Design and Synthesis of Hydroxyethylene-Based BACE-1 Inhibitors Incorporating Extended P1 Substituents

Novel BACE-1 inhibitors with a hydroxyethylene central core have been developed. Modified P1´ and extended P1 substituents were incorporated with the aim to explore potential interactions with the S1´ and the S1-S3 pocket, respectively, of BACE-1. Inhibitors were identified displaying IC50 values in the nanomolar range, i.e. 69 nM for the most potent compound. Possible inhibitor interactions with the enzyme are also discussed.


INTRODUCTION
Alzheimer's disease (AD) is a serious and fatal condition affecting tens of millions of individuals worldwide [1]. This form of dementia is primarily believed to be caused by the formation of insoluble polypeptides which form neurodegenerative plaques within the brain. One of the key enzymes involved in this formation is the human aspartic protease BACE-1 [2]. This knowledge together with the fact that no amyloid plaques are found in knock-out mice lacking BACE-1, while they seem to be vital and fertile, make inhibition of BACE-1 an interesting approach for targeting AD [3,4].
Earlier studies of aspartic proteases such as the malaria plasmepsins [5] and human renin [6] have shown that the S1 pocket of these enzymes can accommodate large P1 residues, many of which can reach into the S3 pocket. In a previous study we explored the possibility of extending deep into the BACE-1 S1 pocket by employing statine-based inhibitors with phenyloxymethyl residues in the P1 position Fig. (1) [7]. Some of these inhibitors displayed substantial activity against BACE-1 (12-40 nM), in the enzyme assay, but they were found to be inactive in a cell-based assay, likely due to poor permeability.
In the present work we have further investigated the possibility of improving cell permeability by reducing the polarity of the inhibitors. We also decided to focus on a hydroxyethylene (HE) central core in place of the statine-based isostere. An HE core with an O-methyl P1´ functionality was previously and successfully employed in our laboratories to synthesize potent BACE-1 inhibitors [8]. The reason for this change was that the carboxylic acid functionality of statine-*Address correspondence to this authors Department of Chemistry, Linköping University, S-581 83 Linköping, Sweden; Tel: +46 13 281000; Fax: +46 13 281399; E-mails: versa@ifm.liu.se; andda@ifm.liu.se based isosteres, which is a critical feature for high activity for this class, render inhibitors with poor cell-based permeability [7]. Further, the HE isostere contains one less amide linkage compared to the statine core, which decreases the peptidic character of these inhibitors. In addition, various substituents at the P1´ position have been evaluated.
Based on these observations, the synthesis of a series of hydroxyethylene-based inhibitors containing different P1´ residues and P1 substituents of variable sizes was undertaken. The main objective was to further investigate potential interactions with the S1-S3 binding pocket of BACE-1 Fig. (1).

Chemistry
A general structure of the inhibitors that were synthesized, along with the various R substituents used, is illustrated in Fig. (2). R 4 amines M-O were synthesized from commercially available precursors using standard coupling and deprotection procedures (see the experimental section) and amine P was synthesized from commercially available precursors according to a three step protocol (including coupling, reduction and deprotection) (see the experimental section). The phenols C and D were commercially available. The P1 R 2 residues E and F were synthesized from their corresponding benzyl ethers, as shown in Schemes 1 and 3. An electrophilic aromatic substitution on 3-benzyloxy-phenol (C), utilizing N-bromosuccinimide (NBS) in DCM at -15 °C, produced aryl halide 1 in 71% yield (Scheme 1) [9]. Compound 1 was subsequently subjected to Mitsunobu-like conditions using 3-methoxy-1-propanol, PPh 3 , and DIAD in THF to give compound 2 in 90% yield. We then applied microwaveassisted palladium coupling conditions, which in this case involved treatment of 2 with 4-fluorophenylboronic acid or phenylboronic acid, K 3 PO 4 , and PEPPSI ™ -IPr catalyst in EtOH, H 2 O, and DMF. Microwave irradiation at 125 °C for 15 min produced biphenyl derivatives 3 and 4 in 70% and 81% yield, respectively. The corresponding benzyl ether of compound G was synthesized from C by applying the same conditions utilized to prepare compound 2, and compound H was obtained from commercially available 4-bromophenol using the same Suzuki protocol employed to synthesize compound 3. Compounds A and B were prepared according to reported procedures [10,11].   Fig. (1). A statine-based central core (I) described in a previous report on potent BACE-1 inhibitors [7]; and a general illustration of the hydroxyethylene (HE) isostere (II) (with R, X and Y substituents) that was used in this work. Different P1´ residues were incorporated into the HE isostere with the aim to find improved inhibitor interactions. The purpose of the X and Y substituents on the P1 residue was to protrude deeper into the S1 pocket and to extend into the S3 pocket, respectively.  The synthesis of compounds 12a-d is depicted in Scheme 2. 1,2:5,6-Di-O-isopropylidene--D-glucofuranose was used as a precursor from which compound 5 was synthesized in 49% yield over five steps using a method described in the literature [12][13][14][15]. Treatment of 5 with MeI and Ag 2 O in DMF furnished the methyl ether 6 in 89% yield. Hydrolysis of 6 using H 2 SO 4 in dioxane gave hemiacetal 7a in 85% yield, and the subsequent oxidation with PDC in DCM produced lactone 8a in 77% yield [8]. Lactone 7b was synthesized according to a literature procedure [13] and was used as a precursor for synthesizing compounds 8b-d. Stereoselective alkylation at the C2 position of 7b was achieved using LDA, tripyrrolidinophosphine oxide, and EtBr or 2iodopropane in THF at -68 °C to furnish the alkylated lactones 8b and 8d in 32% and 41% yield, respectively. Compound 8c was produced in 63% yield from 7b using a similar procedure with LDA and BnBr in THF at -78 °C. Next, compounds 8a-d were exposed to catalytic hydrogenolysis to prepare the diols 9a-d in 62-98% yield. Selective primary benzylation [16] was performed by converting 9a-d into their corresponding tin acetals by reacting the diols with Bu 2 SnO in refluxing toluene, followed by treatment with tetrabutylammonium bromide and 4-methoxybenzyl bromide to furnish the monobenzyl ethers 10a-d in 70-92% yield over two steps. Employing a Mitsunobu-like protocol with PPh 3 , DIAD, and DPPA in dry THF provided azides 11a-d in 57-99% yield. Finally, an oxidative removal of the pmethoxybenzyl group using 2,3-dichloro-5,6-dicyano benzoquinone (DDQ) in H 2 O and DCM, delivered compounds 12a-d in 91-100% yield [17].
In Scheme 3 the synthesis of target compounds 15 and 16 is presented. Initially, compound 3 was converted to the cor- responding phenol by catalytic hydrogenolysis using H 2 /Pd-C in ethanol. Alcohol 12a was then reacted with Tf 2 O and pyridine in DCM to yield the corresponding triflate. The phenol and triflate were used without further purification in a substitution reaction, using Cs 2 CO 3 in DCM containing 4 Å molecular sieves, to give compound 13 in 47% yield. It should be noted that we first applied Mitsunobu conditions in an attempt to produce 13. However, the use of this protocol led to significant elimination problems whereas the former approach described above reduced this side reaction considerably. This could be achieved by using fresh Cs 2 CO 3 in dry DCM, and by letting the reaction proceed for a short time at 40 °C under N 2 atmosphere. The lactone 13 was opened using amine O and 2-hydroxypyridine in DIPEA and DMF at 75 °C, which gave 14 in 83% yield. Reduction using PPh 3 in H 2 O and methanol afforded target compound 15 in 84% yield. Amine 15 was coupled with carboxylic acid A using HATU and DIPEA in DMF to provide target compound 16 (61%). It should be noted that the steps described in Scheme 3 are a representative final route for synthesizing the target compounds. All inhibitors were prepared following this general procedure with some variations, i.e. different phenols were used in step c, opening of the lactone was performed with amines M-P and the final peptide coupling step was only performed in the synthesis of final products 16, 18, 19, 24, and 25 (see the experimental section). Attempts to co-crystallize inhibitor 15 with BACE-1 were performed, but these were not successful.

Structure activity relationships
The target compounds are summarized in Table 1. They were synthesized from compounds 12a-d according to Scheme 3 using appropriate R substituents shown in Fig. (2). Enzyme activities were measured against BACE-1, and the IC 50 values are presented in Table 1. In addition, percent inhibition in a cell-based assay was determined at the concentration 1 μM for the four most potent inhibitors (compounds 25, 27, 28, and 30).
Initially, m-benzyloxy insertion in the P1 position of the general structure outlined in Table 1 gave the inactive amine 17 and the modestly active sulfonamide 18 (IC 50 values > 10 and 2.2 μM, respectively). The P1 p-benzyloxy-modified sulfonamide 19 proved to be nearly equipotent (IC 50 4.6 μM) to the corresponding P1 m-benzyloxy compound 18. Nonetheless, the results for 18 and 19 were promising and indicated that more advanced P1 substituents could provide inhibitors with increased activity. Further development led to BACE-1 inhibitors 20 and 15 containing the P1 R 2 substituent E and P2´-P3´ R 4 substituents N and O Fig. (2) with a pchloro substituent in the P3´ part of the molecule. Compound 15 has an isoleucine in the P2´ position and an IC 50 value of 1.2 μM, making it slightly more active than the corresponding P2´ valine-containing inhibitor 20 (IC 50 2.7 μM). In addition, the fluorine substituent on the P1 moiety seems to contribute somewhat to the activity, which can be tentatively concluded when looking at compound 21 (IC 50 2.5 μM), lacking the fluorine. The activity of this class of inhibitors most likely depends on interactions from both the para and the meta groups on the P1 substituent, as observed for targets 22 and 23 (IC 50 values > 10 μM) lacking the para and the meta substituent, respectively. Notably, 15 is a moderately more potent inhibitor than 18, even though it does not contain a P2 substituent. This can probably be explained by an interaction effect of the more advanced P1 substituent. Unfortunately, introduction of the P2 sulfonamide substituent A into 15 and 21 rendered compounds highly hydrophobic in character, which caused precipitation problems during the activity measurements and thus no inhibition data were obtained for 16 and 24. Compound 25 contains the larger P2-P3 portion B and has a notable IC 50 value of 69 nM, which is very intriguing, since it might indicate that the active site possibly can accommodate such a large compound having both the P3 portion and the sizable P1 residue. Further exploration of the P1´ and P3´ positions were done to see if a more drug-like compound could be found, with the P2-P3 substituent excluded. Modification in the P1´ position resulted in compound 26 which proved to be almost equipotent with compound 15. Also, compound 29 proved to be roughly equally potent as 15. Compound 27 has a benzyl group in the P1´ position and an IC 50 value of 0.47 μM, making it somewhat more active than inhibitor 15. To investigate if the amide-bond between P2´ and P3´ contributes to biological activity, compounds 28 and 30 were synthesized. Interestingly, 28 and 30 have IC 50 values of 0.30 and 0.54 μM, respectively, which are equal to the activities seen for compounds 27 and 29, indicating that the P2´-P3´ amide portion is noncritical for BACE-1 activity in this series of inhibitors.
Even though we were able to exclude the P2-P3 portion of the inhibitors and still maintain promising enzyme activities in the nanomolar range, the size and the overall polarity of the compounds unfortunately resulted in poor activities in the cell-based assay with 3%, 19%, 14%, and 28% inhibition of BACE-1 at the concentration 1 μM for the four most active inhibitors 25, 27, 28, and 30, respectively. The fact that compound 25 shows such a low inhibition as 3% can most likely be explained by its size and peptide-like structure. As expected, decreased size and less peptidic character lead to increased inhibition in cells.

CONCLUSION
We have synthesized a series of BACE-1 inhibitors based on a hydroxyethylene core. Although our focus has been on the P1 part of the target molecules, other positions of the inhibitors have also been varied and studied. Enzyme measurements (to give IC 50 values) have been performed on the inhibitors and percent inhibition in a cell-based assay was measured for the four most active inhibitors. Various side chain substituents have been employed and the best inhibitor (target 25) displayed an enzyme IC 50 of 69 nM. Furthermore, target molecules 27-30 all had IC 50 values in the nanomolar range. Interestingly, though they lack a P2 substituent, they still performed well potency-wise. Moreover, compounds 28 and 30 possess only one amide bond, which make them more drug-like when compared with other published inhibitors [18]. The cell-based assays revealed that, as could be foreseen, smaller molecular size and less peptidic character led to increased activity in this regard. Attempts to co-crystallize inhibitor 15 with BACE-1 were done, but were unfortunately unsuccessful. X-ray data could have given us valuable information about interactions between our compounds and BACE-1, which hopefully would have provided us with new ideas on further optimizations. Further research should therefore be aimed at obtaining X-ray structures showing the interactions of our best inhibitors in the active site of BACE-1.

General
NMR-spectra were recorded on a Varian 300 MHz instrument using CDCl 3 and CD 3 OD as solvents. TLC was carried out on Merck precoated 60 F 254 plates using UV-light and charring with ethanol/sulfuric acid/p-anisaldehyde/acetic acid 90:3:2:1 or a solution of 0.5% ninhydrin in ethanol for visualization. Flash column chromatography was performed using silica gel 60 (0.040-0.063 mm, Merck). Organic phases were dried over anhydrous magnesium sulfate. Concentrations were performed under diminished pressure at a bath temperature of 40 °C. Optical rotations were measured using a Perkin-Elmer 141 polarimeter.
Gradient LC-MS was performed on a Gilson system (column: Phenomenex C-18 250 x 15 mm, 10 μm and Phenomenex C-18 150 x 4.6 mm, 5 m for preparative and analytical runs, respectively; pump: Gilson gradient pump 322; UV/VIS-detector: Gilson 152; MS detector: Thermo Finnigan Surveyor MSQ; Gilson Fraction Collector FC204) using methanol with 0.1% formic acid and deionized water with 0.1% formic acid as mobile phases. High resolution mass spectra (HRMS) were recorded on a Waters Synapt HDMS instrument equipped with an electrospray interface.

BACE-1 Enzyme Assay
The BACE-1 assay was performed as previously described [7]. The inhibition of BACE-1 was determined in a homogeneous time resolved fluorescence (TRF) assay (True Point kit, Perkin-Elmer). The assay buffer contained sodium acetate, CHAPS, Triton-X 100, and EDTA, pH 4.5. The substrate used was the Swedish mutant sequence Eu-EVNLDAEFK-Quencher.

BACE-1 Cell Assay
The cell-based assay was performed as previously described [13]. Human embryonic kidney 293 cells (HEK-293) stably expressing Swedish mutant APP (swAPP751) were cultured in DMEM supplemented with 10% FCS, PEST (50 U penicillin/50 mg/mL streptomycin) and 200 mg Hygromycin/mL at 37 °C, 5% CO 2 . After 24 h cell culture media was har-vested and analysed by Ab 1-40 ELISA according to the manufacturer instruction (The Genetics Company, Schwitzerland). The results were treated with the curve fitting package in GraphPad Prism 5.

Peptide Coupling
To a cooled (0 °C) solution of the acid (4.95 mmol) in DMF (20 mL) were added the amine (5.44 mmol), DIPEA (14.84 mmol) and HATU (5.94 mmol). The solution was stirred at 0 °C for 0.5 h and for an additional 2 h at room temperature. EtOAc and brine were added and the organic phase was separated and washed once with brine, dried and concentrated. The crude peptide mixture was purified by flash column chromatography.

Boc Deprotection
To a solution of the protected compound (0.167 mmol) and triethylsilane (0.419 mmol) in DCM (3 mL) was added TFA (1 mL). The solution was stirred for 2 h at room temperature and then concentrated and co-evaporated with toluene.

General Procedure A. Debenzylation
To a solution of 3 (121 mg, 0.330 mmol) in EtOH (95%, 4 mL) was added Pd-C (10%, 60 mg) and the mixture was hydrogenolyzed (atmospheric pressure) at room temperature for 3h. The suspension was filtered through Celite and concentrated. The resulting phenol was used without any further purification in the following step.

General Procedure B. Substitution
To a cooled (0 °C) solution of 12a (60 mg, 0.298 mmol) and dry pyridine (46 μL, 0.571 mmol) in dry DCM (3 mL) was added Tf 2 O (84 μL, 0.498 mmol) and the mixture was stirred at 0 °C for 45 min. The reaction was quenched with saturated aqueous NH 4 Cl and the organic phase was separated, dried, filtered and concentrated.
The phenol and the triflate (obtained from the general procedures A and B, respectively) were dissolved in dry DCM (3 mL) and stirred together with 4 Å powdered molecular sieves (50 mg) for 10 min before the addition of

General Procedure E. Peptide Coupling
To a cooled (0 °C) solution of acid A (5.0 mg, 0.022 mmol), amine 15 (17 mg, 0.025 mmol) and DIPEA (12 μL, 0.069 mmol) in DMF (1 mL) was added HATU (10 mg, 0.026 mmol). The solution was stirred for 0.5 h at 0 °C and for an additional 2 h at room temperature. EtOAc and brine were added and the organic phase was separated and washed once with brine, dried and concentrated. Purification using LC-MS gave 16 (