De novo Design of Macrocyclic Molecular Glues

Inside every living cell, a vast network of machines works to keep things running smoothly, but sometimes these machines malfunction, causing disease. One way scientists try to fix a broken machine is by building a tiny tool that latches onto it and stops it from working. For decades, the standard approach has been to design a tool that fits perfectly into a specific hole on the target machine, blocking its function. However, many of the most dangerous malfunctioning parts in our cells do not have these holes, making them impossible to stop with traditional tools. A newer strategy has emerged that changes the rules entirely: instead of blocking a part, scientists try to force two different parts to stick together. This concept, known as induced proximity, relies on a small molecule acting as a bridge, or a glue, to bring a target protein close to a cellular cleanup crew. When the target is held next to this crew, the cell recognizes it as waste and destroys it. While this method has shown great promise, finding these glue molecules has mostly been a matter of luck, with scientists stumbling upon them by accident and then trying to improve them after the fact.

A team of researchers has now developed a way to design these molecular glues from scratch, using only the genetic code of the proteins involved. They created a computer system called EvoBind-multimer, which acts as a designer for these bridges. Unlike older methods that require a detailed 3D map of how two proteins fit together, this new system starts with just the sequence of letters that make up the proteins. It then generates entirely new, ring-shaped chains of amino acids, known as macrocyclic peptides, specifically shaped to connect two chosen proteins. The researchers tested this system by asking it to design a bridge between a cellular cleanup enzyme called VHL and two difficult targets: KRAS and BRD4. These targets are known to drive cancer growth and have historically been very hard to treat. The computer produced new macrocyclic molecules, and when the team tested them in living human cells, the designs worked exactly as intended. The molecules successfully pulled the cleanup enzyme and the cancer targets together, forming a three-part complex that triggered the cell to break down the unwanted proteins and stop their harmful signals.

The results went beyond simple destruction, revealing a surprising nuance in how these glues behave. When the researchers tested the same designed molecules in tumor samples taken from different patients, the outcome depended entirely on the specific environment of the cell. In one patient model, the molecules acted as powerful degraders, successfully eliminating the target protein. In another patient model, the exact same molecules did not destroy the target. Instead, they held the target and the cleanup enzyme together in a stable lock, effectively trapping the target in place without breaking it down. The researchers call these stabilizing agents "LOCKTACs." This finding shows that the success of these molecular glues is not just about the design of the glue itself, but also about the context of the cell it enters. By proving that they can design these bridges directly from genetic sequences without needing prior knowledge of the protein shapes, the researchers have opened a path to creating new ways to control protein function. This approach moves the field away from relying on accidental discoveries and toward a future where scientists can deliberately engineer new biological functions to treat diseases that were previously considered untouchable.