A Short and Efficient Synthesis of Neodysiherbaine A by Using Catalytic Oxidative Cyclization
Isolated a decade ago from a marine sponge, neodysiherbaine A is an excitatory amino acid and a powerful convulsant and glutamate receptor antagonist. There've been four syntheses to date; two by Sakai and coworkers in 26 and 23 steps respectively, one by Hatakeyama in 21 steps, and a considerably shorter one by Lygo and coworkers in 15 steps. Tim Donohoe and his group at Oxford saw room for improvement again, and published yesterday a sweet seven step synthesis showcasing their Osmium based methodology for the oxidative formation of cis-tetrahydrofuran rings.
I'm usually fairly unenthused by syntheses starting from carbohydrate building blocks as there's usually a lot of protecting group juggling, but things are remarkably streamlined here. The starting material is commercial, or 1 step from D-ribose, which falls into the 'basically free' category of commercial chemicals. Treatment with the allyl silane shown below in the presence of tin tetrabromide gave the vinyl bromide as a 2:1 mixture of C-1 epimers in favour of the desired compound, presumably via the oxocarbenium ion. This preference is rationalised by a chairlike transition state in which 1,3-diaxial interactions between the acetoxy groups destabilise the conformer leading to the undesired product.
This vinyl bromide was then Negishi cross coupled with the alkyl iodide derived from N-Boc protected serine t-butyl ester (by standard Appel-type conditions with I2, imidazole and PPh3). Zinc/copper couple was used to form the organozinc reagent, which then underwent coupling in excellent yield using a little tetrakis(triphenylphosphine) palladium as the catalyst. A look at the SI shows that the reaction works well on gram-scale (91% with 2.5g bromide and 3 equivalents of the iodide/organozinc) but uses benzene/HMPA as the solvent. Well, you can't have everything. The acetate esters were then cleaved with catalytic sodium methoxide in methanol to give the triol, ready for the crucial THF formation. At this point I remembered the last Donohoe paper I read, where stoichiometric Osmium tetroxide was used for the syn-selective dihydroxylation of cyclic allylic alcohols, a methodology eventually applied in Kumamoto's synthesis of methyl-kinamycin C. I was therefore pleasantly surprised to see that a low loading of K2OsO2(OH)4 was able to do the job to give the cis-THF ring in excellent yield. Furthermore, using pyridine-N-oxide (PNO) as the oxidant, the dreaded OsO4 is never generated as PNO is only strong enough to go from the Os(IV) spat out of the catalytic cycle to the Os(VI) required for the next turnover, and not to Os(VIII). This is good, firstly because OsO4 is a little unpleasant, but also because if it was formed then dihydroxylation of the 1,1-disubstituted olefin might compete with the desired THF formation. Although previous reports from the group used a buffered solution, as in their 2009 synthesis of (+)-silvaticin, they found here that using plain water (resulting in a slightly acidic reaction mixture ~pH 5) was not detrimental to the acid labile protecting groups and greatly decreased the reaction time while also improving the yield.
Unfortunately, it was found necessary to tie up the vic-diol as the acetonide in order to manipulate the primary alcohol. Using rather a lot of TPAP, this was oxidised to the lactam (via the hemiaminal), and a final acid hydrolysis of this, the Boc group and the t-butyl ester gave the natural product in great yield.
Although this isn't the group's first showing of their oxidative approach to cis-THF rings it is an elegant application, leading to an interesting and brief synthesis of the amino acid neodysiherbaine A in just seven linear steps and an almost unbelievably good 24% overall yield.