White, Stang. Nature Chem., 2009, AOP. DOI: 10.1038/nchem.351.
Nice to see a bit of total synthesis in Nature Chemistry this month – but to be fair, this is more about the organometallics than the macrolide. But if you’re going to prove some methodology – especially, as in this case, macrolactonisation, this is a good target to work on. 6-deoxyerythronolide B (6-dEB) is a typical polyketide macrolide, and I mean that chemically, as the configuration of the stereocenters is very much the norm. This meant two things – 1) the group could use typical polyketide chemistry to build the stereocenters – and 2) this would serve as an excellent example for macrocyclisation chemistry.
The first task, of course, was to build the linear polyketide, unsurprisingly using an aldol-tastic route. Working right-to-left, the group first used a Myres alkylation to append an allyl group in a stereoselective manner. Two Evans-type boron syn-aldols allowed construction of the central chunk, whilst a further Myres alkylation allowed the C-6 stereocenter to be installed along with an aldehyde, ready for the next aldol. This final aldol was between two larger, and quite highly functionalised units – with several competing stereochemical interactions. Using standard conditions (see this Tetrahedron), a bit of tickle-four returned a reasonable yield of the product (49%, 7:1 d.r.). I’d have been fairly happy with that, but the group went that bit further, improving matters considerably with Ti(Oi-Pr)Cl3.
Barring a little reduction and removal of the chiral auxilliary, the linear precursor was done. With a free seco-acid, they were all set to macrocyclise – except they were lacking the C-13 hydroxyl with which to form the ester. However that was whole point – to do a stereoselective C-H oxidation and form the lactone in situ. Quite a lofty goal, but one which the group have worked on for some time.
Throwing the starting material in with a bit of catalyst (10 mol%) returned only a very low yield. Boosting the loading, and also the concentration, allowed the group to isolate a 34% yield, along with 45% recovered SM. Apparently, efforts to boost this result were unsuccessful, as they resorted to recycling the SM. However, this allowed isolation of a 56% yield, and a near perfect d.r. Damn nice work.
The paper then goes on to discuss the rationale for this stereochemical result, using a rather interesting method. Presuming that the reaction proceeds via a macrocyclic tether, the group using a source of flouride to deactivate this selectivity. The fluoride presumably attacks the catalyst, disrupting the tether, and is evident in the dramatic drop in diastereoselectivity. The suggestion is that it is the conformation of this tether, and the product-like transition-state that account for the stereoselectity, bourne-out by computational studies which suggest a 3 kcal mol-1 preference for the desired stereochemistry. When forming the pi-allyl intermediate, this complex can epimerise back and forth, with the desired conformation leading to product.
An interesting further study was to take the SM, and hydroxylate it. The two hydro-acids were then reacted under typical (Yamaguchi) conditions; whilst the SM with the desired stereochemistry cyclised nicely (87%), the C-13 epimer did not. It’s nice when nature works in your favour!
I should mention that a further three steps were needed to get to the close, but after this, do you really think it troubled them?