To date, the detailed structure-function studies on glycosyltransferases, in particular on β 1,4-galactosyltransferase, have shown that in the vicinity of their catalytic pocket are one or two flexible loops that, upon binding of the nucleotide sugar donor substrate and the metal ion (when required as a cofactor), change conformation often from open to closed, creating an acceptor binding site. The loop acts as a lid covering the bound donor substrate. After the transfer of the glycosyl unit to the acceptor, the product is ejected, and the loop reverts to its native conformation to release the remaining nucleotide moiety. To diversify the catalytic activity towards less preferred substrates, such as sugar acceptors or proteins or lipids or aglycons, the catalytic domains of glycosyltransferases either interact (1) with an additional protein, or (2) have acquired add-on domains at the C-terminous Defining and determining the structure of these add-on domains provide structural basis of understanding the specificities of glycosyltransferases towards specific acceptor substrates. The specificity of the sugar donor in many glycosyltransferases is determined by a few residues in the sugar-nucleotide binding pocket of the enzyme, which are conserved among the family members from different species. Mutating these residues can generate novel glycosyl-transferases with broader donor specificity. Several mutant glycosyltransferases have been generated that can transfer a sugar residue with a chemically reactive functional group to N-acetylglucosamine, galactose and xylose residues of glycoproteins, glycolipids and proteoglycans (glycoconjugates). These mutant glycosyltransferases allow us to modify the oligosaccharide moieties of glycoproteins so that they can be linked via glycan chains, thereby assisting in the assembly of bio-nanoparticles that are useful for the development of the targeted-drug delivery system and contrast agents for MRI.