New aspects of glycoside bond formation*

Glycoside bond formation generally requires activation of the sugar at the anomeric center. To this end, anomeric oxygen exchange reactions, resulting in the Koenigs±Knorr method and variations or, alternatively, activation through retention of the anomeric oxygen, resulting in the trichloroacetimidate method and in the phosphite method, have been proposed. The successful application of the trichloroacetimidate method to the total synthesis of GPI anchors is particularly worth mentioning. a[2-3]-Sialylation can be based on sialyl phosphites as glycosyl donors and on the nitrile effect for anomeric stereocontrol. This is exhibited for a preparative synthesis of ganglioside GM2 which is required for tumor vaccine studies. For the generation of a[2-8]-linkage between neuraminic acid residues anchimeric assistance by a 3-thiocarbonyloxy group is introduced. The required sialyl donor can be ef®ciently prepared and a-linkage to 8-O-unprotected neuraminic acid derivatives is almost quantitative. The limitations of chemical glycopeptide synthesis encourage to employ the protein biosynthesis machinery in combination with an expanded genetic code. This is exhibited for the synthesis of glycosylated hARF-1 protein. GLYCOSIDE BOND FORMATIONÐGENERAL ASPECTS Glycoside bond formation in order to gain chemically de®ned oligosaccharides and glycoconjugates remains an important task, though generation of the anomeric linkage has seen years of dynamic progress, mainly based on highly reactive glycosyl donors, on versatile building block strategies, and on advanced protective-group design [1±3]. The endeavour was stimulated by the eminent role of carbohydrates and especially of glycoconjugates in various ®elds of modern biology [4,5]. A particularly prominent area in this regard is glycoconjugate synthesis. Activation of the sugar residue through exchange of the anomeric hydroxy group by bromine and chlorine, respectively, has led to the well-known Koenigs±Knorr method (Scheme 1). Thus an a-haloether is generated that can be readily activated in the glycosylation step by halophilic promoters, typically heavy-metal salts, thus resulting in an irreversible glycosyl transfer to the acceptor. This valuable method has been extensively reviewed [1.3]. It has been continually developed and widely applied. In spite of its generality, the requirement of at least an equimolar amount (often up to 4 eq) of metal salt as promoter (often incorrectly termed `catalyst') and the problems concerning disposal of waste material (e.g. mercury salts) could be limiting factors for large-scale preparations. Therefore, alternative methods are of great interest. Other anomeric oxygen-exchange reactions in the activation step have been quite extensively investigated. Closely related to the classic Koenigs±Knorr method is the introduction of ̄uorine as leaving group (see Scheme 1, Xˆ F) [6] which, owing to the stability of the C±F bond, leads to stable glycosyl donors. Because of the generally lower glycosyl donor properties and because also at least equimolar amounts of promoter are required, the ̄uorides exhibit no real advantages over the corresponding glycosyl bromides or chlorides. *Lecture presented at the 19th International Carbohydrate Symposium (ICS 98), San Diego, California, 9±14 August 1998, pp. 719±800. 2Corresponding author. Anomeric oxygen-exchange reactions by thio groups (see Scheme 1, Xˆ SR) have recently attracted considerable attention for the generation of glycosyl donors [1,7]. Thioglycosides offer suf®cient temporary protection of the anomeric center, thereby enabling various ensuing chemical modi®cations of the glycosyl donor without affecting the anomeric center. Additionally, they present several alternative possibilities for the generation of glycosyl donor properties: besides various thiophilic heavy-metal salts, also iodonium, bromonium, and chloronium ions are highly thiophilic; yet, at least equimolar amounts are required and with counterions, such as bromide and chloride, halogenoses are formed. Therefore, a poor nucleophile is required as counterion, for instance use of N-iodoor N-bromosuccinimide or tri ̄uoromethanesulfonic acid, in order to enable a reaction with the acceptor as nucleophile. Obviously, the basic drawbacks of the activation through anomeric oxygen exchange reactions are also associated with this promoter system. Activation through retention of the anomeric oxygen. The requirement for glycoside synthesisÐhigh chemical and stereochemical yield, applicability to large-scale preparations, with avoidance of large amounts of waste materials, by having a glycosyl transfer from the activated intermediate through a catalytic processÐwere not effectively met by any of the methods described in the foregoing for the synthesis of complex oligosaccharides and glycoconjugates. However, the general strategy for glycoside bond formation seems to be correct: 1 The ®rst step should be activation of the anomeric center under formation of a stable glycosyl donor (activation step). 2 The second step (glycosylation step) should consist of a sterically uniform, high yielding glycosyl transfer to the acceptor; it should be a catalytic process, where diastereocontrol is derived from the glycosyl donors' anomeric con®guration (inversion or retention), or anchimeric assistance, in ̄uence of the solvent, thermodynamics, or any other effects. This led us to the concept of sugar activation through retention of the anomeric oxygen (Scheme 1). This concept is based on simple base treatment of sugars, thereby generating from a pyranose or a furanose at ®rst an anomeric oxide structure. Immediate alkylation (or arylation) leads to the anomeric O-alkylation (O-arylation) method for glycoside bond formation, which turned out to be highly 730 R. R. SCHMIDT et al. q 1999 IUPAC, Pure Appl. Chem. 71, 729±744 Scheme 1 valuable for the synthesis of various glycosides [1,8]. However, extension of this methodology to the synthesis of glycoconjugates is hampered by the limited generality. As an alternative the anomeric oxide can be used to generate glycosyl donors by addition to appropriate triple bond systems A;B (or cumulenes AˆBˆC or by condensation with Z±AˆBH systems, where Z represents a leaving group). The most successful methods developed thus far using these types of reactions are trichloroacetimidate (see Scheme 1 D, ±AˆBH ˆ ±C(CCl3)ˆNH [1±3,8] and phosphate and phosphite formation (AˆBHˆ PO(OR)2, P(OR)2) [3,9], also sulfoxide formation has been investigated [10]. The analogous glycosyl sulfates, sulfonates, and sul®tes have not yet been as successful and not as extensively investigated. All these methods are particularly tempting because nature has a similar approach for generating glycosyl donors, namely glycosyl phosphate formation in the activation step and (Lewis) acid catalysis in the glycosylation step. For the trichloroacetimidate method, the activation step consists of a simple base-catalyzed addition of the anomeric hydroxy group to trichloroacetonitrile, and the glycosylation step requires only catalytic amounts of a simple (Lewis) acid for the generation of strong glycosyl donor properties, thus leading to the desired glycosides in an irreversible manner. The water liberated on glycoside bond formation is thereby transferred in two separate steps to the activating species A;B ˆ CCl3CN under formation of stable, nonbasic trichloroacetamide (see Scheme 1) providing the driving force for the glycosylation reaction. Thus, a very economic and ef®cient glycosylation procedure is available. Because of the very low basicity of the liberated trichloroacetamide, the (Lewis) acid required for activation of the basic Oglycosyl trichloroacetimidate is released and is ready for further activation of unreacted glycosyl donors. This is exhibited in Scheme 2 for the formation of N-acetyllactosamine from protected O-galactosyl trichloroacetimidate as donor and 4-O-unprotected N-acetylglucosamine as acceptor [11]. The process very much resembles enzymatic N-acetyllactosamine generation [12]; however, the trichloroacetimidate-based process is obviously more simple, although protection and deprotection steps have to be taken into account. RECENT APPLICATIONS OF THE TRICHLOROACETIMIDATE METHODÐGPI ANCHOR SYNTHESIS The general signi®cance of O-gycosyl trichloroacetimidates lies in their ability to act as strong glycosyl donors under relatively mild acid catalysis. This has been overwhelmingly con®rmed in various Glycoside bond formation 731 q1999 IUPAC, Pure Appl. Chem. 71, 729±744 Scheme 2 laboratories, and the scope and limitations of this method can be readily derived from these investigations [1±3,13,14]. Some representative examples of the successful application of the trichloroacetimidate method to various important glycoside bond formations from our group are the synthesis of: X glycosphingolipids of the lacto-, the lactoneo-, the globo-, and the ganglio-series, X various glycopeptides, X glycosyl phosphatidyl inositol (GPI) anchors,