Many a synthetic campaign begins with a keystone idea, in which the skeleton of a natural product swiftly flips into being. Clearly, taking that brainwave directly to the fumehood and cracking on with the synthesis is risky in the extreme - especially if the key step lies towards the end. A common strategy is to model that step; generally stripping the substrate back to an undecorated core and optimising the conditions. But the draw of the ’model study’ has unfortunately left the field of total synthesis strewn with key step conundrums - where the perfectly modelled step has suddenly turned feral. Simplifying the substrate inevitably changes how the chemistry behaves, so striking the balance of complexity and ease of synthesis can be tricky.

Those issues aside, the power of modelling is too large to ignore, so teams like that led by Paul Wender of Stanford University in California, US, often use it to narrow the odds of a long synthesis.1 The case in point is that of yuanhuapin, a fabulously complex member of the daphnane diterpene orthoester class of natural products, bearing an astonishing twelve contiguous stereogenic centres around its seven rings (look closely!).2 A key feature to the team is the epoxide nestling on the cycloheptane ring, a feature the group was keen to install in the late stages of the synthesis, working from the appropriate cycloheptene. However, the group still had to find a strategy to install the remaining eight oxygenated positions.

The team made a great start by working from a derivative of tartaric acid, giving them a pair of stereocentres and a diol for free. The group converted the two esters at either end of the molecule to allyl alcohols, broke the symmetry by protecting just one of the two hydroxyls and then coupled on a pyranone fragment. Of course, all this unsaturation is just desperate to perform the chemical equivalent of a song and dance routine, which Wender’s team encouraged by heating briefly in a microwave reactor. The resulting Claisen rearrangement was followed swiftly by a oxidopyrylium [5+2] cycloaddition, generating two rings in a single step, and increasing the stereocentre count by three - all in fantastic yield (figure 1).


Figure 1

Functionalising this product through further allylation and alkylation allowed the group to yet again decorate their intermediate with an array of unsaturated functionality. In a more delicate and modern protocol, the group again encouraged ring formation, this time using palladium catalysis to persuade reaction (figure 2). Having completed all three of the carbocyclic rings, the group then turned their attention to the heterocyclic functionality. However, with eight hydroxyl groups, they required both patience and finesse with their protecting group strategy. A series of subtle tweaks ultimately completed a key intermediate for this extended molecular family.


Figure 2

So near, and yet so far 

Tantalisingly close to the target, the team must have been frustrated by the number of steps required for relatively small transformations. Indeed, reductive removal of an alcohol and elimination of another, along with formation of the unusual orthoester moiety required 19 steps. But using a less robust protecting group strategy would most likely have ended in disaster, so their effort should be applauded. With all other functionality in place, the team needed only to install the remaining epoxide, and it’s here that the model studies let them down. After proving that the inherent substrate facial selectivity bias could be overcome on a smaller and simpler seven membered ring, their learnings simply didn’t transfer to their key substrate, delivering the oxygen to the opposite face of the alkene (figure 3). And as they were at the final step of their synthetic campaign, lack of material meant that their work came to a pause.  


Figure 3

The team may well have been disappointed that they didn’t quite get to the named target, but their groundbreaking strategy is much more valuable than just one target. More important was the synthesis of the ’gateway precursor’, granting access to the whole family. 

Paul Docherty is a science writer and blogger based in Reading, UK