Totally Synthetic by Paul H. Docherty, 7 December 2011
Total Synthesis of Taxadienone & Taxol
A. Mendoza, Y. Ishihara, P. S. Baran, Nat. Chem. 2012, 4, 21-25
What we’ve got here is a sort-of meta-synthesis; Baran doesn’t actually make any of the taxane natural products, but demonstrates an extremely neat synthesis of a potential common precursor. The theory behind the work is related to previous Baran diatribes, where he points out that too many steps in syntheses simply oxidise and reduce the same carbons. However, in this case Baran describes the Taxanes (which are of course heavily oxidised) as being parented by a simple (almost) un-oxidised ‘precursor’.
This compound is known as ‘Taxadienone’ - and as the name suggests, it contains a pair of alkene groups and a sole carbonyl as its only oxidation. However, deeming it a ‘precursor’ is perhaps a little ambitious, as to get to taxol or related natural products, a lot of oxidation of unactivated C-H bonds is required. In the hands of your average chemist, this seems a pretty insurmountable challenge, but with Baran, perhaps…
Baran’s approach to the complex 6,8,6-ring system, as ever, appears fairly simple. Start with a cyclohexenone, do a pair of carbon-carbon bond formations about the alkene, and then unite these sidechains in a neat cycloaddition. And the interesting thing is that Baran makes it work just as easily as that.
The first, and more complex, sidechain was appended using an organocopper 1,6-addition. This combination of diene with dienone looks set to generate a black tar in my hands, but team Baran got a remarkable 86%. Then using some of Alexakis’ chemistry, they methylated this product in a 1,4-addition to introduce asymmetry. Using a chiral phosphoramidite ligand, this was achieved in a very reasonable 89% yield and 93% e.e., using pretty low loading of the copper thiophenecarboxylate and ligand. I really like the simplicity of this approach - keeping the system symmetrical until this point reduces complexity considerably.
Having trapped the product of the methylation as it’s TMS ether, the group then did a Mukaiyama-type aldol coupling with Acrolein. This was a little awkward, as the group often found the ketone product of desilylation rather than their desired aldol product, even when rigorously excluding water. Bizarely, adding water helped - and by raking to the back of the Lewis acid cupboard in the lab, they found success with the unusual gadolinium triflate. These are apparently Kobayashi’s conditions, and gave them a yield of over 85%, but as a 2:1 mixture of diastereomers. However, they didn’t separate the isomers - rather, they chucked the lot into the Diels-Alder below.
Under more typical reaction conditions, these reaction fairly nicely, giving a reasonable yield of product, but perhaps more importantly (in a sense), the ability to separate the isomers with ease. They weren’t quite done - a triflation of the cyclohexenone carbonyl at the only acidic methylene position in the molecule generated a enol triflate. Reaction of this under Negishi conditions with dimethyl zinc provided them with the required methyl cyclohexene moiety - and in a satisfying 84% yield.
Synthesis done, and like me, you may well be wondering what the excitement is about. However, continue reading the paper into the discussion, and the chemistry takes on a different note, as Baran describes various strategies that were uncooperative. These included a fail aldol approach to closing the medium ring as a later stage, and a stereochemical nightmare involving a Shapiro reaction.
More importantly, Baran believes that site-selective oxidations of taxadienone are possible, and may lead to a total synthesis of taxol form this intermediate. Now that’s a paper worth getting excited about!