Total Synthesis of Quinolizidine Alkaloid (−)-217A. Application of Iminoacetonitrile Cycloadditions in Organic Synthesis

Abstract

An intramolecular iminoacetonitrile [4 + 2] cycloaddition functions as the key step in an efficient total synthesis of the quinolizidine alkaloid (−)-217A.

The importance of substituted quinolizidines and indolizidines as synthetic targets is well established. The skeletons of a number of bioactive natural products incorporate these structures, and many of these compounds are available in very limited amounts from their natural source.¹ Highly toxic quinolizidine and indolizidine alkaloids isolated from the skin of poisonous amphibians have attracted much interest as research tools for neurophysiological investigations, and recently quinolizidine alkaloids obtained from marine sources have been identified as lead compounds for the development of anticancer, anti-inflammatory, and cardiovascular drugs. A number of ingenious methods have been developed in response to the synthetic challenge posed by these molecules, and these alkaloids have served as a popular testing ground for methods for the construction of pyrrolidines, piperidines, and various azabicyclic systems.²

Recently, we reported a strategy for the synthesis of azabicyclic compounds that should provide the basis for efficient new routes to a variety of substituted indolizidines and quinolizidines.³ As outlined in Scheme 1, this strategy involves the intramolecular [4 + 2] cycloaddition of a new class of activated imines,⁴ iminoacetonitriles. Iminoacetonitriles are conveniently prepared from alcohols by Mitsunobu coupling with TfNHCH₂CN followed by elimination of trifluoromethanesulfinate, and readily undergo intramolecular [4 + 2] cycloadditions to afford α-amino nitriles of general type 2.^3,5 The utility of α-amino nitriles as versatile intermediates for the synthesis of substituted nitrogen heterocycles is well documented.⁶ Metalation provides opportunities for alkylation and other carbon-carbon bond-forming processes, while exposure to Lewis acids furnishes iminium ions which can be intercepted with Grignard reagents (i.e., 2→4, the “Bruylants reaction”) or engaged in a variety of other useful carbon-carbon bond forming reactions. Stereoelectronic effects control the outcome of these transformations (vide infra), allowing for the stereoselective elaboration of cycloadducts as outlined in Scheme 1.^3,6

Scheme 1.

In order to further test and refine this methodology, we have undertaken the synthesis of several bioactive quinolizidine alkaloids such as clavepictine (5),^7,8 halichlorine (6),^9,10 and alkaloid 217A (7).¹¹ Herein we report the application of the iminoacetonitrile cycloaddition as a key step in the total synthesis of quinolizidine (−)-217A, an amphibian alkaloid isolated by Daly in 1993 from skin extracts of the Madagascan frog Mantella baroni. Previously, elegant total syntheses of quinolizidine 217A have been reported by the groups of Pearson and Panek.¹² Our goal was the development of an approach to the synthesis of quinolizidine 217A considerably more efficient than these earlier syntheses, and capable of supporting the preparation of significant quantities of the target alkaloid.

Scheme 2 outlines our retrosynthetic strategy, which features the intramolecular iminoacetonitrile cycloaddition 10→9 as a pivotal step. Alkylation of 9 would then be employed to install the enynylmethyl side chain, and stereoelectronic control in the subsequent reductive decyanation step was expected to deliver the desired stereochemistry at C-4. Control of the stereochemistry at C-1 would be established by epimerization of the ketone intermediate 8 derived from the silyl enol ether cycloadduct. In this first generation synthesis, we elected to employ resolution to provide access to the natural (−)-isomer as well as the unnatural isomer, deferring for future study the possibility of employing chiral Lewis acids to catalyze an asymmetric version of the cycloaddition. This decision was also influenced by the fact that for the synthesis of more complex targets such as 5 and 6, we envisioned strategies involving chiral cycloaddition substrates in which the stereogenic centers in the “connecting tether” would dictate the stereochemical outcome of the reaction.

Scheme 2.

Our first synthetic subgoal was the development of an efficient route to cycloaddition substrate 10. Based on our previous studies, we anticipated that 10 would be available from sulfonamide 16 by elimination of sulfinate on exposure to a weak base such as carbonate. Scheme 3 outlines our efficient 4-step route to 16. Mitsunobu coupling¹³ of commercially available 5-hexenol (11) with TfNHCH₂CN provided the expected sulfonamide, and ozonolysis then furnished aldehyde 13 in excellent yield. Wittig olefination of 13 using the acylphosphorane 14¹⁴ produced the desired (E)-α,β-unsaturated ketone 15 in 85–87% yield after purification by column chromatography.¹⁵ Finally, conversion to the desired enol ether was achieved using the general procedure of Dunogues et al.¹⁶ to afford 16 in excellent yield after purification by column chromatography on acetone-deactivated silica gel. Attempts to employ a cross-metathesis strategy¹⁷ for the conversion of 12 to 15 by reaction with 3-methyl-3-buten-2-one were frustrated by the homodimerization of 12, which produced a dimer unreactive in secondary metathesis.

Scheme 3.

As shown in Scheme 4, exposure of 16 to the action of cesium carbonate led to the elimination of trifluoromethanesulfinate and formation of iminoacetonitrile 10 as the expected mixture of E and Z imine isomers. The stereochemistry of this intermediate is not crucial, as we have previously demonstrated that iminoacetonitrile isomers interconvert under the conditions of the [4 + 2] cycloaddition.³ In the event, heating iminoacetonitrile 10 at 130 °C for 36 h produced the desired α-amino nitrile 9 in good yield. Addition of BHT was found to be beneficial in suppressing decomposition of the desired product. As expected, the isomer with an exo-oriented (axial) cyano group was isolated as the exclusive product of the reaction as a consequence of the “α-amino nitrile anomeric effect.”^6,18

Scheme 4.

For the next stage of the synthesis, alkylation of α-amino nitrile 9, we initially focused our attention on the enynylmethyl compounds 17, 18, 19, and 20.¹⁹ Surprisingly, the desired enyne 21 was obtained at best in only 30% overall yield after reductive decyanation (Scheme 5). Although alkylation with model alkylating agents such as allyl bromide proceeded smoothly, complex mixtures resulted from the reaction of 9 with enynylmethyl derivatives 17–20. Although we have been unable to characterize any of the byproducts of this reaction, we speculate that electron transfer to the enynylmethyl halide from the metalated nitrile (thus generating a capto-dative stabilized amino nitrile radical) may be complicating this alkylation.²⁰ We therefore turned our attention to a less unsaturated allylic halide, (Z)-3-bromo-1-chloropropene, with the idea of later elaborating the full enyne moiety via a Sonogashira coupling reaction.

Scheme 5.

In the event, we were pleased to find that alkylation of 9 with (Z)-3-bromo-1-chloropropene proceeded cleanly, and reductive decyanation of the crude alkylation product with sodium cyanoborohydride then afforded the desired quinolizidine 22 in 74–77% overall yield (Scheme 6). As predicted, axial delivery of hydride to the intermediate iminium ion leads to the formation of the desired diastereomer as the exclusive product of the reaction.²¹

Scheme 6.

Treatment of silyl enol ether 22 with 1.1 equiv of n-Bu₄NF in THF then generated ketone 23 as a single diastereomer with the C-1 methyl group in the desired equatorial orientation. Finally, resolution of quinolizidine (±)-23 using (R)-(−)-1,1′-binaphthyl-2,2′-diylphosphoric acid²² provided enantiomerically pure (−)-23 in 44% overall yield from the racemate.

The last stage of the synthesis, involving reductive excision of the carbonyl group and Sonogashira coupling, proved unexpectedly difficult. Initial attempts to effect deoxygenation of ketone 23 (as well as derivatives of the corresponding alcohol) were complicated by the formation of a byproduct ultimately identified as the tricylic amine 24.²³ Since this side product appeared to arise from cyclization of C-2 radical intermediates onto the vinyl chloride appendage, we focused our attention on strategies in which the reduction step could be carried out in the presence of efficient hydrogen atom transfer agents so as to more effectively intercept the intermediate radical prior to cyclization. Success was finally achieved by means of the one-pot protocol outlined in Scheme 7.

Scheme 7.

Thus, reduction of the tosylhydrazone derivative of (−)-23 with NaBH₃CN in the presence of excess t-butyl mercaptan completely suppressed the undesired radical cyclization and furnished vinyl chloride 26 in 63–66% overall yield. Sonogashira coupling with trimethylsilylacetylene proceeded smoothly, provided that the acetylene was added slowly to suppress competing alkyne dimerization. Finally, desilylation with K₂CO₃ in methanol afforded quinolizidine (−)-217A (7) ([α]²²_D -14 (c 0.8, CHCl₃), lit.¹² [α]²⁰_D -13.75 (c 0.4, CHCl₃)) with spectral characteristics identical with those reported for the natural product.^11,12

In summary, the intramolecular iminoacetonitrile [4 + 2] cycloaddition functions as a key step in an efficient assembly of the quinolizidine core of the amphibian alkaloid (−)-217A, enabling the total synthesis of this natural product in only 12 steps. The application of iminoacetonitrile cycloadditions in the synthesis of other bioactive alkaloids is under investigation.