The cytoskeletal proteins talin and vinculin form part of a macromolecular complex on the cytoplasmic face of integrin-mediated cellular junctions with the extracellular matrix. Recent genetic, biochemical and structural data show that talin is essential for the assembly of such junctions, whereas vinculin appears to be important in regulating adhesion dynamics and cell migration.
- focal adhesions
The integrin family of cell adhesion molecules are coupled to the actin cytoskeleton  with the exception of integrin α6β4, which is linked to the intermediate filament network in hemidesmosomes . The link between integrins and F-actin is mediated by a series of actin-binding proteins that include talin , filamin , α-actinin  and tensin ; integrin-linked kinase (ILK)  can fulfil a similar role, although here, the link is indirect, and is mediated by the actin-binding protein actopaxin. Although it is evident that a connection between integrins and the actomyosin contractile apparatus is important for cell migration, the physiological significance of the several pathways by which integrins can be coupled with F-actin remains to be fully explored. Each linker protein has a different structure, and may support different types of actin cytoarchitecture. For example, filamin is a Y-shaped parallel homodimer and leads to the assembly of orthogonal actin networks , whereas α-actinin is a short rod-shaped anti-parallel dimer that promotes actin bundling . Clearly, different types of actin organization may be required in different types of cellular junction. Each linker protein may also serve to recruit different proteins to the integrin/cytoskeletal interface some of which may be involved in stabilizing the structure, whereas others may mediate integrin signalling and the regulation of the dynamic properties of the adhesion complex. It has also become apparent that at least one of these linker proteins, talin, is important in integrin activation [10,11], and indeed talin will be the focus of this review.
The structure and function of talin
Talin (270 kDa, 2541 amino acids) is an elongated (approx. 60 nm) flexible anti-parallel dimer , which appears as a series of beads on a string under the electron microscope . Limited proteolysis and N-terminal sequencing shows that there are protease-sensitive regions between the talin head and rod domains, and also within the rod domain (Figure 1). The globular talin head contains a FERM domain (residues 86–400) with binding sites for β-integrin cytodomains , the cytodomain of layilin  (a C-type lectin which acts as a hyaluronan receptor), and for two signalling proteins, FAK  and the PIP kinase (type 1γ isoform of phosphoinositide 4,5-kinase) [16–18]. The FERM domain also contains an actin-binding site , and is flanked by membrane insertion sequences . The rod responsible for dimer formation  (Kd=0.26 μM ) contains a highly conserved C-terminal actin-binding site  similar to that in HipR and yeast Sla2p , a second integrin-binding site  and several binding sites for the cytoskeletal protein vinculin , which itself has multiple binding partners including F-actin . The talin head binds acidic phospholipids , and PtdIns(4,5)P2 has been shown to activate the integrin-binding sites in talin . Whether talin is regulated by an intramolecular head–tail interaction similar to other ERM proteins is unclear (Figure 1), although talin can adopt a globular conformation in low-salt buffers in a manner that depends on the talin head .
Evidence that talin is a key player in integrin-mediated adhesion comes from a variety of sources: (i) microinjection of talin antibodies  or an N-terminal talin polypeptide  into cells in culture disrupted cell–extracellular matrix junctions (focal adhesions); (ii) down-regulation of talin using antisense RNA inhibited focal adhesion assembly and also cell spreading ; interestingly, part of this effect is due to altered integrin processing in the Golgi, and talin is associated with β1 integrins in membrane fractions enriched in Golgi and endoplasmic reticulum ; (iii) disruption of the talin gene compromises cell spreading and focal adhesion assembly in ES cells , and talin1 (−/−) ‘fibroblasts’ derived from talin1 (−/−) ES cells cannot support the initial weak 2 pN bond between fibronectin–integrin complexes and actomyosin  or the assembly of focal adhesion-like structures in response to external force ; and (iv) gene knockout studies in Caenorhabditis elegans , Drosophila  and mouse  confirm that talin is essential for a variety of integrin-mediated developmental events. Immunocytochemical studies show that talin is localized in a wide variety of structures (Table 1). However, interpretation of these data is complicated by the discovery of a second talin gene that encodes a closely related protein (74% identity) recognized by several of the commonly used talin antibodies. Northern-blot analysis shows that talin2 has a more restricted pattern of expression than the ubiquitous talin1 .
Integrin signalling via FAK and Src promotes binding of talin to a PIP kinase [16,18]. Complex formation activates a PIP kinase and promotes translocation of talin to the plasma membrane, although the latter is independent of kinase activity. This suggests a model (Figure 2) in which talin is activated at the plasma membrane by the localized production of PtdIns(4,5)P2, exposing the integrin-binding site . Talin might then activate integrins [10,11] and also provide the link to the actin cytoskeleton. Since binding of the integrin cytodomain and PIP kinase to the talin F3 FERM subdomain is mutually exclusive , it is unclear whether the assembly of integrin–talin complexes requires the displacement of PIP kinase. Indeed PIP kinase is itself localized in focal adhesions [17,18]. Additional components might then be recruited to the complex, some of which may facilitate maturation into more stable focal adhesions, whereas others might be involved in the regulation of the dynamic properties of the complex and, therefore, cell motility.
Vinculin and focal adhesion dynamics
One of the best-characterized focal adhesion proteins is vinculin , which is also considered to be activated by PtdIns(4,5)P2 . However, the role of vinculin therein has proved difficult to establish, although it is clear that vinculin is not required as such for focal adhesion assembly, since vinculin null cells still assemble such structures . Interestingly, vinculin null mouse embryo fibroblasts are less well spread, have fewer and smaller focal adhesions and are more motile in both wound closure and Boyden chamber assays than wild-type cells. They also show increased FAK and paxillin tyrosine phosphorylation [40,41], and both FAK and paxillin signalling are known to be important in focal turnover. This suggests that vinculin negatively regulates cell motility. Moreover, vinculin null cells are resistant to apoptosis . These observations may explain why vinculin behaves as a tumour suppressor in model systems . Using real-time interference reflection microscopy, we have shown that focal adhesions in vinculin null mouse embryo fibroblasts turn over about twice as fast as those in wild-type cells (with M. Holt and G. Dunn, unpublished work). The phenotype is largely rescued by expression of the full-length vinculin cDNA, although it is difficult to achieve physiological levels of vinculin expression in the null cells, probably because vinculin resensitizes the cells to apoptosis.
Since vinculin is supposed to cycle between active and inactive states, we sought to establish the phenotype of cells expressing vinculin mutants in which this cycle was misregulated. The crystal structure of vinculin has recently provided a molecular explanation for the two conformations . In the closed state, the C-terminal Vt (vinculin tail) interacts with the N-terminal Vh (vinculin head), burying most of the ligand binding sites within vinculin. The Vt is comprised of a five-helix bundle stabilized by the C-terminal 14 amino acids that extend from the base of the bundle . Removing the C-terminal arm significantly reduces the ability of the Vt to bind PtdIns(4,5)P2, probably because small conformational changes in the VtΔC mutant (detected by NMR; with I. Barsukov, unpublished work) disrupt the basic collar, a cluster of basic residues that are distant in the primary sequence, and which are implicated in PtdIns(4,5)P2 binding . However, the Vh–Vt interaction (with A. Bobkov and R.C. Liddington, unpublished work), and binding of the Vt to F-actin is unaffected by the VtΔC mutation. Expression of a vinculinΔC mutant in vinculin null mouse embryo fibroblasts resulted in a marked defect in cell spreading, probably because the focal adhesions in these cells turn over much less rapidly than in cells expressing wild-type vinculin (with M. Holt, G. Dunn and R.M. Saunders, unpublished work). These results confirm that vinculin plays a key role in regulating focal adhesion dynamics, and implicate PtdIns(4,5)P2 in this process, although Vt can bind to other acidic phospholipids including PtdIns(3,4)P2 and PtdIns-(3,4,5)P3 .
How vinculin exerts these effects remains to be elucidated. It is interesting that the binding sites for F-actin  and inositol phospholipids  in Vt partially overlap, and binding of PtdIns(4,5)P2 and F-actin to Vt has been shown to be mutually exclusive . The results with the vinculinΔC mutant, which is deficient in inositol phospholipid binding, suggest a model in which vinculin is recruited to stabilize nascent integrin–talin complexes by cross-linking talin to F-actin. This interaction might be inhibited by transient increases in PtdIns(4,5)P2 or PIP3, allowing the complex to be remodelled. Interestingly, PDGF-induced remodelling of focal adhesions requires PI3′-kinase, and can be mimicked by adding PIP3, but not PtdIns(4,5)P2, to the cell culture medium . In the absence of this mode of regulation, the vinculinΔC mutant would suppress focal adhesion turnover and, hence, inhibit cell spreading and cell migration. Evaluation of this hypothesis requires the identification of point mutations in vinculin that ablate talin and actin binding, and a more detailed understanding of the specificity of Vt for individual phospholipids. The recent determination of the crystal structure of the N-terminal region of the Vh (Vh′ residues 1–258) complexed to talin peptides should help in this regard .
Alternative explanations for the role of vinculin as a negative regulator of cell motility include the idea that it sequesters paxillin, inhibiting its interaction with FAK and suppressing FAK and paxillin signalling . To test this model, we have attempted to map the paxillin-binding site(s) in vinculin. Paxillin-binding proteins such as FAK, p95PKL and actopaxin all contain short PBS (paxillin-binding sequence) motifs, the mutation of which disrupts paxillin binding . However, mutation of putative PBS sequences in the Vt (VTRL and LLQVCE), similar to those in FAK, was without effect. Interestingly, both the X-ray  and NMR structures  of FAK show that these PBS motifs are not directly involved in paxillin binding, raising uncertainty about their significance as determinants of paxillin binding. Vinculin has also been shown to bind to (i) VASP that reportedly antagonizes barbed-end capping  and (ii) the Arp2/Arp3 actin nucleation complex , although whether this relates to the ability of vinculin to stabilize focal adhesions has not been explored. Vinculin also binds to vinexin, which is implicated in ERK (extracellular-signal-regulated kinase)-mediated cell spreading .
The talin–vinculin interaction
To address the role of the talin–vinculin interaction in focal adhesions, we have previously mapped three VBSs (vinculin-binding sites) in talin [25,55] (residues 607–636, 853–878 and 1944–1969), each of which predicted to be an amphipathic α-helix. We have now determined the structure of the N-terminal region of the talin rod (residues 482–789) that contains VBS1 . The region comprises two domains that stack against each other via hydrophobic interactions. Domain 1 is a bundle of five amphipathic helices whereas domain 2 contains four helices. Studies on the isolated domain 1 (482–655) show that the VBS in helix 4 is cryptic, and studies with mutant VBS peptides identify six key hydrophobic residues involved in vinculin binding which are buried within, and also contribute to the stabilization of, the amphipathic five-helix bundle. Interestingly, the VBS can be activated either by removal of helix 5 or by introducing mutations in two of the six key residues. NMR studies show that binding of Vh′ to such ‘activated’ talin mutants results in a marked conformational change in the talin polypeptide. Indeed, when Vh′ was co-crystallized with the ‘activated’ talin four-helix bundle (residues 482–636), only the structure of Vh′ complexed to talin helix 4 was discernable in the electron density – the other helices appear to have been proteolysed during crystallization.
Notice that the topology of the talin 5-helix bundle (residues 482–655) and the Vh′/talin VBS helix 4 complex are remarkably similar. We propose a model in which the VBS helix is extracted from the talin five-helix bundle to form a new five-helix bundle with the N-terminal region of Vh′ which is itself composed of four helices. What happens to the other talin helices has not been established. One possibility is that the hydrophobic surfaces of the exposed helices associate with the lipid bilayer, and it is interesting in this regard that talin is able to associate with liposomes . How VBS1 is activated also remains to be determined although the idea that the tension exerted on an integrin–talin–actomyosin complex might play a role is attractive. Indeed, stretching has been shown to expose binding sites for a variety of focal adhesion proteins in Triton X-100 cytoskeletons . Another puzzle is what defines a VBS in talin, the rod domain of which is composed of approx. 60 amphipathic helices. Indeed, recent results suggest that there are more potential VBSs in the talin rod than previously recognized (with R. Frank, W.H. Ziegler and J. Emsley, unpublished work). Whether some are constitutively active whereas others are stretch-activated, or are never exposed, is the subject of ongoing research.
Regulation of talin by phosphorylation
A substantial body of literature indicates that PKC (protein kinase C) is involved in integrin-mediated cell adhesion and spreading, focal adhesion assembly and cell migration. However, the prolonged activation of PKC in a variety of adherent cells leads to the disruption of focal adhesions and actin-stress fibres, and an associated increase in talin serine/threonine phosphorylation [59–61]. A similar effect is elicited by interleukin-1β . In platelets, activation by thrombin leads to translocation of talin from the cytosol to the membrane and is associated with talin phosphorylation again on serine/threonine residues that are predominantly in the talin head . Talin has also been shown to be tyrosine-phosphorylated in L6 myoblasts treated with PDGF . Interestingly, although this caused the rapid loss of vinculin from focal adhesions and the disruption of actin stress fibres, talin remained localized in small focal complexes. Talin is also tyrosine-phosphorylated in chick cells expressing v-src , suggesting that it is probably a substrate for src-family of kinases, which regulate focal adhesion turnover . We have begun to map the PKC sites in talin in vitro using a series of recombinant talin fragments spanning the entire length of the protein (with B. Patel, unpublished work). Remarkably, only fragments containing the talin head were phosphorylated by PKC in vitro, a result that fits well with the in vivo observations. In contrast, we found no Src kinase sites in the intact head. Since the talin head contains binding sites for integrins, layilin, PIP kinase and FAK (Figure 1), it seems probable that phosphorylation will be one of the mechanisms that directly or indirectly regulates the interaction between talin and its various ligands, and thereby integrin-mediated adhesion.
In summary, recent studies have provided strong evidence that talin plays a key role in activating integrins and coupling them with the actin cytoskeleton. The major challenge now is to use the new structural and biochemical information on talin and vinculin to investigate the mode of action of these proteins within the cell.
The author is grateful to B. Patel, R.M. Saunders, L. Jennings, I. Barsukov, J. Emsley, A. Gingras, E. Papagrigoriou (University of Leicester), to M. Holt and G. Dunn (King's College London), to R. Frank (German Research Centre for Biotechnology, Braunschweig, Germany) and W.H. Ziegler (Technical University, Braunschweig, Germany) and to A. Bobkov, R.C. Liddington and E. Adamson (Burnham Institute, La Jolla, CA, U.S.A.) for their collaboration on these projects. This work was partially supported by the Wellcome Trust, BBSRC and CR-UK.
Signalling Outwards and Inwards: A Focus Topic at BioScience2004, held at SECC Glasgow, U.K., 18–22 July 2004. Edited by J. Challiss (Leicester, U.K.), A. Harwood (University College London, U.K.), M. Humphries (Manchester, U.K.), C. Isacke (Institute of Cancer Research, London, U.K.), R. Liddington (Burnham Institute, La Jolla, CA, U.S.A.), T. Palmer (Glasgow, U.K.), K. Siddle (Cambridge, U.K.), C. Sutherland (Dundee, U.K.), H. Wallace (Aberdeen, U.K.) and M. Welham (Bath, U.K.).
Abbreviations: PBS, paxillin-binding sequence; PIP kinase, the type 1γ isoform of phosphoinositide 4,5-kinase; PKC, protein kinase C; VBS, vinculin-binding site; Vh, vinculin head; Vt, vinculin tail
- © 2004 The Biochemical Society