| Abstract: | The hypothesis which explains the diastereoselectivity of the 1,3-dipolar cycloaddition of the N-glycosylnitrones 1 – 3 leading to the 5,5-disubstituted isoxazolidines 4 – 6 on the basis of a kinetic anomeric effect predicts that nucleophiles should add to N-glycosylnitrones with a high degree of diastereoselectivity. To test this prediction, the nucleophilic addition of lithium and potassium dialkylphosphites to the crystalline (Z)-nitrone 11 , prepared from oxime 9 and (benzyloxy)acetaldehyde has been examined. The addition of lithium phosphites gave the N-glycosyl-N-hydroxyaminophosphonates 12 – 16 (d. e. 78–92%) in high yields (Scheme 4). The addition of potassium phosphites showed a much lower diastereoselectivity. Glycoside cleavage, hydrogenolysis, and dealkylation of 12 – 16 gave (+)-(S)-phosphoserine (+)- 19 (34–45% from 9 ). Its absolute configuration was confirmed by an X-ray analysis of the N-(3,3,3-trifluoro-2-methoxy-2-phenylpropionyl) derivative 24 . Similarly, the crystalline nitrone 25 gave the N-glycosyl-N-hydroxyaminophosphonate 26 , which was transformed into (+)-(S)-phosphovaline (+)- 31 (42% from 9 ). The diastereoselectivity of the nucleophilic addition and the enantiomeric purity of (+)- 31 were determined by the analysis of the derivative 30 (d.e. 92%) and 32 (d.e. 93%), respectively. The addition of lithium diethyl phosphite to the nitrone 33 , prepared in situ, gave the N-glycosyl-N-hydroxyaminophosphonate 34 , (41%; d.e. 91%), which was transformed in (+)-(S)-phosphoalanine (+)- 37 (21% from 9 ). |
