The structures of three nine-residue peptide substrates that show diff
erential kinetics of O-linked glycosylation catalyzed by distinct reco
mbinant uridine diphosphate-N-acetylgalactosamine:polypeptide N-acetyl
galactosaminyltransferases (GalNAc transferases) were investigated by
NMR spectroscopy. A combined use of NMR data, molecular modeling techn
iques, and kinetic data may explain some structural features required
for O-glycosylation of these substrates by two GalNAc transferases, Ga
lNAc-T1 and GalNAc-T3. In the proposed model, the formation of an exte
nded backbone structure at the threonine residue to be glycosylated is
likely to enhance the O-glycosylation process. The segment of extende
d structure includes the reactive residue in a beta-like or an inverse
gamma-turn conformation and flanking residues in a beta-strand confor
mation. The hydroxyl group of the threonine to be glycosylated is expo
sed to solvent, and both the amide proton and carbonyl oxygen of the p
eptide backbone are exposed to solvent. The exchange rate of the amide
proton for the reactive threonine correlated well with substrate effi
ciency, leading us to hypothesize that this proton may serve as a dono
r for hydrogen bonding with the active site of the enzyme. The oxygens
of the residue to be glycosylated and several flanking residues may a
lso be involved in a set of hydrogen bonds with the GalNAc-T1 and -T3
transferases.