Cutinase-peptide fusions in thermoseparating aqueous two-phase systems. Prediction of partitioning and enhanced tag efficiency by detergent addition.

It is of increasing importance to develop efficient purification methods for recombinant proteins where the number of steps can be minimised. The aim has been to establish a method for predicting the partitioning of the wild-type target protein in an aqueous two-phase system, and with this as basis, develop fusion tags and optimise the phase system for enhanced partitioning of the target protein. The surface of the lipolytic enzyme cutinase from Fusarium solani pisi was investigated with a computer program, Graphical Representation and Analysis of Surface Properties (GRASP). The accessible surface areas for the different amino acid residues were used together with peptide partitioning data to calculate the partition coefficient for the protein. The separation system was composed of a thermoseparating random copolymer of ethylene oxide and propylene oxide. Breox PAG 50A 1000, as top phase forming polymer and a hydroxypropyl starch polymer, Reppal PES 200, as bottom phase polymer. The calculated partition coefficient for the wild-type protein (K= 1.0) agreed reasonably well with the experimentally determined value (K=0.85). Genetic engineering was used to construct fusion proteins expressed in Saccharomyces cerevisiae based on cutinase and peptide tags containing tryptophan, to enhance the partitioning in aqueous two-phase systems. The partitioning of the cutinase constructs could qualitatively be predicted from peptide partitioning data, i.e. the trends in partitioning could be predicted. A spacer peptide introduced between protein and tag increased the partitioning of the protein towards the ethylene oxide-propylene oxide (EOPO) copolymer top phase. The aqueous two-phase system was modified by addition of detergent to increase the partitioning of the cutinase variants towards the EOPO copolymer phase. Triton and a series of C12En detergents selectively increased the partitioning of cutinase constructs with (WP)4-based tags up to 14 times compared to wild-type cutinase. The protein partition could almost quantitatively be predicted from the peptide partition data.