Disposable Bioreactors for Cells and Microbes

Abstract

The focus in single-use bioreactor development is effective oxygen transfer, and the cultivation of cell culture processes and microbial cultures. Cell densities and productivities similar to conventional stirred tanks can easily be achieved in single-use bioreactors with the introduction of real process control. The Biostat CultiBag RM can reach KLa values of 43.2 h-1 and 12.9 h-1, in 2-L and 20-L cultures, respectively, at small scale. This article will show the productivities of Chinese hamster ovary cells (CHO), E. coli, and C. diphtheriae in the Biostat CultiBag RM using disposable sensor technology.

Early disposable bioreactors designs did not have good control capability, and growth was generally equivalent to shake flasks. Simple cultivation of cells and product for small-scale operations were possible, but a reusable stirred-tank system was required for real process optimization studies, where data could be logged and process parameters could be controlled and modelled. This was mainly because of the lack of good pH and dissolved oxygen (DO) control or at least a level of control similar to traditional stirred systems. Another limitation of rocking disposable bioreactors have been the limited aeration and agitation rates that could be achieved in systems that mix the culture by rocking back and forth and provide gassing by surface aeration alone. Such systems were mainly suited to cell cultures that exhibit low biomass concentrations and oxygen uptake rates. Also, mammalian cells are more fragile, which prevents the use of vigorous aeration and agitation strategies, making a rocking platform an ideal cultivation vessel. However, many modern cell culture processes require more strenuous aeration and agitation, and cells are becoming more robust. Therefore, the ability to have a good gas mixing strategy with feedback control is important. Microbial processes have a high demand for oxygen and are processed to high biomass concentrations. Most disposable systems cannot offer high processing rates. However, disposable rocking devices can be just as effective at the seed stage when good gas mixing and controls are available.

Sartorius Stedim North America

Recently, disposable stirred-tank bioreactor designs have been introduced that mimic traditional stirred-tank bioreactors (STR) and therefore gain more market acceptance. Most of these designs use reusable sensor technology with standard feedback control loops. The insertion of reusable sensors into a disposable system involves time-consuming tasks such as cleaning, sterilizing, and calibrating the sensors before aseptic insertion into the bioreactor. Disposable sensors are relatively new to the market and the market acceptance of such systems is slow because it requires detailed evaluation and validation. The operation of disposable sensors is different because no cleaning, sterilizing, or calibration is required at start-up. The sensors consist of membrane patches with an immobilized fluorescent dye that is able to detect the respective analyte (H+ or O₂). The sensor patches are already part of the bag assembly. They come gamma irradiated and there is no break in the bag seal. An optical fiber transmits light of a specific wavelength to the sensor and returns the luminescence response from the sensor back to the measuring amplifier. This fiber optic works through the bag film, thereby maintaining sterility at all times. Standard feedback control loops are possible and recalibration functions to correct drift over time.

The use of superior gas mixing systems with disposable sensors, and tighter process control, can be used to cultivate both modern cell culture and seed stage microbial cells successfully. Process monitoring and control of cell cultivation processes are required for culture reproducibility, modeling scale-up parameters, increasing achievable cell densities, extending batch age, and increasing productivity and yield. This article shows that using disposable sensor technology, comparable cell densities, viabilities, and titers were achieved in several cell lines, and in some cases, were even higher than in stirred tanks.

Materials and Methods

Determination of KLa values. The KLa-values for the CultiBag RM were determined by the gassing-out method for typical rocking speeds, angles, and gas flow rates. Ambient air and pure oxygen were used as process gasses.

Cultivation of Escherichia coli. To prepare the seed culture, E. coli BL21 (DE3), streaked out on LB agar plates, was used to inoculate two Sartorius-Stedim Biotech CultiFlask 50 disposable bioreactors (Sartorius-Stedim Biotech DF-050MB-SSH-4), each filled with 20 mL LB medium (10 g/L tryptone, 5 g/L yeast extract, 5 g/L NaCL). The seed culture was grown overnight at 37 °C and 150 rpm in an incubator (Incubator Sartorius Certomat). The BIOSTAT CultiBag RM 20 OPTICAL, part number DH-020LORM-1 using Cultibag RM 20L optical, part number DBO020L was filled with 1 L of LB medium, preheated to 37 °C, and inoculated with the preculture to reach an optical density (OD600) of 0.15. Cultivation was started with the following process parameters: temperature 37 °C, rocking speed 42 rpm, rocking angle 10° and airflow 0.5 lpm. Ambient air was used for oxygen supply. Growth was monitored by measuring the optical density in regular intervals. For comparing the growth characteristics, a baffled Erlenmeyer flask filled with 200 mL sterile LB medium was inoculated with E. coli BL21 (DE3) to a final OD600 of 0.15 and incubated at 37 °C and 150 rpm in an incubator.

Cultivation of Corynebacterium diphtheriae. The Biostat CultiBag RM 20 (Sartorius-Stedim Biotech DH-020-L-0-RM-1) was used for a feasibility study to replace a stainless-steel reactor for seed stage fermentation in vaccine production. The CultiBag RM 20-L (Sartorius-Stedim Biotech DBO020L) was filled with 10 L of CY-media (casamino/yeast) and inoculated with approximately 130 mL of a C. diphtheriae culture grown in an aspirator bottle to an OD590 of 8.66 resulting in an optical density (OD590) of 0.123 at the start of fermentation. Cultivation was started using following process parameters: temperature 32 °C, rocking speed 12 rpm, rocking angle 5.9° airflow 0.3 lpm. During the course of the cultivation, the rocking rate was raised to 42 rpm, the angle to 10°, and the airflow to 0.55 lpm to enable high mass transfer. Ambient air was used for oxygen supply. The optical density was measured in regular intervals throughout the process. For comparison, a stainless-steel reactor filled with 20 L of medium was inoculated with approximately 270 mL of a C. diphtheriae culture grown in an aspirator bottle to an OD590 of 8.66 resulting in an optical density (OD590) of 0.179 at the start of fermentation. The fermentation initial process parameters were: back pressure 5.0 psig, airflow 4.2 SLPM, and agitation 70 rpm to control the DO to ≥25%. During the fermentation, the cascade control increased the agitation to 250 rpm to maintain DO.

Cultivation of CHO-X. A commercial CHO cell line cultivated in a serum-free fed-batch stirred-tank culture and compared with CultiBag RM using disposable sensors for process control, referred to as the X cell line for the purposes of this paper. Temperature, pH, and dissolved oxygen were controlled at constant, product-specific set points throughout the production cycle. The details of these cultivation parameters are not given because of confidentiality.

Cultivation of CHO-A. A commercial process of a CHO cell line, referred to as the A cell for the purpose of this paper. Details of the cultivation parameters are not given because of confidentiality. The CultiBag was compared to four 2-L stirred-tank bioreactors and Wave Cellbags on Wave rocking machines. Seed cultivation was performed in shake flasks and Wave Cellbags. The inoculum conditions and feed-control strategy in the CultiBag and the Wave bioreactors was similar to the conventional stainless-steel tank bioreactors. The operational parameters of the CultiBag were similar to those for the Wave Bioreactor. The pH and DO set points for the CultiBag were similar to those for the stirred-tank bioreactors. For the stirred-tank bioreactors, pH was controlled by automatic addition of 1 M NaOH and sparging CO₂ on demand. DO was controlled by sparging pure O₂ on demand. For the CultiBag bioreactor, pH was controlled by continuous automatic surface aeration of CO₂ automatic addition of 1 M NaOH. DO was controlled by continuous automatic surface aeration with the four-stage cascade control system. Cell culture performance was analyzed using CEDEX Cell Counter (Innovatis AG) for cell density and viability; YSI analyzer 2700 Select (YSI, Inc.) for cell culture metabolites: glucose/lactate and L-glutamine/L-glutamate; NOVA BioProfile 100 Analyzer (Nova Biomedical) for ammonium level; ABL-5 radiometer (Radiometer Medical) for pH/pO₂/pCO₂; and RP HPLC for productivity (titer).

Cultivation of CHO-S and CHO-K1. Two different CHO sub clones, CHO-S and CHO-K1, were compared for their growth characteristics and metabolic profiles. The cells were cultured in repeated batch mode in the reusable bioreactor (Sartorius-Stedim Biotech Biostat B-DCU 2; 10-L UniVessel with pitched three-blade impellers and ring sparger) as well as in the single-use CultiBag RM 20-L. The CHO-S cultivations were run in a head-to-head comparison, while for the comparison of CHO-K1, two similar bioreactor runs were assessed. The CHO-S seed culture was grown in a stirred-tank bioreactor and 2,000 mL of the seed were used to inoculate the Biostat B-DCU reusable stirred tank bioreactor as well as the Biostat CultiBag RM. The final volume inside both bioreactors was 10 L of Power-CHO 2 (0.1% Pluronic, 6 mM L-glutamine, Lonza) media. The initial cell density was ~1 x 10⁶/mL. The process parameters used are listed in Table 1.

Table 1. Process parameters used to compare B-DCU STR and CultiBag RM20

After 72 h of cultivation, the culture was divided 1:5 in the same cultivation vessel, i.e., a repeated batch process was carried out. Eight L of media containing the cell suspension were harvested and replaced with 8 L of fresh media. Samples were taken in regular intervals and the viable cell number was determined. Lactate and glucose levels were measured using the Glucose/Lactate Analyzer YSI 7100. The cultivation of CHO-K1 was carried out in a similar manner. The only difference was that the B-DCU STR and the CultiBag RM were inoculated from individual seed cultures to an initial cell density of 5 x 10⁵/mL in the STR and 1 x 10⁶/mL in the CultiBag, respectively. The temperature in the CultiBag was set to 37 °C.

Figure 1

Results and Discussion

The kLa-values for the Biostat CultiBag RM at full-rocking speed, angle, and gas flow using air were 22.0 h-1 for the 2-L system and 6.0 h-1 for the 20-L system. Using pure oxygen as process gas, the KLa-values increased to 43.2 h-1 and 12.9 h-1, respectively.

Figure 2

Growth of the C. diphtheria was assessed (Figure 1). The optical densities of the cultures were measured and the DO was monitored in the CultiBag RM using the disposable optical DO probes. During the course of the cultivation, the DO dropped to around 25% because of the limited oxygen transfer by headspace aeration. However, the culture reached an OD590 of five in the CultiBag RM and 7.3 in the stainless-steel fermenter after an 8-h cultivation period, indicating that the BIOSTAT CultiBag RM is well suited for cultivation of C. diphteriae for seed inoculum.

Figure 3

CHO-X evaluation. A commercial CHO cell line was evaluated using standard rocking and cultivation parameters. The cell growth shown in Figure 2 was slightly lower than the STR but the viability was higher at the end (Figure 3). Figure 4 shows that the overall production was greater than stirred-tank bioreactors.

Figure 4

To demonstrate the reliability of disposable sensor technology, the real time data plot is shown in Figure 5. The pH control is shown to be within +/–0.1 pH units and equivalent to traditional sensors.

Figure 5

CHO-A cell line evaluation. Overall cell culture performance of the CultiBag bioreactor was comparable to the STR (Table 2). The final RP HPLC titer value and the average specific productivity of CultiBag were higher than that of the STR. Temperature control was comparable to a conventional bioreactor (within +/- 0.5 °C). Cell culture performance of the CultiBag bioreactor was comparable to the STR's. Some early stage proportional-integral-derivative controller (PID) loops were tuned to achieve better control of pH and DO. The average specific productivity of CultiBag was higher than that of the STR's. Without pH and DO control, the Wave bioreactor demonstrated different cell culture performance compared to the CultiBag bioreactor in terms of higher final cumulative viable cell count (CVC) value, lower RP HPLC titer, and lower average specific productivity.

Table 2. Cell culture performance of the CultiBag bioreactor compared to the STR

CHO-S and CHO-K1 evaluation. The growth characteristics of the CHO-S and CHO-K1 were also similar, with both cell lines reaching comparable levels in both types of bioreactor (Figure 6). The CHO-S showed better growth, reaching higher cell densities then the CHO-K1, probably reflecting the better adaptation to suspension culture.

Figure 6

Figure 7 shows the glucose and lactate profile of the CHO-S cultivation. In this instance, the glucose consumption of the cells grown in the B-DCU STR was higher. Consequently, the lactate build up was also higher. This observation is congruent with the slightly higher viable cell density reached in the B-DCU.

Figure 7

Conclusion

The Biostat CultiBag RM is shown to reach KLa values of 43.2 h-1, and 12.9 h-1, respectively, at small scale for 2-L and 20-L systems. The reactor easily provides oxygen transfer rates sufficient for medium cell densities in seed stage microbial fermentation. E. coli, still the preferred microbial host for recombinant protein production, was shown to be successfully cultivated to medium cell densities. In addition, growth of the C. diphtheria reached an OD590 of 4.95 in the CultiBag RM and 7.26 in the stainless-steel fermenter after an 8 h cultivation period, indicating that it was successfully cultured in preparation of a seed inoculum despite a drop in dissolved oxygen concentration. A second trial confirmed these results. Despite the high demand for aeration and agitation in microbial cultures, we have demonstrated that the Biostat CultiBag RM is suitable for microbial seed stage cultivation. The superior gas mixing capability optimizes headspace aeration and creates a suitable vessel for microbial cultivation.

When comparing a rocking bioreactor with and without process control as demonstrated by the CHO-A process, a different cell culture performance is demonstrated. The rocking device with control was the CultiBag system with feedback control loops for pH and DO to disposable sensors, and the rocking device without any pH and DO control was the Wave bioreactor. Both systems exhibit similar rocking motion mixing. The Wave bioreactor exhibits a rocking motion, which starts and stops, forcing the culture to create turbulent waves for mixing. The Wave bioreactor was gassed with a constant flow rate of air and CO₂, without any feedback control mechanisms. A constant flow of CO₂ can cause pH fluctuations to occur as the uncontrolled CO₂ dissolves into the media and converts to carbonic acid. The CultiBag offers a gentler and continuous rocking motion with an inbuilt feedback control mechanism, adding CO₂ on demand and only when the culture needs it. The constant flow of air at the liquid surface prevents CO₂ accumulation and pH fluctuation. Both systems had similar rocking and temperature control. The Wave bioreactor was shown to have a higher final CVC value, a lower RP HPLC titer, and a lower average specific productivity, thereby indicating that process control offers many advantages for cell cultivation.

When comparing the CultiBag side-by-side with a conventional stirred-tank bioreactor system in the CHO-A process, the overall cell culture performance of the CultiBag bioreactor was comparable to the STR. In this case, both systems had good control systems in place, but the STR used reusable sensors and the CultiBag used disposable sensors. This demonstrates the comparability between conventional and disposable sensor technology in a live process. Additionally there was no oxygen limitation in the rocking bioreactor compared with the stirred tank. In fact, the final RP HPLC titer value and the average specific productivity of CultiBag were higher than that of the STR.

This is further demonstrated by the CHO-S and CHO-K1 process, which also shows comparable growth and productivity to the STR. Additionally, real time data plots show the online process control of the disposable sensors in the CHO-X evaluation, demonstrating that the process control of disposable sensors are similar to reusable sensors. We can therefore conclude that, not only do disposable sensors provide adequate control of processes, but that these rocking bioreactors can exhibit cell growth productivities similar to a traditional STR and, in some cases, even higher. With the added benefit of disposable systems in terms of cleaning, sterilizing, reduced cross contamination, and flexibility, it is obvious to see why the industry is focused on disposable systems.

Millie Ullah, PhD, is a product manager, disposable bioreactors, at Sartorius Stedim North America, Edgewood, NY, 631.254.4249, millie.ullah@sartorius-stedim.comTerry Burns is a senior technology scientist at Wyeth Vaccines, Sanford, NC. Amardeep Bhalla, PhD, is a senior scientist at Schering-Plough, Summit, NJ. Hans-Wilhelm Beltz is manager of recombinant technology at CSL Behring, Preclinical R&D, Marburg, Germany. Gerhard Greller, PhD, is a scientist of R & D pharma process and Thorsten Adams, PhD, is an application specialist of fermentation technology, both at Sartorius Stedim Biotech GmbH, Goettingen, Germany.