SAR Studies of Indole-5-propanoic Acid Derivatives To Develop Novel GPR40 Agonists

Abstract

G-protein coupled receptor 40 (GPR40) has been considered to be an attractive drug target for the treatment of type 2 diabetes because of its role in free fatty acids-mediated enhancement of glucose-stimulated insulin secretion (GSIS) from pancreatic β-cells. A series of indole-5-propanoic acid compounds were synthesized, and their GPR40 agonistic activities were evaluated by nuclear factor of activated T-cells reporter assay and GSIS assay in the MIN-6 insulinoma cells. Three compounds, 8h (EC₅₀ = 58.6 nM), 8i (EC₅₀ = 37.8 nM), and 8o (EC₅₀ = 9.4 nM), were identified as potent GPR40 agonists with good GSIS effects.

Free fatty acids (FFAs) are an important source of energy for the body and play a role in signal transduction for insulin secretion and other cellular effects.^1,2 G-protein coupled receptor 40 (GPR40) was identified as an orphan G-protein coupled receptor with an unknown function.² Recent studies revealed that GPR40 is abundantly expressed in the pancreas and regulates insulin secretion from beta cells of the pancreas via long-chain FFAs.³ Takeda has developed the GPR40 agonist TAK875 (Figure 1), which activates the glucagon-like peptide-1 (GLP-1) pathway as well as GPR40 and increases the secretion of insulin.⁴ In 2011, a phase III clinical trial of TAK875 was initiated.^5,6 The trial was the first attempt to develop a GPR40 agonist as a drug for type 2 diabetes mellitus. However, its development was discontinued due to potential hepatotoxicity.⁷ Clinical developments of some other GPR40 agonists were also discontinued for undisclosed reasons. Recent studies suggested that the potential liver toxicity of TAK875 and other GPR40 agonists is associated with their analogous hydrophobic pharmacophore. Smaller and less lipophilic compounds have improved the drug-likeness properties.^8,9 As shown in Figure 1, the common feature of most GPR40 agonists is a phenyl propanoic acid group and a benzyloxy or benzyl amine group in the linker unit,¹⁰⁻¹⁵ which are very important for the GPR40 agonistic effect.¹⁶⁻¹⁸ The GPR40 agonist TUG-770 has a rigid alkyne group in the linker part,¹⁰ which should provide a different ligand receptor binding interaction compared to flexible benzyloxy or amino linker.¹⁹

Figure 1.

Reported GPR40 agonists.

Another GPR40 agonist includes a derivative of benzofuran that was designed as a constrained planar derivative of benzyloxy linker. Its agonistic activity is weaker than other derivatives having benzyloxy or alkyne moieties in the linker part.²⁰ In a similar way to develop TAK875, various bicyclic acid moieties including 2-(1H-indol-1-yl)acetic acid or 3-(1H-indol-1-yl)acetic acid are adopted in the acid tail part of GPR40 agonists with benzyloxy linker. However, their activities are weaker than those with a phenyl propanoic acid tail (EC₅₀ values ranging between 1 and 10 μM).²¹ Based on these reports, we designed novel GPR40 agonists by employing an indole moiety in which a phenyl ring is fused to a pyrrole ring as a bioisostere of two-atom linker unit²² and a small-size aryl group to reduce the hydrophobicity (Figure 2). A series of 3-(2-aryl-1H-indol-5-yl) propanoic acid derivatives was prepared for the present structure–activity relationship study. We expected binding modes of these molecules would be similar to that of TAK875 to give an agonistic effect on GPR40,²³ although their indole linker unit is more rigid than the benzyloxy linker.

Figure 2.

Design of indole propanoic acid.

The synthesis of key intermediate 5 was achieved as shown in Scheme 1. The esterification of commercially available 3-(4-aminophenyl) propanoic acid 1 under sulfuric acid provided compound 2. To prepare 2-ethynylaniline 5, the iodination of compound 2 was carried out first. The Sonogashira coupling reaction was conducted to replace the iodo group with ethynyl group (compound 4) and then followed by TMS (trimethylsilyl ether) deprotection using TBAF (tetrabutylammonium fluoride) directly to afford compound 5. Compound 5 is an important intermediate for the synthesis of indole propanoic acid derivatives, as presented in Scheme 2, because it can employ various aryl or alkyl groups at the C2 position of indole ring. Among the various methods for the synthesis of indole, we used a method that cyclizes the acetylenic substituent onto the amino group.

Scheme 1.

Reagents and Conditions: (a) sulfuric acid, EtOH, reflux, 98%; (b) NIS, DMF, r.t., 90%; (c) TMS-acetylene, PdCl₂(PPh₃)₂, Cul, TEA, THF, 60 °C, 73%; (d) TBAF (1 M in THF), THF, r.t., 82%.

Scheme 2.

Reagents and Conditions: (a) aryl halide, PdCl₂(PPh₃)₂, CuI, TEA, DMF, 80 °C; (b) AuCl₃, EtOH, 60 °C; (c) NaOH, MeOH, 40 °C.

The synthesis of indole propanoic acid derivative was accomplished as summarized in Scheme 2. Various phenyl groups were introduced to the key intermediate 5 by the Sonogashira reaction. In this reaction, the bulkiness of the ortho-substituent of aryl halide affected the yield, and the diyne compound was obtained as a main side product. For the synthesis of indole, gold chloride (AuCl₃) was used, and the indole compound 7 was easily obtained under mild reaction conditions.²⁴ Finally, the resulting indole compounds were hydrolyzed to give the final products, 8a–8r.

We conducted a nuclear factor of activated T-cells (NFAT)-luciferase reporter assay²⁶ in Chinese hamster ovary (CHO) cells to evaluate the GPR40 agonistic activities of the compounds in comparison with TAK875, respectively (Figure S1 in the Supporting Information). Our reporter assay system was validated using the reference molecule TAK875, giving the EC₅₀ value of 14 nM (Table 1), which is similar to that from the calcium influx assay.¹¹ In terms of lipophilicity, the calculated logP (ALogP) values for all compounds, except 8m, were less than that of TAK875 (5.06), indicating that hydrophobicity is reduced in indole propanoic acid derivatives as expected.

Table 1. 2-Aryl Substituted Indole Propanoic Acid.

^aLuciferase reporter assay.

^bAlogP²⁵ was predicted by Schrödinger Maestro.

^cPercent activity compared to 1 μM rosiglitazone.

^dPPARγ EC₅₀ = 2.4 μM. N.A., no activity (>25 μM).

Since another antidiabetic target PPARγ is also activated by fatty acids, it is plausible that GPR40 agonists may have PPARγ agonistic activity.²⁷ Several indole-5-acetic acid and α-alkoxy-indole-5-propionic acid derivatives, which are structurally similar to our compounds, were also reported as PPAR agonists.^28,29 Thus, we examined the PPARγ transactivation activity of our compounds at 10 μM (Tables 1 and 2), and most of them did not show significant PPARγ agonist activity, except 8i. The EC₅₀ values for PPARγ were determined for 8i, 8k, and 8r, and they were 2.4, 236.4, and 56.9 μM, respectively (Figure S2 in the Supporting Information). Although 8i showed some PPARγ agonist activity, it is about 60 times more selective for GPR40.

Table 2. Effect of Ortho-Substiuted Aryl 2-Indole Propanoic Acid.

^aLuciferase reporter assay.

^bAlogP²⁵ was predicted by Schrödinger Maestro.

^cPercent activity compared to 1 μM rosiglitazone. N.A., no activity (>25 μM).

To examine the effects of substituents on the phenyl ring, the alkyl- (methyl, methoxy, ethyl, or isopropyl) substituted derivatives did not show good GPR40 agonistic activity, while the halogen substitution resulted in significant agonistic activity (Tables 1 and 2). Chloro substitution (8h–8j) provided good activity in the order meta ≈ ortho > para. At first, we examined if the meta-substitution is important for activity, meta-bromophenyl analogue 8k was synthesized, but its activity was about 50-fold lower than the meta-chloro derivative 8i. Therefore, we focused on the ortho-substitution assuming that the steric effect of the ortho-substituent may adjust the dihedral angle between the indole ring and the phenyl ring optimal for receptor binding (Table 2). Various ortho-substituted derivatives were prepared, and their agonistic effects were evaluated. Among the alkyl substituted analogues, only o-methyl phenyl compound 8b showed submicromolar activity. Derivatives with o-CF₃ (8l), o-OCF₃ (8m), and o-F (8n) exhibited moderate micromolar potency. The compound having o-Br substituent (8o) showed the highest agonistic activity (EC₅₀ = 9.4 nM), while the m-Br derivative showed almost 200 times lower activity (Table 1). These results suggest that a substituent at the ortho position of the aryl group may not only be a simple steric blocker to adjust the conformation of the ligand but may also be involved in interaction with the receptor. The halogen atom is preferred at the ortho-position, and the agonistic activity decreases in the order Br > Cl > F > H.

To understand the dramatic difference in GPR40 agonistic activity between o-Br (8o) and m-Br (8k) based on their binding modes, docking simulations were carried out using the X-ray structure of TAK875-GPR40 complex (pdb id 4PHU).²³ Most GPR40 agonists have a carboxylic acid moiety as a key pharmacophore. The analysis of the X-ray structure revealed a polar interaction network between the carboxylate group of TAK875 and Arg183, Arg258, Tyr91, and Tyr240, which is crucial for GPR40 agonistic activity. Likewise, in the docked models of GPR40 with 8o and 8k within the TAK875-binding pocket (Figure 3), the carboxylate moiety also forms hydrogen bonds with Arg183 and Arg258. In addition, indole N–H interacts with backbone amide carbonyl oxygen of Leu138 by forming a H-bond, suggesting that indole NH group may affect the receptor binding affinity. To verify the significance of indole NH for GPR40 agonistic effect, we prepared and tested several N-methylated indole derivatives of 8i, 8j, and 8o and found that they almost lost activities with EC₅₀ > 10 μM (8s, 8t, and 8u in the Supporting Information Table S1).

Figure 3.

Computational analysis of the binding mode of 8o (A) and 8k (B) in GPR40. The carboxylate moiety is highly coordinated by several key residues. Hydrogen bond interactions (<3 Å) are depicted as dashed cylinders. Gray capped sticks represent key amino acid residues within the binding site and the ribbon is the backbone of hGPR40. The binding site is rendered in MOLCAD lipophilic potential surface and colored from blue (hydrophilic) to dark brown (hydrophobic).

We also identified the unique Br···π interaction^30,31 between the Br substituent of drug and Phe142 in the binding site, as a possible key factor for GPR40 agonistic activity (Figure 3A). In the binding pose of 8o, the distance from o-Br to the centroid of the phenyl ring in Phe142 is 3.71 Å, and that to the nearest carbon of phenyl ring is 3.33 Å, giving a difference of 0.38 Å. Since this difference is bigger than 0.3 Å, the Br···π interaction can be formed with an “edge-on” geometry. However, for the m-Br analogue (8k), the distance from m-Br atom to the centroid of the phenyl ring in Phe142 is 5.25 Å (Figure 3B), and that to the nearest carbon of the phenyl ring is 4.46 Å. It means the Br···π interaction cannot be formed because both distances are longer than 4.2 Å. Thus, the Br···π interaction might contribute to provide higher binding affinity toward GPR40 for 8o.

In most of the reported GPR40 agonists with a propanoic acid functional group, some substituents on its beta position were employed to improve the metabolic stability by avoiding β-oxidation. Therefore, we inspected a liver microsomal stability of compound 8h, 8i, and 8o, and they exhibited good metabolic stability (Table 3).

Table 3. Metabolic Stability of Compounds 8h, 8i, and 8o.

Compd	CLint. (human/mouse)a
8h	9.2/14.3
8i	4.3/11.3
8o	7.1/10.4

^aMicrosomal stability: clearance intrinsic (μL/min/mg).

To detect insulin secretion effects of derivatives having GPR40 agonistic activity in the NFAT luciferase assay, we performed a glucose-stimulated insulin secretion (GSIS) assay in MIN6 cells.¹ The insulin secretion effect was verified at a glucose concentration of 17 mM. For compounds, 8h, 8i, and 8o, insulin secretion was elevated by 1.4-, 1.7-, and 1.6-fold compared to 17 mM glucose, respectively (Figure 4). Compounds 8i and 8o yielded GSIS effects similar to that of the reference compound TAK875 (1.7-fold). In contrast, the compound 8a with low GPR40 agonist activity (EC₅₀ > 25 μM) did not induce insulin secretion. The result suggests that the GSIS effect of 8h, 8i, and 8o are related to selective GPR40 activation.

Figure 4.

Glucose-stimulated insulin secretion test of derivatives 8a, 8h, 8i, and 8o in MIN6 cells. The data is represented as the mean ± standard error (n = 2). ^###p ≤ 0.01 by Student’s test; *p ≤ 0.05, **p ≤ 0.025, ***p ≤ 0.01 vs 17 mM glucose alone by Student’s test.

In conclusion, for development of novel GPR40 agonists, we utilized indole propanoic acid moiety as a simplified novel scaffold for a potent GPR40 agonist. Various derivatives were prepared, and their GPR40 agonistic activities were evaluated by an NFAT luciferase assay. Among these derivatives, potent GPR40 agonists were identified. These agonists, 8h, 8i, and 8o, showed good microsomal stability as well as an excellent insulin secretion effect in MIN6 cells. Based on these results, we continue to perform further in vivo efficacy studies for the development of these GPR40 agonists as antidiabetic candidates.

Supporting Information Available

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acsmedchemlett.7b00460.

• Syntheses and characterization data for compounds 8a–8u, assay protocols, and computational modeling method (PDF)

Author Contributions

^∥ Theses authors contributed equally to this work.

The authors declare no competing financial interest.