Mathematical model relating gene expression to glycoform structure

DESCRIPTION (provided by applicant): Glycans are highly variable and structurally diverse sugar chains that, when attached to membrane proteins and lipids, are the dominant feature of the mammalian cell surface. In many diseases, including cancer, the distribution of glycan structures often differs from normal cells leading to the enticing possibility that disease- specific glycan structures and their functions can be used diagnostically and therapeutically. Development of glycan-based clinical procedures, however, has been hindered by the difficulty of analyzing the thousands of oligosaccharide structures found on cells, which are usually present in minute quantities that make structural characterization a formidable challenge. These difficulties are compounded by the fact that, in contrast to protein structures, glycan structures are not directly encoded in the genome and no template is therefore available that would allow PCR-type amplification. Instead, the structure of glycans is determined by a complex interaction of many glycosylation enzymes with proteins and other metabolites. The objective of this project is to develop mathematical modeling technology to predict 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. Linking glycan structure distributions to genetic modifications or easily obtained gene expression data in this manner will greatly facilitate development of glycan-based diagnostic markers or therapeutic interventions; the goal of this project is to develop a computer model to predict disease-specific glycans and to make this technology commercially available to medical researchers. The project will primarily utilize publicly available experimental data sources on enzyme properties, glycogene expression levels and glycan profiles to develop and refine the mathematical model. Full advantage will be taken of web based compilations of these data in addition to literature resources. The project will also collaborate with ongoing experimental glycobiology programs at Johns Hopkins in order to evaluate the use of a prototype implementation interactively for research guidance. The technology includes both a qualitative and quantitative aspect. The qualitative aspect involves predicting the possible glycan structures that can result from a given starting structure and a given list of glycosylation enzymes. For example, the de novo appearance of one (or a small number) of structures is a powerful indicator of the malignant status of certain cancers. The quantitative aspect relates the abundances of the various glycan structures produced to the set of expression levels of the glycosylation enzymes; this level of sophistication will be valuable for assessing conditions such as prion disease where subtle differences in the abundances of dozens, or hundreds, of structures may be related to pathogenicity. The final product of the project would be a software system, made available to the public as a commercial software package to map the relationship between gene expression and glycan structure in either direction. The results of this project will provide a system of computer tools to help researchers determine new diagnostic methods for serious diseases, such as cancer, and develop new therapies for treating these diseases. The tools will make a link between fundamental cellular processes and the chemical structures displayed on cell surfaces, which control, for example, interactions between normal and malignant cells. The final product of the project would be a software system that contains both qualitative and quantitative tools to map the relationship between gene expression and glycan structure in either direction. The system would be made available to the public as a commercial software package.