Separation Albumin-PEG: Transmission of PEG Through Ultrafiltration Membranes Sandrine Lentsch,' Pierre Aimar,'* Jose Luis Orozco2 zyxwv 'URA CNRS de Genie Chimique, Laboratoire de Genie Chimique et Nectrochimie, Universite P. Sabatier 118, route de Narbonne, 31062 Toulouse Cedex, France; and *Groupement de Recherches de Lacq, B. P. 34 64 zyxwvutsr I70 Artix, France Received April 6, 1992/Accepted December 2 1, 1992 zyxwvu Transmission of polyethylene glycol (PEG) through ul- trafiltration membranes has been studied under vari- ous operating conditions of pressure, crossflow, and concentration, using different membranes cut-offs and two module designs with the aim of understanding the separation of PEG from BSA. The influence of protein adsorption and fouling on the choice of a membrane has also been considered. Retention depends in general on the molecule to average pore size ratio, as expected, but also on concentration polarization. Accordingly, all operating and design parametersfavoring concentration polarization lead to higher transmission. At high fluxes, flexible macromolecules can pass through the mem- brane, even if the random coil is larger than the apparent average pore. From a process selectivity point of view, the best way to separate PEG from BSA would be to use a membrane totally retaining BSA and to enhance concentration polarization of PEG. Unfortunately, such conditions also increasefouling and concentration polar- ization by BSA, which limits flux and thus PEG concen- tration polarization and transmission. Consequences of such conditions on separation efficiency are discussed. zyxwvu 0 1993 John Wiley & Sons, Inc. Key words: polyethyleneglycol albumin ultrafiltration separation INTRODUCTION zyxwvutsrq Protein precipitation is one of the preliminary purification steps at both laboratory and industrial scale. This can be achieved either by addition of organic solvents (ethanol, chloroform), of salts (sodium or ammonium sulfate), or of polymers. One of these polymers, polyethylene glycol (PEG), is successfully used for the separation of proteins from human plasma12 or enzyme from culture broth." The major advantage of PEG is its good water solubility, which avoids any further change of solvent. In addition, this low- toxic and bi~degradable'~ compound can be used in food and pharmaceutical industries. However, a major problem seems to be elimination of PEG once the precipitation step has been achieved. Among the possible separation processes, centrifugation is limited by the high viscosity and density of the fluids to be processed.25 Gel filtration chromatography has been considered, but a preparative scale would require very large columns and a significant dilution of the mixture: * zyxwvutsrqponm To zyxwvutsrqpon whom all correspondence should be addressed. productivity would then be 10w.l~ Other techniques involv- ing selective precipitation to separate PEG from proteins are expensive" or rather hazardous (solvents). In most cases, ion exchange is used at industrial scale and affinity chromatography at laboratory scale, because they allow the recycling of the pre~ipitant,~~ despite the high cost of chromatographic media. Membrane separation processes are often used in biotech- nology to concentrate protein solutions or separate solutes when their sizes are significantly different (proteins from cells or from salts or alcohol). As a general rule, the ratio of their molecular weight has to be larger than 10. While membrane separation provides mild chemical, thermal, and mechanical conditions of separation, allowing preservation of protein activity, several drawbacks (fouling, concentration polarization, solute-solute interactions, etc.) lead to the relatively poor resolution of the process. In this context, our aim was to study limiting phenomena in the separation of PEG from bovine serum albumin (BSA) by porous membranes. Our goal is a 100% retention of BSA and the maximum flux of PEG through the membrane. Several attempts to separate proteins from precipita- tion agents (PEG, polyacrylic acid, etc.) by ultrafiltration have already been published; i.e., Papamichael and Kula,22 Busby and Ingham? and Bozzano and Glatz? The study of the separation of BSA (67 kD) from PEG 4 kD by Busby and Ingham7 has shown that the retention of PEG 4 kD increases during a run mainly because of BSA adsorption. Adsorption, as a limitation to the separation, was also men- tioned by Bozzano and Glatz5 who tried to extract various proteins (bovine serum albumin, conalbumin, lysozyme, ovalbumin) from polyacrylic acid solutions. Further, they outlined the positive influence of concentration polariza- tion on solute transmission, as also observed by Bottino et al.? whereas Papamichael and Kula,22 working in such conditions that solute transport through the membrane was diffusion controlled, did not observe such dependence. Again, these authors suspected protein-precipitant inter- actions to be responsible for the reduction in separation efficiency. Another drawback is the molecular volume of PEG due to its random coil structure. From capillary viscosity data of Tam and Tremblay,26 PEG 20 kD would be of about the same radius as BSA (67 kD). Biotechnology and Bioengineering, Vol. 41, Pp. 1039-1047 (1993) 0 1993 John Wiley & Sons, Inc. CCC 0006-3592/93/01101039-09