Glycomics : integration of lectin and gene expression microarray data

Glycomics is the systematic study of glycosylation in the context of a whole cell or organism. Glycosylated proteins are estimated to make up 50% of all proteins and cover the outside of the cell. Functional roles in glycosylation have been noted in pathogenesis, metastasis, and embryogenesis. However, the structure of these carbohydrates has been difficult to study due to the chemical nature of carbohydrates. Lectins, carbohydrate binding proteins excluding antibodies and enzymes, can be utilized to study glycosylation in a high throughput manner using a microarray format. Glycans, the carbohydrates attached to a protein or lipid, are not synthesized from a template. They are added co- or post-translationally by a concerted set of enzymes in the secretory pathway. In addition, the glycan structures may be altered by metabolism or trafficking. Cell type specific glycosylation has long been hypothesized due to observations of bacteria homing to tissues. We use lectin microarray technology to define the glycosylation in a subset of the NCI-60, a set of cell lines from different tissues. Using a customized gene expression microarray, we identify cell type dependent glycosylation genes and observe evidence of cell type dependent spliceforms for an O-glycosylated mucin. Data from the lectin microarray and a published gene expression data set were integrated using Generalized Singular Value Decomposition (GSVD), a linear matrix decomposition method. We have successfully decomposed the data into 3 cell type dependent meta patterns that segregate by glycosylation family. Correlation projection of the genes and subsequent gene ontology enrichment suggests that genes in different pathways covary with the types of glycosylation. An inverse relationship was revealed for the N- glycosylation pattern between the SVD of the lectins and the GSVD of the genes and lectins together. Whereas, the relationship was correlative for O-glycosylation, which was clearly illustrated in biplots. This work argues that types of glycosylation are regulated by different mechanisms in different cell types.