Diastereoselective Additions of Ethynyl Grignard Reagent to Erythrulose Derivatives Shoji Kobayashi, Parthasarathi Das, Guang Xing Wang, Takashi Mita, Martin J. Lear, and Masahiro Hirama  Department of Chemistry, Graduate School of Science, Tohoku University and CREST, Japan Science and Technology Corporation (JST), Sendai 980-8578 (Received November 30, 2001; CL-011214) Through systematic changes in reaction conditions and in the use of Ti(OiPr) 4 , the typical stereochemical outcome of ethynylmagnesium bromide on ,-O-isopropylidene-erythru- lose derivatives has been reversed with exceptional levels of control. During the course of our recent studies into the total and analog synthesis of nine-membered enediyne antibiotics, we have been continuously improving our tactical approach in the synthesis of key fragments. In particular, we have shown 1,2-O- isopropyridene-3-ethynylerythrulose derivatives (cf. 1, 2, ent-1 and ent-2) to be valuable intermediates in our synthetic studies of N1999-A2 (3), the neocarzinostatin chromophore, C-1027, and model chromophores of kedarcidin. 1 To date only a few methods to obtain 1 and 2 have been reported, especially with regard to attaining 2,3-syn-isomers (1). Specifically, Nagano 2 and Terashima 3 achieved the synthesis of 1 and 2 by treating ethynyl Grignard reagent or lithium acetylides with ketones 5a or 5c, respectively, but no notable selectivity for 2,3-syn-products (1) over 2,3-anti-isomers (2) were observed. Although the 2,3-syn-selective addition with the corresponding aldehyde (4) has been studied extensively 4 and some useful conditions 4b,c including the use of organocopper reagents 4d have been reported, these are not always applicable to functionalized ketones or to the low reactivity of ethynyl copper reagents. Indeed, treatment of the TBS-protected ketone (5b) with ethynyl copper reagent in THF-Me 2 S 4d resulted in the complete recovery of 5b. The best cases of organometallic additions to systems like 5 which divert from an anti-selective outcome involve carbonyl- substrates with -benzyloxy or non-cyclic alkoxy protective groups, that are readily capable of forming a strong metal chelate between the ether oxygen and carbonyl groups. 5;6 To the best of our knowledge, only the groups of Nagano 2 and Marco 7 have reported 2,3-syn-selective Grignard additions to the cyclic 1,2-O- isopropylidenes (5), but still the ratios are rather poor. 8 In this letter, we disclose stereodivergent conditions to generate either the syn-adducts (1) or the anti-adducts (2) with high levels of stereoselectivity. Results of the addition of ethynyl Grignard reagent to the ketones (5b-d) are summarized in Table 1. Anti-adduct (2b) was obtained predominantly in the case of the TBS-ether (5b) in various solvents without additives (entry 1–4). Notably, the reaction in THF or Et 2 O gave the anti-adduct (5b) in high yield and selectivity (entry 1, 2). Addition of ZnBr 2 and MgBr 2  OEt 2 was ineffective in trying to reverse the anti-selectivity, 4;9 and only a decrease in both the reaction rate and the chemical yield resulted (entry 5–9). Interestingly, the syn/anti ratio changed dramatically in favor of the syn-adduct (1b) when the reaction was conducted in CH 2 Cl 2 or THF in the presence of Ti(OiPr) 4 (entry 10, 11). 4b,5a,6a,7b,10 In contrast to that observed for the TBS-ether (5b), reactions of the pivalate (5c) in the absence of additives gave the anti-adduct (2c) in poor ratios (entry 12–15); in fact, entry 15 even gave a marginal preference for the syn-adduct (ent-1c). Remarkably, the addition of Ti(OiPr) 4 was exceedingly effective in this series, and the reaction in THF solely gave the syn-adduct (1c) as a single stereoisomer (entry 17), as proved by X-ray crystallography. 11 This marked tendency was also observed with the MPM-ether (5d) (entry 18), and all reactions were free from racemization. In summary, we have succeeded in the stereodivergent addition of the acetylide group to ketones (5) by simple Table 1. Reaction of ketones (5) with ethynyl Grignard reagent a entry ketone solvent additive b syn/anti c yield/% c 1 5b THF none 10 : 90 96 d 2 5b Et 2 O none 9 : 91 95 d 3 5b Toluene none 20 : 80 28 4 5b CH 2 Cl 2 none 15 : 85 71 5 5b Et 2 O ZnBr 2 13 : 87 20 6 5b CH 2 Cl 2  ZnBr 2 21 : 79 12 Et 2 O(2:1) 7 5b Et 2 O MgBr 2  OEt 2 19 : 81 10 8 5b CH 2 Cl 2  MgBr 2  OEt 2 21 : 79 11 Et 2 O(2:1) 9 5b CH 2 Cl 2 MgBr 2  OEt 2 18 : 82 63 10 5b CH 2 Cl 2 Ti(OiPr) 4 80 : 20 76 d 11 5b THF Ti(OiPr) 4 93 : 7 90 12 5c THF none 47 : 53 81 13 5c Et 2 O none 28 : 72 78 14 5c Toluene none 34 : 66 42 15 ent-5c CH 2 Cl 2 none 55 : 45 97 d 16 5c CH 2 Cl 2 MgBr 2  OEt 2 43 : 57 73 17 5c THF Ti(OiPr) 4 syn only 93 d 18 5d THF Ti(OiPr) 4 syn only 75 d a Reactions were performed at 78  C to 10  C for 2–4 h in entry 1–9 and entry 12–16, but at 78  C to room temperature for 1–2 h in entry 10, 11 and entry 17, 18. Ethynylmagnesium bromide (1.5–4.0 equivalents, 0.5 M solution in THF) purchased from Aldrich chemical co., inc. was used. b 1.5– 4.0 equivalents of additives were used. c Ratio of syn-(1)/anti- (2) and yield were determined on crude by 200 MHz 1 H-NMR analysis, unless noted otherwise. d Isolated yield after column chromatography. 300 Chemistry Letters 2002 Copyright Ó 2002 The Chemical Society of Japan