Evaluation of Single-Use Fluidized Bed Centrifuge System for Mammalian Cell Harvesting

November 1, 2012
Hsu-Feng Ko

Ravi Bhatia

BioPharm International, BioPharm International-11-01-2012, Volume 25, Issue 11

This article discusses the evaluation of a novel single-use fluidized bed centrifuge for harvesting of antibodies.


This article discusses the evaluation of a novel single-use fluidized bed centrifuge (FBC) for harvesting of antibodies. An FBC that contains four single-use 100-mL chambers was used to harvest Chinese Hamster Ovary (CHO) cell cultures. Optimal operating parameters were defined by performing preliminary studies to determine the maximum chamber capacity and the feed flow rates into the centrifuge. Results of the preliminary studies showed the maximum capacity was approximately 10x109 cells/chamber, and the initial and process feed flow rates used for the studies were 80 and 140 mL/min/chamber, respectively. Five simulated cell-harvesting runs followed the preliminary studies. Three of the five simulated runs utilized healthy cell cultures with viabilities > 90%, and the remaining two runs were with cell cultures with viabilities < 50%. Results showed that the FBC was efficient in separating cells from the product, with low cell density and turbidity detected in the centrate. Average clarification efficiencies were between 88–93% without the use of depth filtration post-clarification. There was no increase in lactate dehydrogenase (LDH) and residual DNA levels indicating that minimal amount of shear stress was induced by the centrifugation. The results, therefore, suggest that the FBC is a promising alternative for cell-harvesting applications.

The first step in the recovery of a secreted product from a mammalian cell culture is to separate the cells and cell debris from the product in the supernatant of the cell culture. A disc-stack centrifuge or tangential flow filtration and microfiltration (TFF–MF) is, generally, used to separate cells as a primary recovery step, followed by depth filtration as a secondary step to remove remaining cell debris and other impurities before the product in the clarified liquid (centrate) can be loaded into a chromatography column (1–3). More information on these primary technologies can be found in a review paper (4).

Image courtesy of kSep Systems

Most conventional cell harvesting technologies are not single-use systems and, therefore, require extensive cleaning and sterilization between batches during production. Single-use technologies have been widely used in the biotechnology industry due to their distinct advantages. Within the past decade, a variety of cell-culture processes has adapted single-use systems, ranging from the use of WAVE technology for processes up to the 500 L scale as well as single-use stirred-tank bioreactor (SUBs) technology up to the 2000 L scale as the production bioreactor vessel for therapeutic protein production (5, 6). Several single-use cell harvesting technologies have also been utilized, including but not limited to the Pod Filter system from Millipore, TFF bioprocess systems from SciLog, and Sartoclear depth filter systems from Sartorius Stedim.

In the past two years, single-use centrifuge technologies have also emerged as a potential alternative for primary recovery processes. The Unifuge, made by Pneumatic Scale Angelus, is a fully automated centrifuge system that uses a single-use insert inside the centrifuge bowl. The kSep technology, a FBC made by kSep Systems Inc., uses balancing centrifugal and fluid flow forces to capture cells in up to four single-use chambers. Both technologies involve retaining cells inside the centrifuge while the centrate that contains the product is continuously separated and discharged. Because of the FBC's advantages (e.g., washing capability, low shear, and continuous mode of operation), the authors selected the FBC for the evaluation of mammalian cell harvesting. The FBC and its single-use set are shown in Figure 1.

Figure 1: FBC (left) and single-use set (right) loaded into the centrifuge rotor (Image courtesy of kSep Systems). (FIGURE 1 COURTESY OF kSEP SYSTEMS.)

Compared with cell harvesting alternatives that rely on the use of conventional centrifuges such as disc-stack or filtration devices to separate cells from the supernatant, the FBC has several advantages over both conventional methods. First, the FBC eliminates the need for cleaning and sterilizing cycle validations. Second, there is no risk for cross-contamination between batches because the product-contact components are single-use only. Third, there is lower shear stress induced by the FBC during operation compared with conventional centrifugation and filtration. Because of the establishment of the fluidized bed during operation, FBC's g-force does not result in cell lysis because cells are not packed against the centrifuge wall. Finally, the washing option available with the FBC provides maximal recovery of the product without diluting the centrate. In addition to these advantages, historical clarification efficiency data from disc-stacks have been shown to be comparable with the expected clarification efficiency performance of the FBC (7).

The FBC is based on a balancing act of two counteracting forces within the system, the centrifugal force (Fg) versus the fluid flow (VQ). The physics of the FBC follows Stokes' Law for spherical particles in a continuous fluid (Equation 1), or in the case of bioprocessing, a cell suspended in a medium:

where Vs is the settling velocity of the particle, ρp is the density of the particle, ρf is the density of the fluid, μ is the dynamic viscosity of the fluid, D is the diameter of the particle, and g is the gravitational acceleration of the particle. The centrifugal force (Fc) and fluid flow velocity (VQ) are defined in the following equations (Equations 2 and 3):

where ω is the angular velocity, R is the radius, Q is the flow rate, and A is the cross-sectional area of the chamber. When the settling velocity (VS) due to the centrifugal force is balanced against the fluid flow velocity (VQ), a fluidized cell bed is created within the chambers while the supernatant is discharged from the chamber as the centrate. Figure 2 shows a schematic of the inner workings of the FBC.

Figure 2: FBC schematic demonstrating principles of FBC’s fluidized bed. (FIGURES 2–6 ARE COURTESY OF THE AUTHORS)

The overall preparation time for the FBC is relatively short compared to the cycle times for cleaning and sterilizing stainless-steel equipment. A single-use set that contains four interconnected chambers and the valve set are first inserted into the FBC prior to an operation. After loading the single-use set into the FBC, individual tubing ends from the single-use set can be sterile-connected to respective vessels such as the bioreactor, buffer, centrate vessel for harvested product, and waste container. Pinch valves and bubble sensors are built-in as part of the FBC and used to automate the cell-harvesting process. The built-in human-machine interface (HMI) screen on the FBC allows run parameters such as centrifuge speed, flow rates, and harvest volumes to be set in individual recipes. Cell-harvesting runs can be nearly fully automatic once the recipes are configured, but manual controls are also available if required.

Figure 3 shows the process flow diagram of a typical harvest clarification recipe. Step one in the process is to prime the single-use set with a buffer. All tubing and chambers are filled with the buffer to displace air from the system. Step two in the recipe directs the system to displace the buffer initially in the single-use chamber(s) into the waste vessel. In doing so, the buffer is not introduced into the centrate vessel and will not dilute the product. Harvest clarification begins in step three, where cells and cell debris are retained in the chamber(s) and the clarified liquid is separated into the centrate vessel. When the chamber(s) are nearly filled with cells, in step four, the bioreactor feed temporarily pauses, and the buffer is used to flush the chamber(s) to recover the product that is still present in the cell bed. In step five, after the wash, pump directions are reversed to discard the cell bed into the waste vessel. In the case where the bioreactor volume is larger than what all four single-use chambers can handle in one batch, steps two through five can be repeated in multiple cycles to finish processing the full volume of the bioreactor. At the end of the process, the system purges all liquids from the single-use set and prompts the operator to seal all tubing prior to the disposal of the single-use set. The remainder of this article discusses the results obtained from the evaluation runs conducted using the FBC for cell harvesting and clarification.

Figure 3: Process flowchart for harvest clarification using the FBC.



All studies used CHO cells cultured in shake flasks (Corning) followed by expansion into WAVE bioreactor (GE Healthcare). Cells were cultured in Janssen R&D's proprietary, chemically-defined medium. 1X phosphate-buffered saline (PBS, Gibco) was used for priming and rinsing during the runs.

The FBC (model kSep 400) and the single-use sets were purchased from kSep Systems, Inc. Cells were placed on an orbital shaker (VWR International) set at 100 RPM during operations to keep cells in suspension.

Cell culture

Because of the simulated nature of the study, CHO cells were only cultured up to four days and up to a cell density of approximately 5x106 cells/mL. To simulate cell cultures with lower viabilities such as those found in 18–20 day cultures to harvest antibodies or therapeutic proteins, a second batch of cells were intentionally starved/asphyxiated by turning off nutrient and air supply after day four and then mixed with healthy cultures to lower the viability.

Determination of maximum chamber capacity

The maximum number of cells each chamber can hold is of great significance because, in most applications, the total number of cells from a bioreactor will exceed what the four chambers can hold. The maximum capacity can also vary from cell line to cell line and is dependent on factors such as cell size. It is, therefore, crucial to determine the maximum chamber capacity at which breakthrough of the cell bed will occur. For this experiment, one chamber was used and the centrifugal force was maintained at 1000g. Cell suspensions were continuously processed by the FBC at 140 mL/min/chamber and samples were taken from the centrate to examine the amount of cells escaping from the chamber. Samples were taken until the breakthrough threshold of the chamber capacity had been reached. Cell count was performed using Cedex cell counter (Roche).

Determination of initial feed flow rate

The initial flow rate when cell suspensions are first introduced into the FBC was determined. It is hypothesized that prior to the formation of the cell bed in the chambers, the flow rate should be kept lower to minimize potential escape of cells from the chambers due to initial instability of the cell bed. To test this hypothesis, 100 mL/min/chamber and 80 mL/min/chamber were compared in two separate runs. Two chambers were used in each of the runs, and the centrifugal force was maintained at 1000g. Cell counts samples were taken and performed on the Cedex to determine the cell density.

Determination of optimal process feed flow rate

Once the fluidized bed is established in the single-use chamber and is stable, the feed flow rate can then be increased. Consecutive runs were performed where process flow rates were increased incrementally from 140 to 225 mL/min/chamber while maintaining the centrifugal force at the maximum of 1000g. Cell counts samples were taken from the centrate to determine the cell density. One chamber was used in each of the runs.

Cell harvesting with FBC

Five total cell-harvesting runs were completed using the FBC. The first three runs were conducted using CHO cell cultures with > 90% viability, followed by two remaining runs using CHO cell cultures with < 50% viability. For all runs, the centrifugal force was kept at the maximum g-force of 1000g. The initial feed flow rate was 80 mL/min/chamber during initial establishment of the cell bed, and increased to 140 mL/min/chamber after 5 min into the process. Samples were taken to measure cell density, turbidity (NTU), LDH level, and residual DNA content.

Cell density and viability were determined using the Cedex cell counter. Turbidity was measured using the HACH 2100AN Turbidimeter. Cell density and turbidity were also used to calculate the FBC's clarification efficiency using Equations 4 and 5:

where CDCentrate is the final cell density in the final harvest vessel, CDBioreactor is the starting cell density in the bioreactor, NTUCentrate is the turbidity in the final harvest vessel, and NTUBioreactor is the starting turbidity in the bioreactor.

LDH levels and residual DNA from samples before and after the FBC process were used as a measure of shear stress and cell lysis during the process. LDH and residual DNA were measured using Johnson & Johnson Vitros Chemistry System DTSC Module and Applied Biosystems Prism 7500 Sequence Detection System, respectively. Antibody titer was also determined using Agilent 1100 Series HPLC.

Theoretical calculations to estimate cell harvesting process time

Finally, because the studies discussed in this article were simulations on a small scale, theoretical calculations were performed to estimate the total process time required for larger bioreactor scales. Calculations were performed using both the kSep 400 and the process scale version, kSep 4000, for 50-L, 25-0L, and 1000-L bioreactors, assuming all four single-use chambers were run at 1000g and at various feed-flow rates.


Determination of maximum chamber capacity

In Figure 4, the measured cell density in the centrate is displayed in blue, and the estimated total cell number based on the flow rate and time is displayed in red. The cell densities in the centrate were initially low until approximately 9–10 billion cells were retained in the chamber. After this point, cell density in the centrate became exponential, as the cell density began to approach the cell density of the bioreactor (~2.6x106 cells/mL). The study was completed before the centrate cell density actually reached the cell density of the bioreactor. Otherwise, it would have been observed where the cells entering the chamber would directly exit into the centrate, and the cell density in the centrate would be equal to the cell density of the bioreactor. Based on the results of this study, to minimize the amount of cells from escaping into the centrate, the maximum amount of cells per chamber should be kept below 10 x 109. This limit was factored into the remainder of the studies.

Figure 4: Determination of maximum chamber capacity: cell density and estimated total cells.

Determination of initial feed flow rate

Based on cell count results from the Cedex cell counter, when running the FBC at 100 mL/min/chamber initially, a sharp spike (result not shown) was observed in the first few minutes of the run. When the initial flow rate was reduced to 80 mL/min/chamber, the spike was not observed. Hence, a slower flow rate of 80 mL/min/chamber was more optimal and was used in remaining studies during the formation of the cell bed.

Determination of optimal process feed flow rate

Cell densities, both viable and total, that correspond to the amount of cells in the centrate at various flowrates are shown in Figure 5. As the flow rates increased from 140 mL/min/chamber to 225 mL/min/chamber, both viable and total cell density in the centrate increased exponentially as the centrifugal force alone became increasingly insufficient to retain cells inside the chamber. The trend exhibited in Figure 5 suggests that the process feed flow rate can potentially be set at 160 or 180 mL/min/chamber with only minimal amount of cells lost into the centrate. The authors elected to use 140 mL/min/chamber as the process feed flow rate for all studies reported in this article because this flow rate is adequate to accomplish all of the objectives in the studies. Increasing the process feed flow rate up to 160 or 180 mL/min/chamber can be a part of future studies if reducing process time becomes critical.

Figure 5: Centrate cell density over time at various process flow rates.

Cell harvesting with FBC

Table I lists the pre-FBC (bioreactor) and post-FBC (centrate) data from three cell-harvesting runs with high cell viability (> 90%) and two cell-harvesting runs with low cell viability (< 50%).

Table I: Starting bioreactor and final centrate parameters for cell harvesting runs. LDH is lactate dehydrogenase. NTU is turbidity.

Cell density data collected from the centrate indicated that the FBC was efficient in separating cells from the supernatant as shown by the low cell counts. Starting with cell densities of 2.3x106 cells/mL (Run #1), 5.4x106 cells/mL (Run #2), 4.8x106 cells/mL (Run #3), 4.3x106 cells/mL (Run #4), and 4.0x106 cells/mL (Run #5) in the bioreactor, none of the cell densities measured from the centrate exceeded 0.215x106 cells/mL. Comparing the cell counts in the final harvest vessels to the starting bioreactor cell counts for each run, the efficiencies of cell removal were in the range of 95.7–98.7% for all runs.

Turbidity data of the centrate samples, measured in NTU, also reflected a similar outcome as the cell densities. It was shown by the large reduction in NTUs that the FBC was efficient in separating cells from the supernatant. Starting with NTUs of 38.4 (Run #2), 33.3 (Run #3), 63.7 (Run #4), and 57.4 (Run #5) in the bioreactor, the FBC effectively separated cells from the supernatant, resulting in significantly reduced NTUs in the range of 2.47–6.91.

The clarification efficiencies for each of the runs, based on NTU measurements, were 93.4%, 90.2%, 89.7%, and 88.2% for runs #2 through #5, respectively. Similar results on clarification efficiencies have been reported using the disc-stack centrifuge technology (7).

LDH levels measured pre- and post-FBC are shown in Figure 6. The average LDH levels for > 90% viability cultures were 3027±313 U and 3459±785 U for pre-FBC and post-FBC samples, respectively, and the LDH levels for the < 50% viability culture were 43,611±2782 U and 37,287±10,419 U for pre-FBC and post-FBC samples, respectively. Post-FBC LDH levels did not increase compared to pre-FBC levels, a clear indication and confirmation that no cell lysis occurred during the process. This is a unique advantage of using the FBC, because cells are suspended in a fluidized bed rather than having a high g-force packing them against the centrifuge wall. A similar outcome was confirmed by the analysis of residual DNA content. Residual DNA for the > 90% viability culture were 12.9 mg and 12.1 mg for pre-FBC and post-FBC samples, respectively, and the residual DNA for the < 50% viability culture were 49.6 mg and 51.4 mg for pre-FBC and post-FBC samples, respectively, again showing minimal cell lysis in the FBC.

Figure 6: LDH level comparison between bioreactor and centrate.

Centrate samples from both Run #3 and Run #5 were analyzed to determine the antibody titers. Minimal antibody titer loss or dilution was observed after processing with FBC due to FBC's efficient washing capabilities.

Theoretical calculations to estimate cell harvesting process time

Table II shows the estimated time for a typical cell harvesting process based on theoretical calculations using both scales of the FBCs. Assuming all four single-use chambers are used at 1000g and 180 ml/min/chamber, approximately 1.2 hr and 5.8 hr are required to harvest a 50-L and 250-L bioreactor, respectively, using kSep 400. The estimated times of 42 min and 2.8 hr are required at 1.5 L/min/chamber to harvest a 250 L and 1000-L bioreactor, respectively, using kSep 4000. Combined with the advantages the FBC possesses over other cell harvesting alternatives, the FBC is emerging as a promising option for cell harvesting.

Table II: Theoretical calculations showing estimated total process time using fluidized bed centrifuges.


The single-use FBC was evaluated for cell-harvesting applications. After determining the maximum capacity of each single-use chamber and the initial and processing flow rates, five total cell-harvesting runs were completed. Cell density, NTU, LDH levels, and residual DNA content from the centrate were collected for evaluating the system. Low cell density and turbidity in the centrate indicated that clarification efficiency was high, and little to no change in the LDH level tle cell lysis during processing. Compared to disc-stack centrifuges and other current standard technologies, it can be concluded that the FBC system is an attractive single-use alternative to current options for cell harvesting.


The authors would like to acknowledge the following individuals for their contributions: Sunil Mehta and Tod Herman for providing their expertise on the FBC; Divya Harjani and Nikhil Patel for maintaining CHO cell cultures and assistance in carrying out some of the FBC runs; Meredith Rice for her assistance with the turbidimeter; and the Janssen R&D Process Analytical Support group for analyzing the residual DNA and antibody titer samples.

Hsu-Feng Ko is a research scientist and Ravi Bhatia is associate director, both at Janssen Research & Development, Spring House, PA. hko@its.jnj.com


1. R. Kempken, A. Preissmann, and W. Berthold, Biotechnol. Bioeng. 46 (2), 132–138 (1995).

2. R. Van Reis, L.C. Leonard, C.C. Hsu, and S.E. Builder, Biotechnol Bioeng. 38 (4), 413–422 (1991).

3. Y. Yigzaw, R. Piper, M. Tran, and A.A. Shukla, Biotechnol Prog. 22 (1), 288-296 (2006).

4. D.J. Roush and Y. Lu, Biotechnol Prog. 24, 488–495 (2008).

5. V. Singh, Cytotechnology. 30 (1–3), 149–158 (1999).

6. R. Bhatia, C. Wood, N. Richardson, and S. Ozturk, ACS National Meeting (San Francisco, CA, 2006).

7. M. Iammarino et al., Bioprocess International, 5 (10) 38–50 (2007).