Anion-exchange membranes were evaluated for the capture of a small protein (α-lactalbumin, 3.5-nm diameter) and a large protein (thyroglobulin, 20-nm diameter). The static binding capacity equaled the dynamic binding capacity and increased with increasing protein size. This result was in agreement with calculations based on monolayer coverage on the membrane surface and an absence of mass-transfer limitations. In contrast, for anion-exchange beads, the static capacity was the same for both proteins, and the dynamic capacity decreased strikingly with increasing protein size. These observations were attributed to very slow intrapore diffusion for large proteins in the beads, resulting in surface binding only. This work has important applications in the selection of chromatography media for the purification of viruses and plasmid DNA. Specifically, membranes with a high capacity for large biomolecules (20−300 nm) and a low capacity for small host-cell proteins and endotoxin contaminants are preferable to beads for the purification of such biomolecules.

The Langmuir, steric hindrance, and spreading equations were evaluated separately in a mathematical model of protein purification using ion-exchange membranes. The spreading equation provided the best fit to experimental breakthrough curves (BTCs) for α-lactalbumin (ALA) and thyroglobulin (THY), followed by the steric hindrance equation, and finally the Langmuir equation. The intrinsic rate of protein adsorption to the membrane was found to be rate-limiting, whereas effects of liquid-phase mass transfer and flow nonidealities were negligible. An adsorption rate constant that decreased with increasing surface coverage was required to fit BTCs that were sharp initially and then broadened dramatically as the membrane approached saturation. Predictions using the spreading equation agreed with literature reports on the spreading of blood proteins such as fibrinogen on polymer surfaces.

The performance of ion-exchange membranes for protein purification is analyzed using numerical solutions of different mathematical models. The models incorporate nonlinear sorption isotherms and mass-transfer coefficients based on either the overall or local solid-phase and liquid-phase driving forces. The numerical solutions are compared to analytical solutions which use overall mass-transfer coefficients only and, in general, are theoretically incorrect for nonlinear isotherms. The numerical solutions are fit to experimental breakthrough curves from the literature. The models allow the determination of the rate-controlling mass-transfer phenomena and solid-phase concentration, and prediction of the operating and membrane-design parameters needed to obtain sharp breakthrough curves.

We previously reported that in potato chip and French fry models, the formation of acrylamide can be reduced by controlling pH during processing steps, either by organic (acidulants) or inorganic acids. Use of phytate, a naturally occurring chelator, with or without Ca++ (or divalent ions), can reduce acrylamide formation in both models. However, since phytate itself is acidic, the question remains as to whether the effect of phytate is due to pH alone or to additional effects. In the French fry model, the effects on acrylamide formation of pH, phytate, and/or Ca++ in various combinations were tested in either blanching or soaking (after blanching) steps. All treatments significantly reduced acrylamide levels compared to control. Among variables tested, pH may be the single most important factor for reducing acrylamide levels, while there were independent effects of phytate and/or Ca++ in this French fry model. We also developed a mathematical formula to estimate the final concentration of acrylamide in a potato chip model, using variables that can affect acrylamide formation: glucose and asparagine concentrations, cut potato surface area and shape, cooking temperature and time, and other processing conditions.

Sorption of the four subclasses of human immunoglobulin G (hIgG) to recombinant protein G immobilized to microporous membranes was examined to further the understanding and characterization of this medicallyimportant system. Using batch incubation, sorption of the individual hIgG subclasses was measured in competitive and noncompetitive experiments. Individually, all subclasses had very similar sorption rates and equilibrium capacities. In contrast, for mixtures of the subclasses, binding was distinctly different, with strong competitive binding occurring:  as the system approached equilibrium, significantly more hIgG1 and hIgG3 bound to the membrane than did hIgG2 and hIgG4. A kinetic and equilibrium model was able to successfully simulate the binding of hIgG1, hIgG2, and hIgG4 but not hIgG3. The results are relevant to the healthcare and biotechnology industries:  (1) the diagnosis and treatment of autoimmune diseases, (2) mitigation of rejection in recipients of organ transplants, and (3) production of monoclonal antibodies for use as therapeutic biopharmaceuticals.