Tlic first asymmetric synthesis of a phytosphingosine, (ZS,38,4R)-2-amino-1.3.4-hexadecanetriol (Z), was accomplishd by kinetic resolution and asymmetric cpoxidation. The absolute stereochcmislry of the phytosphingosine from starfishes (Amithasier plniici iind Astcrinrr prciin~jkrrr) was cstnblishcd to be the Samc as Tor 2. The phytosphingosinc anhydro basc 19 was also asymmctricaUy synthesized, and the structure was determined on the basis of NMR analyses involving 2D-COSY. NOESY, and difference NOE experimenls.In continuation of our study of biologically active glycosides from two starfishes, Acanthaster planci and Asterina pectinifera, we isolated 6 pure cerebrosides, acanthacerebroside A (l), B, C, D, E, and F2). Acid hydrolysis of 1 gave Dglucose, (2R)-2-hydroxytetracosanoic acid, and phytosphingosine (4-hydroxysphinganine) 2b,3). However, the relative or absolute structure of the phytosphingosine was not determined exactly because the optical rotation, the NMR chemical shifts, and the coupling constants were not described in all previous reports4'. Acanthacerebroside A Reverse-phase HPLC was successfully used for the isolation of the cerebrosides. In the case of the mixture of isomeric cerebrosides, each of which has the same molecular mass, the chromatographic separation was very difficult2).In order to determine the absolute configuration of the phytosphingosine and to develop synthetic routes leading to pure cerebrosides and more complicated glycosphingolipids such as gangliosides'), we started the asymmetric synthesis of phytosphingosine.Optically active sphinganine and sphingosine (Csphingenine) were already synthesized by use of an enzyme6), Synthesis of phytosphingosine (4-hydroxysphinganine) was only accomplished from a chiral precursor, ~-glucosamine~",~~), ~-g a l a c t o s e~~,~~) , and natural ~-erythro-sphingosine~~~~. The asymmetric synthesis of phytosphingosine has not been reported.We chose (2S,3S,4R)-2-amino-1,3,4-hexadecanetriol (2), the configuration of which is the most common among naturally occurring phytosphingosines, as the synthetic target. Our synthesis of focused on how chirality can be introduced into a synthetic intermediate at the initial step of the synthesis and on how high diastereomeric excess can be obtained by stereoselective and regioselective reactions. We utilized the Katsuki-Sharpless asymmetric epoxidation ' ' I to introduce the chirality into an intermediate. The epoxidation has two different characters. One is the asymmetric epoxidation of a primary allylic alcohol lla,llb) and the other is kinetic resolution of a secondary allylic alcohollld).Our synthetic design was as follows. The kinetic resolution gave the (4R)-allylic alcohol (4) and subsequent epoxidation gave the 3,4-anti-epoxy alcohol (14 and 15), diastereoselectively. The high diastereomeric excess (de) with the 3,4-anti configuration can be predicted in terms of a matched pair for double asymmetric induction by Masamune et al.'*) An amino group was introduced regioselectively by intraLiebigs Ann. Chem. ...