Direct catalytic asymmetric synthesis of anti-1,2-amino alcohols and syn-1,2-diols through organocatalytic anti-Mannich and syn-aldol reactions.

Chiral 1,2-amino alcohols and 1,2-diols are common structural motifs found in a vast array of natural and biologically active molecules.1 Recently, significant efforts have been applied toward the development of direct catalytic asymmetric approaches to the construction of these units based on the addition of unmodified R-hydroxyketones to imines or aldehydes in Mannich-type and aldol reactions, respectively.2,3 Although the elegant studies of Shibasaki and Trost have provided routes to both synand anti-1,2-amino alcohols and diols using metal-based catalysis,2 highly enantioselective organocatalytic approaches have been limited to syn-1,2amino alcohols and anti-1,2-diols.3 Here we describe simple and efficient routes to highly enantiomerically enriched anti-1,2-amino alcohols and syn-1,2-diols through direct asymmetric Mannich, Mannich-type, and aldol reactions catalyzed by primary aminecontaining amino acids. To generate anti-1,2-amino alcohols and syn-1,2-diols, we sought to design novel catalysts. In the reactions of R-hydroxyketones with (S)-proline, products form via a reaction involving an (E)-enamine A for both Mannich-type and aldol reactions3 (Scheme 1). With pyrrolidine-derived catalysts or secondary amines, (E)-enamine intermediates predominate because of steric interactions in (Z)enamine B. The stereochemistry of the product can be explained by transition state C or D because the si face of the (E)-enamine reacts (Scheme 1a). To selectively form anti-Mannich products in reactions involving alkylaldehydes and alkanone-derived nucleophiles, we previously designed catalysts (3R,5R)-5-methyl-3pyrrolidinecarboxylic acid and (R)-3-pyrrolidinecarboxylic acid ((R)-â-proline), respectively.4,5 With the latter catalyst, reactions proceed through transition state E, and the reaction face of the (E)enamine is reversed from that of the (S)-proline-catalyzed reaction (Scheme 1b). These catalysts were, however, less than optimal for reactions of R-hydroxyketones.6 For reactions of R-hydroxyketones, we reasoned that the use of a (Z)-enamine in the C-C bond-forming transition state should generate anti-Mannich and syn-aldol products. In our early studies of aldol reactions involving unmodified hydroxyacetone mediated by antibody catalysis, we noted preferential reaction of a (Z)enamine of hydroxyacetone formed with the primary amine of the lysine side chain, the key catalytic residue of the aldolase, rather than reaction through an (E)-enamine as we had observed with cyclic ketones.7 We reasoned that, with primary amines, (Z)enamines of R-hydroxyketones F should predominate over (E)enamines G.8 When (Z)-enamine F reacts in the C-C bond-forming transition state (H or I), anti-Mannich or syn-aldol products should form predominately (Scheme 1c). Studies of direct asymmetric aldol and Mannich-type reactions catalyzed by primary amine-containing amino acids have been reported.9 However, within these studies, reactions of R-hydroxyketones were either not tested or, when tested, enantioselectivities of the products were moderate. On the basis of our design considerations, we first evaluated a variety of natural acyclic amino acids and their derivatives, including amino acids 1-3, for the Mannich-type and aldol reactions of hydroxyacetone that afforded 4 and 5, respectively (Figure 1 and Table 1). In accord with our hypothesis, primary aminecontaining amino acids predominantly provided anti-Mannich product 4 or syn-aldol product 5, but the anti/syn ratios and ee’s were varied. For the Mannich-type reaction, reactions catalyzed by L-Trp (1) and O-tBu-L-Thr (3) afforded anti-4 with high dr and ee (entries 1 and 4). N-Methyl-L-Trp catalysis provided only trace amounts of product. For the aldol reaction, the reaction catalyzed by 3 afforded syn-5 with the best dr and ee (entry 9). The L-Thr (2)-catalyzed aldol reaction provided the next best syn-selectivity and enantioselectivity (entry 8). Other natural amino acids did not provide significant syn-selectivity or enantioselectivity (data not shown). With all catalysts tested, C-C bond formation with hydroxyacetone selectively occurred at the carbon bearing the hydroxyl group. Conditions were optimized for the 1and 3-catalyzed Mannichtype reactions. Using the optimized conditions, Mannich and Mannich-type reactions of hydroxyacetone with a variety of imines were performed in DMF for catalyst 1 or N-methylpyrrolidone (NMP) for catalyst 3 at 4 °C (Table 2). The reaction with catalyst 1 was faster than that of catalyst 3. Reaction time was 16-20 h with 1 and 48 h with 3. The desired anti-amino alcohols 4, 6-8 were obtained in good yields with excellent diastereoselectivities (up to >15:1) and enantioselectivities (90-98% ee) in most cases. Significantly, reaction of unmodified 1-hydroxy-2-butanone provided the anti-product regioselectively with excellent dr and ee Figure 1. Structures of catalysts studied.