Design, Synthesis and Evaluation of 3-(2-Aminoheterocycle)-4-benzyloxyphenylbenzamide Derivatives as BACE-1 Inhibitors

Three series of 3-(2-aminoheterocycle)-4-benzyloxyphenylbenzamide derivatives, 2-aminooxazoles, 2-aminothiazoles, and 2-amino-6H-1,3,4-thiadizines were designed, synthesized and evaluated as β-secretase (BACE-1) inhibitors. Preliminary structure-activity relationships revealed that the existence of a 2-amino-6H-1,3,4-thiadizine moiety and α-naphthyl group were favorable for BACE-1 inhibition. Among the synthesized compounds, 5e exhibited the most potent BACE-1 inhibitory activity, with an IC50 value of 9.9 μΜ and it exhibited high brain uptake potential in Madin-Darby anine kidney cell lines (MDCK) and a Madin-Darby canine kidney-multidrug resistance 1 (MDCK-MDR1) model.


Introduction
More than 25 million people in the world are suffering from dementia, the most common form among which is Alzheimer's Disease (AD), characterized by progressive memory loss and cognitive decline [1]. The major pathological hallmark for AD research is the unfolding of amyloid plaques and OPEN ACCESS neurofibrillary tangles. The amyloid cascade hypothesis, which stated that accumulation of β-amyloid (Aβ) in the brain is the leading factor in the pathogenesis of Alzheimer's disease [2], has been supported by abundant genetic and pathological evidence. Amyloid precursor protein (APP) is sequentially processed by β-secretase (also known as BACE-1) and γ-secretase to liberate Aβ. It has been revealed that BACE-1 −/− mice are devoid of cerebral Aβ production without any sign of significant dysfunction [3]. In addition, the rescue of memory deficits in BACE-1 −/− APP bigenic mice suggested that BACE-1 inhibition would improve Aβ-dependent cognitive impairment in AD patients [4]. Therefore, BACE-1 has been considered to be an attractive target for the therapy of AD with lots of small molecular inhibitors discovered in the past few years. Due to low oral bioavailability, metabolic instability, and poor ability to penetrate brain barrier of peptidomimetic inhibitors, recent attention has been mainly focused on non-peptidomimetic scaffolds, including 1,3,5-trisubstituted aromatics, isophthalamides, acylguanidines, piperazines, macrocycles, amino heterocycles and so on [5][6][7][8][9]. Among compounds with these different scaffolds, aminoheterocyclic derivatives, such as 2-amino-thiazole, 2-aminopyridine, 6-aminoimidazopyrimidine, 2-aminoquinazoline and 2-amino-pyrimidinone, have been of great interest in recent years due to their simple structures and specific binding modes with BACE-1 [10][11][12][13][14][15][16].
2-Aminopyridine derivative 1 (Figure 1), prepared via a fragment based drug design strategy by Congreve and coworkers was reported as a BACE-1 inhibitor with an IC 50 value of 0.69 μM [13]. The X-ray structure of 1/BACE-1 complex revealed that 1 occupied two hydrophobic pockets (S 1 and S 2 '), and the 2-aminopyridine moiety directly interacted with two catalytic aspartic acids (Asp32 and Asp228) via two hydrogen bonds ( Figure 1).   Stachel and coworkers revealed that the 2-aminothiazole derivative 2 showed similar binding mode with BACE-1 enzyme in the active site with the amino group interacting directly with Asp32 and Asp228 through a bidentate interaction [14]. Compound 2 exhibited a brain-to-plasma ratio value of 3.9 when it was administered to mice at a 20 mg/kg iv dose (t = 30 min; [brain] = 15 μM). The similar binding features of 1 and 2 with BACE-1 and the desirable brain-barrier penetrating characteristics of compound 2 prompted us to design new amino-heterocyclic derivatives as potent BACE-1 inhibitors by using the following drug design strategies: (1) the 1,2,4-trisubstituted benzene moiety from compound 1 was taken as the skeleton and the 1-benzyloxy moiety was retained to make hydrophobic interactions with S 2 ' binding pocket; (2) 2-aminothiazole, 2-aminooxazole and 2-amino-6H-[1,3,4]thiadiazine moieties were respectively introduced into the 2-position of the benzene based on the  bioisosterism principle to form exquisite hydrogen bonds with the two catalytic Asp32 and Asp 228 of  BACE-1; (3) various substituted phenyl, pyridyl, phenylalkyl and naphthyl groups were incorporated into the 4-position of the benzene with an amide linkage to fit into the S 1 or S 3 binding pocket of BACE-1.
To confirm our hypothesis and predict the binding model of the designed compounds in the active pockets of BACE-1, molecular docking studies of 2-aminothiazole derivative 3a, 2-aminooxazole derivative 4a and 2-amino-6H-1,3,4-thiadizine derivative 5a with BACE-1 were performed by using the 2.1/CDOCKER protocol within Discovery Studio package (Figure 2A-D). The crystal structure of ligand/BACE 1 complex (PDB ID: 1W51) was used as the template [15]. As shown in Figure 2A,B, compounds 3a and 4a shared similar binding modes with BACE-1. The 1-benzyloxy group was located in the S 2 ' pocket and the benzamide moiety fit within the critical S 1 pocket. The amino groups of 3a and 4a formed hydrogen bonds with Asp32, with distances of 1.99 Å, 2.80 Å and 2.66 Å, respectively. The oxygen atoms of the benzyloxy linkage in 3a and 4a formed additional hydrogen bonds with Arg235, with distances of 3.02 Å, 3.03 Å and 2.80 Å, respectively. Interestingly, Figure 2C,D revealed that 2-amino-6H-1,3,4-thiadizine derivative 5a bound to BACE-1 in a different manner. The 4-benzamide moiety extended into the critical S 3 pocket rather than the S 1 pocket. The amino group of 5a not only formed bidentate hydrogen bonds with Asp32 with distances of 2.44 Å and 2.89 Å, but also formed a hydrogen bond with Gly34 with a distance of 3.18 Å. Additionally, the two nitrogen atoms of 6H-1,3,4-thiadizine formed bidentate hydrogen bonds with Asp228. Besides, another hydrogen bond was presented between the oxygen atom of the benzyloxy linkage and Thr231. Based on the docking analysis, three series of 3-(2-aminoheterocycle)-4benzyloxy-phenyl-benzamide derivatives 3a-i, 4a-e and 5a-e was synthesized and evaluated for their BACE-1 inhibitory activities.

BACE-1 Inhibition Activity
The obtained target compounds were tested for their BACE-1 inhibitory activities using a fluorescence resonance energy transfer (FRET) assay, with OM99-2, a potent peptidomimetic inhibitor, as the positive control [16]. Compounds with a BACE-1 inhibition rate higher than 50% at 20 μg/mL were tested for their IC 50 values. The results are summarized in Table 1. As shown in Table 1, most of the tested compounds demonstrated moderate to good BACE-1 inhibition at 20 μg/mL, 13 compounds exhibited more than 30% inhibition and five compounds showed more than 50% inhibition. Preliminary structure-activity relationships could be concluded as follows: (1) The variation of the heterocycle moiety affected the BACE-1 inhibitory activities significantly. 2-Amino-6H-1,3,4-thiadizine derivatives were more potent than 2-aminothioazole and 2-amino-oxazole derivatives. Four of the 2-amino-6H-1,3,4-thiadizine derivatives (compounds 5a, 5b, 5d and 5e) showed more than 60% inhibition against BACE-1 at 20 μg/mL (64.2, 76.0, 84.9 and 60.0%, respectively). Among the synthesized compounds, 5a and 5e are the two most potent BACE-1 inhibitors, with IC 50 values of 16.7 and 9.9 μM, respectively. The 2-aminooxazole derivatives 4a-e demonstrated similar BACE-1 inhibitory activities as the 2-aminothioazoles 3a-e, with inhibition rates ranging from 34.4% to 49.9% in comparison with 20.4% to 55.3% at 20 μg/mL. The previous docking study of compounds 3a, 4a and 5a with BACE-1 reveals that the enhanced potency of 2-amino-thiadiazine derivatives 5a-e may be attributed to their different binding modes with BACE-1 than the 2-aminothiazole and 2-aminooxazole derivatives, with the amino group forming two additional hydrogen bonds with Gly34 and Asp228, and the benzamide moiety fitting into the S 3 pocket instead of the S 1 pocket (Figure 2A,B versus Figure 2C,D).
(2) A glimpse of different substituents at the R 1 -position implied that α-naphthyl group introduction was favorable for BACE-1 inhibition. For example, 3e, 4e and 5e exhibited 55.3%, 42.6% and 60.0% of BACE-1 inhibition at 20 μg/mL, respectively, and 5e showed the lowest IC 50 value (9.9 μM) among all the compounds. This implied that a bulky naphthyl ring was better accommodated into the critical S 1 or S 3 binding pockets and an electron-rich ring had a better hydrophobic interaction with amino acids. The existence of chloro, methoxy or trifluoromethyl phenyl ring substituents had a limited effect on BACE-1 inhibitory activities, for example, 2-aminothiazole derivatives 3a-d showed similar BACE-1 inhibition rates, ranging from 20.4% to 33.0%.

In Vitro Blood-Brain Barrier Permeability
Blood-brain barrier (BBB) permeation is critical for any AD therapeutic drug. Many previously synthesized potent BACE-1 inhibitors displayed poor brain barrier penetration, which restricted their further development. For example, the highly potent BACE-1 inhibitor GSK188909 (IC 50 = 5.0 nM) showed poor blood-brain barrier permeability, and it need to be combined with Pgp inhibitor GF120918 to exert its Aβ reducing activity in the brain of mice [17]. In order to investigate the BBB permeability of the newly synthesized aminoheterocyclic derivatives, the most potent compound 5e was picked out to evaluate its transport efficient (P app values) in Madin-Darby canine kidney cell line (MDCK) and Madin-Darby canine kidney-multidrug resistance 1 (MDCK-MDR1) monolayer cells (in vitro cell culture model of BBB). The results are summarized in Table 2. Concentration of 5e was 55.6 μM, transport efficient (P app value) are presented as the mean ± SD; n = 3.
As shown in Table 2, compound 5e exhibited high apparent permeability coefficients P app (A-B) and P app (B-A) in the MDCK cell model, with values of 28.20 × 10 −6 cm/s and 27.66 × 10 −6 cm/s. Similarly, 5e also exhibited P app (A-B) and P app (B-A) values of 31.78 × 10 −6 cm/s and 22.23 × 10 −6 cm/s in the MDCK-MDR1 cell model, respectively. It has been reported that compounds with P app (A-B) values > 3 × 10 −6 cm/s in the MDCK-MDR1 model would have high brain uptake potential [18]. These results suggested that compound 5e had a good penetration ability through the blood-brain barrier. The efflux ratios of 5e in the MDCK and MDCK-MDR1 models were 0.98 and 0.70, respectively. The net efflux ratio of 5e was 0.71, less than FDA's recommendation of 2 for P-gp substrate, which indicated that compound 5e was not a P-gp substrate [19].

General
All reagents and solvents used were analytical grade and purchased from common commercial suppliers. Melting points were determined with a B-540 Buchi melting-point apparatus and are uncorrected. 1 H-NMR was performed on a Brüker Advance DMX 400 MHz spectrometer with TMS as internal standard. Proton chemical shifts were expressed in parts per million (ppm) and coupling constants in Hz. HRMS spectra were measured with an Agilent 6224 TOF LC/MS. Mass spectra (ESI-MS, positive) were recorded on a Finnigan LCQ DecaXP ion trap mass spectrometry. Molecular docking studies were performed using Discovery Studio 2.1. (7). To a warmed (90-95 °C) mixture of 4-nitrophenol (6, 2.78 g, 0.02 mol) in aqueous NaOH solution (20 mL, 1.5 mol/L) was added Ac 2 O (2.83 mL, 0.03 mol). The mixture was stirred and cooled to room temperature. The formed precipitate was collected by suction filtration, washed with water and dried in vacuo to afford 7 as a pale yellow solid (3.52 g, 97.2%), m.p. 78-80 °C (lit. 77-79 °C) [20].

General Procedure for the Synthesis of 11a-i
To a mixture of 2-benzyloxy-5-aminoacetophenone (10, 0.72 g, 3.0 mmol) and Et 3 N (0.87 mL, 6.0 mmol) in CH 2 Cl 2 (10 mL), a solution of aromatic acyl chlorides (4.5 mmol) in CH 2 Cl 2 (10 mL) was added in dropwise within 30 min. The mixture was stirred for 6-8 h at room temperature. Then, saturated NaHCO 3 solution (20 mL) was added and organic layer was separated, washed with brine and dried over anhydrous sodium sulphate. The solvent was removed under vacuum and the residue was purified by silica gel chromatography eluting with PE-EtOAc (5:1) to give 11a-i.

General Procedure for the Synthesis of 12a-i
To a refluxed solution of N-(3-acetyl-4-(benzyloxy)phenyl)amide derivatives 11a-i (2.0 mmol) in 1:1 EtOH-CH 3 Cl (15 mL) was added CuBr 2 (4.0 mmol) in three portions within 2 h. The mixture was cooled to room temperature and filtered. The filtrate was concentrated to dryness and the residue was extracted with EtOAc twice. The organic layer was combined, washed with brine, dried with Na 2 SO 4 , evaporated in vacuo to dryness. The residue was purified with silica gel chromatography eluting with PE-EtOAc (5:1) to afford 12a-i. The mixture of bromoacetophenone derivative 12a-i (0.2 mmol) and thiourea (0.22 mmol) in EtOH (10 mL) was refluxed for 6-8 h until the substrate 12a-i had disappeared. After cooling to room temperature, the mixture was filtered and the filtrate was concentrated to dryness. The residue was purified by silica gel chromatography eluting with PE-EtOAc-Et 3 N (100:50:1) to afford white to pale yellow solids.   13  added to the apical side as a donor chamber, 1.5 mL fresh incubation medium was added to the basolateral side as a receiver chamber. For the secretion study (Basolateral to Apical), 1.5 mL incubation medium containing drug was added to the basolateral side as a donor chamber, 0.5 mL fresh incubation medium was added to the apical side as a receiver chamber. Transport studies were conducted at 37 °C in a humidified incubator with shaking (50 rpm) for 1 h, and then the collected samples were analyzed by HPLC. Apparent permeability coefficients (P app(A-B) , P app(B-A) ) and efflux ratio (P ratio = (P app(B-A) /P app(A-B) ) were used to evaluate the permeability and absorption profiles of compounds. P ratio is an important parameter to denote if a compound is a substrate of P-gp or not.

Conclusions
Three series of 3-(2-aminoheterocycle)-4-benzyloxyphenylbenzamide derivatives were designed, based on the binding mode between aminoheterocyclic derivatives and BACE-1, synthesized and evaluated as BACE-1 inhibitors. The results showed that most of these compounds demonstrated promising BACE-1 inhibitory activities and a preliminary SAR study revealed that a 2-amino-6H-1,3,4-thiadizine moiety and α-naphthyl group were favorable for BACE-1 inhibition, which was supported by a molecular docking study of 5a with BACE-1. Compound 5e exhibited the most potent BACE-1 inhibitor activity, with an IC 50 value of 9.9 μΜ, and also displayed favorable blood-brain barrier permeability in the MDCK and MDCK-MDR1 monolayer cell model. Our work revealed that the 2-amino-6H-1,3,4-thiadizine-4-benzyloxyphenylbenzamide would be a promising structural template for the development of BACE-1 inhibitors.