(552d) Prediction of Intracellular Nucleotide Sugar Variation Impact on Monoclonal Antibodies (mAbs) Glycan Distribution Using Glycosylation Reaction Network Models

Prediction of
intracellular nucleotide sugar variation impact on monoclonal antibodies (mAbs)
glycan distribution using glycosylation reaction network models

Sha Sha^a and Seongkyu Yoon^a

^aDepartment of
Chemical Engineering, University of Massachusetts, Lowell, MA, USA

Glycosylation is a
critical post-translational process on monoclonal antibodies (mAbs) produced by
mammalian cells as the glycan profile on mAbs plays significant impact on the pharmaceuticals'
stability, efficacy and half-life. Glycosylation central reaction network (CRN)
in cellular Golgi compartment is the branching paths for mAb to be
attached with various intermediate or terminal glycan structures via the mediation
of a series of enzymes. The recent development of reaction network models is an
approach to simulate glycosylation process in Golgi CRN and predict the distribution
of glycan on final monoclonal antibodies (including a number of intermediate or
terminal glycan structures) through simulation using the model, and thus are
potentially of great value in the application in the industrial bioprocess. Reaction
network models are formulated by describing the kinetics of reactions in the
network and involve various intracellular factors associated with glycosylation
as model parameters. The capability of models being built to reflect key
intercellular factor variation is one of the priorities for models to ensure
proper performance. However, the performance of model prediction for glycan
distribution out from intracellular factor (e.g. nucleotide sugar) variation has
not been sufficiently evaluated in the past studies. In this work, a reaction
network model was developed, describing the CRN by resembling Golgi as a four-compartment
continuous stirred-tank reactor (CSTR) and the glycan distribution was
estimated by solving the model based on mass balance of all glycan structures
in the network at a steady state. The impact of intracellular nucleotide sugar
concentration variation on the glycan distribution output was then simulated
using the model and further compared with nucleotide sugar- glycan variation
correlation being examined by published literature. The model was then
optimized to minimize the gap between model prediction and experiments-derived
data, therefore becoming a more reliable tool to be utilized in the future for
glycosylation prediction.