A new approach to mapping co-localization of multiple glycosyl transferases in functional Golgi preparations

We have developed a new method to co-localize multiple glycosyl transferases in different Golgi compartments. The approach relies on the proven ability of intact, sealed rat liver Golgi preparations to concentrate exogenous labeled sugar nucleotides into the lumen where they glycosylate either endogenous or artificial acceptors. The premise is that if two glycosyl transferases are co-localized within the same compartment, they will compete for the limited amount of transported donor. If the donor is consumed in glycosylating a permeable artificial glycoside within a Golgi compartment, it will be unavailable to glycosylate endogenous products within that same compartment. The greater the degree of transferase co-localization, the greater the potential decrease in glycosylation of endogenous acceptors. We provide an example consistent with these predictions. Adding 1 microM UDP[3H]Gal to Golgi preparations followed by a chase with a cocktail of unlabeled sugar nucleotides labels mostly endogenous N-linked oligosaccharides containing both beta 1,3- and beta 1,4[3H]Gal residues with and without sialic acid. Addition of increasing amounts of 4-methylumbelliferyl-beta-xyloside (Xyl beta MU) produces [3H]Gal1 beta, 4Xyl beta MU and leads to a reciprocal decrease in labeling of a restricted set of the endogenous acceptors. This decrease is preferential for [3H]Gal beta 1-->3GlcNAc beta 1-->R and, to a lesser extent, [3H]Gal beta 1-->4GlcNAc beta 1-->R structures in neutral and mono-sialylated oligosaccharides; synthesis of these structures in di- and tri-sialylated oligosaccharides was unaffected. These preferential decreases are not seen in detergent permeabilized, sugar nucleotide transport-independent Golgi incubations, and are not due to inhibition by the Gal beta 1,4Xyl beta MU product. These results argue that there is significant overlap in the functional co-localization of sialyl and galactosyltransferases in rat liver Golgi preparations and that GAG chain core specific Galactosyltransferase I is co-localized with subsets of N-glycan Gal beta 1,3 and Gal beta 1,4 transferases. This approach can be used with other glycosides and sugar nucleotides to map and co-localize other glycosyl transferases. The functional compartments defined by this approach may or may not correspond entirely with morphologically defined Golgi domains.