Report

Talin Binding to Integrin ß Tails: A Final Common Step in Integrin Activation

See allHide authors and affiliations

Science  03 Oct 2003:
Vol. 302, Issue 5642, pp. 103-106
DOI: 10.1126/science.1086652

Abstract

Control of integrin affinity for ligands (integrin activation) is essential for normal cell adhesion, migration, and assembly of an extracellular matrix. Integrin activation is usually mediated through the integrin β subunit cytoplasmic tail and can be regulated by many different biochemical signaling pathways. We report that specific binding of the cytoskeletal protein talin to integrin β subunit cytoplasmic tails leads to the conformational rearrangements of integrin extracellular domains that increase their affinity. Thus, regulated binding of talin to integrin β tails is a final common element of cellular signaling cascades that control integrin activation.

Cellular control of integrin activation is essential for normal development because it controls cell adhesion, migration, and assembly of an extracellular matrix (1, 2). Integrin activation can be controlled by signaling pathways that are thought to act by regulating specific interactions between cytoplasmic proteins and the integrin α- or β-subunit cytoplasmic tail (3). However, the critical integrin tail-binding protein(s) required for integrin activation have not been identified.

Talin is a major cytoskeletal protein that colocalizes with activated integrins and binds to integrin β cytoplasmic domains, and overexpression of talin's N-terminus activates integrins (4). To establish whether talin is required for integrin activation, we used small interfering RNAs to knock down talin expression in cultured cells (5). A Chinese hamster ovary (CHO) cell line stably expressing integrin αIIbβ3 (6) was transfected with plasmids encoding short hairpin RNAs (shRNAs) (7) targeting talin-1. At various times after transfection, talin expression was assessed by intracellular flow cytometric analysis with an antibody that recognizes talin-1 and talin-2. Talin knockdown (Fig. 1A) was directly related to shRNA levels, as monitored by the expression of cotransfected green fluorescent protein (GFP). Knockdown was maximal (55 to 70%) 72 hours after transfection, and this time point was used in all further experiments. The residual talin signal may represent incomplete talin-1 knockdown or talin-2 expression. Talin shRNA did not reduce surface expression of integrin αIIbβ3 (Fig. 1B; fig. S1), αVβ3 (Fig. 2C), or α5β1 (Fig. 2D), nor the expression of the other focal adhesion proteins vinculin, filamin, α-actinin, or RACK1 (fig. S1). Furthermore, talin knockdown was not observed in cells transfected with control or integrin β3 shRNA plasmids, and β3 shRNA reduced αIIbβ3, but not talin, expression (Fig. 1, A and B).

Fig. 1.

Knockdown of talin inhibits integrin activation. Sequence-specific and dose-dependent inhibitory effect of shRNA. αIIbβ3-expressing CHO cells were transfected with shRNA and GFP. GFP expression and intracellular expression of talin (A) or cell surface expression of αIIbβ3 (B) were determined by flow cytometry 72 hours later. Dot plots represent the degree of transfection (vertical) and protein expression (horizontal). Bar charts represent specific antibody binding to highly transfected cells (GFP fluorescence > 600) normalized to control shRNA-transfected cells. (C) Effect of shRNA on αIIbβ3 activation. Binding of the ligand-mimetic αIIbβ3 mAb, PAC-1, to αIIbβ3-expressing CHO cells transfected with shRNA was assessed, in the absence or presence of the activating antibody, anti-LIBS6. All data are the mean ± SE; n = 3.

Fig. 2.

Knockdown of talin inhibits cellular activation of multiple integrins. CHO cells were transfected with plasmids encoding shRNA, and wild-type or activated αIIb and β3. The specific binding of PAC-1 (A) or PAC-1 Fab (B) was then determined by flow cytometry. (C) Talin modulates the affinity of αVβ3. αVβ3-CHO cells were transfected with shRNA-expressing vectors, and binding of the activation-specific anti-αVβ3 Fab, WOW1 (left), and intracellular talin or cell surface αVβ3 expression (right) was determined by flow cytometry. (D), Talin modulates activation of β1 integrins. NIH-3T3 cells were transfected with GFP and shRNA-expressing plasmids. Specific binding of the α5β1 ligand FN9-11, talin expression, and β1 expression on GFP-positive cells was assessed. All data are the mean ± SE; n = 3.

Binding of the αIIbβ3 monoclonal antibody (mAb) PAC-1, a ligand mimetic, provides a measure of integrin activation (5, 6). Talin knockdown reduced PAC-1 binding (Fig. 1C). However, an αIIbβ3-activating antibody, anti-LIBS6 (8), increased PAC-1 binding to similar levels in both control and talin knockdown cells (Fig. 1C), showing that even in the absence of talin, integrins can be activated by an antibody acting on the extracellular domain. β3 knockdown reduced PAC-1 binding in both the absence and presence of anti-LIBS6 (Fig. 1C). Thus, knockdown of talin inhibits integrin αIIbβ3 activation without altering integrin expression.

As in resting platelets, αIIbβ3 expressed in CHO cells is normally in a low activation state and thus binds only small amounts of PAC-1 (6). To test the role of talin in supporting high levels of integrin activation, we assessed the effect of talin knockdown on activated variants of αIIbβ3. Substitution of Asp723 in the β3 cytoplasmic tail with Arg (αIIbβ3D723R) or substitution of the αIIb tail with that of α5 (αIIbα5β3) results in a metabolic energy-dependent activation of αIIbβ3 (9). When expressed in CHO cells, these activated integrins bound higher levels of PAC-1 than did wild-type αIIbβ3; nonetheless, talin knockdown still reduced PAC-1 binding to these integrins (Fig. 2A). Again, anti-LIBS6 increased PAC-1 binding to similar levels in control and talin knockdown cells (Fig. 2A). Another constitutively activated variant, which does not require metabolic energy for high-affinity ligand binding, replaces the αIIb cytoplasmic tail with a membrane-proximal deletion mutant of the αL tail (αIIbαLΔβ3) (10). Talin knockdown had no effect on PAC-1 binding to this variant (Fig. 2A). Thus, knockdown of talin expression reduced energy-dependent integrin activation but had no effect on energy-independent activation. Talin is therefore required for normal cellular activation of integrins.

Monovalent PAC-1 Fab fragments bind specifically to the active conformation of αIIbβ3, and binding is insensitive to the effects of integrin clustering (11). Talin knockdown inhibited PAC-1 Fab binding to αIIbβ3, αIIbα5β3, and αIIbβ3D723R (Fig. 2B; fig. S2). Thus, reduction of talin expression inhibits the conformational rearrangements that increase integrin affinity.

Because integrin αVβ3 also undergoes affinity regulation (12), we assessed the role of talin in its activation. Binding of WOW-1, a ligand-mimetic anti-αVβ3 Fab (12), was reduced by talin knockdown (Fig. 2C). Talin knockdown did not affect αVβ3 expression levels, and direct exogenous activation with MnCl2 induced similar levels of WOW-1 binding to talin knockdown and control cells (Fig. 2C). Thus, talin is also required for activation of αVβ3 integrins.

To assess the generality of the talin requirement for integrin activation, we tested the effect of talin knockdown on β1 integrins. Talin knockdown reduced binding of both PAC-1 (13) and PAC-1 Fab (Fig. 2B) to αIIbα5β3β1A, a high-affinity, energy-dependent, integrin chimera in which the cytoplasmic domains of αIIbβ3 were replaced with those of α5β1A. Furthermore, talin knockdown in NIH 3T3 fibroblasts (Fig. 2D) and CHO cells (13) inhibited binding of a fibronectin fragment (FN9-11) (5) to endogenous α5β1 integrins without reducing cell surface β1 levels. The β1-activating antibody, 9EG7, stimulated FN9-11 binding in both control and talin knockdown cells (13). Therefore, talin is generally required for cellular activation of endogenously and exogenously expressed integrins.

The requirement for talin suggests that it may be a downstream target of cellular signaling pathways that activate integrins. In platelets and their precursors, megakaryocytes, intracellular signals initiated by adenosine 5′-diphosphate (ADP) or protease-activated receptor 4 (PAR4) agonists, acting via distinct heterotrimeric GTP-binding proteins (G-proteins), result in αIIbβ3 activation (14). Expression of talin shRNA, but not control shRNA, blocked ADP or PAR4 agonist-stimulated fibrinogen binding to embryonic stem cell–derived megakaryocytes (Fig. 3A). Talin knockdown also inhibited αIIbβ3 activation stimulated by expression of β3-endonexin (15) (Fig. 3B). Furthermore, neither expression of activated R-Ras nor CD98 heavy chain reversed talin shRNA-mediated suppression of integrin activation (Fig. 3C), despite their abilities to reverse H-Ras–, Raf–, or free integrin β tail–mediated suppression (16, 17). However, expression of an integrin-activating fragment of talin (18) did reverse the effect of talin knockdown (13). Hence, talin is required for integrin activation downstream of a number of physiologically relevant signaling pathways, and expression of several other integrin regulators cannot compensate for loss of talin.

Fig. 3.

Integrin activation by multiple agonists and signaling pathways requires talin. (A) GFP expression and fibrinogen binding to murine embryonic stem cell–derived megakaryocytes infected with virus encoding shRNA and GFP were assessed by flow cytometry following stimulation, as indicated. Addition of EDTA was used to estimate nonspecific binding. The mean fluorescence intensity of fibrinogen binding to GFP-positive cells is indicated. Specific PAC-1 binding to αIIbβ3 in CHO cells transfected with plasmids encoding shRNA in combination with either GFP-β3 endonexin or GFP (B) or R-Ras(G38V) or CD98 (C) was assessed by flow cytometry. All data are the mean ± SE; n = 3.

A major integrin β cytoplasmic tail-binding site is localized to a phosphotyrosine-binding (PTB) domain–like subdomain within the talin FERM domain (18), and expression of this domain increases αIIbβ3 activation (4, 18). To determine if interaction between talin and integrin β tails is required for integrin activation in vivo, we identified mutations that disrupt this interaction and assayed their effects on activation.

Talin binds amino acids Trp739 to Lys748 of the integrin β3 tail by a variant of the canonical PTB domain–NPxY interaction (19) (fig. S3). Substitution of Tyr747 with Ala inhibited binding of talin and a number of other β tail–binding proteins (4, 20) (Fig. 4A). Mutations of Asp740 or Lys748 did not affect talin binding; however, alanine substitutions at positions Trp739 or Leu746 inhibited talin binding without affecting binding of filamin A or Syk (Fig. 4, A and B). Furthermore, Coomassie blue staining of all proteins bound to the β tails suggested that talin was the only protein whose binding was modified by these mutations (fig. S3).

Fig. 4.

Integrin mutations that inhibit talin binding prevent integrin activation. (A) Affinity chromatography using wild-type and mutant β3 model proteins was performed with platelet lysates. Binding of talin, Syk, and filaminA (FLNa) was assessed by Western blotting. Loading of each model tail protein was judged by Coomassie blue staining. (B) Talin binding was quantified by densitometry and normalized for binding to wild-type β3 (mean ± SE; n = 5). (C) CHO cells were transfected with plasmids encoding αIIbα5 and wild-type or mutant β3, as indicated. Integrin expression and PAC-1 binding were assessed by flow cytometry in the absence (top row) or presence (bottom row) of anti-LIBS6. Dot plots represent integrin expression (vertical) and PAC-1 binding (horizontal). (D) CHO cells were transfected with αIIbα5, αIIbα6A, αIIbα6B, or αIIb and wild-type or mutant β3, and the specific binding of PAC-1 to transfected cells was assessed (mean ± SE; n ≥ 3). (E) The specific binding of PAC-1 to CHO cells transfected with αIIbαLΔ and wild-type or mutant β3 was assessed (mean ± SE; n ≥ 3). (F) CHO cells expressing αIIbβ3 were transfected with wild-type or mutant talin F23 (amino acids 206 to 405), and specific PAC-1 binding to transfected cells was measured (mean ± SE; n = 5).

CHO cells were transfected with cDNA encoding either wild-type αIIb or an activating chimeric αIIb subunit along with a wild-type or mutant β3 subunit. αIIbα5β3 integrins containing mutations that abrogate talin binding [β3(L746A) or β3(W739A)] exhibited reduced PAC-1 binding (Fig. 4C). In contrast, point mutations in the β3 tail that did not interfere with talin binding did not affect PAC-1 binding. Anti-LIBS6 increased PAC-1 binding to all the mutant integrins, regardless of their ability to bind talin (Fig. 4C). Similar results were observed for αIIbα6Aβ3, αIIbα6Bβ3, and αIIbβ3 integrins (Fig. 4D). However, the energy-independent activated state of αIIbαLΔβ3 was not affected by β3 mutations (Fig. 4E). Thus, talin binding to integrin β tails is required for cellular activation of integrins.

We also generated point mutations within the talin PTB-like domain that reduce binding to β3 tails (R358A, W359A, and A360E) and one mutation that does not (K357A) (13, 19). When expressed in CHO cells, only the wild-type and K357A talin fragments stimulated PAC-1 binding to αIIbβ3, whereas talin fragments containing mutations that disrupt integrin binding did not (Fig. 4F). In all cases, anti-LIBS6 stimulated PAC-1 binding, confirming that αIIbβ3 was still expressed and could be activated externally (13). Thus, talin is required for integrin activation, and mutations within the talin PTB-like domain that prevent integrin β tail binding block its ability to induce integrin activation.

We have established that talin binding to integrin β tails is the final step leading to integrin activation. Regulation of this step, perhaps by phosphorylation (21), proteolysis (22), or phosphoinositide binding (23), may be a final common element in signaling pathways that control integrin activation. Although integrin residues critical for talin binding are localized within the W739 to K748 region, membrane-proximal β3 tail residues are also perturbed upon binding of activating fragments of talin (19, 24). Associations between the membrane-proximal regions of α and β tails are thought to prevent integrin activation (3, 9, 24, 25), and talin binding may induce integrin activation by disrupting their interaction (24). Deletion of the membrane-proximal regions generates integrins that remain activated in the absence of talin or metabolic energy (10). Therefore, talin-integrin interactions mediate integrin activation, possibly through effects on the membrane-proximal regions of integrin β tails.

Supporting Online Material

www.sciencemag.org/cgi/content/full/302/5642/103/DC1

Materials and Methods

Figs. S1 to S3

References

References and Notes

View Abstract

Navigate This Article