The Use of the ATF System to Culture Chinese Hamster Ovary Cells in a Concentrated Fed-Batch System

Published on: 
, , , ,
BioPharm International, BioPharm International-06-01-2011, Volume 24, Issue 6

Cell concentrations and resulting protein concentrations are higher in a concentrated fed-batch process than in a standard fed-batch culture system.

The Chinese hamster ovary cell line (CHO cell) is considered by many as the workhorse of the biotechnology industry. It is a preferred cell for production of an ever-expanding array of engineered products that have therapeutic, diagnostic or other applications. The cell line not only provides a well characterized laboratory for performing genetic manipulations, but also offers an excellent expression host for production of engineered products. The glycosylation pattern of this mammalian cell is an advantage not offered by potentially more productive microbial expression systems. Inherent in its mammalian nature, however, is its relatively slow growth rate and the constraints on its cultivation.

Although great improvements have been made in the culture of most animal cells, resulting in increases in their productivity, their use in large scale production has been limited in the past to batch or fed-batch processes. In both processes, cells are inoculated into a fixed culture volume and the more product needed, the larger the culture system. Cell growth is rapid in the fresh medium. As the nutrients are consumed, and waste products accumulate, growth rate declines, resulting in a culture decay phase. In a fed-batch system, the viability and productivity of the culture can be extended somewhat by supplementing it with growth promoting additives. In either case, however, declining growth and decay is a consequence of consumption of essential nutrients and accumulation of growth limiting cellular byproducts. The final productivity of such cultures may typically vary from hundreds of mg/L to a few mg/L.

Alternatively, it is possible to extend and maintain the viability of a culture and its productive phase by continuously removing the inhibitory waste byproducts while replenishing the culture with fresh nutrient medium, in a process known as perfusion. Normally, this process is done using a filter or other cell retention device, where the protein product is collected with the waste stream while the cells remain in the reactor. Continuous removal of product through the waste stream prevents its accumulation in the vessel. Rapid removal and storage offers an important advantage with some labile products, but generally, the large volume collected is not desirable. The ATF System used in a concentrated fed-batch (CFB) process takes advantage of the high cell concentrations and resultant high levels of product formation offered by perfusion and at the same time eliminates dilution of the product. The ATF System CFB process is realized simply by using a standard molecular weight cut-off filter, typically, 50 kDa. The filter allows waste products or any compound smaller than the indicated pore size to flow through, but not larger molecules. Antibodies and many other potential proteins or products larger than 50 kDa are retained by the filter. Both cells and product can thus be concentrated to high specific concentrations. A comparison between a standard fed-batch and CFB was undertaken by Biovian to evaluate the differences between these two processes.



ATF System: Refine Technology Alternating Tangential Flow (ATF2) System using the F2 Hollow Fiber Module 50 KDa pore, PS, 1 mm ID, 0.14 m2

Wave reactor: Wave Cellbase 20 PS with control unit Wave Biotech BWC

Wave bag: Sartorius-Stedim Cultibag RM 2 L Optical ATF and Opta SFT sterile connector

Medium: HyClone SFM4CHO

CHO cells: CHO cell line expressing human immunoglobulin G (hIgG)

Analysis of product: hIgG immunoassay.


The CHO cell seed culture was prepared in spinner flask. The cells were transferred to the wave reactor, at a starting concentration of approximately 0.4x106 cells/mL (Day 1). The bag culture-volume was approximately 1 L. The bag was equipped with optical sensors for dissolved oxygen (DO) and pH on-line measurements. Aeration was not initiated until Day 2 in order to maintain pH. Afterwards, aeration was maintained at 100 mL/min. The oxygen concentration was maintained as instructed by the manufacturer by increasing the rock and angle parameters.


In the fed-batch reference run, nutrients, such as CD-feed and glucose, were fed batch-wise at four different points in time during the process. The pH was automatically controlled.

In the ATF CFB experiments, the ATF System was connected to the culture bag. The flow was started on Day 2 at a constant flow rate of 0.5 L/min. The perfusion (fresh medium in and waste medium out) was started on Day 3. In Run 1, the perfusion rate was initiated at 0.8 L/d and increased to 1.2 L/d. In Run 2, it was initiated at 0.8 L/d and increased to 2.5 L/d by Day 7. The medium addition and waste outflow rates were determined by weight change of the respective containers.

Samples of approximately 2 mL were removed at least once a day from the cultivation bag with a syringe. DO and pH were registered during the sampling. An aliquot of the sample was immediately frozen at –20 °C for product analysis performed after the end of the actual experiment. Glucose and lactate concentrations were determined subsequent to sampling. The cell concentration and cell viability were analyzed by diluting the sample in phosphate buffered saline and tryphan blue before automatic counting of living and dead cells by a cell counter. The product was analyzed by a hIgG immunoassay.



The results of the fed-batch culture shown in Figures 1–6 are typical of such cultures. There was rapid loss of viability following a growth phase (see Figures 1, 2), continued increase in product concentration after the culture began to decline (see figure 3) and accumulation of waste products (lactate, see Figure 4), accompanied by a steady decline in glucose concentrations and DO (see Figures 5, 6). Supplementing the fed batch culture four times with nutrient media during the course of the culture resulted in a peak cell concentration of 5.8x106 cells/mL at 191 h (see Figure 1). As expected, the highest product concentration was somewhat delayed, reaching 0.4 g/L, at 260 h (see Figure 3).

Figure 1: Number of viable cells in fed-batch and concentrated fed-batch runs. (ALL FIGURES ARE COURTESY OF THE AUTHORS)

First CFB run

The benefits of addition of fresh medium while removing waste media is also illustrated in Figures 1–6. pH was set by the process without the need for a separate pH control (see Figure 7). The perfusion rate in this run varied between 0.8 and 1.5 vv/d (see Figure 8). The first run was less than ideal. The weight measurement for the Wave bag and associated control loop to keep the level constant in the bag had a problem resulting in a working volume that was too high. A manual adjustment was made by increasing the filtrate flow rapidly to remove this excess media. However, this adjustment was too fast for the ATF System's cleaning action and the essentially dead-end filtration performed led to filter failure. Viability started to decline early in the run, at <150 h, and the viable cell count also leveled off at about the same time (see Figures 1,2). The perfusion was ended at 239 h and in spite of the dificulties, a peak cell concentration of 16.4x106 cells/mL, was reached at 195 h (see Figure 1), whereas the highest product concentration, 1.9 g/L, was reached at 239 h (see Figure 3).

Figure 2: Percent of viable cells in fed-batch and concentrated fed-batch runs.

Second CFB run

The second CFB run with the ATF System differed somewhat from the first CFB run. It was performed with a more rapid increase in perfusion rate to about twice the final rate of the first run. The perfusion rate in this run varied between 0.8 and 2.5 vv/d (see Figure 8). The cell growth was sustained longer than in the first run, achieving a significantly higher cell concentration. The highest cell concentration, 34.5x106 cells/mL, was reached at 284 h (see Figure 1), whereas the highest product concentration, 3.8 g/L, was reached at 332 h (see Figure 3). The viability of the culture was sustained above 85% for the duration of the culture (see Figure 2). After an initial adjustment, glucose and lactate were sustained at about constant levels for most of the culture duration (see Figures 4,5).

Figure 3: Product concentration in fed-batch and concentrated fed-batch runs.


The results in Table I show that cell, and more importantly, product concentration can be greatly enhanced with the ATF System CFB process. Even in the first CFB run which was a less than optimal culture, nearly three–fold higher cell concentration was achieved and more than a 4–fold increase in product concentration was obtained than in fed-batch, in about the same time period. The results improved further in the second CFB run, increasing product concentration to nearly ten–fold. Media feed rate appears to have had a great affect on cell growth rate and product levels. Further enhancement would be possible with enrichment of oxygen transfer and media development.

Figure 4: Lactate concentration in fed-batch and concentrated fed-batch runs.

Despite the increases in product concentration in CFB over fed batch, the ultimate productivity and the economical aspects of both process modes need to be assessed and understood. Some considerations are reactor volume capacity, and the amount of product needed, as well as the time taken to produce the amount of product.

Figure 5: Glucose concentration in fed-batch and concentrated fed-batch runs.

Using the data obtained in these experiments, fed-batch is further compared to CFB. Three production scenarios, (Case A, B and C in Table II), are offered for producing 3.8 g of hIgG, which was achieved in the second ATF run, and 100 g of hIgG.

Figure 6: Dissolved oxygen (DO) in fed-batch and concentrated fed-batch runs.

Table II can be used to evaluate requirements for production of a specified amount of product and to further analyze whether those requirements correspond to a user's needs or capabilities. For example, if the user has 1-L reactors, but wishes to produce 4 g of product, then it will require over 3 months and 9 batches to do so, using the fed-batch mode (Case B), but will only need 1 batch and two weeks in concentrated fed-batch mode (Case C). Given the risks of running 9 batches versus 1, and the 3 month extra overhead cost, there is a significant advantage in adopting concentrated fed-batch, at the cost of spending as little as 2.5 times more on media.

Figure 7: pH in fed-batch and concentrated fed-batch runs.

The relationship above is obviously relevant for any size reactor: If the user has a 100-L reactor, perhaps a single use bioreactor, and needs 400 g of product, it will require about 10 batches and four months with a fed batch process (Case B). The user may consider buying a new 1000-L reactor, which would do the job in one batch, but will also require associated accessories, downstream equipment and facility space. Alternately, the user may use the ATF System, which will deliver the 400 g required in one batch in the existing 100-L (Case C).

Figure 8: Flow rate in concentrated fed-batch runs.

If the user has three reactors, 25-L, 100-L, and a 250-L to choose from, and 100 g of product is needed, then the 250-L fed-batch option would seem to come out slightly better than the 100 L fed-batch, with a longer production cycle, or the 25-L CFB system with a similar production timeframe but costlier media use. However, when considering seed expansion for this 250-L reactor, which is not factored into the examples in Table II, it appears that the 25-L and 100-L reactors are needed anyway; therefore, when assessing risk versus the gain in using the 25-, 100- or 250-L system, it appears to be similar in each case. Other factors such as downstream processing of 25-L versus 250-L harvest need also to be taken into account.

Table I: Summary of results.

In yet another scenario: What if the user has several 25-L reactors and only one 250-L reactor but wishes to produce multiple products in a limited time? Clearly the concentrated fed-batch approach can most readily achieve that goal. The multiple 25-L reactors can be used simultaneously, rather than waiting for the main reactor to become free. The 25-L reactors will also not be tied up as seeds for the 250-L reactor.

Table II: Production of human Immunoglobulin G (hIgG) using fed-batch and concentrated fed-batch.


The use of the ATF System to achieve exceptionally high productivity when used in a concentrated fed-batch process lends itself to more efficient transition to upstream operation. Additionally, in the case of a midsize or smaller facility, the requirement of expanding capacity through reactor scale-up is removed. The ATF System allows existing equipment to be more flexibly adjusted to increased production requirements.

Because the ATF System can also be used to produce high density, large volume cell banks in disposable bags, which are used to inoculate small or even pilot sized reactors directly from the freezer, the facility can be further optimized. Similarly, for seed expansion, the ATF system can reduce the reactor train lengths, freeing up existing reactors for other production requirements.

For a facility using or planning to use disposable reactors, the ability to reassign capacity to new projects could be very valuable in reducing costs and risks in biomanufacturing.

John Bonham-Carter* is vice-president, Jerry Shevitz is president, and Edi Eliezer is vice-president, all at Refine Technology LLC, Pine Brook, NJ, Jan Weegar and Antti Nieminen are both senior scientists at Biovian, Turku, Finland.