Increasing Efficiency in Protein and Cell Production by Combining Single-Use Bioreactor Technology and Perfusion

May 2, 2011

BioPharm International

Volume 2011 Supplement, Issue 4

An off-the-shelf, single-use perfusion system.


Perfusion processes in mammalian cell cultivations are currently experiencing a renaissance, and are making their way into modern single-use bioreactor systems. In this article, the authors review the common challenges and main applications for perfusion processes. The authors present experimental evidence for reaching extreme cell densities of 150 ×106 cells/mL in an easy to use, high-performance, single- use bioreactor that provides an efficient way to generate cell mass for cell banking or seed trains, as well as production of considerable amounts of monoclonal antibodies (mAbs), even from cell lines with mediocre expression levels.

During the past decade, the pharmaceutical industry has increasingly switched focus from small molecule-based drugs to recombinant therapeutic proteins, mAbs, vaccines and more recently, cell therapy products to find novel treatments for unmet medical needs. Considerable effort has been dedicated to improving the efficiency of cell lines, vectors, culture media and the production process itself. Today, a number of platform approaches to high-titer expression, especially of mAbs, have been reported (1, 2). These platforms consist of a host cell line, vectors, media, and feeds that are crucial to reach titers of >2–5 g/L in a 14 day fed-batch process.



Recently, there has been renewed interest in perfusion processes. In perfusion culture, fresh medium is continuously fed to the culture while the cells are retained in the bioreactor, and metabolic waste products are removed. Depending on the chosen cell retention device, the large molecular product can either be retained in the bioreactor (concentrated fed-batch) or removed from the culture (perfusion). The latter production mode is typically used when the recombinant protein is prone to degradation, toxic, or inhibitory to the cells. In perfusion processes, cell densities are typically 5– to 10–fold higher than in batch or fed-batch processes. Process optimization of perfusion processes is intended to maintain cell-specific productivity by optimizing the composition of the perfused medium and/or applying a certain cell bleed stream to control cell density and growth rate. It is then possible to achieve 5 to 10 times higher total amount of product compared with batch and fed-batch (3). An economic comparison of the different process modes has recently been published (4). Due to significantly increased productivity, a number of companies have identified concentrated fed-batch and perfusion as tools to reduce production scale and ultimately make commercial production amenable to single use bioreactors. This growing interest in continuous perfusion is further fueled by the increased number of recombinant therapeutic proteins in the development pipeline.

Several cell-retention devices have been developed during the past 20 years. Today, there are a handful of systems that have proven suitable for suspension cell culture at process scale: continuous centrifugation, inclined settlers, hydrocyclones, and alternating tangential filtration (3, 4).

In addition to improved yields in protein production, high cell-density culture also offers advantages for seed production by reducing the bioreactor scale and number of steps necessary to generate the seed for the final production bioreactor. A 10–fold increase in cell density reduces the required inoculum volume by the same factor. To increase flexibility and reduce the time from thawing a cell bank vial to inoculation of the production bioreactor, studies have been performed examining the use of large volume, cryopreserved cultures in cryobags, where the concentrated seed culture has been produced using a perfusion process in order to replace shake flask technology (5, 6). This approach not only reduces the number of bioreactors and steps involved, but also the necessary footprint for seed bioreactors, and investment in equipment.


Perfusion often requires expensive investment in cell-retention equipment, certain infrastructure, and personnel with specialised skills. Therefore, it has not been widely adopted for the production of proteins that are used for research purposes, such as target proteins or therapeutic lead proteins used in early preclinical research. At the same time, early protein supply is often characterised by low expression rates of the cell lines used and the need to produce many different proteins in a given period of time to fuel the development pipeline. We have therefore developed a novel, easy to use, benchtop perfusion bioreactor that is based on rocking motion agitation (BIOSTAT CultiBag RM 20 perfusion, Sartorius Stedim Biotech). In this study, we present data on the high cell-density culture of a mAb-producing Per.C6 model cell line.


A Per.C6 cell line expressing a monoclonal anti–EpCAM (epithelial cell adhesion molecule, a cancer marker) immunoglobulin G (IgG) antibody (licensed from Crucell) and commercially available, chemically-defined serum-free medium, CDM4 PerMab (Hyclone), was used for the batch and perfusion phases of the cultivation. Inoculum was produced in two 1 L polycarbonate Erlenmeyer flasks (Corning) with 300 mL culture, each incubated in a Certomat CT plus shaker incubator (Sartorius Stedim Biotech) at 85% humidity, 37 °C, 5% CO2 and a shaker rate of 240 rpm.

The bioreactor bag (CultiBag RM10 perfusion, Sartorius Stedim Biotech), equipped with single-use optical pH and DO probes, was filled with 5 L of medium and equilibrated to process temperature. After calibration of the optical sensors, the reactor was inoculated with a seeding density of 5 x 105 cells/mL. The initial process parameters were 19 rocks/min, a rocking angle of 6° and a temperature of 37 °C. The pH was controlled at pH 7.1 with CO2 and 1M Na2CO3. Dissolved oxygen (DO) was controlled at 40% by automatically adjusting the gas flow rate (range 0.3 to 3 L/min) and the composition of N2, air, and O2. For comparison to the perfusion process, a batch culture was performed using the above mentioned parameters.


An empty bag was welded to the perfusion outlet of the bioreactor bag, and a 50 L media bag was welded to the perfusion inlet and placed on a balance. Perfusion was started at an exchange rate of 0.5 reactor volumes/d, which corresponds to a feed and harvest flow of 0.105 kg/h. The perfusion rate was subsequently increased based on the measured glucose and glutamine concentration as detailed below. As the bioreactor controller automatically increased the gas flow rate and oxygen content in the process gas, the angle and rocking rate were adjusted manually from initially 6° to 10°, and 19 rocks/min to 23 rocks/min. Due to the increased angle and rocking rate, oxygen transfer from the headspace to the liquid phase could be further improved.

Samples were taken at regular intervals for offline pH measurement (Sartorius PB-11 pH meter) and monitored for cell count (Nucleocounter, Sartorius Stedim Biotech), glucose, glutamine, and lactate concentration (Bioprofile 100plus Analyzer, Nova Biomedical) and antibody concentration (ELISA).


In batch mode, a cell density of 17 × 106 cells/mL at a viability of 98% was reached (see Figure 1). This result is typical for a pH and DO controlled bioreactor batch culture of this specific Per.C6 clone and the PerMAB medium used. Once the substrate was depleted, cells went into the stationary phase. In contrast, in an uncontrolled process, e.g., in a shaker flask, the cell density typically reaches about 12 × 106 cells/mL under similar conditions (data not shown).

Figure 1: Viable cell density and viability of batch and perfusion process.

In perfusion culture using the BIOSTAT CultiBag RM 20, we were able to reach a very high cell density of 150 × 106 cells/mL at a viability of 99%. Up to now, the highest reported cell density of a wave-type, single-use bioreactor has been 60 × 106 cells/mL (7). Typically, we achieve about a 10-fold higher cell density in perfusion culture using this bioreactor set up compared to the batch results (data not shown). The total cell yield was 7.5 × 1011 cells derived from 5 L of reactor culture.


It was possible to obtain very high cell densities without the need for sophisticated equipment, fed-batch strategies or process optimization by using off-the-shelf equipment with powerful control capabilities. The L–glutamine and glucose concentrations were used to initiate the perfusion mode and to determine the medium exchange rate. When the glutamine concentration dropped below 1 g/L, which was 66 h after inoculation, addition of fresh medium started. In order to maintain the glutamine concentration at above 0.5 g/L, the perfusion rate was increased step-wise. After 138 hours, the perfusion rate was increased to 1.5/d (0.315 kg/h), and after 163 hours, to an exchange rate of 3/d (0.63 kg/h) (see Figure 2).

Figure 2: Glucose, glutamine and lactate concentrations over the course of the culture.

The perfusion process was terminated after 9 days when the perfusion membrane started to block; the perfusion control system detected the reduced harvest flow rate and shut down the pumps. Consequentially, the remaining substrate was consumed, lactate accumulated, and the logarithmic growth phase ended. Typically, in a production setting, one would directly move to harvest of the supernatant. In our set up, we wanted to test the limit and robustness of the bioreactor system. In the case of cell production, e.g., for inoculation of a production bioreactor or large scale cell banking in cryobags, the culture would be terminated before reaching the limit of the perfusion membrane.

The signals from the disposable optical DO and pH sensors that are integrated into the single use bioreactor bag were used to control the respective process parameters. pH was controlled mainly by adjustment of CO2 in the gas stream. Disposable optical sensor technology is a rather new, but versatile, technology which is suitable even for challenging applications like this high cell density perfusion culture. Regular off-line measurement using a conventional pH meter served to check the precision of the optical pH sensor, and to determine if recalibration of the optical pH sensor would be necessary to compensate for a potential drift. Also, changes in ionic strength might affect the accuracy of the sensor readout. During culture, the ionic strength might be influenced by accumulation of lactate or ammonia or other metabolic by-products. A comparison of the pH values obtained from the disposable optical sensors and the off-line values is shown in Figure 3. At the beginning of the perfusion process, one recalibration of the optical pH sensor based on the off-line value was performed. During the course of the high density culture, pH could be maintained at the set point despite a change of the perfusion rate, which temporarily led to an increased pH in the bioreactor.

Figure 3: Biorector pH control and comparison between online and offline pH data.


Process parameters affecting DO in a rocking motion bioreactor are shown in Figure 4. The DO could be maintained throughout the whole high cell density culture at approximately 40%. The rocking rate was increased manually from 19 rocks/min to 21 rocks/min, and finally to 23 rocks/min. The angle was increased from 6° to 7° and finally to 10°. As the rocking rate and angle increased, the wave formation in the bag became stronger, hence increasing the surface exchange rate at the gas-liquid interface and ultimately the oxygen transfer rate (OTR) of the bioreactor. The OTR could further be increased by adding pure oxygen to the process gas. Since we worked with moderate rocking rates, which leave room for further increase for most common cell lines, it can be concluded that we are still far away from the upper limit of the oxygen transfer capacity that can be achieved in this type of bioreactors. It should be noted that the increase of the rocking angle from 7° to 10° at process time 160 h reduced the requirement of pure oxygen in the process gas dramatically.

Figure 4: Dissolved oxygen concentration (DO), gas flow, percentage of pure oxygen (O2) in gas stream, rocking rate and angle.


We have compared the productivity of recombinant anti–EpCAM in the bioreactor system operating in batch and perfusion mode. During the perfusion run, approximately 50 L of medium had been consumed, which was 10 times the amount that was used in batch mode. The maximum viable cell concentration was increased by a factor of 10. The total IgG yield in the perfusion run was 5900 mg, which was approximately 30 times greater than in the batch mode, where the total amount was 205 mg (see Figure 5). This illustrates that the perfusion system is perfectly suited for the fast production of high amounts of proteins such as monoclonal antibodies.

Figure 5: Total Immunoglobulin (IgG) amount in batch and perfusion cultivations.


The authors demonstrated that an easy to use, off-the-shelf, single-use, bench-top perfusion bioreactor is capable of producing and supporting very high cell density cultures of up to 150 × 106 cells/mL. This approach can be applied to mediocre cell lines in order to quickly produce the amounts of recombinant proteins typically required during drug discovery or early development (e.g., for high throughput screening, or early preclinical studies in animals), and formulation development. Furthermore, this is a simple approach to improve the efficiency of seed production in large scale production. A typical seed train for a 1000 L bioreactor consists of a 50 L and 200 L bioreactor applying a dilution factor of 1:5 to 1:10 at maximum. Assuming an inoculum cell density of 5 × 105 cells/mL at 1000 L scale, which is equivalent to 5 × 1011 cells, this density can be achieved by approximately 20 L of a high cell density culture grown in a rocking-motion perfusion bioreactor. Certain limitations of this technology could be expected in larger bag systems, such as 100 L or 200 L rocking motion bags with integrated membranes for cell retention. The oxygen transfer rate in these systems is usually lower, and the ratio of membrane surface area to culture volume can not be scaled in a linear fashion due to size restrictions. While cell densities of 30-40 × 106 cells/mL have been reported for these larger scale rocking motion systems, it is likely that extreme high cell densities as reported in this article may not be reached.

THORSTEN ADAMS* is product manager, single use bioreactors, UTE NOACK is head of process engineering, upstream technology, TANJA FRICK is in upstream technology, GERHARD GRELLER is director, upstream technology, and CHRISTEL FENGE is vice-president, marketing and product management, fermentation, all at Sartoius Stedim Biotech, Göttingen, Germany,


1. J.M. Coco-Martin, and M.M. Harmsen, BioProcess Int. 6 (S6), 28–33 (2008).

2. A. Salmen et. al., BioProcess Int. 7 (11), 34–39 (2009).

3. C. Fenge and E. Lüllau, "Cell Culture Bioreactors" in Cell Culture Technology For Pharmaceutical and Cell-Based Therapies, S. Ozturk, and W-S Hu, Eds. (Taylor & Francis, New York 2006), pp. 155–224.

4. J. Lim, et. al., BioPharm Int. 24 (2), 54–60 (2011).

5. R. Heidemann et. al., Cytotechnol. 38 (1–3), 99–108 (2002).

6. A. Amanullah et. al., Cell Culture Engineering IX conference, (Cancún, Mexico, March 2004).

7. R. Ohashi et. al., Biotechnol. Prog. 18 (6), 1292–1300 (2002).