(–)-Scabrolide B (and friends)

In the early decades of the field, total synthesis was inextricably linked with structural elucidation. Long before chemists were making targets for fun, or to showcase their new reactions, they were making them because they had no choice. Textbooks talk endlessly about classic syntheses like Marshall Gates’s mastery of morphine or R B Woodward’s conquest of strychnine, but what’s often elided are the years-long arguments about the structures of these targets. In those days, degradation, derivatisation and reconstitution demanded grams of material, and the final ‘proof’ always came from synthesis. Now, in the modern age of high-field NMR, DFT calculations and tandem mass spectrometry, structure elucidation is often carried out by specialised (non-synthetic) research groups and rarely involves anything as messy and imprecise as bench chemistry.

Yet, a reader of the current natural products literature will see that mistakes over structure still occur relatively frequently. Even the gold-standard technique of x-ray crystallography is not infallible. Indeed, several important natural products such as diazonamide A and the kinamycins were famously misassigned despite crystallographic data (unluckily, my PhD advisor had the misfortune to work on both!). And even today, the determination of absolute configuration can still be a thorny problem, especially for natural products that are not crystalline.

In 2022, Alois Fürstner and coworkers at the Max Planck Institute in Germany completed the second total synthesis of the interesting natural product scabrolide A. Unlike previous routes to the target, Fürstner’s was based on its putative biosynthesis and proceeded via scabrolide B. The group’s biomimetic endgame was a clever strategy as the two targets were thought to be isomers differing only in the position of a double bond. But surprisingly, while the data for synthetic scabrolide A matched previous reports, the data for scabrolide B did not!

In the aftermath, the team’s own detective work, along with reisolation and reassignment by another group, led to a new structure that they set out to confirm experimentally.¹ The route starts with nor-carvone (look closely – it’s missing a methyl) along with a densely functionalised vinyl iodide that contains most of the molecule’s stereogenic centers (figure 1). Surprisingly, it’s the latter of these that’s chiral-pool derived, but neither is as simple as it appears; preparing both takes 16 steps between them. With the legwork done, the now-central seven-membered ring is quickly assembled through a Mukaiyama–Michael reaction followed up by a palladium-catalysed α-vinylation-isomerisation of the ketone. This very rapidly puts together the molecule’s carbon skeleton, with just a few oxidation-state adjustments to make. Hydroxylation proves tough, but the team is eventually able to get the atoms installed, albeit in the wrong place, next to the ketone. Fortunately, transposition is accomplished using catalytic rhenium (an element that’s still unchecked on my own periodic table reaction bingo card).

Allylic oxidation of the new alcohol with manganese dioxide completes scabrolide B, confirming the corrected structure and giving the team the chance to further probe possible biosynthetic relationships within the family. Simple dehydration completes sinuscalide C but treatment with base is more dramatic, initially producing ineleganolide via epimerisation and oxa-Michael. However, the team noticed that this new natural product appears to be metastable under these conditions, slowly converting to yet another natural product, horiolide (figure 2). Interestingly, while the conversion of ineleganolide into horiolide does fit with the proposed biosynthesis, the former was not thought to be formed from scabrolide B, again underscoring the importance of doing the labwork!

I have really enjoyed following this story over the last couple of years and congratulations to the Fürstner group for finally confirming the structures and relationships between these targets.