Protein Evolution in Mammalian Cells May Improve Therapeutics

To develop better protein-based drugs, scientists can force these molecules to evolve in the lab within a much-shortened time scale. Due to technical limitations, this synthetic biology approach, collectively known as directed evolution, mostly relies on single-celled organisms, such as bacteria or yeast, even when the end-products are intended for humans.¹ However, because the intracellular environments of mammalian cells are dramatically different from those of bacteria and yeast, this compromise often leads to the production of nonfunctional proteins, which defeats the very purpose of directed evolution.

“There are many, many examples, both published and unpublished, where people who do protein engineering got this great thing, but then it didn’t work at all when they brought it to the system they actually want to use it in,” said Matthew Shoulders, a biochemist at the Massachusetts Institute of Technology.

Recently, researchers developed a method, called Protein Evolution Using Selection (PROTEUS), to evolve proteins in mammalian cells.² In a proof-of-concept experiment, Daniel Hesselson and Gregory Neely, synthetic biologists at the University of Sydney, and their team showed that a protein optimized for mammalian cells didn’t have its same improved function in bacteria. This demonstrates the tool’s capability to evolve molecules specifically for use in mammalian systems. Their findings were published in Nature Communications.

“I’m very excited about this method,” said Shoulders, who was not involved in the study. “It holds a lot of promise.”

In directed protein evolution, scientists carry out multiple cycles of gene diversification—typically by introducing mutations, followed by the selection and amplification of genes that encode mutant proteins with optimized function. This iterative, multi-step workflow is tricky to establish in mammalian cells, which are inherently complex, and further optimizations are time-consuming and expensive.

One way to introduce mutations quickly is through viruses. Using PROTEUS, Hesselson and Neely’s team expressed genes of interest in the genome of the Semliki Forest Virus (SFV), an RNA virus that’s distantly related to SARS viruses. SFV naturally makes many errors when replicating its genome and doesn’t have any error-correcting capacity, which makes it a robust way to introduce a lot of mutations into proteins.

Another group of researchers previously developed a mammalian directed evolution method, called Viral Evolution of Genetically Actuating Sequences (VEGAS), using a different virus in the same family as SFV.³ However, Hesselson and Neely’s team found that VEGAS did not work in their hands.⁴

The main breakthrough, Hesselson said, was when his team realized that in VEGAS, the virus’s protective shell, or capsid, was interacting with parts of the viral genome. These interactions contributed to the production of non-functional viral particles, also known as cheater particles. These don’t contain the complete viral genome, thus preventing the virus from propagating properly.

“We then set about trying to fix those issues,” Hesselson said.

To get around VEGAS’s problem with viral propagation, PROTEUS relies on an engineered SFV that can’t make capsids. Instead, the researchers outsourced genome packaging to an unrelated virus that couldn’t interact with SFV’s RNA. By eliminating this genome-capsid interaction, the researchers stopped the production of cheater particles and allowed viral genome replication to proceed efficiently.

“I really liked how they were able to get away from using intact viruses and make this happen,” Shoulders said. “That was a key innovation here, and I think that’s really compelling.”

As a proof-of-concept, Hesselson, Neely, and their team used PROTEUS to improve the sensitivity of rtTA-3G, a protein that can induce gene expression in an antibiotic dose-dependent manner. Researchers commonly use the rtTA system to control the extent of exogenous gene expression, and many of those who work on directed evolution, even in the early 2000s, have tried to optimize the protein.^5,6

In their study, the team used PROTEUS to evolve rtTA-3G in hamster cells such that the protein was about six-fold more sensitive to the antibiotic doxycycline and could induce gene expression in a 3D human cell culture at up to eight-fold lower concentrations compared to its unevolved counterpart. However, when the researchers introduced the evolved protein to bacteria, they did not observe the same increase in sensitivity to doxycycline.

“That shows that our system found mutations that are only beneficial in the mammalian environment and could not have been discovered in a bacterial system,” Hesselson said. “That really validates the reason for doing [directed evolution] work in a mammalian system.”

In the future, Hesselson hopes to use PROTEUS to improve protein-based drugs for various diseases, such as cancer, diabetes, and cardiovascular disorders.

“The reason we don’t widely do [directed evolution] in mammalian cells yet is because it’s still very hard, and this is a step in the right direction,” Shoulders said.