Structure–Activity Relationship Studies of 3- or 4-Pyridine Derivatives of DS-6930

Abstract

Derivatization efforts were continued to discover backups for a potent selective PPARγ modulator, DS-6930. In this Letter, the replacement of 2-pyridine ring in DS-6930 with 3- or 4-pyridyl group is reported. As the introduction of substituents on the pyridine ring did not provide potent partial agonists, modifications of benzimidazole ring were explored to discover potent intermediate agonists. 4′-Alkoxy substituted benzimidazoles failed to show potent efficacy in vivo, whereas 7′-fluoro benzimidazole 3g (DS19161384) was found to result in robust plasma glucose reductions with excellent DMPK profiles.

Thiazolidinedione (TZD)-based PPARγ full agonists, pioglitazone hydrochloride (I, Figure 1) and rosiglitazone maleate (II, Figure 1), demonstrated to have robust pharmacological efficacy in clinical settings.¹ However, they attracted much attention owing to their adverse effects. The short-term usage of TZD drugs often causes peripheral edema.² It has also been reported that their long-term use is often associated with bone fracture, carcinogenicity, and cardiovascular risks.^3,4 Such adverse effects have limited their usage because of safety concerns.

Figure 1.

Chemical structures of pioglitazone hydrochloride (I), rosiglitazone maleate (II), and DS-6930 (III).

Although it was reported that the inhibition of PPARγ phosphorylation by Cdk5 is the underlying mechanism of antidiabetic efficacy,⁵ we believe there is still scope for the development of safer PPARγ modulators through the selective regulation of cofactors. Such a conventional approach led to the discovery of DS-6930 (III, Figure 1).⁶ DS-6930 demonstrated potent plasma glucose (PG) reduction with fewer PPARγ-related adverse effects than rosiglitazone in preclinical studies.⁶ Based on the cocrystal structure of III bound to PPARγ ligand binding domain (LBD), the avoidance of direct interactions with Tyr473 on helix 12 and the additional lipophilic interactions of dimethylpyridyl group are suggested to be related to the intermediate agonist activity of DS-6930.⁶ The superior safety profile of DS-6930 was attributable to this distinct binding mode through the selective recruitment of cofactors.⁶ We continued our exploration of structure–activity relationships (SARs) of this series of compounds to discover backup clinical candidates for DS-6930. Herein, we report SAR studies of 3- or 4-pyridines to identify novel PPARγ intermediate agonist 3g (DS19161384), which demonstrated robust pharmacological efficacy with excellent DMPK properties in preclinical studies.

We reported that the high lipophilicity caused the elevation of liver enzyme activities, and the suppression of lipophilicity was achieved by the replacement of phenyl ring with 2-pyridyl group, which led to the discovery of DS-6930.⁶ As the derivatization of 3- or 4-pyridines has generally been avoided owing to their poor potencies in vitro, we have focused on them as backups.

The SAR results of 3-pyridines with in vitro ADME profile are summarized in Table 1. Unsubstituted 3-pyridine 1a possessed poor agonist potency. Unlike the previous SAR of the 2-pyridine series,⁶ the introduction of a methyl group did not result in good potency in vitro, despite the improvement of membrane permeability (compounds 1b–1e). Compounds 1a–1e exhibited comparable aqueous solubility and microsomal stability as those of III. Accordingly, further modifications were explored owing to the attractive in vitro ADME profile of 1a–1e. When a chloro group was incorporated in each position of the pyridine ring, none of them had EC₅₀ < 1 μM (data not shown). Improved in vitro potency was observed for 4- or 5-methoxy derivatives (1g and 1h), whereas the introduction of 2-methoxy group did not result in high potency (compound 1f). Although 4-methoxy analog 1g exhibited the best in vitro potency among 3-pyridines, 1g lost partial agonist activity. Full agonists were also found in methyl analogs (1c and 1d). Further, introduction of a substituent in 1f or 1g led full agonists 1i–1k. In contrast, dimethyl derivative 1l exerted modest potency, with partial agonist activity in vitro (E_max = 36%). In terms of PPARα selectivity, only 1k showed relatively high PPARα activity.

Table 1. PPAR Transcriptional Activities and in Vitro ADME Profiles of 1 and 2.

cmpd	X	Y	PPARγ EC50 (nM)a	PPARγ Emax (%)a	PPARα EC50 (nM)a	PPARα Emax (%)a	Log D	PAMPA Papp (10–6 cm/s)	solubility (μg/mL)	MS (%, h/r/mon)b
III			41 ± 10d	68 ± 8.0d	>10000	13 ± 2.0d	1.4	13 ± 3.4d	31	NTc/92/80
1a	H		6831	66	>10000	5	0.5	<2.0	23	100/NTc/NTc
1b	2-Me		>10000	77	>10000	3	1.0	2.6	54	98/NTc/NTc
1c	4-Me		543	110	>10000	4	0.8	5.9	71	87/NTc/NTc
1d	5-Me		779	104	>10000	5	1.0	2.4	43	NTc
1e	6-Me		>10000	45	>10000	9	0.9	4.1	13	100/NTc/NTc
1f	2-OMe		1193	59	>10000	8	0.8	3.0	46	NTc
1g	4-OMe		155	101	>10000	17	1.3	10.5	1.9	NTc
1h	5-OMe		231	86	>10000	5	0.9	2.8	58	NTc
1i	2-Me, 4-OMe		255	106	>10000	2	1.7	10.1	5.1	NTc
1j	2-OMe, 4-Me		436	131	>10000	12	1.4	10.4	39	NTc
1k	4-OMe, 5-Me		315	125	NDe	61	2.0	21.7	11	100/82/98
1l	2-Me, 4-Me		273	36	>10000	–2	1.3	4.2	46	100/100/100
2a	3-Me	CH	5947	44	>10000	3	0.9	<2.0	87	100/NTc/NTc
2b	3-Et	CH	509	99	>10000	3	1.3	4.2	71	88/NTc/NTc
2c	3-OMe	CH	188	110	>10000	9	1.3	6.8	38	98/100/100
2d	3-OEt	CH	182	83	>10000	4	1.8	12.6	21	94/NTc/NTc
2e	3-Cl	CH	5947	83	>10000	7	1.1	4.9	6.2	NTc
2f	3-Me, 5-Me	CH	2745	99	>10000	7	1.0	3.0	45	NTc
2g	3-OMe, 5-Me	CH	100	112	>10000	21	1.7	12.2	6.5	96/96/85
2h	3-OMe, 5-Me	N	166	74	>10000	20	1.2	8.5	18	91/98/100

^aPPAR activity was assessed in COS-7 cells transfected with a chimeric human PPAR-Gal4 receptor expression plasmid and a pG5luc reporter plasmid.

^bHuman/rat/monkey microsomal stability.

^cNot tested.

^dValues represented as mean ± SEM. Values on two independent experiments.

^eNot determined.

We then focused on 4-pyridines (compounds 2 in Table 1). 3-Methyl derivative 2a lost in vitro potency, whereas 3-ethyl analog 2b showed 10-fold higher potency. Further enhancement of potency was achieved by the introduction of an alkoxy group. 3-Methoxy and 3-ethoxy derivatives (2c and 2d) exhibited EC₅₀ = 188 and 182 nM, respectively. The 3-alkoxy group is essential for potency of 4-pyridine series because 3-chloro and 3,5-dimethyl analog (2e and 2f) lost their high potency in vitro. Although 3-methoxy-5-methyl derivative 2g showed the best potency among the 4-pyridine series, full agonist activity was observed. The modification of benzimidazole ring of 2g led to the discovery of intermediate agonist 2h (E_max = 74%). Imidazopyridine 2h exhibited adequate solubility and high microsomal stabilities against three species, with high PPARα selectivity.

As described, the introduction of substituents on 3- or 4-pyridine ring enhanced in vitro potency to discover partial agonist 1l and intermediate agonist 2h. As the enhancement of in vitro potency is often associated with full agonist activity, other modifications were required.

X-ray crystal structure of PPARγ LBD complexed with 1l was investigated to seek other chemical modifications. The binding mode of 1l superimposed with DS-6930 is shown in Figure 2a. As expected, 1l displayed the same binding mode as DS-6930.⁶ As no direct interaction with Tyr473 on helix 12 and the lipophilic interactions of dimethylpyridyl group were confirmed, these interactions relates for 1l to show partial agonist activity. A space around the 4′-positoin of benzimidazole ring surrounded by Phe264-Gly284-Arg288 was observed (Figure 2b). The existence of such a space urged us to introduce 4′-substituents on benzimidazole ring as shown in Table 2 (compounds 3).

Figure 2.

X-ray crystal structures of 1l (PDB 6IZM) and 3g (PDB 6IZN) bound to PPARγ-LBD. (a) Details of the binding mode of 1l (blue) superimposed with DS-6930 (magenta). Hydrogen bonds are marked as red dotted lines. (b) Compound 1l in PPARγ-LBD. PPARγ-LBD is shown in surface representation, and residues F264, G284, R288, and 1l are shown in stick presentation. (c) Details of the binding structure of 3g to PPARγ-LBD. Hydrogen bonds are marked as red dotted lines.

Table 2. PPAR Transcriptional Activities and in Vitro ADME Profiles of 3, 4, and 5.

cmpd	X	PPARγ EC50 (nM)a	PPARγ Emax (%)a	PPARα EC50 (nM)a	PPARα Emax (%)a	Log D	PAMPA Papp (10–6 cm/s)	solubility (μg/mL)	MS (%, h/r/mon)b
3a	4′-Me	811	116	>10000	4	1.7	6.0	3.2	100/NTc/NTc
3b	4′-OMe	249	99	>10000	1	1.2	2.7	50	NTc
3c	4′-OEt	104	103	>10000	5	1.6	4.8	130	100/NTc/NTc
3d	4′-OPr	32	94	>10000	1	2.2	11.8	34	NTc
3e	4′-Oi-Pr	31	103	>10000	4	1.9	9.1	630	91/65/81
3f	7′-Me	244	92	>10000	10	1.8	7.0	14	NTc
3g	7′-F	246	76	>10000	3	1.6	9.2	28	100/100/100
4a	4′-OEt	58	84	>10000	2	1.7	18.6	12	100/NTc/NTc
4b	4′-Oi-Pr	57	88	>10000	9	2.0	24.5	7.6	96/13/84
5a	4′-Me	57	83	>10000	13	1.7	12.6	36	100/NTc/NTc
5b	4′-OEt	58	71	>10000	5	1.7	15.5	120	100/NTc/NTc
5c	4′-Oi-Pr	32	100	>10000	14	2.0	17.3	53	97/91/94
5d	7′-F	60	95	>10000	18	1.7	14.0	2.1	97/NTc/NTc

^aPPAR activity was assessed in COS-7 cells transfected with a chimeric human PPAR-Gal4 receptor expression plasmid and a pG5luc reporter plasmid.

^bHuman/rat/monkey microsomal stability.

^cNot tested.

The introduction of 4′-methyl group in the bicyclic core structure led to disappointing results (compound 3a), whereas 4′-methoxy derivative 3b retained potency in vitro. Consequently, several 4′-alkoxy groups were incorporated. The in vitro potency was found to be correlated with the lipophilicity of molecules (3b–3e). For example, 4′-methoxy derivative 3b had EC₅₀ = 249 nM, whereas 4′-propoxy analog 3d exhibited EC₅₀ = 32 nM. This modification enhanced the activity by 7.8-fold, with an increase in Log D value by 1.0. Because the membrane permeability progressively increased, the improvement of membrane permeability might contribute to the high potency in vitro. This enhancement of PPARγ potency was not accompanied by the PPARα activation. Isopropyl analog 3e exhibited high potency in vitro with enhanced aqueous solubility compared with that of 1l. The improvement in solubility achieved was over 10-fold, even though 3e was more lipophilic than 1l. Compound 3e maintained robust stability against human liver microsomes. However, 3e showed full agonist activity. In fact, all 4′-substituted analogs 3a–3e showed full agonist activity (E_max = 94–116%).

Accordingly, we sought another solution. Several small substituents were incorporated at 7′-position on benzimidazole ring due to the limited space around this position (Figure 2b). A methyl or fluoro group was introduced into this position, and both analogs retained in vitro potency. In particular, 7′-fluoro analog 3g demonstrated intermediate agonist activity (E_max = 76%), with acceptable solubility and robust microsomal stability. The binding mode of 3g is shown in Figure 2c. 7′-Fluoro analog 3g exhibited the same binding mode as 1l.

The introduction of substituents on the benzimidazole ring in III was then explored (compounds 4 in Table 2). 4-Pyridine 2c was also modified (compounds 5 in Table 2). When 4′-alkoxy group was introduced in III, 4′-ethoxy and 4′-isopropoxy derivatives 4a and 4b retained high potency in vitro. However, the introduction of such substituents resulted in full agonist efficacy. The enhancement of in vitro potency was also achieved for 4-pyridines. When 4′-methyl, 4′-ethoxy, or 4′-isopropoxy group was introduced into benzimidazole ring, all compounds exerted high potency in vitro (compounds 5a, 5b, and 5c). In particular, ethoxy analog 5b demonstrated strong potency in vitro with an intermediate agonist activity (EC₅₀ = 58 nM, E_max = 71%). Compound 5b also exhibited improved aqueous solubility. The introduction of 7′-fluoro group resulted in full agonist 5d.

Several potent compounds were advanced to in vivo pharmacological profiling. These compounds were assessed for their ability to reduce PG in Zucker diabetic fatty (ZDF) rats at 3 mg/kg p.o. for 14 days as shown in Figure 3a. Plasma exposures of the test compounds (AUC) are shown in Figure 3b. DS-6930 exerted a statistically significant reduction in PG (57%, p < 0.05). 3-Pyridine 1l exhibited poor efficacy in vivo owing to low potency in vitro, whereas better results were observed with full agonist 2c (37% PG reduction, p < 0.05). Further enhancement of the efficacy was achieved in 4-pyridines 2g and 2h with statistical significance (p < 0.01). Compound 2g exerted 61% reduction of PG, whereas 66% PG reduction was observed in imidazopyridine 2h. Note that 2g is a full agonist. Owing to the excellent plasma exposure, 7′-F substituted analog 3g exhibited excellent efficacy (73% PG reduction, p < 0.01). Of 2-pyridines, 4′-ethoxy derivative 4a showed modest efficacy, whereas 4′-isopropoxy analog 4b lost the potent efficacy in vivo. As both compounds showed the same range of strong potency in vitro, the difference in the plasma exposure of the compounds resulted in such a pharmacological gap. 4′-Alkoxy substituted 4-pyridines, 5a and 5b, lost potent efficacy in vivo. The reasons for this poor efficacy are unclear because 5a showed sufficient plasma exposure with high potency in vitro. Overall, 4′-alkoxy substituted benzimidazoles tended to show lower efficacy in vivo. 4′-Alkoxy analog 3e resulted in lower PG reduction than 7′-F analog 3g, owing to the lower plasma exposure of the compound. Unsubstituted benzimidazole III exerted higher efficacy than 4′-alkoxy analogs 4a and 4b, and the same relationship was observed between unsubstituted analog 2c and 4′-alkoxy derivatives (5a and 5b).

Figure 3.

Pharmacological effects and PK parameters of PPARγ modulators. (a) PG reduction (% change in PG level vs vehicle control) in ZDF rats after the oral administration of 3 mg/kg of the test compounds in 0.5% methylcellulose on day 14 (n = 5). Data are represented as mean ± SEM. Statistical significance compared to vehicle treatment is denoted by *p < 0.05 and **p < 0.01. (b) AUC_0–24 h (h·μg/mL) was acquired by the administration of compounds to ZDF rats on day 15 (n = 5). Each value is the mean ± SD.

Based on pharmacological results combined with ADME profile, 2h and 3g were selected as candidates for further evaluations. PK study of these compounds was performed by oral administration to male cynomolgus monkeys at 3 mg/kg as shown in Table 3. The compounds (1 mg/kg) were intravenously administered to the same monkey to calculate the total body clearance (CL), distribution volume at steady state (V_ss), and F value. 4-Pyridine 2h retained excellent CL comparable to III. Further improvement of CL was achieved in 7′-fluoro derivative 3g. 7′-Fluoro analog 3g demonstrated excellent PK parameters, including the lowest CL as well as the highest AUC and F even compared to III. The PK parameters of 3g in rodents at the same dose (3 mg/kg p.o., 1 mg/kg i.v.) are summarized (Table 3). 7′-Fluoro analog 3g showed lower AUC and higher CL in rodents than in monkeys. Although 3g exerted the highest CL and the lowest bioavailability in mice, these PK parameters were acceptable for clinical candidate selection.

Table 3. Pharmacokinetic Parameters and Bioavailability of Compounds III, 2h, and 3g at 3 mg/kg^a.

cmpd	species	Cmax (μg/mL)	Tmax (h)	T1/2 (h)	AUClast (h·μg/mL)	F (%)	CL (mL/min/kg)	Vss (L/kg)
III	monkey	2.25 ± 0.72b	5.00 ± 4.2b	13.5 ± 0.42b	23.5 ± 5.9b	89 ± 15b	2.06 ± 0.21b	0.36 ± 0.0071b
2h	monkey	0.50 ± 0.37b	4.50 ± 4.9b	7.30 ± 2.8b	5.12 ± 2.4b	18 ± 8.7b	2.08 ± 1.1b	0.35 ± 0.049b
3g	monkey	28.8 ± 10b	0.50 ± 0.0b	9.34 ± 3.3b	128 ± 73b	76 ± 34b	0.40 ± 0.23b	0.15 ± 0.042b
3g	rat	3.49 ± 0.93c	2.25 ± 1.75c	3.45 ± 0.70c	28.0 ± 6.65c	74 ± 18c	1.29 ± 0.18b	0.27 ± 0.035b
3g	mouse	1.29 ± 0.43c	0.50 ± 0.0c	4.35 ± 2.49c	2.07 ± 0.26c	21 ± 2.6c	4.91 ± 0.66b	0.62 ± 0.22b

^aThe test compounds in 0.5% methylcellulose were administered to male cynomolgus monkeys at 3 mg/kg (p.o.). Total body clearance (CL), distribution volume at steady state (V_ss), and F value were calculated after intravenous administration of the test compounds (1 mg/kg). Each value represents the mean ± SD.

^bn = 2.

^cn = 3.

In summary, the 2-pyridine ring of DS-6930 was replaced with a 3- or 4-pyridyl group to identify backup clinical candidates for DS-6930. Although the introduction of substituents on 3- or 4-pyridine ring resulted in improvements of the in vitro potency, potent agonists showed full agonist activity. The introductions of 4′-substituent on benzimidazole ring also enhanced potency in vitro, while these compounds exhibited modest in vivo efficacy in ZDF rats. Most 4′-substituted compounds including 3e exerted full PPARγ agonist activity. Other modifications of the benzimidazole ring led to the discovery of 7′-fluoro analog 3g and imidazopyridine 2h, which demonstrated potent PG reduction in vivo with an intermediate agonist activity in vitro. Among them, 7′-fluoro analog 3g (DS19161384) exhibited excellent DMPK profile.

Glossary

ABBREVIATIONS

PPAR

peroxisome proliferator-activated receptor

TZD

thiazolidinedione

Cdk5

cyclin-dependent kinase 5

PG

plasma glucose

LBD

ligand binding domain

SAR

structure–activity relationship

PAMPA

parallel artificial membrane permeation assay

ZDF rat

Zucker diabetic fatty rat

Supporting Information Available

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acsmedchemlett.8b00645.

• General synthetic procedure of all compounds; experimental procedures and characterization data of selected compounds; procedures for pharmacological activities (PDF)

The authors declare no competing financial interest.