Modeling partitioning of proteins in aqueous two-phase system

Advances in biotechnology have made possible the production of a large number of new proteins important to the food, pharmaceutical, and chemical industry. The significance of aqueous two-phase systems (ATPSs) in downstream processing with commercial application has been reported for the separation of a large number of biopharmaceutical products. For both industrial- and laboratory-scale purposes it is essential to know the partitioning behaviour of biomolecules and which parameters lead to an optimal separation [1]. ATPS process design and implementation at a large scale is limited by the poor understanding of the molecular mechanism behind the solute partitioning. A better understanding of the molecular mechanism of the factors, which govern partitioning, contributes for the prediction of the partitioning behaviour in ATPSs. The approach is based on a thermodynamic cycle where the polar and nonpolar contributions to the free energy of transfer are assumed to be additive. The total free energy of transfer was experimentally obtained by determination of the proteins partitioning coefficient. Electrostatic energy contributions were estimated using a Poisson-Boltzmann computational method. The nonpolar contribution was obtained as a difference between the total free energy of transfer and the electrostatic contribution and related with the solvent accessible surface area. This methodology was already successfully applied to the partitioning of amino acids in ATPSs [2]. In this study partitioning of a 14 globular proteins, was examined in 4 different polymer/polymer ATPSs. The model performance was analysed regarding to the influence of different PDB structure and protein charges. Nonpolar energy contribution was related with the solvent accessible surface area, by linear relationship. Using a semi-empirical model, it was possible to predict the partition coefficient of several proteins in four different polymer/polymer systems. The model shows improved prediction capacity when applied to the group of protein with the same electrostatic characteristic (positive or negative charge proteins).