Total Synthesis of (-)-Anaferine: A Further Ramification in a Diversity-Oriented Approach.

The piperidine ring is a widespread motif in several natural bioactive alkaloids of both vegetal and marine origin. In the last years, a diversity-oriented synthetic (DOS) approach, aimed at the generation of a library of piperidine-based derivatives, was developed in our research group, employing commercially available 2-piperidine ethanol as a versatile precursor. Here, we report the exploration of another ramification of our DOS approach, that led us to the stereoselective total synthesis of (–)-anaferine, a bis-piperidine alkaloid present in Withania somnifera extract. This natural product was obtained in 9% overall yield over 13 steps, starting from a key homoallylic alcohol previously synthesised in our laboratory. Therefore, the collection of piperidine-derivatives accessible from 2-piperidine ethanol was enriched with a new, diverse scaffold.


Introduction
Piperidine alkaloids are widespread natural products in the plant kingdom, but they are also present in the animal one, in particular in marine microorganisms [1]. They are historically known as potent poisons and teratogens: some well-known examples include coniine (from hemlock), lupinine (from lupin flour) and cytisine, contained in Laburnum spp. pods [2]. These effects are mainly the result of the nicotinic acetylcholine receptors (nAChRs) desensitisation [3]. However, piperidine alkaloids are characterised also by a wide range of interesting biological properties, including anthelminthic (e.g., isopelletierine), antioxidant and neuroprotective activities (e.g., lobeline) and the abilities of increasing drugs bioavailability and of contrasting multidrug resistance (e.g., piperine, lobeline) [1]. The structures of the mentioned alkaloids are depicted in Figure 1.
Also several bis-piperidine alkaloids, in which the two heterocycles are embedded in complex macrocyclic systems, are diffused in marine microorganisms. These compounds proved to possess antifungal, antimicrobial properties and cytotoxic activity against different cancer cell lines. Some examples belonging to this class are tetradehydrohalicyclamine A and B, 22-hydroxyhalicyclamine A, haliclonacyclamine A and F, arenosclerins D and E, madangamine F and neopetrosiamine A [6][7][8][9][10]. In this context, we realized that some of the cited natural products shared a common structural motif, constituted by a piperidine ring substituted in the 2-position, generating in this way a defined stereocenter (shown in pink in Figure 1).
These features seemed quite familiar to us, because they characterised a library of compounds previously synthesised in our laboratory. In fact, the importance of the piperidine ring in bioactive natural products, prompted us to devise a diversity-oriented approach (DOS), aimed at the generation of a diversified library of derivatives containing this key heterocycle. In this project, the ideal starting point to achieve structural diversification proved to be the 2-piperidine ethanol (1), a cheap and commercially available racemic product that incorporates the fundamental highlighted 2-substituted-piperidine moiety. Our library is composed by natural products (such as aloperine, coniine, boehmeriasin A and sedum alkaloids [11][12][13][14]), and synthetic derivatives [15][16][17][18], as depicted in Figure 2. We are currently trying to further enrich our library with new scaffolds. To this end, here we report the synthesis of (-)-anaferine starting from 1, as a further ramification of our DOS approach.
(-)-Anaferine is a bis-piperidine alkaloid, in which the two rings are connected through their C-2 atoms by a 2-propanone bridge. Being C2-symmetrical, this framework exists as two enantiomers and a meso-form. The diastereomeric mixture was isolated for the first time in 1962 [19] from the root extract of Withania somnifera, a plant widely exploited in the Ayurvedic medicine on the basis of several biological properties, including anti-inflammatory, anti-stress, immunomodulatory, adaptogenic, anticancer and neuroprotective activities [20]. ORD (optical rotary dispersion) analysis led to the determination of the absolute configuration of the two enantiomers: (-)-anaferine proved to be the (R,R)-enantiomer, and (+)-anaferine the (S,S)-one [21].
Most of the studies and the synthetic efforts on this topic, date back to the 70s, but in the last years this alkaloid has attracted once again the interest of the scientific community, thanks to in silico studies that suggested possible applications in the treatment of neurodegenerative diseases [22,23].
The last total synthesis of (-)-anaferine was accomplished in 2002 by Stapper and Blechert [24], while del Pozo et al. recently reported the preparation of the (+)-enantiomer [25] (Figure 3). Stapper and Blechert's strategy took advantage of an enantiopure cycloheptene precursor, accessed in five steps from the commercially available tropone. Then, after its conversion into the corresponding diamine, a tandem ring rearrangement metathesis, consisting in a ring-opening metathesis, followed by two ring-closing metatheses, afforded the fundamental carbon skeleton, that was transformed into the final product upon reduction of the double bonds, cleavage of the protecting groups and oxidation of the secondary alcohol to ketone ( Figure 3A).
On the other hand, del Pozo's approach involved the double use of N-sulfinyl amines, as both chiral auxiliaries and nucleophiles in an intramolecular aza-Michael reaction, as appreciable in Figure 3B. A bidirectional cross metathesis (CM), between the required sulfinyl amine and divinyl ketone gave an intermediate that underwent a double intramolecular aza-Michael reaction, forming the bis-piperidine core and generating at the same time the two stereocenters. Acidic cleavage of the sulfinyl groups led to (+)-anaferine synthesis completion.

Results and Discussion
The idea of enriching our DOS library with (-)-anaferine was suggested by compound 3 (Scheme 1A), previously obtained in our laboratory as an intermediate in the synthesis of the pironetin-dumetorine hybrids shown in Figure 2 [15]. Compound 3 is characterised by a piperidine core, deriving from the 2-piperidine ethanol, by a 2-methoxy propane bridge and by a homoallylic alcohol. The synthesis of 3, starting from 2-piperidine ethanol is summarized in Scheme 1A, and involves as key step the conversion of 1 into the homoallylic alcohol 2a. Notably, 2a was recently exploited also for the synthesis of an α,β-unsaturated lactam (compound 5), as appreciable again in Scheme 1A [18]. Therefore, comparing the two synthetic routes, it's possible to deduce the genesis of our synthetic plan. In fact, we reasoned that an analogue (8) of compound 3, bearing an O-silylether instead of the methoxy group, could have been easily obtained starting from 2a, taking advantage of the same synthetic strategy. The homoallylic alcohol of 8 could be then converted into lactam 12, exploiting the same approach employed for the synthesis of 5 from 2a. The synthesis of (-)-anaferine could be completed reducing the α,β-unsaturated lactam of 12 to the corresponding piperidine, and oxidizing the secondary alcohol to ketone, as depicted in the forward synthetic Scheme 1B. Of course, the synthesis of (-)-anaferine requires a strict control over the stereochemistry, in order to access the correct (R,R)-enantiomer. To this extent, two asymmetric Brown allylations were considered. The proposed synthetic strategy started from the homoallylic alcohol 2a, previously obtained in our research group from 1 in three steps, involving the protection of the piperidinic nitrogen, oxidation of the alcohol to aldehyde and its asymmetric allylation under Brown's conditions, leading to two diastereomeric alcohols 2a and 2b [18] (Scheme 1A). The relative anti-configuration of 2a was established by X-ray analysis, while the desired absolute (R,S)-configuration resulted from the use of the proper allylating agent enantiomer (in this case the (-)-B-allyl diisopinocampheylborane). In this way, 2a was obtained with an e.e. = 84% (Supplementary Information, Figure S1) and was then protected at the hydroxyl group as tert-butylsilyl ether (6).
Oxidative cleavage of the double bond afforded aldehyde 7, that was allylated again exploiting a Brown allylation. Also in this case, the stereocontrol was exerted mainly by the chirality of the Ipc (isopinocampheyl) substituents on the borane, rather than by the stereocenters already present in the substrate. The use of (-)-B-allyl diisopinocampheylborane, secured the formation of the desired stereoisomer of compound 8, that was accessed with a d.r.: 85:15. The diastereomeric ratio was determined through 1 H-NMR analysis. In fact, the two possible diastereomers 8 and 8b, characterized by an opposite configuration at the C-10 stereocenter, were independently synthesised. Their 1 H-NMR spectra in CDCl 3 were almost completely superimposable, rendering impossible the determination of the diastereomeric composition of the reaction mixture obtained under asymmetric conditions. However, we realized that using deuterated benzene a satisfactory signals separation for the two single diastereomers was achieved (Figure 4a, green and red spectra). In this way, once identified the signals of each diastereomers, the d.r. of the asymmetric reaction (blue spectrum in Figure 4a) was calculated on the basis of signals integrals, as reported in Figure 4b. With the correct stereoisomer of compound 8 in our hands, we proceeded with the following step, that was a Mitsunobu reaction in the presence of diphenylphosphorylazide, that converted the alcohol into the corresponding azide 9, with inversion of configuration. In this way, the correct (R)-configuration of C-10 stereocenter was accessed. Azide reduction under mild Staudinger conditions afforded homoallylic amine 10. Acylation with the commercially available acryloyl chloride, followed by a ring closing metathesis, led to the formation of the desired α,β-unsaturated lactam 12. As reported in our previous studies, Umicore M73 SIMes catalyst proved to be the most efficient to favour the ring enclosure [18,26]. In this way, the fundamental (-)-anaferine carbon skeleton, characterised by two nitrogen-containing hexacycles, connected by the 2-hydroxypropane linker and maintaining the (R,R)-configuration of the two key stereocenters, was secured. The next goal was the reduction of the α,β-unsaturated lactam into the corresponding piperidine. First of all, the double bond was selectively reduced through catalytic hydrogenation in the presence of Pd/C. Then, the amidic nitrogen of 13 was protected as Boc [27] and the piperidone was reduced to piperidine 15, using BH 3 ·SMe 2 . This mild reducing agent was preferred to the most common LiAlH 4 , to avoid the conversion of the two Boc protecting groups into methyl amines [28]. The TBS protecting group was selectively cleaved using TBAF, affording the known intermediate 16, reported in Blechert and Stapper's synthesis of (-)-anaferine [24].
The use of TBAF instead of acid conditions was fundamental to preserve the two Boc protecting groups: this was a mandatory requirement, because two basic nitrogen atoms would have interfered with the following oxidation of the hydroxyl group to ketone. Dess-Martin periodinane (DMP) was employed instead of pyridinium chlorochromate, that was the oxidizing agent of choice in the previously reported synthesis. DMP is a less toxic reagent, easier to handle and that gave, after a basic work-up, the desired product 17 in quantitative yield and without need of further purification. Finally, the cleavage of the two Boc protecting groups in acid environment, led to the desired (-)-anaferine dihydrochloride (18), whose identity was confirmed comparing analytical and spectroscopic properties with data reported in literature. The whole synthetic route is depicted in Scheme 2. In conclusion, 2-piperidine ethanol proved to be once again a flexible synthon for the obtainment of diversified piperidine-containing compounds. In this work, the collection of derivatives previously synthesised in our research group was enriched with (-)-anaferine, that was accessed as a new branching of our diversity-oriented approach, with an overall 9% yield over 13 steps. Notably, this strategy could open the possibility of affording interesting analogues of piperidine-based marine and plant-deriving natural products, characterised by a polyketide side chain in C-2 position.

General Remarks
Unless otherwise stated, reagents and solvents were purchased from Sigma Aldrich (St. Louis, MO, USA), Fluorochem (Zentek, Milano, Italy) or TCI (Zentek, Milano, Italy) and used without further purification. Unless otherwise stated, all reactions were carried out in oven-dried glassware and dry solvents, under nitrogen atmosphere and were monitored by thin layer chromatography (TLC) on silica gel (precoated 60F254 plates, Merck, (Darmstadt, Germany, with detection by UV light (254 nm) or by solutions of potassium permanganate stain, or ninhydrin, or p-anisaldehyde stain. Flash chromatography was performed using silica gel (240−400 mesh, Merck) as stationary phase. 1 H-NMR spectra were recorded on an Avance (400 MHz) spectrometer (Bruker, Billerica, MA, USA) and are reported relative to residual CDCl 3 or CD 3 OD. 13 C-NMR spectra were recorded on the same instruments (100 MHz) and are reported relative to residual CDCl 3 or CD 3 OD. All 1D and 2D NMR spectra (Bruker, Billerica, MA, USA) were collected using the standard pulse sequences available with Bruker Topspin 1.3. Chemical shifts (δ) for proton and carbon resonances are quoted in parts per million (ppm) relative to tetramethylsilane (TMS), used as an internal standard. Data for 1 H-NMR are reported as follows: chemical shift (δ/ppm) (multiplicity, coupling constant (Hz), integration).
Multiplicities are reported as follows: s = singlet, d = doublet, t = triplet, m = multiplet, bs = broad singlet. Data for 13 C-NMR are reported in terms of chemical shift (δ/ppm). Mass spectra were registered exploiting the electrospray ionisation (ESI) technique on a Q-Tof micro mass spectrometer (Waters, Milford, MA, USA).
Specific rotation values were measured on a P-1030 polarimeter (Jasco, Europe, Cremella, LC, Italy), using 1 mL cells, with path length of 10 cm. Measures were collected at 20-25 • C, using sodium D line wavelength λ = 589 nm.  (6) TBSCl (0.574 g, 3.81 mmol) was added to a solution of 2a (0.790 g, 2.93 mmol) and imidazole (0.400 g, 5.86 mmol) in anhydrous CH 2 Cl 2 (15 mL), previously cooled at 0 • C. The reaction mixture was then diluted with CH 2 Cl 2 and washed with brine. The aqueous layer was extracted with CH 2 Cl 2 and the collected organic phases were dried over Na 2 SO 4 , filtered and concentrated under vacuum. The crude product was purified by column chromatography on silica gel (hexane: EtOAc = 9:1), to give 6 (1.010 g, 90%) as a yellow oil. [