Platensimycin (1) is a novel broad-spectrum antibiotic (against Gram-positive bacteria) which was isolated from Streptomyces platensis by scientists from Merck: it inhibits bacterial growth by selectively inhibiting the condensing enzyme FabF of the bacterial fatty acid synthesis pathway. [1] Platensimycin (1) shows no cross-resistance to methicillinresistant Staphylococcus aureus, vancomycin-intermediate S. aureus, and vancomycin-resistant enterococci. As a result of its remarkable biological profile and challenging structure, platensimycin (1) has been the focus of intense synthetic activity, [2] and herein we describe the results of our recent efforts towards the synthesis of this intriguing compound.At the outset, a synthesis of the pivotal tetracyclic intermediate 2 from the cagelike ketone A was envisioned (Scheme 1). Rhodium(II)-catalyzed decomposition of a diazoketone D would lead to the formation of A and/or B through [3+2] cycloaddition [3] of the corresponding carbonyl ylide with conformations C and/or C'. This type of cycloaddition in the presence of rhodium(II) acetate is known to favor the formation of B (R = H; A/B/cyclopropanes = 6:41:36), [4] and a reversal of this product distribution seemed necessary to ensure the success of our synthetic approach.In practice, preparation of the quaternary-substituted diazoketone D was problematic.[5] After considerable experimentation, we found that the reaction sequence was successful with R = CN. Thus, diazoketone 6 was prepared from ethyl cyanoacetate (3) in a straightforward manner by sequential alkylation (Scheme 2). In the presence of rhodium(II) acetate, decomposition of diazoketone 6 proceeded smoothly to yield the cagelike ketone 8, accompanied by only a trace amount of the desired product 7, and a small amount of cyclopropane products 9. The use of rhodium(II) trifluoroacetate led to a clean conversion of 6 into 8.At first glance these results were disappointing, but we recognized that the overall cyclization-cycloaddition process was more competitive than the cyclopropanation reaction, and that the LUMO-dipole/HOMO-dipolarophile (type III) interaction [6] clearly dominated in the cycloaddition process because of the introduction of a nitrile group. The desired regioselectivity would be attained through the reversal of the olefin (dipolarophile) HOMO coefficient, for which halogen substitution emerged as an appropriate possibility. Accordingly, the halogenated substrates 10-13 were prepared from 4 for further regioselectivity studies.Rhodium(II)-catalyzed decomposition of 10 led to the relatively clean production of ketone 14 (Scheme 3). Under similar conditions, the (Z)-bromide 11 produced a relatively complex mixture from which 15 and 16 [7] were isolated in low yields. Gratifyingly, the (E)-bromide 12 was converted into Scheme 1. Retrosynthetic analysis of platensimycin (1).Scheme 2. Prototype carbonyl ylide [3+2] cycloaddition. a) NaOMe, MeCOCH 2 Cl, MeOH; b) NaH, CH 2 CHCH 2 Br, THF; c) 1 n KOH, MeOH; d) ClCO 2 iBu, TEA, diethyl ether, 0 8C; ...