B.R.S.M. Pain is temporary, publications are eternal



Total Synthesis of Gelsemoxonine [PDF][SI][GROUP]

Fukuyama et al. J. Am. Chem. Soc., 2011, ASAP

DOI: 10.1021/ja208617c

After what feels like a long dry spell in exciting total syntheses, Fukuyama and coworkers disclosed the first asymmetric synthesis of the natural product gelsemoxonine last week. Although not much is known about the biological activity of gelsemoxonine itself, the related natural products gelsedine and gelsenicine have been shown to exhibit potent cytotoxic activities and moreover, the densely functionalised oxabicyclo[3.2.2]nonane core with its azetidine and indolinone quaternary centres makes for an exciting synthetic target.

The route began with a reaction that was a key part of my 4th year undergraduate project; the somewhat hard-to-pronounce Achmatowicz oxidative expansion of furfuryl alcohol. The group then went on to use lipase AK to perform a dynamic kinetic resolution and obtain the corresponding acetate in an enantioenriched form. Unfortunately, this didn't give a particular high ee and so a second lipase was then used perform solvolysis of the unwanted enantiomer, increasing the ee to 99%. The enone was then converted to the corresponding cyclopropane upon treatment with diethyl bromomalonate in the presence of DBU. Attack of the malonate anion was directed to the α-face due to the presence of the β-acetoxy group, and the product was obtained as a single diastereomer. The acetate was then excised using triethylsilane and TMSOTf, and the group then set about reduction of the diester to the diol. Although LiAlH4 could get the job done, the resulting triol was difficult to isolate due to its coordinating ability and high water solubility. After what I imagine was a lot of experimentation, the group found that the reduction could be effected using sodium borohydride in refluxing THF. When the reaction was quenched with methanol and acetic acid then simple concentration gave a mixture of the triol and sodium acetate, from which the product could easily be separated.[1] The less hindered alcohol on the convex face was then protected selectively as the pivaloate ester, and the remaining  alcohols were oxidised the respective ketone and aldehyde.

The group had originally planned to unite this ketoaldehyde with the known indolinone by means of a Knoevenagel-type condensation, but this proved unsuccessful and so a two step aldol reaction-dehydration sequence was used. The ketone was then converted to the silyl enol ether and when this was heated in toluene then the planned divinylcyclopropane-cycloheptadiene Cope rearrangement occurred, forming the oxabicyclo[3.2.2]nonane ring system in excellent yield.

The pivaloate ester was then cleaved using sodium methoxide,[2] and the resulting alcohol oxidised using the Piancatelli conditions. Next followed a rather interesting sequence to install the nitrogen atom which would ultimately be used to form the azetidine present in the natural product. Thus, when the enal was treated with TMSCN and DBU the cyanohydrin TMS ether was formed, which then underwent isomerisation to the α-trimethylsiloxy vinyl cyanide before conversion to the acyl cyanide. This was quenched with allyl alcohol to give the expected ester, obtained as 4:1 mixture of epimers in favour of the desired axial product. Deallylation using Pd(PPh3)4 and pyrrolidine as the nucleophile gave the acid that was then converted, via the acid chloride, to the acyl azide. Heating of this compound in benzyl alcohol effected a Curtius rearrangement to give the Cbz protected amine in good yield.

This protected amine was then heated with Bredereck’s reagent in toluene, resulting in conversion of the ketone to the vinylogous amide and installing an extra carbon atom. This amide then underwent the expected reaction with oxalyl chloride to give the β-chloroenal, which was subjected to palladium catalysed dehalogenation. The resulting enal was then converted to the ketone by treatment with ethylmagnesium bromide, followed by reoxidation with IBX. Exposure of the new α,β-unsaturated ketone to t-butylhydroperoxide resulted in epoxidation exclusively from the convex face, and the Cbz group was then removed with iodotrimethylsilane. Finally, simply heating the resulting amine in refluxing ethanol led to azetidine formation in high yield, a step the group performed on over 300 mg scale, marking an impressive end to a very interesting total synthesis.

1. The only other byproduct of this procedure is trimethyl borate, which was easily removed under vacuum without the need for an aqueous workup.

2. Prepared from the action of sodium hydride on methanol, a popular protocol for people with no patience.

Comments (2) Trackbacks (0)
  1. “Prepared from the action of sodium hydride on methanol, a popular protocol for people with no patience” HAHAHAHA sooooo true. I’m nearly always tempted to make it this way just cause the sodium metal way takes so god damn long. Excellent writeup of a beautiful synthesis!

    • Thanks! This week I had to dissolve a full mole of sodium metal in around 300 mL of methanol, which took pretty much an entire afternoon even with heating. It’s fast at first then it slows right down!

Leave a comment

No trackbacks yet.