Protein evolution occurs slowly via small incremental mutations—usually. However, dramatic changes can happen: for instance, a fragment of one protein getting copied into the middle of another. While this kind of structural shake-up can sometimes be disastrous for protein stability, in other situations it spurs novelty and functional innovation.
One such event occurred early in chordate evolution when a member of the integrin protein family picked up a new domain. It then repurposed, or exapted, this insertion I domain to expand its ligand binding repertoire.
Integrins are transmembrane proteins that link the exterior environment of the cell to the interior and cytoskeleton. They are two-part, or heterodimeric, complexes of one alpha and one beta subunit. Humans encode 18 alpha and 8 beta integrins—mixed-and-matched to form 24 unique combinations—which are crucial in processes like focal adhesion and cell migration.
The activity of integrins depends on their shape. In the “off” conformation, integrins fold in a compact structure with low ligand affinity. In the active conformation with high ligand affinity, the integrin headpiece folds upward and the legs extend outward in a process likened to a switch-blade knife unfolding.
Despite their importance in health and disease, “integrins are enigmatic in their ability to translate structural conformation to cellular signaling,” says Jeremy Hollis, a graduate student in the Campbell Lab and one of the 2025 recipients of the Harold M. Weintraub Graduate Student Award. He is the lead author of a recent study in Science Advances that integrates cryoEM with evolutionary analyses to reveal fascinating insights about integrin biology.
“CryoEM has greatly expanded our ability to probe these conformations, as more classical structural methods like crystallography fail to capture the extensive integrin range of motion,” he explains. “We hope this work elevates our field by giving integrin conformation its day in the sun.”
Hollis is fascinated by the I domain present in about half of the alpha subunits. This domain insertion “in some ways fundamentally changed how integrins worked, but in others totally preserved their critical functions,” says Hollis.
Integrins lacking the I domain bind their ligand at an interface between the alpha and beta subunits. Many ligands possess an amino acid motif RGD (arginine-glycine-aspartate) that glues to this interface and activates the integrin’s conformational change. However, the I domain is inserted into the alpha protein such that it blocks this binding interface, seemingly preventing canonical integrin activation. It is hypothesized that the I domain structurally stabilizes the ⍺β interface to compensate for the loss of the ancestral binding pocket.
The team focused on two integrins: α4β7 and αEβ7. While both alphas complex with integrin subunit β7, α4 does not have the I domain while αE does. Comparing structures of both complexes using cryoEM allowed the team to understand how the I domain impacts ligand binding and integrin function.
Hollis and co-authors began by imaging α4β7 bound to its cognate ligand, mucosal addressin cell adhesion molecule-1 (MAdCAM-1). They found that MAdCAM-1 bridges the αβ interface with an RGLD motif (like an RGD motif but with a leucine added for flavor). These contacts are stabilized by a region in MAdCAM-1 that coordinates a metal ion-dependent adhesion site in the β loop (βMIDAS).
Next, the authors profiled αEβ7 without ligand in high-calcium buffers that should support the inactive conformation. Unexpectedly, they saw two populations of molecules: many were in the expected compact, inactive shape, but some were in an open conformation despite no ligand being present.
The explanation lies in an EG motif in the I domain which coordinates with the βMIDAS to engage β7. This extends β7 into an active state with a high degree of leg flexibility. In other words, the I domain may allow the integrin to “sample” an open state even in the absence of ligand.
So, what happens when αEβ7 does bind a ligand? To test this, the authors profiled αEβ7 in complex with its own cognate ligand, E-cadherin. They found that E-cadherin binds the distal side of the I domain, pushing the internal ligand to consistently engage with β7. This stabilizes the leg into the open position but does not completely restrict β7 leg movement.
It is amazing how similar ligand-bound α4β7 and αEβ7 are despite their different mechanisms of ligand binding and activation. This is due to the impressive structural mimicry of the EG in the I domain that simulates external ligand binding.
So, where did the I domain come from? The authors hypothesized that its ancestor must have had the ability to engage the ligand binding pocket, or its incorporation wouldn’t have been evolutionarily advantageous. They tracked the origin of the I domain to similar domains found in collagen proteins.
However, one mystery remained: the origins of the essential EG motif. They suggest that perhaps the ancestral form of the I domain did indeed contain this motif or a similar one (DG was found in the closest related animal). To test this, they profiled the degree of integrin activation with different motifs swapped into the I domain. The ancestral reconstruction of the I domain loop as well as the DG domain were able to activate integrin, albeit to a lesser degree than the canonical EG motif.
“I'm personally attached to this project because I think it can teach us some principles about how proteins evolve on the broader scale,” says Hollis. “Here this domain serendipitously co-opted, or exapted, machinery that was already present for new function.”
Hollis points out that this study only focused on one of nine integrins that possess the I domain. “We think the next step is to see how many of the concepts we probe in this work are universally applicable and how many are unique to αEβ7 by surveying a larger sample of this integrin family with cryoEM,” he says.
“It's becoming clearer that each integrin has its own idiosyncrasies that make it unique,” he adds. “We were fortunate to jump a few technical hurdles with this work that make answering some of these next questions much more achievable.”
Fred Hutch/University of Washington/Seattle Children’s Cancer Consortium Members Melody Campbell and Harmit Malik contributed to this research.
The spotlighted research was funded by a National Science Foundation Graduate Research Fellowship, a Pew Biomedical Scholars Award, the Mahan Fellowship, the National Institutes of Health, and the Fred Hutch Cancer Center.
Hollis JA, Chan MC, Malik HS, Campbell MG. 2025. Molecular exaptation by the integrin αI domain. Science Advances. doi: 10.1126/sciadv.adx9567.
Hannah Lewis is a postdoctoral research fellow with Jim Boonyaratanakornkit’s group in the Vaccine and Infectious Disease Division (VIDD). She is developing screens to find rare B cells that produce protective antibodies against human herpesviruses. She obtained her PhD in molecular and cellular biology from the University of Washington.
For the Media