Stereocontrolled Construction of Quaternary Carbons by Photosensitized Decarboxylative Michael Addition: Applications to Total Synthesis of Complex Terpenoids

            Almost 25 years ago, Keiji Okada and Masaji Oda of Osaka University demonstrated that a variety of tertiary radicals could be generated from N-phthalimidoyloxy esters using visible-light photoredox catalysis and that these trialkyl-carbon radicals add to Michael acceptors (a representative example is shown above). Importantly, this C-C bond-forming reaction generates a new quaternary carbon, with extrusion of CO₂ gas and phthalimide as by-products. The photodecarboxylation of Okada proceeds by initial single-electron transfer by a stoichiometric reducing agent (1-benzyl-1,4-dihydronicotinamide, BNAH) to the triplet state of Ru(bpy)₃²⁺ to give Ru(bpy)₃⁺. The Ru(I) species then reduces the N-phthalimidoyloxy ester substrate to the corresponding radical anion, which, upon subsequent homolytic cleavage of the N-O bond and decarboxylation, is converted to a tertiary carbon-centered radical. The nucleophilic tertiary radical then adds to the electron-deficient alkene coupling partner, thus affording an intermediary a-acyl radical that is finally quenched by hydrogen atom abstraction from BNAH radical cation. As stated above, this sequence furnishes a Michael adduct bearing a new all-carbon quaternary stereogenic center. Stereocontrol in the radical process is governed by steric properties inherent to the N-phthalimidoyloxy ester substrate.

            Larry Overman’s group at the University of California, Irvine has introduced some methodological advancements to the Okada Michael addition that have greatly extended the substrate scope of this arguably underutilized C-C bond forming photoredox process. The reaction has proven itself particularly useful for Overman’s group as a natural product fragment coupling, wherein a relatively complex polycyclic framework is converted to a trialkyl-carbon radical and joined with an oxacyclic side chain electrophile (see above) or a cyclic enone (see below) as the Michael acceptor. In the case of the trans-clerodane plant diterpenoid solidagolactone, a convergent synthetic approach involving 1,6-addition of a decalin-type tertiary radical to a vinyl-substituted butenolide was envisioned. In this pivotal coupling step, the decalin radical, generated from its corresponding N-phthalimidoyloxy ester under photocatalysis conditions, engaged the electron deficient butenolide alkene exclusively from the less-hindered equatorial face, presumably due to destabilizing 1,3-diaxial interactions in the diastereomeric axial-coupled product. Their reaction conditions are essentially identical to those reported by Okada in 1991 except for replacement of BNAH with a Hantzsch ester as the stoichiometric reductant. The product was obtained as a single epimer at the newly formed quaternary stereocenter, also bearing a mixture of two side chain olefin regioisomers. Exposure of the mixture to DBU equilibrated the double bond configuration to the thermodynamically favored a,b-unsaturated butenolide and a final rhodium-catalyzed olefin isomerization then yielded fully synthetic (-)-solidagolactone. A similar photoredox-mediated fragment union was utilized by the Overman laboratory in the course of their total synthesis of the complex diterpenoid, aplyviolene (see below). In this case, a bicyclic N-phthalimidoyloxy ester susbtructure was coupled to an a-chlorocyclopentenone to fashion adjacent quaternary and tertiary stereocenters in a stereoselective fashion.

            In view of the mild chemical nature and broad substrate scope of the Okada/Overman decarboxylative Michael reaction, it is perhaps surprising that this process is not widely applied for the construction of quaternary all-carbon stereocenters. This may be attributed to a perceived difficulty in controlling the stereochemical course and outcome of reactions involving trigonal planar carbon-centered radicals. Overman’s group also hypothesizes that this underutilization is due to lack of convenient methods for generating the required synthetic intermediates. They developed a simple protocol for the preparation of tert-alkyl N-phalimidoyl oxalates from a variety of readily available tertiary alcohols. The tert-alkyl N-phalimidoyl oxalate substrates are generated by acylation of the corresponding tertiary alcohol with the reagent chloro N-phthalimidoyl oxalate (generated in situ from oxalyl chloride and N-hydroxyphthalimide). Synthesis of the analogous Okada substrates is more challenging, requiring non-trivial esterification of a hindered neopentyl carboxylic acid with N-hydroxyphthalimide. Overman’s oxalates undergo photoredox-mediated fragmentation to give nucleophilic tertiary radicals through a fragmentation mechanism similar to that described above, but involving two decarboxylation events rather than one (Overman’s published mechanism is reproduced below). The reaction is tolerant of a variety of functional groups (including N-heterocycles) and chiral oxalates such as the estrone-derived precursor depicted below couple with high diastereoselectivity (dr > 20:1).

            Given the demonstrated compatibility of the mild photoredox coupling conditions of Okada/Overman with ornate polycyclic systems, it is tempting to wonder if this method might be suitable for closure of the B-ring in the context of a hypothetical application to steroid total synthesis (for previous discussions of steroid B-ring closure in total synthesis, see here and here). Such a process, if conducted in an intermolecular sense, would require a pro-C10 tert-alkyl N-phalimidoyl oxalate (shown below) wherein the A-ring is adjoined to the C-ring via an a,b-unsaturated ketone tether. This substrate could potentially be synthesized from an alkyne (or, more specifically, enyne) precursor, itself produced by a Sonogashira coupling reaction. Enynes of this nature have been shown to undergo triple bond hydration in a regioselective fashion under acidic conditions in the presence of palladium(II). The pendant alcohol functionality is participatory in the hydrolysis mechanism, which involves formation of a dihydrofuran intermediate. Photoredox-mediated oxalate fragmentation with concomitant intramolecular Michael addition of a pro-C10 tertiary radical to the C-ring enone then forges the intact steroid nucleus and establishes the all-carbon quaternary stereocenter at C10. The reaction also installs new stereogenic positions at C8 and C9, ultimately furnishing the C7-oxo steroid shown below. The stereochemical outcome of the final transformation is far from certain. A chairlike transition state (depicted below in brackets) wherein the olefin approaches the tertiary radical from the equatorial face would lead to the all-trans relative diastereomeric configuration found in the natural androstane carbon skeleton. Retrosynthetic dissection of the polycyclic steroid framework using new bond disconnections enabled by novel synthetic methodologies is always an interesting thought exercise.