Discovery of an Orally Bioavailable Benzimidazole Diacylglycerol

May 12, 2017 - †Global Discovery Chemistry, ‡Cardiovascular and Metabolism, ∥PK Sciences, and §Translational Medicine, Novartis Institutes for ...
18 downloads 11 Views 2MB Size
Article pubs.acs.org/jmc

Discovery of an Orally Bioavailable Benzimidazole Diacylglycerol Acyltransferase 1 (DGAT1) Inhibitor That Suppresses Body Weight Gain in Diet-Induced Obese Dogs and Postprandial Triglycerides in Humans Katsumasa Nakajima,*,† Ricardo Chatelain,‡ Kevin B. Clairmont,‡ Renee Commerford,‡ Gary M. Coppola,† Thomas Daniels,‡ Cornelia J. Forster,† Thomas A. Gilmore,† Yongjin Gong,† Monish Jain,∥ Aaron Kanter,† Youngshin Kwak,† Jingzhou Li,† Charles D. Meyers,§ Alan D. Neubert,† Paul Szklennik,† Vivienne Tedesco,‡ James Thompson,‡ David Truong,‡ Qing Yang,‡ Brian K. Hubbard,‡ and Michael H. Serrano-Wu† †

Global Discovery Chemistry, ‡Cardiovascular and Metabolism, ∥PK Sciences, and §Translational Medicine, Novartis Institutes for Biomedical Research, 100 Technology Square, Cambridge, Massachusetts 02139, United States S Supporting Information *

ABSTRACT: Modification of a gut restricted class of benzimidazole DGAT1 inhibitor 1 led to 9 with good oral bioavailability. The key structural changes to 1 include bioisosteric replacement of the amide with oxadiazole and α,α-dimethylation of the carboxylic acid, improving DGAT1 potency and gut permeability. Since DGAT1 is expressed in the small intestine, both 1 and 9 can suppress postprandial triglycerides during acute oral lipid challenges in rats and dogs. Interestingly, only 9 was found to be effective in suppressing body weight gain relative to control in a diet-induced obese dog model, suggesting the importance of systemic inhibition of DGAT1 for body weight control. 9 has advanced to clinical investigation and successfully suppressed postprandial triglycerides during an acute meal challenge in humans.



INTRODUCTION Excess accumulation of triglycerides in the body may result in obesity or contribute to the pathogenesis of cardiovascular disease1 and pancreatitis.2 Pharmacological inhibition of triglyceride synthesis has great therapeutic potential toward treatment of these diseases. Of several enzymes involved in triglyceride synthesis, two diacylglycerol acyltransferases, DGAT13 and DGAT2,4 catalyze the final committed step. While DGAT2 is essential for survival in mice,5 DGAT1 knockout mice have an attractive metabolic phenotype, being healthy, resistant to diet-induced obesity and with improved insulin sensitivity relative to wild-type.6 DGAT1 has emerged as an attractive drug discovery target for a number of research groups.7−9 While DGAT1 is expressed ubiquitously, it is most abundant in the enterocytes of the small intestine where it is involved in resynthesizing triglycerides from fatty acids and monoacylglycerol, the hydrolysis products of dietary triglycerides present in the gastrointestinal tract.3,10 Consistent with this expression pattern, a recent genetic study11 in mice suggests that loss of © 2017 American Chemical Society

intestinal DGAT1 is responsible for the phenotype of DGAT1 whole-body knockout mice. Toward recapitulating this phenotype pharmacologically, we have reported benzimidazole inhibitor 1 that selectively targets intestinal DGAT1 with low oral bioavailability (Figure 1).12

Figure 1. Benzimidazole DGAT1 inhibitor 1. Received: February 17, 2017 Published: May 12, 2017 4657

DOI: 10.1021/acs.jmedchem.7b00173 J. Med. Chem. 2017, 60, 4657−4664

Journal of Medicinal Chemistry

Article

Table 1. DGAT1 IC50 Data of Compounds 2−5a

Despite exposure of 1 being restricted to the small intestine after oral administration, postprandial triglycerides were significantly suppressed compared to control during an oral lipid challenge in rats. A robust suppression of postprandial triglycerides was similarly observed in dogs with very low plasma exposure of 1 after oral administration at 1 mg/kg.12 1 was then dosed orally to diet-induced obese dogs at 5 mg/kg q.d. for 21 days. Unexpectedly, both treatment and control groups (n = 6 per group) showed statistically similar body weight gains over baseline. Systemic exposure of 1 on the last day of the study was measured to be low (AUC0−24h = 70 nM· h), suggesting that, as expected, DGAT1 inhibition by 1 was restricted to the intestine. In a separate study using a mouse model of diet-induced obesity, the group treated with 1 did show reduced body weight gain compared to control (P ≤ 0.05) after oral administration at 30 mg/kg q.d. for 14 days. However, the higher dose led to significant plasma exposure of 1 (Cmax = 350 nM, Cmin = 71 nM, AUC0−24h = 3081 nM·h), so we could not conclude that the body weight reduction was based solely on inhibition of intestinal DGAT1. These observations raise the possibility that systemic exposure is required for a DGAT1 inhibitor to be efficacious against dietinduced body weight gain, suggesting a need for a potent DGAT1 inhibitor from the benzimidazole series which has good oral bioavailability. Here we report details of the medicinal chemistry effort that led to the discovery of such an inhibitor and its effectiveness in a chronic body weight study in dogs as well as in an acute meal challenge study in humans.

compound DGAT1 IC50 (nM)

2

3

4

5

280 (1)

132 ± 2 (2)

47 ± 12 (4)

23 ± 5 (2)

Average values ± SD. The number of independent experiments is shown in parentheses. Each experiment was run in duplicate.

a

Table 2. In Vitro Data of Compounds 6 and 7a



compound

RESULTS AND DISCUSSION Compound 1 may have a suboptimal membrane permeability based on a 10-fold decrease in potency from DGAT1 measured biochemically (IC50 = 39 nM) to inhibition of triglyceride synthesis in a C2C12 mouse myoblast cellular assay (IC50 = 390 nM) as well as a low A−B value (0.7 × 10−6 cm/s) in the Caco-2 permeability assay. (B−A)/(A−B) ratio (12.9) in the Caco-2 assay also suggests 1 is an efflux substrate in the gut, resulting in low bioavailability after oral administration. To improve membrane permeability and reduce efflux, our optimization strategy focused on replacing the amide and the carboxylic acid. Oxazole and oxadiazole are widely utilized bioisosteres of amides in drug design.13,14 In particular, they lack a N−H hydrogen, which can lead to reduced solvation and improved membrane permeability.15 The effect of these replacements on DGAT1 inhibition was first examined in benzimidazoles devoid of a propionic acid (2−5)16,17 (Table 1), showing that both oxazole (3, IC50 = 132 nM) and oxadiazole (4, IC50 = 47 nM) were in fact more potent than the corresponding amide 2 (IC50 = 280 nM). When the 4substituent on the western phenyl ring was changed from a chloride to a methoxy group, a slightly more potent analog 5 (IC50 = 23 nM) was obtained. The impact of oxadiazole on membrane permeability was investigated later when the propionic acid was attached on the eastern phenyl ring. In a related subseries with a quinolineamide side chain, α,αdimethylation of the propionic acid was found to be well tolerated (DGAT1 IC50: 6, 14 nM; 7, 32 nM)16 (Table 2). Both 6 and 7 showed similar IC50 values in C2C12 cells (6, 913 nM; 7, 954 nM), so improvement of membrane permeability of 7 over 6 was not evident in C2C12 cells. However, increase of A−B values from 6 (1.4 × 10−6 cm/s) to 7 (3.5 × 10−6 cm/s) as well as reduced efflux ratio from 6 to 7 ((B−A)/(A−B): 6, 10.0; 7, 3.5) in the Caco-2 assay suggested improved gut

DGAT1 IC50 (nM) C2C12 cells IC50 (nM) Caco-2: A−B (×10−6 cm/s) Caco-2: B−A (×10−6 cm/s) (B−A)/(A−B) ratio

6

7

14 ± 9 (752) 913 (1) 1.4 ± 0.3 (4) 14.1 ± 2 (4) 10.0 ± 2 (4)

32 ± 4 (2) 954 (1) 3.5 (1) 12.4 (1) 3.5 (1)

Average values ± SD. The number of independent experiments is shown in parentheses. Each experiment was run in duplicate for DGAT1 IC50 and quadruplicate for C2C12 cells IC50.

a

permeability of 7 over 6. These Caco-2 data nicely predicted increased oral bioavailability (6, 2%; 7, 23%) in a rat pharmacokinetic (PK) study (Table 3). Our next molecular design combined the oxadiazole and α,αdimethylated propionic acid groups in one molecule (9) (Table 4). The corresponding α-unsubstituted propionic acid analog 8 was evaluated to confirm the impact of α,α-dimethylation of the carboxylic acid on oral bioavailability in the oxadiazole series. Good DGAT1 potency was obtained with both of the compounds (IC50: 8, 10 nM; 9, 17 nM). Potency for inhibition of triglyceride synthesis in C2C12 cells by 8 and 9 (IC50 of 181 and 151 nM, respectively) was also improved relative to 1 (IC50 = 390 nM). Potency shifts between the two assays for 8 (18fold) and 9 (9-fold) were, however, still comparably high to 1 (10-fold), implying that neither oxadiazole ring (8, 9) nor α,αdimethylation of carboxylic acid (9) contributed to improved membrane permeability in C2C12 cells. However, improvement of gut permeability from 8 to 9 was suggested based on increase of A−B values from 8 (1.4 × 10−6 cm/s) to 9 (4.2 × 10−6 cm/s) as well as reduced efflux ratio ((B−A)/(A−B): 8, 6.6; 9, 2.4) in the Caco-2 assay. This improved in vitro data for 9 over 8 were again reflected in a rat PK study, where 9 had improved oral bioavailability (90%) over 8 (2%) (Table 5). The extent of the improvement was more dramatic than that of the 4658

DOI: 10.1021/acs.jmedchem.7b00173 J. Med. Chem. 2017, 60, 4657−4664

Journal of Medicinal Chemistry

Article

Table 3. Rat PK Data of Compounds 6 and 7a

compound b

iv dose (mg/kg) CL (mL min−1 kg−1) Vss (L/kg) T1/2 (h) AUC0−8h (nM·h) po dose (mg/kg)b Cmax (nM) Tmax (h) AUC0−8h (nM·h) F (%)

6

7

1 6.9 ± 1 0.4 ± 0 1.7 ± 0.4 5232 ± 706 5 333 ± 169 0.4 ± 0.1 405 ± 250 2±1

1 5.5 ± 0.7 0.4 ± 0.07 0.8 ± 0.07 6111 ± 676 3 1717 ± 283 0.5 ± 0 4234 ± 1423 23 ± 8

Average values ± SD of two rats (iv dose) or three rats (po dose). bSolution administration, vehicle: 10% NMP, 30% PG, 6% ETPGS (for compound 6); 10% NMP, 40% PG, 5% ETPGS (for compound 7). a

Table 4. In Vitro Data of Compounds 8 and 9a

triglycerides robustly over the following 4 h, confirming successful inhibition of DGAT1 in vivo. In a subsequent chronic efficacy study, 9 was orally dosed to dogs at 5 mg/kg q.d. for 21 days (Figure 3). Upon high fat ad libitum feeding (free access to food and water), the group treated with 9 (n = 6 animals) showed an 11% decrease in body weight gain compared to the vehicle group (n = 6 animals) (P < 0.005). Plasma levels of 9 were measured at 90 min after the first and twentieth doses [738 ± 272 (n = 6) and 930 ± 38 nM (n = 3), respectively]. A full 24 h pharmacokinetic profile was obtained in a separate group of three dogs by the same dosing method, confirming high concentration of 9 in plasma (AUC0−24h = 5270 ± 842 nM·h). Compared to the chronic study of 1 described earlier, it is reasonable to speculate that it is the significant plasma exposure of 9 which contributed to decreased body weight gain in the dog model. However, further investigation is required to understand the relationship of in vitro potency to in vivo efficacy since free concentration of 9 in plasma is low (plasma protein binding 99.1 ± 0.7% in the dog) with unbound Cave of 9 as 2 nM, which is below in vitro potency value of 9 in C2C12 cellular assay (IC50 = 151 nM). Gastrointestinal adverse events such as diarrhea and steatorrhea were not observed in the treatment group with 9. During selectivity and in vitro safety profiling, 9 was found to be a poor inhibitor of DGAT2 in vitro (IC50 > 10 μM). It also showed a low risk of mutagenicity in the Ames assay19,20 as measured by increase of reverting mutations in histidinedepleted Salmonella typhimurium strains (TA98 and TA100) in the presence and absence of S9 metabolic activation at concentration of 9 up to 1000 μg/well. This finding was particularly satisfying as we had previously identified that a benzimidazole substituted with 2,6-dimethylphenyl group at C2, a partial structure present in 6, can serve as a toxicophore for such a risk via generation of a reactive metabolite of proposed

compound DGAT1 IC50 (nM) C2C12 cells IC50 (nM) Caco-2: A−B (×10−6 cm/s) Caco-2: B−A (×10−6 cm/s) (B−A)/(A−B) ratio

8

9

10 ± 4 (4) 181 (1) 1.4 (1) 9.2 (1) 6.6 (1)

17 ± 15 (42) 151 ± 31 (4) 4.2 (1) 10 (1) 2.4 (1)

Average values ± SD. The number of independent experiments is shown in parentheses. Each experiment was run in duplicate for DGAT1 IC50 and quadruplicate for C2C12 cells IC50.

a

carboxamide analogs 6 and 7 (2−23%), suggesting that 9 might be more soluble in the intestinal environment at the site of absorption than 7, though both compounds showed poor solubility in vitro (