Development of a Novel Rolling-Tube Cell Culture Platform and Demonstration of System Feasibility

August 1, 2010

Rolling-tube system balances scale-up accuracy and thoroughput.


This article describes and demonstrates the performance of a novel disposable culture platform, the rolling tube, which we have developed, tested, and found superior, in many aspects, compared with current platforms being used in process development applications such as media screening and clone selection. The rolling-tube platform provides a balance between throughput and scale or volume (necessary to perform analytical assays); it also is robust, easy to set up, and ergonomic to use, as well as versatile, saving time and money. Importantly, results obtained with this system are applicable to larger platforms used in commercial production. This platform is now being implemented in our laboratories in California and Europe.

Throughput, robustness, and low cost are major system requirements in cell culture process development projects, ranging from clone selection to media screening. A desirable system also should allow for sufficient sample volumes to perform a comprehensive analysis for testing metabolites, viable cell growth, product titer, and quality attributes. Ideally, the system would closely mimic the outcome in production bioreactors and thus provide high confidence in obtaining a beneficial outcome following scale-up and implementation.


Disposability is now common across a growing spectrum of platform scales.1 Traditionally, 2-L roller bottles on a rolling device placed in a humidified, temperature and CO2-controlled incubator served as a common screening system for the aforementioned applications, toward implementation in perfusion, batch, and fed-batch bioreactors. They are inexpensive and disposable, provide a low-shear culture environment, and often demonstrate predictability of culture behavior following scale-up, making them a good choice compared to other small-scale platforms such as spinner and shake flasks. However, their size, rolling equipment space requirements, and cumbersome handling are major drawbacks that limit throughput. They also are wasteful because smaller volumes of culture and spent-media usually are required for titer, potency, and spent-media and metabolite analysis.

The volume of the 2-L roller bottles is not large enough to allow purification and a comprehensive characterization of the product or impurities. Smaller, low-cost platforms such as T-flasks and plate systems have advantages in throughput, space use, and ease of use while allowing for sufficient sample volume for analysis and being predictive of performance at large scale. In previous experiments, we have found that T-flasks performed similarly or even better than the large roller bottles (not shown). However, both flask and plate systems have drawbacks such as ergonomics and gas transfer and condensation problems resulting from their design combined with lower volume (especially multi-well plates) that can significantly affect culture performance. Therefore, we set out to develop a platform that would preserve the advantages of the traditional roller bottle along with those of the more advanced flask and plate systems. Here, we describe the performance of the roller tube platform compared to two small-scale platforms: T-flasks and 6-well plates.


The rolling-tube platform combines two low cost, off-the-shelf products to create a scaled-down roller bottle platform. The platform comprises 50-mL tubes with vented caps, allowing for superior gas exchange (available from TPP and Sartorius) placed on a rolling device with speed control. These tubes (trade name Cultiflask 50) have been used to grow cells by mounting them vertically on an orbital shaker platform.2,3 Mixing using a rolling motion provides good aeration but low shear, which is important for fragile, agitation-sensitive cells, and certain media formulations. The tubes are angled to prevent the media from wetting the filter on the cap, and the rolling device is placed inside a humidified, temperature- and CO2-controlled incubator. The device can operate continuously inside the incubator. The rolling-tube platform can operate at a throughput of 16 tubes per platform, and because the device is stackable, it requires a small footprint. The 50-mL growth vessel also can serve as the harvest vessel, facilitating media exchanges (e.g., during cell splitting) or cell harvest through standard centrifugation. The current set up allows for fill volumes of up to ~14 mL. This volume may be increased in a future improvement of tube design by centralizing the vents on the tube's cap. The rolling device may be improved as well. The current model is available as a "rock and roll" model. We had to request that the rocking motion be eliminated, requiring customization by the manufacturer to a "roll-only" model to prevent filter wetting during culture.


Proprietary commercial suspension-adapted mammalian production cells and media were used throughout the study. Incubator settings (temperature, % CO2, and humidity) were set to correspond to commercial manufacturing settings used in perfusion bioreactors. Intervals between cell passaging and target seeding density were optimized to correspond to the cell-specific perfusion rate of the production-scale bioreactors. Data are presented for a four-day batch culture. Up to 16 vented-cap 50-mL tubes can be placed on the 9-roller platform that was operated at 30 rpm. No significant difference in culture performance was observed up to the maximum speed of 55 rpm. T-75 flasks and 6-well plates, both nontissue culture treated to prevent adhesion of the suspension-adapted cells, were placed on a standard orbital shaker operated at 35 rpm.

Both the rolling platform and shaker were placed in the same incubator. A blood gas analyzer (BGA) was used to determine pH, pCO2, and pO2. Cell count, viability, nutrients, metabolites, and osmolality were determined using a NOVA BioFLEX instrument. Daily samples of 1.3 mL were taken for BGA and NOVA analysis. Potency (also used for specific productivity calculation) was determined using a product-specific assay. To set up the experiments, cells passaged and scaled up using T-75 flasks were used to inoculate the three platforms at 0.35 x 106 viable cells/mL, as follows: two T-75 flasks (duplicates) with 20 mL working volume, two 6-well plates with 2.9 mL working volume/well, and five 50-mL rolling tubes (two duplicates and three for individual sampling) with 14-mL working volume. This setting was meant to test robustness of the rolling tube platform across different working volumes. One set decreased in working volume along the culture days because of daily BGA and NOVA sampling, whereas in the other set of tubes, one tube was sampled each day, so the working volume remained constant.

To determine the effect on culture performance in the rolling tubes of volume reduction caused by daily sampling, (considering the greater percentage change in volume when sampling from small working volume platforms) three rolling tubes were included to be sampled only once, each on a different day, throughout the four-day batch culture. In this set, a 1.3-mL sample was taken each day from a different tube on day 2, 3, or 4. Potency samples (1.5 mL) were taken on days 3 and 4 for each culture to test for productivity.

Figure 1. On each of the four days of the study, samples were taken from: A) the same 50-mL rolling tubes (volume reduced along study), B) a single 50 mL rolling tube sacrificed for sampling (constant volume along study), C) T-flasks, and D) 6-well plates. Error bars represent the average of two samples (duplicates). The same cell pool was used to seed the cultures on day 0, therefore the values are the same across all platforms. A and B values for day 1 are the same because the tube sampling regimens were shared at the first sampling point.


Growth and viability profiles of the different platforms are shown in Figure 1a and 1b. The rolling tubes (A and B) and T-flask (C) platforms demonstrated equivalent growth rates and cell densities. No difference in growth rate was observed between the roller tubes being sampled successively throughout the study (A) and roller tubes sampled once, a different rolling tube each time on a different day (B). Six-well plates (D) demonstrated very little growth during the study, which apparently resulted from high osmolality, as explained below. Viability was very high (>96%) in all culture platforms throughout the length of the study.

Figure 2. On each of the four days of the study, samples were taken from: A) the same 50-mL rolling tubes (volume reduced along study), B) a single 50 mL rolling tube sacrificed for sampling (constant volume along study), C) T-flasks, and D) 6-well plates. Error bars represent the average of two samples (duplicates). The same cell pool was used to seed the cultures on day 0, therefore the values are the same across all platforms. A and B values for day 1 are the same because the tube sampling regimens were shared at the first sampling point.

Nutrient and metabolite profiles are shown in Figure 2, and included: glutamine, glutamate, glucose, lactate, pH, pO2, pCO2, and osmolality. Rolling tubes of 50 mL (with either the same or different tubes being sampled each day) and T-flasks demonstrated similar profiles for all analytes tested. Glutamine (2a) and glucose (2c) were consumed during the culture in all platforms but the consumption rate was markedly lower for the 6-well plates. Likewise, lactate (2d) production was higher for the 6-well plates. Glutamate (2b) also was produced during the batch culture in all platforms, but the levels varied between experimental cases. Again, pH (2e) and gas (2f, 2g) levels were similar between the rolling tube and T-flask platforms but were relatively lower in the last two days for the 6-well plates. Osmolality (2h) was similar between the 50-mL rolling tubes (with either the same or different tubes being sampled each day). Slightly higher osmolality was observed with the T-flasks. However, 6-well plates demonstrated increasingly higher osmolality levels along the culture period. We have determined that these differences resulted from higher evaporation. Condensation was noticeable on the lids of the flask and plate culture systems. Media volume reduction was most significant in the 6-well plate culture platform (data not shown). The data demonstrate that media volume loss is minimal for the rolling tube platform. Because the culture contacts the entire surface of the rolling tubes, condensation is drawn back into the culture, whereas small amounts of condensation will accumulate over time on the surfaces of the T-flask that do not contact media.

Figure 3. Potency (a) and specific productivity (b) results on days three and four, based on a product specific assay. Samples were collected from 50 mL rolling tubes (A), T-75 flasks (B), and 6 well plates (C). Data were normalized to T-75 flasks taken as 100%

As shown in Figure 3, potency and product-specific productivity were similar for both 50 mL rolling tube and T-flask platforms. In contrast, potency was lower for the 6-well plate although specific productivity was apparently higher.

Performance of the rolling-tube platform is robust across a wide range of volumes. The data demonstrate that the rolling tubes performed similarly whether or not volume was gradually reduced as samples were drawn out over the course of the analysis (compare A and B in Figures 1 and 2). In group A, 1.3 mL of media was drawn from the (duplicate) 50-mL rolling tubes per day on days 1 and 2, plus an additional 1.5-mL samples for submission to potency assay on days 3 and 4. From 14 mL of initial culture, there was approximately 8.6 mL or 61% original volume left in group A tubes by day 4. Thus, decreasing volume has no apparent impact on growth and viability (Figure 1), nutrients, or gas transfer (Figure 2).

Rolling tubes and T-flasks demonstrated comparable culture performance and metabolite profiles, as well as product potencies (Figures 1, 2, and 3). In comparison, 6-well plates demonstrated inferior growth to either T75 flasks or 50-mL rolling tubes. Cell density (Figure 1) reached a maximum of ~0.7 x 106 vc/mL compared with ~2.0 x 106 vc/mL for the other two platforms. Glutamine consumption (Figure 2a) was lower, although no significant difference in glucose consumption was observed (Figure 2c). Lactate levels increased (Figure 2d) possibly because of less efficient gas transfer in the 6-well platform, causing a shift toward anaerobic metabolism. The significant increase in osmolality over time (Figure 2h) was notable. This phenomenon was investigated using 6-well plates with uninoculated media.

Evaporation was the apparent reason for the significant rise in osmolality and in glutamine and glucose metabolites in the 6-well plate platform (data not shown). Total production of the desired protein (Figure 3a) also was much lower in 6-well plates than in the other platforms, though per cell specific productivity (Figure 3b) was ~40% greater. This could have resulted from the much lower viable cell densities (Figure 1a) in the 6-well plate platform caused by the higher osmolality resulting from evaporation or a shift in metabolism toward production (rather than growth) because of osmotic stress.4

The performance of the small-scale systems also was representative of commercial production on the basis of spent media analysis performed on cultures from both scales using the same cell line and media. It should be noted that seeding density and passage schedule have been designed to emulate cell-specific perfusion rate in production bioreactors. In Figure 4, the amino acid profile of small scale is compared to production-scale culture. The similarity in the profile (i.e., in the extent of consumption, or for some amino acids, the extent of production) between the scales strongly suggests that the optimized small-scale systems are able to predict the performance in bioreactors in applications such as media or clone screening.

Figure 4. Amino acid analysis of small scale versus production-scale spent media. The percent change in levels of different amino acids in spent media relative to the starting fresh media is shown for production scale runs and for representative screening scale runs (T-flask).


In this study, we assessed the performance of the rolling tube platform compared to the T-75 flask and the 6-well plate platforms. Although T-flasks recently have replaced the large roller bottles in many laboratory-scale applications on the basis of culture performance (internal data, not shown), they also have major drawbacks including throughput and ease of operation. Other small-scale systems used in a variety of screening applications, such as micro-bioreactors, are not covered here because they have major drawbacks, most notably cost.5 This study demonstrates that the rolling-tube system is a platform of choice for development projects because it performs similarly to the T-flask (and the roller bottle, data not shown) but is much more user friendly and has higher throughput. Evaporation minimization, pH control, and gas transfer rates are superior in the rolling tube system as a result of the mixing through rolling motion where media is in greater contact with the vessel surface and because of the geometry of the tubes that contain vented caps.

On the other hand, the 6-well plate platform demonstrated inferior performance resulting from evaporation and gas transfer problems. Data including metabolite analysis, cell growth rate, viability, specific productivity, and product titer and potency show the performance of rolling tubes is comparable to the T-flask and superior to 6-well plates. The rolling tubes also demonstrated robust and consistent performance at a broad range of working volumes. This allows for the designing of fill volumes tailored to the volumes and frequency required for sampling (e.g., for metabolites and titer determination) in a given experimental and screening design. Seeding density and passage schedule have been optimized to emulate perfusion reactor conditions.

Samples submitted to spent media analysis (e.g., amino acids, Figure 4) demonstrated a similar profile to that of production samples, indicating scalability and the ability to predict performance in production bioreactors. The platform is ideal for suspension-adapted cell lines because the cell tested did not adhere to the tube's surface, and because passaging and harvest is greatly facilitated by the possibility of using the growth vessels (i.e., the 50-mL tube) as the harvest vessel by centrifugation, eliminating volume transfer steps. This significantly reduces labor handling and the risk of contamination, and increases efficiency. Ergonomics and handling also are made easier by the ability to line up tubes in a relatively small footprint inside a biological hood; volume culture transfer is facilitated by the wide opening of the tubes.

Additional benefits of the rolling-tube system include: low material and operation costs (for purchase of the disposable tubes and rolling device), low media volume usage, and the ability to run up to 16 tubes per platform simultaneously; this number can be increased by using additional stackable platforms. The platforms can be placed on a shelf inside a humidified temperature controlled CO2 incubator. The platform withstands continuous operation in the incubator's environment. Given the advantages of the rolling-tube system in performance and handling, it is a preferred platform for a variety of cell culture applications.


The rolling-tube platform presents an ideal balance between predictability of performance at production scale, throughput, robustness, and ease of use, while also providing sufficient sample volumes for product and analytical testing. Its gentle agitation with mixing through rolling also emulates the cumbersome but traditional roller bottle "small-scale" system. We currently use the rolling-tube platform in cell line selection and process development projects such as media screening using design of experiment approaches.

Yuval Shimoni is principal engineer, Carmen Chin is a senior associate process engineer II, Teng Liu is a laboratory technician, Veronica Hernandez-Rodriguez is a laboratory assistant, Peter Kramer is director, and Jin Wang is manager, all in manufacturing sciences, product supply biotech at Bayer HealthCare, Berkeley, CA, 510.705.5775, yuval.shimoni@bayer.comVolker Moehrle is a senior scientist at Bayer Technology Services, Leverkusen, Germany.


1. Eibl R, Kaiser S, Lombriser R, Eibl D. Disposable bioreactors: the current state-of-the-art and recommended applications in biotechnology. Appl Microbiol Biotechnol. 2010;86:41–9.

2. De Jesus MJ, Girard P, Bourgeois M, Baumgartner G, Jacko B, Amstutz H, Wurm FM. TubeSpin satellites: A fast track approach for process development with animal cells using shaking technology. Biochem Engin J. 2004;17:217–23.

3. Jordan M, Jenkins N. Tools for high-throughput medium and process optimization. Methods in Biotechnology: methods and protocols, 2nd Ed. Portner R (Editor) 2007;24:193–202.

4. Yi X, Sun X, Zhang Y. Effects of osmotic pressure on recombinant BHK cell growth and von willebrand factor (vWF) expression. Proc Biochem. 2004;39:1817–23.

5. Amanullah A, Otero JM, Mikola M, Hsu A, Zhang J, Aunins J, Schreyer HB, Hope J, Russo P. Novel micro-bioreactor high throughput technology for cell culture process development: reproducibility and scalability assessment of fed-batch CHO cultures. Biotechnol Bioeng. 2010;106:57–67.