“Who would want to build the future of biomanufacturing on tech build during WW2?” Shannon Hall is the CEO and co-founder of Pow.Bio, a Berkeley-based startup with a mission to revolutionize the biomanufacturing industry and the fermentation process altogether. And she thinks that the industry needs a complete change of mindset to achieve the cost and production efficiencies to make a true bioeconomy possible.

The topic of using new technologies for bioproduction will be discussed extensively during the SynbioBeta 2024 Conference, where Shannon Hall will be the chair of the Biomanufacturing Scale-Up track.

‍

From Cattle Stomachs to Steel Tanks… and to the Cloud?

Microbial fermentation is an ancient process closely tied to the development of civilization. About 12,000 years ago, settlements in northern Africa used bacteria to ferment milk into a yogurt-like product. 6-7,000 years ago, the Chinese and the Mesopotamians discovered how to brew alcoholic beverages. 2,500 years ago, people in China discovered the antimicrobial properties of moldy soybean curd and used the secondary metabolites of the mold to treat wounds, albeit without understanding the rationale behind it.

A true revolution of biomanufacturing began when Louis Pasteur established the microbial nature of fermentation and a process to preserve food using temperature treatment (pasteurization). The 20th century saw an explosion of fermentation applications, where different microbes produced amino acids, organic acids, medicines, food products… and the list goes on. This gave birth to the steel bioreactor, a significant technological advancement allowing the growth of microbes in a controlled environment in very high volumes.

‍

The rapid development of synthetic biology expanded the array of products microbes can produce and also gave rise to a new challenge: scaling up needs to become faster, cheaper, more efficient, and accommodate more varied bioprocesses. Typically, a microbial strain generated using synthetic biology is selected and screened for function in the lab at small volumes. The strain then grows into small bioreactors (0.5-10L) before testing on a pilot scale (~100-200L). The testing will let the fermentation engineers determine the maximum productivity and optimize the cultivation conditions before committing to augmenting the volumes to production sizes (typically thousands or even millions of liters).

This process is costly and carries a lot of uncertainties. The capacity of pilot bioreactors and the economics of testing and scaling are probably going to become the new “valley of death” for innovative synbio startups. There might be a solution, though: making part of the process digital.

‍

Digital Twins for Bioreactors

“The transition to fully digitalized biomanufacturing processes is not just an aspiration but a necessity for the future of sustainable and efficient bioproduction,” Esteban Marcellin, Professor at the University of Queensland, Australia, said. His lab is modeling the bioproduction process both at the microbial strain level and as process optimization. He is particularly interested in gas fermentation, a technology successfully deployed by companies such as Lanzatech. He explained that he turned to gasses as growth feedstock to reduce production costs. Marcellin said that when using expensive sugars for growth, “approximately 40% of the carbon is lost as CO2, making the value proposition difficult to justify”.

Using bioreactor datasets, researchers worldwide create digital twins to understand the fermentation process better and generate in silico models of bioproductions. Digital twins are digital representations of a system or a process and are in heavy use in the majority of engineering disciplines. A 2023 McKinsey insight mentions that 70% of C-suite technology executives at large enterprises are exploring digital twins, while the investments might reach $48B by 2026. Meanwhile, technology has revolutionized other manufacturing disciplines. For example, automobile engineers run digital simulations of how a car behaves when driven, identifying additions they can do to increase safety before the vehicle leaves the design board. There is no reason we shouldn’t do that with biology.

‍

“Model-driven approach provides a faster and more cost-efficient means to evaluate many different conditions in silico, reducing the need for expensive and time-consuming experiments,” explained Chueh Loo Poh, Associate Professor at the National University of Singapore. His lab is developing models and tools to integrate insights from lab experiments to large-scale bioproduction. The task is not trivial, as cells behave very differently in lab scale and in high production volumes. “We are utilizing the models to gain insights about the bottlenecks of the process and to identify the desired operating conditions,” Poh explained.

‍

What to Measure and How to Put It In Use

Hall, Marcellin, and Poh all acknowledge the importance of large production datasets that will train and optimize the digital models. The new AI tools certainly help accelerate the process. I asked them what parameters they usually measure in their bioreactor experiments, and they mentioned pH, temperature, feeding rates, biomass accumulation, and productivity. A challenge shared between research groups is the lack of standardization in data collection and protocols. This means that the researchers find it difficult to combine information from different sources or that they have to convert data between formats.

However, collecting data is of little use if the algorithms cannot generate useful outcomes or new ways to approach bioproduction. Marcellin is deploying his modeling approach to enhance precision fermentation to create novel proteins, fats, and flavors for the food industry. “At the University of Queensland's Food and Beverage Accelerator, we focus on the development of innovative ingredients that promise not only to enhance food quality and sustainability but also to redefine consumer experiences,” he said. “By collaborating with researchers, startups, and industry leaders, Faba.au acts as a catalyst for bringing these groundbreaking ingredients from the lab bench to the marketplace.”

Hall’s Pow.Bio envisions a paradigm shift in bioproduction by optimizing continuous fermentation. Traditionally, bioproduction uses batch fermentation, where a bioreactor filled with media is inoculated, cells grow, the product is produced, and the cells are harvested before the next production batch starts. In continuous fermentation, cells and media are fed constantly into the bioreactor, and production also happens in an unremitting manner. Hall mentioned that their system has achieved productivity using 1/10 of the volume that traditional fermentation approaches would require. Their approach integrates production with data collection and real-time adjustments, creating a self-optimizing fermentor with enhanced performance. 

The ultimate goal of fully digital bioprocess, able to predict strain behavior and production outcomes with minimal experimental data input, may be difficult, but not impossible. The new computational tools, innovative processes, and advanced sensors allow innovative synthetic biology ideas to test themselves faster and, hopefully, successfully scale up!

‍