Use and elucidation of biochemical data in the prediction of the membrane separation of biocolloids

The successful design and operation of large–scale processes in the biotechnological industries requires effective methods for the prediction of process performance. Such predictive methods should be based on a knowledge of the physicochemical properties of the materials being processed. A major challenge for prediction in biological process engineering is to make use of the sophisticated three-dimensional structural information that is now available for many biological materials. The present paper develops such an approach for an important type of separation: the membrane ultrafiltration of a recombinant protein, in this case human lactoferrin. The protein is treated as a biocolloid. From the available amino acid sequence and structural data for the protein, together with the experimentally measured zeta potential, a charge–regulation model for the lactoferrin–solution interface is developed that takes into account chloride–ion binding. This surface model is then incorporated into a predictive process model for the rate of membrane ultrafiltration of proteins, which takes into account, in a sophisticated manner, the protein–protein interactions (electrostatic, London-van der Waals and entropic) that take place close to the membrane surface and which control the rate of the process. By comparison of the experimentally obtained process data with model predictions it is found that a further attractive interaction between lactoferrin molecules is present, which can be quantified through an exponential decay term. Incorporation of such an interaction in the process model allows successful prediction of the rate of ultrafiltration over a wide range of solution conditions. The existence of the further attractive interaction has been quantitatively verified by osmotic pressure measurements. Hence, the paper shows how detailed structural information may be successfully used for process prediction. It also shows how analysis of process data may itself provide quantitative information on the interactions of biocolloids. This type of approach is likely to become increasingly important as biological process engineers incorporate the latest findings in the biosciences into the development of successful industrial processes.