Researchers from the University of Sydney have developed a system that mimics natural selection inside living mammalian cells but in a very controlled setting. They call it PROTEUS, short for PROTein Evolution Using Selection. And it might just change how we design everything from gene therapies to disease-fighting proteins.
“PROTEUS can be used to generate new molecules that are highly tuned to function in our bodies,” said co-senior author Professor Greg Neely. “We can use it to make new medicine that would be otherwise difficult or impossible to make with current technologies.”
For some time, scientists have used directed evolution — an approach that imitates Darwin’s process of mutation and selection in a lab setting — to evolve enzymes, antibodies, and other molecules for certain applications. But nearly all of those experiments have taken place in simple organisms like bacteria or yeast. These systems are fast and easy to manipulate, but they lack the complexity of human cells.
What Neely’s team has done is bring the entire machinery of evolution inside mammalian cells for the first time, successfully and at scale.
Mimicking Natural Selection with Synthetic Viruses
At the heart of PROTEUS is a clever repurposing of viruses. But instead of relying on live, replication-competent viruses that can cause disease, the researchers engineered chimeric virus-like vesicles. These are essentially harmless packages of genetic material.
These vesicles are based on a modified version of Semliki Forest Virus, an alphavirus that can enter mammalian cells. But the team removed the viral capsid, the protein shell that can cause infections, and replaced it with an envelope protein from a completely different virus. The hybrid design allowed PROTEUS to operate safely and robustly.
The result is a stable, self-contained evolutionary circuit. Each cycle begins when a vesicle carrying a target gene enters a host mammalian cell. Inside, error-prone enzymes introduce mutations. If the mutated gene improves the cell’s survival or activity — say, by switching on a key protein — the cell helps that version of the gene spread. It’s evolution, accelerated towards a certain goal. The original development of directed evolution, performed first in bacteria, was recognized by the 2018 Noble Prize in Chemistry.
“The invention of directed evolution changed the trajectory of biochemistry,” said Dr. Cristopher Denes. “Now, with PROTEUS, we can program a mammalian cell with a genetic problem we aren’t sure how to solve.”
Evolving Better Tools for the Human Body
To prove the system worked, the team put PROTEUS through a gauntlet of tests.
First, they took on a classic challenge: evolve resistance to doxycycline, a broad-spectrum antibiotic that normally switches off a synthetic gene in cells. Within four rounds of evolution, the system had selected double-mutant versions of a protein called tTA that kept working even in the presence of the drug.
Next, they turned to a more subtle problem: Could PROTEUS improve an already highly optimized gene switch?
Yes, it could. The researchers took a widely used gene regulator known as rtTA-3G, which activates genes in response to doxycycline. Then they let PROTEUS run for 30 rounds.
By the end, the system had evolved a new version, rtTA-4G, with two subtle mutations: D5N and M59I. These changes boosted the protein’s sensitivity nearly sixfold. In lab-grown human embryoid bodies, the upgraded switch activated genes more efficiently than its predecessor, even at low drug doses.
Critically, these changes only worked in mammalian cells, not in bacteria. That’s something earlier systems never achieved.
All the evolution experiments were conducted in BHK-21 cells, which are derived from baby hamster kidneys. These cells are widely used in virology because they lack a strong antiviral interferon response, making them a safe and permissive environment for replicating RNA-based systems like PROTEUS.
This design choice was deliberate. Building a stable directed evolution system requires high mutation rates and multiple rounds of replication. In human cells, those very features would trigger defensive shutdowns.
Still, while the evolution itself occurred in hamster cells, the molecules it produced were tested and worked in more human-like contexts.
A Biosensor That Sees DNA Damage
One of the most compelling demonstrations came when the team used PROTEUS to evolve an intracellular biosensor. Specifically, this was a nanobody that could detect when DNA had been damaged. This is a very valuable early signal in cancer and aging.
The original nanobody, known as Nb139, had trouble finding its target inside the cell nucleus. But after 35 rounds of evolution, PROTEUS had produced a version with a single mutation (S26P) that gave it a sharp eye for p53, the famous tumor-suppressor protein.
In cells exposed to cisplatin, a chemotherapy drug that damages DNA, the improved biosensor lit up inside the nucleus, forming tiny glowing foci.
“This campaign demonstrates that the intracellular function of Nb139 can be further improved through evolution within a mammalian cell,” the researchers wrote in Nature Communications.
A Biological AI — But for Molecules
What makes PROTEUS remarkable is not just that it works, but how it works.
Much like how generative AI tools explore millions of possibilities to find useful answers, PROTEUS tests millions of never-before-seen mutations, iterating rapidly toward better solutions. Except it does this in real, living cells. Although no machine learning, deep learning, or any other kind of AI was used here, the system can explore millions of protein variants created through random mutations. For this reason, PROTEUS has been likened to a sort of biological AI.
This system allows scientists to find a solution that would normally take a human researcher years to solve, if at all.
In one campaign, they showed that adding molnupiravir, an antiviral drug, to the cells could further increase the mutation rate, expanding the evolutionary search space.
Importantly, PROTEUS is open source.
“We made this system open source for the research community,” Neely added. “We are excited to see what people use it for. Our goals will be to enhance gene-editing technologies, or to fine-tune mRNA medicines for more potent and specific effects.”
What’s Next?
The implications of PROTEUS are sweeping. If the technology can be adapted to work in human cell types beyond hamster cells (BHK-21), it could provide researchers with tissue-specific or even disease-specific evolution environments.
That would be a game-changer for fields like gene therapy, cancer treatment, synthetic biology, and personalized medicine.
As with any powerful technology, there are challenges ahead. The system’s current mutation bias favors certain genetic changes, and more work is needed to evolve biomolecules with completely unbiased diversity. But the team has already begun tackling this with small molecule tweaks.
“By applying PROTEUS,” said Denes, “we hope to empower the development of a new generation of enzymes, molecular tools and therapeutics.”
