The mechanism of kindlin-mediated activation of integrin αIIbβ3

Abstract

Increased ligand binding to cellular integrins ("activation") plays important roles in processes such as development, cell migration, extracellular matrix assembly, tumor metastasis, hemostasis, and thrombosis. Integrin activation encompasses both increased integrin monomer affinity and increased receptor clustering and depends on integrin-talin interactions. Loss of kindlins results in reduced activation of integrins. Kindlins might promote talin binding to integrins through a cooperative mechanism; however, kindlins do not increase talin association with integrins. Here, we report that, unlike talin head domain (THD), kindlin-3 has little effect on the affinity of purified monomeric αIIbβ3, and it does not enhance activation by THD. Furthermore, studies with ligands of varying valency show that kindlins primarily increase cellular αIIbβ3 avidity rather than monomer affinity. In platelets or nucleated cells, loss of kindlins markedly reduces αIIbβ3 binding to multivalent but not monovalent ligands. Finally, silencing of kindlins reduces the clustering of ligand-occupied αIIbβ3 as revealed by total internal reflection fluorescence and electron microscopy. Thus, in contrast to talins, kindlins have little primary effect on integrin αIIbβ3 affinity for monovalent ligands and increase multivalent ligand binding by promoting the clustering of talin-activated integrins.

Figure 1

Kindlins selectively increase the binding of multivalent ligands to recombinant integrin αIIbβ3 cells but not to monomeric integrins in nanodiscs. (A) Activation of αIIbβ3 in nanodiscs was assayed by PAC1 binding as described in the supplemental information. Increase in integrin activation is calculated as (L_i-L_0-), where L_i is the Eptifibatide inhibitable PAC1 binding (luminescence) in the presence of THD(1 μM), kindlin-3(2.5 μM) or THD+kindlin-3, L₀ is PAC1 binding to integrin nanodiscs alone, which was 535,000 arbitrary units. The concentration of THD was that inducing half the maximal response observed at 5 μM THD. Error bars indicate ±SEM of three independent determinations. (B) HEK293 cells stably expressing αIIbβ3 were transfected with cDNA encoding kindlin-1, THD, or kindlin-1+THD. Twenty-four hr. after transfection, cells were harvested, split into 3 groups, and stained separately with PAC1, PAC1Fab, or D57 (anti-αIIbβ3). Cells were then fixed in 3.7% formaldehyde. The quantity of PAC1 (decavalent), PAC1Fab (monovalent) or D57 bound was assayed by APC-conjugated anti-IgM (μ chain specific) and anti-IgG, and analyzed by flow cytometry. The αIIbβ3-specific ligand binding was calculated as described in the supplemental information. Right panel shows the western blots confirming the abundance of expressed proteins in the transfected cells. Error bars indicate ± SEM of 3 independent experiments. PAC1 Fab was used at a concentration at least 25 fold higher than that used by Bunch [48]. (C) HEK293 cells stably expressing αIIbβ3 were transfected with cDNA encoding kindlin-1, THD, or kindlin-1+THD. The cells were processed as described in (B), and the binding of PAC1 (decavalent), D57, or 3FN10 (GST-removed, monovalent) bound was assayed. The αIIbβ3-specific ligand binding was calculated as described in the supplemental information. Error bars indicate ± SEM of 4 independent experiments. (D) HEK293 cells, stably expressing αIIbβ3, were transfected with cDNA encoding kindlin-1, THD, or kindlin-1+THD. The cells were harvested and then the binding of D57, biotinylated 3FN10 monomer (with GST removed), GST-3FN10, or oligomeric GST-3FN10 was assayed. Depicted is the αIIbβ3-specific ligand binding as described in supplemental information. Error bars indicate ± SEM of four independent experiments.

Figure 2

Depletion of kindlins primarily inhibits the binding of multivalent ligands to integrin αIIbβ3 in nucleated cells and in mouse platelets. (A) Cells expressing an active αIIbβ3(D723R) mutant were transduced with lentivirus expressing a GFP marker and either a scrambled shRNA (control) or kindlin-2 shRNA1. To confirm the specificity of previously characterized kindlin-2 knock-down constructs [17], shRNA1-transduced cells were transiently co-transfected with an shRNA-resistant human kindlin-2 and tdTomato as a transfection marker (at a ratio kindlin-2:tomato=50:1). The cells were then harvested and binding of anti-αIIbβ3 (D57), monomeric 3FN10 (with GST removed) or GST-3FN10 oligomers was used to measure the αIIbβ3-specific ligand binding as described in the supplemental information. Kindlin2 expression level was assessed by immunoblotting as shown in the right panel. The western blot was performed with total cells whereas the only transduced (GFP positive) and transfected (tomato positive) cells were analyzed for ligand binding by flow cytometry. (B) Generation of chimeric mice as described in the supplemental information. For intracellular flow cytometry, platelets were fixed with 2% formaldehyde, permeabilized and stained with rabbit anti-kindlin-3 polyclonal antibody or biotinylated anti-talin monoclonal antibody (clone 8d4). After washing, bound antibodies were detected with either FITC-anti-rabbit secondary antibody or APC-strepavidin respectively. The dot plots show the reduced expression of kindlin-3 and talin1 in DsRed negative Fermt3^−/−, and Tln1^−/− platelets respectively. (C) PAR4-stimulated binding of 3FN10 to platelets. Platelets from irradiated mice reconstituted with mixtures of DsRed-expressing wild-type (Wt) and either Fermt3^−/− or Tln1^−/− hematopoietic stem cells were incubated in the presence of the indicated concentration of PAR4 agonist peptide or buffer. Eptifibatide-inhibitable specific binding of 3FN10 (with GST removed) to αIIbβ3 on DsRed-negative Fermt3^−/− or Tln1^−/− or DsRed positive wild-type platelets was assessed by FACS. Cells were separately stained with anti-CD41 (αIIb) to calculate ligand binding as described in the supplemental information. 3FN10 at 100 μg/ml (9.4 μM) was used. (D) PAR4-stimulated binding of fibrinogen to platelets. In an experiment conducted on the same day as that described in (C), 3FN10 was replaced by FITC-conjugated fibrinogen. In both C and D, error bars indicate standard error of the mean (7 chimeric mice for each genotype).

Figure 3

Kindlins promote clustering of occupied integrins. (A) Cells expressing αIIb-GFPβ3 transduced with lentivirus encoding both kindlin-2 shRNA and DsRed were mixed with uninfected cells and plated on fibrinogen-coated cover slips. The left panels are TIRF images of the distribution of αIIb-GFPβ3 at the cell-substrate interface in uninfected cells (DsRed negative) and shRNA-transduced cells (DsRed positive, white arrows). DsRed epi-illuminated fluorescence images in the middle panel indicate shRNA lentiviral transduction. GFP epi-illuminated fluorescence images in the right panels indicate comparable αIIbβ3 fluorescence in uninfected control cells and kindlin-2 shRNA-expressing cells. Scale bar is 10 μm. (B) Fluorescence intensities of integrin puncta were measured and averaged as described in supplemental information. Error bars indicate standard error of n=1223, 1204 and 1008 puncta in uninfected cells or those transduced with kindlin-2 shRNA1 or shRNA2 respectively. (C) Cells from the experiment depicted in Panel (A) were stained with anti-αIIbβ3 (D57) and analyzed by FACS to assess αIIbβ3 surface expression. Data are expressed as percent of αIIbβ3 expressed in control shRNA-infected cells. (D) DsRed-positive shRNA-transduced cells were isolated by FACS and kindlin-2 expression was assessed by western blotting.

Figure 4

Loss of kindlin-2 reduces clustering of αIIbβ3 integrins. (A) EM images of colloidal gold-labeled αIIbβ3 in adherent cell ventral membrane. CHO cells stably expressing αIIbβ3 were transduced with lentivirus expressing control shRNA or kindlin-2 shRNA1, adhered to fibrinogen-coated EM grids, swollen in a hypotonic buffer and subjected to a flow of low salt buffer to remove the cell body. The remaining ventral membrane sheet was stained with an anti-β3 tail antibody followed by colloidal gold adsorbed 2° antibody. Left, middle and right columns are control shRNA cells, kindlin-2 knockdown cells, and irrelevant IgG-stained control shRNA cells at differing magnifications. Many clusters of colloidal gold containing >5 particles per cluster are present in control shRNA cells and such clusters are absent in kindlin-2 silenced cells. Red circles indicate examples of integrin clusters. Scale Bars are 200 nm for the images in top and middle row and 2 μm for the images in bottom row. The images in the middle and lower rows are ventral membranes at different magnifications. Images in the top row show enlarged views of selected areas from those in the middle row. (B) Cells in (A) were lysed and analyzed with anti-kindlin-2 or anti-GAPDH by western blotting.