Systems analysis of N-glycan processing in mammalian cells

Abstract

N-glycosylation plays a key role in the quality of many therapeutic glycoprotein biologics. The biosynthesis reactions of these oligosaccharides are a type of network in which a relatively small number of enzymes give rise to a large number of N-glycans as the reaction intermediates and terminal products. Multiple glycans appear on the glycoprotein molecules and give rise to a heterogeneous product. Controlling the glycan distribution is critical to the quality control of the product. Understanding N-glycan biosynthesis and the etiology of microheterogeneity would provide physiological insights, and facilitate cellular engineering to enhance glycoprotein quality. We developed a mathematical model of glycan biosynthesis in the Golgi and analyzed the various reaction variables on the resulting glycan distribution. The Golgi model was modeled as four compartments in series. The mechanism of protein transport across the Golgi is still controversial. From the viewpoint of their holding time distribution characteristics, the two main hypothesized mechanisms, vesicular transport and Golgi maturation models, resemble four continuous mixing-tanks (4CSTR) and four plug-flow reactors (4PFR) in series, respectively. The two hypotheses were modeled accordingly and compared. The intrinsic reaction kinetics were first evaluated using a batch (or single PFR) reactor. A sufficient holding time is needed to produce terminally-processed glycans. Altering enzyme concentrations has a complex effect on the final glycan distribution, as the changes often affect many reaction steps in the network. Comparison of the glycan profiles predicted by the 4CSTR and 4PFR models points to the 4PFR system as more likely to be the true mechanism. To assess whether glycan heterogeneity can be eliminated in the biosynthesis of biotherapeutics the 4PFR model was further used to assess whether a homogeneous glycan profile can be created through metabolic engineering. We demonstrate by the spatial localization of enzymes to specific compartments all terminally processed N-glycans can be synthesized as homogeneous products with a sufficient holding time in the Golgi compartments. The model developed may serve as a guide to future engineering of glycoproteins.

Figure 1. Glycan reaction network and its symbolic representation.

A: Schematic of the formation of the terminally processed N-glycans; dashed lines denote multiple reaction steps. B: The corresponding network representing the various pathways connecting the glycans shown in A; least, early, late, and terminally processed glycans denoted on the network, as plotted by GlycoVis. C: The starting glycan and the most fully processed N-glycan considered in the model, with sugar linkages labeled for reference.

Figure 2. Schematic of the Golgi as a system of four reactors in series.

Figure 3. Pathway maps with reactions highlighted according to the enzyme catalyzing the reaction.

Figure 4. Percentage of the terminally processed N-glycans after adjusting the enzyme levels in a single PFR model (dashed line corresponds to a completely equal mixture).

Figure 5. Effect of normalized distance (holding time) on glycan flux in a PFR model.

Figure 6. Visualization of the glycan profile and relative reaction rates as a function of normalized distance.

A,B: 0.25 C,D: 0.50 E,F: 1.0

Figure 7. Sensitivity coefficients of the 12 terminal glycans with respect to the ten enzymes.

Figure 8. The effect of GalT amplification on N-glycan fluxes.

A: Terminal glycan fluxes with basecase GalT concentration. B: Terminal glycan fluxes with 10X higher GalT concentration. C: Controlling branch point in N-glycan biosynthesis. D: Comparison of fluxes at branch point shown in C, at different GalT concentrations, and normalized distances. E: Ratio of fluxes at the branch point, at different GalT concentrations. F: Fluxes of all GalT catalyzed reactions, at different GalT concentrations and normalized distances.

Figure 9. Competing reactions utilizing glycan 15.

Figure 10. Comparison of N-glycan fluxes between 1PFR and 4PFR models.

A: Terminal glycan fluxes. B: Nonterminal glycan fluxes.

Figure 11. Comparison of N-glycan fluxes between 4PFR and 4CSTR models.

A: Terminal glycan fluxes with a 40 min residence time. B: Terminal glycan fluxes with a 20 min residence time. C: Nonterminal glycan fluxes with a 40 min residence time; least, early, and late processed glycans as shown in Figure 1B. D: Nonterminal glycan fluxes with a 20 min residence time; least, early, and late processed glycans as shown in Figure 1B.

Figure 12. Active and divergent reactions in each compartment for the generation of uniform glycan 99.

A: Cis Golgi, B: Medial Golgi, C: Trans Golgi, D: TGN Golgi. Arrows: blue for reactions with active flux; gray for inactive fluxes; red for divergent fluxes.