|Home | About | Journals | Submit | Contact Us | Français|
Lycoperine A was synthesized through a highly convergent route in which a double alkylation of 2,6-dicyano-N-benzylpiperidine with the octahydroquinoline moiety gave the lycoperine skeleton. The octahydroquinoline was prepared by a desymmetrization reaction of 5-methylcyclohexane-1,3-dione. Hydrolysis, reductive amination and cyclization gave lycoperine A in 13 steps and 3% overall yield. The absolute configuration of lycoperine A was assigned as 6R, 6′R, 8R, 8′R, 13S, 17R.
Lycopodium alkaloids are an important class of natural products due to their diversity and their varied biological activities.i Several of these alkaloids inhibit acetylcholinesterase (AChE), which is responsible for the breakdown of the neurotransmitter acetylcholine. After the discovery of this significant biological activity around 1986ii there was a surge of interest in Lycopodium alkaloids, and subsequently numerous new alkaloids in this class were discovered and characterized.i In the time span from 1993 to 2004, 81 new Lycopodium alkaloids had been reported.1 Of the Lycopodium alkaloids discovered in the 1980’s, huperzine A (HupA) demonstrated the greatest inhibition of acetylcholinesterase and its synthesis was first reported by Qian and Ji in 1989.iii Recent studies subjecting rats to HupA have demonstrated increased efficiency in learning and memory and have been considered potential lead compounds for the treatment of Alzheimer’s disease.iv
Lycopodium alkaloids have also attracted significant interest from synthetic chemists due to their unique and complex structures. The four classes of Lycopodium alkaloids are the phelgmarine, lycodine, lycopodine, and fawcettimine (Figure 1).i,v,vi Lycopodine was the first of this class discovered in 1881 by Bodecker.vii Phlegmarineviii is a representative of the most recently discovered class, described as the “miscellaneous group”.v They are structurally related through their common quinolizine, quinoline, pyridine, or α-pyridone ring systems. They generally contain 16 carbons or as dimers 32 carbons.
Lycoperine A (1) was recently isolated from the club moss Lycopodium hamiltonii in Japan by Kobayashi and co-workers from Hokkaido University.ix This compound exhibited a moderate inhibitory activity against acetylcholine esterase from bovine erythrocyte. The structure and the relative stereochemistry of lycoperine A were determined by detailed 2D NMR studies after conversion to the N-diethyl analogue by the reduction of the acetamide moieties, but its absolute configuration was not determined. This compound has a characteristic pseudosymmetric structure in which two identical octahydroquinoline rings are linked to a central 2,6-cis-disubstituted piperidine ring. We report here the total synthesis of lycoperine A and its absolute configuration.
Our retrosynthetic analysis for Lycoperine A is shown in Scheme 1. The plan called for the generation of Lycoperine A by reductive decyanation of the dinitrile 2 and is based on precedent from Husson’s program.x The dinitrile 2 would arise from a double alkylation of 2,6-dicyano-N-benzylpiperidine (3) with the octahydroquinoline 4. The synthesis of 4 was envisioned to come from elaboration of vinylogous amide 5, which itself would be prepared by a desymmetrization reaction. The starting material for the synthesis was 5-methylcyclohexane-1,3-dione (6). The configuration of the natural product would be derived from a chiral amino alcohol auxiliary.
Preparation of the octahydroquinoline segment was based on precedent from Lhommet and from Bosch.xi,xii Diketone 6 was combined with acrolein to produce hemiacetal 7, a cyclic version of the underlying symmetric dione. Desymmetrization by condensation with amino alcohol auxiliaries 8 and 9 led to the formation of vinylogous amides 10 and 11, respectively (Table 1). The condensation with 8 was efficient using several different Lewis acids, but the desymmetrization to set the C8 methyl center resulted in very modest selectivities (entries 1–3). Switching to the amino alcohol 9 led to much more effective desymmetrization, with ca. 95:5 dr at the C8 stereogenic center. Copper triflate was the best catalyst, and 3 mol% of the catalyst was effective (entry 5). Minor diastereomers were produced with both auxiliaries, but in the case of 9 they could be separated on silica gel chromatography to deliver the vinylogous amide 11 in 96% yield. The relative configuration of compound 11 was confirmed by X-ray crystallography.xiii This unusual desymmetrization strategyxiid allowed the C8 stereogenic center to be introduced efficiently from a racemic precursor.
Synthesis of the octahydroquinoline 4 is presented in Scheme 2. Optimized conditions for the desymmetrization led to vinylogous amide 11 with excellent diastereoselectivity. Attempts to reduce the tetrasubstituted double bond by hydrogenation were unsuccessful, and so we turned to dissolving metal conditions. Based on Palmieri’s precedent, we investigated sodium and isopropanol.xiv Optimized conditions delivered the amine 12 in 67% yield with >95:5 dr. The relative configuration of amine 12 was established by x-ray crystallography.xiii Hydrogenolysis of the auxiliary was accomplished under forcing conditions with 20% Pd(OH)2/C in acidic methanol. The resulting amine was directly acylated to give an acetamide in 70% yield over two steps. Oxidation with Dess-Martin periodinane produced the crystalline ketone 13 as a single diastereomer. The relative configuration of 13 was confirmed by X-ray crystallography.xiii The octahydroquinoline 13 was assembled in six steps and 27% overall yield from commercially available diketone 6.
The next goal was to introduce the tetrasubstituted double bond and a one-carbon substituent onto ketone 13. A thermodynamic enolate is required, and a number of conditions were investigated to generate it. KHMDS and PhNTf2 gave predominantly the less substituted enol triflate.xv Potassium tert-butoxide led to more of the desired enol triflate, but it was accompanied by epimerization adjacent to nitrogen, presumably due to a reversible Michael addition of the amide. Selection of the solvent and reaction temperature turned out to be important, and going from THF to 2:1 DMF/THF gave predominantly the more hindered enol triflate 14 with no detectable epimerization adjacent to nitrogen.xvi Stille carbonylation was uneventful, and reduction delivered the allylic alcohol 15.xvii Generation of the alkylating agent 4 used standard PPh3 and CBr4 conditions.xviii Allyl bromide 4 was set up to couple with the central piperidine ring.
The original plan called for the introduction of the two identical octahydroquinoline segments by alkylation of a piperidine dinitrile 3, followed by reductive decyanation and deprotection. A single reductive decyanation has ample precedent from Husson’s piperidine assembly.x Model studies for the double alkylation and reductive decyanation are outlined in Scheme 3. Dinitrile 3 was prepared from glutaraldehyde following Husson’s procedure.xix After some optimization, the double alkylation using just 2.2 equiv of allyl bromide worked extremely well, delivering predominantly the cis-16 diastereomer in >90% yield.xx Surprisingly, the reductive decyanation was not straightforward. Numerous attempts to cleave the nitriles did not produce the expected cis-2,6-disubstituted piperidine (17 cis), but rather led to complex mixtures of products. A representative example is shown in Scheme 3, where the expected decyanation product 17 was produced in only 4% yield as a mixture of stereoisomers. Reductive coupling of the two nitriles generated diketone 18, and addition of lithium amide and cyclization led to compound 19. Clearly the reduction does not proceed one nitrile at a time, but rather the two nitriles interact and generate a variety of side products. The alkylation reaction was very effective, but the reductive decyanation was an utter failure.
An alternative approach to the central piperidine ring would be a reductive amination of a 1,5-diketone. A number of different routes to the desired 1,5-diketone 20 can be envisioned, but we decided to build upon the successful alkylation reaction in Scheme 3. The double alkylation was accomplished in good yield by mixing 1.0 equivalent of 3 with 2.2 equivalents bromide 4 in THF/DMPU and adding an excess of LiHMDS at reduced temperature. The dinitrile piperidine 2 was isolated in good yield. Deprotection to give the diketone 20 was initially attempted by treatment with HCl in MeOH, but the conditions led to subsequent alkene isomerization. Using AgNO3 as a Lewis acid to promote the hydrolysis gave the desired diketone 20 in good overall yield.xxi
Reductive amination was first attempted using sodium cyanoborohydride as described by Abe and co-workersxxii but the reaction was very slow and provided a mixture of the desired product and an undesired side product, the tetrahydropyran ring corresponding to piperidine 1, which arose from reduction of the ketone and cyclization. Ketone reduction could be avoided by using the more selective reducing agent, sodium triacetoxyborohydride.xxiii The reductive amination was still slow, gave incomplete conversion, and produced 28% yield of the piperidine 1 along with 50% of recovered diketone 20. The NMR spectra and other physical data for piperidine 1 matched that reported for lycoperine A, including optical rotation.xxiv LAH reduction of diamide 1 gave the diethyl compound, and its spectra compared favorably with that reported by Kobayashi’s.xxv
Synthesis of lycoperine A was accomplished using a desymmetrization reaction that led to an efficient synthesis of the octahydroquinoline 4. A double alkylation reaction and subsequent piperidine ring formation completed the synthesis and allowed the configuration of the natural product to be assigned as 6R, 6′R, 8R, 8′R, 13S, 17R.
This work was supported by a gift from the Schering-Plough Research Institute and by the S. T. Li foundation. Financial support was also provided by Daiichi-Sankyo (Y. N.) and the Japan Society for the Promotion of Science (S. K.).