For generations, the mutations most responsible for driving cancer have hidden inside the cell, beyond the reach of the antibody therapies that transformed oncology. Researchers at KAIST in South Korea have now crossed that threshold, engineering an antibody that borrows the perceptual logic of T cells to recognize fragments of a mutated protein displayed on the cancer cell's surface — giving precision medicine a new way of seeing. Their target, KRAS(G12D), fuels some of the world's deadliest cancers and was long deemed untreatable by antibody-based means. If the underlying computational metho
Computationally designed antibody targets previously untreatable KRAS cancer mutations
Antibodies that can read the cancer's own fingerprints
Mark
So the core problem was that antibodies work on the cell surface, but KRAS mutations are inside the cell. How does this new antibody actually get around that?
Mimi
It doesn't go inside the cell. Instead, it exploits something cells do naturally. When cancer cells break down their mutated KRAS proteins, they chop them into fragments. Some of those fragments get transported to the cell surface as part of normal cellular housekeeping. The antibody recognizes those fragments—the evidence of the mutation—sitting on the outside.
Mark
So it's reading a fingerprint of the internal mutation rather than accessing the mutation itself.
Mimi
Exactly. And the fingerprint is specific enough that the antibody can tell the difference between a cell with KRAS(G12D) and a normal cell. That specificity is what makes it useful as a therapy.
Mark
Why is this computationally designed approach better than what researchers tried before?
Mimi
The traditional way to develop antibodies is through screening—you generate thousands of variants and test them. With computational design, you're building the antibody from first principles, modeling how it will bind to the target before you even make it. For something as precise as recognizing a single point mutation on a fragment, that modeling saves enormous time and increases the odds of success.
Mark
And the TCR-like part—that's the real innovation?
Mimi
That's where the insight lives. T cell receptors naturally read protein fragments on cell surfaces. By designing an antibody that works the same way, they gave antibodies a new capability they never had before. It's borrowing the immune system's own logic.
Mark
What happens next?
Mimi
Animal models have shown it works. Now comes the long road to human trials, and then the question of whether this approach scales to other mutations. But the fundamental proof is there.
Il Polso
- KRAS(G12D), a mutation driving pancreatic, colorectal, and lung cancers, has resisted antibody therapy for decades because it hides inside the cell — a blind spot that has cost countless lives.
- KAIST researchers discovered that cancer cells inadvertently betray themselves, displaying fragments of their mutated proteins on their surface like unwitting flags — a vulnerability the team moved quickly to exploit.
- By computationally designing a TCR-like antibody that reads these surface fragments the way a T cell would, the team effectively gave a conventional antibody an entirely new sensory capability.
- In animal models, the engineered antibody eliminated KRAS(G12D)-carrying cancer cells with high selectivity, leaving normal tissue unharmed — a precision that distinguishes it from blunter therapeutic approaches.
- The method is not locked to a single target: the same computational design platform could be turned toward other intracellular mutations, potentially unlocking treatments for cancers that have long resisted every available tool.
For generations, the mutations most responsible for driving cancer have hidden inside the cell, beyond the reach of the antibody therapies that transformed oncology. Researchers at KAIST in South Korea have now crossed that threshold, engineering an antibody that borrows the perceptual logic of T cells to recognize fragments of a mutated protein displayed on the cancer cell's surface — giving precision medicine a new way of seeing. Their target, KRAS(G12D), fuels some of the world's deadliest cancers and was long deemed untreatable by antibody-based means. If the underlying computational method proves as versatile as its creators believe, this may mark the beginning of a broader reckoning with the 'undruggable' category itself.
For decades, antibodies have served as precision instruments in cancer treatment — but they have always carried a critical blind spot. The mutations that most aggressively drive cancer tend to hide inside the cell, where conventional antibodies cannot follow. This has left certain cancers, including some of the deadliest, effectively beyond the reach of antibody-based medicine.
A team at KAIST, working alongside the startup Therazyne, has now found a way through that barrier. Their insight was biological: when cells break down damaged or old proteins as part of normal housekeeping, they sometimes fragment mutated proteins and transport those pieces to the cell surface. There, the fragments sit exposed — small molecular flags that, in principle, could announce a cell's malignant nature to the right observer.
The researchers built that observer. Using computational design, they engineered an antibody modeled on the T cell receptor, the immune system's own tool for reading protein fragments on cell surfaces. The result is an antibody with what might be called T cell vision — capable of recognizing the specific fragment produced by the KRAS(G12D) mutation and distinguishing it from the normal protein with striking accuracy. In animal models, it cleared cancer cells carrying the mutation while leaving healthy tissue untouched.
KRAS(G12D) is no minor target. It appears across significant fractions of pancreatic, colorectal, and lung cancers — malignancies that together claim enormous numbers of lives each year. The mutation was long labeled 'undruggable' not because treatment was theoretically impossible, but because no available tool could reach it. Published in June in Molecular Therapy and led by Professor Byung-Ha Oh, this work suggests that label may need revision.
Perhaps most consequentially, the computational method behind the antibody is not specific to KRAS. The same design logic could be applied to other intracellular mutations that have resisted treatment — transforming what began as a single breakthrough into a potential platform for reaching cancers that have long seemed beyond medicine's grasp. Human trials remain years away, but the foundational barrier has been crossed.
For decades, antibodies have worked like precision weapons in cancer treatment—proteins engineered to seek out and destroy malignant cells with remarkable specificity. But they have always had a critical blind spot. The mutations that actually drive cancer, the genetic errors that tell a cell to grow without limit, often hide inside the cell where antibodies cannot follow. This limitation has made certain cancers nearly impossible to treat with antibody-based therapies, no matter how sophisticated the design.
A team at KAIST, South Korea's premier science and technology university, has now found a way around this barrier. Working with researchers from Therazyne, a startup founded by the lab's director, they have created an antibody that can recognize and attack cancer cells carrying KRAS(G12D), one of the most common cancer-driving mutations in the world. The breakthrough hinges on a clever biological insight: cancer cells don't keep their secrets perfectly. When cells break down old or damaged proteins—a normal housekeeping process—they sometimes chop up the mutated proteins into fragments. Some of these fragments get transported to the cell surface, where they sit like flags announcing the cell's malignant nature to the immune system.
The researchers used computational design to engineer an antibody that could read these flags. They modeled it after a T cell receptor, the sensor that allows immune cells to identify infected or cancerous cells by examining protein fragments on their surface. By giving an antibody this same ability to recognize mutant protein fragments, they essentially gave it what one might call T cell vision. The resulting antibody, developed through a combination of computer modeling and laboratory screening, can distinguish cancer cells carrying the KRAS(G12D) mutation from healthy cells with striking precision. When tested, it showed little to no reaction with normal tissue or unrelated proteins.
KRAS(G12D) is not an obscure target. The mutation appears in a significant fraction of pancreatic cancers, colorectal cancers, and lung cancers—some of the deadliest malignancies. For years, it was considered an "undruggable target," a phrase that meant not impossible to treat, exactly, but impossible to treat with the tools that had worked elsewhere. The protein sits inside the cell, where conventional antibodies cannot reach it. Small-molecule drugs have been developed to attack KRAS directly, but they work differently and come with their own limitations. An antibody therapy that could harness the immune system's own machinery to eliminate these cells would represent a fundamentally different approach.
The team's work, published in June in Molecular Therapy, a leading journal in gene and cell therapy, showed that their antibody could do exactly that. In animal models, it effectively eliminated cancer cells carrying the mutation while leaving normal cells untouched. The researchers, led by Professor Byung-Ha Oh, emphasized that this selectivity matters enormously. Precision in cancer treatment means not just killing cancer cells but doing so without collateral damage to healthy tissue. An antibody that can distinguish a single point mutation from the normal version of the same protein represents a significant step toward that ideal.
What makes this work particularly significant is its potential as a platform. The computational design method the team developed is not specific to KRAS(G12D). The same approach could theoretically be applied to other intracellular cancer mutations, other "undruggable targets" that have resisted conventional treatment. If that proves true, this single breakthrough could open pathways to therapies for cancers that have long seemed beyond reach. The next phase will be moving from animal models to human trials, a process that typically takes years. But the fundamental barrier—the question of whether antibodies could ever reach inside a cell to recognize a cancer mutation—has now been breached.
Citazioni salienti
The antibody developed in this study can selectively identify only cancer cells carrying the KRAS(G12D) mutation, demonstrating the potential for precision antibody therapeutics that minimize damage to normal cells.— Professor Byung-Ha Oh, KAIST Department of Biological Sciences
Vuoi la storia completa?
Leggi l'originale su News-Medical ↗
Health & Medicine