Mathematical Modeling of Glycosylation Pathways

Glycans are highly abundant and diverse biological molecules that bind proteins and lipids constituting a dominant feature of the cell surface. As glycans play important roles in several complex human diseases, the need to understand their functions and predict their structures arises in both the biotechnology industry and in the clinic. Towards these goals, we are developing a broad based mathematical model that aids in representing the biosynthetic pathways for eukaryotic glycosyslation and predicting the impact of variations on the pathways while linking glycan structures to cellular processings.  In the past years our research has been focused in applying the model to modify glycoform distributions towards understanding how to enhance glycoproteins and identification of glycan structures from MALDI TOF mass spectra of normal and disease cells in terms of activities of the enzymes responsible for their processing. Lately, we have been focusing in using the model to link glycan structure distributions to genetic modifications to assess   how changes in the expression of glycosylation genes – now readily obtained through DNA microarray experiments – affect the highly diverse repertoire of glycan structures found on cells.

For example, one of our works (Krambeck and Betenbaugh, 2005) has focused on metabolic engineering of N-linked oligosaccharide biosynthesis to predict glycoform distributions of recombinant human thrombopoietin (rhTPO) in CHO cells, the results were validated with experimental data (Inoue et al., 1999)- table 1. The model was used to analyze how glycan structures change as protein productivity is increased i.e. increase of total glycan concentration. In Figure 1, a general decrease in the degree of sialylation is observed (S2, S3, and S4) as the total glycan concentration increases suggesting a bottleneck in the sialylation processing.

The capability of the model to make connections between glycan structure and metabolite levels is critical to make predictions. For example if a target glycan structure is desired,  this can be increased, then the model will allow to predict what are the effects of increasing such structure over other structures, and which are the metabolites associated to that change. This information has significant utility in development of glycoprotein pharmaceuticals, where specific glycan structures confer desired properties to proteins, making them effective to treat diseases.

As envisioned this model could be not only used to facilitate development of pharmaceuticals with improved bioactivity but also to discover glycan structures or glycan patterns associated with unhealthy cells.

References
Krambeck, F. J. and M. J. Betenbaugh (2005). "A mathematical model of N-linked glycosylation." Biotechnol Bioeng 92(6): 711-28

Inoue, N., T. Watanabe, et al. (1999). "Asn-linked sugar chain structures of recombinant human thrombopoietin produced in Chinese hamster ovary cells." Glycoconjugate Journal 16(11): 707-718.