Prediction and validation of monoclonal antibodies separation in aqueous two-phase system using molecular dynamic simulation

Abstract

In order to predict how mAbs partition in 20% ethylene oxide/80% propylene oxide (v/v) random copolymer (EO20PO80)/water aqueous two-phase system (ATPS), a molecular dynamic simulation model was developed using Gromacs and then validated by experiments. The ATPS was applied with seven kinds of salt, including buffer salt and strong dissociation salt that were commonly employed in the purification of protein. Na2SO4 was shown to have the best effects on lowering EO20PO80 content in the aqueous phase and enhancing recovery. The content of EO20PO80 in the sample solution was decreased to 0.62%±0.25% and the recovery of rituximab increased to 97.88%±0.95% by adding 300 mM Na2SO4 into back extraction ATPS. The viability determined by ELISA was 95.57% at the same time. A strategy for constructing a prediction model for the distribution of mAbs in ATPS was proposed in consideration of this finding. Partition of trastuzumab in ATPS was predicted by the model created using this method and the prediction result was further validated by experiments. The recovery of trastuzumab reached 95.63%±2.86% under the ideal extraction conditions suggested by the prediction model.