Sialic acids--why always alpha-linked?

Abstract

A variety of naturally occurring oligosaccharides are found in glycoproteins, glycolipids, and glycosaminoglycans. The elucidation of their cellular functions is an expanding area of research in glycobiology. Especially glycoconjugates of the cell surface containing sialic acid have been implicated in biological recognition (Zeller and Marchase, 1992; Varki, 1993). While other monosaccharides can occur in both anomeric configurations, a or p, the sialic acids found so far in mammalian glycolipids and glycoproteins are only a-configured. This corresponds to an equatorial orientation of the anomeric C-O-bond in the sialic acid residue. In this communication we correlate this observation with a hypothesis about the mechanism of sialyltransferases. The biosynthesis of glycoconjugates requires the action of glycosyl transferases that catalyze the transfer of a monosaccharide unit from a glycosyl donor to an appropriate acceptor molecule. The monosaccharide is activated either as a lipid or a nucleotide sugar. The most common nucleotide-activated sugars in mammals are the nucleoside diphosphates UDP-otD-glucose (UDP-Glc), UDP-a-D-galactose (UDP-Gal), UDPN-acetyl-a-D-glucosamine (UDP-GlcNAc), UDP-N-acetyl-aD-galactosamine (LTDP-GalNAc), UDP-a-D-glucuronic acid (UDP-GlcA), UDP-ot-D-xylose (UDP-Xyl), GDP-P-L-Fucose (GDP-Fuc), GDP-a-D-Mannose (GDP-Man), and the nucleoside monophosphate CMP-p-sialic acid. The sialic acid moiety occurring in CMP-sialic acid is a N-acetyl neuraminic acid residue (CMP-NeuAc), but N-glycolyl derivatives are also found (Schauer et al., 1995). Activated sialic acids, the CMPsialic acids, are unique among the glycoconjugates in that the sialic acid is P-glycosidically linked to the aglycon CMP (Haverkamp et al., 1979). In contrast to many other enzymes, neither the three-dimensional structure nor the mechanism of glycosyltransferases are known. The crystal structure of a DNA modifying fJ-glucosyltransferase has been solved, but conclusions on the transferase mechanism or the role of divalent metal ions cannot be drawn from the data (Vrielink et al, 1994). Mechanistic information is only available from the analysis of enzyme kinetics (Kadowaki and Grant, 1994). It can be expected, however, that transferase activity is mechanistically related to the glycosidase reaction with the difference that the sugar residue is not transferred to a water molecule but to the nucleophilic center of another glycosyl acceptor. In both cases it can be assumed that a carboxonium-ion intermediate is stabilized by the enzyme (Sinnott, 1990). Though nucleotide activated monosaccharides occur only in one configuration each, monosaccharides in other glycoconjugates are found in both anomeric configurations, a and p. With the exception of GDP-fucose, the anomeric oxygen in nucleotide activated monosaccharides is in an axial orientation. Fucose is generally found in the a-configuration, but also pglycosidically linked fucosides are known (Flowers, 1981). In contrast, the sialic acid residues observed so far (with the exception of CMP-sialic acid) are all a-glycosidically linked. We assume that the reason for this exclusive configuration might be based on the specificity of the sialyltransferases (Tsuji, 1996) that utilize CMP-sialic acids as glycosyl donor. The other glycosyl donors mentioned above contain a diphosphate building block which serves as ligand for divalent metal ions at the active site of the glycosyl transferases (Tsoponakis and Herries, 1978; Murray et al, 1996). The biochemistry of nucleoside diphosphates in general and especially of phosphoryl transfer reactions depends on the presence of divalent metal ions (Cherfils et al, 1994; Strater et al, 1996). In contrast to the other nucleotide-activated sugars, CMP-sialic acid occurs as a monophosphate. Since sialic acid and sialic acid residues bear a negative charge under physiological conditions, it can be expected that both the carboxylate and phosphate groups of CMP-sialic acid occur complexed to a divalent metal ion forming a seven member chelate complex (Schmidtbaur et al, 1990). The consequence might be—by comparison to other nucleotide-activated sugars—that in the active site of sialyltransferases the sialic acid residue and the leaving group are linked by the metal ion. Therefore, independent of the glycosylation mechanism the CMP leaving residue should prevent nucleophilic attack of the aglycon from the p-direction and thereby exclude the formation of P-ketosidically linked sialic acid residues. The hypothesis offered here is not supported by few reports about preparations of sialyltransferases with no apparent divalent metal ion requirement or with no inhibition by EDTA (Beyer ef al, 1981). However, the presence of low concentrations of tightly bound metal ions in these preparations cannot be ruled out It is also conceivable that in some sialyltransferases a cationic amino acid side chain substitutes functionally for the divalent metal ion. The correlation of a-configuration and monophosphateactivated sugar is supported by the observation that in glycoconjugates the structurally related sugars, Kdn (2-keto-3deoxy-D-glycero-D-galacto-nonulosonic acid) (Wilson et al., 1996) and Kdo (3-deoxy-D-manno-2-octulosonic acid) (Lind-