Predicting the Solubility of Amino Acids and Peptides with the SAFT-γ Mie Approach: Neutral and Charged Models.

Modeling approaches that can be used to predict accurately the solubility of amino acids and peptides are of interest for the design of new pharmaceutical processes and in the development of new peptide-based therapeutics. We investigate the capability of the SAFT-γ Mie group-contribution approach to predict the aqueous and alcohol solubility of glycine, alananine, valine, leucine, and serine and of di- and tripeptides containing these amino acids. New SAFT-γ Mie group interactions are characterized using experimental thermodynamic and phase-equilibrium data of compounds and mixtures that contain groups relevant to the amino acids and peptides, but no solubility data (except for the case of glycine). Once all the group interaction parameters are developed, predictive solid-liquid solubility calculations are carried out. Neutral and charged models are considered to account explicitly for the zwitterionic nature of the molecules in aqueous solution, and the solubility of the solution is presented as a function of pH. A detailed discussion of the molecular models and Helmholtz free-energy expressions used to represent the ionic and zwitterionic forms of the amino acids, together with their speciation in solution is also provided. Overall, very good agreement with available data is shown, with an absolute average deviation (AAD) in mole fraction of 0.0038 over 283 solubility data points for the amino acids studied and an AAD in mole fraction of 0.02128 over 141 peptide-solubility points when the systems are studied at their isoelectric point and neutral models are used. The solubility as a function of pH for a range of temperatures is also predicted accurately when charged models are incorporated. These results confirm the predictive accuracy of the SAFT-γ Mie method and pave the way for future studies involving larger peptides.