Prediction of glycosylation across the human proteome and the correlation to protein function

摘要

The addition of a carbohydrate moeity to the side-chain of a residue in a protein chain influences the physicochemical properties of the protein. Glycosylation is known to alter proteolytic resistance, protein solubility, stability, local structure, lifetime in circulation and immunogenicity. Of the various forms of protein glycosylation found in eukaryotic systems, the most important types are N-linked, O-linked GalNAc (mucin-type) and O-β-linked GlcNAc (intracellularuclear) glycosylation. N-linked glycosylation is a co-translational process involving the transfer of the precursor oligosac-charide, GlcNAc_2Man_9Glc_3, to asparagine residues in the protein chain. The asparagine usually occurs in a sequon Asn-Xaa-Ser/Thr, where Xaa is not Proline. This is however, not a specific consensus since not all such sequons are modified in the cell. O-linked glycosylation involves the post-translational transfer of an oligosaccharide to a serine or threonine residue. In this case, there is no well-defined motif for the acceptor site other than the near vicinity of proline and valine residues. We have developed glycosylation site prediction methods for these three types of glycosylation, using artificial neural networks that examine correlations in the local sequence context and surface accessibility. In this paper, we have used glycosylation site information on human proteins to illustrate the contribution of glycosylation to protein function and assess how widespread this modification is across the human proteome.