(5R,8S,8aS)-(-)-8-methyl-5-pentyloctahydroindolizine

Background: Prior work from these laboratories has centred on the development of enaminones as versatile intermediates for the synthesis of alkaloids and other nitrogen-containing heterocycles. In this paper we describe the enantioselective synthesis of indolizidine and quinolizidine analogues of bicyclic amphibian alkaloids via pyrrolidinylidene-and piperidinylidene-containing enaminones.

At this stage, however, our fears of the discrepant behaviour of five-and six-membered enaminones proved to be all too well founded. In the indolizidine series, the robust enaminone 17 survived reduction with lithium aluminium hydride, leaving only the saturated ester to be reduced. With the six-membered analogue 28, the enaminone unit was far more susceptible to reduction, and despite many attempts to modify conditions, over-reduction led to a plethora of basic products that could neither be separated nor properly characterised. Although the desired alcohol (+)-29 containing an intact enaminone system could be isolated on occasion, the best yield obtained was 29% when the reaction was not allowed to go to completion. Thus a change of strategy was required to produce 29, the pivotal intermediate from which the quinolizidine nucleus needs to be constructed.
The reduction of the tert-butyl ester clearly needed to be performed at an early stage of the synthesis before the introduction of other incompatible functional groups (lactam, thiolactam, enaminone). The only feasible option was to go back to the chiral amine (+)-14, reduction of which with lithium aluminium hydride gave the unstable amino alcohol (+)-30 in 97% yield as long as the amine was added slowly to a stirred suspension of the hydride in diethyl ether (Scheme 4). If the order of addition were reversed, the best yield obtained was 48%. The amino alcohol was protected as its tert-butyl(dimethyl)silyl ether (-)-31 (99%) before hydrogenolysis of the benzyl groups over Pearlman's catalyst in glacial acetic acid gave the free amine (-)-32 in quantitative yield. Treatment with 5-bromopentanoyl chloride as described above afforded the unstable bromoamide 33 as an orange oil in 89% yield. In this case, cyclisation of the crude intermediate to the lactam (+)-341 was most successfully effected by adding potassium tert-butoxide to a solution of the bromoamide in dry tetrahydrofuran at room temperature, a yield of 81% being obtained by keeping the reaction time short (25 min). To our dismay, however, the attempted thionation of 34 with Lawesson's reagent under a variety of conditions was uniformly unsuccessful, apparently because the silyl ether failed to survive the reaction conditions.
Inelegant though it was, we were forced at this stage to change protecting groups on the alcohol. Fortunately, the drop in yield was not too serious when desilylation of 34 with aqueous hydrofluoric acid to give the free alcohol (+)-35 was followed by acetylation with acetic anhydride in pyridine (Scheme 5). The lactam (+)-36, obtained in an overall yield of 89%, was then successfully thionated with Lawesson's reagent in boiling toluene to give the thiol- iii 92% iv, v 75% vi 29% HO actam (+)-37 in 94% yield. Finally, reaction with ethyl bromoacetate followed by treatment with triphenylphosphine and triethylamine in acetonitrile give the vinylogous urethane (+)-38 in 80% yield. Hydrolysis of the acetate with potassium carbonate in methanol then afforded the pivotal alcohol (+)-29 (70%). The scene was now set for cyclisation to the quinolizidine system. Immediate conversion of the unstable free alcohol into the corresponding iodide with iodine, triphenylphosphine and imidazole in a mixture of toluene and acetonitrile [29] and heating the reaction mixture under reflux gave the desired 3,4,6,7,8,9-hexahydro-2H-quinolizine-1-carboxylate (-)-39 in 70% yield. In order to introduce the remaining stereogenic centres of the target system, the alkene bond of the bicyclic vinylogous urethane 39 needs to be reduced stereoselectively. Based on our previous success with the indolizidine analogue 19, we opted for catalytic hydrogenation, which is expected to produce not only a cis-relationship between C-1 and C-9a, but also a cis-relationship between C-4 and C-9a. The developing chair conformation of the six-membered ring in the transition state should result in an equatorial preference for the pentyl side chain, which in turn should bias the approach of the reductant towards the more remote face of the double bond. Gratifyingly, hydrogenation of intermediate 39 over platinum oxide catalyst in ethanol at a pressure of five atmospheres produced the quinolizidine (-)-40 as a single diastereomer in 97% yield. The diastereoselectivity is manifestly better than in the indolizidine case. Support for the cis-relationship of the hydrogen atoms at positions C-4 and C-9a and the trans-ring junction in the product was once again provided by Bohlmann bands in the FTIR spectrum at ca. 2790 cm -1 . However, further confirmation of the relative stereochemistry by consideration of the 1 H NMR spectrum was not feasible because overlap of signals prevented the extraction of coupling constants for 1-H and 9a-H.
Finally, reduction of the ester to the primary alcohol (-)-41 was accomplished in moderate yield (65%) with lithium aluminium hydride. Again, coupling constants could not be determined for 1-H and 9a-H. In this case, however, there is good precedent for assigning the relative stereochemistry of the hydroxymethyl substituent at C-1 on the basis of 13 C chemical shifts. For example, the chemical shift of C-1 in lupinine 12, which possesses an axial hydroxymethyl substituent, is 38.8 ppm; whereas the corresponding chemical shift in epilupinine 13, the equatorial hydroxymethyl epimer, is 43.8 ppm. [30] The chemical shift difference of about 5 ppm between the C-1 equatorial and axial hydroxymethyl epimers appears to be general for quinolizidines. [31] A similar effect has been reported for 8-hydroxymethylindolizidine epimers, for which the chemical shift difference is even larger (ca 10 ppm). [24] In the present case, the observed chemical shift of 38.4 ppm for 41 is consistent with an axial disposition of the C-1 substituent, and thus with the expected cishydrogenation of 39.
While it would have been desirable to conclude this investigation by preparing (1S,4R,9aS)-4-pentyloctahydro-2Hquinolizine 42, the ring homologue of 8-epi-indolizidine 209B, this target eluded us. Attempts to reduce the corresponding methanesulfonate of 41 with Raney nickel in boiling ethanol gave ambiguous results no matter how we modified the reaction conditions.

Conclusion
Few approaches to 1,4-cis-disubstituted quinolizidines and 5,8-cis-disubstituted indolizidines of amphibian origin have been reported in the literature. Because the route we have devised proceeds through bicyclic enaminone intermediates in which the alkene bond is located between the bridgehead position and the adjacent site, we have a convenient and dependable method for introducing the correct relative stereochemistry at these two sites by means of catalytic hydrogenation. However, the differences in behaviour of pyrrolidinylidene-and piperidinylidene-containing enaminones that we have come to expect