Identifying glycan motifs using a novel subtree mining approach

摘要

As one of the four major classes of biomolecules, carbohydrates are present in all organisms and play crucial roles in biomolecular interactions. Organisms polymerise simple sugars to yield oligo- and polysaccharides, which are typically termed glycans when attached to proteins and lipids. Glycans may be composed of several sugar residues with various glycosidic linkages, often forming branched structures. Consequently, there are a myriad of glycan structures that have arisen in organisms, with distinct glycosylation patterns observed between evolutionary clades. Glycoforms can even differ between individuals. Aberrant glycosylation is a hallmark of cancer, and a body of research has focused on the identification of glycan biomarkers as diagnostic and prognostic tools for use in oncology [ , ]. Additionally, carbohydrate determinants are frequently involved in host–pathogen interactions. Notable examples of this include the attachment of influenza virions to host sialic acid residues and the recognition of pathogens by mannose receptors and anti-carbohydrate antibodies [ , ]. The mannose receptor, along with DC-SIGN, is an example of a C-type lectin present on the surface of immune cells. Lectins can be defined as ‘proteins that possess at least one noncatalytic domain that binds reversibly to a specific mono- or oligosaccharide’, excluding enzymes (e.g. glycosyltransferases) and carrier proteins [ ]. Due to their broad selectivities, lectins are also distinct from other glycan-binding proteins that recognise specific carbohydrate antigens, such as antibodies and T-cell receptors. The carbohydrate-binding properties of plant lectins have been exploited by scientists for a number of laboratory techniques, including histochemical staining, affinity chromatography, and identification of biomarkers. For example, agglutinin (LCA)-reactive -fetoprotein (a glycoform termed ‘AFP-L3’) is an FDA-approved biomarker for the risk assessment of hepatocellular carcinoma [ , ]. However, the selectivities of lectins for glycan motifs are often poorly defined, which undermines confidence in glycan profiling.