Current, Future Opportunities in Bioprocessing

By Chris Anderson

August 30, 2023 | At CHI’s Bioprocessing Summit earlier this month, three plenary presentations delved into current methods of enhancing bioprocessing, the future of bioprocessing, and how it may change in the coming years.

Antibody Production

Glen Bolton, executive director of Late Stage Bioprocess Development at Amgen, kicked off the program focusing on bioprocessing of antibody therapies.

Bolton detailed work Amgen is doing in the area of developing biosimilars and high-throughput methods to help the company choose the best high-expressing, high-growing clones to choose the best cells for a regulatory filing—all supported by images of the cells with their application. The future will likely see the company expand the number of attributes it uses for cell selection, as well as leverage machine learning methods to further optimize cell selection and cell line development.

A significant challenge Bolton noted downstream relates to the limitations of chromatography.  While the systems he uses to employ robotics for high-throughput processing allow for relative comparison, screening for buffers, and pH, they can only use the columns about 20 times before the seal breaks, can’t get resin for viral clearance, and can only run two cycles per day.

Modeling cycles plays a large role for process development and characterization, Bolted noted, and his team has used modeling to support an FDA filing. “It's great for predicting directionality: what would happen if I had higher flow, a higher bed height, or wider column diameters,” he said. “But the challenge there is we still had to do dozens of experiments to get individual parameters for host cell protein DNA, everything we wanted to bind, everything we wanted to remove, and then tune the models as well.”

To move forward, the company needed to focus on cell line development and bioprocess improvement.

For cell line development, Bolton’s team created a novel, high-expressing cell vector adapted to lean media. This cell line lacks some growth factors and can tolerate high lactate levels. These genetically characterized cells allow gene manipulation for antibody and bispecific antibody production.

For optimizing the bioreactors, the mandate was to focus on one attribute—titer. Through the use of new amino acids, optimizing pH, optimizing temperature, and employing novel perfusion methods, the team significantly improved productivity. They pushed resin cycling beyond 300 cycles and with high titers, which optimized manufacturing facilities for additional downstream trains.

Challenges of New Modalities

Konstantin Konstantinov, CTO of Ring Therapeutics, noted in his plenary address that the increasing complexity of new therapeutics including small molecules, antibodies, and cell therapies and novel modalities of treating disease also require innovation to develop new manufacturing methods that can produce them efficiently and at scale. For the past eight years, he has focused on developing and optimizing new manufacturing modalities, he said, because the old metrics of optimization don’t hold up.

“If you're an antibody field, you can work hard on optimizing a well-established process, and perhaps increase the productivity by 5%. This is going to be incredible, and you get a lot of recognition for this,” he said. “However, in new modalities there is the possibility—and also the need—to increase productivity and process performance by 10x, by 100x. We need to do that, and this is actually happening.”

Konstantinov provided two examples of how to develop continuous manufacturing methods, using innovative approaches or leveraging existing technology to boost productivity for three therapeutic modalities: secreted products, antibody proteins, including exosomes; secreted products that accumulate in cells like AAVs; and cell therapies, where the cells themselves are the therapy.

In the first example, taking a cell engineering approach, he detailed how to leverage a cascade of two bioreactors in a way that decouples cell growth from productivity. In the first bioreactor, the cells are grown, then transferred to a second bioreactor where production may occur by infecting the cells, for instance. Increased efficiency in production can be achieved via the use of a perfusion bioreactor as opposed to using a hemostat to produce biomass at very high cell density and cell viability.

A second approach Konstantinov described was biological: to produce AAVs by engineering exosomes that could encapsulate AAVs within a cell. When the exosomes are secreted, the lipid bubble releases the AAVs.

Konstantinov encouraged tapping other industries for innovation. He pointed to auto manufacturer Tesla’s work with Giga Press, an aluminum die casting technology that is allowing the company to eliminate production of multiple components by die casting them as a single unit. “Now, what is happening is that instead of manufacturing 200 parts to assemble the frame of the car, they manufacture only two parts,” Konstantinov said. “This is very, very efficient.”

Citing Elon Musk’s statement on efficient manufacturing that “the best part is no part,” Konstantinov thinks the same mindset can be applied to bioprocessing and mused: “the best unit operation is no unit operation.”

“This resonates with me a lot,” he added. “Because if we can eliminate unit operations from the process—chromatography or whatever—this is the way to improve, streamline the process, make it more robust, make it more efficient, and cheaper. That's the guiding principle.”

Digitizing Systems

In the final plenary session, Richard D. Braatz, a professor of Chemical Engineering at Massachusetts Institute of Technology (MIT), looked toward future trends of bioprocessing including strategies to improve efficiencies and simplify complex processes such as glycosylation.

As Konstantinov did, Braatz looked outside the industry to a method of manufacturing called Industry 4.0, which is sometimes called “smart factory” or “factory of the future”—a current trend of automation and data exchange leveraging the Internet of Things and cloud computing to create cyber-physical systems. Biopharma 4.0 is Industry 4.0’s analog, with a manufacturing focus on process development workflows with an eye toward automating as many development processes as possible.

Braatz outlined five strategies: enhancing understanding and optimization of unit operations, using process intensification including continuous manufacturing, employing microscale technology for data collection, creating plug-and-play modules for seamless integration, and using mechanistic models for unit operation simulation and control design.

As in other industries, automating as many processes as possible will come to the forefront to drive increases in productivity, Braatz noted. “Humans… carry germs, carry microbes, make mistakes, and they're slow,” he said. “We’re trying to make things very fast.”

While people can never be taken out of the process entirely, leveraging microscale technology for low-cost data collection can provide a wide variety of benefits to digitize the manufacturing platform. It allows physical modules and software to communicate, automatically design control and monitoring systems for the broader integrated system and use these data for mechanistic models and control design. Having a focus on the potential to optimize each unit operation can help drive successful bioprocess intensification.

In one example of successful mechanistic modeling, Braatz recounted work with the yeast pichia pastoris to produce a recombinant protein. Using data collected from published research and comparing it with experimental results, the team was able to develop a continuous process that included building a compact seed perfusion reactor to feed into a production reactor. Detailed calculations and modeling provided information on the optimal rate of seeding the reactor to produce the highest output.

Braatz also detailed ways to use the latest technologies employed in the commercial setting to build heavily instrumented process development systems that collect relevant data for building models of bioprocessing at scale and perform risk-based analysis. This approach allows bioprocess designers to manipulate an array of variables to find the most efficient processing method—all performed digitally.

While this approach can work for production of small molecules and biologics for which there is ample data, Braatz acknowledged that for some emerging modalities—exosomes was one example—it is much more difficult. And in some instances, standard mechanistic modeling won’t always lead directly to the best answer, as real-world variation and other challenges can’t always be fully anticipated.

This highlights the need for real-time monitoring of the performance of any new systems that have been developed digitally, especially for new modalities, to capture real-world data to monitor if the system is performing as predicted or needs to be adjusted accordingly.