Predicting N-glycan processing based on enzyme-glycan accessibility

Session: 
S2.2 Protein N-glycosylation
Code: 
OL2.2.2
Location (hall): 
Mannose
Start/end time: 
Monday, July 1, 2019 - 15:00 to 15:15
Oliver
Grant

Oliver Grant1, Robert Woods1

1University Of Georgia, Athens, United States

Background:
In this work, computer simulation, glycoproteomics and crystallographic data are combined to show that glycoprotein glycoform distributions depend on the accessibility of N-glycans to the relevant glycosidases in the ER. We illustrate this for three systems: a protein disulfide isomerase precursor named Pdi1p, a hemagglutinin (HA) from influenza A, and the HIV envelope glycoprotein.

Methodology:
We leverage the recently solved 3D structure of ER mannosidase I (ERManI) with molecular dynamics simulations of the glycoproteins, where the enzyme’s substrate, Man9GN2, is present at each site. We calculate the percentage of simulation time that ERManI is physically able to bind the glycan, as it samples different shapes throughout the simulation.

Results:
In the case of Pdi1p, the correlation between accessibility to ERManI and the degree of processing is striking. The modeling also predicted that a domain deletion would expose a glycosylation site on a neighboring domain to processing, which was confirmed experimentally. For influenza HA, we were able to rationalize why certain sites remained as Man9GN2. Further, the modeling work was able to propose a 3D model for how the pulmonary collectin SP-D would bind these Man9GN2 glycans and thus neutralize the virus. When applied to the HIV envelope protein, our modeling approach provides insight into formation of the so-called “high-mannose patch”.

Conclusion:
The presentation will illustrate the degree to which site-specific glycan processing can be predicted on the basis of 3D structure. 

Sponsors

Sponsors