stability as will be discussed later in greater detail. In addition, the incorporation of spiropiperidines moiety into these compounds minimized the risk of activity against PPARs. Because ligands to PPAR α, δ, and γ are known to have therapeutic beneﬁts but with signiﬁcant risks,36 we decided to remove all PPAR activity from the GPR40 clinical candidates to avoid preclinical and potentially clinical adverse events. Compounds in Tables 1−4 were tested in all PPARα,δ,γ binding and functional assays at multiple concentrations up to 30 μM. The data demonstrated lack of signiﬁcant PPAR binding or functional activities. These observations were also conﬁrmed by an adipogenesis assay in 3T3-L1 cells. Replacement of the spiro[1H-indene-1,4′-piperidine] in 1 with 2,3-dihydrospiro[1H-indene-1,4′-piperidine] to minimize any potential risk associated with the formation of reactive intermediate such as glutathione conjugation resulted in compound 24, which maintained both GPR40 binding aﬃnity and agonist potency. Combining the 2,5-disubstituted thiophene central linker element with this spirocyclic tail piece aﬀorded compound 2, which maintained the in vitro activities and demonstrated an outstanding in vivo eﬃcacy (Table 4). Replacement of the dihydrospiro[1H-indene-1,4′-piperidine] with dihydro-1′Hspiro[indole-3,4-piperidine] provided compounds 25, 26, and 27 of equivalent potency and improved solubility. Substitution of the phenyl ring of the indoline moiety with small size electronwithdrawing lipophilic groups is preferred over electronreleasing (OMe) or neutral (H) substituents. More lipophilic and bulkier substituents such as isopropyl or phenyl (26 and 27) were also well tolerated. The incorporation of more polar group such as dihydro-1′H-spiro chromene-2,4′-piperidine in compound 32 (clogP = 2.7) to test the polarity boundaries was accompanied by slight decreases in binding aﬃnity (4-fold), Ca2+ mobilization (7-fold), and human β-arrestin (3-fold).

examples. Substitution at the para position of 6 was critical. Alternative (o- or m-) substitutions leading to nonlinear orientations of head and tail groups either dramatically reduced or eliminated the GPR40 activity (data not shown). Substituted phenyl with medium sized lipophilic groups showed comparable aﬃnity in the human GPR40 binding assay. For instance, addition of chlorine, methyl, or triﬂuoro methyl groups to the aryl ring resulted in compounds 13, 15, and 16, which demonstrated enhanced human β-arrestin potency compared to compound 1 while showing a slight potency decrease in the human GPR40 Ca2+ release assay. The 3-methoxy phenyl analogue (14) showed signiﬁcant improvement in physicochemical properties but also demonstrated signiﬁcant (Ki = 6.3 μM) activity at the human ether-a-go-go (hERG) channel. Furthermore, eﬀort to introduce polarity to the center linker in these systems was not compatible with GPR40 potency. It was therefore decided to limit the SAR to lipophilic groups in this area. A number of heterocycles were also evaluated at the center linker. The most interesting were thiazoles, thiophenes, and furans which reduced the extent of O-dealkylation while maintaining on-target potency (Tables 2 and 4). The incorporation of a thiophene linker (2, 18, 19) appeared to suppress O-dealkylation, however N-dealkylation then was observed as a signiﬁcant metabolic pathway. Compounds 1, 2, and 18 were investigated in rat, dog, and human hepatocyte incubations, and the abundance of metabolites were compared based on their relative MS peak intensities. Compound 1 showed moderate metabolism in human, rat, and dog hepatocytes with the O-dealkylation (carboxylic acid B shown in Figure 4) product being most abundant, whereas compounds 2 and 18 showed very minor metabolism in rat and human hepatocytes and moderate metabolism in dog with the glucuronidation, N-dealkylation, and taurine formation being the major products. Thiazoles 20, 21, 22, and 23 and furan 17 were found to be signiﬁcantly less potent in the human β-arrestin recruitment assay compared to corresponding thiophenes 18 and 19, although they have comparable potencies as determined employing the human GPR40 Ca2+mobilization format. At this point, it became apparent that human Ca2+ FLIPR response did not correlate with either βarrestin or binding aﬃnity. Although there was a trend between β-arrestin and binding aﬃnity, an adequate correlation was lacking. The question of which of these in vitro parameters might best predict in vivo pharmacology will be discussed later in greater detail. After identiﬁcation of the optimal structural features of the acidic headgroup and the center linker needed for GPR40 agonist potency, we returned to optimization of the tail portion. We found that the tail group can be used to eﬀectively modulate the overall physicochemical properties of these ligands. To introduce polarity into the hydrophobic tail, we explored a broad range of structurally distinct spiropiperidines. Furthermore, it was earlier hypothesized that the spirocyclic substructure might preclude these molecules from adopting a planar conformation that would result in reducing the aﬃnity of these compounds to serum proteins. Indeed, a range of spiropirperidines with various ring size showed good activity in human, mouse, and rat GPR40 in vitro assays (Table 3), with no signiﬁcant shift in potency being observed in a modiﬁed Ca2+ mobilization assay in which 0.1% of serum albumin was present. In spite of the minimal shift observed with and without serum albumin, it should be noted that all GPR40 compounds tested were highly protein bound in plasma (>99%) based on equilibrium dialysis. Because plasma protein

■

MECHANISTIC STUDIES: SIGNALING PATHWAYS, LIGAND BINDING, AND DRUG−TARGET RESIDENCE TIME During the course of this program, an extensive exploratory eﬀort was directed toward an increased understanding of the parameters that inﬂuence GPR40 pharmacology. The aim was to establish a robust in vitro/in vivo correlation that would enable the rational design of a potent and selective GPR40 agonist with preclinical eﬃcacy, safety, and PK proﬁles to allow full exploration of human eﬃcacy and safety. To this end, our initial SAR was guided by compound activity in Ca2+ mobilization assay using HEK293 cells overexpressing GPR40. Tables 1−4 show the eﬀect of compounds on this measure of agonist potency. As the eﬀort progressed, a lack of correlation between in vitro calcium measurements and in vivo activity became evident, illustrated by examples in Table 4. Compound 11 had Ca2+ mobilization potency and in vivo exposure similar to a number of other compounds (e.g., 1, 2, and 18). While 1, 2, and 18 demonstrated glucose-dependent insulin secretion (GDIS) activity in insulin secreting cells and robust in vivo activities in a mouse intraperitoneal glucose tolerance test (IPGTT; ED90 = 0.58, 3.67, and 5.6 mg/kg, respectively), compound 11 was H

DOI: 10.1021/acs.jmedchem.6b00892 J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry

Drug Annotation

Figure 5. Radioligand binding characterization of GPR40 agonists. (A) Percent speciﬁc inhibition of [3H]-4 binding to GPR40 with unlabeled compound 1, 2, or 3. Shown is the average of three independent runs (error bars = SEM). (B) Dissociation of [3H]-11 bound to GPR40 was initiated with unlabeled compound 4 at various concentrations. The percent of the maximum response prior to dissociation was measured over time. Shown are the results from a single representative experiment. (C) Log binding aﬃnities (pKi) for compounds (n = 126) tested in [3H]-4 (y-axis) or [3H]-11 (x axis). The black solid line is the best ﬁt linear correlation (R2 = 0.87), and the red dashed line is the 45° line of agreement. Data points represent single tests for each compound. (D) Homologous kinetic dissociation (percent maximum response) of [3H]-11 binding to GPR40 with unlabeled compound 11 (●) or [3H]-4 with unlabeled compound 4 (■). Data were modeled to single phase dissociation curve ﬁts to determine the dissociation rate constant (koff) and calculate the drug−target residence time (1/koff). Data points are the mean of three replicates within a single experiment (error bars = SEM). (E) Kinetic association of [3H]-11 in the presence of 0 nM (▼), 8 nM (▲), 25 nM (■), or 75 nM (●) unlabeled compound 4. Shown are single data points for each time measurement. These data are ﬁt to the kinetics of competitive binding equation (GraphPad Prism, v6.03) to determine the drug− target residence time for the added competitor, as previously described.38

glucose lowering and ∼4 logs for β-arrestin). Agonists with potent β-arrestin activity demonstrated strong glucose lowering, whereas those with weak β-arrestin activity were inactive in vivo regardless of their calcium ﬂux properties. Several diﬀerences between β-arrestin and calcium ﬂux assays could explain this trend. In addition to the G-protein versus non-G-protein signaling aspects, kinetics of these measurements are signiﬁcantly diﬀerent. While calcium ﬂux measurements are transient and conducted within 3 min of compound addition, β-arrestin activity is measured after longer compound incubation periods (90 min) and is less transient in nature and nearer to actual equilibrium conditions. To investigate whether compound treatment time was responsible for the observed apparent ligand-biased signaling, we developed an additional Gq-coupled assay that measured IP-1 accumulation 2 h after agonist stimulation. Compounds listed in Table 4 that were eﬀective in cellular insulin secretion (GDIS) or in vivo glucose lowering (IPGTT) also had potent IP-1 responses (∼1 nM EC50). Similar to β-arrestin activity trends, compounds that were inactive in insulin secretion or in vivo glucose lowering had higher EC50 values in the IP-1 accumulation assay. However, IP-1 potency did not correlate directly with in in vivo activity. Compounds active in vivo showed a narrow range of potent IP-1 responses that were not signiﬁcantly diﬀerent from each other. Conversely, agonist responses measured by β-arrestin recruitment were more consistently aligned with in vivo activity. This observation

devoid of in vivo activity (IPGTT; ED90 > 100 mg/kg). Likewise, the potencies of 6 and 4 in the Ca2+ mobilization assay were comparable (EC50 = 180 and 160 nM, respectively), but 6 was not active in cell-based GDIS studies or in vivo IPGTT (IPGTT; ED90 > 100 mg/kg), while 4 exhibited GDIS in cell-based assays and showed glucose lowering with an ED90 = 29.4 mg/kg. Similar disconnects were observed with a broader range of SAR compounds (Pearson’s correlation coeﬃcient r near 0 for glucose AUC % lowering and Ca2+ mobilization EC50 based on ∼140 compounds covering a range of ∼80% for glucose lowering and ∼3 logs for Ca2+ mobilization). This has triggered our extensive mechanistic studies and the development of a number of additional cell-based assays to understand the predictability of in vitro to in vivo activity. To evaluate potential ligand-biased signaling at GPR40, we developed assays that measure β-arrestin activity. We found that compounds had a high level of potency for the three species tested (human, rat, and mouse), suggesting that GPR40 can signal through both G-protein and a non-Gprotein mediated pathways.7−9 Moreover, compounds that lacked GDIS or in vivo mouse IPGTT eﬃcacy had lower βarrestin potency, an indication that β-arrestin measurement could be used to predict in vivo activity. A correlation analysis of a broader range of in vivo and in vitro data suggested that βarrestin yielded the best prediction of in vivo glucose lowering (Pearson’s r = 0.66 for glucose AUC % lowering and β-arrestin EC50 based on ∼30 compounds covering a range of ∼70% for I

DOI: 10.1021/acs.jmedchem.6b00892 J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry

Drug Annotation

Figure 6. In vivo tracer microdose diﬀerential distribution proﬁle for 5 (A) and 18 (B) single 3 μg/kg IV dose in 25% HP-BCD in fed male C57BL/6 mice, 40 min survival interval at which point mice were sacriﬁced by cervical dislocation. Trunk blood collected in EDTA coated Eppendorf tubes and stored on wet ice until study completion. Target organs rapidly removed and lightly rinsed with ice-cold sterile water. Target tissues dissected, weighed, stored in 1.5 mL Eppendorf tubes, and placed on wet ice until tissue extraction.

to calculate the duration of the binding interaction directly as the reciprocal of the equilibrium dissociation constant (koff), also known as the drug−target residence time (τR). This value may be useful in predicting sustained in vivo eﬃcacy.37 We found that the residence time for 4 (88 ± 10 min, n = 6) was signiﬁcantly longer than for 11 (22 ± 3 min, n = 4) and sought to determine and compare the residence time for our key molecules. To do this, association curves using [3H]-11 in the presence of several concentrations of unlabeled test compound were ﬁt to equations previously developed.38 An example of the resulting competitive association curves for [3H]-11 in the presence of unlabeled 4 is shown in Figure 5E. After analyzing a broad range of compounds (data not shown), we concluded that although there was no perfect correlation between residence time and in vivo pharmacology, a threshold τR of 30 min was required for a compound to exhibit in vivo eﬃcacy. For analogues within this series in which τR < 30 min, no GDIS was observed (Table 4, compounds 6 and 11). In general, compounds with potent βarrestin responses had residence time longer than 30 min, whereas compounds with residence times shorter than 30 min had weaker β-arrestin responses. With this additional knowledge, our GPR40 agonist SAR strategy focused on optimizing the βarrestin response and drug−target residence time along with PK to deliver a compound with an acceptable human eﬃcacious dose with once daily oral treatment. Despite having identiﬁed a trend for in vivo potency in the IPGTT assay with agonist potency measured by β-arrestin recruitment, exceptions were noted. We hypothesized that these deviations might be due to the diﬀerential tissue distribution of compounds based on their lipophilicity. We evaluated the tissue exposure in C57BL/6 mice for two representative GPR40 agonists: compound 5 (cLogP = 6.9; mouse GPR40 β-arrestin EC50 = 1.5 nM; mouse IPGTT ED90 = 1 mg/kg; plasma AUC at 1 mg/kg = 33859 ng·h/mL) and compound 18 (cLogP = 3.98; mouse GPR40 β-arrestin EC50 = 0.46 nM; mouse IPGTT ED90 = 3.7 mg/kg, plasma AUC at 3.7 mg/kg = 1055 ng·h/mL). Tracer microdosing allowed for subpharmacological dosing of com-

could be explained by the position of these responses relative to receptor activation. IP-1 activation is farther downstream and has the potential for signal ampliﬁcation, while β-arrestin measurement is closer to receptor activation and may reﬂect the agonist response of the assay more accurately. Thus, kinetics of the receptor/ligand interaction rather than biased signaling was potentially responsible for the diﬀerences observed in downstream eﬃcacy. Therefore, we became interested in understanding the ligand binding characteristics of the receptor and the drug−target residence time and their possible inﬂuences on in vivo activity. Toward this goal, we ﬁrst developed a radioligand binding assay with [3H]-4 (Kd = 4.2 nM) to ensure receptor interaction and to further characterize GPR40 agonists. Aﬃnities (Ki) of compounds in Tables 1−4 were determined by measuring competitive inhibition of [3H]-4 binding to membranes prepared from cells overexpressing recombinant human GPR40, illustrated by concentration response curves for clinical candidates 1, 2, and 3 (Figure 5A). However, when using recombinant overexpressed receptors, binding aﬃnity may not connect exactly with functional potency due to receptor expression levels, receptor occupancy, or ampliﬁcation of response. For example, the binding aﬃnities for 20, 21, and 22 did not parallel calcium mobilization, β-arrestin, or IP-1 functional responses (Table 2). Therefore, molecules from structurally distinct chemical classes were radiolabeled to better understand GPR40 pharmacology. We found that compound 4 did not alter [3H]-11 equilibrium dissociation kinetics at multiple concentrations (Figure 5B), indicating that both ligands bound the same site on the receptor. Binding aﬃnities for 127 compounds were determined using separate competitive binding assays with [3H]-4 or [3H]-11 (Figure 5C). The high linear correlation (R2 = 0.87) provided additional conﬁdence that both ligands could be used as probes for GPR40 binding. While investigating GPR40 binding kinetics, we noticed that 4 and 11 had diﬀerent steady-state dissociation responses (Figure 5D). Because both compounds were radiolabeled, we were able J

DOI: 10.1021/acs.jmedchem.6b00892 J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry

Drug Annotation

Figure 7. X-ray crystal structure of (A) GPR40 agonist, compound 4, EC50 = 159 nM, PDB code 4PHU.41 (B) A potent PPARγ agonist, EC50 = 3 nM; PDB code 2P4Y.42 Binding site surface is colored by electrostatic potentials. Predicted binding poses of 1 (green), 2 (yellow), and 3 (brown) by docking them to the crystal structure of GPR40-4 complex: (C) shows the whole structures of agonists, and (D) just highlights the head groups. Binding site surface is colored by electrostatic potentials. All drawings were done using MOE.43

4 (Figure 7D). Additional insights can be derived from inspection of the docking pose of analogue 6. Lack of substituents near the acid group led to a more than 10-fold weaker binding potency and near 20-fold decrease in EC50 compared to compound 1 (Table 4). Moreover, the rigidity of the propargyl may contribute to the slow koff rate of compound 1 (0.00822/min) versus compound 6 (0.0866/min) by making it more diﬃcult energetically for compound 1 to move out of the binding pocket. The openness of the GPR40 binding pocket allowed the incorporation of bulky groups such as spiro structure into the tail region (Figure 3a). In contrast, the large and wide binding pockets of PPARs are usually relatively ﬂat and enclosed inside the protein (Figure 7B); thus they cannot accommodate bulky and conformationally restricted polar groups. This observation helped to formulate our SAR strategy of introducing polar conformational restrained groups into the tail region (Figure 3). The strong selectivity of our compounds against the PPARs conﬁrmed our hypothesis. Unlike 4 with a polar tail group, compounds from the present series with more lipophilic tail groups appear to bind toward to the middle of the transmembrane domain or the center of the lipid membrane. Furthermore, our GPR40 agonists were highly selective when tested in selectivity panels against other GPCRs at DiscoveRX and CEREP.

pounds to determine organ exposures and plasma binding proﬁles. As shown in Figure 6, analogue 18 distributed diﬀerentially toward the GPR40 target rich spleen (a surrogate for the islet) and ascending colon relative to skeletal muscle and plasma at the two time points tested (20 and 40 min), indicating potential speciﬁc binding. In contrast, compound 5 demonstrated high plasma protein binding and no diﬀerential distribution toward the spleen or ascending colon. Using LCMS/MS to measure tracer levels in discrete tissue areas allowed an assessment of receptor occupancy of these two compounds. Also, LC-MS/MS detection eliminated the need for radiotracers to determine target occupancy; the resolution and selectivity of LC-MS/MS quantiﬁed only the compound of interest and eliminated the possibility of measuring metabolites that can be problematic in peripheral tissue analysis.39 To shed additional light on how GPR40 agonists may interact with the receptor, we developed a docking model for GPR40 agonists, initially by building a homology model of GPR40 based on the crystal structures of other class A GPCR targets40 and later reﬁned by the crystal structure of GPR40 ligated with 4 (PDB code 4PHU, Figure 7A). Our binding assay results indicated that compounds reported herein were allosteric agonists bound in the same pocket as compound 4 despite the presence of the much more hydrophobic tail group and diﬀerent headgroup in our series. Docking studies showed that compounds 1−3 can also ﬁt very tightly into this pocket (Figure 7A,C). As with compound 4, the carboxylic acid group in all three compounds formed strong H-bond interactions with Arg183, Arg258, Tyr91, and Tyr240. The alkyne moiety appeared to play important role in binding by making favorable van der Waals contacts and ﬁlling the small pocket near Ala182 and Leu186 not occupied by compound 4, consistent with higher potency of these compounds compared to

■

ADME, PHARMACOKINETICS, AND PHYSICOCHEMICAL PROPERTIES OF ADVANCED COMPOUNDS With the in vitro/in vivo connectivity established and selectivity achieved, we were able to use our testing paradigm to eﬃciently predict and meet the in vivo potency goals. Attention was then turned to the selection of a spiropiperidine that could be pursued K

DOI: 10.1021/acs.jmedchem.6b00892 J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry

Drug Annotation

Figure 8. Evolution of spiropiperidines: second-generation GPR40 agonists.

glucuronidation being the major pathway. No O- or Ndealkylation metabolites were detected when compound 3 was incubated in rat, dog, or human hepatocytes. Despite the modest improvement in solubility, 3 showed signiﬁcant improvement in the fraction of dose absorbed (DABS = 845 mg). Physical chemical properties of 1, 2, and 3 were assessed to identify potential issues that may inﬂuence absorption, formulation, and future development of the compounds. The solid-state properties of a zwitterionic form for each of the compounds were evaluated by thermal analysis, vapor pressure isotherms, and XRPD. The forms evaluated were anhydrous and nonhygroscopic, with melting onsets ranging from 117 to 167 °C (Table 5). In solution, the compounds were chemically stable when stressed at 40 °C for 24 h in phosphate buﬀer at pH 2 and 8. Solubility data generated on crystalline material indicated low aqueous solubility, with 3 exhibiting the highest solubility among the three compounds. The calculated human passive permeability (cPassive) for each compound was considered high relative to model compounds. Consistent with solubility trends, the estimated human absorbable dose in the fasted state based on microscopic mass balance (MiMBa) modeling ranged from 127 to 845 mg (Table 6). Fraction absorbed was predicted to be sensitive to particle size and particle size control would likely be warranted depending on dose. The potential clinical candidates 1, 2, and 3 were administered to mice, rats, and beagle dogs for determination of their pharmacokinetic properties. Blood was collected at selected time points with EDTA as the anticoagulant and processed to plasma for LC/MS/MS bioanalysis. Pharmacokinetic properties were determined by noncompartmental analysis. Only minor diﬀerences in the pharmacokinetic properties of the test compounds were observed. Low clearance and volume of distribution were observed in all species (Table 6). Oral bioavailability across species ranged from 64 to 74% for 1, 17 to 46% for 2, and 55 to 95% for 3. Plasma elimination half-lives ranged from approximately 2 to 5 h for compound 1, 2 to 4 h for compound 2, and approximately 4 to 6 h for compound 3. In rat and dog toxicology studies, exposures increased with increasing dose but these were less than dose proportional for all compounds. The compounds show gender diﬀerences in rats, with higher exposures in females than in males. Gender diﬀerences in exposure were not observed in dogs. Compound 3 exhibited a more sustained exposure and prolonged half-life across species compared to its spiro analogues 1 and 2, which suggested a

in clinical studies as well as the identiﬁcation of secondgeneration compounds (Figure 8). This process began with a full pharmacokinetic characterization of compounds 1, 2, 18, and 25 in three species along with the identiﬁcation of their associated metabolites. The in vitro metabolism of GPR40 compounds was determined in liver microsomes and hepatocytes. Metabolic pathways identiﬁed from LC/MS/MS analysis include oxidation at multiple sites, dealkylation (N and O), and conjugation (glucuronide and taurine), with hepatocyte metabolism being the rate-limiting step. Therefore, hepatocyte clearance measurements were a paramount consideration during the selection and prioritization of compounds. Compound 1 and 2 combined excellent in vivo potency with relatively lower eﬃcacious exposure compared to the reference compounds in Table 4 (IPGTT; ED90 = 0.58 and 5.6 mg/kg, plasma AUC at ED90 = 218 and 898 ng·h/mL, respectively) and acceptable rates of clearance. However, they suﬀered from higher than desired clearance to allow for the development of a compound with a sustained exposure to support once daily oral treatment in humans. With this in mind, we investigated a broad range of spiropiperidines and their direct analogue nonspiropiperidine compounds in rat pharmacokinetic studies and characterized their metabolites. This exercise allowed building a database to help select compounds with a lower projected hepatic clearance. In an attempt to reduce the extent of clearance related to N- and O-dealkylation and improve the solubility of the advanced spiropiperidines 1 and 2, a nonspiropiperidine related analogue 3 was designed while maintaining the same acidic headgroup and center region of 2. Compound 3 with 5-[(3,4-dihydro-8methoxy-1(2H)-quinolinyl moiety showed equivalent potency in the β-arrestin assays across species, acceptable residence time (τR = 46 min) that translated into an excellent pharmacology in mouse and rat as will be discussed later in greater detail (e.g., IPGTT in mouse, ED90 = 5 mg/kg; plasma AUC at 5 mg/kg = 4926 ng·h/mL) and improved metabolic stability (Table 5). Furthermore, the incorporation of quinolinyl with pKa2 = 6 contributes to reducing the zwitterionic character of spiropiperidines whose pKa2 ranges from 7.5 to 8.5 and increasing the polar surface area (tPSA = 50 for 1 and 2 vs tPSA = 59 for 3). Compound 3 was investigated in rat, dog, and human hepatocyte incubations, and the abundance of metabolites was compared based on the relative MS peak intensities. In contrast to 1 and 2, compound 3 showed only minor metabolism in rat and human hepatocytes and moderate metabolism in dog, with the L

DOI: 10.1021/acs.jmedchem.6b00892 J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry

Drug Annotation

Table 5. Drug Metabolism and Developability Assessment of Optimized GPR40 Agonists

a-c β-Arrestin recruitment with human, mouse and rat GPR40. d-fHuman PPARα functional agonism. eHuman PPARδ functional agonism. fHuman PPARγ functional agonism. gHuman Kv11.1 voltage gated potassium channel inhibition. hMetabolite identﬁcation in mouse, rat, and dog hepatocytes. iMetabolic stability; % metabolized following a 30 min incubation in mouse or human microsomes. jMicrosomal clearance in mouse and human microsomes. kHepatocyte clearance in mouse and human hepatocytes. lDiﬀerential scanning calorimetry. mCalculated human passive permeability. nCalculated human absorbable dose, fasted state.

greater potential for qd dosing (Table 6). In vivo metabolism of compounds 1, 2, and 3 was investigated in dog feces, plasma, and urine. The parent drug 3 was shown to be the major circulating component observed in plasma, and only one plasma metabolite was observed (parent + glucuronide). Compound 3 was also the major component observed in the urine and fecal samples. All

metabolites in urine and bile were minor based on MS peak intensities. This diﬀerentiated from compound 2, in which the parent drug was the major component observed in feces and plasma and the N-dealkylation metabolite was the major component observed in urine. Compound 1 showed that parent drug was the major component in plasma and the O-dealkylation M

DOI: 10.1021/acs.jmedchem.6b00892 J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry

Drug Annotation

Table 6. Mean Pharmacokinetic Parameters of 1, 2, and 3 in Mice, Rats, And Dogs 1

2

3

parametera

mouse

rat

dog

mouse

rat

dog

mouse

rat

dog

mg/kg dose iv/po Clp (mL min−1 kg−1) Vdss (L/kg) T1/2 (h) Cmax (ng/mL) Tmax (h) Foral (%)

1/10 20.4 1.4 2.1 1103 0.5 64

3/10 11.6 0.4 2.1 11203 0.25 71

1/5 4.9 0.8 5.05 2860 3.0 74

1/3 9.3 1.6 1.7 184 2 21

1/3 20 0.8 1.6 829 0.5 46

1/3 2.2 0.2 3.8 1100 2 17

1/3 15.4 3.9 4.2 552 1 62

1/3 3.8 1.2 5.6 4135 0.6 95

1/3 4.6 1.0 4.4 2290 0.5 55

a

Clp, plasma clearance; Vdss, volume of distribution at steady state; T1/2, terminal half-life; Cmax, maximum plasma concentration following oral dosing; Tmax, time to maximum concentration; Foral, oral bioavailability

as the major metabolite. This result was also consistent with the hepatocyte incubation results that were discussed earlier. In addition to their in vitro and in vivo potency, and good pharmacokinetic proﬁles, compounds 1, 2, and 3 were tested in selectivity panels against over 100 diﬀerent GPCRs, kinases, enzymes, and nuclear receptors and showed minimal activity (98%) and overall yield (55% over 4 steps) on 14 kg scale. The process to prepare 2 also required slight modiﬁcations to the discovery synthesis to enable larger quantities of API (Scheme 2). Nucleophilic substitution between bromothiophene 3947,48 and spirocyclic amine 4047 in MeCN using i-Pr2NEt as base gave 41 in good yield. The methyl ester was reduced using LiAlH4 in THF and then reacted with SOCl2 to provide chlorothiophene 43, which was isolated as an HCl salt. Another nucleophilic substitution between 43 and the resulting phenolate of 37 using Cs2CO3 in DMSO gave rise to 44, which was isolated from the reaction mixture as an oxalate salt. Hydrolysis using NaOH (aq), followed by recrystallization, completed the preparation of 2 in excellent purity (>98%) and overall yield (36% over 5 steps) on 34 kg scale. The process to prepare 3 was also modiﬁed to enable larger quantities of API (Scheme 3). Nucleophilic substitution between

Figure 12. GLP levels in lean mice following administration of 1, 2, and 3.

In preclinical pharmacology studies, compounds 1, 2, and 3 demonstrated the necessary potency, eﬃcacy, durability, and selectivity to justify clinical testing. All three compounds produced glucose-dependent insulin secretion in cell based systems and potent glucose lowering and insulin elevation in mouse IPGTT studies, normalizing glucose levels and conﬁrming the GPR40 mechanism. The compounds performed as potent, eﬃcacious, and durable treatments during OGTTs in insulin resistant rats. In addition, signiﬁcant GLP-1 secretion was seen in a rodent L-cell line and in normal mice following administration of 1, 2, or 3. These three clinical candidates were predicted to have an acceptable projected human eﬃcacious dose and sustained exposure to support a safe once daily oral Scheme 1. Synthesis of 1 on 14 kg Scalea

Reagents and conditions: (i) MeOH/THF, NaHCO3 (aq), 0 °C, 89%; (ii) LiAlH4, THF, 2-MeTHF, 10 °C, 93%; (iii) PPh3, DIAD, toluene, rt, 81%; (iv) NaOH, EtOH/H2O, recrystallized from n-PrOH, 82%.

a

P

DOI: 10.1021/acs.jmedchem.6b00892 J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry

Drug Annotation

Scheme 2. Synthesis of 2 on 34 kg Scalea

Reagents and conditions: (i)i-Pr2NEt, CH3CN, 50 °C, 82%; (ii) LiAlH4, THF, 0 °C, 89%; (iii) SOCl2, EtOAc/H2O, 94%; (iv) Cs2CO3, DMSO, 73%; (v) NaOH (aq), THF/EtOH, recrystallized from EtOH, 71%.

a

Scheme 3. Synthesis of 3 on 50 kg Scalea

Reagents and conditions: (i)i-Pr2NEt, NaHCO3, i-PrOAc/CH3CN, 45 °C, 69%; (ii) LiAlH4, THF, 0 °C, 95%; (iii) SOCl2, EtOAc/H2O; Cs2CO3, DMSO, 79%; (iv) NaOH (aq), THF, i-PrOH, recrystallized from acetone/n-heptane, 82%.

a

bromothiophene 3947,48 and tetrahydroquinoline 4548 in iPrOAc and CH3CN using i-Pr2NEt as base gave 46 in good yield. The methyl ester was reduced using LiAlH4 in THF and reacted with SOCl2 to provide chloro-thiophene 48. A second nucleophilic substitution between 48 and the resulting phenolate of 37 with Cs2CO3 in DMSO gave rise to 49, which was not isolated from the reaction mixture but reacted with NaOH (aq) to furnish 3. Crystallization from i-PrOH, followed by recrystallization from acetone and n-heptane, completed the preparation of 3 in excellent purity (>98%) and overall yield (42% over 4 steps) on 50 kg scale.

increased with increasing dose (Figure 13). Key pharmacokinetic parameters of the compounds in healthy volunteers and patients with T2DM are summarized in Tables 7 and 8. All three molecules were rapidly absorbed after oral administration. Peak plasma concentrations were observed soon after oral administration with Tmax values occurring at approximately 3−6 h. Compound 2 showed the highest apparent clearance, and 3 had the longest terminal half-life. The PK parameters at individual doses in T2DM patients with 3 are listed in Table 8. The analysis of the plasma glucose lowering during the single dose study with 3 in patients with T2DM suggests 12 h pharmacodynamics eﬀect (Figure 14). Compound 3 was prioritized to progress to a multiple ascending dose (MAD) study in patients with T2DM, as it appeared to provide the best 24 h PK and PD coverage and therefore the most likely to be administered as a once daily oral treatment (Figure 14). In the MAD study, 3 was well tolerated and dose escalation proceeded from doses of 60 to 1200 mg.

■

CLINICAL TRIAL RESULTS The safety proﬁle and tolerability of 1, 2, and 3 were assessed in humans in single ascending dose studies (SAD), followed by 4week multiple ascending dose studies (MAD) for 3. There were no dose-limiting adverse events. The pharmacokinetics in the SAD was dose proportional; the drug exposures in plasma Q

DOI: 10.1021/acs.jmedchem.6b00892 J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry

Drug Annotation

Table 8. Noncompartmental Pharmacokinetic Parameters Following a Single Oral Dose of 3 to Type 2 Diabetic Mellitus (T2DM) Subjects analyte = plasma 3 3 geometric mean (CV%)d dose (mg) Ne Cmaxb (ng/mL) Tmaxg (h) T1/2f (h) AUC(0−∞)a (ng·h/mL) CL/Fc (L/h) Vz/Fh (L)

270 6 2000 (37) 2.75 (1.50−6.00) 12.1 (39) 19400 (27) 13.9 (27) 244 (28)

540 6 3780 (56) 4.00 (1.50−4.00) 14.5 (47) 41400 (34) 13.0 (34) 273 (22)

1080b 5 7950 (30) 4.00 (4.00−4.00) 22.4 (18) 97300 (39) 11.1 (39) 358 (47)

a

AUC(0−∞) = area under the plasma concentration−time curve from time zero to inﬁnite time. bCmax = maximum observed drug concentration. cCL/F = apparent total body clearance following oral administration. dCV = coeﬃcient of variation. eN = number of subjects. fT1/2 = half-life. gTmax = time of maximum observed drug concentration. hVz/F = apparent volume of distribution during the terminal elimination phase.

Figure 13. Mean plasma concentrations of 3 following a single dose in T2DM subjects (top panel = linear scale, bottom panel = log scale). SD = standard deviation, Conc. = concentration, hr = hours.

Table 7. Mean Pharmacokinetic Parameters of 1, 2, and 3 in Humana geometric mean

1

2

3

Cl/F (L/h) T1/2 (h) Tmax (h)

10.6 7.27 3.25

30.6−43.4 8.89−14.1 4−6

11.1−13.9 12.1−22.4 2.75−4

a Cl/F, apparent plasma clearance; T1/2, terminal half-life; Tmax = time of maximum observed drug concentration.

Figure 14. Change from baseline arithmetic mean (+SD) pharmacodynamics glucose (mmol/L) concentrations on day 1 following a single dose in T2DM subjects.

Twice daily dosing was investigated in two cohorts of patients with T2DM. Statistical analysis results of change from baseline (day-1 AUC) blood glucose AUC(0−24) are shown in Table 9. There is a trend toward decreased glucose from baseline compared to placebo at dose levels of 200 mg and above and statistically signiﬁcant reductions on day 8 for the 500 mg of 3 qd treatment group and on day 28 for the 500-mg of 3 qd, 150-mg of 3 b.i.d., and 400 mg of 3 b.i.d. treatment groups (demonstrated by the 90% CI for the diﬀerence from the model being less than zero). There were no episodes of symptomatic hypoglycemia in the study. Overall, the clinical data from the single and multiple ascending dose studies with 3 demonstrated proof of concept of glucose lowering in patients with T2DM.

GPR40 has attracted signiﬁcant interest over the past decade and was shown to be an excellent target for developing eﬃcacious therapies for use in the treatment of T2DM. We have described herein our eﬀort in this area that culminated in the identiﬁcation of 1, 2, and 3 with the adequate pharmacology and pharmacokinetic properties to explore the potential of this target. Hypothesis driven structural modiﬁcations to the endogenous ligand FFAs, coupled with in depth investigation of the pharmacology and mechanism of action, led to the identiﬁcation of a novel class of GPR40 agonists with unique pharmacology, selectivity, and pharmacokinetic properties. During the course of these programs, a number of key challenges were addressed to enable the design of better agonists. We investigated the SAR of these GPR40 ligands to better understand the role of G protein- and β-arrestin-mediated

■

CONCLUSION Because of its expression and function in both pancreatic β-cells and the enteroendocrine cells of the gastrointestinal tract, R

DOI: 10.1021/acs.jmedchem.6b00892 J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry

Drug Annotation

this novel class of GPR40 agonists might oﬀer promising strategies to explore the full potential opportunities of this new mechanism for the treatment of TD2M.

Table 9. Statistical Analysis of the Change from Baseline Blood Glucose AUC(0−24) a Multiple Ascending Dose (MAD) in Patients with T2DMa

■

variable = CFB (day-1 AUC) glucose AUC(0−24) mg·h/dL

treatment

least squares means [N]

diﬀerence of least squares means (3− Placebo)

90% CI for the diﬀerence (lower, upper)

Chemistry. Unless otherwise noted, all materials were obtained from commercial suppliers and used as obtained. Organic solvents were also purchased from commercial sources on scale, without additional puriﬁcation. Reactions were monitored using Agilent 1100 or 1200 series LCMS with ultraviolet light (UV) detection at 215 and 254 nm and a low resolution electrospray mode (ESI). HRMS data were recorded on an Agilent LC-MS TOF (time-of-ﬂight), model G1969A instrument. Purity was measured using Agilent 1100 or 1200 series high performance liquid chromatography (HPLC) with UV detection at 215, 254, and 280 nm (15 min; 1.5 mL/min ﬂow rate), eluting with a binary solvent system A and B using a gradient elution (A, water with 0.1% TFA; B, MeCN with 0.1% TFA). Unless otherwise noted, the purity of all compounds was ≥98%. Enantiomeric excess for compounds bearing a stereogenic center were determined using analytical HPLC. 1HNMR spectra were recorded on a Bruker AV-400 (400 MHz)spectrometer at ambient temperature. Chemical shifts are reported in parts per million (δ) and are calibrated using residual signal from undeuterated solvent as an internal reference. Data for 1H NMR spectra are reported as follows: chemical shift (δ ppm) (multiplicity, coupling constant (Hz), integration). Multiplicities are reported as follows: s = singlet, d = doublet, t = triplet, q = quartet, m = multiplet, br = broad, or combinations thereof. Diﬀerential scanning calorimetry (DSC) was performed on a TA Instruments Q100 calorimeter in an aluminum Tzero pan under dry N2 ﬂowing at 50 mL/min. Preparation of Compound 1. Methyl 4-(spiro[indene-1,4′-piperidin]-1′-ylmethyl)benzoate (35). To a 2000 L glass-lined reaction vessel at ambient temperature with stirring was added MeOH (363 L), spiro[indene-1,4′-piperidine] hydrochloride (33) (36.2 kg, 163.4 mol), and saturated NaHCO3 (aq) (186 L). The resulting suspension was cooled to 0 °C, and a solution of methyl 4-(bromomethyl)benzoate (31) (36.7 kg, 160.3 mol) in THF (36 L) was added. The mixture was stirred for 4 h at 0 °C, then warmed to rt and stirred for an additional 8 h. The reaction mixture was warmed to 45 °C, treated with 5 N HCl (114 L), and the resulting suspension was stirred for 1 h. The mixture was cooled to rt, stirred for an additional 12 h, and then ﬁltered to provide 35 (52.7 kg, 142.5 mol, 89% yield) as a white solid. 1H NMR (400 MHz, chloroform-d): δ 8.13 (d, J = 8.1 Hz, 2H), 7.90 (d, J = 8.3 Hz, 2H), 7.54 (d, J = 7.1 Hz, 1H), 7.38−7.21 (m, 3H), 6.86 (d, J = 5.6 Hz, 1H), 6.75 (d, J = 5.8 Hz, 1H), 4.38 (d, J = 5.1 Hz, 2H), 3.94 (s, 3H), 3.62 (d, J = 10.6 Hz, 2H), 3.17−3.01 (m, 4H), 1.47 (d, J = 12.1 Hz, 2H). LRMS (ESI): m/z (M + H) calcd, 334.1; found, 334.2. (4-(Spiro[indene-1,4′-piperidin]-1′-ylmethyl)phenyl)methanol (36). To 640 L glass-lined reaction vessel at ambient temperature with stirring was charged 2-methyl-THF (284 L), followed by 35 (28.0 kg, 75.7 mol). The resulting suspension was cooled to 10 °C, and LiAlH4 (3.0 kg, 79.1 mol) in THF (30 L) was added slowly over 2 h. The reaction mixture was allowed to reach ambient temperature and stirred for 2 h, and MeOH (5 L) was added. An aqueous solution of Na-Ktartrate (156 L) was added, the phases were separated, and the aqueous phase was re-extracted with 2-Me-THF (81 L). The organic layers were combined and sequentially washed with aqueous Na-K-tartrate (80 L), water (80 L), and saturated aqueous NaCl (80 L). The organic phase was concentrated to an approximate volume of 50 L, and toluene (135 L) was added. The solution was concentrated again to furnish a solution of 36 (21.5 kg, 70.4 mol, 93% yield) in toluene (50 L), which was used in the subsequent step without further puriﬁcation. A concentrated sample is included for characterization purposes. 1H NMR (400 MHz, DMSOd6): δ 7.43 (d, J = 6.8 Hz, 1H), 7.32−7.27 (m, 5H), 7.22−7.14 (m, 2H), 6.95 (d, J = 5.6 Hz, 1H), 6.78 (d, J = 5.6 Hz, 1H), 5.15 (dd, J = 6.0, 6.0 Hz, 1H), 4.48 (d, J = 5.6 Hz, 2H), 3.57 (s, 2H), 2.87 (d, J = 11.6 Hz, 2H), 2.37−2.29 (m, 2H), 2.11−2.04 (m, 2H), 1.18 (d, J = 12.8 Hz, 2H). LRMS (ESI): m/z (M + H) calcd, 306.2; found, 306.1. Ethyl (S)-3-(4-((4-(Spiro[indene-1,4′-piperidin]-1′-ylmethyl)benzyl)oxy)phenyl)hex-4-ynoate (38). To a 600 L glass-lined reaction

P-value

Day 8 placebo 3 60 mg qd

−177.15 [12] 56.98 [5]

3 200 mg qd

−288.79 [8]

−111.64

3 500 mg qd

−632.15 [7]

−455.00

3 1200 mg qd

−374.24 [9]

−197.09

3 150 mg b.i.d.

−523.41 [8]

−346.26

3 400 mg b.i.d.

−454.13 [9]

−276.98

234.13

(−258.00, 726.27) (−528.86, 305.58) (−884.11, −25.89) (−601.20, 207.01) (−762.41, 69.89) (−680.15, 126.18)

0.431 0.657 0.081 0.419 0.170 0.256

Day 28 placebo 3 60 mg qd

23.25 [10] 105.81 [5]

3 200 mg qd

−18.71 [8]

−41.95

3 500 mg qd

−505.65 [8]

−528.89

3 1200 mg qd

−355.87 [8]

−379.11

3 150 mg b.i.d.

−525.43 [8]

−548.67

3 400 mg b.i.d.

−538.35 [8]

−561.59

82.57

(−419.63, 584.76) (−471.27, 387.37) (−957.05, −100.74) (−805.86, 47.64) (−976.86, −120.48) (−987.54, −135.64)

EXPERIMENTAL SECTION

0.785 0.871 0.043 0.143 0.036 0.031

a

AUC = area under the concentration versus time curve; b.i.d. = twice daily administration; CFB = change from baseline; CI = conﬁdence interval; N = number of patients; qd = once daily administration. Model: parameter = baseline + treatment + day + treatment × day + subject + error. Baseline is shown in parentheses in the variable name.

signaling and drug−receptor residence time on the in vivo eﬃcacy. We examined the correlation between in vivo results and the in vitro assays that we had used, and the best in vitro to in vivo correlation was seen when β-arrestin and glucose lowering were compared, most likely driven by binding kinetics on the receptor and the equilibrium properties of the β-arrestin assay. Calcium ﬂux measurements, due to their transient nature, lacked correlation with in vivo activities while IP-1 assays, with equilibrium kinetic characteristics, had better connectivity. Compounds 1, 2, and 3 were identiﬁed as potent GPR40 agonists that demonstrate selectivity and GDIS in MIN6 cells and in primary islets from mice, rats, and humans. Potent, eﬃcacious, and durable dose-dependent reductions in glucose levels along with signiﬁcant increases in insulin secretion were seen during GTT studies in normal mice and insulin resistant rats. Immediate and prolonged increases in GLP-1 secretion were seen when compounds were administered orally to normal mice, plus a dose-dependent increase in GLP-1 secretion was seen in the mouse gut tumor cell line, STC-1. The combination of GDIS plus GLP-1 secretion supported further development of these compounds as potential glucose-lowering drug candidates. A clinical study of multiple doses over 28 days provided proof of concept for 3 in patients with T2DM. Further development of S

DOI: 10.1021/acs.jmedchem.6b00892 J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry

Drug Annotation

was cooled to 5 °C, and the resulting suspension was ﬁltered to provide 42 (41.3 kg, 131.9 mol, 89% yield) as a white, crystalline solid. 1H NMR (400 MHz, chloroform-d): δ 7.25−7.13 (m, 4H), 6.86 (d, J = 3.6 Hz, 1H), 6.80 (d, J = 3.6 Hz, 1H), 4.79 (s, 2H), 3.74 (s, 2H), 2.93−2.86 (m, 4H), 2.27−2.21 (m, 2H), 2.01−1.92 (m, 5H), 1.53 (d, J = 11.6 Hz, 2H). LRMS (ESI): m/z (M + H) calcd, 314.2; found, 314.1. 1′-((5-(Chloromethyl)thiophen-2-yl)methyl)-2,3-dihydrospiro[indene-1,4′-piperidine] Hydrochloride (43). To a 2000 L reaction vessel was added (5-((2,3-dihydrospiro[indene-1,4′-piperidin]-1′-yl)methyl)thiophen-2-yl)methanol (42) (40.1 kg, 127.9 mol), EtOAc (800 L), and water (1.5 L) with stirring at ambient temperature. The solution was cooled to 0 °C, and SOCl2 (27.9 L, 383.7 mol) was added dropwise and then stirred for 4 h. The solution was concentrated, and EtOAc was added (110 L), followed by stirring for 2 h. The resulting solids were collected by ﬁltration to provide 43 (44.3 kg, 120.3 mol, 94% yield) as an oﬀ-white solid. 1H NMR (400 MHz, DMSO-d6): δ 7.64−7.10 (m, 6H), 5.06 (s, 2H), 4.58 (d, J = 4.4 Hz, 2H), 3.40 (m, 1H), 3.20−3.08 (m, 2H), 2.87 (dd, J = 14.8, 14.8 Hz, 2H), 2.10−2.03 (m, 2H), 1.75−1.65 (m, 2H). LRMS (ESI): m/z (M + H) calcd, 332.1; found, 332.1. Ethyl (S)-3-(4-((5-((2,3-Dihydrospiro[indene-1,4′-piperidin]-1′-yl)methyl)thiophen-2-yl)methoxy)phenyl)hex-4-ynoate Oxalate (44). To a 2000 L reaction vessel was added ethyl (S)-3-(4-hydroxyphenyl)hex-4-ynoate (37) (26.8 kg, 115.3 mol), DMSO (780 L), and Cs2CO3 (194.0 kg, 595.4 mol) with stirring at ambient temperature. 1′-((5(Chloromethyl)thiophen-2-yl)methyl)-2,3-dihydrospiro[indene-1,4′piperidine] hydrochloride (43) (44.9 kg, 121.9 mol) was added, and the reaction mixture was stirred for 16 h. MTBE (900 L) was added, followed by the addition of water (900 L) over 1 h. The phases were separated, and the aqueous layer was extracted again MTBE (300 L). The organic layers were combined, washed with water (450 L), and then passed over silica gel (90.0 kg) using MTBE (300 L) as eluent. The solution was concentrated under reduced pressure at 45 °C, activated carbon (4.5 kg) was added, and the solution ﬁltered over Celite at 45 °C, using an additional MTBE (120 L) rinse of the ﬁlter cake. The ﬁltrate was cooled to ambient temperature, and oxalic acid (22.6 kg, 251.0 mol) in water (250 L) was added dropwise. The solution was stirred for 12 h, and the resulting solids were collected by ﬁltration and recrystallized from EtOH to give 44 (52.0 kg, 84.2 mol, 73% yield) as a white, crystalline solid. 1H NMR (400 MHz, DMSO-d6): δ 7.64−7.10 (m, 6H), 5.06 (s, 2H), 4.58 (d, J = 4.4 Hz, 2H), 3.40 (m, 1H), 3.20−3.08 (m, 2H), 2.87 (dd, J = 14.8, 14.8 Hz, 2H), 2.10−2.03 (m, 2H), 1.75−1.65 (m, 2H). LRMS (ESI): m/z (M + H) calcd, 332.1; found, 332.1. (S)-3-(4-((5-((2,3-Dihydrospiro[indene-1,4′-piperidin]-1′-yl)methyl)thiophen-2-yl)methoxy)phenyl)hex-4-ynoic Acid (2). To a 3000 L glass-lined reaction vessel at ambient temperature with stirring was added ethyl (S)-3-(4-((5-((2,3-dihydrospiro[indene-1,4′-piperidin]-1′-yl)methyl)thiophen-2-yl)methoxy)phenyl)hex-4-ynoate oxalate (44) (60.1 kg, 97.3 mol), EtOH (620 L), and THF (200 L). The solution was cooled to 10 °C, and NaOH (18.0 kg, 450.0 mol) in water (152 L) was added dropwise. The reaction mixture was warmed to ambient temperature and stirred for 5 h, and CH2Cl2 (600 L) was added. Additional water (580 L) was charged, and the solution cooled to 5 °C. Then 1 N HCl (604 L) was added slowly, and the layers were separated. The aqueous phase was extracted again with CH2Cl2 (300 L), and the organic layers were combined, washed with water (300 L), and then concentrated. Fresh CH2Cl2 (300 mL) was added, followed by seed crystals 2 (190.0 g, 380 mmol). The solution was concentrated to a ﬁnal volume of approximately 150 L before cooling to 10 °C. The suspension was stirred for 14 h, ﬁltered, and recrystallized from EtOH to provide 2 (34.3 kg, 68.6 mol, 71% yield) as a white, crystalline solid. 1H NMR (400 MHz, DMSO-d6): δ 7.64−7.10 (m, 6H), 5.06 (s, 2H), 4.58 (d, J = 4.4 Hz, 2H), 3.40 (m, 1H), 3.20−3.08 (m, 2H), 2.87 (dd, J = 14.8, 14.8 Hz, 2H), 2.10−2.03 (m, 2H), 1.75−1.65 (m, 2H). LRMS (ESI): m/z (M + H) calcd, 332.1; found, 332.1. Preparation of Compound 3. Methyl 5-((8-methoxy-3,4-dihydroquinolin-1(2H)-yl)methyl)thiophene-2-carboxylate (46). To a 2000 L glass-lined reaction vessel was added CH3CN (240 L), 8-methoxy1,2,3,4-tetrahydroquinoline (45) (46.9 kg, 235.1 mol), and NaHCO3 (33.7 kg, 401.2 mol) at ambient temperature with stirring. Neat i-Pr2NEt (84.5 L, 485.1 mol) was added dropwise, and the reaction vessel was

vessel at ambient temperature with stirring was added 36 (20.5 kg, 67.1 mol) as a solution in toluene (40 L) from the previous step. Ethyl (S)-3(4-hydroxyphenyl)hex-4-ynoate (37) (18.9 kg, 81.4 mol, 98.5%ee) was added to the reactor, followed by additional toluene (150 L). To the resulting solution was added PPh3 (17.7 kg, 67.5 mol), followed by DIAD (diisopropyl azodicarboxylate) (14.3 kg, 70.7 mol), and the mixture was stirred for 1 h at ambient temperature. Heptanes (299 L) was added, the solution cooled to 0 °C, and the resulting suspension was stirred for 14 h and ﬁltered. The ﬁltrate was concentrated at 70 °C under reduced pressure to arrive at a ﬁnal volume of approximately 50 L, then puriﬁed by passing the solution through silica gel (63 kg) and eluting with toluene (620 L) to arrive at 38 (28.3 kg, 54.5 mol, 81% yield) as a solution in toluene (63 L) that was used in the subsequent step without additional puriﬁcation. A concentrated sample is included for characterization purposes. 1H NMR (400 MHz, DMSO-d6): δ 7.44−7.38 (m, 5H), 7.37−7.33 (m, 1H), 7.29−7.20 (m, 2H), 7.19−7.15 (m, 2H), 6.96−6.94 (m, 3H), 6.93−6.91 (m, 1H), 5.06 (s, 2H), 4.06−3.97 (m, 3H), 3.59 (s, 2H), 2.88−2.85 (m, 2H), 2.67−2.65 (m, 2H), 2.38−2.33 (m, 1H), 2.12−2.05 (m, 1H), 1.77 (d, J = 2.4 Hz, 3H), 1.20−1.11 (m, 7H). LRMS (ESI): m/z (M + H) calcd, 520.3; found, 520.4. (S)-3-(4-((4-(Spiro[indene-1,4′-piperidin]-1′-ylmethyl)benzyl)oxy)phenyl)hex-4-ynoic Acid (1). To 400 L stainless steel reaction vessel at ambient temperature was added 38 (18.1 kg, 34.8 mol) in toluene (50 L) from the previous step, and the solution was concentrated at 70 °C under reduced pressure. EtOH (105 L) was added, the solution was concentrated, and this was repeated two additional times to remove residual toluene. Water (170 L) was added, followed by 5 M NaOH (13 L, 65.0 mol), and the resulting cloudy solution was heated to 80 °C. After 19 h, a premixed solution of water (20 L), AcOH (0.9 L) and ethanol (6 L) was added to the reaction mixture at 60 °C, resulting in a pH of 8−9. Seed crystals (60 g, 122 mmol) were added and stirred for 1 h at 60 °C. A premixed solution of water (40 L), AcOH (1.8 L), and EtOH (12 L) was added slowly over 2 h at 60 °C to arrive at a pH of 6−7. The suspension was cooled to ambient temperature, ﬁltered, and recrystallized from n-propanol to provide 1 (14.0 kg, 28.5 mol, 82% yield) as a white, crystalline solid. 1H NMR (500 MHz, DMSO-d6): δ 12.40−12.10 (bs, 1H), 7.44−7.39 (m, 5H), 7.28 (d, J = 6.8 Hz, 1H), 7.27 (d, J = 8.8 Hz, 2H), 7.23−7.14 (m, 2H), 6.96−6.94 (m, 3H), 6.78 (d, J = 6.0 Hz, 1H), 5.06 (s, 2H), 3.94 (ddd, J = 9.6, 7.2, 2.4 Hz, 1H), 3.60 (s, 2H), 2.88 (d, J = 11.6 Hz, 2H), 2.59 (d, J = 7.6 Hz, 2H), 2.40−2.34 (m, 2H), 2.12 (m, 2H), 1.77 (s, 3H), 1.20 (d, J = 13.2 Hz, 2H). LRMS (ESI): m/z (M + H) calcd, 492.3; found, 492.7. Preparation of Compound 2. Methyl 5-((2,3-dihydrospiro[indene1,4′-piperidin]-1′-yl)methyl)thiophene-2-carboxylate (41). To a 1000 L glass-lined reaction vessel was added 2,3-dihydrospiro[indene-1,4′piperidine] hydrochloride (40) (40.0 kg, 178.8 mol), DIPEA (109 L, 625.8 mol), and acetonitrile (200 L) with stirring at ambient temperature. The mixture was heated to 50 °C, and a solution containing methyl 5-(bromomethyl)thiophene-2-carboxylate (39) (43.6 kg, 185.3 mol) in acetonitrile (140 L) was added slowly over 1 h. After 5 h, the reaction was deemed complete, and water (320 L) was added over 2−3 h, then the reaction mixture was cooled to 10 °C. The solids were collected by ﬁltration, stirred again in water (400 L), ﬁltered, and recrystallized from MeOH (140 L) and CH2Cl2 (32 L) to give 41 (50.1 kg, 146.7 mol, 82% yield) as a white crystalline solid. 1H NMR (400 MHz, chloroform-d): δ 7.68 (d, J = 3.6 MHz, 1H), 7.23−7.14 (m, 4H), 6.93 (d, J = 3.6 Hz, 1H), 3.88 (s, 3H), 3.79 (s, 2H), 2.93−2.87 (m, 4H), 2.28 (ddd, J = 12.0, 12.0, 2.0 Hz, 2H), 2.01−1.94 (m, 4H), 1.54 (d, J = 13.2 MHz, 2H). LRMS (ESI): m/z (M + H) calcd, 342.1; found, 342.3. (5-((2,3-Dihydrospiro[indene-1,4′-piperidin]-1′-yl)methyl)thiophen-2-yl)methanol (42). To a 2000 L glass-lined reaction vessel containing LiAlH4 (5.9 kg, 158.1 mol) in THF (330 L) with stirring was added a solution of 41 (50.6 kg, 148.2 mol) in THF (330 L) at 0 °C. After 3 h, the reaction was deemed complete and MeOH (50 L) was added slowly, followed by stirring for 3 h. Na2SO4·10H2O (45.0 kg, 139.7 mol) was added in portions, the mixture was ﬁltered over Celite (37 kg), and the ﬁltrate was concentrated. To the concentrated reaction mixture was added toluene (300 L), and the solution was heated to 55 °C with stirring for 2 h. Heptane (420 L) was added slowly, the solution T

DOI: 10.1021/acs.jmedchem.6b00892 J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry

Drug Annotation

warmed to 45 °C. Methyl 5-(bromomethyl)thiophene-2-carboxylate (39) (48.1 kg, 204.6 mol) in i-PrOAc (150 L) was added slowly, and the reaction was stirred at 45 °C for 16 h. The reaction was cooled to ambient temperature, and i-PrOAc (400 L) was added, followed by water (500 L). The layers were separated, the aqueous phase was again extracted with i-PrAc (230 L), and the organic layers were combined. The combined organic phase was washed with 5% NaCl (aq) (500 L) and then concentrated to approximately 600 L. Activated carbon was added, and the solution was heated to 55 °C, ﬁltered over Celite, and concentrated to approximately 150 L. Additional i-PrOAc (200 L) was added, the solution was concentrated to approximately 150 L and heated to 65 °C, and n-heptane (440 L) was added dropwise. The solution was cooled to 45 °C, and 46 crystal seed (500.0 g, 1.6 mol) was added. The solution was cooled to 0 °C over 3 h and held at this temperature with stirring for 12 h. The solids were collected by ﬁltration and recrystallized in EtOH to provide 46 (44.8 kg, 141.1 mol, 69% yield) as a white, crystalline solid. 1H NMR (400 MHz, chloroform-d): δ 7.67 (d, J = 3.6 Hz, 1H), 6.94 (d, J = 3.6 Hz, 1H), 6.87 (dd, J = 8.0, 8.0 Hz, 1H), 6.72 (d, J = 8.0 Hz, 1H), 6.69 (d, J = 7.6 Hz, 1H), 4.34 (s, 2H), 3.88 (s, 3H), 3.83 (s, 3H), 3.09−3.06 (m, 2H), 2.76 (t, J = 6.4 Hz, 2H), 1.82−1.76 (m, 2H). LRMS (ESI): m/z (M + H) calcd, 318.1; found, 318.1. (5-((8-Methoxy-3,4-dihydroquinolin-1(2H)-yl)methyl)thiophen-2yl)methanol (47). To a 3000 L glass-lined reaction vessel containing methyl 5-((8-methoxy-3,4-dihydroquinolin-1(2H)-yl)methyl)thiophene-2-carboxylate (46) (60.1 kg, 189.4 mol) in THF (380 L) was added LiAlH4 (8.1 kg, 213.4 mol) in THF (460 L) dropwise with stirring at 0 °C. After 3 h, MeOH (65 L) was added slowly and the mixture was warmed to ambient temperature. After stirring for 3 h, Na2SO4·10H2O (55.0 kg, 170.3 mol) was added in portions and stirring was continued for 2 h. The reaction mixture was ﬁltered over Celite (13 kg) and concentrated to approximately 150 L, and toluene (320 L) was added. The reaction mixture was heated to 60 °C, and n-heptane (500 L) was added dropwise. The reaction mixture was cooled to 5 °C and the resulting solids collected by ﬁltration to furnish 47 (51.8 kg, 179.0 mol, 95% yield) as a white solid. 1H NMR (400 MHz, chloroform-d): δ 6.89− 6.81 (m, 3H), 6.73 (d, J = 8.0 Hz, 1H), 6.69 (d, J = 7.6 Hz, 1H), 4.77 (d, J = 5.6 Hz, 2H), 4.31 (s, 2H), 3.91 (s, 3H), 3.11−3.08 (m, 2H), 2.76 (t, J = 6.4 Hz, 2H), 1.87 (dd, J = 5.6, 5.6 Hz, 1H), 1.82−1.76 (m, 2H). LRMS (ESI): m/z (M + H) calcd, 290.1; found, 290.3. Ethyl (S)-3-(4-((5-((8-Methoxy-3,4-dihydroquinolin-1(2H)-yl)methyl)thiophen-2-yl)methoxy)phenyl)hex-4-ynoate (49). To a 3000 L glass-lined reaction vessel containing 47 (49.7 kg, 171.7 mol) in CH2Cl2 (380 L) at 0 °C was added SOCl2 (18.3 L, 252.2 mol) dropwise. The reaction mixture was stirred for 1 h, EtOH (100 L) was added slowly, and the solution was concentrated to approximately 200 L. DMSO (250 L) was added, the solution was concentrated, and a previously prepared solution containing ethyl (S)-3-(4-hydroxyphenyl)hex-4-ynoate (37) (37.4 kg, 161.0 mol) and Cs2CO3 (450.4 kg, 1382 mol) in DMSO (500 L) was charged. The reaction was stirred at ambient temperature for 16 h, and water (1250 L) was added, followed by MTBE (1000 L), and the layers were separated. The organic phase was washed sequentially with water (250 L), 0.5 N HCl (500 L × 2), and additional water (250 L). To the organic phase was added activated carbon (10.0 kg) and the mixture ﬁltered over Celite (10.0 kg). The ﬁltrate was concentrated to approximately 150 L, and EtOH (500 L) followed by THF (340 L) were added to arrive at 49 (62.6 kg, 124.3 mol, 79% yield) as a solution that was used in the subsequent step without further puriﬁcation. A concentrated sample was included for characterization purposes. 1H NMR (400 MHz, DMSO-d6): δ 7.26 (d, J = 8.8 Hz, 2H), 6.98 (d, J = 3.2 Hz, 1H), 6.94 (d, J = 8.8 Hz, 2H), 6.86 (d, J = 3.6 Hz, 1H), 6.79−6.72 (m, 2H), 6.60 (dd, J = 6.8, 2.5 Hz, 1H), 5.18 (s, 2H), 4.21 (s, 2H), 4.04 (dd, J = 6.4, 3.6 Hz, 1H), 4.01 (dd, J = 6.8, 2.4 Hz, 1H), 3.99−3.93 (m, 1H), 3.76 (s, 3H), 2.94−2.88 (m, 2H), 2.67−2.62 (m, 4H), 1.76 (d, J = 2.0 Hz, 3H), 1.71−1.63 (m, 2H), 1.11 (t, J = 6.8 Hz, 3H). LRMS (ESI): m/z (M + H) calcd, 504.2; found, 504.3. (S)-3-(4-((5-((8-Methoxy-3,4-dihydroquinolin-1(2H)-yl)methyl)thiophen-2-yl)methoxy)phenyl)hex-4-ynoic Acid (3). To a 3000 L glass-lined reactor was added (49) (64.4 kg, 127.9 mol) in THF (575 L), followed by the slow addition of 1.5 N NaOH (265 L, 397.5 mol) at ambient temperature with stirring. After 5 h, EtOAc (650 L) was added,

followed by 2 N HCl (aq) (230 L), and the layers were separated. The organic layer was washed with a solution containing saturated 1.5% NaHCO3 (530 L) and 25% NaCl (395 L) in water (300 L) and again with 1.5% NaCl (530 L). The organic phase was treated with activated carbon (7 kg), ﬁltered over Celite (7 kg), and concentrated under reduced pressure. To the reaction vessel was added i-PrOH (330 L), and the solution was concentrated to a ﬁnal volume of approximately 130 L. The addition of i-PrOH (330 L) followed by concentration of the reaction mixture was repeated two additional times to remove residual EtOAc, and a ﬁnal charge of i-PrOH (330 L) was added. The resulting solution was heated to 60 °C, water (103 L) was added, and the solution was stirred for 1 h. The solution was cooled to ambient temperature, seed crystals (3) (170 g, 357 mmol) were added, and the solution stirred for 3 h. The solution was further cooled to 15 °C and stirred for 12 h. The resulting solids were ﬁltered and recrystallized from acetone and nheptane to give 3 (49.8 kg, 104.7 mol, 82% yield) as a white, crystalline solid. 1H NMR (400 MHz, DMSO-d6): δ 12.50−12.20 bs, 1H), 7.28 (d, J = 8.4 Hz, 2H), 7.00 (d, J = 3.2 Hz, 1H), 6.96 (d, J = 8.4 Hz, 2H), 6.88 (d, J = 3.2 Hz, 1H), 6.81−6.74 (m, 2H), 6.62 (d, J = 6.4 Hz, 1H), 5.19 (s, 2H), 4.22 (s, 2H), 3.95−3.92 (m, 1H), 3.78 (s, 3H), 2.93−2.91 (m, 2H), 2.66 (dd, J = 6.0 Hz, 2H), 2.59 (d, J = 7.6 Hz, 2H), 1.78 (d, J = 1.6 Hz, 3H), 1.75−1.63 (m, 2H). LRMS (ESI): m/z (M + H) calcd, 476.2; found, 476.2. Competitive Radioligand Binding. Crude cell surface membranes were prepared from human embryonic kidney (HEK) 293 cells stably transfected with full length recombinant human GPR40 cDNA using diﬀerential centrifugation methods. First, 10 μL of compound diluted in 100% DMSO, and 90 μL of assay buﬀer (50 mM Tris-HCl, pH 7.5, 5 mM CaCl2, 5 mM MgCl2, 0.1% w/v fatty acid-free BSA) were added to a deep 96-well polypropylene assay plate (Beckman Coulter). Then 200 μL of [3H]-4 (52 Ci/mmol, Quotient Bioresearch Radiochemicals, Ltd.; 5 nM ﬁnal concentration) and 200 μL of GPR40 membranes (5 μg/ well), both diluted in assay buﬀer, were added to the assay plate, followed by a 1 min shake and a 2 h incubation at rt (22 °C). Assays were terminated by ﬁltration through GF/C glass ﬁber ﬁltermats (PerkinElmer) presoaked in 50 mM Tris-HCl, pH 7.5, using a Mach III cell harvester (TomTec). Filtermats were washed 2 times with 5 mL of ice-cold 50 mM Tris-HCl, pH 7.5 buﬀer, dried 1 h in a convection oven at 60 °C, and embedded with Meltilex A solid scintillant (PerkinElmer). Radioactivity was determined as counts per minute (CPM) using a Trilux Microbeta plate scintillation counter (PerkinElmer). The equilibrium dissociation constant (Ki) was calculated from the relative IC50 value based upon the equation Ki = IC50/(1 + L/Kd), where L equals the concentration of radioligand used in the experiment and Kd equals the equilibrium binding aﬃnity constant of the radioligand, determined from saturation analysis (6.2 nM). Competitive radioligand binding assays using [3H]-11 (83 Ci/ mmol, Quotient Bioresearch Radiochemicals, Ltd., 5 nM ﬁnal concentration) were performed using the same methods described above. Binding Kinetics and Receptor Residence Time. The association (at rt) of [3H]-11 to GPR40 was initiated by adding 5 μg of human GPR40 membranes diluted in 100 μL of assay buﬀer (50 mM Tris-HCl, pH 7.5, 5 mM CaCl2, 5 mM MgCl2, 0.1% w/v fatty acid-free BSA) at various time points into a deep 96-well plate. Each well contained 200 μL [3H]-11 in assay buﬀer and either 100 μL of assay buﬀer to deﬁne total binding or 100 μL of unlabeled 11 in assay buﬀer (10 μM ﬁnal concentration) to deﬁne nonspeciﬁc binding. After all time points at 0−3 h were completed on the plate, separation of bound and free radioligand followed by quantiﬁcation of radioactivity (CPM) was performed using the rapid-wash ﬁltration/counting methods described above. Data were ﬁt to a one-phase association curve using GraphPad Prism software.56 Dissociation experiments (at rt) were performed by adding 200 μL of [3H]-11 or [3H]-4 (5 nM ﬁnal concentration) and 200 μL of human GPR40 membranes (5 μg) to all wells of a deep 96-well plate. The plate was preincubated for 2 h to allow the radioligand and receptor to reach steady state. Radioligand dissociation was initiated by the addition of 100 μL unlabeled 11 or 4 (10 μM ﬁnal concentration) to wells at various times. After all time points were completed on the plate, separation of U

DOI: 10.1021/acs.jmedchem.6b00892 J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry

Drug Annotation

complementation of the β-galactosidase (β-gal) enzyme fragments, forming a β-gal enzyme that generates a chemiluminescent signal using the DiscoveRx PathHunter detection kit. Cells were incubated overnight at 5000 cells/well in 384-well plates in culture media (DiscoveRx) containing 1% heat inactivated FBS. Serial diluted compound in DMSO (2× dilutions to generate 20 concentrations) were step-down diluted in culture media containing 1% FBS and added to cells with a ﬁnal top concentration starting of 100 μM and 1% DMSO. After addition of compound, cells were incubated for 90 min at 37 °C in a 5% CO2 incubator, and DiscoveRx kit detection reagents were added. Measurement of the chemiluminescent signal was ascertained with the Envision reader after a 1 h incubation at rt. Percent activity was measured versus maximum response to 1 μM of 15. EC50 values were calculated by plotting test compound concentration versus percent stimulation using a 4-parameter logistic curve ﬁtting equation. Peroxisome Proliferator-Activated Receptor (PPAR) α, δ, and γ Binding and Functional Assays. Binding aﬃnities of compounds for the PPAR α, δ, and γ receptors were assessed using scintillation proximity assay (SPA) technology. Biotinylated oligonucleotide direct repeat 2 (DR2) was used for binding the receptors to yttrium silicate streptavidin-coated SPA beads (PerkinElmer). PPAR α, δ, γ, and retinoid X receptor (RXR) α were expressed using a baculovirus system and SF9 cell lysates containing the speciﬁc receptors were used in the individual assays. The DR2 was attached to the SPA beads over a 30 min period in a binding buﬀer containing 10 mM HEPES, pH 7.8, 80 mM KCl, 0.5 mM MgCl2, 1 mM DTT, 0.5% 3[(3-cholamidopropyl)dimethylammonio]-propanesulfonic acid (CHAPS) and 4.4% bovine serum. The cell lysates were incubated in each well with one of 11 concentrations of compound in the presence of a radiolabeled PPAR α/ δ dual agonist reference compound for the α and δ receptor assays (∼0.033.8 μCi 3H) or a radiolabeled PPARγ agonist reference compound for the γ receptor assays (∼0.037.3 μCi 3H), 110 μg of yttrium SPA streptavidin coated beads, 0.126 nM HD oligo DR2, and either 0.3 μg PPARα with 0.5 μg RXRα, 0.5 μg PPARδ with 0.5 μg RXRα, or 1.25 μg PPARγ with 3.03 μg RXRα in the binding buﬀer above plus 14% glycerol and 5 μg of sheared salmon sperm DNA. Nonspeciﬁc binding was determined in the presence of 10000 nM of the unlabeled PPAR α/δ dual agonist reference compound for the α and δ receptor assays and the PPARγ agonist reference compound for the γ receptor assay. The binding reaction (100 μL per well in a 96-well [Costar 3632] plate) was incubated for 10 h and CPM determined on a microbeta plate scintillation counter. Receptor binding aﬃnity (IC50) for the compounds was determined by ﬁtting an 11-point concentration− response curve with a 4-paramater logistic equation. Ki values were determined from the IC50 using the equation described above with the Kd determined by saturation binding. CV1 cells, African green monkey kidney cells, were transfected with various receptor and reporter plasmids using Fugene. For the Gal4 PPARα and PPARδ assays, a reporter plasmid containing ﬁve tandem copies of the yeast transcription protein Gal4 response element, cloned upstream of a ﬁreﬂy luciferase gene driven by the major late promoter of adenovirus, was transfected together with a Simian Virus 40 (SV40) driven plasmid constitutively expressing a hybrid protein containing the Gal4 DNA binding domain (DBD) and either the PPARα or PPARδ ligand binding domain. For the PPARγ assay, plasmids encoding PPARγ and RXRα, both driven by a cytomegalovirus (CMV) promoter, were transfected together with a plasmid containing luciferase reporter cDNA driven by the TK promoter and a receptor response element (2× PPRE). Cells were transfected in T225 cm2 cell culture ﬂasks in DMEM media with 5% charcoal-stripped FBS. After an overnight incubation, transfected cells were trypsinized, plated in opaque 96-well plates (15000 cells/well) in DMEM media containing 5% charcoal-stripped FBS, incubated for 4 h, and exposed to 0.17 nM to 10 μM of test compounds or reference compound in half log dilutions. After 24 h incubation with compounds, cells were lysed and luciferase activity was determined as a measure of receptor activation by luminescence. The percent stimulation was determined versus maximum stimulation obtained with 10 μM of an appropriate PPAR agonist reference compound. Data were ﬁt to a 4 parameter logistics model to determine EC50 values.

bound and free radioligand followed by quantiﬁcation of radioactivity was performed using the rapid-wash ﬁltration/counting methods described above. Total binding and nonspeciﬁc binding controls were included on the plate to ensure that steady-state conditions were maintained throughout the dissociation time period (0−3 h post preincubation). Data were ﬁt to a one-phase dissociation curve using GraphPad Prism software. The time of duration that an unlabeled compound was complexed to GPR40 (residence time, τR), was quantiﬁed using kinetics of competitive radioligand binding methods.38 Binding to GPR40 was measured at various time points (as described above) using [3H]-11, in the absence or presence of unlabeled test compound at several concentrations. The resulting parameters were ﬁt to a model that uses the association and dissociation rates for [3H]-11 to GPR40, determined with the independent kinetic experiments described above. Calcium Flux Assay. HEK293 cells overexpressing human GPR40 (hGPR40) were plated (25K cells/well) into 384-well microtiter plates using Dulbecco’s Modiﬁed Eagle’s Medium (DMEM) plus F12 medium in 3:1 ratio supplemented with 0.2% certiﬁed fetal bovine serum (FBS), 18.9 mM HEPES, 0.9 mM sodium pyruvate, 94.5 U/mL penicillin:94.5 μg/mL streptomycin, and 800 μg/mL geneticin. The cells were incubated overnight at 37 °C and 5% CO2. Calcium 4 dye (Molecular Devices) diluted in assay buﬀer (HBSS and 19.6 mM HEPES) was added (20 uL/well) to the cell plates, followed by a 2.5 h incubation in the dark at 25 °C and 5% CO2. Test compounds were serially diluted 2fold in 100% DMSO and immediately diluted in assay buﬀer. Diluted compounds were immediately added to the cell plates using the liquid handling capabilities of a FLIPR to achieve ﬁnal test compound concentrations of 40 μM to 0.1 nM (20-point concentration response curve) at a ﬁnal DMSO concentration of 1%. Receptor activation was immediately measured as an increase in intracellular calcium using the FLIPR with an excitation ﬁlter of 470−497 nm and emission at 515−575 nm, over 3 min. To determine agonist responses, maximum minus minimum relative ﬂuorescence units (RFUs) over 55 reads were calculated per well and used to calculate percent stimulation relative to 100 μM of the natural ligand, linoleic acid, response. EC50 values were calculated by plotting test compound concentration versus percent stimulation using a 4-parameter logistic curve ﬁtting equation. IP-1 Agonist Assay. HEK293 cells overexpressing human GPR40 (hGPR40) were plated (20K cells/well) into 384-well microtiter plates using Dulbecco’s High Glucose Modiﬁed Eagle’s Medium (DMEM) supplemented with 5% qualiﬁed fetal bovine serum (FBS), 18.9 mM HEPES, 0.9 mM sodium pyruvate, 94.5 U/mL penicillin, 94.5 μg/mL streptomycin, and 800 μg/mL geneticin. The cells were incubated overnight at 37 °C and 5% CO2 and then washed two times, once with assay media without FBS and the other time with IP1 stimulation buﬀer (Cisbio IP-One HTRF assay kit). After washing, IP1 stimulation buﬀer was added back to the cell plates. Test compounds were serially diluted 2-fold in 100% DMSO and then intermediately diluted in IP1 stimulation buﬀer. Diluted compounds were immediately added to the cell plates containing IP1 stimulation buﬀer to achieve ﬁnal test compound concentrations of 25 μM to 0.05 nM (20-point concentration response curve) at a ﬁnal DMSO concentration of 0.6%. The cells and compounds were incubated for 120 min at 37 °C and 5% CO2. To measure compound induced IP-1 generation, a competitive immunoassay was performed by adding terbium cryptate-labeled antiIP1Mab and d2-labeled IP-1 (Cisbio IP-One HTRF assay kit) followed by a 60 min incubation at ambient temperature. The plates were read on a PerkinElmer Envision, and the ﬂuorescence ratio at 665 nM:620 nM was quantiﬁed. A standard curve was used to convert raw data to IP-1 concentration. Percent stimulation was calculated versus the response of 38 nM of GPR40 agonist 5. EC50 values were calculated by plotting test compound concentration versus percent stimulation using a 4parameter logistic curve ﬁtting equation. Human, Mouse, and Rat β-Arrestin Agonist Assays. HEK293− hGPR40 cells were purchased from DiscoveRx. Human osteosarcoma (U2OS) cells expressing mouse GPR40 or rat GPR40 were developed by DiscoveRx. These cells coexpress the Prolink (PK)-tagged GPR40 and the Enzyme Acceptor (EA)-tagged β-arrestin fusion proteins. If activation of the GPR40 stimulates β-arrestin recruitment, it would force V

DOI: 10.1021/acs.jmedchem.6b00892 J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry

Drug Annotation

hERG Selectivity. Compounds were sent to Cerep (France) to generate a selectivity assessment against hERG activity using an astemizole binding assay available through their services. In Vivo Diﬀerential Distribution LC-MS/MS Assay. Trunk blood was collected in EDTA coated 1.5 mL conical polypropylene centrifuge tubes and stored on wet ice until study completion. Total and null binding peripheral tissues were dissected, rinsed with cold saline, placed in 1.5 mL conical polypropylene centrifuge tubes, minced to ﬁne pieces using ﬁne point scissors, weighed, and stored on wet ice until completion of live phase study. Tissues were homogenized ﬁrst by bead lysis using the Bullet blender (Next Advance, Inc.). An equal weight of 0.9−2 mm stainless steel beads to wet tissue weight was added to tubes, and the instrument was run at the maximum speed setting for 5 min. Four volumes of acetonitrile + 0.1% formic acid (FA) (based on the weight of wet tissue collected) were added, and all tubes were vortexed. The partially homogenized samples were then homogenized a second time using a probe-tip ultrasonic dismembrator (Fisher Scientiﬁc model 100) at medium power until an even slurry was obtained (15−30 s). All samples were allowed to sit 5 min at rt and then centrifuged at 20000g for 20 min. A 100 μL aliquot of clear supernatant containing the tracer was diluted with 300 μL of water to bring the acetonitrile content to 25% acetonitrile + 0.1% FA. A 10 μL sample was injected by auto sampler onto a Zorbax SB-C18 column (Agilent Technologies), 3.5 μm, 2.1 mm × 50 mm, that is maintained at 30 °C with Shimadzu Prominence uHPLC. The tracer was detected with an ABSciex 4000 QTRAP LCMS/MS triple quadrupole mass spectrometer using multiple reaction monitoring (MRM) methods to monitor the transition from parent to daughter ion with mass to charge ratios and quantiﬁed by comparison to a standard curve generated by extracting a series of target tissue samples from nontreated animals and processed as described above to which known quantities of analyte had been added. Tracer levels in tissue are represented in units of ng/g of tissue. Trunk blood samples that had been kept on ice in EDTA coated 1.5 mL polypropylene tubes were centrifuged at 14000 rpm for 16 min. After centrifuging, 50 μL of supernatant (plasma) from each sample were added to 200 μL of acetonitrile + 0.1% FA. For standard curve (0.1−100 ng/mL) samples, a calculated volume of standard reduces the volume of acetonitrile. Samples are thoroughly vortexed and then centrifuged for 30 min at 16000 rpm. Then 100 μL of supernatant were removed from each sample vial and placed in a new LC autosampler vial, followed by the addition of 400 μL with sterile water (pH 6.5). Tracers were detected by LC-MS/MS as previously described for tissues. Tracer levels in plasma were represented in units of ng/mL of plasma. All procedures in this protocol are in compliance with the U.S. Department of Agriculture’s (USDA) Animal Welfare Act (9 CFR Parts 1, 2, and 3), the Guide for the Care and Use of Laboratory Animals (Institute of Laboratory Animal Resources, National Academy Press, Washington, DC, 1996), and the National Institutes of Health, Oﬃce of Laboratory Animal Welfare. Whenever possible, procedures in this study are designed to avoid or minimize discomfort, distress, and pain to animals. Glucose Dependent Insulin Secretion (GDIS) Assays. Because activation of GPR40 is known to result in insulin secretion which is dependent on high glucose concentrations, two separate assay systems (insulinoma cell line and primary rodent islets) were developed to further characterize compounds that are known to increase intracellular calcium in the GPR40 primary assay discussed above. GDIS assays were performed using the mouse insulinoma cell line MIN6. MIN6 cells were maintained in DMEM containing nonessential amino acids, 10% FBS, 50 mM 2-mercaptoethanol, and 1% penicillin and streptomycin at 37 °C plus 5% CO2. On the day of the experiment, the cells were washed twice with 200 μL of prewarmed Krebs−Ringer buﬀer without glucose. Addition of 200 μL of prewarmed Krebs−Ringer buﬀer containing 2.5 mM glucose was used to starve the cells followed by the addition of compounds in the presence of a high concentration of glucose (25 mM). The plate was incubated at 37 °C for 2 h. At the end of the 2 h incubation, the supernatant was gently transferred into a Millipore ﬁlter plate and spun at 200g for 3 min. Insulin was assayed using a Mercodia Insulin estimation kit.

GDIS assays were also performed in primary islets. Pancreatic islets of Langerhans were isolated from male SD (Sprague−Dawley) rats or C57Bl6 mice by collagenase digestion and Histopaque density gradient separation. The islets were cultured overnight in RPMI-1640 medium with GlutaMAXn to facilitate recovery from the isolation process. Insulin secretion was determined using 90 min incubation in EBSS (Earle’s Balanced Salt Solution) buﬀer in a 48-well plate. Brieﬂy, islets were ﬁrst preincubated in EBSS with 2.8 mM glucose for 60 min and were then transferred to a 48-well plate (four islets/well) containing 150 μL of 2.8 mM glucose and incubated with 150 μL of EBSS with 2.8 or 11.2 mM glucose in the presence or absence of test compounds for 90 min. The buﬀer was removed from the wells at the end of the incubation period and assayed for insulin levels using a rat or mouse insulin ELISA kit (Mercodia). A similar protocol was used to investigate GDIS in human islets, which were procured from a commercial provider of human donor material (Asterand Inc.). Insulin levels were determined using a human insulin ELISA kit. GLP-1 Secretion In Vivo. C57BL/6 mice (Harlan, Indianapolis, IN) were orally administered either compound of interest at 30 mg/kg or the vehicle control (20% Captisol). GLP-1 (X-36 amide) levels were determined at 0.25, 0.5, 1.5, and 3.0 h after administration of the GPR40 agonist or the vehicle control. A group of six mice was used for each time point. GLP-1 levels were measured using an ELISA kit (MesoScale) with antibodies to an internal sequence and the 36 amide C-terminal portion of the peptide; thus, both GLP-1 fragments 7-36 amide and 9-36 amide were measured. The use of animals was in accordance with international guidelines (NIH 85-23) and was approved by the local animal ethics committee at Lilly Research Laboratories. Intraperitoneal Glucose Tolerance Test (IPGTT). Male Balb/c (albino mice) mice (8−9 weeks of age) were single housed and fed a normal rodent chow diet and allowed water ad libitum. Animals were orally administered compound for 4 days, with an IPGTT performed on the fourth day. On the night before the IPGTT study, animals were fasted overnight in clean cages. On the morning of the IPGTT, animals were dosed orally with compound of interest or vehicle alone 60 min prior to the IPGTT (glucose 2g/kg, ip). Blood glucose levels were determined from tail bleeds taken at 0, 3, 7, 15, 30, and 60 min after glucose challenge. Plasma was isolated and used to estimate respective insulin levels. The blood glucose excursion proﬁle from t = 0 to t = 60 min was used to integrate an area under the curve (AUC) for each treatment group. Percent lowering in glucose is calculated from the AUC data of the compounds with respect to the AUCs of the vehicle group and positive control. The ED90 was determined by ﬁtting a fourparameter logistic model to the dose−response curve in GraphPad Prism. The use of animals was in accordance with international guidelines (NIH 85-23) and was approved by the local animal ethics committee at Lilly Research Laboratories. Oral Glucose Tolerance Test (OGTT) in Zucker fa/fa Rats. OGTTs were performed in Male Zucker fa/fa rats (10 weeks of age), a rodent model of insulin resistance, after 1 and 21 days of orally administered. Compounds were administered orally at various doses to provide a dose response eﬃcacy curve. Rosiglitazone served as the positive control and reference standard for the study. OGTTs were performed 1 h post compound administration with blood samples taken for determination of glucose and insulin levels at 0, 10, 20, 40, and 60 min post glucose administration (2 g/kg). The use of animals was in accordance with international guidelines (NIH 85-23) and was approved by the local animal ethics committee at Lilly Research Laboratories. Clinical Trials. All clinical studies were approved by local institutional review boards and were performed in compliance with the principles of good clinical practice and in accordance with the provision of the Declaration of Helsinki. All subjects provided signed informed consent. W

DOI: 10.1021/acs.jmedchem.6b00892 J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry

■

Drug Annotation

(6) Shapiro, H.; Shachar, S.; Sekler, I.; Hershfinkel, M.; Walker, M. D. Role of GPR40 in fatty acid action on the beta cell line INS-1E. Biochem. Biophys. Res. Commun. 2005, 335, 97−104. (7) Briscoe, C. P.; Tadayyon, M.; Andrews, J. L.; Benson, W. G.; Chambers, J. K.; Eilert, M. M.; Ellis, C.; Elshourbagy, N. A.; Goetz, A. S.; Minnick, D. T.; Murdock, P. R.; Sauls, H. R., Jr.; Shabon, U.; Spinage, L. D.; Strum, J. C.; Szekeres, P. G.; Tan, K. B.; Way, J. M.; Ignar, D. M.; Wilson, S.; Muir, A. I. The orphan G protein-coupled receptor GPR40 is activated by medium and long chain fatty acids. J. Biol. Chem. 2003, 278, 11303−11311. (8) Itoh, Y.; Kawamata, Y.; Harada, M.; Kobayashi, M.; Fujii, R.; Fukusumi, S.; Ogi, K.; Hosoya, M.; Tanaka, Y.; Uejima, H.; Tanaka, H.; Maruyama, M.; Satoh, R.; Okubo, S.; Kizawa, H.; Komatsu, H.; Matsumura, F.; Noguchi, Y.; Shinohara, T.; Hinuma, S.; Fujisawa, Y.; Fujino, M. Free fatty acids regulate insulin secretion from pancreatic beta cells through GPR40. Nature 2003, 422, 173−176. (9) Kotarsky, K.; Nilsson, N. E.; Flodgren, E.; Owman, C.; Olde, B. A human cell surface receptor activated by free fatty acids and thiazolidinedione drugs. Biochem. Biophys. Res. Commun. 2003, 301, 406−410. (10) Prentki, M.; Tornheim, K.; Corkey, B. E. Signal transduction mechanisms in nutrient-induced insulin secretion. Diabetologia 1997, 40 (Suppl 2), S32−S41. (11) Edfalk, S.; Steneberg, P.; Edlund, H. Gpr40 is expressed in enteroendocrine cells and mediates free fatty acid stimulation of incretin secretion. Diabetes 2008, 57, 2280−2287. (12) Hira, T.; Mochida, T.; Miyashita, K.; Hara, H. GLP-1 secretion is enhanced directly in the ileum but indirectly in the duodenum by a newly identified potent stimulator, zein hydrolysate, in rats. Am. J. Physiol. Gastrointest. Liver Physiol. 2009, 297, G663−671. (13) Xiong, Y.; Swaminath, G.; Cao, Q.; Yang, L.; Guo, Q.; Salomonis, H.; Lu, J.; Houze, J. B.; Dransfield, P. J.; Wang, Y.; Liu, J.; Wong, S.; Schwandner, R.; Steger, F.; Baribault, H.; Liu, L.; Coberly, S.; Miao, L.; Zhang, J.; Lin, D. C. H.; Schwarz, M. Activation of FFA1 mediates GLP1 secretion in mice. Evidence for allosterism at FFA1. Mol. Cell. Endocrinol. 2013, 369, 119−129. (14) Araki, T.; Hirayama, M.; Hiroi, S.; Kaku, K. GPR40-induced insulin secretion by the novel agonist TAK-875: first clinical findings in patients with type 2 diabetes. Diabetes, Obes. Metab. 2012, 14, 271−278. (15) Brown, S. P.; Dransfield, P. J.; Vimolratana, M.; Jiao, X.; Zhu, L.; Pattaropong, V.; Sun, Y.; Liu, J.; Luo, J.; Zhang, J.; Wong, S.; Zhuang, R.; Guo, Q.; Li, F.; Medina, J. C.; Swaminath, G.; Lin, D. C.; Houze, J. B. Discovery of AM-1638: A Potent and Orally Bioavailable GPR40/FFA1 Full Agonist. ACS Med. Chem. Lett. 2012, 3, 726−730. (16) Lin, D. C.; Zhang, J.; Zhuang, R.; Li, F.; Nguyen, K.; Chen, M.; Tran, T.; Lopez, E.; Lu, J. Y.; Li, X. N.; Tang, L.; Tonn, G. R.; Swaminath, G.; Reagan, J. D.; Chen, J. L.; Tian, H.; Lin, Y. J.; Houze, J. B.; Luo, J. AMG 837: a novel GPR40/FFA1 agonist that enhances insulin secretion and lowers glucose levels in rodents. PLoS One 2011, 6, e27270. (17) Negoro, N.; Sasaki, S.; Mikami, S.; Ito, M.; Suzuki, M.; Tsujihata, Y.; Ito, R.; Harada, A.; Takeuchi, K.; Suzuki, N.; Miyazaki, J.; Santou, T.; Odani, T.; Kanzaki, N.; Funami, M.; Tanaka, T.; Kogame, A.; Matsunaga, S.; Yasuma, T.; Momose, Y. Discovery of TAK-875: A Potent, Selective, and Orally Bioavailable GPR40 Agonist. ACS Med. Chem. Lett. 2010, 1, 290−294. (18) Poitout, V.; Lin, D. C. Modulating GPR40: therapeutic promise and potential in diabetes. Drug Discovery Today 2013, 18, 1301−1308. (19) Sasaki, S.; Kitamura, S.; Negoro, N.; Suzuki, M.; Tsujihata, Y.; Suzuki, N.; Santou, T.; Kanzaki, N.; Harada, M.; Tanaka, Y.; Kobayashi, M.; Tada, N.; Funami, M.; Tanaka, T.; Yamamoto, Y.; Fukatsu, K.; Yasuma, T.; Momose, Y. Design, synthesis, and biological activity of potent and orally available G protein-coupled receptor 40 agonists. J. Med. Chem. 2011, 54, 1365−1378. (20) Shimada, T.; Ueno, H.; Tsutsumi, K.; Aoyagi, K.; Manabe, T.; Sasaki, K.; Kato, H. Spiro-ring compound and use thereof for medical purposes. PCT Int. Appl. WO 2009054479 A1, April 30, 2009. (21) Tan, C. P.; Feng, Y.; Zhou, Y. P.; Eiermann, G. J.; Petrov, A.; Zhou, C.; Lin, S.; Salituro, G.; Meinke, P.; Mosley, R.; Akiyama, T. E.;

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.jmedchem.6b00892. Molecular formula strings and the associated biochemical and biological data (CSV) Accession Codes

The 3D coordinates of three compounds, 1, 2, and 3 were from molecular modeling and do not have PDB ID codes. Authors will release the atomic coordinates and experimental data upon article publication.

■

AUTHOR INFORMATION

Corresponding Author

*Phone: 317-612-4096. E-mail: hamdouchi_chaﬁ[email protected]. Author Contributions

The manuscript was written through contribution of all authors. All authors have given approval to the ﬁnal version of the manuscript. Notes

The authors declare no competing ﬁnancial interest.

■

ACKNOWLEDGMENTS We thank the GPR40 team at both Lilly and Jubilant for their dedicated eﬀorts. We thank Dr. Ruth Gimeno, Dr. Guoxin Zhu, Dr. Michael Coghlan, Dr. Thomas Hardy, and Dr. Timothy Grese for helpful discussions. We thank the Lilly GPCR Platform for their helpful suggestions. Dr. James Monn and Dr. Alan Warshawsky are gratefully acknowledged for their helpful comments during the preparation of this manuscript.

■

ABBREVIATIONS USED GPR40, G protein-coupled receptor 40; FFAR1, free fatty acid receptor 1; GDIS, glucose dependent insulin secretion; PK, pharmacokinetics; AUC, area under curve; CL, plasma clearance; T2DM, type 2 diabetes mellitus; SAR, structure−activity relationship; PPAR, peroxisome proliferator activator receptor; FLIPR, ﬂuorescence imaging plate reader

■

REFERENCES

(1) Global Report on Diabetes; World Health Organization: Geneva, 2016; http://apps.who.int/iris/bitstream/10665/204871/1/ 9789241565257_eng.pdf (accessed May 26, 2016). (2) Briscoe, C. P.; Peat, A. J.; McKeown, S. C.; Corbett, D. F.; Goetz, A. S.; Littleton, T. R.; McCoy, D. C.; Kenakin, T. P.; Andrews, J. L.; Ammala, C.; Fornwald, J. A.; Ignar, D. M.; Jenkinson, S. Pharmacological regulation of insulin secretion in MIN6 cells through the fatty acid receptor GPR40: identification of agonist and antagonist small molecules. Br. J. Pharmacol. 2006, 148, 619−628. (3) Fujiwara, K.; Maekawa, F.; Yada, T. Oleic acid interacts with GPR40 to induce Ca2+ signaling in rat islet beta-cells: mediation by PLC and L-type Ca2+ channel and link to insulin release. Am. J. Physiol. Endocrinol. Metab. 2005, 289, E670−E677. (4) Nagasumi, K.; Esaki, R.; Iwachidow, K.; Yasuhara, Y.; Ogi, K.; Tanaka, H.; Nakata, M.; Yano, T.; Shimakawa, K.; Taketomi, S.; Takeuchi, K.; Odaka, H.; Kaisho, Y. Overexpression of GPR40 in pancreatic beta-cells augments glucose-stimulated insulin secretion and improves glucose tolerance in normal and diabetic mice. Diabetes 2009, 58, 1067−1076. (5) Schnell, S.; Schaefer, M.; Schofl, C. Free fatty acids increase cytosolic free calcium and stimulate insulin secretion from beta-cells through activation of GPR40. Mol. Cell. Endocrinol. 2007, 263, 173−180. X

DOI: 10.1021/acs.jmedchem.6b00892 J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry

Drug Annotation

Einstein, M.; Kumar, S.; Berger, J. P.; Mills, S. G.; Thornberry, N. A.; Yang, L.; Howard, A. D. Selective small-molecule agonists of G proteincoupled receptor 40 promote glucose-dependent insulin secretion and reduce blood glucose in mice. Diabetes 2008, 57, 2211−2219. (22) Tsujihata, Y.; Ito, R.; Suzuki, M.; Harada, A.; Negoro, N.; Yasuma, T.; Momose, Y.; Takeuchi, K. TAK-875, an orally available G proteincoupled receptor 40/free fatty acid receptor 1 agonist, enhances glucosedependent insulin secretion and improves both postprandial and fasting hyperglycemia in type 2 diabetic rats. J. Pharmacol. Exp. Ther. 2011, 339, 228−237. (23) Burant, C. F.; Viswanathan, P.; Marcinak, J.; Cao, C.; Vakilynejad, M.; Xie, B.; Leifke, E. TAK-875 versus placebo or glimepiride in type 2 diabetes mellitus: a phase 2, randomised, double-blind, placebocontrolled trial. Lancet 2012, 379, 1403−1411. (24) Kaku, K.; Araki, T.; Yoshinaka, R. Randomized, double-blind, dose-ranging study of TAK-875, a novel GPR40 agonist, in Japanese patients with inadequately controlled type 2 diabetes. Diabetes Care 2013, 36, 245−250. (25) Naik, H.; Vakilynejad, M.; Wu, J.; Viswanathan, P.; Dote, N.; Higuchi, T.; Leifke, E. Safety, tolerability, pharmacokinetics, and pharmacodynamic properties of the GPR40 agonist TAK-875: results from a double-blind, placebo-controlled single oral dose rising study in healthy volunteers. J. Clin. Pharmacol. 2012, 52, 1007−1016. (26) Takeda Announces Termination of Fasiglifam (TAK-875) Development; Takeda Pharmaceutical Company Limited: Osaka, December 27, 2013; https://www.takeda.com/news/2013/20131227_6117.html (accessed May 26, 2016). (27) Defossa, E.; Wagner, M. Recent developments in the discovery of FFA1 receptor agonists as novel oral treatment for type 2 diabetes mellitus. Bioorg. Med. Chem. Lett. 2014, 24, 2991−3000. (28) Garrido, D. M.; Corbett, D. F.; Dwornik, K. A.; Goetz, A. S.; Littleton, T. R.; McKeown, S. C.; Mills, W. Y.; Smalley, T. L., Jr.; Briscoe, C. P.; Peat, A. J. Synthesis and activity of small molecule GPR40 agonists. Bioorg. Med. Chem. Lett. 2006, 16, 1840−1845. (29) Sum, C. S.; Tikhonova, I. G.; Neumann, S.; Engel, S.; Raaka, B. M.; Costanzi, S.; Gershengorn, M. C. Identification of residues important for agonist recognition and activation in GPR40. J. Biol. Chem. 2007, 282, 29248−29255. (30) Smith, D. A.; Di, L.; Kerns, E. H. The effect of plasma protein binding on in vivo efficacy: misconceptions in drug discovery. Nat. Rev. Drug Discovery 2010, 9, 929−939. (31) Lin, D. C.; Guo, Q.; Luo, J.; Zhang, J.; Nguyen, K.; Chen, M.; Tran, T.; Dransfield, P. J.; Brown, S. P.; Houze, J.; Vimolratana, M.; Jiao, X. Y.; Wang, Y.; Birdsall, N. J.; Swaminath, G. Identification and pharmacological characterization of multiple allosteric binding sites on the free fatty acid 1 receptor. Mol. Pharmacol. 2012, 82, 843−859. (32) Luo, J.; Swaminath, G.; Brown, S. P.; Zhang, J.; Guo, Q.; Chen, M.; Nguyen, K.; Tran, T.; Miao, L.; Dransfield, P. J.; Vimolratana, M.; Houze, J. B.; Wong, S.; Toteva, M.; Shan, B.; Li, F.; Zhuang, R.; Lin, D. C. A potent class of GPR40 full agonists engages the enteroinsular axis to promote glucose control in rodents. PLoS One 2012, 7, e46300. (33) Wang, Y.; Liu, J. J.; Dransfield, P. J.; Zhu, L.; Wang, Z.; Du, X.; Jiao, X.; Su, Y.; Li, A. R.; Brown, S. P.; Kasparian, A.; Vimolratana, M.; Yu, M.; Pattaropong, V.; Houze, J. B.; Swaminath, G.; Tran, T.; Nguyen, K.; Guo, Q.; Zhang, J.; Zhuang, R.; Li, F.; Miao, L.; Bartberger, M. D.; Correll, T. L.; Chow, D.; Wong, S.; Luo, J.; Lin, D. C.; Medina, J. C. Discovery and Optimization of Potent GPR40 Full Agonists Containing Tricyclic Spirocycles. ACS Med. Chem. Lett. 2013, 4, 551−555. (34) Ma, Z.; Lin, D. C.; Sharma, R.; Liu, J.; Zhu, L.; Li, A. R.; Kohn, T.; Wang, Y.; Liu, J. J.; Bartberger, M. D.; Medina, J. C.; Zhuang, R.; Li, F.; Zhang, J.; Luo, J.; Wong, S.; Tonn, G. R.; Houze, J. B. Discovery of the imidazole-derived GPR40 agonist AM-3189. Bioorg. Med. Chem. Lett. 2016, 26, 15−20. (35) The synthesis of compound 11 will be reported elsewhere. (36) Cheang, W. S.; Tian, X. Y.; Wong, W. T.; Huang, Y. The peroxisome proliferator-activated receptors in cardiovascular diseases: experimental benefits and clinical challenges. Br. J. Pharmacol. 2015, 172, 5512−5522.

(37) Copeland, R. A. The drug-target residence time model: a 10-year retrospective. Nat. Rev. Drug Discovery 2016, 15, 87−95. (38) Motulsky, H. J.; Mahan, L. C. The kinetics of competitive radioligand binding predicted by the law of mass action. Mol. Pharmacol. 1984, 25 (1), 1−9. (39) Chernet, E.; Martin, L. J.; Li, D.; Need, A. B.; Barth, V. N.; Rash, K. S.; Phebus, L. A. Use of LC/MS to assess brain tracer distribution in preclinical, in vivo receptor occupancy studies: dopamine D2, serotonin 2A and NK-1 receptors as examples. Life Sci. 2005, 78, 340−346. (40) Rasmussen, S. G.; Choi, H. J.; Fung, J. J.; Pardon, E.; Casarosa, P.; Chae, P. S.; Devree, B. T.; Rosenbaum, D. M.; Thian, F. S.; Kobilka, T. S.; Schnapp, A.; Konetzki, I.; Sunahara, R. K.; Gellman, S. H.; Pautsch, A.; Steyaert, J.; Weis, W. I.; Kobilka, B. K. Structure of a nanobodystabilized active state of the beta(2) adrenoceptor. Nature 2011, 469, 175−180. (41) Srivastava, A.; Yano, J.; Hirozane, Y.; Kefala, G.; Gruswitz, F.; Snell, G.; Lane, W.; Ivetac, A.; Aertgeerts, K.; Nguyen, J.; Jennings, A.; Okada, K. High-resolution structure of the human GPR40 receptor bound to allosteric agonist TAK-875. Nature 2014, 513, 124−127. (42) Einstein, M.; Akiyama, T. E.; Castriota, G. A.; Wang, C. F.; McKeever, B.; Mosley, R. T.; Becker, J. W.; Moller, D. E.; Meinke, P. T.; Wood, H. B.; Berger, J. P. The differential interactions of peroxisome proliferator-activated receptor gamma ligands with Tyr473 is a physical basis for their unique biological activities. Mol. Pharmacol. 2008, 73, 62− 74. (43) Molecular Operating Environment (MOE), version 2015.11; Chemical Computing Group, Inc.: 1010 Sherbooke St. West, Suite 910, Montreal, QC, Canada, H3A 2R7, 2016. (44) Kava, R.; Greenwood, M. R. C.; Johnson, P. R. Zucker (fa/fa) Rat. ILAR J. 1990, 32, 4−8. (45) Deeks, E. D.; Keam, S. J. Rosiglitazone: a review of its use in type 2 diabetes mellitus. Drugs 2007, 67, 2747−2779. (46) Parker, H. E.; Habib, A. M.; Rogers, G. J.; Gribble, F. M.; Reimann, F. Nutrient-dependent secretion of glucose-dependent insulinotropic polypeptide from primary murine K cells. Diabetologia 2009, 52, 289−298. (47) Hamdouchi, C.; Lineswala, J.; Maiti, P. Spiropiperidine compounds and pharmaceutical use thereof for treating diabetes. PCT Int. Appl. WO 2011046851 A1, April 21, 2011. (48) Hamdouchi, C.; Lineswala, J.; Maiti, P. Novel spiropiperidine compounds. PCT Int. Appl. WO 2011066183 A1, June 3, 2011. (49) Hamdouchi, C. A novel 1,2,3,4-tetrahydroquinoline derivative useful in the treatment of diabetes. PCT Int. Appl. WO 2013025424 A1, February 21, 2013. (50) Houze, J. B.; Zhu, L.; Sun, Y.; Akerman, M.; Qiu, W.; Zhang, A. J.; Sharma, R.; Schmitt, M.; Wang, Y.; Liu, J.; Liu, J.; Medina, J. C.; Reagan, J. D.; Luo, J.; Tonn, G.; Zhang, J.; Lu, J. Y.; Chen, M.; Lopez, E.; Nguyen, K.; Yang, L.; Tang, L.; Tian, H.; Shuttleworth, S. J.; Lin, D. C. AMG 837: a potent, orally bioavailable GPR40 agonist. Bioorg. Med. Chem. Lett. 2012, 22, 1267−1270. (51) Chambers, M. S.; Baker, R.; Billington, D. C.; Knight, A. K.; Middlemiss, D. N.; Wong, E. H. Spiropiperidines as high-affinity, selective sigma ligands. J. Med. Chem. 1992, 35, 2033−2039. (52) Efange, S. M.; Khare, A. B.; Foulon, C.; Akella, S. K.; Parsons, S. M. Spirovesamicols: conformationally restricted analogs of 2-(4phenylpiperidino)cyclohexanol (vesamicol, AH5183) as potential modulators of presynaptic cholinergic function. J. Med. Chem. 1994, 37, 2574−2582. (53) Hayashi, S.; Ohashi, K.; Nakata, E.; Emoto, C. Efficient design, synthesis and structure−activity relationship studies of 1-(3′-substituted propyl)-4-arylpiperidines as non-peptide antagonists of nociceptin/ orphanin FQ receptor: biological activities, metabolic stabilities and hERG channel bindings. Med. Chem. Res. 2014, 23, 526−552. (54) Cui, S.; Walker, S. D.; Woo, J. C.; Borths, C. J.; Mukherjee, H.; Chen, M. J.; Faul, M. M. Practical asymmetric conjugate alkynylation of Meldrum’s acid-derived acceptors: access to chiral beta-alkynyl acids. J. Am. Chem. Soc. 2010, 132, 436−437. Y

DOI: 10.1021/acs.jmedchem.6b00892 J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry

Drug Annotation

(55) Enantomeric excess (ee) was determined using analytical HPLC and forward processed to provide the ﬁnal target structure. X-ray crystallography data conﬁrmed the isomer was of the R conﬁguration. (56) GraphPad Prism Software for Windows, version 6.00; GraphPad Software: La Jolla, CA, 2016.

Z

DOI: 10.1021/acs.jmedchem.6b00892 J. Med. Chem. XXXX, XXX, XXX−XXX

The Discovery, Preclinical, and Early Clinical Development of Potent

Recommend Documents