September 15, 2006 (Vol. 26, No. 16)
Revised Flow Design Allows for Significantly Improved Cell-Harvesting Processes
Advances in fermentation and cell culture techniques have greatly increased the titers of target proteins in culture fluid. This increase in upstream efficiency has led to a bottleneck in downstream processing at the cell-harvest stage. Scale up of current tangential flow filtration (TFF) systems has not been successful in terms of efficiency and economics.
Cell harvesting, or clarification, is an important process in nearly all downstream purifications of biotech-based products. When the product is internal to the cells, cell harvesting is used to decrease the liquid volume of cells to be processed in the product extraction steps. An example of this process type is the isolation of inclusion bodies from E. coli lysate prior to solubilizing the inclusion body product.
When the product is extracellular, cell harvesting is used to separate the product from the cells and cellular debris. An example of this process type is the isolation of an extracellular antibody from mammalian cell culture.
Commercially available methods for cell harvesting include TFF and the combination of centrifugation with depth filtration. Current installations of TFF systems, as well as centrifugation offer significant opportunities for improvement.
TFF is a well-established technology for cell harvesting. Traditional TFF modules for biopharmaceutical applications use flat-sheet formats that are assembled into cassettes or hollow fibers that are assembled into modules. Originally designed with a small membrane area for laboratory filtrations, cassette and hollow-fiber systems were enlarged to meet pilot- and production-scale operations without consideration of the fluid flow dynamics required for the larger membrane incorporated in the scaled-up processes.
Overall, the advantages of TFF include relatively low capital costs and the ability to reuse the filter modules. However, as companies scaled up traditional TFF formats, they encountered low yield, short filter life, and inconsistent results. Reasons for the low yield and decreased performance included filter clogging during the process.
SmartFlow™ Technology
The performance limits of traditional cassette and hollow fiber TFF systems have recently been expanded with the introduction of SmartFlow™ technology developed by NCSRT (www.ncsrt.com).
Modules incorporating SmartFlow (SF-TFF) technology feature a patented ribbed configuration. Uniform flow is created by the combination of discrete retentate channels within the module and diagonally opposed inlet and outlet ports in the holder. This design is also linearly scalable from the laboratory to full production systems without a drop in efficiency or yield.
The enhanced fluid dynamics of these new modules results in reduced gel layer buildup, decreased fouling, and increased membrane throughput. Therefore, these modules exhibit increased permeate rates, improved process yield, enhanced cleaning, and sustained permeation rates that are maintained over longer periods of time.
Lack of clogging is especially useful when a large volume of material containing smaller particles is processed, such as when harvesting a bioreactor. Finally, the channel height can be specified in these modules allowing for efficient retentate distribution even with viscous or turbid solutions. Thus even in high density concentrated cultures a module with an optimal channel to prevent clogging is available.
Several studies focused on harvesting mammalian cell cultures, microbial fermentations, and module cleaning are reviewed in this tutorial to demonstrate the platform capabilities of SF-TFF.
Figure 1: Comparison of the regenerated cellulose membrane with a 0.45-um pore size and a modified membrane.
CHO Cell Harvest
In the first study, for the separation of CHO cells from an extracellular therapeutic Mab, the OPTISEP® 800 filter module was utilized to optimize membrane performance and protein passage. Two different membrane chemistries, modified polysulfone (MPS) and regenerated cellulose (RC), were evaluated.
Figure 1 illustrates the flux performance for 0.45-µm MPS and RC membrane chemistries. The average Lm2H was 300 Lm2H for the MPS membrane and 130 Lm2H for the RC membrane. The antibody passage was found to be 99% at 9X concentration for the MPS membrane and 90% at 10X concentration for the RC membrane (Figure 2). Because it had a higher flux rate and higher antibody passage, the MPS membrane was selected for further scale-up studies.
The volume processed to membrane area was analyzed; specifically 62.5 L/m2 and 125 L/m2 of membrane were compared. The results of these trials demonstrated that the initial process volume to the membrane-area proportion is a critical factor in process scale-up. Results demonstrated that doubling the fluid-to-membrane proportion halved the flux rate of the process.
Figure 2: Protein passage at the beginning and end of a CHO cell harvest for the RC and MPS membranes.
Mouse Mammary Cell Perfusion Harvest
An OPTISEP 3000 filter module with a 0.45-µm MPS membrane was used to assess its performance with mouse mammary perfusion cell culture and verify the CHO cell culture results with a second cell type. The proportion of starting volume to membrane area with the highest flux rate from the CHO cell experiments was 125 L/m2. Using this volume-to-membrane area proportion in the mouse mammary cell culture experiment, the average Lm2H during a 10X concentration was over 500 Lm2H. Additionally, the product protein passage remained at >97% even after multiple runs (Figure 3).
Figure 3: The membrane flux and protein passage for harvest of a mouse mammary cell line using an OPTISEP 3000
Insect Cell Harvest
An OPTISEP 7000 containing 1 m2 of MPS membrane was used to harvest 25 L of Drosophila S2 culture. The products of interest, dengue vaccine components, were secreted into the culture medium. Two target proteins, a 47-kD monomer and a 68-kD dimer, were isolated. Low shear rate and pressure were maintained during the course of the experiment to prevent cell lysis during the clarification.
For both proteins, the 25 L of culture was clarified to a concentration factor over 10X in less than 20 minutes. This represents an average performance of over 80 Lm2H. Additionally, the protein quality obtained from the SF-TFF isolation was compared to proteins obtained from centrifugation using SDS-PAGE. As Figure 4 indicates, the retentate and permeate samples obtained using the OPTISEP SF-TFF module were of comparable quality to those obtained using traditional centrifuge methods.
Figure 4: SDS Page of insect cell harvest.
Pichia pastoris Harvesting
In addition to harvesting mammalian and insect cells, modules with SmartFlow technology are also effective for clarification of microbial fermentations. An OPTISEP 7000 filter module was used to concentrate a 50-L Pichia pastoris fermentation.
At the end of the concentration, a wet cell concentration of over 80% was reached (Figure 5). Membrane flux performance averaged 50 Lm2H for the Pichia concentration from 30% wet cell solids to 80% wet cell solids. A final concentration of up to 90% wet cell solids has been achieved using a higher (1.0-mm) channel height.
Figure 5:The membrane flux and a function of percent weight cell solids for a harvest of P.pastoris
Bacteria Harvesting
A common method for harvesting E. coli inclusion bodies (IBs) is to first lyse the cells and then purify the IBs from the cell debris. Using an OPTISEP 800 filter module containing a 0.2-µm membrane, 2.5 L of resuspended E. coli lysate was added to a PUROSEP™ LT-2Q system. To purify the IB product (Figure 6), the resuspended cells were washed with four diavolumes of wash buffer containing 2.5% Triton X-100 and then washed with eight diavolumes of buffer.
The final retentate was concentrated to a minimum hold-up volume. The module demonstrated flux rates between 40 and 80 Lm2H with a process average of 50.8 Lm2H.
Figure 6: Membrane flux during the diafiltration of E.coli lysate
Module Cleaning
The ability to clean TFF modules can be determined by comparing the clean-water flux rate before use to the clean-water flux rate after using the module and performing a cleaning step. The ability to clean an OPTISEP 3000 module was demonstrated by customer trials, which harvested five successive CHO cell cultures.
After each harvest, the membrane was first flushed with 15 minutes of warm DI water, then cleaned with 60 minutes of 0.2-M NaOH, and then flushed with warm DI water to neutral pH. The OPTISEP SmartFlow modules final clean-water flux after the five trials was the same as the initial clean-water flux (Figure 7).
These trials demonstrated the ability to clean and reuse Optisep filter modules for CHO cell culture harvest.
Figure 7: Shown are the initial (purple data points) and final (red data points) clean water fluxes after five CHO-cell harvest.
James Kacmar, Ph.D., is scientist, process development, Hank Kopf is CTO, and Mark Vander Hoff is vp marketing, NCSRT. Web: www.ncsrt.com. Phone: (919) 387 8460. E-mail: [email protected].