(606d) Predicting Protein Solubility and Crystallization Behavior Based On the Second Osmotic Virial Coefficient
AIChE Annual Meeting
2013
2013 AIChE Annual Meeting
Separations Division
Crystallization of Pharmaceutical and Biological Molecules II
Thursday, November 7, 2013 - 9:45am to 10:10am
The biocatalytic production of complex pharmaceutical proteins (e.g. monoclonal antibodies) has gained an increased industrial interest. Within these processes, the total production costs are often dominated (up to 80 %) by the downstream processing. In state-of-the-art engineering, protein purification is usually achieved by a series of cost-intensive chromatographic steps. One interesting alternative to the conventional processing is protein precipitation, as a first capture step, and crystallization for final product polishing.
In order to develop a precipitation or crystallization step, the solubility of the target protein in aqueous solutions and the type of solid formed (crystal or precipitate) are required as crucial information. In order to supply this data, within this work, an innovative and simple solubility model to measure and predict the protein solubility and crystallization behavior based on the second osmotic virial coefficient (B22) is developed. By use of a potential-of-mean-force model (DLVO) with parameters fitted to B22 data at low precipitant concentration, B22 at high precipitant concentrations can be efficiently predicted, thus decreasing the experimental effort. By relating B22 to the asymmetric activity coefficient, the solid-liquid-phase equilibrium of a protein in solution, consisting of solvent, protein, and precipitant, can be calculated. This combination now enables for the prediction of the protein solubility over a broad concentration range of precipitant with a minimal set of experimental data.
The value of B22 herein serves as an ideal measure, as it combines information on the crystallization or precipitation behavior of the protein under the applied conditions, e.g. temperature, pH, type of precipitant used (salt or polymer e.g. PEG) and precipitant concentration.
In order to show the applicability of the model for varying protein sizes, experimental data of B22 for two proteins (14 and 144 kDa) in water at varying conditions were measured by static light scattering. The results were compared to experimental and literature data. The results point to the possibility of qualitative and quantitative prediction of the protein solubility in aqueous solution based on two B22 and two solubility data at varying precipitant concentrations. This modeling strategy supports a cost-efficient development of crystallization steps.