Formal Synthesis of (−)‐Aphanorphine Using Sequential Photomediated Radical Reactions of Dithiocarbamates

Abstract

Aphanorphine (1), an alkaloid isolated from the freshwater blue-green alga Aphanizomenon flos-aquae, has attracted considerable attention from the synthetic community owing to its structural similarity to natural and non-natural analgesics such as morphine, eptazocine, and pentazocine (Scheme 1). Approaches to aphanorphine developed to date have all relied on the formation of the B or C ring to complete theC-norbenzomorphan skeleton, typically exploiting the rigidity of the bridged tricyclic 3-benzazepine structure to set the second stereocenter from a preexisting quaternary benzylic stereocenter at C1, or from an abranched amine at C4. In this communication we present a complementary strategy for the synthesis of aphanorphine which is characterized by the late-stage incorporation of the aromatic A ring, and formation of the pyrrolidine C ring through a novel carbon–carbon bondforming reaction. We have recently reported a new method for the generation of carbamoyl (aminoacyl) radicals from dithiocarbamate precursors, and their subsequent intramolecular addition—dithiocarbamate group-transfer reactions with alkenes. Application of this methodology to the synthesis of the core 6-azabicyclo[3.2.1]octane ring system of aphanorphine was envisaged based upon a regioselective 5-exo-trig cyclization of carbamoyl radical 2 followed by dithiocarbamate group transfer to give the functionalized bicyclic lactam 3 (Scheme 1). It was further envisaged that the dithiocarbamate group in 3 would provide a handle for phenol annulation. Critical to the success of such an approach is the ability of carbamoyl radicals generated from dithiocarbamate precursors to undergo potentially difficult cyclizations onto unactivated alkenes. 4] Previous work by Quirante, Bonjoch, et al. had shown that a-amino radicals undergo analogous cyclizations onto alkenes carrying electron-withdrawing groups at C9a (aphanorphine numbering); however, unactivated alkenes did not undergo cyclization. The effect of a further alkene substituent at C1, which may also disfavor 5-exo-trig cyclization, was not evaluated. An asymmetric synthesis of the requisite secondary cyclohexenylamine is outlined in Scheme 2 and relies on Ellman6s sulfinamide auxiliary to set the amino-substituted stereocenter destined to be C4 of aphanorphine. 8] Condensation of enantiomerically pure (R)-tert-butanesulfinamide (5) with commercially available cis-4-heptenal gave the expected (E)-sulfinimine 6 (Scheme 2). Addition of 2methylallylmagnesium chloride gave rise to sulfinamide 7 in excellent yield as a 83:17 mixture of diastereoisomers. The configuration of the major stereoisomer 7 was predicted to be R on the basis of the Ellman model and was ultimately proven through a formal synthesis of ( )-aphanorphine. Separation of the two diastereomers could not be achieved at this stage, and so the mixture was carried through the following steps. Following N-methylation of 7, 1,7-diene 8 was subjected to ring-closing metathesis using the Grubbs second-generation catalyst, which furnished the trisubstituted alkene 9 in excellent yield. Finally removal of the sulfinyl auxiliary under acidic conditions gave the hydrochloride salt 10. At this stage, a single recrystallization of 10 gave enantiomerically pure material. Scheme 1. Analgesics structurally related to aphanorphine and a retrosynthesis of ( )-aphanorphine (1).