This thesis is concerned with calculating the thermodynamic properties of proteins dissolved in aqueous electrolyte solutions. These solutions are commonly used for the preliminary purification step of target protein molecules from fermentation broth...
This thesis is concerned with calculating the thermodynamic properties of proteins dissolved in aqueous electrolyte solutions. These solutions are commonly used for the preliminary purification step of target protein molecules from fermentation broths due to the selectivity and low cost of the process. In addition, salt-induced protein crystallization is a popular method for obtaining high quality crystals suitable for x-ray diffraction. To optimize these processes, we need to develop specific criteria for choosing conditions favorable for reducing the solubility of the target protein. Here, we seek to develop a molecular-thermodynamic description of protein solubility.
Most molecular-thermodynamic models of aqueous protein solutions are based on a potential of mean force that is an effective protein-protein interaction where the positions of solvent molecules are averaged out. To model protein-protein interactions in concentrated salt solutions, we propose a potential of mean force analogous to the solvation potential used in protein folding. The potential is given by the free energy to desolvate the part of the protein surfaces buried by the protein-protein interaction. In this work, we measure osmotic second virial coefficients for ovalbumin, for lysozyme, and for a mutant lysozyme, as a function of salt concentration and pH. By fitting the proposed solvation potential of mean force to the measured osmotic second virial coefficients, we show that the primary effect of salt is to enhance hydrophobic forces between protein molecules.
Protein solubility can be determined from simplified forms of the potential of mean force. To demonstrate this, we use an anisotropic potential of mean force determined from fitting osmotic second virial coefficients to correlate protein solubility. We extrapolate this potential to determine the solid-phase properties where we also include a term to account for the entropy change of crystallization. Quantitative agreement between the model and experimental data is obtained.
Potential-of-mean-force models are cast in the McMillan-Mayer framework where the independent variables are temperature, protein concentration and the set of solvent chemical potentials. This is in contrast to the experimentally-friendly variables of the Gibbs framework (temperature, pressure, protein concentration, and salt molality). To match experimental data, the results of the models need to be converted to the Gibbs framework. Here, we perform the conversion for a sample calculation of protein cloud-point temperature curves.