The integrin lymphocyte function-associated antigen-1 (LFA-1, αLβ2) participates in a wide range of adhesive interactions including antigen recognition, emigration from the vasculature, and migration of leukocytes within tissues1,2. Integrin ectodomains assume three global conformational states (Fig. 1a) with the extended-open conformation binding ligand with ~1,000-fold higher affinity than the bent-closed and extended-closed conformations2. Binding of LFA-1 to intercellular adhesion molecule (ICAM) ligands by the αI domain in the integrin head is communicated through the β-subunit leg, transmembrane, and cytoplasmic domains to the actin cytoskeleton via adaptors such as talins and kindlins that bind specific sites in the β-subunit cytoplasmic domain3. As reviewed4,5, measurements of traction force on substrates and more specific measurements of force within ligands and cytoskeletal components have suggested that integrins are important in force transmission between extracellular ligands and the actin cytoskeleton. Forces on the cytoplasmic domain of the LFA-1 β2-subunit have been measured in the 1-6 pN range and associated with binding to ligand and the cytoskeleton6.
Tensile forces exerted through integrins between the actin cytoskeleton and extracellular ligands have the potential to align integrins; such alignment would in turn provide the opportunity to test alternative models of integrin activation. Some models suggest that binding of the cytoskeletal adaptor protein talin to the integrin β-subunit cytoplasmic domain is fully sufficient to activate high affinity of the extracellular domain for ligand7,8. Other models, supported by steered molecular dynamics and measurements in migrating cells, have proposed that tensile force stabilizes the high-affinity, extended-open integrin conformation because of its increased length along the tensile force-bearing direction compared to the other two integrin conformations (Fig. 1a)6,9,,-12. These models propose that inherent in the three conformational states of integrins is a mechanism by which integrin adhesiveness can be activated when the integrin simultaneously binds the actin cytoskeleton and an extracellular ligand that can resist cytoskeleton-applied force. Thus, the same intracellular effectors that regulate actin dynamics can simultaneously and coordinately regulate cell adhesion to provide the traction for cellular chemotaxis and migration. Furthermore, directional migration is a critical aspect of immune cell function, and alignment of integrins by activation would provide a mechanism for directional sensing. While this model is appealing to structural biologists, it requires validation in cells.
Here, we test a key prediction of the cytoskeletal force model of integrin activation: that the tensile force exerted through integrins between the actin cytoskeleton and extracellular ligands as they function in cell migration causes them to assume a specific orientation and tilt on the cell surface relative to the direction of pulling on the integrin by actin retrograde flow; actin flow is known to be locally aligned for migrating fibroblast and epithelial cells13,14. Measuring integrin orientation on cell surfaces also provides an opportunity to correlate crystal structures of integrins at the Å length scale with microscopic measurements on integrin-bearing cells at the micron length scale. Integrating measurements at such different length scales is a long-standing goal of many fields of biological research. While integrins like other membrane proteins are generally free to rotate in the plane of the membrane, tensile force would cause an integrin to orient in the same direction as the pulling force. Like most membrane proteins, integrins are drawn in cartoons (as in Fig. 1a) as projecting with their leg-like domains normal to the plasma membrane; however, resting integrins are free to tilt15 and force could tilt the integrin far from the membrane normal. In general, despite a wealth of structures for membrane protein ectodomains, little is known about their orientation on cell surfaces.
In this work, we make use of previous structural studies on integrins2,16,17, and orient these structures in a reference frame that corresponds to the plasma membrane of a migrating lymphocyte. In addition to general structural knowledge on many integrin families, we make specific use of crystal structures for the αI domain of LFA-1 bound to ICAMs, the LFA-1 headpiece, and two states of the bent ectodomain of the LFA-1 (αLβ2) relative, αXβ2. We also use negative stain EM class averages showing the bent-closed, extended-closed, and extended-open conformations of the αLβ2 and αXβ2 ectodomains. These structures together with those of green fluorescent protein (GFP) have guided development here of constrained integrin-GFP fusions and prediction using Rosetta18 of the orientation of GFP and its fluorescent excitation/emission transition dipole relative to the integrin. Two different types of fluorescent microscopes provide similar measurements of the orientation of the transition dipole relative to the direction of actin flow. Integrin-ligand engagement in combination with cytoskeletal force results in spatially ordered organization of LFA-1 in the protrusive lamellipodial region and is dependent on the movement vector of the underlying actin cytoskeletal framework19,20. The results show that actin flow from the leading edge dictates a specific molecular orientation on the cell surface of LFA-1 and support the cytoskeletal force model of integrin activation.