ACYCLIC DIASTEREOSELECTIVE SYNTHESIS USING TARTRATE ESTER MODIFIED CROTYLBORONATES - DOUBLE ASYMMETRIC REACTIONS WITH ALPHA-METHYL CHIRAL ALDEHYDES AND SYNTHESIS OF THE C(19)-C(29) SEGMENT OF RIFAMYCIN-S

Double asymmetric reactions of the tartrate ester modified crotylboronates 1 and 2 and α-methyl chiral aldehydes are described. The reactions of the appropriate enantiomers of 1 and 2 with β-alkoxy-α-methyipropionaldehydes 11 provide adducts 12, 13, and 14 with a minimum diastereoselectivity of 90%, provided that the optimal hydroxyl protecting group is selected for 11. Thus, TBDMS protected aldehyde 11a is the optimal substrate for the matched double asymmetric reactions leading to 12a and 14a, while the TBDPS protect 11b is the optimal precursor to 13b and 15b via mismatched double asymmetric reactions. Asimilar dependence of stereoselectivity on the protecting group is seen in the reactions of 11 and chiral allylboronate 16. It is inferred from these and other data (c.f., ΣΔΔG⋆ data provided in Table IV) that β-alkoxy aldehyde substituents have a significant, negative impact on the diastereoselectivity of the double asymmetric reactions of the tartrate allylboronates, especially those involving 2 and 16. Additional insight into the existence of the “alkoxy effect” is provided by the double asymmetricreactions of 1, 2, and 16 with aldehyde 20 that lacks an offending β-alkoxy group. These experiments (Table V) show that the diastereoselectivity of the reactions of 20 especially with 2 and 16 (ΣΔΔG⋆ = 1.7-1.8 kcal mol−1) are significantly improved relative to those with 11 (typically ΣΔΔG⋆ = 1.1-1.4 kcal mol−1). Improvements in stereoselectivity of the allyl- and (E)-crotylborations of both 11 and 20 are also possible by using reagents 28 and 29 incorporating the more highly enantioselective N,N′-dibenzyl-N,N′-ethylenetartramide auxiliary previously developed in these laboratories (Table VI). Adduct 23 deriving from these studies has been converted into lactone 27, a known precursor of the Prelog-Djerassi lactonic acid. An empirical model is presented that enables one to predict the situations in which 1 and 2 will be maximally effective in complex synthetic problems. Thus, dipropionate substructures 7 and 9 with anti relationships between branching methyl groups can be prepared with very high diastereoselectivity via matched double asymmetric reactions with the appropriate-methyl chiral aldehyde substrate, while substructures 8 and 10 with syn relationships between methyl branches are more difficult to prepare viamismatched double asymmetric reactions. Moreover, the ease of preparation of 7 and 9, and the difficulty with 8 and 10, is expected to increase as the intrinsic diastereofacial preference of the chiral aldehyde increases. Accordingly, the number of bond constructions leading to 1,3-anti branching methyl relationships should be maximized when applying this technologyin total synthesis, and the more difficult 1,3-syn branching methyl units should be introduced as early as possible. These principles are illustrated in a highly diastereoselective synthesis of the C(19)-C(29) segment of the ansa bridge of rifamycin S. This synthesis featuresfour C-C bond forming reactions involving the chiral crotyl- and allylboronate technology and proceeds in 15% yield and with 78% stereoselectivity for the 16-step sequence originating from (S)-llb. © 1990, American Chemical Society. All rights reserved.