CRISPR’s Newest Weapon Doesn’t Edit Genes. It Shreds the Genome of Every Cancer Cell It Finds.

For nearly half a century, the protein p53 has been the white whale of cancer biology.

Called the “guardian of the genome,” p53 normally patrols cells for DNA damage and triggers self-destruction when it finds too much. Mutations that disable p53 appear in roughly half of all human cancers. In the deadliest forms (ovarian: 96%, pancreatic: 72%, high-grade serous carcinoma: 96%), the fraction is far higher. No approved drug directly targets mutant p53. The protein’s flat, featureless surface offers no groove or pocket for a small molecule to grab. Phase III trials of the most advanced candidate, eprenetapopt (APR-246), failed to improve survival in myelodysplastic syndromes. p53 earned the label “undruggable” not as hyperbole but as a description of structural reality.

Now two groups, working with overlapping teams at UC Berkeley, the Gladstone Institutes, Utah State University, and the University of Utah, have demonstrated a fundamentally different strategy. Instead of trying to fix the broken protein or block its effects, they engineered a CRISPR system that listens for the RNA signature of mutant p53, and when it hears it, detonates the cancer cell’s entire genome from within.

Two Papers, One Month, One Enzyme

On May 6, Ryan Jackson’s lab at Utah State and Yang Liu’s lab at the University of Utah published in Nature the first systematic demonstration that a CRISPR enzyme called Cas12a2 can kill human cells by sensing their RNA. On June 8, Jingkun Zeng and Jennifer Doudna’s team at the Innovative Genomics Institute published in Nature the cancer-specific application: Cas12a2 programmed to recognize mutant p53 transcripts, selectively destroying tumor cells in mouse models of lung and liver cancer. Jackson, his doctoral student Kadin Crosby, and Liu appear as co-authors on both papers. Doudna won the 2020 Nobel Prize for CRISPR-Cas9; this work uses a different, more recently discovered CRISPR system.

The pairing is not coincidental. The first paper establishes the weapon. The second aims it at cancer’s most fortified target.

How Cas12a2 Kills

Standard CRISPR-Cas9 is a scalpel. It finds a specific DNA sequence, cuts it, and lets the cell’s own repair machinery patch the break. The goal is precision editing: fix one letter, insert a gene, knock out a target. For diseases caused by a single known mutation (sickle cell disease, beta-thalassemia), this works. For cancer, where each tumor carries hundreds to thousands of mutations across a chaotic genome, the logic breaks down. You cannot fix a cancer cell one edit at a time.

Cas12a2 is not a scalpel. It is a bomb with a very specific fuse.

The enzyme uses a guide RNA to scan the cell’s transcriptome, the complete set of RNA molecules the cell is producing at any moment. If the guide RNA finds and binds its complementary target, Cas12a2 undergoes a conformational change that activates indiscriminate double-stranded DNase activity. It shreds every piece of chromatin DNA it can reach. The result is catastrophic: hundreds of double-stranded breaks across the genome, a massive DNA damage response, cell cycle arrest, and apoptosis. The cell dies.

If the guide RNA does not find its target, or finds an RNA that differs by even a single nucleotide from the expected sequence, the enzyme remains inert. The cell lives.

“Its goal is not to correct anything,” said Yang Liu, assistant professor of biochemistry at the University of Utah. “Instead, it’s to destroy anything it sees. The enzyme that we’re working with is extremely specific. It does not touch healthy cells. That was striking to us.”

The Numbers

I assembled a comparison to put Cas12a2’s mechanism in context against existing cancer cell-killing approaches.

Approach	Target Specificity	Can Target Mutant p53?	Selectivity for Cancer Cells
Cisplatin / Etoposide (chemo)	None (all dividing cells)	No	Low: kills healthy tissue
ADCs (e.g. Enhertu)	Surface protein (HER2, etc.)	No	Moderate: bystander effects
CAR-T cell therapy	Surface antigen (CD19, BCMA)	No	High for blood cancers; poor in solid tumors
CRISPR-Cas9 gene editing	DNA sequence	In theory; editing too slow	Not designed for cell killing
CRISPR-Cas13 (RNA targeting)	RNA transcript	In theory	Poor: 37% depletion, collateral RNA damage
CRISPR-Cas12a2	RNA transcript (single-nucleotide resolution)	Yes: demonstrated	86% depletion; no effect on non-target cells

The Jackson lab reported 86% cell depletion in HeLa cells expressing the target transcript, with 5.2-fold more DNA double-strand breaks than controls. For comparison, Cas13a achieved only 37% depletion targeting the same site. The DNA damage was comparable in magnitude to cisplatin and etoposide, two of the most widely used chemotherapy drugs, but with a critical difference: Cas12a2 caused damage only in cells expressing the target RNA. Cells lacking the target showed background-level DNA breaks.

In mice, a single treatment reduced tumor volume by approximately 50%. “We demonstrate Cas12a2 can selectively kill cells containing a single-point mutant that causes cancer while leaving cells without the mutant unaffected, with no observable side effects,” said Crosby.

Why This Matters for p53

The Doudna lab’s paper takes the mechanism established by the Jackson group and points it at oncology’s biggest target. By programming Cas12a2 with guide RNAs complementary to the most common p53 hotspot mutations, the team demonstrated selective killing of cancer cells harboring those specific mutations. Because the enzyme discriminates at the level of individual RNA nucleotides, it can distinguish between a cell producing wild-type p53 mRNA and one producing the R175H, R248W, or R273H mutant transcripts that dominate clinical datasets.

This specificity matters because p53 is not just mutated. It is essential. Wild-type p53 is active in every cell in the body, and losing it entirely causes Li-Fraumeni syndrome, a condition of universal cancer predisposition. Any therapy that targets p53 must distinguish between the mutant and the wild-type form with near-perfect precision. Small molecules cannot do this because they interact with the protein’s three-dimensional structure, which is similar across many mutants and the wild type. Cas12a2 bypasses the protein entirely and reads the RNA sequence directly, where each mutation produces a distinct, targetable transcript.

The Addressable Population

The World Health Organization reported 20 million new cancer diagnoses globally in 2022, the most recent year with complete data. If approximately 50% carry p53 mutations, the theoretical addressable population for a p53-targeted therapy is roughly 10 million patients per year. Even restricting the scope to the cancers with the highest p53 mutation rates (ovarian, pancreatic, esophageal squamous cell, small-cell lung) yields a population in the low millions, all of which have median survival times measured in months and limited therapeutic options.

No existing precision oncology platform addresses this population. Antibody-drug conjugates require surface proteins. CAR-T cells require surface antigens. Kinase inhibitors require druggable enzymes. p53 offers none of these handles. What it does produce is a mutant mRNA transcript, in every cancer cell, continuously, for the lifetime of the tumor. Cas12a2 is the first technology demonstrated to exploit that transcript as a kill signal.

Limitations

Five substantial barriers separate these results from a clinical therapy. First, delivery. Both papers relied on electroporation or lipid nanoparticle (LNP) packaging to get Cas12a2 into cells. LNPs accumulate naturally in the liver (which is why the Doudna lab chose liver tumors as one model), but reaching solid tumors in the lung, pancreas, or ovary with sufficient payload remains an unsolved problem across the entire gene therapy field. Second, immune response. Cas12a2 is a bacterial protein foreign to the human body. A single administration may be tolerated, but repeat dosing, likely necessary for tumors that are not fully eliminated, could trigger neutralizing antibodies. Third, tumor heterogeneity. Not every cell in a tumor carries the same p53 mutation. Clonal evolution could generate subpopulations with different mutations or wild-type p53 reversion, escaping a single guide RNA. Fourth, the mouse-to-human gap. A 50% tumor reduction in mice is encouraging but far below the bar for clinical efficacy, and mouse immune systems, tumor microenvironments, and pharmacokinetics differ substantially from those in humans. Fifth, manufacturing. Clinical-grade Cas12a2 protein at scale has not been produced, and the regulatory pathway for a CRISPR-based cell-killing therapy (as opposed to a gene-editing therapy like Casgevy) is entirely uncharted.

Strongest Counterargument

The most compelling competitor is not another CRISPR system but the class of drugs that recently cracked a different “undruggable” target: RAS. In June 2026, Revolution Medicines reported that daraxonrasib, a RAS(ON) inhibitor, extended median survival in pancreatic cancer by 7. months in a Phase III trial, receiving a standing ovation at ASCO. The drug is a small molecule, orally administered, manufacturable at scale, and already in late-stage clinical development. If the pharmaceutical industry can crack RAS with chemistry, the argument goes, eventually it will crack p53 with chemistry too, rendering the biological complexity of Cas12a2 unnecessary.

Where this argument breaks down is in the structural biology. RAS proved “undruggable” for decades not because the protein lacked binding pockets but because the relevant pocket (the nucleotide-binding site) was too small and too high-affinity. Researchers eventually found a different pocket. p53 presents a harder problem: gain-of-function mutations cause the protein to misfold into diverse, unstable conformations. There is no single druggable site because there is no single mutant structure. A molecule that binds R175H p53 may have no affinity for R248W p53. Cas12a2 sidesteps structural biology entirely by targeting the mRNA, where every point mutation is a distinct, unambiguous sequence.

The Bottom Line

CRISPR revolutionized biology as a gene editor. Cas9 cuts DNA to fix mutations. Cas13 cuts RNA to silence genes. Cas12a2 does something no CRISPR system has done before in a therapeutic context: it reads a cell’s RNA, and if that RNA carries a cancer mutation, it destroys the cell’s entire genome. The cell is not edited. It is executed.

Whether this reaches patients depends on delivery, manufacturing, and a regulatory framework that does not yet exist for CRISPR-based cell killing. The timeline is measured in years, probably a decade or more. But the principle demonstrated across these two Nature papers is significant: for the first time, a programmable molecular system can discriminate cancer cells from healthy cells at the resolution of a single RNA nucleotide and kill only the ones it identifies as mutant.

What you can do with this: If you work in oncology drug development, evaluate Cas12a2 as a platform alongside ADCs and CAR-T for cancers driven by intracellular mutations that lack surface markers. The delivery problem is shared with the entire gene therapy field, but the targeting mechanism is unique. If you are evaluating CRISPR-focused biotechs (Intellia, Editas, CRISPR Therapeutics, Caribou), note that Cas12a2 represents a new modality entirely separate from their Cas9-based editing pipelines; the companies closest to this work are Akribion Therapeutics and the Innovative Genomics Institute. If you follow p53 biology, this is the first credible in vivo demonstration that the mutant transcript itself can be weaponized, not just detected. The implications for liquid biopsy-guided therapy selection are substantial: sequence the tumor’s p53 mutation from a blood draw, program a Cas12a2 guide RNA to match, and deliver a personalized cell-killing payload.