Synthesis and biological evaluation of novel thienopyrimidine derivatives as diacylglycerol acyltransferase 1 (DGAT-1) inhibitors

Abstract A novel series of thieno[3,2-d]pyrimidine derivatives were synthesised and their inhibitory effects against diacylglycerol acyltransferase 1 (DGAT-1) were assessed. cis-Isomer 17a showed potent and selective inhibitory activity against DGAT-1 in SF9 cells. In addition, 17a had an acceptable pharmacokinetic profile and accumulated mainly in the small intestine and liver. Oral administration of 17a led to a significant reduction in plasma triacylglycerol level during an oral lipid tolerance test (OLTT) in murine and canine models. Taken together, 17a is a high-quality candidate that deserves further investigation.


Introduction
Obesity is characterised by abnormal fat accumulation due to a systemic dysfunction of energy homeostasis. It is a significant risk factor for diabetes, hypertension, cardiovascular disease, and nonalcoholic fatty liver disease 1 . Excessive accumulation of triglycerides (or triacylglycerol, TAG) in adipocytes and non-adipocytes, including in the liver, skeletal muscle, myocardium, and pancreas, is a crucial feature of obesity 2 . In particular, abnormal TAG levels increases the risk of metabolic syndrome, characterised by as insulin resistance, dyslipidemia and cardiomyopathy 3 .
Therefore, preventing excessive accumulation of TAG could be beneficial in the treatment of metabolic diseases such as obesity and type 2 diabetes. Two routes of TAG biosynthesis have been identified: the glycerol phosphate pathway, and the monoacylglycerol pathway 4 . Both routes have a common intermediate, diacylglycerol (DAG), which is then converted to TAG by acyl CoA: diacylglycerol acyltransferase (DGAT) 5 . DGAT exists as two isoforms, DGAT-1 and DGAT-2, which share minimal sequence homology 6 . These enzymes catalyse the final dedicated step in TAG synthesis: i.e. the esterification of a fatty acid moiety to DAG to generate TAG. DGAT-1 is highly expressed in the small intestine, adipose tissue, and liver 6 . DGAT-1 deficient mice showed a significant reduction in the postprandial increase of plasma TAG and were resistant to diet-induced obesity due to increased energy expenditure. Moreover, the DGAT1 knockout mice had enhanced sensitivity to insulin and leptin. Indeed, DGAT-1 inhibitors had significant pharmacologic effects, including decreased TAG, which were consistent with the DGAT1 knockout mice phenotypes 7 . However, DGAT2 knockout mice had extremely low levels of TAG, and their skin could not protect against moisture, so these mice died after birth 8 . Therefore, selective DGAT-1 inhibitors have been developed to manage metabolic diseases such as obesity and type 2 diabetes [9][10][11] . Pharmaceutical companies, including AstraZeneca (1) 12 , Novartis (2) 13 , Pfizer (3) 14 , and Abbott (4) 15 , and Hoffmann-La Roche (5) 16 are developing selective DGAT-1 inhibitors, whose structures are described in Figure 1. Other novel DGAT1 inhibitors have also been reported [17][18][19] .
Herein, we described the discovery of novel thieno [3,2-d]pyrimidine derivatives as DGAT-1 inhibitors. Target molecules were designed based on a bioisosteric replacement strategy ( Figure 2). Introduction of a heterobicylcle instead of a phenyl linker was attempted. Thienopyrimidines showed various pharmacologic activities as a privileged scaffold 20 , and the introduction of thieno [3,2-d]pyrimidine was the first approach to designing novel inhibitors based on the structural features of reported DGAT-1 inhibitors.

Chemistry
All commercial chemicals and solvents were reagent grade and were used without further purification. Completion of the reactions was monitored by analytical thin layer chromatography (TLC) using precoated glass-backed plates (E-Merck, silica gel 60 F 254 0.25 mm). For normal pressure and flash column chromatography purifications, Merck silica gel 60 (size 70-230 and 230-400 mesh, respectively) was used. 1 H and 13C NMR spectra were recorded with an Avance DPX-300 NMR spectrometer (Bruker, Germany) and Jeol 600 MHz spectrometer (Jeol Resonance ECZ 600 R, USA). All the 1 H and 13 C NMR spectra were recorded in deuterated chloroform (CDCl 3 ) with tetramethylsilane (TMS) as an internal standard or deuterated dimethyl sulfoxide (DMSO)-d6 as solvents; chemical shifts are reported in d values (ppm) relative to the residual solvent peak. Multiplicity was indicated as follows: s (singlet); d (doublet); t (triplet); m (multiplet); dd (doublet of doublet); brs (broad singlet). Mass spectra were obtained on Waters Aquity UPLC/QTOF (Waters Corporation, USA). HPLC was used Agilent, 1200 series using capcellpak MGII (4.6 Â 150 mm, 5 lm) eluted with a 30 min gradient from 20-70% acetonitrile in water.

DGAT-1 inhibition assay (IC 50 )
The activity of DGAT-1 inhibitors in vitro was evaluated by using a human recombinant DGAT1 enzyme expressed in insect cells (SF9 cells). SF9 cells were homogenised by washing them with DPBS (Dulbecco's phosphate-buffered saline) and then suspending cell pellets with a tris buffer (250 mM sucrose; 10 mM Tris-HCl [pH 7.4]; proteinase inhibitor). The resulting mixture was centrifugally separated at 10,000 Â g for 30 min to remove the cell debris remaining in the lower layer thereof, and was centrifugally separated at 100,000 Â g for 60 min to obtain a microsomal membrane. Further, membrane fractions were resuspended by the tris buffer, and then stored at -80 C.
The activity of DGAT1 was measured according to the reported method 20 . Specifically, 0.0001 -10 lM (final concentration, FC) of the test compounds were cultured at room temperature (25 C) for 15 min with a 10 lM of SF9 microsomal protein solution and 100 mM of MgCl 2 solution, and were then further cultured at room temperature (25 C) for 30 min after the addition of 100 lM (FC in 12.5% EtOH) of 1,2-dioleyl-sn-glycerol and 30 lM (final conc.) of 14 C-oleyl coenzyme A. The reaction was completed by the addition of 300 lL of a mixed solution of 2-propanol/heptane (7:1), and radioactive triglyceride was separated from an organic solvent layer by using 200 lL of heptane and 200 lL of a 0.1 M carbonate buffer (pH 9.5). The amount of triglyceride was measured by liquid scintillography (Perkin Elmer, USA) after mixing with an organic solvent and an equivalent amount of scintillation solvent (MicroScint-O, Perkin Elmer, USA). The effects of inhibiting DGAT1 were calculated as percent in comparison with a control material.

Statistical analysis
Statistical analysis was performed using one-way ANOVA, followed by Dunnett's multiple comparison test using Prism 6.0 (GraphPad, USA). The Kruskal-Wallis test was applied when data did not pass the normality test.

Results and discussion
We found that thieno [3,2-d]pyrimidine could be a core scaffold connecting an aryl moiety at one end to a cyclo-aliphatic carboxylic acid group at the other. Synthetic schemes for a series of thieno [3,2-d]pyrimidine derivatives are outlined in Schemes 1 and 2.
Desired compounds 14a-14l were synthesised from commercially available 7-bromo-4-chlorothieno [3,2-d]pyrimidine (6), as shown in Scheme 1. Consecutive Suzuki coupling reactions gave 8 and 10. Compound 9 was prepared using the reported method 19 . Hydrogenation of 10, followed by de-protection of the Boc group with trifluoroacetic acid afforded the key intermediate 12. The aniline 12 was converted to amides 14a-14h or ureas 14i-14l by coupling with acid chlorides or aryl isocyanates, respectively, followed by hydrolysis with sodium hydroxide or lithium hydroxide.
The activities of synthesised compounds were evaluated using SF9 insect cells expressing human DGAT-1 22 . The IC 50 values are summarised in Table 1. A potent DGAT-1 inhibitor, LCQ-908 (2 or pradigastat) was used as a positive control. To optimise the alkyl chain length of the acid moiety, compound 14b (n ¼ 1) was more potent than short chain 14a (n ¼ 0) and long chain 14c (n ¼ 2). The calculated lowest energy conformation model of 14c provided that the distance between the terminal carboxylic acid and NH of urea was over 16 Å, which may result in reduced activity. 14a exhibited a loss of activity, which suggested that the terminal carboxylic acid should be placed one-carbon away from cyclohexane. Next, structure-activity relationships for the aryl moiety on the left side were investigated, with derivatives bearing chain lengths the same as that of 14b. 5-chloro-2-nitro phenyl (14d), 2-trifluoromethylpyridinyl (14f), 4-chloropyridinyl (14k), and 5-bromopyridinyl (14l) derivatives had no activity (IC 50 >1 lM), while 14b, 14c, 14e, 14h, and 14j had good activity (IC 50 values of 0.3 lM-0.4 lM). Compounds 14g and 14i were equipotent to the known compound 2. The most potent compound, 14i, was subjected to further investigation. Cis and trans isomers of 14i were separated to determine the activity and properties of each isomer. As shown in Scheme 2, the racemic mixture 13 was conveniently separated into cis-isomer 15a and trans-isomers 15b by the treatment with ethyl acetate. 15a and 15b exhibited greatly different solubility for ethyl acetate, and were soluble and insoluble for ethyl acetate, respectively. Hydrolysis of 15a and 15b afforded cis isomer 17a and trans isomer 17b, respectively.
Interestingly, the isomers 17a and 17b were more potent than mixture 14i, and even compound 2, in SF9 cells (Table 2).
Considering potency, selectivity, and pharmacokinetic profiles, 17a was investigated for further pharmacokinetic properties in four species, and results are summarised in Tables 4 and 5. 17a exhibited low to moderate clearance and acceptable oral bioavailability in rat and dog.
17a exhibited durable metabolic stabilities, as shown in Table 5. In particular, 17a was metabolically stable in four species: human, dog, rat, and mouse. Further evaluation of 17a was carried out for inhibition of clinically relevant cytochrome 450 (CYP) isoforms (3A4, 1A2, 2C9, 2C19, 2D6). No significant inhibition by 17a was observed (Table 5). Thus, 17a would be very unlikely to affect the pharmacokinetics of other drugs.
To determine in vivo efficacy, we evaluated the activity of 17a against serum TAG level induced by the oral administration of corn oil (oral lipid tolerance test, OLTT; Figures 3 and 4), since intestinal targeted DGAT-1 inhibition results in a reduced serum TAG level 14 . 17a was orally administered to ICR mice at 3 mg/kg or to dogs at 1 mg/kg before administration of corn oil. As shown in Figure 3, 17a showed significant activity, as potent as compound 2, in reducing plasma TAG level.
In addition, tissue distribution of 17a was investigated. 17a was orally administered to mice at 30 mg/kg, and 2 was administrated as a positive control ( Figure 5). The concentration levels of 17a in the liver and small intestine were much higher than in other tissues, indicating that liver and small intestine are the main target organs of 17a, while 2 showed much high concentrations in the liver and large intestine. Analysis of tissue-to-plasma ratios indicated that 17a had favourable distributions in the liver and small intestine. Significant activity of 17a in the OLTT appeared to result from high concentration in the small intestine, even though bioavailability of 17a was not great in mice. This result could be proof of a pharmacokinetic-pharmacodynamic correlation.

Conclusion
We found that a novel series of thieno[3,2-d]pyrimidine derivatives had potent DGAT-1 inhibitory activity. A representative compound, cis-isomer 17a showed potent DGAT-1 inhibitory activity, metabolic stability in four species, an acceptable pharmacokinetic profile in rodents and dogs, and a significant reduction of TAG in the OLTT in mice and dogs. In a tissue-distribution study, 17a was mainly distributed to the liver and small intestine. Therefore, 17a Serum TG (or TAG) concentration was measured before and 1, 2, and 4 h after oral administration of corn oil in ICR mice treated with 17a (3 mg/kg) or 2 (1 mg/ kg). Vehicle for 2: 0.5% methylcellulose/0.5% Tween80 in distilled water, vehicle for 17a: 0.5% Tween80 in distilled water. . Effect of 17a in a canine oral lipid tolerance test (n ¼ 2, Ã p < .05; ÃÃÃ p < .001). The dog was treated with 17a (po. 1 mg/kg) or 2 (po. 1 mg/kg) before oral administration of corn oil. Vehicle for 2: 0.5% methylcellulose/0.5% Tween80 in distilled water, vehicle for 17a: 0.5% Tween80 in distilled water. could be a candidate with considerable potential and deserves for further investigation.

Disclosure statement
No potential conflict of interest was reported by the authors.

Funding
This research was supported by the Advanced Research Centre Programme (2015R1A5 A1008958) funded by the Korean Government.