Some proteins do not settle into one stable shape on their own. Yet when two such proteins meet, they can form a precise structure together. A study of protein 4.1G and NuMA, proteins involved in cell division, used experiments and Google’s AlphaFold AI software to reveal how that can happen.
When a protein’s flexibility is part of its function
For many proteins, a stable three-dimensional shape is central to what the protein does. Its folds and pockets can help drive chemical reactions or bind a particular chemical inside a cell. Scientists have developed several ways to determine these shapes.
Intrinsically disordered proteins complicate that picture. Some regions appear to move continually, while other proteins resist structural analysis altogether. Researchers now understand that this apparent disorder can reflect the protein’s actual behavior, and that it can matter to its function.
Disorder does not have one universal role. A flexible region might adopt a structure when it binds another molecule, take different shapes with different partners, or remain disordered while active. Working out which behavior applies to a particular protein can be difficult.
Testing a pair involved in cell division
The researchers in Hefei, China, focused on protein 4.1G and NuMA. Their interaction is essential for cell division, and the regions that connect the two proteins had been identified. Both regions are intrinsically disordered, leaving open the question of what they look like when they bind.
To find which parts mattered for the interaction, one group of researchers linked the two regions to separate halves of a protein that catalyzes a chemical reaction. If the regions bound, they would bring those halves together and allow the reaction to proceed. The team introduced mutations and measured the reaction rate, identifying locations where changes disrupted the interaction.
A separate approach modeled a disordered region’s possible shapes using physical features such as charge interactions and interactions with water. Over 200 nanoseconds, the region shifted among 15 conformations. That result illustrated the challenge: a flexible protein can sample many arrangements, making its functional binding shape hard to isolate.
AlphaFold predicted a shared structure
The team then used AlphaFold to predict protein complexes, pairing one disordered region with pieces of the other protein. The predictions repeatedly showed the same configuration. It also fit the earlier mutation results: the predicted arrangement depended on locations that experiments had shown were important for the interaction.
In the proposed structure, three amino-acid sheets run antiparallel to one another. Two sheets come from one protein and the third from its partner. This arrangement helps explain why the regions can remain disordered separately: the structure depends on both proteins being present.
The researchers sought experimental support for the prediction. A second AI package suggested mutations that might stabilize the complex. Tests identified a mutant that made the complex stable enough to obtain a crystal of the proteins, confirming the AlphaFold-predicted structure.
The researchers also used AlphaFold to search for other proteins with regions that might interact in a similar way. Of the 38 potential partners tested, seven interacted. That result points to additional possible examples, while leaving open how common this mechanism is.
Flexibility may help protein 4.1G connect partners
A stable structure could form this kind of complex, too, so the structure alone does not explain why protein 4.1G is intrinsically disordered. The study offers a functional possibility: protein 4.1G acts like a bridge, interacting with many proteins and bringing them together. Its flexibility may let it connect with a wider range of partners.
The findings do not show that all intrinsically disordered proteins form ordered structures when they bind. They establish an example in which two disordered regions create a specific arrangement together, and show how an AI prediction can help guide experiments on a difficult biological question.