Improving the expression of therapeutic proteins in mammalian or microbiologic cultures has become a leading bioprocess troubleshooting objective. Given the goal of some predefined combination of yield and quality, debottlenecking may be directed narrowly at the culture stage through, for example, feed or medium optimization strategies, or earlier and more broadly, during gene-focused discovery.
Where the money is...
When asked why he robbed banks, highwayman Willie Sutton is alleged to have replied, “because that’s where the money is.” One could say the same for bioprocess unit operations involving cell culture—the most visible and obvious area of yield improvement—which is where protein expression occurs.
Search Protein expression Search Now Search our directory to find protein expression related products and services.
Here, optimization hinges on overcoming two primary challenges, according to Atul Deshpande, Senior Product Manager at Promega: ensuring the accurate incorporation of post-translational modifications (PTMs), and preventing protein misfolding and aggregation.
“Mammalian and yeast expression systems provide more human-like PTMs, particularly glycosylation, while glycoengineered microbial hosts are emerging as cost-effective alternatives,” Deshpande says.
“Strategies to prevent aggregation and misfolding include co-expressing chaperons, optimizing cell culture conditions through temperature shifts or adjusting inducer levels, and using solubility-enhancing fusion tags. Computational tools and kinetic modeling further aid in predicting and mitigating aggregation risks.”
Artificial intelligence (AI) and machine learning (ML) are emerging as powerful aids in engineering protein sequences for improved solubility, stability, and expression efficiency, Deshpande tells Biocompare. “By leveraging large datasets, ML models can also identify mutations that enhance folding or minimize aggregation, reducing the need for extensive trial-and-error optimizations.”
He believes future progress on AI- or ML-enhanced protein expression strategies will depend on integrating them with conventional approaches like real-time monitoring and adaptive bioprocessing.
Back to basics
At its most basic level then, improving protein expression becomes a question of nurture vs. nature, with the former approach focusing on culture conditions and the latter on the protein’s molecular pedigree—in other words, the gene. This is possible because, due to the direct correspondence between gene and protein building blocks, they and their expression is almost universally treated as synonymous or equivalent.
Compatibility between the expressed protein and the host also come into play.
“Scientists control gene/protein expression for different reasons,” says Jean Peccoud, Ph.D., Founder of GenoFAB, a genetic tools and methods company. “For example, in metabolic engineering, where enzymes are expressed to control a metabolic pathway to produce a chemical, the protein is not the final product.”
Yet processes for many recombinant biologic drugs reveal that biomanufacturing yield does not necessarily correlate with protein expression. “For example, very highly expressed proteins negatively affect the fitness and even the viability of host cells,” Peccoud adds. “Because such cells are difficult to grow at high density in a bioreactor, the process yield will be low even though expression is high.”
But can we produce it?
Manufacturability may also be addressed at the gene expression stage, Peccoud says.
A common complaint about small molecule drugs was that discovery groups did not fully appreciate the complexities of manufacturing at scale, or the potential disconnect between a molecule’s “ideal” chemical environment and its eventual formulation. Similar issues affect drug discovery involving genes or proteins, according to Peccoud.
“Most scientists involved in discovery have little understanding of biomanufacturing. They purify small quantities of protein for preclinical experiments using fast and easy processes, but clinical trials require proteins manufactured in a GMP environment using a scalable process that meets FDA quality standards.”
Given the complexity of recombinant proteins used in clinical application, the production processes define the product (or as the U.S. Food and Drug Administration would put it, “the product is the process”). Peccoud, therefore, advises customers to design a preclinical strategy around proteins produced through a process as similar as possible to the process used for cGMP manufacturing.
“Because the manufacturability of a drug candidate is almost as important as its biological properties, we encourage clients to perform preclinical work with the end game in mind,” he says. “The sooner they optimize gene expression for manufacturing, the faster and cheaper the transition to the clinic will be.”
Peccoud argues for what amounts to a lifecycle approach to drug discovery, encompassing not just instructions for generating an amino acid sequence from a collection of nucleic acid sequence, but the design of other critical characteristics at the genetic “concept” stage.
“Protein expression is controlled at transcription and translation, with multiple ways to control these two stages,” he explains. “Transcription can be controlled by plasmid copy number, promoter strength, enhancers, or genomic location. Translation can be controlled by codon usage, translation initiation sites, and the architecture of the 5' and 3' UTR untranslated regions. Optimizing gene expression often requires finding an optimal combination of parameters, not simply maximizing all the parameters. It's more about balancing the different elements.”
At a broader level, the selection of expression hosts and growth conditions may also influence the selection of genetic elements controlling gene expression. For biologic drugs, choice of hosts is governed by several factors, according to Peccoud:
- Cost and speed of the fermentation process. Microbial fermentations are faster and cheaper than mammalian cultures. Between these two extremes exists a broad range of possible hosts, including yeasts, insect cells, plants, and milk-producing animals.
- Post-translational modifications of the proteins can significantly affect their function or immunological properties. PTMs are completely controlled by the host cells, so the desired PTMs determine the host, and thereby the expression system.
- Finally, the host organism will affect purification. E. coli are easier to grow but often generate unwanted protein aggregates in inclusion bodies. This requires a refolding step that may not be required using mammalian cells. Alternately, for example, purifying recombinant proteins from milk is easier than from cell cultures.
- Regulatory implications: Using non-standard hosts would lead to regulatory uncertainties. So, people tend to stick to proven solutions even though other approaches may be cheaper or faster.
Although the profitability of bioprocesses is most strongly associated with yield or volumetric productivity, there is more to the story than simply focusing on attamoles of product per cell. Paying gene expression its due can improve quantities of harvested protein to be sure, but can also help streamline formulation and purification strategies.