In recent years biopharmaceutical manufacturing has demonstrated major improvements in MAb production, exhibiting product titers as high as 25 g/L often associated with very high cell densities (1). High-density cell cultures with >150 million cells/mL pose a great challenge in clarification and further downstream processing because of a need to remove a large amount of biomass and increased levels of contaminants from cell debris generated during cell culture and harvesting. Production of biological substances (MAbs, in particular) usually involves processing a complex cell culture broth from which desired biological substances must be isolated and purified while maintaining high overall product recovery and quality.
The eXtreme-Density (XD) cell culture process is a continuous process in which both cells and product are retained in a stirred-tank bioreactor using suspension culture of Crucell’s (www.crucell.com) PER.C6 human cells (1,2). This is accomplished by the use of a modified alternating tangential-flow perfusion system from Refine Technology (www.refinetech.com) in which fresh medium is continuously supplied and waste by-products are continuously removed and discarded. Cell densities of >150 million viable cells per milliliter of culture and product titers of >25 grams of MAb per liter of culture are possible. Because it retains product inside a bioreactor, the XD process produces a much lower harvest volume (only that contained in the bioreactor), which allows for less downstream processing than with traditional perfusion processes (1).
PRODUCT FOCUS: Monoclonal antibodies and other biologics
PROCESS FOCUS: Production and early downstream processing
WHO SHOULD READ: Manufacturing and process development
KEYWORDS: HIGH-DENSITY CELL CULTURE, CLARIFICATION, ANION EXCHANGE, PURIFICATION, PER.C6 CELLS
LEVEL: INTERMEDIATE
Traditionally, centrifugation and a combination of filtration techniques (tangential-flow filtration and depth filtration) have been widely accepted as workhorses for clarifying these complex cell culture broths (3,4,5). However, improvement of mammalian cell culture processes is providing for total cell densities far beyond traditional levels of 20 × 106 cells/mL for CHO cells (6) to >150 × 106 cells/mL for PER.C6 cells (1,2). Thus, limitations of both centrifugation and filtration techniques are apparent by the high (≤40%) solids content of such harvests.
Centrifugation can be applied to process feed streams with high levels of solids, for instance. However, product recovery can be low because of increased pellet volumes and a need to desludge frequently (especially in large-scale continuous centrifugation). Additionally, cell disruption from shear forces generated during centrifugation can further decrease the efficiency of harvest clarification and potentially cause product damage and/or entrapment (4,7).
Depth filters are advantageous because they remove contaminants (8), and many come in single-use format, reducing the need for cleaning and validation (9). However, depth filters are currently unable to handle high-solids feedstreams and are often used in series with centrifugation. TFF can handle high solids loading, but this technique can exhibit poor yield because of polarization of solids at the membrane surface when processing highly dense feed streams. Excessive product dilution and cell lysis caused by shear forces can also limit the utility of TFF.
Flocculation of cell culture harvests has also been widely used to enhance clarification throughput and downstream filtration operations (10,11,12). Current techniques include the use of soluble polyionic polymers (such as DEAE dextran, acryl-based polymers, and polyethylene amine) and inorganic materials such as diatomaceous earth and perlites, which remove cells and cell debris (13). However, polymers must subsequently be removed from process streams, which requires monitoring and quantification by in-process and product-release assays. If IEX chromatography is included as a purification step in the downstream process, binding capacities will be greatly affected by the charged nature of flocculants. The high viscosity of polycation stock solutions presents an additional process challenge.
The goal of our work was to develop an alternative method to centrifugation and known flocculation techniques and apply commonly used micro- or depth-filtration steps for clarification of high-density cell harvests. The method presented here is advantageous over classical flocculation because the polyionic polymer is attached to an insoluble matrix and consequently removed along with cells and cell debris. In this approach, IEX matrices are used to induce and enhance the settling of cells in situ. With a much lower cell density than the starting material, the partially clarified supernatant is recovered for further processing (e.g., by depth filtration). Because the matrices have ionogenic groups, this method also potentially reduces contaminants such as HCP and DNA. Reduction of impurities at this early stage of downstream processing can greatly increase the efficiency of subsequent unit operations (e.g., affinity or IEX chromatography) and thus reduce the overall number of downstream steps.
We developed an enhanced cell settling (ECS) technique for clarification of high-density cell cultures of PER.C6 cells in four distinct stages. First, we screened a range of IEX matrices in terms of their chemical functionality, particle size, and particle density to identify those matrices with the best performance. Second, we investigated the optimal initial cell density (Xt) and the amount of matrix needed to observe ECS. Third, because the technique is based on IEX, we had to determine the effect of cell culture broth conductivity on the extent of settling and contaminant removal. Last, we scaled up the technique to harvest 2-L bioreactors and depth-filtered the subsequent process stream.
Materials and Methods
We generated our cell culture harvest in-house using the XD cell culture technique (1) with PER.C6 cells producing a representative MAb (IgG1, pI = 8.3, 150 kDa). These are E1-immortalized human embryonic retinal cells described elsewhere (14). Using the XD technique, they typically reach a total density of 150–170 × 106 cells/mL (1) in 14–21 days.
Table 1 summarizes the IEX matrices we used. They were prepared by washing with four to five volumes of PBS. We used GIBCO Dulbecco’s PBS 1× (catalog #14190) purchased from Invitrogen (www.invitrogen.com) as received or diluted to the desired conductivity with MilliQ water from Millipore (catalog #ZMQS6V00Y, www.millipore.com). The particle density (dry grams/mL) was determined pycnometrically. Matrices were added as a 50% (v/v) slurry in buffer unless otherwise noted.
Table 1: Summary of IEX matrices used in ECS development
For depth filtration we used two-stage filter trains from Millipore and Cuno (www.cuno.com). The Millipore train involved different grades of Millistak+HC media: a LabScale Pod primary filter containing 540 cm2 of D0HC media (catalog #MD0HC054H1) and a LabScale Pod secondary filter containing 540 cm2 of X0HC media (catalog #MX0HC054H1). Millipore depth filtration was followed by sterile filtration with two parallel AcroPak 20 filters from Pall Corporation (www.pall.com) containing 20 cm2 of 0.8/0.2-µm Supor membrane.
NOMENCLATURE
APAC: analytical protein A chromatography
CM: carboxymethyl
DEAE: diethylaminoethyl
DNA: deoxyribonucleic acid
ECS: enhanced cell settling
ELISA: enzyme-linked immunosorbent assay
g: gravitational force
HCP: host cell protein
IEX: ion exchange
MAb: monoclonal antibody
ND: not determined
PBS: phosphate-buffered saline
PEI: polyethyleninimine
pI: isoelectric point
rt-PCR: real-time polymerase chain reaction
SDS-PAGE: sodium dodecylsulfate polyacrylamide gel electrophoresis
Si-PEI: Bakerbond wide-pore PEI
TBE: Tris borate ethylenediaminetetraacetic acid (EDTA)
TFF: tangential-flow filtration
TMP: transmembrane pressure
TP: Toyopearl brand
%V: percent viability
Vpool: volume of pooled supernatant and washes
Vwork: working volume of bioreactor
XD: extreme-density (perfusion bioreactor)
Xt: total cell density
The Cuno filtration train involved different grades of ZetaPlus media: a 25-cm2 BioCap primary filter containing 10M02 media (catalog #BC0025L10M02) and a 25-cm2 BioCap secondary filter containing 60ZA05A media (catalog #BC0025L60ZA05A). Sterile filtration followed, with a Pall Acrodisc PF filter containing 2.8 cm2 of 0.8/0.2 µm Supor membrane (catalog #4187).
We chose the area of each filter based on the volume of partially clarified media to be processed and operated the filters in series, with the TMP across each filter monitored by pressure transducers. First we flushed the depth filters with ≥100 L of reverse-osmosis water for each square meter of filter area at 600 L/m2/h to wet the filter media and flush out extractables, after which we attached the sterile filters. Then we loaded partially clarified harvest at 50 L/m2/h until the TMP across any one filter reached 15 psig (or until the partially clarified harvest was fully processed). The filter trains were then flushed with 40 L of 7-mS/cm PBS for each square media of filter area and blown down with air to recover the hold-up volume.
We measured antibody titers in cell culture media samples using APAC with a PA immunodetection cartridge (catalog #2-3001-00) from Applied Biosystems (www.appliedbiosystems.com) run on a Waters 2695 separations module (www.waters.com). The mobile phase was 10 mM sodium phosphate from EMD Chemicals (catalog #SX0720, www.emdchemicals.com), 150 mM sodium chloride from Mallinckrodt (catalog #7581, www.mallinckrodt.com) at pH 8.5. The elution buffer was 12 mM hydrochloric acid from Sigma (catalog #H1758, www.sigmaaldrich.com) with 150 mM sodium chloride. HCP levels were determined with a proprietary ELISA at the quality control department of DSM Biologics (www.dsm.com). We determined DNA levels using rt-PCR with customized primers from Applied Biosystems on a StepOne instrument from the same company (catalog #4376600).
We performed several small-scale experiments to develop a method for clarifying XD cell culture by ECS. Unless otherwise noted, 20 mL of cell culture broth was added to a 50 mL conical tube. In most cases, the XD culture was diluted to a cell density of 70–80 × 106 cells/mL using 5- to 7-mS/cm PBS. We added the IEX matrix and gently mixed the suspension end-over-end for 30 seconds. After the tubes settled for 0.5–2.0 hours, we analyzed aliquots of the supernatant by APAC, HCP ELISA, and rt-PCR. Cell densities and viability were measured with a Vi-Cell XR instrument from Beckman-Coulter (www.beckmancoulter.com). Then the remaining supernatant was decanted and cell pellets washed two times with isotonic PBS.
We pooled the initial supernatant and supernatants from both washes for analysis by APAC, HCP ELISA, and rt-PCR. The percent product recovery was determined using Equation
1, the percent reduction of HCPs with Equation
2, and the percent reduction of DNA by Equation
3. Because of the high cell density of the culture, it was necessary to correct MAb, HCP, and DNA concentrations for the percent biomass. We determined biomass through centrifugation of a known volume of culture at 10,000g for five minutes and taking the biomass volume from the pellet volume after centrifugation.
Reagents for agarose-DNA gels came from Invitrogen. We prepared agarose gels (0.8%) by combining 10 mL of 10× TBE (Invitrogen catalog #15581), 0.8 g Ultrapure agarose (Invitrogen catalog #15510-019), 90 mL of MilliQ water from Millipore, and 10 µL of SYBR safe DNA gel stain (Invitrogen catalog #S33102). The mixture was heated for two minutes in a microwave oven until it was homogeneous. Then the solution was cooled for 15 minutes and poured into a gel tray. The gel solidified for an hour at room temperature.
We prepared samples by mixing a desired volume with 10× BlueJuice sample buffer (Invitrogen catalog #10816-015) to a final buffer concentration of 2×. For these experiments, we made 50-µL harvest samples by combining 10 µL of BlueJuice buffer, 10 µL of XD harvest, and 30 mL of MilliQ water. Samples from the partially clarified material were prepared similarly to the harvest samples; however, we incorporated a dilution factor to ensure that equivalent volumes would be added to each gel lane for a direct comparison. We placed the gel in a Gel XL plus cuvette from Labnet International (www.labnetlink.com) and added 1× TBE to cover the gel and the electrodes. A 50-V electrical gradient was applied for one to two hours. We used a 1-kB DNA ladder from New England Biolabs, Inc. (www.neb.com, catalog #N3232L) as a reference and imaged the gel using an Alpha Innotech (www.alphainnotech.com) Fluorchem FC2 instrument at 302 nm, exposing the gel for five seconds.
Results and Discussion
IEX Matrix Screening: We evaluated a total of seven IEX matrices on the basis of the extent of settling (pellet volume, supernatant cell density), product recovery, and clearance of HCP and DNA impurities. The matrices span a range of particle densities, diameters, functionality, and structural backbone. In each case, 2 mL (10% v/v) of matrix was added to 20 mL of cell culture media diluted to 70–80 × 106 cells/mL with PBS (unless otherwise noted).
We evaluated Si-PEI (15 and 40 µm) from JT Baker (www.jtbaker.com), Toyopearl SuperQ (35 and 65 µm) and Toyopearl DEAE (35 and 65 µm) from Tosoh Bioscience (www.toyopearl.com), and Pall DEAE HyperD media with cell cultures of an initial total cell density (Xt) of 98.8 × 106 cells/mL (%V = 94%). The cell culture was diluted to 75 × 106 cells/mL with 5 mS/cm Dulbecco’s PBS to a final volume of 20 mL. We added each IEX matrix as a 50% (v/v) slurry to a final concentration of 10% v/v (matrix volume per dilute cell culture volume) and allowed the cells to settle until the pellet volume was constant (one hour). Then we decanted the supernatant and washed the pellets two times by adding isotonic PBS and allowing cells to settle again until the pellet volume was constant (30 minutes for each wash). Supernatants were pooled from the initial settling and two washes.
Figure 1 shows the supernatant volume over time for each IEX matrix. The Si-PEI matrices produced the largest amount of supernatant volume and corresponded to the most compact pellet (40% of total volume compared with 81% in the control). According to Table 1, Si-PEI matrices are more dense than methacrylate- and agarose-based matrices, which allows for more compact pellets and faster settling times. Ceramic HyperD media have an intermediate density with corresponding intermediate settling times and pellet volumes.
Table 2 summarizes cell density of the pooled supernatants resulting from the addition of IEX matrix and the control. We observed enhanced cell settling (compared with the control) in each case. The smaller particle sizes appear to decrease pool cell densities slightly more than larger particle sizes in each case.
Table 2: Summary of ECS results for various IEX matrices
Table 2 also shows product recovery and HCP reduction. Addition of the IEX matrices increased product recovery and significantly reduced HCP levels in partially clarified pooled supernatants. These results show that the 15-µm Si-PEI performs better than the 40-µm in recovery and HCP clearance; however, the extent of settling is similar. For this reason, we used Si-PEI (15 µm) for subsequent experiments in developing our ECS technique.
Unlike previously evaluated matrices, the CM HyperD product is a cation-exchange IEX matrix. Because cells have a negative charge distribution on their surfaces due to their lipid bilayers, addition of an IEX matrix with the same charge had a negligible effect on the extent of cell settling (15). Additionally, because DNA and a certain HCP population are negatively charged at this pH level, we saw no impurity removal (data not shown). In the HyperZ case, the particle density was too large to enhance settling. The matrix particles immediately settled to the bottom of the conical tube, and no ECS was observed.
Effect of Percent IEX Matrix Added and Initial Cell Density: We evaluated the amount of IEX matrix needed to observe ECS using Si-PEI (15 µm). Different amounts were added to individual conical tubes containing 20 mL of diluted cell culture. The initial Xt was 174 × 106 cells/mL (%V = 79%), which we diluted to 63 × 106 cells/mL. For these experiments, the amount of Si-PEI ranged 0.05–0.1 mL Si-PEI per milliliter of dilute harvest (5–10% v/v). We allowed the cells to settle for two hours, then washed cell pellets with isotonic PBS two times before pooling the supernatants.
Figure 2 shows the extent of settling and the product recovery as a function of the amount of Si-PEI added. Increased amounts of Si-PEI resulted in decreased cell densities in the supernatant pool. In each case, product recovery remained >99% and appeared to be independent of the amount of Si-PEI added. A large HCP reduction was observed in all cases over the range of experimental conditions. For the conductivity and pH of this cell culture, we expect that only a certain population of HCPs (those with pI values below the culture pH) will be removed.
At cell densities >100 × 106 cells/mL, decreased settling was observed due to viscosity effects and cell–cell repulsion. Therefore we determined a more desirable cell density to minimize dilution and maximize ECS by diluting the cell culture with isotonic PBS. For these experiments, 10 mL of cell culture (Xt = 150 × 106 cells/mL, %V = 94%) was diluted 1:1 with PBS to a final volume of 20 mL, and 10% (v/v) Si-PEI was added (2 mL). Figure 3 shows the corresponding supernatant volumes.
In this experiment, we observed very little settling with the undiluted cell culture even after adding Si-PEI, and the recovery and impurity removal could not be determined. However, the diluted cell culture settled to a great extent. The protein recovery was 93% after two pellet washes, and the HCP level was reduced by 28%. Although significant settling occurred in the case of the diluted control without Si-PEI, the cell density of the resulting supernatant was 29 × 106 cell/mL (compared with 10 × 106 cells/mL in the diluted cell culture with Si-PEI). After we pooled supernatants of the Si-PEI treated material, the cell density was 6.1 × 106 cells/mL. Diluting to a cell density of 70–80 × 106 cells/mL with PBS is advantageous in the ECS method with respect to product recovery, impurity removal, and the extent of settling.
Effect of Conductivity: In this method, removal of impurities is based on IEX properties and is inherently affected by the conductivity of cell culture media. Using varying PBS concentrations, we diluted cell cultures with Xt = 127 × 106 cells/mL (%V = 87%) to 75 × 106 cells/mL to reach a final conductivity of 5–10 mS/cm. Si-PEI (15 µm) was added (10% v/v), and the cells settled for an hour. After that initial settling, we washed the resulting pellet twice with isotonic PBS and pooled the supernatants. Table 3 summarizes the results. We found no discernible trend in terms of product recovery and pool cell density. HCP reduction increased as conductivity decreases, which is expected based on the mechanism of IEX.
Table 3: Effect of conductivity on ECS for Si-PEI (15 µm)
Scale-Up: We scaled up our ECS technique to harvest two 2-L XD bioreactors (reactor A and reactor B, both Sartorius Univessel models with Biostat B controllers, www.sartorius.com) with respective working volumes (Vwork) of 1.7 L and 1.8 L. We diluted the harvests 1:1 with 7-mS/cm PBS and then added 10% (v/v) Si-PEI. After 1.0–1.3 hours of settling, the cell density in the resulting supernatant was substantially lower in each case. Figure 4 shows supernatant volumes as a function of time for both reactors. As before, the supernatant was decanted and the pellet washed twice with isotonic PBS, then the pellet was allowed to settle for 30 minutes after each wash.
Table 5 summarizes the results from our scale-up work with reactors A and B. Figure 5 shows SDS-PAGE and DNA gels for both reactors. We pretreated the samples with MAbSelect SuRe protein A resin from GE Healthcare (catalog #17-5438, www.gelifesciences.com) to remove antibody so impurity profiles could be easily discerned. These results indicate that addition of Si-PEI greatly reduces the levels of both HCP and DNA in partially clarified harvests.
Table 4: Summary of scale-up results for reactors A and BTable 5: Summary of impurity removal during depth filtration of reactor B material that was treated by ECS and reactor C material that was not treated
The goal of our work was to develop an initial clarification method that would enable further clarification by conventional methods such as depth filtration. Therefore, we evaluated two filter trains for product recovery, HCP removal, and throughput using Si-PEI treated material from reactor B. The first filter train consisted of a ZetaPlus 10M02 primary filter with 10/2-µm nominal pore size and a 60ZA05A secondary filter with 6/0.5-µm nominal pore size, both from Cuno. The second filter train consisted of a Millistak+HC D0HC primary filter with 5/1-µm nominal pore size and an X0HC secondary filter with 2/0.1-µm nominal pore size, both from Millipore. In each case, product recovery was >90%. Table 5 summarizes HCP reduction across each filter, and Figure 6 shows hydraulic data for both filter trains. The Cuno and Millipore filter trains were comparable in terms of hydraulic performance, with throughputs >550 grams of MAb per square meter of filter, with the Millipore train having a small advantage in HCP reduction.
By comparison, XD cell culture not treated by the ECS method was depth filtered (reactor C) and evaluated for HCP removal. In these experiments, first the culture (150 × 106 cells/mL) was centrifuged at 15,000g for 32 minutes at 4 °C, then the supernatant was loaded onto a Cuno ZetaPlus 60ZA05A (6/0.5-µm) depth filter. Table 5 summarizes the filter performance in removing impurities. From this comparison, it is evident that ECS is instrumental in achieving large reduction of impurities during clarification of high-density cell harvests.
A Good Platform Process Step
Our work demonstrates that addition of weak anion-exchange matrices to cell culture harvest enhances cell settling and offers an effective alternative to centrifugation for clarifying XD PER. C6 cell culture harvests. We screened a wide range of anion-exchange matrices, observing enhanced cell settling in all cases. Si-PEI in particular had a favorable particle density and ligand chemistry that offered superior performance for the ECS technique to remove cells and cell debris as well as host cell related impurities such as HCP and DNA. Cell settling was affected marginally in cultures with conductivities of 5–10 mS/cm. Moreover, increased conductivities are expected to hinder the reduction of impurities. The ECS technique can be readily scaled up while maintaining high product recoveries and reduction of HCP and DNA to enable subsequent downstream operations.
The ECS technique could be adapted for other platform processes in which cell cultures are grown to high densities. Additionally, ECS can be part of an overall strategy to use only single-use technologies for the entire processing of recombinant molecules. When using single-use bioreactors, cells and cell debris can be easily disposed of along with the single-use bioreactors themselves after in-situ clarification.
PER.C6 is a registered trademark of Crucell Holland BV Corporation; XD is a registered trademark of DSM NV; and ECS is a registered trademark of DSM NV. All other brand names are trademarks of their respective owners.
REFERENCES