Abstract
Enantioselective syntheses of tert-butyl ((S)-2-(3,5-difluorophenyl)-1-((S)-oxiran-2-yl)ethyl)carbamate and ((S)-2-(3,5-difluorophenyl)-1-((R)-oxiran-2-yl)ethyl)carbamate are described. We utilized asymmetric syn- and anti-aldol reactions to set both stereogenic centers. We investigated ester-derived Ti-enolate aldol reactions as well as Evans’ diastereoselective syn-aldol reaction for these syntheses. We have converted optically active ((S)-2-(3,5-difluorophenyl)-1-((S)-oxiran-2-yl)ethyl)carbamate to a potent β-secretase inhibitor.
Keywords: Aldol reaction, Asymmetric, Protease Inhibitor, Fluoroisostere, Stereoselective
Graphical Abstract

The design of aspartic acid protease inhibitors continues to be an important area for drug development and discovery against a variety of human diseases.1 These include renin inhibitors for hypertension, HIV protease inhibitors for HIV/AIDS and β- and γ-secretase inhibitors for Alzheimer’s disease.1–3 The β-secretase inhibitors are particularly receiving much attention due to their potential for the treatment of Alzheimer’s disease (AD).4 BACE1 is a membrane anchored aspartic acid protease, responsible for the initial cleavage of amyloid precursor protein to neurotoxic amyloid-β-peptides in the brain. The amyloid-β-peptide is the main component of amyloid plaques, the neuropathological hallmark of AD.5,6 Since the discovery of BACE1 in 1999, extensive research efforts led to the evolution of a variety of small-molecule peptidomimetic and non-peptide BACE1 inhibitors with therapeutic potentials.7,8 Over the years, many potent and selective peptidomimetic BACE1 inhibitors have been designed based upon incorporation of traditional hydroxyethylene and hydroxyethylamine transition-state isosteres at the cleavage site of BACE1.4,9,10 The majority of early BACE1 inhibitors containing phenylalanine and leucine side chains as the P1 ligand showed low nanomolar BACE1 Ki values and they also exhibited good reduction of cellular Aβ production.4,7,8 At present, there are at least three small molecule BACE1 inhibitors advanced to clinical trials.11,12
A clinically effective BACE1 inhibitor, however, needs to have the ability to cross the blood-brain barrier (BBB) and the neuronal membrane.13 In this context, traditional transition-state isosteres were made more lipophilic by inserting fluorines in an effort to improve membrane permeability, metabolic stability, and enzyme-inhibitor interactions.4,7,8 As exemplified in Figure 1, peptidomimetic inhibitors 1 (BACE1 IC50 = 5 nM; Cell IC50 = 3 nM) and 2 (BACE1 IC50 = 30 nM; Cell IC50 = 3 μM) incorporated hydroethylamine and hydroxyethylene dipeptide isosteres, respectively, with a 3,5-difluorophenylmethyl group as the P1 ligand.14,15 These difluoro-dipeptide isosteres have been utilized in the design and synthesis of many other potent and selective BACE1 inhibitors with BBB permeability.4,7,8
Figure 1.

Structures of BACE-1 inhibitors 1 and 2.
In general, BACE1 inhibitors containing hydroxyethylamine isosteres show potent BACE1 inhibitory activity and cellular activity. The synthesis of inhibitors involves the opening of an aminoalkyl epoxide with an appropriate amine followed by the functionalization of the N-terminus with suitable P2 ligands.2,4 As shown in Figure 2, BACE1 inhibitors containing a 3,5-difluorophenylmethyl side chain as the P1 ligand in a general inhibitor 3, can be synthesized from tert-butyl-((S)-1-(S)-oxiran-2yl)-2-phenylethyl carbamate 4. Enantioselective synthesis of such epoxide is typically carried out with optically active 3,5-difluorophenyl alanine 5.16,17 In our continuing interest in the design and synthesis of novel BACE1 inhibitors containing fluorines at P1-ligand, we have investigated the synthesis of difluoroepoxide 4 utilizing an asymmetric syn-aldol as the key step.18,19 In principle, aldol product such as 6 would provide access to epoxide by removal of chiral auxiliary (XC) followed by Curtius rearrangement of the resulting acid to install the amine functionality.20,21 Furthermore, anti-aldol product such as 8, obtained from a diastereoselective anti-aldol reaction,22 could provide access to diastereomeric oxirane derivative 7 for the synthesis of BACE1 inhibitors with hydroxyethylene isosteres. Herein, we report a highly diastereoselective synthesis of (2R, 3S)-1,2-epoxy-3-(Boc-amino)-4-(3,5-difluorophenylmethyl)-butane 4 and (2R, 3R)-1,2-epoxy-3-(Boc-amino)-4-(3,5-difluorophenylmethyl)butane 7 (Figure 2) utilizing an asymmetric aldol reaction as the key step. The overall route is amenable to quantities of difluoroepoxide cores 4 and 7 in high optical purity.
Figure 2.

Structures of epoxides 4 and 8 and their respective aldol precursors
We first investigated ester-derived titanium-enolate based aldol reactions to afford both syn- and anti-aldol products for the stereoselective synthesis of aminoalkyl oxiranes 4 and 7. These reactions have been utilized in the synthesis of a number of dipeptide isosteres.19,20,23 The synthesis of oxirane 4 using syn-aldol reaction is shown in Scheme 1. 3,5-Difluorohydrocinnamic acid 9 was prepared in multigram scale as described in the literature.24,25 Reaction of this acid with commercially available N-tosyl-1-aminoindan-2-ol 10 with DCC in the presence of DMAP in CH2Cl2 at 23 °C for 18 h afforded ester 11 in 76% yield. Treatment of ester 11 with 1M TiCl4 in the presence of diisopropylethylamine (DIPEA) in CH2Cl2 at 0 °C to 23 °C for 2 h provided the corresponding Ti-enolate. Reaction of this Ti-enolate with (benzyloxy)acetaldehyde precomplexed with TiCl4 in CH2Cl 2(1 M CH2Cl2) at −78 °C for 2 h provided the syn-aldol product 12 in 60% yield. The 1H-NMR analysis revealed the presence of a single diastereomer. Saponification of aldol product 12 with aqueous lithium hydroperoxide at 0 °C to 23 °C for 12 h provided the corresponding acid which was subjected to Curtius rearrangement20,26 with diphenylphosphorazidate in the presence of triethylamine in dry benzene at 90 °C for 12 h to afford oxazolidinone derivative 13 in 70% yield over 2-steps. Oxazolidinone was converted to Boc-derivative 14 in a two-step sequence. Hydrolysis of 13 with aqueous KOH provided aminoalcohol which was reacted with di-tert-butyl dicarbonate in a mixture (1:1) of CH2Cl2 and water at 23 °C for 4 h to affod the Boc-derivative. Catalytic hydrogenation of benzyl ether over Pearlman’s catalyst using a hydrogen-filled balloon in ethyl acetate removed the benzyl group and diol 14 was obtained in 77% yield over 2-steps. The diol 14 was converted to epoxide 4 by a regioselective monotosylation of the primary alcohol with p-toluenesulfonylchloride and triethylamine in the presence of a catalytic amount (25 mol %) of dibutyltin oxide at 23 °C for 4 h to provide the corresponding tosylate. Exposure of this tosylate to K2CO3 in MeOH at 0 °C to 23 °C for 1 h afforded (S)-2-(3,5-difluorophenyl)-1(S)-oxiranyl ethyl carbamate 4 in 67% yield over two-steps. The overall route is straightforward and provided fluorine substituted chiral epoxide for the synthesis of BACE1 inhibitors.
Scheme 1.

Synthesis of 3,5-difluorobenzyl epoxide 4
For the synthesis of 3,5-difluorobenzyl oxirane 7, we carried diastereoselective anti-aldol reaction of ester 11 with cinnamaldehyde as shown in Scheme 2. Ester 11 was converted to Ti-enolate with 1M TiCl4 in CH2Cl2 in the presence of DIPEA as described above. This enolate was reacted with trans-cinnamaldehyde precomplexed with 1.3 equivalents of TiCl4 at −78 °C for 3 h to provide anti-aldol product 15 as a single isomer (by 1H-NMR and 13C-NMR analysis). Saponification of ester 15 with aqueous lithium hydroperoxide followed by Curtius rearrangement of the resulting acid as described above resulted in oxazolidinone derivative 16 in 96% yield over 2-steps. Ozonolytic cleavage of the double bond in compound 16 was achieved by passing ozone in a mixture (4:1) of CH2Cl2 and MeOH at −78 °C followed by reduction with NaBH4 providing alcohol 17 in 90% yield. Exposure of alcohol 17 to aqueous KOH followed by reaction of the resulting aminoalcohol with (Boc)2O in CH2Cl2 afforded Boc-derivative 18 in 67% yield. Diol derivative 18 was converted to (S,R)-3,5-difluorophenyl oxirane 7 by selective tosylation followed by treatment of the resulting tosylate with K2CO3 in MeOH as described for compound 14 in Scheme 1 to provide oxirane 7 in 67% yield over 2-steps.
Scheme 2.

Synthesis of 3,5-difluorobenzyl epoxide 7
The synthesis of (S)-2-(3,5-difluorophenyl)-1-(S)-(oxiran-2-yl)ethyl carbamate 4 using Evans’ syn-aldol reaction is shown in Scheme 3. Difluorohydrocinnamic acid 9 was transformed into a mixed anhydride with pivaloyl chloride and triethylamine at −78 °C. The resulting mixed anhydride was reacted with lithio derivative of chiral oxazolidinone 19 to furnish carboximide 20.27,28 Treatment of this chiral carboximide derivative with dibutylboron trifluoromethanesulfonate (Bu2BOTf) in the presence of N,N-diisopropylethylamine at −78 °C provided the boron enolate. Reaction of this boron enolate with benzyloxy acetaldehyde furnished aldol product 21 in 46% yield after standard work up and silica gel chromatography. The 1H- and 13C-NMR analysis showed the presence of a single diastereomer. In an effort to improve the overall efficiency of this asymmetric aldol route, we further explored the Evan’s asymmetric aldol reaction of chiral carboximide 20 with relatively inexpensive cinnamaldehyde. As shown in Scheme 3, formation of boron enolate with Bu2BOTf and aldol reaction with cinnamaldehyde proceeded very well providing syn-aldol product 22 in near quantitative yield. Aldol product 22 was obtained as a single diastereomer by 1H- and 13C-NMR analysis. This is a significant improvement over aldol reaction with benzyloxyacetaldehyde for aldol product 21. Exposure of aldol product 21 to lithium hydroperoxide in aqueous THF at 0 °C to 23 °C for 10 h to provide the corresponding carboxylic acid. This acid was subjected to Curtius rearrangement20,21 with diphenyl phosphorazidate in the presence of triethylamine in toluene at 90 °C for 16 h to provide the corresponding oxazolidinone derivative. Catalytic hydrogenation of benzyl ether over Pearlman’s catalyst provided alcohol 23 in 82% yield over 3-steps. This was converted to epoxide 4 as described above.
Scheme 3.

Synthesis of 3,5-difluorobenzyl epoxide 4
Aldol product 22 was also converted to epoxide 4. The removal of the chiral auxiliary by exposure to lithium hydroperoxide followed by Curtius rearrangement of the resulting acid with DPPA as described above furnished oxazolidinone derivative 24 in 68% yield after silica gel chromatography. Ozonolysis of the double bond in a mixture (4:1) of CH2Cl2 and methanol at −78 °C followed by a reductive workup with NaBH4 furnished alcohol 23 in 90% yield. Alcohol 23 was readily converted to Boc-derivative 14 by exposure to aqueous KOH in ethanol followed by protection of the resulting crude amine as Boc derivative 14 as described previously.
We demonstrated the utility of difluorophenylethyl oxirane derivative 4 in the synthesis of known BACE1 inhibitor 27.14 As shown in Scheme 4, reaction of oxirane 4 with 3-methoxybenzylamine in isopropanol at 80 °C for 12 h provided the corresponding aminoalcohol. Deprotection of the Boc-group by exposure to trifluoroacetic acid (TFA) in CH2Cl2 at 23 °C for 2 h afforded aminoalcohol 25. Coupling of the primary amine with known14 isophthalic acid derivative 26 in the presence of HATU and triethylamine in CH2Cl2 at 23 °C for 16 h furnished inhibitor 27 in 43% yield over three steps.
Scheme 4.

Synthesis of BACE1 inhibitor 27
In conclusion, we accomplished convenient syntheses of aminoalkyl oxiranes 4 and 7 containing 3,5-difluorobenzyl side chain using an asymmetric aldol reaction as the key step. The stereochemistry of both stereogenic centers was set by highly diastereoselective syn- and anti-aldol reactions. The removal of the chiral auxiliary followed by Curtius rearrangement of the resulting acid installed the amine functionality. This was readily converted to epoxides 4 and 7 efficiently. These epoxides are important building blocks for the synthesis of a variety of BACE1 inhibitors incorporating hydroxyethylamine and hydroxyethylene isosteres. The overall route is quite efficient, scalable and provides facile access to diverse inhibitors. We have converted epoxide 4 to BACE1 inhibitor 27. Further application of these epoxides in the synthesis of novel protease inhibitors is in progress in our laboratory.
Supplementary Material
Highlights.
Enantioselective syntheses of difluorinated aminoalkyl epoxides are reported.
The synthesis utilized Ti-enolate and boron-enolate asymmetric aldol reactions.
Aldol products were obtained with high diastereoselectivites and in good yields.
Curtius rearrangement of β-hydroxy acids installed the amine functionalities.
A fluorinated aminoalkyl oxirane was converted to a potent BACE1 inhibitor.
Acknowledgments
Financial support by the National Institutes of Health (GM53386) is gratefully acknowledged. We would also like to thank the Purdue University Center for Cancer Research, which supports the shared NMR and mass spectrometry facilities.
Footnotes
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Supplementary Data
Supplementary data associated with this article can be found in the online version.
References and notes
- 1.Ghosh AK, editor. Aspartic acid protease as therapeutic targets. Wiley-VCH; Weinheim, Germany: 2010. [Google Scholar]
- 2.Ghosh AK, Osswald HL, Prato G. J Med Chem. 2016;59:5172–5208. doi: 10.1021/acs.jmedchem.5b01697. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Ghosh AK, Cárdenas EL, Osswald HL. Top Med Chem. ASAP; 2016. [Google Scholar]; Ghosh AK, Tang J. ChemMedChem. 2015;10:1463–1466. doi: 10.1002/cmdc.201500216. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Ghosh AK, Osswald HL. Chem Soc Rev. 2014;43:6765–6813. doi: 10.1039/c3cs60460h. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Hardy J, Selkoe DJ. Science. 2002;297:353–356. doi: 10.1126/science.1072994. [DOI] [PubMed] [Google Scholar]
- 6.Selkoe DJ, Schenk D. Annu Rev Pharacol. 2003;43:545–584. doi: 10.1146/annurev.pharmtox.43.100901.140248. [DOI] [PubMed] [Google Scholar]
- 7.Citron M. Trends in Pharma Sci. 2004;25:92–97. doi: 10.1016/j.tips.2003.12.004. [DOI] [PubMed] [Google Scholar]
- 8.Iserloh U, Cummins JN. In: Peptidomimetic BACE1 Inhibitors for Treatment of Alzheimer’s Disease: design and evolution in Aspartic Acid proteases as Therapeutic Targets. Ghosh AK, editor. Wiley-VCH; 2010. pp. 441–479. [Google Scholar]
- 9.Cole CD, Bursavich M. Nonpeptide BACE1 inhibitors: Design and Synthesis. In: Ghosh AK, editor. Aspartic acid proteases as therapeutic targets. Wiley-VCH; 2010. pp. 481–509. [Google Scholar]
- 10.Ghosh AK, Brindisi M, Tang J. J Neurochem. 2012;120(Suppl 1):71–83. doi: 10.1111/j.1471-4159.2011.07476.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Vassar R, Kovacs DM, Yan R, Wong PCJ. Neurosci. 2009;29:12787–12794. doi: 10.1523/JNEUROSCI.3657-09.2009. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Vassar R. Alzheimer’s Res Ther. 2014;6:89. doi: 10.1186/s13195-014-0089-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Ghosh AK, Bilcer G, Hong L, Koelsch G, Tang J. Curr Alzheimer Res. 2007;4:418–422. doi: 10.2174/156720507781788864. [DOI] [PubMed] [Google Scholar]
- 14.Maillard MC, Hom RK, Benson TE, Moon JB, Mamo S, Bienkowski M, Tomasselli AG, Woods DD, Prince DB, Paddock DJ, Emmons TL, Tucker JA, Dappen MS, Brogley L, Thorsett ED, Jewett N, Sinha S, John V. J Med Chem. 2007;50:776–781. doi: 10.1021/jm061242y. [DOI] [PubMed] [Google Scholar]
- 15.Hom RK, Gailunas AF, Mamo S, Fang LY, Tung JS, Walker DE, Davis D, Thorsett ED, Jewett NE, Moon JB, John V. J Med Chem. 2004;47:158–164. doi: 10.1021/jm0304008. [DOI] [PubMed] [Google Scholar]
- 16.Greenlee W. J Med Res Rev. 1990;10:173–236. doi: 10.1002/med.2610100203. [DOI] [PubMed] [Google Scholar]
- 17.Ghosh AK, Bilcer G, Schiltz G. Synthesis. 2001;15:2203–2229. doi: 10.1055/s-2001-18434. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Evans DA, Takacs JM, McGee LR, Ennis MD, Mathre DJ, Bartroli J. Pure & Appl Chem. 1981;53:1109–1127. [Google Scholar]
- 19.Ghosh AK, Dawson Z. Synthesis. 2009;17:2992–3002. doi: 10.1055/s-0029-1216941. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Ghosh AK, Fidanze S. J Org Chem. 1998;63:6146–6152. doi: 10.1021/jo980159i. [DOI] [PubMed] [Google Scholar]
- 21.am Ende DJ, DeVries KM, Clifford PJ, Brenek SJ. Org Proc Res & Dev. 1998;2:382–392. [Google Scholar]
- 22.Ghosh AK, Onishi M. J Am Chem Soc. 1996;118:2527–2528. doi: 10.1021/ja9539148. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Akaji K, Teruya K, Aimoto S. J Org Chem. 2003;68:4755–4763. doi: 10.1021/jo030063a. [DOI] [PubMed] [Google Scholar]
- 24.Shi Z, et al. Bioorg Med Chem Lett. 2005;13:4200–4208. doi: 10.1016/j.bmc.2005.04.028. [DOI] [PubMed] [Google Scholar]
- 25.Jin YZ, et al. Green Chem. 2002;4:498–500. [Google Scholar]
- 26.Ghosh AK, Kumaragurubaran N, Hong L, Kulkarni SS, Xu X, Chang W, Weerasena V, Turner R, Koelsch G, Bilcer G, Tang J. J Med Chem. 2007;50:2399–2407. doi: 10.1021/jm061338s. [DOI] [PubMed] [Google Scholar]
- 27.Evans DA, Bartroli J, Shih TL. J Am Chem Soc. 1981;103:2127–2129. [Google Scholar]
- 28.Gage JR, Evans DA. Org Synth. 1990;68:83–91. [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
