Protein purification by polyelectrolyte coacervation: influence of protein charge anisotropy on selectivity

The effect of polyelectrolyte binding affinity on selective coacervation of proteins with the cationic polyelectrolyte, poly(diallyldimethylammonium chloride) (PDADMAC), was investigated for bovine serum albumin/β-lactoglobulin (BSA/BLG) and for the isoforms BLG-A/BLG-B. High-sensitivity turbidimetric titrations were used to define conditions of complex formation and coacervation (pH(c) and pH(?), respectively) as a function of ionic strength. The resultant phase boundaries, essential for the choice of conditions for selective coacervation for the chosen protein pairs, are nonmonotonic with respect to ionic strength, for both pH(c) and pH(?). These results are explained in the context of short-range attraction/long-range repulsion governing initial protein binding "on the wrong side of pI" and also subsequent phase separation due to charge neutralization. The stronger binding of BLG despite its higher isoelectric point, inferred from lower pH(c), is shown to result from the negative "charge patch" on BLG, absent for BSA, as visualized via computer modeling (DelPhi). The higher affinity of BLG versus BSA was also confirmed by isothermal titration calorimetry (ITC). The relative values of pH(?) for the two proteins show complex salt dependence so that the choice of ionic strength determines the order of coacervation, whereas the choice of pH controls the yield of the target protein. Coacervation at I = 100 mM, pH 7, of BLG from a 1:1 (w/w) mixture with BSA was shown by SEC to provide 90% purity of BLG with a 20-fold increase in concentration. Ultrafiltration was shown to remove effectively the polymer from the target protein. The relationship between protein charge anisotropy and binding affinity and between binding affinity and selective coacervation, inferred from the results for BLG/BSA, was tested using the isoforms of BLG. Substitution of glycine in BLG-B by aspartate in BLG-A lowers pH(c) by 0.2, as anticipated on the basis of DelPhi modeling. The stronger binding of BLG-A, confirmed by ITC, led to a difference in pH(?) that was sufficient to provide enrichment by a factor of 2 for BLG-A in the coacervate formed from "native BLG".