page 1109. Integrin activation by internal signals may therefore not depend on the protein assuming a more extended conformation.
To prevent unregulated adhesion, most integrins are expressed in an inactive form. EM and crystal structures showed that the extracellular domain of the integrin αVβ3 without its ligand can assume a bent conformation. 2D EM reconstructions of the same domain in the presence of a small peptide suggested that the bound form, in contrast, was extended. These results supported a switchblade model, in which the binding of cytoplasmic signals initiates the movement of several intra- and extracellular domains that opens and activates the integrin by exposing its ligand-binding site.
But Adair et al. suggest that such large-scale changes may not be necessary. The authors generated 3D EM reconstructions using a larger, physiological ligand (in this case, a piece of fibronectin). Particles to be analyzed were also selected automatically to minimize bias.
The 3D maps showed that most of the fibronectin-bound integrin complexes were in a bent conformation. The maps with and without the ligand looked generally similar. But in the ligand's presence, additional densities consistent with the size of two fibronectin domains were found at the ligand-binding sites.
If, as the findings indicate, no large-scale opening is needed to make room for ligand, then the mechanism of activation by intracellular signals should be revisited. For instance, a slight sideways movement of a membrane-proximal extracellular domain of the β-subunit away from the ligand-binding site may be sufficient. This model involves much smaller conformational changes than does the switchblade model.
It is possible that the switchblade movement has quite a different function. For example, it may be part of the conformational changes induced by ligand binding that culminate in intracellular responses.