Totally Synthetic by Paul H. Docherty, 23 June 2007
Total Synthesis of Isomigrastatin
I. J. Krauss, M. Mandal, S. J. Danishefsky, Angew. Chem. Int. Ed. 2007, 46, 5576-5579.
Yes, it’s another macrolactone, but this one’s a bit special, and there’s not an aldol in sight! Danishevsky and his group have been working with the Migrastatins for some time now, and have made several analogues, promoting that impressive cell-migration-inhibition activity. However, an isomer, the snappily and originally titled Isomigrastatin, has also been isolated, and Danishefsky quickly set to work.
I guess it’s no real surprise that one of the key reactions was a diastereoselective Diels-Alder reaction, and even less surprising that it was with a derivative of his diene, but I really like the synthesis of the dienophile, which comes from some previous work. It’s based on a protected tartaric acid, which was reduced and alkylated with vinyl grignard. The alcohol was then methylated and the acetonide removed, allowing cleavage of the diol to the desired aldehyde. Nice!
The hetero-Diels Alder went well with a spot of lewis-acid catalysis, and removal of the TMS group gave the desired pyrone. This was then reduced and treated with acid to effect a Ferrier rearrangement. The olefin was then epoxidised with mCPBA. This epoxy lactol was presumed to be suitable for a Wittig reaction; however, this was a difficult task, and they had to perform the olefination upon the open-chain aldehyde analogue (requiring reduction-protection-oxidation).
They then set about a conjugate addition to the unsaturated epoxide; work by James Marshall suggested that the stereocontrol for this reaction would be determined by attack anti to the epoxide. However, in this paper, Danishefsky found control from the C-15 silanol.
Thus, a distereoselective reduction of the C-15 ketone with CBS catalyst gave them this control element, and addition of the methylcuperate went with good yield (~13:1 d.r.).
Completion of the synthesis from this point required esterification to provide the lactone unit, and RCM. However, the final olefin in the macrocycle was still absent, and by design. The major decomposition products of the natural product remove the strain from the ring by expanding it and/or isomerising one of the E-olefins. Thus, by leaving this strain-building step to last, they hoped to keep it intact. However, to my eye, this was a calculated risk - they knew that the target was unstable, and this unsaturated ester motif was prone to isomerisation. Could they put it in?
Ultimately, they decided to used a selenide elimination to solve the problem - and the pheynyl selenyl group was already in place in the final fragment. Oxidation then cause the elimination - and (no doubt to some relief) a high yield of the target. Interestingly, they noticed that treatment of the target with isomerising conditions showed complete conversion to the Z-isomer, which says something about these reaction pathways.