Researchers screened a library of 59,319 peptides on a computer and pulled out one that sticks to KRAS G12V, a cancer mutation that the approved generation of pills was not built to hit.
The work was published online June 11 in the Journal of Enzyme Inhibition and Medicinal Chemistry ↗. It is early, preclinical, and computational at its core, but it lands on a target that has frustrated drug makers for forty years.
The mutation the pills miss
KRAS is one of the most frequently mutated genes in human cancer. The protein works as an on-off switch for cell growth, and certain mutations jam it in the "on" position so the cell keeps dividing. In colorectal cancer, the version called G12V is one of the common jams.
The two approved KRAS drugs, sotorasib and adagrasib, were engineered for a different specific jam called G12C. They slot into a pocket that only opens up in the G12C version, and they do not work on G12V. That leaves patients whose tumors carry the G12V mutation with little aimed at the actual driver of their disease.
How they found it
Instead of testing peptides in a dish one at a time, the team used structure-based virtual screening. They docked all 59,319 candidates against the G12V protein in software and ranked them by how well they fit the surface. Four came out as plausible binders. The best, which they call Peptide-1, then went through molecular dynamics simulations and free-energy calculations, which model how a peptide and a protein jostle and hold together over time. Those suggested a stable grip built from hydrogen bonds and oily, water-avoiding contacts.
The screen is the easy part. The harder claim is that the hit behaves the same way off the computer. The team confirmed binding with microscale thermophoresis, a bench assay that measures how tightly two molecules associate, and Peptide-1 bound G12V harder than it bound other KRAS mutants. That selectivity is the point. A useful drug should prefer the cancer's exact version of the protein over its near relatives. The peptide also survived in human serum, a basic durability check, and a second assay (NanoBRET) showed it reaches and engages KRAS inside living cells, not just in a tube.
The cell evidence, including the useful negative
In colorectal cancer cell lines, Peptide-1 slowed growth, and it did less to normal cells. The most persuasive result is a negative one. When the researchers knocked down KRAS G12V, the peptide lost much of its effect. If a drug works by hitting a target, removing the target should blunt it, and it did.
In SW480 cells, a standard colorectal line, the peptide turned down ERK1/2 phosphorylation, a signal that sits directly downstream of KRAS, raised p21, which acts as a brake on the cell cycle, and parked cells in the resting G0/G1 phase instead of letting them divide. Three readouts, all pointing the same way, all consistent with shutting off the switch.
What this is not
It is not a drug. There is no animal data, no dosing, and no disclosed sequence, and "potent" in the title is relative to three sibling peptides, not to anything in the clinic. Virtual screening is good at proposing binders and bad at guaranteeing they will behave in a body. Oncology is littered with compounds that gripped a target in a dish and did nothing in a patient. What earns this one a second look is the pairing of a selectivity result with a knockdown rescue, two independent ways of arguing the effect actually runs through G12V rather than through some off-target nuisance.
Peptides going after KRAS is no longer a fringe idea. peptidemodel covered Chugai's LUNA18 ↗, an oral macrocyclic peptide aimed at the broader RAS family that recently cleared kilogram-scale manufacturing. That program and this one sit at opposite ends of the same bet. LUNA18 is a clinical-stage molecule with a supply chain and a pan-RAS reach. Peptide-1 is a computer hit with a narrow, mutant-specific aim. Both ask whether a peptide can do what small molecules have struggled to do against one of cancer's most wanted targets, and for the patients with a G12V tumor, the narrow aim is the whole point.