Protein fouling during constant-flux virus filtration: Mechanisms and modeling

As biomanufacturers consider the transition from batch to continuous processing, it will be necessary to re-examine the design and operating conditions for many downstream processes. For example, the integration of virus removal filtration in continuous biomanufacturing will likely require operation at low and constant filtrate flux instead of the high (constant) transmembrane pressures (TMPs) currently employed in traditional batch processing. The objective of this study was to examine the effect of low operating filtrate flux (5–100 L/m²/h) on protein fouling during normal flow filtration of human serum Immunoglobulin G (hIgG) through the Viresolve® Pro membrane, including a direct comparison of the fouling behavior during constant-flux and constant-pressure operation. The filter capacity, defined as the volumetric throughput of hIgG solution at which the TMP increased to 30 psi, showed a distinct minimum at intermediate filtrate flux (around 20–30 L/m²/h). The fouling data were well-described using a previously-developed mechanistic model based on sequential pore blockage and cake filtration, suitably modified for operation at constant flux. Simple analytical expressions for the pressure profiles were developed in the limits of very low and high filtrate flux, enabling rapid estimation of the filter performance and capacity. The model calculations highlight the importance of both the pressure-dependent rate of pore blockage and the compressibility of the protein cake to the fouling behavior. These results provide important insights into the overall impact of constant-flux operation on the protein fouling behavior and filter capacity during virus removal filtration using the Viresolve® Pro membrane.     

Interactions between protein molecules and the virus removal membrane surface: Effects of immunoglobulin G adsorption and conformational changes on filter performance

Membrane fouling commonly occurs in all filter types during virus filtration in protein‐based biopharmaceutical manufacturing. Mechanisms of decline in virus filter performance due to membrane fouling were investigated using a cellulose‐based virus filter as a model membrane. Filter performance was critically dependent on solution conditions; specifically, ionic strength. To understand the interaction between immunoglobulin G (IgG) and cellulose, sensors coated with cellulose were fabricated for surface plasmon resonance and quartz crystal microbalance with energy dissipation measurements. The primary cause of flux decline appeared to be irreversible IgG adsorption on the surface of the virus filter membrane. In particular, post‐adsorption conformational changes in the IgG molecules promoted further irreversible IgG adsorption, a finding that could not be adequately explained by DLVO theory. Analyses of adsorption and desorption and conformational changes in IgG molecules on cellulose surfaces mimicking cellulose‐based virus removal membranes provide an effective approach for identifying ways of optimizing solution conditions to maximize virus filter performance. © 2017 American Institute of Chemical Engineers Biotechnol. Prog., 34:379–386, 2018     

Prefiltration enhances performance of sterile filtration for glycoconjugate vaccines

Recent studies have reported very low capacity during sterile filtration of glycoconjugate vaccines due to rapid fouling of the sterile filter. The objective of this study was to explore the potential for significantly increasing the capacity of the sterile filter through the use of an appropriate prefilter. Data were obtained using prefilters with different pore size and chemistry, with the sterile filtration performed at constant filtrate flux using 0.22 μm nominal pore size Durapore® polyvinylidene difluoride membranes. Prefiltration through 5 μm pore size Durapore® or Nylon prefilters nearly eliminated the fouling of the sterile filter, leading to more than a 100-fold reduction in the rate of pressure increase for the sterile filter. This dramatic improvement in sterile filter performance was due to the removal of large components (greater than 1 μm in size) as confirmed by dynamic light scattering. These results demonstrate the potential of using large pore size prefilters to significantly enhance the performance of the sterile filtration process for the production of important glycoconjugate vaccines.     

Virus harvesting in perfusion culture: Choosing the right type of hollow fiber membrane

The use of bioreactors coupled to membrane-based perfusion systems enables very high cell and product concentrations in vaccine and viral vector manufacturing. Many virus particles, however, are not stable and either lose their infectivity or physically degrade resulting in significant product losses if not harvested continuously. Even hollow fiber membranes with a nominal pore size of 0.2 µm can retain much smaller virions within a bioreactor. Here, we report on a systematic study to characterize structural and physicochemical membrane properties with respect to filter fouling and harvesting of yellow fever virus (YFV; ~50 nm). In tangential flow filtration perfusion experiments, we observed that YFV retention was only marginally determined by nominal but by effective pore sizes depending on filter fouling. Evaluation of scanning electron microscope images indicated that filter fouling can be reduced significantly by choosing membranes with (i) a flat inner surface (low boundary layer thickness), (ii) a smooth material structure (reduced deposition), (iii) a high porosity (high transmembrane flux), (iv) a distinct pore size distribution (well-defined pore selectivity), and (v) an increased fiber wall thickness (larger effective surface area). Lowest filter fouling was observed with polysulfone (PS) membranes. While the use of a small-pore PS membrane (0.08 µm) allowed to fully retain YFV within the bioreactor, continuous product harvesting was achieved with the large-pore PS membrane (0.34 µm). Due to the low protein rejection of the latter, this membrane type could also be of interest for other applications, that is, recombinant protein production in perfusion cultures.     

Tangential flow microfiltration and ultrafiltration for human influenza A virus concentration and purification

Large scale purification of viruses and viral vectors for gene therapy applications and viral vaccines is a major separation challenge. Here tangential flow microfiltration and ultrafiltration using flat sheet membranes has been investigated for concentration of human influenza A virus. Ultrafiltration membranes with molecular weight cutoffs of 100 and 300 kDa as well as 0.1, 0.2 and 0.45 microm microfiltration membranes have been tested. The results indicate that use of 300 kDa membranes not only concentrate the virus particles but also lead to a significant removal of host cell proteins and DNA in the permeate. Tangential flow filtration may be used to fractionate virus particles. Human influenza A virus particles are spherical with an average size of 100 nm. Use of a 0.1 microm membrane leads to passage of virus particles less than 100 nm into the permeate and an increase of larger particles in the retentate. These results suggest that control of the transmembrane pressure, membrane pore size and pore size distribution could enable isolation of intact virus particles from damaged virions. Isolation of the virus particles of interest from viral fragments and other particulate matter could result in simplification of subsequent purification steps. Larger pore size membranes such as 0.45 microm that allow the passage of all virus particles may be used to remove host cell fragments. In addition virus particles attached to these fragments will be removed. Careful selection of membrane morphology and operating conditions will be essential in order to maximize the benefit of tangential flow filtration steps in the purification of viral products from cell cultures.     

Investigation of virus retention by size exclusion membranes under different flow regimes

Virus removal by filter membranes is regarded as a robust and efficient unit operation, which is frequently applied in the downstream processing of biopharmaceuticals. The retention of viruses by virus filtration membranes is predominantly based on size exclusion. However, recent results using model membranes and bacteriophage PP7 point to the fact that virus retention can also significantly be influenced by adsorptive interactions between virus, product molecules, and membranes. Furthermore, the impact of flow rate and flow interruptions on virus retention have been studied and responsible mechanisms discussed. The aim of this investigation was to gain a holistic understanding of the underlying mechanisms for virus retention in size exclusion membranes as a function of membrane structure and membrane surface properties, as well as flow and solution conditions. The results of this study contribute to the differentiation between size exclusion and adsorptive effects during virus filtration and broaden the current understanding of mechanisms related to virus breakthroughs after temporary flow interruptions. Within the frame of a Design of Experiments approach it was found that the level of retention of virus filtration membranes was mostly influenced by the membrane structure during typical process-related flow conditions. The retention performance after a flow interruption was also significantly influenced by membrane surface properties and solution conditions. While size exclusion was confirmed as main retention mechanism, the analysis of all results suggests that especially after a flow interruption virus retention can be influenced by adsorptive effects between the virus and the membrane surface. © 2018 American Institute of Chemical Engineers Biotechnol. Prog., 35: e2747, 2019.     

Impact of protein fouling on nanoparticle capture within the Viresolve® Pro and Viresolve® NFP virus removal membranes

Virus filtration is a robust size-based technique that can provide the high level of viral clearance required for the production of mammalian-derived biotherapeutics such as monoclonal antibodies. Several studies have shown that the retention characteristics of some, but not all, virus filters can be significantly affected by membrane fouling, but there have been no direct measurements of how protein fouling might alter the location of virus capture within these membranes. The objective of this study was to directly examine the effect of protein fouling by human immunoglobulin G (IgG) on virus capture within the Viresolve® Pro and Viresolve® NFP membranes by scanning electron microscopy using different size gold nanoparticles. IgG fouling shifted the capture location of 20 nm gold nanoparticles further upstream within the Viresolve® Pro filter due to the constriction and/or blockage of the pores in the virus retentive region of the filter. In contrast, IgG fouling had no measurable effect on the capture of 20 nm nanoparticles in the Viresolve® NFP membrane, and IgG fouling had no effect on the capture of larger 40 and 100 nm nanoparticles in either membrane. These results provide important insights into how protein fouling alters the virus retention characteristics of different virus filters.     

Investigation of fouling mechanisms of virus filters during the filtration of protein solutions using a high throughput filtration screening device

The downstream process development of novel antibodies (Abs) is often challenged by virus filter fouling making a better understanding of the underlying mechanisms highly desirable. The present study combines the protein characterization of different feedstreams with their virus filtration performance using a novel high throughput filtration screening system. Filtration experiments with Ab concentrations of up to 20 g/L using either low interacting or hydrophobically interacting pre-filters indicate the existence of two different fouling mechanisms, an irreversible and a reversible one. At the molecular level, size exclusion chromatography revealed that the presence of large amount of high molecular weight species—considered as irreversible aggregates—correlates with irreversible fouling that caused reduced Ab throughput. Results using dynamic light scattering show that a concentration dependent increase of the mean hydrodynamic diameter to the range of dimers (17 nm at 20 g/L) together with a negative DLS interaction parameter k_D (−18 mL/g) correlate with the propensity to form reversible aggregates and to cause reversible fouling, probably by a decelerated Ab transport velocity within the virus filter. The two fouling mechanisms are further supported by buffer flush experiments. Finally, concepts for reversible and irreversible fouling mechanisms are discussed together with strategies for respective fouling mitigation. © 2019 American Institute of Chemical Engineers Biotechnol. Prog., 35: e2776, 2019.     

Murine leukemia virus clearance by flocculation and microfiltration

Clearance of murine leukemia virus from CHO cell suspensions by flocculation and microfiltration was investigated. Murine leukemia virus is a retrovirus that is recommended by the U.S. Food and Drug Administration for validating clearance of retrovirus-like particles. Due to biosafety considerations, an amphotropic murine leukemia virus vector (A-MLV) that is incapable of self-replication was used. Further, A-MLV is incapable of infecting CHO cells, thus ensuring that infection of the CHO cells in the feed did not result in a reduced virus titer in the permeate. The virus vector contains the gene for the enhanced green fluorescent protein (EGFP) to facilitate assaying for infectious virus particles. The virus particles are 80-130 nm in size. The feed streams were flocculated using a cationic polyelectrolyte. Microfiltration was conducted using 0.1 and 0.65 microm pore size hollow fiber membranes. The level of virus clearance in the permeate was determined. For the 0.1 microm pore size membranes a 1,000-fold reduction in the virus titer in the permeate was observed for feed streams consisting of A-MLV, A-MLV plus flocculant, A-MLV plus CHO cells, and A-MLV plus flocculant and CHO cells. While the flocculant had little effect on the level of virus clearance in the permeate for 0.1 microm pore size membranes, it did lead to higher permeate fluxes for the CHO cell feed streams. Virus clearance experiments conducted with 0.65 microm pore size membranes indicate little clearance of A-MLV from the permeate in the absence of flocculant. However, in the presence of flocculant the level of virus clearance in the permeate was similar to that observed for 0.1 microm pore size membranes. The results obtained here indicate that significant clearance of A-MLV is possible during tangential flow microfiltration. Addition of a flocculant is essential if the membrane pore size is greater than the diameter of the virus particles. Flocculation of the feed stream leads to an increase in the permeate flux.     

Particle-based analysis elucidates the real retention capacities of virus filters and enables optimal virus clearance study design with evaluation systems of diverse virological characteristics

In virus clearance study (VCS) design, the amount of virus loaded onto the virus filters (VF) must be carefully controlled. A large amount of virus is required to demonstrate sufficient virus removal capability; however, too high a viral load causes virus breakthrough and reduces log reduction values. We have seen marked variation in the virus removal performance for VFs even with identical VCS design. Understanding how identical virus infectivity, materials and operating conditions can yield such different results is key to optimizing VCS design. The present study developed a particle number-based method for VCS and investigated the effects on VF performance of discrepancies between apparent virus amount and total particle number of minute virus of mice. Co-spiking of empty and genome-containing particles resulted in a decrease in the virus removal performance proportional to the co-spike ratio. This suggests that empty particles are captured in the same way as genome-containing particles, competing for retention capacity. In addition, between virus titration methods with about 2.0 Log₁₀ difference in particle-to-infectivity ratios, there was a 20-fold decrease in virus retention capacity limiting the throughput that maintains the required LRV (e.g., 4.0), calculated using infectivity titers. These findings suggest that ignoring virus particle number in VCS design can cause virus overloading and accelerate filter breakthrough. This article asserts the importance of focusing on virus particle number and discusses optimization of VCS design that is unaffected by virological characteristics of evaluation systems and adequately reflect the VF retention capacity.     

The effects of buffer condition on the fouling behavior of MVM virus filtration of an Fc-fusion protein

A combined pore blockage and cake filtration model was applied to the virus filtration of an Fc-fusion protein using the three commercially available filters, F-1, F-2, and F-3 in a range of buffer conditions including sodium-phosphate and tris-acetate buffers with and without 200 mM NaCl at pH 7.5. The fouling behaviors of the three filters for the feed solutions spiked with minute virus of mice were described well by this combined model for all the solution conditions. This suggests that fouling of the virus filters is dominated by the pore blockage mechanism during the initial stage of the filtration and transformed to the cake filtration mechanism during the later stage of the filtration. Both flux and transmembrane resistance can be described well by this model. The pore blockage rate and the rate of increase of protein layer resistance over blocked pores are found to be affected by membrane properties as well as the solution conditions resulting from the modulation of interactions between virus, protein, and membrane by the solution conditions.     

Microscopic visualization of virus removal by dedicated filters used in biopharmaceutical processing: Impact of membrane structure and localization of captured virus particles

Virus filtration with nanometer size exclusion membranes (“nanofiltration”) is effective for removing infectious agents from biopharmaceuticals. While the virus removal capability of virus removal filters is typically evaluated based on calculation of logarithmic reduction value (LRV) of virus infectivity, knowledge of the exact mechanism(s) of virus retention remains limited. Here, human parvovirus B19 (B19V), a small virus (18–26 nm), was spiked into therapeutic plasma protein solutions and filtered through Planova™ 15N and 20N filters in scaled-down manufacturing processes. Observation of the gross structure of the Planova hollow fiber membranes by transmission electron microscopy (TEM) revealed Planova filter microporous membranes to have a rough inner, a dense middle and a rough outer layer. Of these three layers, the dense middle layer was clearly identified as the most functionally critical for effective capture of B19V. Planova filtration of protein solution containing B19V resulted in a distribution peak in the dense middle layer with an LRV >4, demonstrating effectiveness of the filtration step. This is the first report to simultaneously analyze the gross structure of a virus removal filter and visualize virus entrapment during a filtration process conducted under actual manufacturing conditions. The methodologies developed in this study demonstrate that the virus removal capability of the filtration process can be linked to the gross physical filter structure, contributing to better understanding of virus trapping mechanisms and helping the development of more reliable and robust virus filtration processes in the manufacture of biologicals.     

Densonucleosis virus purification by ion exchange membranes

Preparative chromatography is widely used in the downstream purification of biopharmaceutical products. Replacement of resins by membranes as chromatographic supports, overcomes many of the limitations associated with resin-based chromatography such as high-pressure drops, slow processing rates due to pore diffusion and channeling of the feed through the bed. In particular, adsorptive membranes may be ideally suited for virus capture. Virus capture is critical in a number of applications. In gene therapy and vaccine production, large-scale purification of virus vectors is often essential. In the manufacture of biopharmaceuticals, validation of virus clearance is critical.Here results for purification of Aedes aegypti densonucleosis virus (AeDNV) using anion and cation exchange membranes are presented. AeDNV is a non-enveloped, single-stranded mosquito-specific parvovirus. Virus particles are around 20 nm in size. AeDNV could find potential applications in integrated vector-borne disease control programs. In addition, capture of parvovirus for validation of virus clearance in the manufacture of biopharmaceuticals is of commercial importance.By adjusting the pH of the feed stream, AeDNV particles may be adsorbed by both anion and cation exchange membranes. However, strongly basic anion exchange membranes were the most effective in adsorbing AeDNV particles. Adsorption and subsequent elution of AeDNV by anion exchange membranes leads to significant virus concentration. Dynamic and static capacities for anion exchange membranes were similar. Further, a sharp elution curve was obtained suggesting that pore diffusional resistances are insignificant. The adsorption of AeDNV particles by anion exchange membranes may be described by a linear isotherm.     

Protein fouling of virus filtration membranes: effects of membrane orientation and operating conditions

The capacity of virus filters used in the purification of therapeutic proteins is determined by the rate and extent of membrane fouling. Current virus filtration membranes have a complex multilayer structure that can be used with either the skin-side up or with the skin-side facing away from the feed, but there is currently no quantitative understanding of the effects of membrane orientation or operating conditions on the filtration performance. Experiments were performed using Millipore's Viresolve 180 membrane under both constant pressure and constant flux operation with sulfhydryl-modified BSA used as a model protein. The capacity with the skin-side up was greater during operation with constant flux and at low transmembrane pressures, with the flux decline or pressure rise due primarily to osmotic pressure effects. In contrast, data obtained with the skin-side down showed a slower, steady increase in total resistance with the cumulative filtrate volume, with minimal contribution from osmotic pressure. The capacity with the skin-side down was significantly greater than that with the skin-side up, reflecting the different fouling mechanisms in the different membrane orientations. These results provide important insights for the design and operation of virus filtration membranes.     

Integration of Planova filters in manufacturing processes of biologicals improve the virus safety effectively: A review of publicly available data

The capacity to remove viruses by Planova filters produced by Asahi Kasei, primarily by small virus-retentive filters, were compiled from data in peer-reviewed publications and, partly, publicly available data from presentations at conferences (Planova workshops). Data from more than 100 publications and presentations at conferences covering Planova filters were assessed. The data were grouped according to the different virus filters regarding mean pore sizes and viruses of different sizes for plasma and cell culture derived products. Planova 15N and 20N filters removed parvoviruses below the limit of detection of viruses in the filtrate in approx. 50% of all studies and mean LRFs (log reduction factors) for viruses detected in the filtrate were above 4, demonstrating effective parvovirus reduction. Parvovirus removal capacity increased for Planova BioEX filters as well as for 2 Planova 20N in series. Large viruses as retroviruses (e.g., HIV and MuLV), herpesviruses, flaviviruses and togaviruses were removed effectively by Planova 15N, 20N and BioEX filters and also by Planova 35N filters. Flow interruption, transmembrane pressure, volume and protein concentration per filter area had had no substantial impact on virus removal capacity at manufacturing specification. In conclusion, the incorporation of Planova filters in manufacturing processes of biologicals remove, depending on the filter pore size, small and large viruses from the feed stream reliably. This virus reduction step with an orthogonal mechanism integrated in the manufacturing processes of biologicals, based primarily on size exclusion of viruses, improves the virus safety of these biopharmaceutical products considerably.     

Effect of solution pH on protein transmission and membrane capacity during virus filtration

Although virus filtration is now an integral part of the overall viral clearance strategy for the purification of many commercial therapeutic proteins, there is currently little understanding of the factors controlling the performance of the virus filters. The objective of this study was to examine the effects of solution pH on protein transmission and capacity during virus filtration. Data were obtained using bovine serum albumin as a model protein with Viresolve 180 membranes oriented with both the skin-side up and skin-side down. Membranes were also characterized using dextran sieving measurements both before and after protein filtration. Membrane capacity and protein yield were minimal at the protein isoelectric point, which was due to the greater degree of concentration polarization associated with the smaller protein diffusion coefficient at this pH. In contrast, the actual protein sieving coefficient was maximum at the protein isoelectric point due to the absence of any strong electrostatic exclusion under these conditions. The yield and capacity were both significantly greater when the membrane was oriented with the skin-side down. These results provide important insights into the effects of solution conditions on the performance of virus filtration membranes for protein purification.     

Impact of virus preparation quality on parvovirus filter performance

Virus‐removal filtration technology is commonly used in the manufacturing process for biologics to remove potential viral contaminants. Virus‐removal filters designed for retaining parvovirus, one of the smallest mammalian viruses, are considered an industry standard as they can effectively remove broad ranges of viruses. It has long been observed that the performance of virus filters can be influenced by virus preparations used in the laboratory scale studies (PDA, 2010 ). However, it remains unclear exactly what quality attributes of virus preparations are critical or indicative of virus filter performance as measured by effectiveness of virus removal and filter capacity consistency. In an attempt to better understand the relationship between virus preparation and virus filter performance, we have systematically prepared and analyzed different grades of parvovirus with different purity levels and compared their performance profiles on Viresolve® Pro parvovirus filters using four different molecules. Virus preparations used in the studies were characterized using various methods to measure DNA and protein content as well as the hydrodynamic diameter of virus particles. Our results indicate that the performance of Viresolve® Pro filters can be significantly impacted depending on the purity of the virus preparations used in the spike and recovery studies. More importantly, we have demonstrated that the purity of virus preparations is directly correlated to the measurable biochemical and biophysical properties of the virus preparations such as DNA and protein content and monodispersal status, thus making it possible to significantly improve the consistency and predictability of the virus filter performance during process step validations. Biotechnol. Bioeng. 2013; 110: 229–239. © 2012 Wiley Periodicals, Inc.     

Virus filtration of high‐concentration monoclonal antibody solutions

The ability to process high‐concentration monoclonal antibody solutions (> 10 g/L) through small‐pore membranes typically used for virus removal can improve current antibody purification processes by eliminating the need for feed stream dilution, and by reducing filter area, cycle‐time, and costs. In this work, we present the screening of virus filters of varying configurations and materials of construction using MAb solutions with a concentration range of 4–20 g/L. For our MAbs of interest—two different humanized IgG1s—flux decay was not observed up to a filter loading of 200 L/m² with a regenerated cellulose hollow fiber virus removal filter. In contrast, PVDF and PES flat sheet disc membranes were plugged by solutions of these same MAbs with concentrations >4 g/L well before 50 L/m². These results were obtained with purified feed streams containing <2% aggregates, as measured by size exclusion chromatography, where the majority of the aggregate likely was composed of dimers. Differences in filtration flux performance between the two MAbs under similar operating conditions indicate the sensitivity of the system to small differences in protein structure, presumably due to the impact of these differences on nonspecific interactions between the protein and the membrane; these differences cannot be anticipated based on protein pI alone. Virus clearance data with two model viruses (XMuLV and MMV) confirm the ability of hollow fiber membranes with 19 ± 2 nm pore size to achieve at least 3–4 LRV, independent of MAb concentration, over the range examined. © 2009 American Institute of Chemical Engineers Biotechnol. Prog., 2009     

Analysis of protein fouling during ultrafiltration using a two-layer membrane model

Protein fouling can significantly alter both the flux and retention characteristics of ultrafiltration membranes. There has, however, been considerable controversy over the nature of this fouling layer. In this study, hydraulic permeability and dextran sieving data were obtained both before and after albumin adsorption and/or filtration using polyethersulfone ultrafiltration membranes. The dextran molecular weight distributions were analyzed by gel permeation chromatography to evaluate the sieving characteristics over a broad range of solute size. Protein fouling caused a significant reduction in the dextran sieving coefficients, with very different effects seen for the diffusive and convective contributions to dextran transport. The changes in dextran sieving coefficients and diffusive permeabilities were analyzed using a two-layer membrane model in which a distinct protein layer is assumed to form on the upstream surface of the membrane. The data suggest that the protein layer formed during filtration was more tightly packed than that formed by simple static adsorption. Hydrodynamic calculations indicated that the pore size of the protein layer remained relatively constant throughout the adsorption or filtration, but the thickness of this layer increased with increasing exposure time. These results provide important insights into the nature of protein fouling during ultrafiltration and its effects on membrane transport.     

Role of membrane structure on the filtrate flux during monoclonal antibody filtration through virus retentive membranes

Virus removal filtration is a critical step in the manufacture of monoclonal antibody products, providing a robust size-based removal of both enveloped and non-enveloped viruses. Many monoclonal antibodies show very large reductions in filtrate flux during virus filtration, with the mechanisms governing this behavior and its dependence on the properties of the virus filter and antibody remaining largely unknown. Experiments were performed using the highly asymmetric Viresolve® Pro and the relatively homogeneous Pegasus™ SV4 virus filters using a highly purified monoclonal antibody. The filtrate flux for a 4 g/L antibody solution through the Viresolve® Pro decreased by about 10-fold when the filter was oriented with the skin side down but by more than 1000-fold when the asymmetric filter orientation was reversed and used with the skin side up. The very large flux decline observed with the skin side up could be eliminated by placing a large pore size prefilter directly on top of the virus filter; this improvement in filtrate flux was not seen when the prefilter was used inline or as a batch prefiltration step. The increase in flux due to the prefilter was not related to the removal of large protein aggregates or to an alteration in the extent of concentration polarization. Instead, the prefilter appears to transiently disrupt reversible associations of the antibodies caused by strong intermolecular attractions. These results provide important insights into the role of membrane morphology and antibody properties on the filtrate flux during virus filtration.