1. Introduction

Farnesoid X receptor (FXR) [1, 2], belonging to the nuclear receptor superfamily, has emerged as a key player in the control of multiple metabolic pathways. It is rich in liver and some extrahepatic tissues such as the kidney and intestine, which were constantly exposed to high concentration of bile acids, the physiological ligands of FXR. It is also highly expressed other cholesterol-rich tissues such as adrenal glands [2, 3].

FXR binds to cis-acting elements in the promoters of series target genes, such as cholesterol 7α-hydroxylase (Cyp7A1), bile salt export pump (BSEP) and Na⁺-dependent taurocholate cotransporting polypeptide (NTCP), and modulate their expressions in response to bile acid metabolites [4]. Apart from its crucial role in maintaining bile acids and cholesterol homeostasis, FXR also regulates the lipid and glucose metabolism [5]. Previous studies demonstrated the regulation of FXR benefits to various diseases such as metabolic disorders, hepatitis, arteriosclerosis and cancer [6-8]. Thus, FXR has become an attractive therapeutic target.

Extensive studies on the discovery of FXR ligands in the past 1-2 decades, since the deorphanization of FXR, has resulted in the report of some steroidal-(natural ligand like) and non-steroidal FXR ligands [9], among which three compounds (obeticholic acid, Px-102 and LJN-452) entered clinic trials. In May 2016, obeticholic acid was approved by FDA for the treatment of primary biliary cholangitis (PBC) in combination with ursodeoxycholic acid (UDCA) in adults with an inadequate response to UDCA, or as a single therapy in adults unable to tolerate UDCA, validating the utility of FXR interacting agents in human [10].

Serious side effects were encountered during the application of FXR full agonists to animals and patients with diabetes and liver steatosis, such as, inhibition of bile acid synthesis and increased levels of low-density lipoprotein (LDL) [11, 12]. So FXR partial agonists and antagonists have been actively pursued as alternatives to full agonists, although their number and structural diversity are still limited.

Most of reported FXR antagonists to date are still steroidal molecules, including the first known antagonist guggulsterone (Fig. 1) [13, 14]. There are also a few synthetic non-steroidal scaffolds reported to date (Fig. 2), i.e., substituted-isoxazole derivatives (2a-2b) [15], 1, 3, 4-trisubstituted-pyrazolone (3) [16], T3 (4) [17], NDB (5) [18], hydroxyacetophenone derivatives (6a-6c) [19] and 1, 3-disubstituted-pyrazole-3-carboxamide (7a-7c) [20]. In addition to guggulsterone, a family of sesterterpene suvanine, were also the rare examples for natural products derived antagonists [21].

In the pursuit of new chemotype of FXR interacting agents, we have discovered chalcones 8a and 8b as moderate FXR antagonists (IC₅₀ at 10^-5 mol/L), by the screening of a nature product like library containing 9a-9l. However, those cellular active chalcones were not active in animal assays, possibly due to their instability in vivo. Then we tried to optimize their PD and PK properties by the conformational restriction of chalcones and found benzopyrenes as FXR antagonists both in vitro and in vivo whereas flavones not [22]. This suggested the right conformation for chalcones to interact with FXR is the more extended, low energy comformation B (Fig. 2). Here we reported an attempt to retain or increase the FXR interaction by cyclizing the different ring from that in benzopyrene formation, but still retain the right conformation of chalcones, leading to the discovery of the benzoxepin-5-one based FXR antagonists.

2. Results and discussion 2.1. Chemistry

As shown in Scheme 1, one phenol in the different substituted dihydroxy-acetophenones 11a-11d were selectively protected in the presence of chloromethoxymethane and Hünig' base [23]. Subsequently, the other phenol group in the mono-protected compounds 12a-12d was treated with 1, 2-dibromoethane and then cyclized to afford the corresponding benzoxepines 14a-14d. In the ether formation step, 12a-12d were only partially converted to 13a-13d in the presence of potassium carbonate as the base, even at high temperature (110 ℃) and prolonged time (20 h). Then a more soluble and stronger base cesium carbonate was used instead, and 13a-13d obtained in good to excellent yield (>95%) at 100 mg scale. However, when the reaction was scaled up to 1 g, only about 30% convertion was observed and byproducts (phenylethynyl ether and others unidentified) began to appear with prolonged reaction time, when a magnetic stirring bar was used. When this reaction was conducted with mechanical stirring (possibly better mixing), 13a-13d was afforded in good yield at gram-scale (1g-10 g). Finally, the condensation of benzoxepines 14a-14d with different benzaldehydes in the presence of catalytic amount of piperidine in solvent-free conditions and subsequent deprotection under acidic conditions in one-pot afforded corresponding benzoxepin-5-ones 10a-10l in good yields. The spectral data of most potent compound 10l was shown below.

6-Hydroxy-4-(4-methoxybenzylidene)-3, 4-dihydrobenzo[b] oxepin-5(2H)-one (10l): ¹H NMR (CDCl₃, 500 MHz): δ 12.13 (s, 1H, OH), 7.55 (s, 1H, CO-C=CH), 7.43 (d, 2H, J = 8.5 Hz, H-2', H-6'), 7.36 (m, 1H, H-8), 6.96 (d, 2H, J=8.5Hz, H-3', H-5'), 6.77(d, 1H, J=8.5Hz, H-9), 6.60 (d, 1H, J=8.0Hz, H-7), 4.39 (t, 2H, J=6.0Hz, O-CH₂-CH₂), 3.85 (s, 3H, O-CH₃), 3.02 (t, 2H, J=6.0Hz, C-CH₂-CH₂); ¹³C NMR (CDCl₃, 125MHz): 199.4 (C, CO), 162.9 (C, C-9a), 160.2 (C, C-4'), 159.7 (C, C-6), 138.2 (CH, CO-C-CH), 136.6 (C, CO-C-CH), 135.9 (C, C-8), 131.5 (2C, C-2', C-60), 127.7 (C, C-1'), 115.3 (C, C-5a), 114.2 (2C, C-3', C-5'), 113.1 (C, C-7), 112.2 (C, C-9), 71.5 (CH₂, O-CH₂-CH₂), 55.3 (CH₃, O-CH₃), 29.2 (CH₂, CH₂-CH₂-C); ESI-MS: [M+H]⁺: 297.1; HRMS: C₁₈H₁₇O₄⁺: m/z 297.1129 (observed), 297.1121 (calculated).

Experimental procedure for the synthesis of all 1-benzoxepin-5-one and their spectral data, were shown in Supplementary material.

2.2. Biological activity

The cellular activities of compounds 10a-10l were screened by using a mammalian one-hybrid FXR coactivator association assay. Encouragingly, three compounds (10j-10l) showed moderate to high antagonistic activities at 10^-5 mol/L against the FXR activation induced by the same concentration of CDCA (Fig. 3). It was found that the 6-hydroxyl group in benzoxepin-5-one (10j-10l) is essential for its FXR antagonistic activity and a 4'-methoxyl group favourable for higher activity (10j-10l vs. 10g-10i).

The benzoxepin-5-one 10l, the most potent FXR antagonists among this series of compounds, was chosen to evaluate its metabolic regulation activity in vivo. It was administrated orally (200mg/kg, qd) in diabetic KKay mice for 28 days. The plasma triglyceride level in 10l treatment group decreased by 33.3% and 27.7% (P < 0.05, P < 0.05, Fig. 4A) after 16 and 28days of treatment, respectively. Hepatic triglyceride level was also decreased by 31.6% (P < 0.05, Fig. 4B) after 28 days' treatment. The increased level of ALT (also named as GPT, glutamic pyruvic transaminase) and AST (also named as GOT, glutamic oxaloacetic transaminase) in plasma is usuallyassociated with damaged liver. Compound 10l was found to reduce the plasma ALT level by 23.1% (P < 0.05, Fig. 4B) and had no effect on the plasma AST level after 28days of treatment, indicating this antagonist does nothavehepatitis damageand even could play a hepatic protection role. Besides, it also did not affect the food and water intake and the body weight of mice.

3. Conclusion

In summary, we reported a new series of FXR antagonists, benzoxepin-5-ones, based on the conformational restriction of chalcones. The most potent antagonist 10l in cells significantly reduce the triglyceride in plasma and hepatic and plasma ALT level upon its treatment after 28days in KKay mice. This research provided a new chemotype to the repertoire of FXR antagonists, and showed the potential of FXR antagonist in the therapeutics of hypertriglyceridemia and in hepatic protection.

4. Experimental

Experimental procedure for the synthesis of 1-benzoxepin-5-one and their spectral data, cellular and animal experiment procedures were shown in Supplementary material.

Acknowledgment

This work was supported by the Hong Kong, Macao and Taiwan Science & Technology Cooperation Program, MOST of China (No. 2012DFH30030).