Direction of actin flow dictates integrin LFA-1 orientation during leukocyte migration

Abstract

Integrin αβ heterodimer cell surface receptors mediate adhesive interactions that provide traction for cell migration. Here, we test whether the integrin, when engaged to an extracellular ligand and the cytoskeleton, adopts a specific orientation dictated by the direction of actin flow on the surface of migrating cells. We insert GFP into the rigid, ligand-binding head of the integrin, model with Rosetta the orientation of GFP and its transition dipole relative to the integrin head, and measure orientation with fluorescence polarization microscopy. Cytoskeleton and ligand-bound integrins orient in the same direction as retrograde actin flow with their cytoskeleton-binding β-subunits tilted by applied force. The measurements demonstrate that intracellular forces can orient cell surface integrins and support a molecular model of integrin activation by cytoskeletal force. Our results place atomic, Å-scale structures of cell surface receptors in the context of functional and cellular, μm-scale measurements.

Fig. 1

Integrins, GFP fusions, and modeling GFP and transition dipole orientation with Rosetta. a Three global conformational states of integrins. Cartoons depict each integrin domain and GFP with its transition dipole (red double-headed arrows). b Ribbon diagram of the integrin headpiece of αL-T bound to ICAM-1. The GFP insertion site in the β-propeller domain is arrowed. Dipole is shown in red. c Cartoon as in a of ICAM-engaged, extended-open LFA-1 showing direction of leading edge motion and actin flow. Large arrows show pull on integrin-β by actin and resistance by ICAM-1. Axes shown in a, c are similar to those in the reference state in Fig. 6. d Sequences and boundaries used in GFP-LFA-1 fusions. Highlighted residues were completely modeled by Rosetta to link GFP to the integrin (yellow) or altered in sidechain orientation only to minimize energy (orange). e Orientation of the transition dipole in GFP-LFA-1 fusions. Integrin domains are shown as ellipsoids or torus and GFP is shown in cartoon for 1 ensemble member. GFP transition dipoles are shown as cylinders with cones at each end for 20 representative Rosetta ensemble members, with the asymmetry of GFP referenced by using different colors for the ends of transition dipoles (which themselves have dyad symmetry)

Fig. 2

Emission anisotropy of GFP-LFA-1 fusions. a Schematic of emission anisotropy TIRF microscopy. The transition dipole of GFP has an excitation dipole (green) that is very close in orientation to its emission dipole (red)^,. I _II and I _⊥ are emission intensities parallel and perpendicular to the direction of polarized excitation. b–d Schematics showing a cell and integrins in same microscope xy plane as in a and different outcomes in emission anisotropy. The excitation (electric) field of polarized light is shown as a blue wave. b. Outcomes using a constrained integrin-GFP fusion. Depending on the orientation of integrins within pixels (1–3) or regions of interest (ROI), the emission anisotropy will change as indicated. c, d Outcomes with aligned integrins with constrained c or unconstrained GFP d. e Representative images from movies of Jurkat T cells stably expressing GFP-LFA-1 fusions migrating on ICAM-1. Each pair of panels (scale bars are 5 µm) shows total GFP fluorescence intensity (upper) and anisotropy (lower, color scale to right). f Emission anisotropy of GFP-LFA-1 fusions, averaged over Jurkat T cells migrating in random directions, in at least five independent experiments. Box plots show the full range (whiskers) of observations with median as line and 25–75 percentile range boxed. Kruskal–Wallis test with multiple comparison correction gave the indicated P values. N (number of cells) from left was 18, 22, 17, 37. *p < 0.05; **p < 0.01; ****p < 0.0001

Fig. 3

Actin flow dynamics, orientation, and relation to GFP-LFA-1 anisotropy in migrating T cells. a Representative frames of structured illumination microscopy movies (above) of the actin cytoskeleton visualized with lifeact-mNeonGreen in Jurkat T cells migrating on ICAM-1 (10 µg/ml), anti-CD43 IgG (10 µg/ml), or their mixture (10 µg/ml each). Optical actin flow vector maps of the same cells are shown below with zoom insets of representative areas. Vectors encode flow direction by color (circular keys in the direction from center of circle to perimeter) and velocity by length. Scale bars encode dimensions in the micrograph (white, 5 µm) and velocity (yellow, 30 nm/s). b Actin flow direction relative to tangent of leading edge membrane. Bins of 15° are shown with Gaussian fit and mean ± SEM. ICAM-1, N = 62; αCD43, N = 63; ICAM-1 + αCD43, N = 67. ROI (N) came from six to eight cells. c Leading edge actin flow velocity from optical flow analysis. Plots show data on each cell. Bars show mean ± s.d. Two-tailed Mann–Whitney tests all show p < 0.0001 (****). ICAM-1, N = 39; αCD43 N = 38; and ICAM-1 + αCD43, N = 25. d, e Jurkat T cells migrating on ICAM-1 (10 µg/ml) expressing αL-T-GFP were treated with DMSO as control, blebbistatin (100 µM), or cytochalasin D (100 nM) and fixed. d Whole-cell emission anisotropy analyzed as in Fig. 2e with error bars showing SEM of three independent experiments. Mann–Whitney test for DMSO (N = 37) vs. blebbistatin (N = 9) was 0.130 and for DMSO vs. cytochalasin D (N = 14) was 0.003. **p < 0.01. e Representative total fluorescence intensity (I _|| + 2I _⊥, left) and anisotropy (right). Scale bars: 5 µm

Fig. 4

Orientation of LFA-1 in the leading edge of migrating cells by EA-TIRFM. a Schematic showing relation between excitation polarization orientation, transition dipole orientation, and emission anisotropy in EA-TIRFM. Their quantitative relationship is described by the cos² function (inset equation) where r is anisotropy, A is the absolute amplitude of angular dependence and reports the degree of GFP dipole alignment; γ is the angle between the membrane normal and the excitation axis; θ _d is the angle between the membrane normal and transition dipole and is equivalent to the phase shift in the cosine function; and C is isotropic background. b Simulation of ability to find a predefined dipole orientation angle relative to membrane normal in an idealized leading edge. Test data were generated with a fixed amplitude (0.05) and five θ _d angles. Left: simulated leading edge images with anisotropy color-coded according to the key. Right: fitted curves. c, d Representative segmentation (cell, edge, protrusion, and leading edge) of a migrating Jurkat T cell expressing αL-T-GFP with corresponding maps of orientation relative to the cell midline c and anisotropy d in the segmented regions (panels 1–4). Scale bar = 5 µm. e Emission anisotropy of cell, edge, protrusion, and leading edge segments for αL-T, cytosolic GFP, and GFP-CAAX (Supplementary Figs. 5–8) fit to the cos² function. Plots show anisotropy vs. orientation relative to the cell midline for each pixel in individual, representative cells (from c and d for αL-T), the running average (blue) and fit (red). Values for the fit parameters (A = absolute amplitude, θ _d = phase shift, R ² = goodness-of-fit) are displayed in each plot. f Absolute amplitudes of emission anisotropy fits to the cos² relationship. Box plots show the full range (whiskers) of observations with median as line and 5–95 percentile range boxed. Kruskal–Wallis test with multiple comparison correction to αL-T leading edge data (N = 206) gave P values from left to right: < 0.0001 (N = 85); <0.0001 (N = 52); <0.0001 (N = 58); 0.0033 (N = 185); <0.0001 (N = 71); <0.0001 (N = 83); <0.0001 (N = 55) with N leading edges. See Supplementary Table 3 for more details. A two-tailed Mann–Whitney test of αL-F in absence and presence of talin head gave a P value of 0.0059. **p < 0.01, ****p < 0.0001

Fig. 5

Integrin alignment and orientation in migrating T cells measured by instantaneous FluoPolScope. a Schematic of Instantaneous FluoPolScope and equation for polarization factor p. b Representative total fluorescence intensity image of αL-T Jurkat T cell migrating on ICAM-1 (10 µg/ml) with overlay of ROIs (white = leading edge, magenta = cell body), normal to leading edge tangent (yellow), and average GFP emission dipole orientation with length proportional to polarization factor (red). Scale bars = 1 µm. Panel below is enlarged from dashed area. c Polarization factors of GFP and GFP-LFA-1 fusions in T cells migrating in random directions. Box plots show the full range (whiskers) of observations with median as line and 25–75 percentile range boxed. Kruskal–Wallis test with multiple comparison correction of leading edges gave the indicated P values to αL-T using N number of cells shown in d. **p < 0.01, ***p < 0.001, ****p < 0.0001. d Radial histograms of GFP transition dipole projection in image plane relative to the membrane normal (θ _d). Each radial histogram wedge shows mean intensity-weighted polarization factor (p) in 15° bins (solid outlines) and is reflected to represent the dyad symmetry axis of the dipole. Dashed propeller-shaped outlines in the lower two panels show the fit of the data to a circular Gaussian. e–g, e A representative αL-T Jurkat cell migrating on ICAM-1 was segmented as shown in b. Segments are color coded in blue-red rainbow around the leading edge and are shown in same color in f, g. Each segment is represented as a reflected wedge at the angle of its experimentally measured emission dipole projection with respect to the x axis f or with respect to the membrane normal (θ _d) g

Fig. 6

Molecular model of LFA-1 orientation on the cell surface. a Schematic of the integrin-microscope reference frame. b, c Details of the integrin-microscope reference frame. xy and xz planes are black and blue grids, respectively. Integrin Cα atoms used to define the origin (red), x axis (gold), and xz plane (silver) are shown as large spheres. The GFP transition dipole (red) and its projection on the image plane (yellow–orange) are shown as cylinders with cones at each end. A spherical coordinate radial marker r (red arrow) is used to compare integrin orientations between the reference state with θ = 0° and ϕ = 90° b and integrin orientation with θ = 0° and ϕ = 45° that fits data well c. Inset shows relation between Cartesian coordinates and spherical coordinates with ϕ measured between the z axis and radial marker (r) and θ measured between the projection of r in the xy plane and the x axis. d, e The image plane with dipole positions of Rosetta ensemble members (lowest 40% in energy) projected from a spherical surface and shown as open gray circles for αL-T d and αL-F e. Projections are shown in the reference frame with θ = 0° and tilts in ϕ ranging from 11.25 to 67.5°. Silver, red, and gold circles show the same integrin reference atoms as in b & c. The calculated ensemble transition dipole is projected as a green line with length proportional to p. The transition dipole orientation determined by FluoPolScope is shown as black line with ± 1 s.d. shown as dashed lines. f Schematic showing integrin and GFP dipole orientation relative to tensile force between the actin cytoskeleton and ICAM (arrows) in migrating cells. g Model of cytoskeletal force acting on an integrin drawn to scale tilted at ϕ = 45°. Structures^,–,,– were assembled and rotated at domain–domain junctions known to be flexible and depicted with PyMol