No abstract

Anion exchange membrane adsorbers are used for contaminant removal in flow-through polishing steps in the manufacture of biopharmaceuticals. This contribution describes the clearance of minute virus of mice, DNA, and host cell proteins by three commercially available anion-exchange membranes: Sartobind Q, Mustang Q, and ChromaSorb. The Sartobind Q and Mustang Q products contain quaternary amine ligands; whereas, ChromaSorb contains primary amine based ligands. Performance was evaluated over a range of solution conditions: 0-200 mM NaCl, pH 6.0-9.0, and flow rates of 4-20 membrane volumes/min in the presence and absence of up to 50 mM phosphate and acetate. In addition contaminant clearance was determined in the presence and absence of 5 g/L monoclonal antibody. The quaternary amine based ligands depend mainly on Coulombic interactions for removal of negatively charged contaminants. Consequently, performance of Sartobind Q and Mustang Q was compromised at high ionic strength. Primary amine based ligands in ChromaSorb enable high capacities at high ionic strength due to the presence of secondary, hydrogen bonding interactions. However, the presence of hydrogen phosphate ions leads to reduced capacity. Monoclonal antibody recovery using primary amine based anion-exchange ligands may be lower if significant binding occurs due to secondary interactions. The removal of a specific contaminant is affected by the level of removal of the other contaminants. The results of this study may be used to help guide selection of commercially available membrane absorbers for flow-through polishing steps.

Membrane adsorbers may be a viable alternative to the packed-bed chromatography for clearance of virus, host cell proteins, DNA, and other trace impurities. However, incorporation of membrane adsorbers into manufacturing processes has been slow due to the significant cost associated with obtaining regulatory approval for changes to a manufacturing process. This study has investigated clearance of minute virus of mice (MVM), an 18-22 nm parvovirus recognized by the FDA as a model viral impurity. Virus clearance was obtained using three commercially available anion exchange membrane adsorbers: Sartobind Q®, Mustang Q®, and ChromaSorb®. Unlike earlier studies that have focused on a single or few operating conditions, the aim here was to determine the level of virus clearance under a range of operating conditions that could be encountered in industry. The effects of varying pH, NaCl concentration, flow rate, and other competing anionic species present in the feed were determined. The removal capacity of the Sartobind Q and Mustang Q products, which contain quaternary ammonium based ligands, is sensitive to feed conductivity and pH. At conductivities above about 20 mS/cm, a significant decrease in capacity is observed. The capacity of the ChromaSorb product, which contains primary amine based ligands, is much less affected by ionic strength. However the capacity for binding MVM is significantly reduced in the presence of phosphate ions. These differences may be explained in terms of secondary hydrogen bonding interactions that could occur with primary amine based ligands.

This contribution describes the preparation of strong anion-exchange membranes with higher protein binding capacities than the best commercial resins. Quaternary amine (Q-type) anion-exchange membranes were prepared by grafting polyelectrolyte nanolayers from the surfaces of macroporous membrane supports. A focus of this study was to better understand the role of polymer nanolayer architecture on protein binding. Membranes were prepared with different polymer chain graft densities using a newly developed surface-initiated polymerization protocol designed to provide uniform and variable chain spacing. Bovine serum albumin and immunoglobulin G were used to measure binding capacities of proteins with different size. Dynamic binding capacities of IgG were measured to evaluate the impact of polymer chain density on the accessibility of large size protein to binding sites within the polyelectrolyte nanolayer under flow conditions. The dynamic binding capacity of IgG increased nearly linearly with increasing polymer chain density, which suggests that the spacing between polymer chains is sufficient for IgG to access binding sites all along the grafted polymer chains. Furthermore, the high dynamic binding capacity of IgG (>130 mg/mL) was independent of linear flow velocity, which suggests that the mass transfer of IgG molecules to the binding sites occurs primarily via convection. Overall, this research provides clear evidence that the dynamic binding capacities of large biologics can be higher for well-designed macroporous membrane adsorbers than commercial membrane or resin ion-exchange products. Specifically, using controlled polymerization leads to anion-exchange membrane adsorbers with high binding capacities that are independent of flow rate, enabling high throughput. Results of this work should help to accelerate the broader implementation of membrane adsorbers in bioprocess purification steps.

The surface-initiated polymerization protocol developed in part I was used to prepare strong anion-exchange membranes with variable polymer chain graft densities and degrees of polymerization for DNA and virus particle separations. A focus of part II was to evaluate the role of polymer nanolayer architecture on DNA and virus binding. Salmon sperm-DNA (SS-DNA) was used as model nucleic acid to measure the dynamic-binding capacities at 10% breakthrough. The dynamic-binding capacity increases linearly with increasing poly ([2-(methacryloyloxy)ethyl]trimethylammonium chloride) chain density up to the highest chain density used in this study. The new membranes yielded threefold higher SS-DNA-binding capacity (30 mg/mL) than a leading commercial membrane with the same functional group chemistry. Elution of bound DNA yielded a sharp peak, and resulted in a 13-fold increase relative to the feed concentration. This concentration effect further demonstrates the highly favorable transport properties of the newly designed Q-type membranes. However, unlike findings in part I on protein binding, SS-DNA binding was not fully reversible. Minute virus of mice (MVM) was used as model virus to evaluate the virus clearance performance of newly designed Q-type membranes. Log reduction of virus (LRV) of MVM increased with increasing polymer chain density. Membranes exhibited >4.5 LRV for the given MVM impurity load and may be capable of higher LRV values, as the MVM concentration in the flow-through fraction of these samples was below the limit of detection of the assay.