Scalability of the Mobius CellReady Single-use Bioreactor Systems

The Mobius CellReady family of products includes the bench-scale (3-L), small-scale (50-L) and pilot-scale (200-L) bioreactor systems that enable cell growth in volumes appropriate for early process development through clinical batch production. The small-scale Mobius CellReady 3-L bioreactor is a rigid, stirred-tank bioreactor (see Figure 1A) while the Mobius CellReady 50-L and 200-L bioreactor systems (see Figures 1B and 1C) are inflatable stirred-tank bioreactor process containers used in stainless-steel vessels. Table I outlines the features of all three. Both the Mobius CellReady 50-L and 200-L bioreactor process containers use the Mobius SensorReady technology for process monitoring and control. The Mobius SensorReady assembly is an external loop that is connected to the bioreactor process container that allows for a configurable number of probes and sensors to be used.

Image courtesy of EMD Millipore

The ability to scale up a biomanufacturing process is essential for process development and the production of biologics. Bioreactor process set points, acceptable ranges, and general operating parameters used at the large scale are commonly based on those developed at the benchtop or small scale where experimentation is more cost effective and efficient. Large-scale performance and production expectations are often established based on results obtained at smaller scales. It is, therefore, important that the process parameters developed at the small scale are readily transferrable to the larger scale.

Figure 1: (A) 3-L Mobius CellReady bioreactor; (B) 50-L; and (C) 200-L Mobius CellReady bioreactor process containers.

Challenges to bioreactor scale-up occur because even when two geometrically similar tanks are used, it is not possible to simultaneously maintain key bioreactor characteristics such as shear, mixing time, and oxygen mass transfer coefficient (k_La) identical in both the large and small tanks (1). Other variables, such as bubble size and distribution, nutrient regulation and delivery, and process control capabilities may also contribute to variable performance results across scales. Ultimately, successful scale-up is determined when comparable process performance endpoints such as cell growth, cell viability, protein production, and product quality are achieved. The probability of meeting these criteria can be increased when the bioreactor systems are well-characterized and the process design space is better understood.

Table I: Comparison of the features of 3-L, 50-L, and 200-L CellReady bioreactors.

In this study, several key engineering parameters k_La, power-per-unit volume, Reynolds number (Re), mixing time, and tip speed were characterized for the three different sized single-use bioreactor process containers. Chinese hamster ovary (CHO) cells were then cultured in each of these bioreactor systems based on maintaining equivalent power per unit volume as the primary scaling parameter. The results of this study define the characterization and process design space offered by the bioreactor systems and demonstrates the capability to achieve expected cell culture performance results across scales, thus demonstrating the scalability of the family of Mobius CellReady bioreactor systems.

EFFICIENT GAS TRANSFER

One of the most crucial scale-up parameters for bioreactors is mass transfer of gases. To provide an optimal environment for the cells, a bioreactor must be able to supply enough dissolved oxygen (O₂) for efficient cellular metabolism as well as maintain an appropriate level of dissolved carbon dioxide (CO₂). Sufficient O₂/air delivery is required not only to support cell growth and protein production but also to prevent excessive CO₂ accumulation in the media that can impact both of these critical performance endpoints (2). To assess the gas transfer efficiency of the Mobius CellReady bioreactor process containers, the volumetric mass-transfer coefficients (k_La) for oxygen were measured in each bioreactor process container using the static gassing out method.

k_La values were determined by filling the bioreactor process containers to the maximum working volume with a mock media (1X phosphate buffered saline (PBS), 2 g/L Pluronic F-68, 50 ppm Anti-foam C), and setting the temperature to 37 °C. The dissolved oxygen was stripped from the bioreactors by supplying nitrogen gas via the microsparger. Once the dissolved oxygen (DO) concentration reached less than 2% air saturation, the nitrogen supply was turned off and air was sparged through the microsparger at flow rates ranging from 0.0025 vvm to 0.5 vvm at two different agitation rates. The DO concentration was recorded over time until the dissolved oxygen concentration in the media plateaued at the fully saturated value (100% air saturation). An air overlay gas was not used in these studies.

k_La Calculation

The k_La value for each trial was calculated from the linear portion of the DO vs. time graph. To avoid subjectivity in determining the linear portion, the uniform DO interval used for the calculation was between 10% and 90% air saturation. The k_La values shown represent the slope of the line created by plotting ln((C*–Ct₁)/(C*–Ct₂)) versus time (t₂–t₁), where C* is fully saturated liquid, Ct₁ is the percent air saturation at the initial time, Ct₂ is the percent air saturation at time 2, and t₁ and t₂ are the initial time and time 2, respectively.

Figure 2A: kLa Scalability of the 50-L and 200-L Mobius CellReady bioreactor process containers. Each bar represents the average kLa value of n=3 at each air-flow rate and impeller agitation rate tested. Error bars represent the standard deviation of triplicate determinations.

As detailed in Table I, the impeller and sparger design and placement in the 50-L and 200-L bioreactor process containers are similar; however, the 3-L bioreactor is significantly different from the 50-L and 200-L bioreactor process containers. For example, in the 50-L and 200-L bioreactor process containers the membrane polyethylene microsparger is located directly beneath the impeller, while in the 3-L bioreactor a sintered polyethylene microsparger is located off to the side of the impeller. Also, the impeller blade shape differs between the bioreactors. A pitched blade impeller is found in the 50-L and 200-L bioreactor process containers while a marine impeller is found in the 3-L bioreactor. As shown in Figure 2A, the 50-L and 200-L bioreactor process containers, at the same power per unit volume, exhibit comparable k_La values. These two bioreactor process containers can achieve k_La values ranging from 4 hr^-1 at the lowest agitation and air flow rates tested to 60 hr^-1 at the highest agitation and air flow rates tested. Given the differences in the volume and design between the 3-L bioreactor and the 50-L and 200-L bioreactor process containers, it is not surprising that, at the same gas flow rates and power per unit volume, the k_La values are not immediately comparable. However as shown in Figure 2B, by adjusting the air flow rates, scalable k_La values between all three bioreactor systems at similar power per unit volumes can be achieved.

Figure 2B: kLa Scalability of the family of Mobius CellReady bioreactor systems. Each bar represents the average kLa value of n=3 at each air-flow rate and impeller agitation rate tested. Error bars represent the standard deviation of triplicate determinations.

MIXING

Mixing is a crucial bioreactor performance characteristic because it is responsible for minimizing gradients and maintaining control within the cell culture environment. Good mixing in a bioreactor strives for sufficient fluid pumping and turnover throughout the system to effectively create a single homogeneous environment which can be accurately monitored and controlled. Good mixing should evenly distribute bioreactor contents, helping to minimize zones of uneven cell density, pH, temperature, dissolved gases, and nutrient or waste concentrations, while minimizing the shear stress imparted on the cells by the fluid dynamics or the mixing element itself.

Table II: Tip speed, Reynolds number, and ungassed power-per-unit volume values for the 3-L Mobius CellReady bioreactor process container.

Although no single parameter can guarantee comparable process performance between stirred tank bioreactor systems, choosing an agitation rate that matches energy dissipation or power per unit volume (W/m³ ) is a common first approach (3, 4). The impeller design, fluid density, and agitation rate are considered in the power per unit volume (P_o/V) equation, P_o/V = Np x ρ x n³ x d⁵ , where P_o is ungassed power, V is liquid volume, Np is impeller power number, d is impeller diameter, n is impeller agitation rate and ρ is fluid density. Using this scaling method, agitation rate is varied to maintain similar ungassed power-per-unit volume across vessels.

Table III: Tip speed, Reynolds number, and ungassed power-per-unit volume values for the 50-L Mobius CellReady bioreactor process container.

The highest shear zones in a stirred tank bioreactor are often described as existing within the impeller zone. Because the outer edge of the impeller blades create shear as they rotate through the liquid, the impeller tip speed is often considered during bioreactor comparisons. The impeller tip speed calculation, π x d x n, where d is impeller diameter and n is agitation rate, however, does not take impeller design into consideration. Reynolds number (Re) can be considered when estimating fluid conditions at a given agitation rate by providing a ratio of the inertial forces to the viscous forces where Re = (ρ x n x d² ) / µ. The fluid density is represented by ρ, n is agitation rate, d is impeller diameter, and µ is dynamic viscosity. The system is considered fully turbulent when Re > 10,000 (5, 6). The Reynolds number calculation is based on the assumption of a cylindrical tank with a centered rotating impeller and does not take impeller design into account. The Mobius CellReady 50-L and 200-L bioreactor process containers contain a bottom mounted, offset, pitched blade impeller and a baffle, which together increase the turbulence and improve overall mixing.

Table IV: Tip speed, Reynolds number, and ungassed power-per-unit volume values for the 200-L Mobius CellReady bioreactor process container.

Tables II, III, and IV outline various tip speed, Reynolds number and ungassed power-per-unit volume values calculated for a range of applicable agitation rates for the 3-L, 50-L, and 200-L Mobius CellReady bioreactor process containers.

Mixing was evaluated for the 50-L and 200-L Mobius CellReady bioreactor process containers by observing conductivity probe response curves measured at four locations (top, middle, bottom, and inserted in the probe port of the Mobius SensorReady loop) within the system. The average of these four probes is considered the system average. A salt solution was introduced at the liquid surface and the T₉₅ mixing time was determined for each of the four probe locations as the time when conductivity profiles had reached 95% of its final value. Each trial was performed in triplicate and results are shown in Figure 3. The Mobius CellReady 3-L bioreactor was evaluated in a similar manner, using just one conductivity probe inserted in the head plate probe port.

Figure 3: System average mixing times of the family of Mobius CellReady bioreactor systems. Data points represent the system average mixing time of n=15 individual trials. Error bars represent 1 standard error for each data point.

Results of these studies show that the Mobius CellReady 50-L and 200-L bioreactor process containers demonstrated similar system average mixing times at equivalent power-per-unit volume. Mixing times for the 3-L bioreactor are significantly shorter than mixing times observed in the larger bioreactor process containers for similar power per unit volume. This is to be expected as larger liquid volumes result in longer fluid paths. To achieve a well mixed condition in the same amount of time, the fluid velocity would have to increase for the condition in the larger volume tank. As a guide, fluid velocity is proportional to the square root of power per unit volume in turbulent conditions (i.e., when Re > 10,000) (5, 6). As a result, an increase in power-per-unit volume would be required to achieve similar mixing times at significantly larger scales. The system average mixing times for the Mobius CellReady bioreactor process containers, as determined by the multipleprobe experimental methods described above, do not dramatically increase with increasing power per unit volume across the agitation range tested. The 50-L and 200-L bioreactor process containers contain a single baffle, which coupled with the off-center, angled position of the bottom mounted impeller, provides good mixing dynamics even at lower power inputs. Empirical results show the mixing time to be 12–13 s from 2 to 15 W/m³ for the 3-L, 38–24 s from 5 to 30 W/m³ for the 50-L and 33–26 s from 1 to 25 W/m³ for the 200-L bioreactor process containers as displayed in Figure 3.

Table V: Comparison of operating parameters for 3-L, 50-L, and 200-L CellReady bioreactors.

CELL CULTURE

Maintaining a homogenous environment within the bioreactor is the most crucial criterion for a successful cell culture process. While different agitation strategies may be used to scale up a biomanufacturing process, power-per-unit volume is the most often used scaling parameter (3, 4). The second most important criterion for a successful cell culture run is efficient delivery of oxygen to maintain cell growth and productivity. Using these two criteria, a CHO cell culture batch process was performed in the 3-L, 50-L, and 200-L bioreactor systems using power per unit volume as the primary scaling parameter to demonstrate cell-culture scalability across the Mobius family of bioreactor systems. Further, gas flow rates were chosen to achieve similar k_La values for each vessel. The process parameters for each scale are outlined in Table V.

Figure 4A: Viable cell density and viability vs. culture time for a batch culture process at the 2-L, 50-L, and 200-L working volumes. Data were obtained daily from the Vi-Cell XR (Beckman Coulter).

To demonstrate scalable cell culture performance, parameters including cell growth, viability and metabolism were compared For this study, samples were analyzed daily on a Vi-Cell XR (Beckman Coulter), a BIOPROFILE FLEX system (NOVA Biomedical) and a Blood Gas Analyzer (Siemens Rapidlab 248). As shown in Figure 4A, the viable cell densities and viabilities were comparable between all three scales. As shown in Figure 4B and 4C, the nutrient profiles and metabolic rates between all three scales were also comparable. The comparable cell growth, viabilities and metabolic profiles demonstrate that the cell culture performance in all three Mobius CellReady bioreactor systems is scalable.

Figure 4B: Glucose and lactate Concentrations vs. culture time for a batch culture process at the 2-L, 50-L, and 200-L working volumes. Data were obtained daily from the BIOPROFILE FLEX system (NOVA Biomedical).

CONCLUSIONS

Successful scale-up of a biomanufacturing process is dependent on several factors including gas mass transfer, mixing efficiency and shear effects. An in-depth understanding of the process design space of the three Mobius CellReady bioreactor systems has beed developed through a series of experiments aimed at characterizing several key engineering parameters of the bioreactors. The three bioreactor systems, despite design and volume differences, are capable of achieving equivalent k_La values by adjusting the air flow rates at equivalent power-per-unit volume set points. In addition, similar system average mixing times can be achieved for the 50-L and 200-L systems within their power per unit volume operating range.

Figure 4C: Metabolic rates for culture days 1–5 for a batch culture process at the 2-L, 50-L, and 200-L working volumes. Data were obtained daily from the BIOPROFILE FLEX system (NOVA Biomedical).

Based on an understanding of the design space of each of the bioreactor systems, process set points were chosen for CHO cell cultivation in each of these. Maintaining equivalent power per unit volume was chosen as the primary scaling parameter and gas flow rates were chosen based on achieving similar k_La values at each scale. During a 10-day batch culture, comparable cell growth, viability, and nutrient metabolism were achieved in each bioreactor system.

The comparable cell culture performance results with all three bioreactor systems demonstrate the scalability of the family of Mobius CellReady bioreactor systems. The comprehensive characterization of several key engineering parameters resulted in a detailed understanding of the design space of each bioreactor. This understanding allows users to choose process set points that enable successful scale-up of biomanufacturing process across the Mobius CellReady family from the 3-L to 200-L scale.

JENNIFER DEKARSKI is global product manager for single-use bioreactors at EMD Millipore, Bedford, MA.

REFERENCES

1. M. Shuler and F. Kargi, F. Bioprocess Engineering Basic Concepts, 2nd ed., (Prentice Hall, 2002).

2. D. M. Marks, Cytotechnology 42, 21-33 (2003).

3. Z. Xing. et al., Biotechnol. and Bioeng. 103, 733-746 (2009).

4. S.D. Kaiser et al., CFD for Characterizing Standard and Single-use Stirred Cell Culture Bioreactors. In Computational Fluid Dynamics Technologies and Applications (InTech, 2011).

5. Dalian University of Technology. Mixing, Chapter 6. http://ceb.dlut.edu.cn/uploads/soft/110415/7-110415154330.pdf.

6. A. W. Nienow, Cytotechnology 50, 9-33 (2006).