Scale-down screening models are important for developing and perfecting cell culture systems under a wide variety of conditions and there are a number of commercial technologies available in the marketplace. GEN interviewed several experts to learn more about these approaches:

Jochen Sieck, PhD, head of perfusion systems R&D, MilliporeSigma
Steve King, president, Artius Bioconsulting
Vince Nguyen, senior upstream process development engineer, Precision Biosciences


GEN: Can you speak to your experiences with scale-down screening model systems? Do you have a preference?

Sieck: The best screening model depends on specific needs. An ideal small-scale model should replicate results at larger scales, but there is often a tradeoff between throughput and predictability, as well as limitations in small-scale modeling of continuous applications.

Jochen Sieck, PhD
Jochen Sieck, PhD

On the smallest scale, 96- or 24-deep-well plate screening is effective for quickly testing a high number of conditions. This high-throughput method can generate a large amount of data, but so far this method is mostly limited to fed-batch systems rather than perfusion modeling. Next up are 50 mL bioreactor tube models, which are effective. They allow for 40–100 conditions depending on the method and can be reasonably predictive in both fed-batch and perfusion.

The most accurate models use miniature bioreactors. These do a good job of replicating results seen in benchtop bioreactors but are limited in the number of conditions tested at once and are more suited for fed-batch modeling.

As a general rule, the scale-down model needs to exactly mimic the critical process parameter to be studied and optimized. Considering this, there are probably still some gaps with regards to scale-down solutions in upstream and downstream process development, particularly for continuous applications.

GEN:  Some studies have compared the process of codification in semiconductor manufacture with biopharmaceutical manufacturing. Both industries have much in common, the principle difference is that the biopharmaceutical industry relies on living organisms as its production factories. Some say because of their complexity and the fact that they cannot be characterized completely means that computer-aided design tools cannot be applied in the manufacturing process for bioprocess development, but rather must be developed empirically for each biomanufacturing situation.

Do you feel this is an accurate description of the challenges facing companies as they seek to improve upstream and downstream performance, and if it is, what strategies to propose to address these issues rapidly and effectively?

Sieck: The biological nature of cellular production does increase variability and can create complications beyond what is seen in nonbiological manufacturing. While there remains a need for significant empirical development, computer design tools and prediction software are already useful in bioprocess development, and the power of these tools will only grow as our understanding increases.

Process development is no longer purely empirical. Today, new research, methods, and technologies enable the use of computer-aided design tools. As our knowledge increases, and with the availability of more high-throughput process development tools, we firmly believe that digitization will play a role in enhancing the biopharmaceutical manufacturing process.

GEN:  As pointed out in numerous publications, the demand for therapeutic proteins continues to grow, in large part due to the high dosing regimens (often more than 1 g per dose) and the approval of more biologics in therapeutic regimes.

One critical issue is the maintenance of appropriate glycosylation profiles in scaleup for protein production. A number of adjustments to culture conditions have been proposed to deal with this challenge. Have you encountered this issue and if so, how have you confronted it?

King: Maintaining close control of protein glycoform characteristics is becoming more important as the industry moves further toward glycoengineered antibodies and other proteins. This requires a thorough understanding of process parameters and how they relate to critical quality attributes (CQA) so that the drug developer can establish a process control system that will result in the desired product quality consistently. This is clearly one of the reasons that the ambr system and other similar systems have become so popular based on their ability to effectively perform DOE studies to understand the process and its impact on product characteristics.

Steve King
Steve King

Alternatively, there are other emerging genetically engineered expression systems that can achieve similar goals, including yeast and plants. While it may seem counterintuitive to turn to yeast and plants as a way to make fully human proteins with desired human glycosylation characteristics, the technologies are quickly emerging and may present an attractive alternative to traditional systems. This is an exciting time of opportunity for companies that embrace not just novel targets but combining the target with the desired CQA.

Nguyen: In the past, I’ve found that the combination of spent media analysis along with process characterization can make it possible to identify process parameters that will increase and decrease the afucoslyation of antibody populations by several hundred percent with a corresponding correlation in potency.­­ Understanding the product and process intersection can be key to achieving the desired results.

I would also like to add that the views I am expressing are my personal opinions and not necessarily those of Precision Biosciences.

Sieck: Product quality is always one of the top priorities in biologic production and normally requires monitoring throughout any development project. Process development, aided by predictive small-scale models, is critical for matching the desired profile. The presence and quantity of individual components in the cell culture medium can also play a role in these protein quality attributes, therefore through holistic media development, we can adjust glycosylation profiles to fit the desired specifications.

In addition to media optimization, MilliporeSigma has developed a supplement to precisely tune galactosylation: EX-CELL Glycosylation Adjust (Gal+) allows users to easily achieve desired N-linked glycosylation by increasing the galactose site occupancy on the oligosaccharide to a higher level.

GEN: Increasing interest is focused towards the use of miniature bioreactors during late clinical phase process characterization studies as a scale-down model of the cGMP manufacturing process. Recent efforts have shown promise in qualifying ambr systems as scale-down models to support more efficient, robust, and safe biomanufacturing processes. Do you have experiences in this area that you can detail?

Vince Nguyen
Vince Nguyen

Nguyen: Once the appropriate scale down studies are embarked upon, there are a large number of parameters that you can test in miniature bioreactors such as feeds, time of feed, temperature, pH, and many more.  My biggest concern would be agitation, which I think can best be characterized in bench-scale reactors (2–5 L). In adeno-associated virus production one additional complication is that the mammalian cells must be transfected in order to produce the virus. This means that scale down to an ambr system scale can be difficult, whereas we have found a robust correlation with transfections from 2 L systems.

This site uses Akismet to reduce spam. Learn how your comment data is processed.