This work is part of the Ph. D. thesis of H . B. [l]. Present address: Ciba Geigy AG, CH4002 Basel b a C R', R2 = suitable protecting groups
A retrosynthetic plan for a first approach is depicted in Scheme 1. The stability of polyoximic acid (3) towards boiling 3~ HC1 during the hydrolysis of polyoxin A [3b] suggested that the isomerization of a 3-vinylazetidine-2-carboxylate a to the corresponding 3-ethylidene derivative b should in principle be possible. Due to the combination of strain and entropic factors, the ring closure to azetidines is a comparatively unfavorable process [12]. Thus, with the selected strategy, additional problems of the ring formation due to the strain exerted by the exocyclic double bond are avoided. Since cyclization of 4-halogeno-substituted 2-(tosy1amino)butyrates is one of the most efficient methods to prepare azetidine-2-carboxylates [ 131, we chose sulfonamide c with a leaving group X as precursor for a3).
A second, more direct way to b was realized by a [2 + 21 cycloaddition.
Results. -Compound c, a y,d-unsaturated c1 -amino-acid derivative, should be accessible by the Ireland-Claisen rearrangement of an appropriate ally1 glycinate 11 513. Our results along these lines are shown in Scheme 2. Glycine protected as tert-butyl carbamate 4 was esterified with an excess of (Z)-2-butene-1,4-diol (5) using dicyclohexylcarbodiimide (DCC) and a catalytic amount of 4-(dimethy1amino)pyridine to give 90 % of monoester 6 together with some bis-glycinate 7 (7 "/.). When (E)-2-butene-1,4-dioI (8), obtained in 80% yield by LiAlH, reduction of 2-butyne-l,4-diol, was reacted analogously, the amount of diester 9 (52%) formed together with mono-glycinate 10 (42%) was larger. However, additional monoester 10 could be obtained by acid-catalyzed transesterification of 9 with 1 equiv. of diol 8. The OH functions of 6 and 10 were then protected by silylation with hexamethyldisilazane. Subsequent Ireland-Claisen rearrangement according to [ 151 gave the epimeric 3-vinylhomoserine-lactone derivatives 11 and 12 in moderate 3,
In another approach to a, ethyl 2-bromo-3-(bromomethyl)-4-pentenoate (1 : 1 epimeric mixture) was reacted with (diphenylmethyl)amine, following the azetidine-2-carboxylate synthesis of Rodebaugh and Cromwell[14]. However, the desired epimeric 3-vinylawtidine-2-carboxylates were formed in 5% yield only, the major pathway of the reaction (43 % of products) being monosubstitution followed by HBr elimination (for details, Another approach to compound c, the ene reaction of (tosy1imino)acetates [16], failed with several 2-butenyl derivatives (cf. [I]). Successful, as reported later by Weinreb et al. [17], was an intramolecular version of this reaction. cf. Ill).
4, ' )
A more direct access to the lactones 13 and 14 would be the rearrangement of allyl N-tosylglycinates, Reaction of the N-tosyl analog of 6 gave, however, only 7 % of 13, even when the highly unstable Li enolate was quenched in situ with chlorotrimethylsilane (cf [I]). Alternatively, reaction of 14 with trichloro(methyl)mlane/NaI according to Oluh et ul. [21] and esterification with CH,N, gave 30% of the iodo methyl ester corresponding to 17 (cf [I]). ' ) 'I) The structure elucidation of 22 mainly relies on the 13C-NMR signals at 72.3 and 79.8 ppm, assigned to the acetylenic C-atoms C(4) and C(5).