Total Synthesis, Stereochemical Revision, and Biological Reassessment of Mandelalide A: Chemical Mimicry of Intrafamily Relationships

Mandelalide A and three congeners had recently been isolated as the supposedly highly cytotoxic principles of an ascidian collected off the South African coastline. Since these compounds are hardly available from the natural source, a concise synthesis route was developed, targeting structure 1 as the purported representation of mandelalide A. The sequence involves an iridium-catalyzed two-directional Krische allylation and a cobalt-catalyzed carbonylative epoxide opening as entry points for the preparation of the major building blocks. The final stages feature the first implementation of terminal acetylene metathesis into natural product total synthesis, which is remarkable in that this class of substrates had been beyond the reach of alkyne metathesis for decades. Synthetic 1, however, proved not to be identical with the natural product. In an attempt to clarify this issue, NMR spectra were simulated for 20 conceivable diastereomers by using DFT followed by DP4 analysis; however, this did not provide a reliable assignment either. The puzzle was ultimately solved by the preparation of three diastereomers, of which compound 6 proved identical with mandelalide A in all analytical and spectroscopic regards. As the entire “northern sector” about the tetrahydrofuran ring in 6 shows the opposite configuration of what had originally been assigned, it is highly likely that the stereostructures of the sister compounds mandelalides B–D must be corrected analogously; we propose that these natural products are accurately represented by structures 68–70. In an attempt to prove this reassignment, an entry into mandelalides C and D was sought by subjecting an advanced intermediate of the synthesis of 6 to a largely unprecedented intramolecular Morita–Baylis–Hillman reaction, which furnished the γ-lactone derivative 74 as a mixture of diastereomers. Whereas (24R)-74 was amenable to a hydroxyl-directed dihydroxylation by using OsO₄/TMEDA as the reagent, the sister compound (24S)-74 did not follow a directed path but simply obeyed Kishi’s rule; only this unexpected escape precluded the preparation of mandelalides C and D by this route. A combined spectroscopic and computational (DFT) study showed that the reasons for this strikingly different behavior of the two diastereomers of 74 are rooted in their conformational peculiarities. This aspect apart, our results show that the OsO₄/TMEDA complex reacts preferentially with electron deficient double bonds even if other alkenes are present that are more electron rich and less encumbered. Finally, in a brief biological survey authentic mandelalide A (6) was found to exhibit appreciable cytotoxicity only against one out of three tested human cancer cell lines and all synthetic congeners were hardly active. No significant fungicidal properties were observed