Perfusion is a means to maintain a cell culture bioreactor in which equivalent volumes of media are simultaneously added and removed from the reactor while the cells are retained in the reactor. This provides a steady source of fresh nutrients and constant removal of cell waste products. Perfusion is commonly used to attain much higher cell density and thus a higher volumetric productivity than conventional bioreactor batch or fed batch conditions. Secreted protein products can be continuously harvested by microfiltration during the process of removing media or retained in the bioreactor by appropriately chosen ultrafiltration during the process of removing media.
With the availability of off-the-shelf perfusion devices that are GMP compliant, such as the XCell ATF® System, perfusion is a viable and accepted manufacturing platform. There are a number of commercial products that are produced using perfusion-based processes since 1990. These include blood clotting factors (Factor VIII Kogenate from Bayer and ReFacto from Wyeth/Pfizer), antibodies (ReoPro, Remicade, Stelera, and Simphoni from Centocor/J&J, and Campath from Genzyme), enzymes (Aldurazyme and Naglazyme of Biomarin), among others.
Attaining high viable cell density (VCD), viability and productivity can improve product quality and impact the production process in multiple ways to achieve significant economic efficiencies. Some examples of how cell culture perfusion can be implemented to improve processes include:
Significant savings can result from each of these processes in a manufacturing production setting. Each of these processes can be assessed to determine the cost savings in the process and/or capital equipment, which can be achieved when implementing perfusion.
Yes, quite often, a client who is new to perfusion will start with their existing basal medium during the beginning of a perfusion run. As cell density increases, a combination of existing basal medium and existing concentrated feed (titrated based on nutrient consumption) is implemented. The relative consumption of medium can be mitigated by changes of the perfusion rate. The ultimate goal is to find a balance between medium composition and the maximum perfusion rate (which is coordinated with the downstream processing and harvest storage capabilities, as well as the cell separation device).
The ideal glucose concentration during steady state perfusion should be low. Cells can have an inefficient metabolism when exposed to a surplus of glucose, which can result in increased lactate production. Therefore, it is recommended that the residual glucose is controlled to a low level, i.e., 0.1 – 0.5 g/L. Unlike a fedbatch process, there can never be a complete depletion of glucose during perfusion, since there is a continuous addition of glucose from the fresh medium. However, at very high cell densities, the glucose present in the medium may not suffice. In this case, a supplemental continuous glucose feed can be implemented, the glucose level in the medium can be increased, or the perfusion rate can be increased if applicable.
Similar impellers applicable to fed batch can be used for perfusion as well, although a dual impeller configuration is recommended for large bioreactors. A combination of marine impeller and Rushton impeller may provide the best combination. kLa studies can be done to determine which impeller configuration is best for your process.
For high density cell culture, a microsparger is recommended. Microspargers create smaller bubbles than drilled tube spargers, and smaller bubbles have higher surface to volume ratio which increase kLa. However, the smaller bubbles are also more stable, and therefore can contribute greatly to CO2 accumulation. And smaller bubbles release more energy during bursting and can be damaging to the cells.
It is recommended that the microsparger is used in combination with a drilled tube sparger, each with its own purpose. The microsparger should be used for O2 sparge only (to minimize gas throughput), while the drilled tube sparger can be used with air or O2 to help minimize the buildup of C02. In addition, the presence of larger bubbles tend to destabilize the foam pack created by the microsparger.
Similar to batch of fedbatch operation, base needs to be used for pH control. The typical choices for base are Sodium Carbonate (Na2C03), Sodium Bicarbonate (NaHCO3), and Sodium Hydroxide (NaOH), but bases used in other processes can also be used. Sodium carbonate is gentler than sodium hydroxide, but it dissociates into more carbonate ions, which contribute to osmolality increase. pCO2 and osmolality are compounded at higher cell densities. Hence, NaOH (e.g. 0.3 – 0.7M) is typical since it does not contribute to pCO2 and it contributes less to osmolality than sodium carbonate. However, because it is a stronger base, it must be added to the system in a way preventing local cell damage, i.e. it should be added at locations where the mixing is the most efficient. It should be added directly above the top impeller so as to dissipate almost immediately.
In order to maintain the level of the bioreactor, the harvest rate should be equal to the predetermined perfusion rate. To prevent any change in level, the feed rate should then be the same as the harvest rate.
The level can be controlled in a variety of ways:
The simplest method is to equalize the harvest pump with the feed pump at the desired perfusion rate, which can be done using a single pump with dual pump heads. This strategy works fine when there is no bleed and there is no base addition.
Another method is to use a conductance probe measuring the level in a feedback loop controlling the feed pump. The tip of the level probe will define the level. As the harvest pump removes the used medium, the level drops and the circuit is broken, which activates the feed pump and medium is added until the circuit is restored. Note: excessive foam build-up can lead to level inaccuracy. If possible, adjust the sensitivity of the level probe to avoid the foam effect.
Weight-based control is the most accurate method to control the level. In this setup, the bioreactor is on a scale (or load cell), which is interfaced with a pump in a feedback loop. This level control loop can be either 1) scale interfaced to feed pump using the bioreactor control loop, or 2) scale interfaced directly to a separate smart feed pump (i.e. SciLog Chemtech pump CP-120). Once ready to start the perfusion, simply start the level control loop on the bioreactor or the smart pump. As the harvest pump removes the medium, the weight drops below the defined set point which triggers the feed pump to bring the weight back to its set point level.
Cell specific perfusion rate (CSPR) equals the perfusion rate/cell density. The ideal CSPR depends on the cell line and the medium. The ideal CSPR should result in optimal growth rate and productivity. 50-100 pL/cell per day may be a reasonable starting range, which can be adjusted to find the optimal rate for your cell line. A lower CSPR can be used to reduce medium use and to increase titer. However, CSPR has been shown to affect product quality, so it is important to determine the ideal CSPR in the context of product quality.
*Note: Once minimum CSPR is determined it is recommended to confirm product quality.
The culture is typically initiated by a batch culture. The perfusion is typically started on day 2-3 after inoculation when the cells are still in exponential growth phase and before nutrient limitation occurs.
The starting day of perfusion is process dependent and will vary with the cell line, inoculate cell density, cell growth rate, metabolism and medium. A strategy can be to initiate the perfusion when the first feed would have taken place in a fedbatch process. Starting perfusion with the XCell ATF® System on day 2-3 is recommended to check/monitor the perfusion equipment (e.g. pumps, level control) and adjust the settings such as the perfusion rate.
There are different methods to increase the perfusion rate, i.e. incremental (manually) or continuous (automatically). Incremental increase is determined by the operator and can be based on the cell density or on a biomarker of the nutrient consumption such as the glucose present in the medium. When the glucose level is a suitable biomarker of the medium nutrient availability, its residual concentration in the bioreactor can be used to adjust the perfusion rate. For an incremental increase, it is typical to start with 0.5 or 1 VVD, and increase by 0.5 or 1 VVD depending on the residual glucose concentration or depending on the cell density. Cell density, growth rate and specific glucose consumption are used to forecast residual glucose. For a continuous increase in perfusion, a biomass probe (e.g. Aber probe) is interfaced with the harvest pump, such that the perfusion rate is increased as a linear function of the cell density determined by the biomass probe, based on a desired CSPR.
The perfusion rate depends on the cell density and the medium. Ideally, it should be minimized to reduce the dilution of the product of interest, i.e. to maximize the harvest titer, while ensuring adequate rates of nutrient addition and by-product removal. Perfusion processes typically range between 1 VVD up to 5 VVD. Lower perfusion rates are preferred in terms of liquid handling and media cost considerations. Higher perfusion rates can be necessary to achieve very high cell densities and decreased POI (Protein of Interest) residence time, but are more challenging in terms of liquid handling. However, if your process relies on high perfusion rate, a supplementary separate XCell ATF® System can be used to concentrate the product from a surge tank to a more reasonable handling volume with the use of an ultrafiltration membrane.
Steady state is a term that refers to the condition where cell density and bioreactor environment remain relatively constant. This can be achieved by cell bleeding, nutrient limitation, and/or potentially by reducing temperature. The cell bleeding method is recommended in case nutrient limitation affects productivity or changes product quality. With the use of an XCell ATF® perfusion system, nutrient supply and waste removal will allow for constant cell growth and productivity.
This will vary depending on the medium, cell line and process. A perfusion rate of 1 VVD commonly supports cell densities of 20 – 30E6 cells/mL. Higher cell densities can be reached by increasing perfusion rate or optimizing medium for use with perfusion. However, there is a point at which too high a cell density is difficult to control within a bioreactor (i.e. pCO2, osmolality, foam, etc.), so it is recommended to reach a steady state at a manageable cell density that can be sustained while productivity remains high.
A cell bleed is simply the removal of cells from the bioreactor. This is typically done through a diptube using a peristaltic pump at a defined flow rate. The choice of tubing should be carefully selected; too narrow and the cells may aggregate and clog while if too large the cells may settle. The cell bleed rate can be determined based on growth rate, thus cell density can be limited to a desired value in a continuous fashion. Alternatively, cells can be removed once a day and replaced by media to maintain cell density within a predictable range.
Ideally, the cell bleed rate is equal to the growth rate to maintain a steady cell density. If there is a significant volume being removed from the bleed with valuable product, then the bleed can be collected into a sterile bag, so protein can be recovered through filtration or centrifugation and then processed to downstream.
Since perfusion cultures operate at higher cell densities, foaming may become an additional factor. However antifoam can be used to control the foam.
Antifoam can be manually added via bulk additions as needed (similar to a fedbatch), or the addition can be automatic using a foam probe. The foam probe is a conductance probe, which is placed higher than the liquid level. As foam increases on the liquid surface, the foam pack rises until it touches the foam probe, creating a circuit. This circuit can be used to trigger a pump which adds antifoam until foam dissipates and the circuit is broken. (Note: a foam probe is a level probe but used in a reversed way.)
In some cases when antifoam is not allowed in the process, foam can be minimized through optimizing the means to achieve efficient oxygen transfer or kLa. Similar to fedbatch, aeration mechanism can be modified to minimize foaming. This can be done by optimizing impeller configuration and speed, sparging only O2 through microsparger, optimizing the pore size of sintered sparger, etc. Notice that if pure O2 sparging (and in particular microsparging) is used, special attention should be paid to the carbon dioxide level in the culture. Note: foaming issues are worse in small scale than in large scale. In large scale, bubble residence time and partial pressure increase with bioreactor height.
It is expected that antifoam will build onto surface of the membrane and it may change the characteristics of the membrane. It may cause some retention of protein and it may have a fouled membrane eventually depending on the amount of days the run is. If fouling (by observation of the permeate flow pressure) is detected the filter can be replaced.