step filtration using clinically approved filters improves titers
Division of Hematology/Oncology, Departments of Medicine, Medical and Molecular Genetics, and Microbiology/Immunology, Indiana University School of Medicine, Indianapolis, IN, USA
Correspondence to: K Cornetta, Bone Marrow Transplantation Program, R4 202, 1044 W Walnut Street, Indianapolis, IN, 46202, USAProduction of retroviral vectors for clinical use requires removal of cells and cellular debris. We combined a series of filters of decreasing pore size using commercially available blood banking filters approved for clinical use. The collection bag and filters can be connec timberland boots ted to create a sterile, closed system using clinically approved tubing and sealing systems. Even when challenged with a large number of vector producer cells (2.38 109cells), no viable cells are passed through the system. The step filtration system developed minimizes the titer reduction associated with filtration, provides rapid flow rates, and was cost effective when filtering volumes in excess of 2 liters. One challenge in processing vector is efficient removal of cells and other debris without significantly decreasing titer. To obtain cell free material, investigators have utilized filter systems ranging from 0.2 to 0.45 micron pore size which are commercially available and well suited for use in processing small to moderate volumes of vector supernate (less than a liter). Unfortunately, these systems are known to decrease titer and are not well suited to larger filtration volumes (liters). We hypothesized that a series of filters with decreasing pore sizes may provide more rapid filtration than single pore size filters and would also improve recovery of vector particles yielding higher titer of the manufactured material. To this end, we evaluated a step filtration system for processing large volumes of supernate. The components of this system are filtration devices currently approved for clinical use and used routinely in blood banking. The components can be joined using a sterile connecting device to form a closed system decreasing the risk of contamination. For vector production runs in excess of 2 liters, our analysis indicates this system provides cell free material of higher titer and at lower cost than filtration through 0.45 filters.
We hypothesized that filtration through a series of filters of decreasing pore size (Figure 1) may decrease filtration time by removing large debris before it can obstruct filtration through small pore filters. Unobstructe timberland boots d filters should generate less shear forces, allowing higher flow rates, quicker freezing, and result in higher titer of the final product. To test this hypothesis, we compared the titer obtained after step filtration with filtration through a 0.45 filter. Forty five roller bottles (Corning, No. One day after confluence, the temperature was dropped to 32 and the medium was changed. We had previously determined that maximal titer production occurred when confluent roller bottles of this vector were harvested at 8 h intervals.1 Sequential harvests were performed and the titers obtained are shown in Figure 2. At optimal harvest conditions for this vector, we noted a median 8.6 fold higher titer when supernate was filtered through the step filtration system compared with filtration though a single 0.45 filter. Based on this encouraging preliminary result, we began a series of experiments to further characterize this filtration method.
It should be noted that both filtration methods will decrease titer compared with unfiltered material. The titer from unfiltered supernate was compared with material passed through a 0.45 filter or treated by the step filtration method. Supernate was obtained from five sequential harvests from a roller bottle production of the PG13 clone 8 retroviral vector. In this experiment, titers obtained after step filtering was approximately 49% (range 39 of that obtained with unfiltered product. For 0.45 filtered product, the titer was approximately 32% (range 26 of the unfiltered product.
To ensure the test filtration system provided a cell free product, three vector production runs were processed through the filtration system and tested for contaminating producer cells. After completion of the clinical production run (ie at the time of maximal cell density), fresh medium was added to each roller bottle and allowed to incubate for 24 h. The supernate was then harvested, passed through the filtration system, centrifuged in 250 ml conical tubes (250 g for 10 min) and the lower 30 ml was collected and placed in tissue culture flasks for 14 days. Flasks were then examined for the presence of colony formation. As shown in Table 1, no producer cells were detected in the filtered product. Unfiltered supernate yielded 48 macroscopic colonies per liter.
To provide a more vigorous test of the filtration system, the filters were challenged with a suspension of producer cells in incre timberland boots asing increments. As shown in Table 2, the challenge consisted of 250 ml of cell suspension beginning at 104 cells per ml and increased to a final challenge volume of 250 ml with a cell density of 107 cells per ml. This represents a total cell challenge of 2.78 109 cells. After filtration, the material was placed into flasks and cell growth was evaluated after 14 days. No cells could be recovered from the filtered supernate. We also determined that a roller bottle culture contains approximately 4 108 cells, and 50 roller bottles are used in a typical 3 liter harvest (total cells per harvest: 2 1010). Therefore, our experiment indicates that even if 10% of cells were to become non adherent, the step filtration system will still provide a cell free vector product.
Another technical challenge for any filtration system is expeditious processing, since delays in freezing harvested supernate may decrease titer. One limitation of the single step filtration method is the tendency for decreased flow rates over time. To evaluate flow rates, five sequential harvests from a roller bottle production of the PG13 clone 8 retroviral vector were centrifuged to remove gross debris (250 g for 10 min), then filtered through a 0.45 filter. Figure 3a illustrates the filtration times for the five harvests and demonstrates an initial rapid rate which significantly decreases as further material is filtered. Three of the filtered products decreased to flow rates less than 50 ml per 10 min, one of which decreased to zero before 200 ml of the product was filtered. If the harvested material was placed directly on the filtration unit without centrifugation, less volume can be filtered before the flow rate decreases and three of the five harvests failed to filter 200 ml before the filtration rate fell to zero (data not shown). This observation suggests that cell and cellular debris can obstruct 0.45 filters and while centrifugation before filtration improves flow rates, for most retroviral vector products the capacity of the filter is less than the 250 ml volume proposed by the manufacturer.
To evaluate the flow rate of the step filtration method, we recorded the flow times during a cell challenge. Even when the system was challenged with a total of 2.78 109 producer cells, the flow rate through the filter system was not significantly decreased (Table 2). min (range 35 median 55). The mean flow rate was 53 ml per minute. The flow rate for the step filtration system is an overestimation of the actual filtration time since the limiting step was often the ability to transfer material from the roller bottle to the filtration system.
Not unexpectedly, filtration rates correlated with the titer of the filtered material. The columns in Figure 3 illustrate the titer obtained in consecutive 50 ml pools of 0.45 filtered material for three production harvests. In harvest 3, the titer was relatively stable throughout filtration and the filtration was completed in 16 min. For harvests 1 and 5, the titer of the end material fell as the flow rate decreased, material obtained after 20 min of filtration appears to be five fold lower in titer than material obtained during rapid filtration. The identical analysis was performed on material that was added to the 0.45 filter without prior centrifugation and the fall in titer was more rapid and of greater magnitude (data not shown). This fall in titer was not seen when using the step filtration method. Harvests of PG13 clone 8 supernate analyzed after 0.45 filtration were also analyzed after step filtration. As shown in Figure 4, the lowest titer was noted in the first 5 ml filtered, but by 50 ml of filtered material the titer was similar to the median titer obtained for the remaining product and did not appear to decrease significantly at the higher volumes.
The decrease in titer associated with poor flow rates of 0.45 filters can be related to degradation of the vector, which is time dependent, or due to other factors. As shown in Table 3, 0.45 filters yield similar titers when the flow rate is high. Specifically, when the titer of the filtered material is measured in the first 30 ml passed though a 0.45 filter the results are similar to titers obtained when 2 liters of material is filtered through the step filtration system. The subsequent loss of titer as additional material is filtered through a 0.45 filter could be related to the known decay of vector particles that is both time and temperature dependent. When we evaluated the titer of vector supernate maintained at room temperature or 4 and titered at times 0, 1, 2, 4 and 8 h, a significant decrease in titer was not observed for the first 4 h (data not shown). Since all processing was performed within this time frame, other factors appear responsible for the loss in titer seen with 0.45 filters. Most likely, mechanical disruption of the vector as it passes through obstructed filter pores is the major factor contributing to the loss of titer.
We have observed that the relative benefit of step filtration is cell line dependent. While most producer cell lines submitted to the NGVL are based on NIH3T3 cells, we have shown that the parent cell lines and the derived vector producer cell lines differ significantly in terms of growth characteristics.1 Vector producer cells release a variety of molecules and debris into the supernate, some of which can affect vector titer.2 We have also observed that the ease with which products filter through a 0.45 filter varies from cell line to cell line. For those products which pass relatively rapidly through the 0.45 filters (such as PG13 clone 8, Figure 3), the benefit from step filtration is modest, about two fold. For products which filter slowly (such as GP+envAM12 MSCVneoMGMT, Figure 2), step filtration markedly improves titer.
At our institution, the costs of filtering 3 liters timberland boots of supernate using 0.45 filters (including pipettes, pooling flask and 20 filters) is $602. Disposables for the step filtration system are $454.60. Even without an improvement in titer, the step filtration system is cost effective when producing volumes of supernate over 2 liter. Given that the titer obtained with step filtration is two to eight fold higher than 0.45 filters, additional saving is realized in the quality of the product generated.
Production of viral vectors suitable for human use is an evolving science where much work remains to define optimal production conditions.3 Generation of clinical grade retroviral vectors by academic centers must balance efforts to maximize titer with the cost of production. For phase I trials submitted to the NGVL, the amount of material requested ranged from 10 to 40 liters of unconcentrated supernate. While bioreactors suitable for very large vector productions can increase titer,4 the amount of cell debris from roller bottle systems may pose an even greater challenge for single filtration methods and suggest our step filtration method may provide added benefit to this method of manufacturing. Concentration and purification methods for vectors can also provide a product with increased titer5,6 and may eliminate the need for filtration.7,8 Unfortunately, the fragile state of the retroviral vectors results in loss of particles during concentration so the actual yield of vector particles is much lower than the amount produced, adding significant cost to production. While improved vector production technology is needed, generation of vector from producer cell lines by the methods described above are still likely to be utilized for many phase I studies. This is related in part to the cost advantages, and to recent improvements in transduction methodology allowing clinically significant transduction of target cells with unconcentrated retroviral vector.9,10
In summary, removal of cells and debris from vector supernate using a step filtration system improves titer and provides for a closed system which easily handles large volumes. Cryopreservation bags used for bone marrow storage provide for easy collection and storage of filtered material and are available in a variety of sizes, so the storage volume can be customized to meet the needs dictated by the intended clinical use.
Two methods of filtration were evaluated. The control method utilized 0.45 filtration flasks (product number 1530045, Nalgene, Rochester, NY, USA). Pooled supernate was transferred into the flasks by pipette and processed by vacuum assisted filtration (maximum volume 200 ml per filter unit). The test method, a step filtration system, is shown in Figure 1. Supernate was pooled into a bone marrow collection kit (product number 4R2107, Baxter, Deerfield, IL, USA) which has a capacity of 1.2 liters. The supernate is passed through a dual in line filter (No. 4C8030), a 20 filter (Baxter, No. 4C7704) and a Sepa cell filter (No. 4C2481). The product is collected into a 2 liter transfer pack which is included with the Baxter bone marrow collection kit. The supernate is distributed into Cryocyte freezing bags (Baxter, No. 4R9955) based on weight (100 g per bag), sealed using the Sebra Sealer model 1100 (Tucson, AZ, USA), and placed in a re usable freezing canister (Baxter, No. 4R54462). The filtration system was connected using the Terumo sterile connecting device (model SCD312; Somerset, NJ, USA) with welding wafers (Baxter, No. 4R4350). The vector submitted by Drs Cassian Yee and Mary Flowers was developed at Targeted Genetics, Seattle, WA, USA.
Murine cell lines were maintained in D10 media (high glucose Dulbecco’s modified Eagle’s medium with 10% fetal bovine serum (Hyclone, Logan, UT, USA), supplemented with 2 mmol glutamine, and 1 mmol Na pyruvate), and incubated at 37 5% CO2. HeLa cells were maintained in MEM with Earle’s salts, 10% fetal bovine serum, and 1% non essential amino acids (Gibco, Grand Island, NY, USA). Vector producer cell lines utilized PG13 cells11 and PA317 cells12 obtained from the ATCC (Nos CCL 92, CRL 10685, CRL 9078, respectively). The GPE+8613 and GPenvAm12 cell lines14 were supplied by Genetix Inc (Tarrytown, NY, USA).
Retroviral producer cell lines containing a drug resistance gene were assayed for vector titer by colony forming assay. 1 105 cells (NIH 3T3 for PA317 and GPenvAm12 packaged vector or HeLa cells for PG13 packaged vector) were plated in 60 mm2 dishes. The following day vector samples were thawed, serial dilutions were prepared in triplicate (using D10 media) and target cells were transduced with 1 ml of vector containing medium. The cells were incubated for 4 h at 37 5% CO2 in the presence of polybrene (8 g after which time the vector containing medium was removed and replaced with fresh serum containing medium. After 48 h, vector transduced cells were incubated with serum containing medium and the appropriate selection agent, either 750 g of active G418 (Life Technologies) or 500 g of Hygromycin B (Boehringer Mannheim, Indianapolis, IN, USA). After 10 days of culture, drug resistance colonies were scored and the colony forming units per milliliter of vector supernate was calculated.
A variety of vectors submitted to the NGVL for clinical grade production were tested for the optimal time for harvesting vector supernate. For clinical production and other roller bottle productions used in this analysis, a Forma incubator (Model No. 3956, Marietta, OH, USA) was used with a Wheaton roller bottle unit (No. 348940 Wheaton Science Products, Millville, NJ, USA). Cells from a single pool of cells are seeded in 850 cm2 bottles (Corning No. 431198, Corning, NY, USA) at a rotation speed of 5 Rotation speed is increased after 12 h to 20%. Cultures are grown with 60 ml of medium and harvested when confluent. Glucose is monitored after each sequential harvest and for clinical production the volume is increased to 120 ml when the harvest glucose falls below 200 mg (titer is volume dependent and optimal titer at the initiation of sequential harvesting is obtained at 60 ml, data not shown).
Titer assessment by flow cytometry
Ecotropically packaged MFG eGFP vector, which expresses the enhanced green fluorescence protein (GFP),15 was used to generate MFG eGFP vector producer cells in the PA317 and PG13 cell lines. Vector expressing producer cells were obtained by cell sorting using a FACStarPlus (Becton Dickinson Immunocytometry Systems, San Jose, CA, USA). Titer was assessed by determining the transduction efficiency of target cells. Undiluted supernate (1 ml) was incubated with 1 105 target cells (NIH 3T3 cells for MFG eGFP cells or HeLa cells for MFG eGFP per well of a six well dish approximately 16 h after plating. Cells were exposed to undiluted and 1:10 diluted vector for 4 h in the presence of 8 g polybrene (5% CO2, 37 Cells were maintained in culture for 2 days then trypsinized, washed twice in phosphate buffered saline, and analyzed for fluorescence using a FACScan (Becton Dickinson Immunocytometry Systems).
1 Reeves L, Smucker P, Cornetta K. Packaging cell line characteristics and optimizing retroviral vector titer: the National Gene Vector Laboratory experience. Hum Gene Ther 2000; 11: 2093 2 Le Doux JM, Morgan JR, Snow RG, Yarmush ML. Proteoglycans secreted by packaging cell lines inhibit retrovirus infection. J Virol 1996; 70: 6468 3 Braas G, Searle PF, Slater NKH, Lyddiatt A. Strategies for the isolation and purification of retroviral vectors for gene therapy. Bioseparation 1996; 6: 211 4 Kotani H et al. Improved methods of retroviral vector transduction and production for gene therapy. Hum Gene Ther 1994; 5: 19 5 Bowles NE et al. A simple and efficient method for concentration and purification of recombinant retrovirus for increased hepatocyte transduction in vivo. Hum Gene Ther 1996; 7: 1735 6 McGrath M, Witte O, Pincus T, Weissman IL. Retrovirus purification: method that conserves envelope glycoprotein and maximizes infectivity. J Virol 1978; 25: 923 7 Lyddiatt A, O’Sullivan DA. Biochemical recovery and purification of gene therapy vectors. Curr Opin Biotechnol 1998; 9: 177 8 Hammar L. Concentration of biomaterials: virus concentration and viral protein isolation. Meth Enzymol 1994; 228: 640 9 Cavazzana Calvo M et al. Gene therapy of human severe combined immunodeficiency (SCID) X1 disease. Science 2000; 288: 669 Article 10 Abonour R et al. Efficient retroviral mediated MDR 1 gene transfer into autologous human long term repopulating hematopoietic stem cells. Nature Med 2000; 6: 652 Article 11 Miller AD et al. Construction and properties of retrovirus packaging cells based on gibbon ape leukemia virus. J Virol 1991; 65: 2220 12 Miller AD, Buttimore C. Redesign of retrovirus packaging cell lines to avoid recombination leading to helper virus production. Mol Cell Biol 1986; 6: 2895 13 Markowitz D, Goff S, Bank A. A safe packaging line for gene transfer: separating viral genes on two different plasmids. J Virol 1988; 62: 1120 14 Markowitz D, Goff S, Bank A. Construction and use of a safe and efficient amphotropic packaging line. Virology 1988; 167: 400 15