Li, Deng, Zhu, Lu, Yu. JACS, 2011, ASAP. DOI: 10.1038/nchem.1196
Just take a quick look at that target and consider that the synthetic route I’m about to summarise took thirteen steps. I’m fairly stunned – I read the top-line number in the abstract, and immediately thought that they must have started with an advanced intermediate or degredation product, but no – the synthetic action (like many natural product syntheses) begins with Citronellal. Sure, only three of the rings are carbocyclic, but there’s a lot going on here, so lets get into it.
As I said, the synthetic starting material is a fairly large pot of tasty-smelling citronellal. It’s not exactly cheap, (about £100 for 5g from S-A), there are more expensive entries to this terpenoid skeleton. They spend the first four steps (which they somewhat cheekily don’t count as it’s all known chemistry) mashing the isoprene group into a protected hydroxyl group, and then do a pair of Horner-Wadsworth-Emmons reactions to firstly build the triene system, and latterly bolt-on the 1,3-dicarbonyl moiety. Interestingly, you’ll note the thioester group – this was to make the prior carbonyl chemistry a bit less dramatic, and make the subsequent chemistry very neat.
A little cooling and addition of everyone’s favourite Lewis acid with weird punctuation resulted in a very neat intramolecular Diels-Alder, building four stereocenters, three tertiary and one quaternary in a yield anyone would be happy with.
Whilst that is a very neat reaction, Diels-Alder cycloadditions are so 19th-century; what happens has to be at least early 20th century! The 1,3-dicarbonyl moiety is, of course, quite acidic, so an allylic displacement reaction could be encouraged by addition of Ag[TFA] and a little base (transesterification conditions). The reaction initially occurs by O-alkylation, forming a furan system, but the equilibrium of C-alkylation to O-alkylation could be transformed by treatment with Pd(OAc)2 and electron-rich ligand PBu3. I’m guessing a little, but I suppose that formation of a Pd-[allyl] intermediate allows this facile interconversion to occur.
Okay, the three carbocyclic rings are in, but we’ve still got two heterocycles to build. The teams next step towards that was formation of an amide by displacement of the fluorinated ester chain with a derivative of D-serine. The remaining terminal olefin was oxidised (after a little experimentation) under Wacker conditions, and the carbonyl reduced with some selectivity to the secondary alcohol. Lastly, this final intermediate was treated with sodium methoxide, promoting a Dieckmann condensation – the base again deprotonates the remaining proton on the 1,3-dicarbonyl, which attacks the methyl ester. The immediate product is the cyclic keto-lactam, but the ketone is attacked by the proximal secondary alcohol, neatly forming the last ring and completing the target.
There was one catch, though – which you may have noticed. The team unfortunately made the enantiomer of the target, but as the spectral data matched with opposing optical rotation, I think we should let them have their prize!