Making proteins for research or therapies often starts with cell-line development. In many applications, genetic engineering is used to optimize a cell line for producing high levels of the intended protein. Researchers and companies often start with Chinese hamster ovary (CHO) cells or human embryonic kidney (HEK) cells, but many other options can be considered. In addition, many emerging tools, especially AI-based ones, can improve the process. To find out more about ongoing advances in developing cell lines for protein expression, GEN talked with three experts from China, Denmark, and the U.S.
Consider the entire process
“With more than 30 years of experience, we always start with the protein,” says Carter Mitchell, PhD, chief science officer at Kemp Proteins—a Frederick, MD-based company that describes itself as a premier provider of customized solutions to protein-related challenges. “The protein is the most important component,” says Mitchell.
Following protein selection, Mitchell and his colleagues apply machine learning (ML) and AI-based workflows. From this, he says, “We can identify liabilities that come from the primary sequence alone, and you can analyze that information to recognize where that protein would be produced in the recombinant host.”
To create an efficient method of collecting this information, scientists at Kemp Proteins developed an automated workflow that starts with the target sequence. Then, that automated workflow searches the literature and patents for information related to that sequence. “We rip out all that information, and then we do our machine-learning process to figure out what post-translational modifications might be present, if secretory tags are present, and where the protein will mature in the host,” Mitchell says. “That gives us a native understanding of the protein.”
The Kemp team then analyzes how the sequence would work in a recombinant system. “We pride ourselves on being agnostic to the expression system,” Mitchell says. “So, we choose the host as a function of the product profile.”
To create the best program for manufacturing a protein, scientists think through the entire process from the very beginning. This means determining the desired product profile and its critical quality attributes.
Scientists at Kemp Proteins also consider how the protein will be used and the intended scale of production. Will it be a therapy, a vaccine, a diagnostic, or an R&D tool? Plus, Mitchell and his colleagues make sure that a program can cover any costs for license agreements that will be required and that milestone payments can be met. As Mitchell says, “It’s really leveraging a whole lot of components to generate stable cell lines with the appropriate profiles.”
In the past, these steps required a lot of manual effort from Mitchell. “Then, I developed a program that does the work for me,” he says. “Now, I just push a button and then drink my coffee and come back to a 100- to 400-page report, and then we come up with some strategies that are based off of logic that’s been applied in the past for us to have success.”
Kemp’s program, though, arose from decades of experience. “If you don’t know the right questions, you’re setting yourself up for failure,” Mitchell says.
Mitchell’s experience in natural products also influences how he goes about expressing proteins, especially for therapeutic use. “I did some structural biology in my graduate career for understanding how proteins can synthesize natural products,” he says. “Working with natural products allows you to understand different options.”
For example, Mitchell explains that his natural product experience helped him understand the importance of post-translational modifications and the maturation process that takes place in the native host. “If you have to have the post-translational modifications, then we need to select a system that’s capable of doing it,” he says. “If there are no recombinant hosts that are capable of doing a sophisticated maturation of the protein, then let’s go to the native source.”
Plus, using a natural host can avoid an investment of tens of thousands of dollars in synthesizing DNA. In such a situation, Mitchell says, the host has “evolved to have this protein, and it couldn’t be more correct than from the native host, right?” Still, he adds that even the geographical location of a host and its stage in life can impact its production of a specific protein.
No matter how a protein is produced, from a native or recombinant host, the product needs to be just right. Even small differences can create big consequences. For example, “In some glycosylations, the carbohydrates in the glycans are very important for the actual function of that molecule,” Mitchell says.
Focusing on fragments
Although complete antibodies are well known for use as therapeutics, they can also be produced from antibody fragments, including the fragment antigen-binding (Fab) region, the single-chain fragment variable (ScFv) region, and heavy-chain variable domain (VHH) fragments. Nonetheless, “Antibody fragments, depending on the specific design, can encounter several CMC (chemistry, manufacturing, and controls) challenges in productivity and quality, including–but not limited to–low productivity, aggregation and truncation, poor purification yield, and poor protein stability,” says Xiaoyue Chen, PhD, head of the cell line development department at WuXi Biologics in Shanghai, China. “Poor developability is usually the root cause for these technical challenges, which may result in future challenges in high cost of goods during manufacturing.”
Addressing these challenges depends on a strong expression system. “WuXi Biologics uses the cell line development platform WuXiaTM and clone-screening approaches to develop antibody fragments, which can best utilize our strengths in in silico sequence analysis and optimization tools, host cell families, and cell culture processes, resulting in an average of 5 grams per liter of antibody-fragment titer on the basis of more than 50 related projects,” Chen says. “Great quality consistency between stable pool and clones can enable an early prediction of quality liabilities of a product and timely intervention from upstream/downstream processes and analytical/formulation support.”
For upstream development, WuXi Biologics relies on methods that boost the titer and inhibit aggregation and truncation. To intensify the process, WuXi Biologics uses its WuXiUITM platform in fed-batch processes or its WuXiUPTM platform for perfusion-based processes. The company also uses simulations of both process types.
For downstream development, WuXi Biologics employs methods that optimize the removal of host-cell proteins (HCPs) and high-throughput pharmacodynamic technologies. The company’s downstream toolbox also includes “effective analytical methods for HCP and protease detection, real-time PAT aided auto-process development and real-time product quality monitoring, and experienced formulation development,” Chen says. This combination of technologies can produce 150 milligrams of product per milliliter.
In 2024, WuXi Biologics launched its WuXia293, which is an HEK-based mammalian production system. This system is “suitable for antibody fragments prone to truncation” and it “can reach comparable productivity to CHO while significantly reducing truncation for certain types of recombinant proteins that are sensitive to CHO HCPs,” Chen says.
Fruitful flies
Instead of using CHO or HEK cell lines to produce proteins, insect cells can also be used. For example, Drosophila melanogaster S2 cells have been used in academic research for decades. These cells, though, pose some challenges, including cell viability and the production of heterogenous proteins.
To improve the commercial use of S2 cells, Max Søgaard, PhD, senior vice president of R&D and technology at ExpreS2ion Biotechnologies in Hørsholm, Denmark says “We utilize stably transfected cells for protein expression, offering significant cost advantages at scale compared to baculovirus-based systems.”
Søgaard adds that his company’s S2 cells are highly viable and produce “homogeneous, well-folded proteins with minimal batch-to-batch variation.” Moreover, these S2 cells grow to very high densities—usually about 30 million cells per milliliter and with some reports of up to 350 million cells per milliliter.” Consequently, Søgaard points out that these cells are “well-suited for industrial fermentation methods, including batch, fed-batch, and perfusion processes.” In addition, he notes that these cells are “particularly resilient, naturally resistant to aggregation, do not require CO2 control, and can grow at room temperature.”
Using genome editing, ExpreS2ion Biotechnologies developed S2 cell lines with a variety of glycosylations that “enhance vaccine immunogenicity or conversely offer more human-like glycosylation for therapeutic applications,” Søgaard says.
Working in collaboration with Adaptvac, a biotechnology company owned by NextGen Vaccines and ExpreS2ion Biotechnologies, Søgaard and his colleagues used S2 cells to express a COVID-19 vaccine (ABNCoV2) that is based on virus-like particles. In a Phase III clinical trial, this vaccine met its primary endpoint of being non-inferior to mRNA vaccines, but it did not meet the secondary endpoint of protection against the newest variant of concern.
Nonetheless, “Its longevity of response far exceeded that of mRNA vaccines, with immunity lasting over a year,” Søgaard says. “To our knowledge, this is the only COVID-19 vaccine to offer such prolonged immunity without the need for an adjuvant.”
ExpreS2ion Biotechnologies is now using similar technology to develop a vaccine for HER2-expressing cancers, currently starting a first-in-human Phase I safety trial in metastatic breast cancer patients in Europe.
From fruit flies to humans, companies can explore a range of cell lines to express research-only or biotherapeutic proteins. To make the most of such systems, it’s worth repeating the cell line development advice from Carter of Kemp Proteins: “Always start with the protein.”
