Stoltz, Liu and Ferreira. JOC, 2007, ASAP. DOI: 10.1021/jo0710883.
An awesome read, this paper is a smart bit of work by the Stoltz group, looking at the synthesis of two members of the Cephalotaxus alkaloid family. They state that the members accessed in this route are biologically inactive, but point out that other members of the family are a bit more pokey (and can presumably be constructed using similar means.
The synthesis begins with one of my favourite reactions, a Johnson-Claisen rearrangement. This uses triethyl ortho acetate (which smells like sweaty shoes) to add two carbons to the allylic alcohol, and then does the [3,3]. I’ve done several, and it’s a top reaction (and quite amenable to microwave chemistry…).
Some straight-forward transformations then give them either the primary amide or the amine, which under aerobic conditions, does an oxidative cyclisation to give the 5,5 fused spiro compound in rather decent yield. I’m especially impressed with the amide case, as I’d imagine that system to be rather stubborn. However, the reaction is not enantioselective, and thus generates a racemic mixture. They state, however, that this inability is useful in this synthesis…
A few more steps, and it’s palladium time again (this time using a rather exotic catalyst). This heck reaction builds that seven member ring rather nicely, completing the carbocyclic skeleton of the natural product, except in a racemic fashion.
However, using a derivative of mandelic acid as the starting material, they were able to do a separation of diastereomers. The stereogenic hydroxyl group that was key to this was then either burned-off to make either enantiomer of celphalotaxine, or used to form the acetal linkage in Drupacine.
There’s lots to read here, and I’d be interested in your opinions about this ideology (using both diastereomers from an unselective reaction to pursue two different targets/enantiomers…)