A Lausanne lab built 15,360 cyclic peptides more or less at random and screened the whole batch for one rare trait: the ability to cross a cell membrane on their own. Most could not. The point of the exercise was the handful that could.
The work, from Christian Heinis's group at EPFL, was published June 1 in Nature Chemical Biology ↗. It is a direct run at the single property that has kept peptide drugs penned in for decades.
The wall peptides cannot get past
Peptides are short chains of amino acids, the same building blocks proteins are made of. That makes them very good at one thing: gripping the large, flat contact patches where two proteins meet, the targets that small conventional pills tend to slide off. The catch is the cell membrane, the oily skin around every cell. Peptides are too big and too water-loving to slip through it.
So the modality has lived with a hard limit. A peptide drug can hit a target that sticks out on the cell surface, which is why nearly every peptide medicine on the market is an injection aimed at a receptor on the outside of a cell. The far larger set of targets, the proteins doing business inside the cell, has been mostly off-limits. Oral dosing runs into the same wall twice, since a pill has to survive the gut and then get into cells.
Screening for the trait instead of guessing it
The usual way to chase membrane permeability is to start from a peptide that already binds something and then tinker with it. Heinis's team inverted that. They synthesized a library of 15,360 fully random cyclic peptides, looped rather than linear, and built the whole library to be small, compact, and relatively greasy, the chemistry that correlates with getting through a membrane. Then they screened for the crossers directly, before worrying about what those peptides might bind.
From that pool they refined a lead they call peptide 30, with a molecular weight of 890.6 daltons, large enough to grip a protein-protein interface and small enough to pass the membrane. As a test target they picked Keap1-Nrf2, a switch that sits inside the cell and governs its response to oxidative stress, the slow chemical damage tied to inflammation, neurodegeneration, and cancer. It is a classic intracellular protein pairing that drug hunters have wanted to pry apart for years and mostly could not reach. In living cells, peptide 30 blocked the interaction, and it did so in proportion to dose, the basic sign that the effect is real rather than incidental.
Why this is a platform result, not a one-off
The molecule is not a drug, and the authors do not claim it is. The result is the route. Building permeability into a library up front, then screening for it without a known starting ligand, means the approach can in principle be pointed at intracellular targets that have no peptide binder yet. That is the half of the proteome that the modality has not been able to touch.
The commercial read is already in motion. Heinis has patented the method and spun out a company, Orbis Medicines, which raised more than 90 million euros in a Series A round to push membrane-permeable macrocycles toward oral drugs. Investors are pricing the property, not the single peptide.
For the peptide field, the constraint this attacks is the same one that shapes the rest of the modality. The reason the blockbuster metabolic peptides are weekly shots rather than daily tablets, and the reason most peptide programs aim at surface receptors, traces back to the membrane. A reliable way to make cyclic peptides that cross it would not change any drug already approved. It would change which targets are worth starting a peptide program against at all. That is the part to watch, and the next test is whether a second target, picked because the biology demands it rather than because it was a clean demo, comes through the same screen.