Total Synthesis of (±)-Chloranthalactone A
Despite being just 12 steps, Liu's synthesis of chloranthalactone A published last week is full of interesting chemistry including some pretty uncommon transformations. The compound itself shows some antifungal activity, and is also structurally very similar to the monomer unit of a number of related dimeric natural products which the Liu group are apparently interested in targeting. The compound also contains a trans-5/6 ring junction with an angular methyl group, which is not at all common in nature, given that such systems much prefer to be cis given half the chance. This dangerously facile epimerisation meant that the group had to be very careful not to put any carbonyl groups near the ring junction at any point during the route, and lead to some creative chemistry.
Anyhow, the route began with commercially available Hagemann's ester (a building block first reported in 1893!). Addition of the cuprate derived from vinylmagnesium bromide and copper(i) bromide dimethyl sulfide complex to the enone proceeded in excellent yield, and the ketone was then protected as the corresponding dioxolane under classic conditions. The ethyl ester was then converted to its Weinreb amide and reacted with LiCH2OMOM (generated from the corresponding stannane by treatment with n-BuLi), both steps again occurring in essentially perfect yield. However, things soon became more challenging as only poor yields could be obtained when attempting to convert the α-alkoxyketone to the corresponding oxirane using Corey-Chaykovsky conditions. In light of this difficulty the group fell back on the somewhat uncommon Matteson epoxidation, using bromomethyllithium as the carbon source, which worked nicely.
Next came the key step of the synthesis, which was to set the crucial trans-5/6 junction. The group had hoped that the chair conformation of the cyclohexane ring would put the vinyl group and epoxide both equatorial, and if a cyclopropanation reaction could be used to link both units then the 5/6 junction would be locked trans with no chance to epimerise. The reaction that the group used to do this was the uncommon, but awesome, Hodgson cyclopropanation. Thus, deprotonation of the epoxide with LiTMP lead to formation of the expected cyclopropane in excellent yield, constructing two rings of the natural product simultaneously. The next challenge was the conversion of the vicinal diol (after MOM deprotection) to the required exo methylene group. Although at first glance one might be tempted to go down the periodate - Wittig route here, the intermediate ketone would introduce the possibility of epimerising the newly formed ring junction, and so an alternative route had to be used. Instead, the group converted the diol to the cyclic thiocarbonate using good ol' thiocarbonyldiimidazole (TCDI) which was then desulfurised/decarboxylated Corey-Winter style when brutalised with some trimethylphosphite. The group completed the synthesis with a clever method for the installation of the γ-alkylidenebutenolide, beginning with the reaction between the ketone enolate and ethylpyruvate. The new hydroxyl was then protected (under somewhat unusual conditions) to prevent a retro-aldol reaction, and treatment of the resulting acetate with DBU effected cyclisation and elimination to give the natural product. In total the route took only 12 steps (14% overall yield) and puts the group in a good position to pursue the more complex members of this family. A well planned (and executed) synthesis of a surprisingly tricky target!
1. Not a lot of people seem to be aware of Tomas Hudlicky's rather useful Synlett paper on the stabilities of bicyclic ketones (which is also reprinted in the Way of Synthesis pages 178-181). It basically draws together the known energy differences of such systems and some new calculations to give a useful picture of the position of the equilibrium between cis and trans forms of different bicycles. For example, here's the 5/6 table from the paper:
T. Hud. et al., Synlett, 2005, 2911-2914
5/5, 5/7 and 5/8 systems are also covered in a similar fashion. What this shows is that putting any group that acidifies adjacent protons next to the ring junction is a very bad idea as epimerisation is likely to be swift and irrevocable. Liu and coworkers are careful not to do this, even though it might be a tempting way to introduce that exo methylene.
2. Mechanistically, this is very similar to the Corey-Chaykovsky reaction. The bromomethyllithium (stable at -78 ºC) just adds to the ketone as expected and the alkoxide generated displaces the bromide, snapping shut the epoxide. I guess that the reagent is a hell of a lot more nucleophilic than the fairly well stabilised sulfur/sulfonium ylides that people usually use. See Synlett, 1991, 631–632 for more!