Ligand binding initiates single-molecule integrin conformational activation

Abstract

Integrins link the extracellular environment to the actin cytoskeleton in cell migration and adhesiveness. Rapid coordination between events outside and inside the cell is essential. Single-molecule fluorescence dynamics show that ligand binding to the bent-closed integrin conformation, which predominates on cell surfaces, is followed within milliseconds by two concerted changes, leg extension and headpiece opening, to give the high-affinity integrin conformation. The extended-closed integrin conformation is not an intermediate but can be directly accessed from the extended-open conformation and provides a pathway for ligand dissociation. In contrast to ligand, talin, which links the integrin β-subunit cytoplasmic domain to the actin cytoskeleton, modestly stabilizes but does not induce extension or opening. Integrin activation is thus initiated by outside-in signaling and followed by inside-out signaling. Our results further imply that talin binding is insufficient for inside-out integrin activation and that tensile force transmission through the ligand-integrin-talin-actin cytoskeleton complex is required.

Keywords: FRET; inside-out signaling; integrin activation pathway; integrin conformational changes; integrin conformational dynamics; ligand; outside-in signaling; single molecule; talin.

Figure 1.. Integrin conformational states and their FRET reporters.

(A) Overall integrin conformational states. (B) Intrinsic ligand binding kinetics and affinity of each state in integrin α5β1^,,. (C) Populations of different forms of α5β1^, and comparison to purified full-length α5β1 in LMNG/CHS calculated from data in Figure 4A. (D) Schematic of full-length α5β1 FRET reporters. Ectodomain reporters are identical except for truncation prior to the transmembrane domains (dashed lines). (E) Conformation-specific Fabs used here. (F) Affinity of TCO*-Lys and Cy3-tetrazine-labeled mutants for cyclic RGD peptide (cRGD, ACRGDGWCG) on transfected Expi293 β1 knockout cells from data in Figure S2E–G.

Figure 2.. Conformational transition rates in the α5β1 ectodomain Closed↔Open reporter.

(A, D, H) Representative time traces with fluorescence intensity (top) and FRET efficiency (bottom) with hidden Markov model fits (black) to data (cyan). (B) Transition density plot under basal conditions (1374 state-to-state transitions from 168 single molecule traces). (C) Single-molecule FRET efficiency histograms from the number of molecules shown in Table S1. (D and F) Dwell time distributions of FRET states with single-exponential fits. The lifetimes here and elsewhere denote the inverse of the rate of transitions for the associated arrows, and would be the average lifetime of the originating state when no other reaction from the state is allowed. (G and J) Scheme for measuring the dwell time of the EO•L state in the absence (G) or presence (J) of extension-stabilizing 9EG7 Fab. (I and K) EO•L transition rates as a function of mAb13 Fab concentration (FigureS3 E–H) with dose-response fit to mAb13 Fab EC₅₀ and maximum transition rate. The concentration of mAb13 Fab was as high as possible considering its solubility and occupation of 60% of the imaging volume by oxygen scavenging buffer. Errors are standard errors from nonlinear least square fits; lifetimes calculated using the thermodynamic cycle (Figure S3I) show errors from error propagation. Conformational transition rates were determined using at least two independent samples and imaging chambers. All transitions, from the number of single molecule traces shown in Table S1, are included in results.

Figure 3.. Conformational transitions in the α5β1 ectodomain Bent↔Extended reporter and coupling to ligand binding.

(A) Representative time trace. (B) Transition density plot under basal conditions. (C) Single-molecule FRET efficiency histograms. (D) Dwell time distributions of FRET states with single-exponential fits. Errors from nonlinear least square fits are as detailed in Figure 2 legend. (E) Representative time traces showing molecules trapped in BC (left panel) and EC (right panel) states. (F) The sequence and rates of conformational transitions in unliganded and liganded α5β1 ectodomain. (G-I) 3-color experiments on the α5β1 ectodomain Closed↔Open reporter with 10 nM Alexa488-Fn3_9–10. (G) Example time trace of Alexa488-Fn3_9–10 (top panel) and FRET reporter (middle and bottom panels). Alexa488, Cy3, and Cy5 were sequentially illuminated (200 msec/frame). (H) Dwell times collected for the EO•L state (mean ± S.E., 10 single molecules from 24 movies, 8 independent measurements). (I) FRET change upon Alexa488-Fn3_9–10 binding. Twenty traces from 15 movies (5 independent experiments) were aligned to the first frame of EO•L state (50 msec/frame).

Figure 4.. Affinities and conformational transition rates of full-length α5β1 in LMNG/CHS.

(A) Upper: Affinity of wild type full-length α5β1 for FITC-cRGD in specific conformational states measured using bulk fluorescence polarization (FP) assays in the presence of Fabs as indicated. Errors in affinity values are SE from nonlinear least square fits except value with “*” is difference from the mean from two experiments on different days and value with “^#” is global fit from three different datasets with different FITC-cRGD concentrations accounting for the effect of high-affinity binding on ligand depletion (Figure S4C). Lower: Representative data; errors in datapoints are standard deviations of triplicates. (B, C) FRET histograms. (D, E) Time traces showing rare transitions. (F) Representative time trace. (G) Dwell time distributions of FRET states with single-exponential fits. Errors are standard errors from nonlinear least square fits, as detailed in Figure 2 legend.

Figure 5.. Conformational transition rates of liganded α5β1 full-length in LMNG/CHS.

(A, C) Representative time traces. (B, D) Transition density plots. (E) Low to high FRET transition rates of FRET reporters at different Fn3_9–10 concentrations do not show concentration dependence as expected because concentrations are far above the K_d for the EO state (Figure 1B) and are fitted with horizontal lines. (F) High to low FRET transition rates of FRET reporters as a function of Fn3_9–10 concentration with dose-response curve fits (maximum rate and EC₅₀ of Fn3_9–10 concentration, with shared EC₅₀). (G) EO•L to EC•L transition rate as a function of mAb13 Fab concentration (FigureS4 H–J) with a dose-response fit (mAb13 Fab EC₅₀ and maximum transition rate). (H) The sequence of ligand induced conformational transitions in full-length α5β1. Errors are as described in Figure 2 legend.

Figure 6.. Effect of talin on integrin activation.

(A) Effect of talin1 head compared to conformation-specific Fab on full-length α5β1 binding to FITC-cRGD in FP assays. Saturating FP value and its error (SE) are from dose-response curve fit of triplicates. B-F, Effect of talin1 head on full-length Closed↔Open reporter. (B) FRET efficiency histograms. (C, E) Dwell time distributions of the low and high FRET states with single-exponential fits. (D) Dependence of the lifetimes of the low and high FRET states on the concentration of talin1 head in the presence of 1μM Fn3_9–10. Lifetime of the low FRET state at each talin1 head concentration (Figure S5D) was fitted to dose-response curve to determine the lifetime in the absence of and at saturating concentration of talin1 head. (F) Lifetimes of the low and high FRET states in various conditions. p-values are from unpaired two-tailed Student’s t test (n= 3 to 5 independent experiments). ns: p >0.05. (G) Population of each state in full-length α5β1. Data is calculated from that in Figure 4A, the thermodynamic cycle (Figure S3I, and the 1.35 fold increase of the EO•L state lifetime with saturating talin 1 head (Panels C-F). Errors are propagated from errors in binding affinities. H-J, Effect of talin1 head transfection on integrin monomeric ligand affinity in intact cells. (H, I) Affinity of α5β1 for cRGD (H) and α4β1 for LDVP (I) in transfected Expi293 cells measured by enhanced binding of 1nM AF647–12G10 Fab. (J) Affinity of αIIbβ3 for eptifibatide on CHO αIIbβ3 stable transfectants measured by enhanced binding of 5nM AF647-MBC319.4 Fab. Mean fluorescence intensity (MFI) of AF647 in EGFP⁺ cells was fitted to a dose-response curve with three parameters: background MFI, maximum MFI, and EC₅₀. EC₅₀ value is identical to K_d. Errors in A-E are SE from nonlinear least square fits, as detailed in Figure 2 legend; errors in H-J are difference from the mean of two independent measurements.

Figure 7.. The pathways for integrin conformational change in the integrin α5β1 machine and the structural requirement for concerted integrin headpiece opening and extension at the knees.

(A) The pathway of α5β1 activation. Lifetimes (timescales) for ligand binding and dissociation and conformational change are shown at 22°C. The transition like state (TS) is described below in Panel C. (B) In bent-closed α5β1 (PDB code 7nxd), interlock at the α and β-subunit knees prevents extension but not headpiece opening. (C) Movement of 1/3 along the trajectory toward open α5β1 separates the knees, but also breaks all other α and β-subunit leg contacts, destabilizing the bent conformation. Therefore, 1/3 opening and 1/3 extension are shown to simulate a transition-like state. (D) Extended-open α5β1 is modeled using open, ligand-bound α5β1 (PDB code 7nxd), with its disordered lower legs replaced with the legs from the bent-closed structure (Methods). (E-I) Leg structures of bent-closed integrins α5β1 (PDB code 7nxd, E), αVβ3 (PDB code 4g1m, F), αIIbβ3 (PDB code 3fcs, G), αxβ2 (PDB code 5es4, H), and αMβ2 (PDB code 7usm, I).