Computer-aided design to select optimal polypeptide tags to assist the purification of recombinant proteins

The addition of a polypeptide fusion partner, called tag, to facilitate purification and detection of recombinant proteins is well recognized. Many different proteins, domains, or peptides can be fused with the target protein and the advantages of using fusion proteins. Nevertheless, the selection of the optimal peptide tag and the right purification system for a specific target protein is difficult. The objective of this work was to develop a mathematical model with decision binary variables, based on MINLP models, which permits the selection of optimal peptide purification tags and optimizes the protein purification process. This model considers a particular set of well-known peptide tags capable of obtaining the required levels of purification. The objective function of the model is the maximization of the profit of the process; this is maximizing the recovery of the desired protein and to minimize the cost of the purification steps. Additionally, a linear relationship between price of the protein and desired purity level was proposed. The mathematical model was evaluated using an example based on cutinase experimental data. The results compare the differences between the sequences with and without tags. In both cases the number of steps is similar, however the recovery level and profit with tag are bigger than the solution without tag. Additionally, the selected peptide tag, for majority cases studied, was FLAG, an 8-amino-acid peptide (DYKDDDDK), which increments the charge and also slightly the hydrophobicity of the protein. Finally, in this model it is simple to introduce new tags for evaluation, in silico, of its possibilities for developing an optimal purification process. Hence, this model could be useful for optimizing purification processes without experimental tests.