Development of CPA-Catalyzed β-Selective Reductive Amination of Cardenolides for the Synthesis and Biological Evaluation of Hydrolytically Stable Analogs

Abstract

This article describes the development of novel, hydrolytically stable cardiotonic steroid analogs featuring a 3β-amine moiety instead of the commonly found 3β-carbohydrates such as oleandrose. To establish the desired 3β-configuration stereoselectively, a new method based on chiral phosphoric acid-controlled diastereoselective reductive amination with Hantzsch esters was developed. This method utilizes readily available unsubstituted (S)-BINOL-based hydrogen phosphate as the catalyst, enabling the synthesis of 13 different 5β-androsterone and digitoxigenin analogs with up to 36:1 β:α diastereoselectivity. Additionally, this strategy was applied to generate two novel oleandrigenin analogs 15a and 15g in 3 steps from readily available gitoxigenin. The synthetic analogs were subjected to the NCI-60 human tumor cell lines screen, and several different digitoxigenin derivatives with tumor cell growth inhibitory power in submicromolar range were identified. The subsequent in vitro evaluation of digitoxigenin and oleandrin derivatives 13a, 13g, 15a, and 15g demonstrated that these four analogs reduced steady-state ATP1A1 levels in T98G cells in the 12-96 nM range. Interestingly, only the oleandrin analog 15g lowered also steady-state levels of the cellular prion protein (PrP^C), the main therapeutic target for the treatment of prion diseases.

Keywords: analog; anticancer activity; cardiotonic steroid; chiral phosphoric acid; diastereoselective; reductive amination.

Figure 1

A) Examples of 3β‐glycosides of digitoxigenin and oleandrigenin aglycones. B) Examples of bioactive digitoxin analogs with a hydrolytically stable 3β‐amine moiety. C) Existing approaches for the synthesis of 3β‐aminocardenolides. D) Diastereoselective CPA‐controlled reductive amination developed in this work.

Scheme 1

Exploring the scope of β‐selective reductive amination of 5β‐androsterone 7 with anilines^{[a] [a]}All reactions were run on 20 mg of 7, with catalyst (S)‐CPA 10j (10 mol%), Hantzsch ester (1.5 equiv), aniline (1.2 equiv), and 3 Å MS as 0.1 M solution in toluene at 50 °C for 14 hours.

Scheme 2

Exploring the scope of β‐selective reductive amination of 3‐oxo digitoxigenin 12 with anilines^{[a] [a]}All reactions were run on 20 mg of 12, with catalyst (S)‐CPA 10j (10 mol%), Hantzsch ester (1.5 equiv), aniline (1.2 equiv), and 3 Å MS as 0.1 M solution in toluene at 50 °C for 14 hours.

Figure 2

A) Oleandrigenin synthesis by Wicha and coworkers. B) Oleandrigenin synthesis by Nagorny and coworkers. C) Semisynthesis of oleandrin analogs from gitoxigenin presented in this work.

Scheme 3

Functionalization of oleandrigenin and synthesis of heterocycle‐containing analogs.

Scheme 4

Computational Exploration of Stereochemical Model (DFT, B3LYP/CC‐PVTZ[RCC‐PVTZ], THF).

Figure 3

Summary of results for the NCI‐60 Human Cancer Cell Line five‐dose screen of compounds 11c and 13a–f. Each compound was evaluated against the specified cancer cell lines using five different doses (10⁻⁴, 10⁻⁵, 10⁻⁶, 10⁻⁷, and 10⁻⁸ M concentrations) that were normalized to control wells. The compounds 13a, 13d, 13e, and 13f were assayed twice while the five doses screen of compounds 11c, 13b, and 13c was performed once. The GI₅₀/TGI/LC₅₀ endpoint values provided are derived from one of the runs and are based on the linear interpolation of growth values above and below the endpoint value. The mean optical densities and the GI₅₀/TGI/LC₅₀ for all experiments are provided in the Supporting Information.

Figure 4

Western blot analysis of concentration dependent steady‐state reduction of PrP levels in T98G cells. T98G cells were exposed to DMSO vehicle solution or compounds 13a, 13 g, 15a, 15 g in DMSO at indicated concentrations for the total duration of 6 days. Following the treatment, the cells were harvested, and cellular proteins were extracted with 0.5% NP‐40, 0.5% deoxycholate, 100 mM Tris/HCl buffer (pH = 8.3), and Complete Protease inhibitor cocktail. The protein concentrations were determined by bicinchoninic acid (BCA) assay and adjusted. The adjustment of total protein levels was validated by Western blot analyses of beta actin (ACTB). Western blot analysis of 20 µg of total protein per lane documented the anticipated concentration‐dependent reduction of steady‐state ATP1A1 for all compounds indicating target engagement. The quantitation of ATP1A1 signals was obtained by first normalizing ATP1A1 signals against signals from the ACTB loading control, followed by assigning a nominal intensity level of 1.00 to the ATP1A1 level of the mock‐treated sample in each compound concentration series. A) Western Blot analysis of oleandrin analog 15 g and digitoxigenin analog 13 g. Both compounds demonstrate the reduction in ATP1A1 levels; however, only oleandrin derivative 15 g caused a concentration‐dependent reduction in steady‐state PrP levels. B) Western blot analysis of oleandrin analog 15a and digitoxigenin analog 13a. Both compounds demonstrate the reduction in ATP1A1 levels; however, no reduction in PrP^C levels was observed. Oleandrin analog 15a was toxic at the 96 nM concentration tested, and the use of 12‐well or 10‐well sodium dodecyl sulfate polyacrylamide gel electrophoresis system Western blots prompted the omission of individual concentrations tested. C) Control panel documenting the exceptional specificity of the 3F4 antibody for human PrP^C, a protein giving rise to a complex band pattern due to its post‐translational modification through the attachment of up to two complex N‐glycan chains. The panel depicts extracts of wild‐type and PRNP ^−/− ReN VM cells as a stand‐in for T98G for this purpose because prion gene‐ablated T98G cells were not available. The specificity of the ab7671 anti‐ATP1A1 antibody was recently verified by us using an siRNA targeting ATP1A1 (S1 Figure, Panel B, cf. Ref. [39]).