Quantitative interpretation of protein breakthrough curves in small-scale depth filter modules for bioprocessing
Graphical abstract
Introduction
Depth filtration is used extensively in downstream bioprocessing for removal of cells and cell debris in the clarification of cell culture fluid [1,2]. Depth filters can also increase the capacity and performance of other downstream unit operations, e.g., chromatography columns [3] and virus filtration membranes [4], by removing aggregates and other foulants. In addition, appropriately designed depth filters can provide high levels of removal of host cell proteins [5], DNA [6], and viruses [7], significantly reducing the required level of purification that must be obtained in the rest of the downstream process.
A number of studies have examined the adsorptive capacities of depth filter media with different physical characteristics, including different filter aids (e.g., diatomaceous earth or perlite), polymeric fibers (e.g., cellulose or polypropylene), and binders. For example, Singh et al. [8] found dynamic binding capacities for a model protein (bovine serum albumin) at low conductivity (5 mS/cm) ranging from <1 mg/mL for the Zeta Plus 60SP05A to more than 50 mg/mL for the Emphaze ST-AEX, both of which are positively-charged media. Similarly, Charlton et al. [9] found DNA binding capacities in salt-free solutions ranging from <0.05 μg/mm2 for the Sartobind Q to more than 1 μg/mm2 for the Zeta Plus 90SP.
In addition to the large range of dynamic binding capacities, many depth filters have very broad breakthrough curves. Yigzaw et al. [5] evaluated the concentration of Chinese Hamster Ovary host cell proteins (CHOP) in the effluent from a Millistak mini-cartridge with A1HC media and found 10% breakthrough at a volumetric loading of 30 L/m2 while 90% breakthrough didn't occur until more than 150 L/m2. A similarly broad breakthrough curve has been reported for DNA, with 10% breakthrough at <40 μg DNA loading and 90% breakthrough at >400 μg for the Zeta Plus 10SP [9]. Because of these broad breakthrough curves, it is very difficult to realize the full binding capacities of these depth filters, particularly if they are sized for breakthrough at around 10% of the feed concentration. Note that the behavior of these depth filters is very different than commercially available membrane adsorbers which typically have very sharp breakthrough curves, with dynamic binding capacities approaching the equilibrium binding capacity due to the minimal mass transfer limitations in these systems [10].
Although there have yet to be any detailed analyses of the breakthrough curves in commercial depth filters, a number of groups have used computational fluid dynamics (CFD) to evaluate the breakthrough behavior in membrane adsorbers. For example, Ghosh et al. [11] developed a CFD model that successfully predicted protein breakthrough curves for membrane chromatography devices from Sartorius with bed volumes ranging from 0.08 to 1200 mL. The large-scale system showed a sharper breakthrough curve than the laboratory device due to the more uniform flow distribution across the membrane. Geerling et al. [12] developed a multi-scale model to optimize the porous structure of depth filtration media, but their analysis was focused entirely on the particle capture behavior without any consideration of the adsorptive removal properties of these depth filters.
The objective of this study was to develop a detailed mathematical description of the protein breakthrough curves for the Pall PDH4 media in a commercial Supracap™ 50 capsule and in a custom-designed stainless-steel holder. The flow distributions in both systems were evaluated using CFD as discussed by Kim et al. [13]. The breakthrough curves were then evaluated assuming Langmuir binding kinetics with the equilibrium binding constants determined from static adsorption experiments with the PDH4 media. The results provide important insights into the factors governing the protein breakthrough curves in depth filtration systems.
Section snippets
Breakthrough experiments
Data were obtained with the PDH4 depth filtration media containing diatomaceous earth and cellulosic fibers (Pall Corp., Port Washington, NY, USA). The PDH4 is a dual-layer media consisting of a lenticular grade 700 filter (K700P with 6–15 μm retention rating) and a grade 50 filter (KS50P with 0.4–0.8 μm retention rating). More details on the structure and properties of the PDH4 media are provided by Nejatishahidein et al. [14].
Experiments were performed with the PDH4 media housed in either a
CFD simulations
Initial simulations were performed in the stainless-steel module due to the much lower computational costs in the 2D-axisymmetric geometry. Full 3D simulations were used for the Supracap™ 50 capsule. The velocity and pressure profiles within the module were evaluated by solving the steady Navier-Stokes and continuity equations for an incompressible Newtonian fluid:where ρ is the fluid density, is the fluid velocity vector, μ is the fluid viscosity, and p is the pressure.
Protein breakthrough
Fig. 2 shows typical breakthrough curves for Ribonuclease A (left panel) and α-chymotrypsin (right panel) in the stainless-steel and Supracap™ 50 modules. Data were obtained at a constant filtrate flux of 0.6 L/m2/min using 0.2 g/L solutions of the individual proteins in 30 mM PBS. There was no evidence of any pressure increase during the depth filtration, with the pressure drop remaining less than 2.8 kPa over the course of the experiment. Two repeat measurements are shown for α-chymotrypsin
Conclusions
The experimental data and CFD simulations presented in this study provide the first detailed analysis of protein breakthrough curves in a commercial depth filter capsule, the Supracap™ 50 from Pall. The results clearly demonstrate that the origin of the very broad breakthrough curves for both Ribonuclease A and α-chymotrypsin are due to the complex flow geometry in the Supracap™ 50 capsule. The presence of multiple flow paths through the capsule also gives rise to distinct inflection points
Author statement
Seon Yeop Jung -- Formal analysis, Investigation, Writing – original draft.
Negin Nejatishahidein – Data curation, Formal analysis, Investigation, Writing – original draft.
Minyoung Kim – Investigation, Writing – review & editing.
Ehsan Espah Borujeni – Funding acquisition, Methodology, Resources, Writing – review & editing.
Lara Fernandez-Cerezo – Investigation, Writing – review & editing.
David J. Roush – Funding acquisition, Methodology, Resources, Writing – review & editing.
Ali Borhan --
Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Acknowledgements
The authors would like to acknowledge Merck & Co., Inc., Kenilworth, NJ, USA for their financial support of this project. We also thank Dr. Kyung Hyun Ahn and Dr. Tae Gon Kang for their support in providing the computational resources used for the binding analysis.
References (15)
- et al.
A direct comparison between membrane adsorber and packed column chromatography performance
J. Chromatogr., A
(2020) - et al.
Model-based analysis and quantitative prediction of membrane chromatography: extreme scale-up from 0.08 ml to 1200 ml
J. Chromatogr., A
(2014) - et al.
Flow and residence time distribution in small-scale dual-layer depth filter capsules
J. Membr. Sci.
(2021) - et al.
Process cost and facility considerations in the selection of primary cell culture clarification technology
Biotechnol. Prog.
(2013) - et al.
Large-scale, single-use depth filtration systems
Bioprocess Int.
(2012) - et al.
Process intensification by frontal chromatography: performance comparison of resin and membrane adsorber for monovalent antibody aggregate removal
Biotechnol. Bioeng.
(2020) - et al.
Achieving high mass-throughput of therapeutic proteins through parvovirus retentive filters
Biotechnol. Prog.
(2010)
Cited by (8)
Synthesis of reverse-selective nanoporous ultrafiltration membranes using dual phase separations of ionic liquid and Poly(ethylene glycol) from the gelating urea-linked covalent network
2023, Journal of Membrane ScienceCitation Excerpt :A major design strategy is the generation of an asymmetric porous structure with a thin active layer of well-defined nanopores on top of the macroporous support [5,23–25]. Three-dimensionally continuous nanoporous structures connected by small bottleneck pores also allowed selective filtration by a depth-filtration mechanism [26–29]. The structural control focused on improving the permeation of smaller solutes through the nanopore while rejecting larger ones with a sharp cut-off.
Role of harvest depth filtration in controlling product-related impurities for a bispecific antibody
2022, Antibody TherapeuticsScale-up issues for commercial depth filters in bioprocessing
2022, Biotechnology and Bioengineering
- 1
currently at Bristol Myers Squibb, Devens, MA
- 2
currently at Dankook University, Department of Chemical Engineering, Yongin-si, Gyeonggi-do, 16890, Republic of Korea