Morimoto, Okita, Kambara. ACIEE, 2009, EarlyView. DOI: 10.1002/anie.200805857.
This isn’t the first poly-THF we’ve examined at Tot. Syn., but it is the most complex (see sylvaticin back in 2006 for the other example). To the casual eye, the molecule looks deceptively symmetrical, but it isn’t. In fact, only the two right-most THFs are alike, with the others differing in substitution, making the synthesis considerably more complex. And to be frank, that’s the only reason we’re looking at it, as there doesn’t appear to be any biological activity. More about the isolation of the compound and this family of laurencia targets can be seen in this Helvetica Chimica Acta, found bearing some rather familiar authors…
A look in that HCA (in fact you need only go as far as the title) indicates it’s relation to squalene, and a hint towards their synthesis. However, rather than simply per-epoxidising squalene and letting the compound zip-up (which is probably what nature does), we chemists have to do be a bit more careful. Morimoto’s retrosynthesis suggests that the three right-most THFs can be formed in a cascade, whilst the left-most could be derived from Hoye’s cyclisation. This in turn results from more epoxidation, meaning that the tub of titanium isopropoxide and fructose required had better be big un’s.
Events are brought into motion by firstly selectively epoxidising the allylic alchol using Sharpless’s finger-licking good asymmetric epoxidation, then Shi’s fructose-DMDO thing to oxygenate the internal alkene. The catch was that the Shi catalyst used is derived from L-fructose – the unnatural variety. However, as pointed out ages ago on this blog, Shi has a rather nice prep of this catalyst from L-sorbose, which is thankfully the naturally available form. I wonder if the lower d.r. in the Shi epoxidation is a result of a conflict between substrate and reagent control… it’s getting kinda close to the diol. Anyway, the group lobbed in a bit of base and went off to tea, fingers crossed that the cyclisation would go as before (actually, do Japanese groups go for tea? I feel I should know more about the Japanese chemist’s life – seeing as Japan is the second-biggest audience for this blog).
It performed like a champ, returning the first THF ‘with high stereospecificity‘; actual numbers are somewhat absent. Oxidation of the primary alcohol and methylenation gave a terminal olefin ready for metathesis to bolt on the other half of the molecule – all-set with three epoxides in place. This one of those points where it’s worth going back over old territory, and mentioning just how awesome metathesis is – both halves of this reaction are rather sensative, and yet metathesis treats them like a gentleman, returning an 87% yield. The freshly installed olefin was the reduced with another smart reaction – a diimide reduction. This bad-boy reduces olefins really selectively – but I’m not entirely sure how the active reductant is formed in this case. Any ideas? Surely it’s not just protonation, decarboxylation and formation of actual diimine?
Anyway, banging this stuff in with a bit of acid in methanol results in reduction, but apparently the protons from the acetic acid aren’t good enough for this cylisation. A bit of CSA did the trick though, causing three consecutive 5-exo-tet cyclisations, and formation of the three remaining THFs in one cascade. Nice, even if the yield isn’t amazing.
The end of the synthesis seems to contain a momentary stumble, though. Removal of the acetonide gave a diol, elimination of which would give the prenyl-style sidechain. However, the paper describes an oxidative cleavage to give a lactol, and then olefination of the lactol. Perhaps a Corey-Winter olefination was intended, but didn’t work?
One critical point – although the supporting information is pretty full when it comes to text and numbers, there aren’t any spectra. Surely a first synthesis of any natural product these days need a comparison between synthetic and natural spectra? Especially when both samples should be in the same fridge?