Predicting protein refolding from a knowledge of protein interactions under renaturation conditions

Recombinant proteins are often produced as inclusion bodies, which are densely packed aggregates of mis-folded proteins. Isolating the native protein requires dissolving the aggregates under highly denaturing conditions followed by a denaturant removal step to allow for protein refolding. During renaturation, aggregation of partially folded intermediates irreversibly traps inactive proteins. At the industrial level, overcoming this problem requires refolding proteins at low concentrations requiring large tanks. Yields can be improved by using certain additives, which enhance refolding or prevent aggregation. The scaled-up process can only be run in a batch or fed-batch mode, in which protein feed concentration and feed flowrate can be optimized. Often refolding yields can still be low (on the order of 10 percent) and in some instances, proteins do not refold at all, whereas in other cases refolding takes much longer than permitted for a scaled-up process. Thus the overall aims of the proposed project are 1) Understand how solvent influences refolding behaviour so as to make informed decisions about choosing renaturation conditions 2) Provide models of protein aggregation for designing large-scale refolding operations 3) Incorporate knowledge of protein-protein interactions into predictions of protein intrinsic refoldability. 4) Develop robust methods for quantifying refolding yield. Meeting these aims requires determining how refolding yield depends upon protein structural properties and the renaturation conditions (variables including denaturants, salts, additives, pH, protein concentration). We will make this connection by determining the molecular origin of protein intra and intermolecular interactions and then either correlating or fitting the interactions to protein refolding behaviour. The intramolecular interactions will be measured in terms of a protein stability fluoresence assay and an osmotic second virial coefficient (SVC) will be used to characterize the protein-protein interactions. An advantage of our methods is their ability to rapidly screen a range of solvent conditions and proteins. Studies will include therapeutically relevant proteins, which might include recombinant growth colony stimulating factor (rGCSF), human growth hormone (HGH), erythropoietin (EPO), insulin, and insulin like growth factor (ILGF). The protein systems are chosen to encompass a broad range of behaviour so as to isolate structure-interaction relationships. Screening solvent conditions (pH, ionic strength, denaturant concentration, oxidation conditions) will allow us to build up simplified interaction models based upon protein physicochemical properties (protein charge, shape and size, hydrophobicity). This knowledge would then allow us to improve upon current methods for predicting intrinsic refoldability from stability. Also, we will determine how solvent components (denaturants, additives) interact with the protein and/or modulate protein-protein interactions. Understanding the mechanism of interaction will guide the choice of renaturation conditions, with emphasis on finding solvents compatible with the downstream operations. As part of the work we will develop methods for quantifying the refolding yield. The current choice is to use RP-HPLC to separate proteins after the refolding step; natively folded proteins are expected to have different hydrophobicities from incorrectly or aggregated proteins. This method will be assessed in our work by characterizing the monomer peak from the RP-HPLC step using a combination of bioactivity assays, alternative HPLC steps, spectroscopic approaches (FTIR, CD), DSC, and light scattering. For a select few cases, we will also use NMR, which is especially sensitive to any incorrectly folded proteins. The results of the characterization will guide us in choosing a good strategy for assessing refolding yield using a small number of experimentally accessible techniques.