Discovery of a Novel Series of Pyridone-Based EP3 Antagonists for the Treatment of Type 2 Diabetes

Abstract

A novel series of pyridones were discovered as potent EP3 antagonists. Optimization guided by EP3 binding and functional assays as well as by eADME and PK profiling led to multiple compounds with good physical properties, excellent oral bioavailability, and a clean in vitro safety profile. Compound 13 was identified as a lead compound as evidenced by the reversal of sulprostone-induced suppression of glucose-stimulated insulin secretion in INS 1E β-cells in vitro and in a rat ivGTT model in vivo. A glutathione adduction liability was eliminated by replacing the naphthalene of structure 13 with the indazole ring of structure 43.

The EP3 receptor belongs to the EP receptor family of four distinct subtypes EP1, EP2, EP3, and EP4 that are activated endogenously by prostaglandin E₂ (PGE2), a lipid mediator produced from arachidonic acid by sequential action of COX-1/COX-2 and PGE synthase. The EP3 receptor is a 7-transmembrane G-protein coupled receptor found in various human tissues including the kidney, uterus, bladder, stomach, and brain. EP3 is a G-protein coupled receptor (GPCR), the activation of which inhibits cAMP production though the G_i alpha signaling pathway. This coupling is distinct from other EP receptor family members that signal through G_q or G_s.¹⁻³ EP3 receptor antagonism has been linked to numerous therapeutic areas including inflammation for the treatment of pain,⁴ overactive bladder (OAB),⁵ diabetes,⁶ and cardiovascular disease, specifically atherothrombosis, which contributes to myocardial infarct (MI) and stroke.^7,8 Clinical studies have provided strong evidence of the role of increased levels of PGE2 as a contributor to defective insulin secretion in diabetic patients. Recently, the functional link between PGE2 suppression of glucose-stimulated insulin secretion (GSIS) and the EP3 receptor was confirmed using β-cell lines and isolated islets.⁶ It is hypothesized that increased PGE2 signaling through the EP3 receptor might be coincident with the development of diabetes and contribute to β-cell dysfunction. Therefore, EP3 receptor antagonists may be an effective treatment for type 2 diabetes by relieving the inhibitory action of PGE2 to partially restore defective GSIS in diabetic patients.

In light of the potential therapeutic utility of EP3 antagonists, there have been a variety of reports on the identification of small molecule EP3 antagonists in the literature. Representative examples of previously reported EP3 antagonists are listed in Figure 1. MK-11b⁹ and DG-041¹⁰ contain an acylsulfonamide as a carboxylic acid isostere. A series of phenyl propionic acids (1) has been reported by ONO Pharmaceutical.¹¹ GSK-29, an oxazolidinedione-containing EP3 antagonist, showed in vivo activity in OAB animal models.¹² Two structurally similar amide-based EP3 antagonists (2) have been described by GSK and Pfizer.^13,14 Recently, an amide-based pyridone (3) was reported to be a potent EP3 antagonist by Pfizer.¹⁵ To date, only DG-041 was reported to have entered phase II clinical trials for atherothrombosis. In pursuit of potent EP3 antagonists with therapeutic potential for type 2 diabetes, we were attracted to the pyridone-based chemotype due to its balanced EP3 activity and drug-like properties.^12,15 Herein, we report our efforts to optimize the in vitro potency and oral bioavailability of scaffold I for the treatment of metabolic diseases.

Figure 1.

Representative EP3 antagonists.

We have recently described our initial exploration on the scaffold I (Figure 1).¹⁶⁻¹⁸ Preliminary structure–activity relationship (SAR) studies indicated: (1) the A-ring of I could be readily modified and EP3 binding affinity and functional activity maintained; (2) the core pyridone ring was essential for EP3 activity with limited tolerance for further modification; (3) the B-ring provided a secondary site for optimization of activity and physicochemical properties.

Given the tendency to maintain hEP3 binding affinity and functional potency, several 5-membered heterocyclic rings (Ia) including oxazolidine-2,4-dione, hydantoin, and thiazolidine-2,4-dione were selected for an SAR study of the A-ring as shown in Table 1. Compounds were made as racemates and screened in an in vitro radioligand displacement binding assay as well as a cAMP functional assay with CHO cells overexpressing human EP3 (hEP3) receptor. We first examined the effect of A-rings linked to the i-propyl and pyridone core through a double bond. While oxazolidine-2,4-diones 4 and 5 only displayed K_i of 140 and 270 nM in the hEP3 binding assay, respectively, thiazolidine-2,4-dione 6 and methyl hydantoin 7 significantly enhanced both hEP3 binding affinity and functional potency, suggesting a hydrophobic A-ring might favor stronger ligand–receptor binding. Compounds (4–7) displayed good metabolic stability as indicated by intrinsic clearance half-life (CL_intT_1/2) values of >180 min in human and rat liver microsomal preparations. We then followed up with the slightly more lipophilic double bond reduced adducts 8 to 12 (higher cLogP values) to improve EP3 activity. In general, switching from a double to a single bond between the A-ring and the linkage to the i-propyl group and the pyridone core greatly improved or maintained hEP3 binding affinity and functional potency. However, the increased flexibility of the single bond also resulted in analogues with reduced metabolic stability, especially in rat liver microsomes, which could present some challenges for in vivo evaluation. As part of the effort to explore the linker between the pyridone core and the A-ring, we synthesized a subset of one-carbon homologues (13–22) of compounds 4–7 and 8–12. These homologues (13–22) possessed both enhanced EP3 binding affinity and functional potency. Compared to the saturated A-ring derivatives (14, 15, 17, 19, 21, 22), the unsaturated analogues (13, 16, 18, 20) exhibited a much longer T_1/2 in human and rat liver microsomes while maintaining comparable EP3 binding affinity and functional potency. Among the compounds, oxazolidine-2,4-dione 13 seemed to strike the best balance of hEP3 binding affinity (K_i of 9.0 nM), antagonistic potency (IC₅₀ of 21 nM), and metabolic stability (T_1/2 > 180 min in human and rat liver microsomes), warranting further pharmacokinetic and pharmacodynamic characterization. Pyridones Ia in Table 1 generally possessed good solubility and permeability with an acceptable range of cLogP values, providing a good starting point for further optimization of this series.

Table 1. SAR of the Acidic Heterocycle R of Pyridones Ia.

^aPTGER3_SPA_ [³H]-PGE2_binding EP3 competition binding assay, average of two tests for K_i < 100 nM.

^bPTGER3_C _cAMP antagonist assay on CHO cells, average of two tests for IC₅₀ < 100 nM.

^cnt: not tested.

^dcLogP is calculated by BioByte software.

Next, we shifted our effort to the R¹–R⁴ substituents of structure Ib shown in Table 2. It is documented in the literature that unsaturated and conjugated oxazolidine-2,4-dione, hydantoin, and thiazolidine-2,4-dione moieties are Michael acceptors that can potentially result in covalent attachment of glutathione (GSH).^19,20 Thus, our strategy to eliminate this potential adverse effect focused on blocking addition to the Michael acceptor with substituents R¹, R², and R³ next to the double bond. Arbitrarily assigned R* enantiomer 13R displayed comparable EP3 binding affinity and antagonistic potency as arbitrarily assigned S* enantiomer 13S. Therefore, we sought to conduct the SAR study using the racemic mixture to avoid labor-intensive chiral separation. Analogues 23 and 24 bearing Cl and Me substituents at the R⁴ position had reduced hEP3 binding affinity and functional potency, suggesting there may be limited steric bulk tolerance in this area of the receptor. In contrast, substituents in the R¹–R³ positions were generally well tolerated, producing analogues with good binding affinity and antagonistic potency, except for Ph (30) or Bn (31) at the R¹ position. Sterically bulky alkyl groups at the R¹ position such as 3-pentyl (26), cyclopentyl (28), and cyclohexyl (29) as well as a methyl group at the R³ position (25) were installed to reduce the potential for GSH adduction. However, metabolic stability tended to decrease with increasing lipophilicity, as indicated by the shorter T_1/2 of compounds 28 and 29 in human liver microsomes. The introduction of geminal substituents at the R¹ and R² positions through bis-alkylation (32 and 33) or cyclization of the R¹ and R² groups into a carbocyclic ring (34 and 35) became an effective way to block potential GSH adduction by increasing steric hindrance. In fact, compounds 32–35 displayed strong binding affinity and good EP3 antagonistic potency while maintaining moderate to good metabolic stability and solubility.

Table 2. SAR of R¹, R², R³, and R⁴ Groups of Oxazolidinedione Containing Pyridones Ib.

								CLintT1/2 (min)		kinetic solubility (μM)
ID	R1	R2	R3	R4	cLogPd	hEP3 binding Ki (nM)a	hEP3 cAMP IC50 (nM)b	h	r	PBS	pH = 2
13	i-propyl	H	H	H	3.13	9.0	21	>180	>180	28.0	8.0
13R	i-propyl (R*)	H	H	H	3.13	11	10	>180	>180	25.4	10.8
13S	i-propyl (S*)	H	H	H	3.13	7.0	9.0	>180	>180	26.4	4.62
23	i-propyl	H	H	Cl	3.79	380	360	>180	>180	21.5	3.20
24	i-propyl	H	H	Me	3.33	23	92	>180	>180	35.7	39.4
25	i-propyl	H	Me	H	3.53	17	39	>180	>180	17.2	12.7
26	3-pentyl	H	H	H	4.19	13	23	>180	>180	33.7	3.68
27	cyclopropyl	H	H	H	2.65	22	33	119	>180	30.8	3.85
28	cyclopentyl	H	H	H	3.77	9.0	7.0	72.7	>180	ntc	nt
29	cyclohexyl	H	H	H	4.33	4.0	8.0	65.6	143.3	nt	nt
30	Ph	H	H	H	3.25	23	190	126	>180	32.8	15.4
31	Bn	H	H	H	3.62	33	200	77.8	177.3	9.1	0.9
32	Me	Me	H	H	2.61	38	57	113.5	>180	36.1	1.66
33	Et	Et	H	H	3.66	13	61	56.9	>180	32.6	13.5
34	cyclopropyl		H	H	2.24	28	55	149	>180	33.4	27.4
35	cyclopentyl		H	H	3.36	50	66	>180	>180	24.5	2.55

^aPTGER3_SPA_ [³H]-PGE2_binding EP3 competition binding assay, average of two tests for K_i < 100 nM.

^bPTGER3_C _cAMP antagonist assay on CHO cells, average of two tests for IC₅₀ < 100 nM.

^cnt: not tested.

^dcLogP is calculated by BioByte software.

Having established the SAR at the R¹–R⁴ positions of structure Ib, we then switched our effort to the R-ring of structure Ic in Table 3. In contrast to the A-ring of structure I, this area was less tolerant of modification according to both reports from the literature and our preliminary study.^12,16−18 Nevertheless, it is essential to optimize the R-ring of structure Ic since the naphthyl group was reported to be a site for GSH adduction.¹² Moreover, we hypothesized that further reducing cLogP values of compounds 32–35 can improve metabolic stability in human liver microsomes. Given the constraints in modifying the R-ring, two strategies were adopted to block potential GSH adduction to the naphthyl ring of Ia and Ib, including polyfluorination and replacement with fused heterobicyclic ring systems. Not surprisingly, most modifications, although extremely minor, either reduced or completely abolished hEP3 binding affinity and functional potency. For example, adding a fluoro substituent at the 6-position of the F-naphthyl ring (36) resulted in only a 2-fold loss of EP3 binding affinity (K_i = 28 nM for 36 vs 13 nM for 33). Quite strikingly, shifting the difluoro substituents from the 6,8- (36) to the 5,7-positions (37) decreased binding affinity by more than 10-fold (K_i = 290 nM for 37 vs 28 nM for 36). All attempts to replace the 8-fluoro substituent with an alkyl group such as ethyl (38), isopropyl (39), or trifluoromethyl (40) failed to maintain either good EP3 binding affinity or antagonistic potency. The metabolic stability of the alkyl naphthyl analogues was also reduced. Further modifications of the R-ring generally did not improve hEP3 binding affinity and functional potency in our preliminary study. Therefore, only selected heterocyclic R-rings were examined consistent with literature precedence.¹⁵ Compound 41 bearing the chromane ring reduced the binding affinity to the low micromolar range and abolished functional potency. In contrast, incorporation of a 4-ethyl-3,4-dihydro-2H-benzo[b][1,4]oxazine ring (42), precedented by Pfizer’s report on another EP3 antagonist chemotype,¹⁴ was tolerated with only a 2-fold drop of the EP3 binding affinity compared to compound 13. Encouraged by this finding, an ethylated fluoro-indazole moiety was introduced to replace the benzoxazine group to give compound 43 with excellent EP3 binding affinity (K_i of 3.0 nM) and EP3 antagonistic potency (IC₅₀ of 33 nM). Indazole 43 exhibited improved metabolic stability and a comparable solubility profile as the naphthyl analogue 33 probably due to reduced lipophilicity as evidenced by the lower cLogP value. A clear preference for the 1-ethyl-4-fluoroindazol-6-yl isomer (43) was evident, as isomers 44 and 47 demonstrated a marked loss of EP3 binding affinity and/or functional potency. Further minor modifications of the indazole ring were not tolerated, as indicated by methyl indazoles 45, 46 and fluoro-iso-indazole 48 with weak EP3 binding affinity and minimal functional activity. Overall, indazole 43 was identified as an effective substitute for naphthene 33 to mitigate potential GSH adduction and to provide improved physical properties for PK and PD evaluation.

Table 3. SAR of R Rings of Oxazolidinedione Containing Pyridones Ic.

^aPTGER3_SPA_ [³H]-PGE2_binding EP3 competition binding assay, average of two tests for K_i < 100 nM.

^bPTGER3_C _cAMP antagonist assay on CHO cells, average of two tests for IC₅₀ < 100 nM.

^cnt: not tested.

^dcLogP is calculated by BioByte software.

Selected compounds in Tables 1–3 were evaluated in a 24 h rat pharmacokinetic study at an oral dose of 10 mg/kg and an iv dose of 2 mg/kg (Table 4). Compound 13 demonstrated good bioavailability (97.7%) and high exposure in plasma (AUC_last = 158570 ng h/mL) after oral dosing. It had low clearance (1.06 mL/min kg), low volume of distribution at steady state (Vss = 0.23 L/kg), and high plasma protein binding, likely due to the acidic oxazolidinedione moiety. Not surprisingly, GSH adducts were observed when compound 13 was incubated with human or rat liver S9 fractions in the presence of NADPH. Subsequent metabolite ID studies suggested that GSH adduction occurred through 1,4-Michael addition of glutathione to the α,β-unsaturated carbonyl moiety of 13 both enzymatically and nonenzymatically. Removal of the Michael acceptor by reduction of the double bond afforded compound 14, which displayed similar pharmacokinetic and plasma protein binding profiles to 13. However, compound 14 still formed a GSH adduct when incubated in human or rat liver microsome in the presence of NADPH. The mass spectra suggested that bioactivation occurred on the fluoronaphthalene moiety of 14 probably through epoxidation, opening of the epoxide with glutathione, and elimination of hydrogen fluoride (see the Supporting Information).^12,13 Increasing the steric bulk around the α,β-unsaturated carbonyl moiety of 13, as in tetra-substituted alkene 25, reduced the oral exposure (C_max = 1227 ng/mL, AUC_last = 12 703 h·ng/mL) and increased the clearance (CL = 15.6 mL/min/kg) in the rat. Introduction of a gem-diethyl moiety in place of the isopropyl group effectively blocked NADPH-dependent GSH conjugate addition to the α,β-unsaturated carbonyl moiety of compound 33. Nevertheless, the fluoro-naphthyl ring of 33 was still subject to NADPH-dependent bioactivation and GSH trapping through a similar mechanism to that of 14. Finally, no NADPH-dependent GSH adduction was detected for compound 43 when using either rat or human liver S9 fractions, which demonstrated that bioactivation liabilities could be mitigated by blocking the α,β-unsaturated carbonyl moiety and replacing the fluoro-naphthyl group with heteroaryl groups. Although highly plasma protein bound, compound 43 exhibited a satisfactory rat PK profile as evidenced by low clearance, long half-life and high oral exposure.

Table 4. Profiles of Selected Compounds.

	rat PK parameters						PPB		NADPH-dependent GSH adduction
ID	CL (mL/min/kg)	Vss (L/kg)	T1/2 (h)	Cmax (ng/mL)	AUC (h·ng/mL)	F (%)	human (% bound)	rat (% bound)	human	rat
13a	1.06	0.23	4.6	24 625	158 570	97.7	99.92	99.99	+e	+e
14b				44 100	217 513		99.68	99.79	+f	+f
25c	15.6	1.96	4.0	1227	12 703	118	ntd	nt	nt	nt
33b				58 100	327 238		nt	nt	+f	+f
43c	0.93	0.24	9.8	8737	100 395	54.9	99.89	99.82	_	_

^aSD full PK: IV, 2 mg/kg in 20% HPbCD (n = 3); PO, 10 mg/kg in 0.5% HPMC (n = 3).

^bSD fast PK: PO, 10 mg/kg in 0.5% HPMC (n = 3).

^cSD full PK: IV, 2 mg/kg in 20% HPbCD (n = 3); PO, 10 mg/kg in 20% HPbCD (n = 3).

^dnt = not tested.

^e1,4-Michael addition of GSH to the α,β-unsaturated carbonyl moiety.

^fThe fluoronaphthalene ring underwent epoxidation and addition of GSH.

Due to its excellent oral bioavailability (F = 97.7%), compound 13 was selected as a tool compound for evaluation in preclinical disease-related models of EP3 agonist-induced inhibition of glucose-stimulating-insulin-secretion (GSIS) as a proof of concept study. As shown in Figure 2, insulin secretion assays were performed in a highly glucose-responsive insulinoma cell line (INS 1E β-cells) with 12 mM stimulating glucose and various doses (0.03–10 μM) of 13 in the presence or absence of 100 nM sulprostone, an EP3-specific agonist. Sulprostone elicited a significant decrease of GSIS in INS 1E β-cells. The inhibitory effect of a 100 nM dose of sulprostone, chosen to be 10 times the EC₅₀ value, could be completely reversed by increasing concentrations of compound 13 from 0.03 to 10 μM. By comparison, DG-041 at 10 μM comparably relieved the inhibitory effect of sulprostone to restore defective GSIS in INS 1E β-cells.

Figure 2.

Effect of compound 13 on reversing EP3 agonist-induced suppression of GSIS in INS 1E β-cells

EP3 receptor-mediated suppression of GSIS has been validated in vivo using a dose-range (3–30 μg/kg/min) of sulprostone in conscious rat intravenous glucose tolerance tests (ivGTTs).²¹ Using this model, we could define the EP3 receptor antagonist plasma exposure required to oppose maximal EP3 receptor-mediated GSIS suppression. Compound 13 had a K_i value of 1 nM in a rat EP3 receptor binding assay and was considered to be a suitable candidate for evaluation of the effect on circulating insulin levels in the sulprostone infusion rat ivGTT model (Figure 3). Intravenous infusion of compound 13 in 20% HPbCD was shown to reverse suppression of GSIS by sulprostone (infused @ 3 μg/kg/min) at plasma concentrations equivalent to 7× K_i = 12.24 μM and 10× K_i = 16.61 μM in SD rats. The robust in vivo activity in the ivGTT indicated that this novel and potent EP3 receptor antagonist series could potentially lead to effective therapeutic agents for treating type 2 diabetes.

Figure 3.

Effect of compound 13 on reversing suppression of GSIS by sulprostone in the rat.

The detailed synthetic routes to 4–48 are described in the Supporting Information. The synthesis of representative compound 13 is outlined in Scheme 1. 2-Bromo-6-methoxypyridine (49) was coupled with 2-(8-fluoronaphthalen-2-yl)-4,4,5,5-tetramethyl-1,3,2-dioxaborolane (50) under Suzuki coupling conditions to afford the corresponding biaryl adduct 51 in 70% yield. ortho-Metalation of 51 with t-BuLi at −78 °C followed by reaction with diethyl oxalate gave the corresponding oxo-acetate adduct, which was then reduced with triethylsilane in TFA at 90 °C to yield ester 52. Alkylation of ester 52 was conducted with t-BuOK and i-PrI at 0 °C, followed by LAH reduction and Dess–Martin periodinane oxidation to give the corresponding aldehyde 53. Aldol condensation of 53 with oxazolidine-2,4-dione and t-BuLi/LiCl followed by SOCl₂ promoted dehydration delivered compound 54, which was then demethylated by in situ generated TMSI to afford compound 13.

Scheme 1.

Reagents and conditions: (1) Pd(dppf)Cl₂·CH₂Cl₂, Na₂CO₃, DME/water (4:1), 90 °C for 16 h (70%); (2) t-BuLi, diethyl oxalate in THF, −78 °C (85%); (3) triethylsilane, TFA, 90 °C for 6 h (79%); (4) t-BuOK, 2-iodopropane in DMF, 0 °C to rt for 3 h (65%); (5) LAH in THF at 0 °C for 1 h (80%); (6) (1,1,1-triacetoxy)-1,1-dihydro-1,2-benziodoxol-3(1H)-one in DCM at rt for 1 h (77%); (7) t-BuLi, LiCl, oxazolidine-2,4-dione in THF at −78 °C for 30 min; (8) SOCl₂, pyridine in THF at 0 °C for 4 h (63% for 2 steps); (9) TMSCl, NaI in ACN, rt 2 h (43%).

In conclusion, we have discovered a novel series of potent pyridone-based EP3 antagonists through extensive SAR studies. Optimization guided by EP3 binding and functional assays as well as by eADME and PK profiling led to several potent EP3 antagonists with good physical properties, excellent oral bioavailability, and a clean in vitro safety profile. Compound 13 was identified as a lead compound with demonstrable EP3 antagonist activity as evidenced by the reversal of sulprostone-induced suppression of GSIS in INS 1E β-cells in vitro and in the rat ivGTT model in vivo. The GSH adduction liability inherent in 13 was eliminated with the design of indazole-based analogue 43. Overall, these novel compounds are potentially useful therapeutic agents for treating type 2 diabetes.

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsmedchemlett.0c00667.

• Experimental procedures for the synthesis and characterization of 4–48 as well as in vitro and in vivo biological protocols (PDF)

The authors declare no competing financial interest.