A second post this weekend - you guys are lucky the weather is awful here!
Total Synthesis of (+)-Omphadiol
Omphadiol is an africanane sesquiterpene isolated for the first time in 2000 from a basidiomycete. No-one knows what it does because of the small amounts isolated, but structurally similar sesqui- and diterpenes exhibit various biological activities including antiviral, anti-inflammatory and anti proliferative effects. It’s also got an interesting 5-7-3 ring system and 6 contiguous stereocentres around the ever popular trans-hydroazulene core. That notwithstanding, it took Daniel Romo and coworkers at Texas A&M University just 10 steps (and not a single protecting group) to prepare the natural product stereoselectively from (R)-carvone via a bicyclic b-lactone intermediate.
The route began with a chemoselective Mukaiyama (Markovnikov) hydration of the carvone enone olefin, followed by oxidative cleavage of the a-hydroxyketone with periodic acid to give the ketoacid shown. TsCl and 4-PPY (which is basically DMAP) were used to form the activated ester, which underwent aldol-type cyclisation followed by b-lactone formation. After some optimisation, using both K2CO3 and Hunig’s base, the reaction could be performed in great yield (83%) and excellent diastereoselectivity (apparently imparted by the isopropenyl group), even on 10 gram scale. The b-lactone was then reduced all the way to the diol with DIBAL. The primary alcohol was selectively tosylated, then converted to the corresponding bromide by a Finkelstein reaction. In the same pot, propionic anhydride was then used to acylate the tertiary alcohol, to give the bromoester in good yield. Another one pot sequence comprising formation of the ester enolate, cyclisation of this onto the newly formed bromide, a second deprotonation of the ester, and quenching with iodomethane, formed the lactone and installed the gem-dimethyl group.
However, opening of the lactone with various vinylmetal reagents to give the RCM precursor to the 7-membered ring was unexpectedly difficult. When the lactone was reduced to the lactol (easily achieved with DIBAL), addition did occur, but gave the undesired stereochemistry at the newly formed stereogenic centre, an outcome attributed to an usual case of 1,5-stereoinduction by the alkoxide in the lactol open form. Numerous attempts to perform monoaddition on the lactone were unsuccessful as after ring opening with the first vinyl nucleophile, 1,4-addition of a second equivalent to the newly formed enone was very rapid. Ultimately the group invented a clever solution to this problem; the lactone was opened using allyllithium (generated, as usual from an allyl stannane) to give the RCM precursor to the undesired 8-membered ring in good yield. However, upon treatment of this compound with Grubbs’ second generation catalyst only the desired 7-membered ring was obtained, with tandem isomerisation-RCM occurring in essentially quantitative yield. The success of this reaction was attributed to the slow formation of 8-membered rings, and the rapid and facile isomerisation of the b,g-olefin into conjugation with the carbonyl. Regio- and stereoselective reduction of the enone carbonyl was then achieved using the rather exciting t-BuLi/DIBAL combo, to give the equatorial alcohol. The final step, although only a Simmons-Smith cyclopropanation, was remarkable for its unusual diastereoselectivity.
The Simmons-Smith reaction is well known to be among the most substrate directable reactions (via OH coordination), but a lot depends on conformation and ring size. For example, 6- and 7-membered rings give cyclopropanation syn to the allylic alcohol with excellent diastereoselectivity, but allylic alcohols in 8-membered rings usually give anti products. The reaction is still hydroxyl directed, but the conformation causes a different outcome (see the review for details).
In the case of this final reaction, however, the authors believe that cyclopropanation is not directed by nonbonding interations, and that the diastereoselectivity arises purely from sterics. DFT modelling and NMR studies apparently indicate a conformation in which the OH is in the plane of the p bond and not able to assist in cyclopropanation. Interestingly, cyclopropanation also occurs from the top face if the epimeric allylic alcohol is used. In any case, the reaction provided the required third ring, and the data and optical rotation matched that of the natural product. Just 10 steps (18% overall yield) were required to access the natural product, and this approach is potentially useful for the synthesis of a number of related compounds. A nice piece of work!
1. Grubbs’ catalysts are actually great for isomerisations. I’ve used them to intentionally isomerise allylnaphthalenes into the corresponding conjugated compounds (with no undesired metathesis), and there are a number of papers on this kind of application. The putative mechanism involves ruthenium hydride intermediates arising from catalyst decomposition (review: Eur. J. Org. Chem. 2004, 1865). If you don’t want isomerisation to occur, Grubbs has shown that adding a trace of benzoquinone (to mop up that troublesome hydride) is an effective way to prevent this. More detail and numerous references can be found in a typically informative All Things Metathesis article.
2. See the references in the paper for some specific examples dealing with allyl alcohols in 7-membered rings. This is also covered generally in some detail in the classic, allstar Evans-Fu-Hoveyda review (Chem. Rev., 1993, 93, 1307-1370; worth reading for a lesson in how not to wear a bow tie from Greg Fu, if nothing else).