Transport and separation of proteins by ultrafiltration through sorptive and non-sorptive membranes

Abstract

Measurement of transmembrane flux and apparent macrosolute rejection during unstirred ultrafiltration of aqueous solutions of bovine serum albumin (BSA) and myoglobin (MYO) and their mixtures have been measured with non-sorptive (regenerated cellulose) and sorptive (polysulfone) membranes of equally rated “cut-off” (30 kDa) as functions of time, solute concentration, pressure, pH, ionic strength, and various conditions of membrane pretreatment. Permeate flux and solute rejection vary with time or permeate volume for both membranes in a manner consistent with accepted models of polarization and protein-layer buildup under these conditions. For the non-sorptive membrane, MYO rejection is low and flux relatively time-independent, while BSA rejection is high, and flux declines rapidly with time. Preexposure of this membrane to MYO or BSA solutions has no significant effect on membrane permeability or rejection. Filtration of binary protein mixtures with this membrane results in virtually unimpeded passage of MYO and nearly complete retention of BSA, allowing efficient separation of these macro-molecules; rejections are roughly the same as observed with single-solute measurements. Failure of the BSA polarization layer to significantly retard MYO permeation was unexpected, since it is at variance with the postulate that the gel layer of retained macrosolute will provide a barrier to transport of the smaller species. With the protein-sorptive polysulfone membrane, on the other hand, BSA rejection is nearly complete, while MYO rejection is low but finite. However, both flux and rejection of both solutes are significantly altered when both solutes are present in solution, and are also functions of such variables as solution ionic strength, protein concentration, pH, and the prior history of membrane-exposure to solutions of either protein. Such changes in operating conditions may either increase or decrease rejection of either protein. This indicates that (largely irreversible) adsorption of protein on and in the pores of such a membrane may control solute and solvent transport. Mechanisms to explain this behavior, based on consideration of protein sorption dynamics and conformational stability, are proposed.