Metabolic glycoengineering for the study of neurodevelopmental and neurodegenerative processes

Abstract

The composition of cell surface carbohydrates varies among different cell types and reflects cellular physiological states. Studying cell surface glycosylation therefore serves as an important quality check of functional biosynthetic pathways. The outermost component of surface glycans is often sialic acid (Sia). This negatively-charged sugar is particularly abundant in the brain, where it regulates neurite development, regeneration and synaptic plasticity. Another specific feature of the brain is the high plasma membrane content of Sia-containing lipids – the gangliosides. Many functions of neuronal Sia are yet to be fully explored. Metabolic glycoengineering (MGE) is a chemical biology method that allows the visualization of certain sugar residues, including Sia, on the cell surface. However, its use in live neurons remains challenging. This thesis, consisting of three manuscripts, focused on MGE approaches to elucidate Sia roles and functions. Manuscript 1 focused on the development of an image analysis program for Superimposing Key Regions (SUIKER). The program was optimized for quantification of colocalized objects within superimposed structures such as complex neurite networks. Based on a set of training images, we showed the ability of the program to simultaneously quantify neurite-specific MGE Sia as well as neuronal viability parameters. SUIKER may also contribute to other spatial co-occurrence studies such as, for instance, assessment of a protein within cell membranes, or larger organelles. Manuscript 2 investigated the metabolic incorporation of an azide-tagged sugar precursor (Ac4ManNAz) into the glycans of mature human neurons (LUHMES) as a potential early functional neurotoxicity readout. We established and optimized MGE neuronal labeling with improved time resolution (from >24 h to 6 h). The SUIKER program was used to quantify viability, neurite integrity and MGE Sia incorporation into neuronal projections after toxicant exposure. Testing the effect of several model compounds, we identified a subgroup (mitochondrial respiratory chain inhibitors) that affected the incorporation of MGE Sia into neurites at much earlier time points (1 h) compared to other viability assays (24 h). The ‘NeuroGlycoTest‘, which was based on these findings, was found to identify subtle neurotoxicity at earlier time points than other test endpoints. In Manuscript 3, we hypothesized that the high content of Sia-lipids in the brain would cause the MGE signal on neurons to be of little specificity for glycoproteins. We examined the effect of compounds affecting glycoproteins on the MGE Sia signal of neurons. Deoxymannojirimycin (dMM), a non-toxic protein glycosylation inhibitor, and N-butyl-deoxynojirimycin (NBdNM) a well-tolerated lipid glycosylation inhibitor, were used to develop a method of differential labeling of sialylated membrane lipids or sialylated N-glycosylated proteins on live neurons. We applied the selective inhibition approach to several cell types and found that in dopaminergic or sensory neurons >60% of Sia is lipid-bound. Different from neurons, most Sia was bound to protein in hepatoma cells or in neural crest cells. The method allowed the visualization of cell-specific sialylation processes for separate classes of membrane constituents. Altogether, the established MGE-based methods contribute to the optimization of neuronal sialylation studies. They allow more sensitive detection of neurotoxicants and point the way towards specific sialoglycan observations on live cells.