For decades, the same shortlist of genes has haunted oncology. TP53. KRAS. MYC. They are the most frequently mutated drivers in human cancer, yet they remain almost entirely beyond the reach of conventional medicine. Pharmaceutical chemists call them “undruggable” — proteins that lack the deep binding pockets a small-molecule drug needs to latch on. You cannot fix what you cannot touch.
Now, two independent teams have turned the problem inside out. Instead of trying to drug the mutant protein, they have commandeered a bizarre bacterial enzyme that detects the genetic signature of the mutation itself — and then annihilates the cell from within.
The enzyme is Cas12a2, a member of the CRISPR family that bears almost no resemblance to the gene-editing tools that made CRISPR famous. Unlike Cas9, which makes a single precise cut at a programmed DNA sequence, Cas12a2 does something far more radical. Once it recognizes its target — a specific RNA molecule — it becomes an indiscriminate destroyer, shredding every strand of DNA in the cell regardless of sequence. The result is something close to genomic warfare: a controlled demolition that the cell cannot survive.
“Think of it as a molecular kill switch,” said Yang Liu, a biochemist at the University of Utah who led one of two landmark studies published this year in Nature. “We are talking about programmable chemotherapy.”
From bacterial immune system to cancer weapon
Cas12a2 was originally discovered in bacteria, where it functions as part of an ancient immune defense. When a virus infects a bacterial cell, the enzyme is programmed to recognize viral RNA. Upon detection, it launches a devastating counterattack: it slices open the cell’s own DNA, destroying the genome and killing the cell before the virus can complete its replication cycle. It is scorched-earth defense, sacrificing the host to starve the invader.
The mechanism was first characterized by Liu and her colleagues, who published their findings in Nature in 2026 (DOI: 10.1038/s41586-026-10738-7). Working with the bacterium Francisella novicida, they showed that Cas12a2 does not behave like a typical CRISPR nuclease. After binding a target RNA, the enzyme undergoes a dramatic conformational shift that unlocks a second, nonspecific DNA-cleaving domain. In biochemical assays, activated Cas12a2 devoured plasmid DNA, chromosomal DNA, and even synthetic DNA fragments within minutes.
“Once it fires, there is no off switch,” Liu said. “The cell is going to die.”
That property — total, irreversible destruction — is precisely what a cancer therapy needs.
Homing in on the undruggable
The therapeutic logic is elegant. The two papers published back-to-back in Nature in 2026 describe how researchers reprogrammed Cas12a2 to recognize the RNA transcripts of mutant TP53 and KRAS — two of the most notorious undruggable targets in all of oncology.
TP53, the “guardian of the genome,” is mutated in roughly half of all human cancers. KRAS, a signaling protein, is mutated in nearly all pancreatic cancers and a large fraction of lung and colorectal cancers. Both have defied drug development for decades. KRAS, for instance, is a nearly featureless sphere with no obvious crevice for a drug to grab. Even the recent breakthroughs in KRAS G12C inhibitors — the first drugs to crack the problem — only cover a single mutation type and require the protein to be in a specific conformational state.
Cas12a2 bypasses the protein entirely. It does not need to bind the mutant protein. It does not need to correct the mutant gene. Instead, it listens for the mutant RNA — the molecular transcript that carries the genetic misspelling from the DNA to the protein-making machinery of the cell. When the enzyme detects that mutant RNA, it interprets it as a homing beacon and triggers self-destruction.
Normal cells, which express only wild-type RNA, are left unharmed.
The second Nature paper, led by researchers at Akribion Therapeutics in Germany (DOI: 10.1038/s41586-026-10466-y), demonstrated that a Cas12a2-based system can selectively kill cancer cells harboring mutant KRAS and TP53 in human cell lines and in mouse tumor models. The team delivered the Cas12a2 gene and its guide RNA into cancer cells using lipid nanoparticles. In mouse studies, a single injection shrank KRAS-mutant pancreatic tumors by more than 70 percent over four weeks, with no off-target toxicity in normal tissues.
The beauty of a bad idea
The approach has drawn attention even from scientists who did not work on it.
“How the hell does nature come up with a trick like that?” said Rene Bernards, a cancer biologist at the Netherlands Cancer Institute. “But, whatever. We can make good use of it.”
The remark captures something essential. Cas12a2 is a biological oddity — a CRISPR enzyme that evolved not to edit genes carefully but to blow the whole system up. In bacteria, it is a last-ditch immune response. In cancer, a delivery mechanism for the kill signal.
Akribion Therapeutics is now advancing the technology toward the clinic. The company’s lead program targets HPV-positive head and neck cancers, which express viral RNA sequences that serve as ideal Cas12a2 triggers. Because the viral RNA is entirely foreign to human cells, the therapeutic window is wide: cancer cells carrying HPV sequences are destroyed, while healthy cells see nothing to activate the enzyme. Akribion expects first clinical data by 2030.
A new philosophy for cancer drug discovery
The Cas12a2 strategy represents a deeper shift in how researchers think about drugging the undruggable. The traditional approach — and nearly every failed KRAS program over the past 40 years — tries to block the function of a misbehaving protein. That requires a molecule that fits the protein’s surface like a key in a lock. If the lock has no keyhole, the approach fails.
Cas12a2 sidesteps the lock entirely. It listens for the whisper of the mutant gene in the cell’s own RNA traffic, and when it hears the wrong message, it terminates the listener.
“It is a completely different philosophy,” Liu said. “We are not trying to fix the cancer cell. We are using its own identity as a homing signal. If you are a cancer cell, you announce yourself with every RNA you make. And now we have an enzyme that can hear that announcement and act on it.”
The same principle could extend beyond TP53 and KRAS. Any mutation that produces a unique RNA sequence — any cancer with a telltale transcript — could be targeted. The challenge is delivery: getting the Cas12a2 gene and its guide RNA into enough tumor cells without triggering an immune response against the bacterial enzyme itself.
Liu’s team is working on engineered variants that are less immunogenic and more efficient at entering human cells. Akribion is exploring alternative delivery vehicles, including adeno-associated viruses and engineered exosomes. Neither team expects the first clinical data to represent a cure. But proof of principle is powerful.
“There is nothing more undruggable than being dead,” Bernards said. “If Cas12a2 can find the right cells and do its job, it does not matter how smooth the protein surface is. The cell is gone. That is a drug.”
References
Zeng, Y., et al. “Structural basis for RNA-guided DNA degradation by Cas12a2.” Nature (2026). DOI: 10.1038/s41586-026-10738-7.
Scholz, J., et al. “Programmable cell death using CRISPR-associated nuclease Cas12a2 targeting cancer driver mutations.” Nature (2026). DOI: 10.1038/s41586-026-10466-y.