Studies on 1,2-migration via the intermediacy of dirhodium (II) carbene and the application towards the synthesis of welwitindolinone family

Abstract

The construction of bridged bicyclo[m.n.1]alkanones based on the 1,2-migration of dirhodium (II) carbenes has been studied. Seven ring-fused bicyclic α-diazo-β-hydroxyketones 2.4, 2.5, 2.6, 2.7, 2.10, 2.11 and 2.13 with different steric and electronic features were synthesized to investigate their reactions with rhodium. The results supported our previous mechanistic understanding that the rhodium carbenes undergo stereoelectronically allowed migrations of the more electron-rich bonds.

The rhodium catalysts have an effect on the reaction outcome. The reactivity of rhodium catalysts should match the reactivity α-diazo-β-hydroxyketones for optimum yields and selectivity. The application of a chiral rhodium catalyst to the reaction of racemic 2.19 was demonstrated to achieve some degree of kinetic resolution. Treatment of α-diazo-β-hydroxyketone 2.11 with rhodium (II) catalysts of lower reactivity afforded 1,2-dione 2.12 along with the typical migration products 2.11a and 2.11b. The formation of the 1,2-dione is proposed to be due to the rhodium catalyst reacting as a Lewis acid instead of generating a carbene.

The reaction of α-diazo-β-hydroxyketone 2.13 generated Wolff rearrangement products 2.14 and 2.15 along with the typical migration product 2.13a, and the distribution was found to vary with the rhodium catalyst.

The results were applied to the synthesis of the welwitindolinone framework 4.47 bearing the indole platform and the gem-dimethyl moiety, via α-diazo-β-hydroxyketone precursors. Commercially available 4-bromoindole was N-methylated, then alkylated with epoxysulfone 4.4 mediated by SnCl4 to afford aldehyde 4.8. A Roskamp reaction and a Pd-catalyzed enolate arylation afforded β-ketoester 4.10.

A first strategy was to synthesize α-diazo-β-hydroxyketone 4.14. for the key migration to afford the welwitindolinone framework. Homologation of 4.10 to 4.11 and debenzylation provided 4.12. Activation and diazomethane addition generated diazoketone 4.13 successfully. However, the hindrance of the gem-dimethyl group prevented the aldol cyclization to form 4.14.

A second strategy converted 4.10 to α-diazo-β-hydroxyketone 4.18 for a ring expansion that successfully constructed the welwitindolinone framework. Ketoester 4.10 was alkylated using a bromoester, then 4.15 was converted to 4.18 via typical transformations. Diazoketone 4.18 was successfully cyclized using TBD to give α-diazo-β-hydroxyketone 4.19. Treatment with rhodium (II) acetate provided bicyclo[4.2.1]alkanone 4.20 smoothly in 77% yield.

Two ring expansion strategies towards achieving the welwitindolinone framework were investigated. Firstly, 4.20 was converted to 4.32 and subjected to dichlorocyclopropanation to provide 4.33, having deactivated the indole ring. A silver (I) assisted ring-opening of the dichlorocyclopropane led to ring opening, but an unintended intramolecular attack diverted the reaction to afford cyclobutane 4.38. Alternatively the treatment of 4.20 with TMSCHN2 in one step afforded 4.48 bearing the welwitindoline framework, in 10 steps from 4-bromoindole.

Key intermediate 4.48 was further oxidized to oxindole 4.50, enolized regioselectively to afford 4.51, and transformed to isothiocyanate 4.56.