Introduction
Regulating adhesion and de-adhesion between cells and their surrounding extracellular matrix (ECM) is critical for cell migration, for example during embryonic development, immune responses, cardiovascular function, angiogenesis, and tumour invasion and metastasis. Moreover, dynamic regulation of focal adhesions and other cell–ECM adhesion types is tightly coordinated with adhesion-dependent intracellular signalling, which in turn regulates survival and cell proliferation1, 2, 3. The integrin transmembrane ECM receptors mediate and control cell–ECM adhesion. Integrin heterodimers themselves are tightly regulated by several different mechanisms, including expression and subunit heterodimerization patterns, clustering and lateral diffusion in the plane of the plasma membrane, as well as by membrane trafficking and interaction with the actin cytoskeleton and in the inside of cells4. Integrin 'activation' is promoted by so-called inside-out signalling that increases the affinity for ECM ligands. Talin is a key participant in integrin activation, essentially acting as an intracellular ligand; the interaction of talin with integrin cytoplasmic regions causes conformational changes within the extracellular domains, which increases binding affinities for ECM ligands at the cell surface5, 6, 7. In addition, talin functions as a molecular bridge to link the cytoplasmic domains of integrins with the actin cytoskeleton, enabling the cell to exert the force on the ECM that is required for cell traction and movement across the substratum8, 9.
Despite decades of work on integrin regulation and the actin cytoskeleton, and on cell–ECM adhesion dynamics, some of the early events during cell spreading and adhesion, and also the relationship between matrix-bound integrin, mechanical force generation and downstream signalling, have remained mysterious. Previous genetic knockout studies implied some role for talin1 in cell adhesion and the organization of the actin cytoskeleton. However, the interpretation of these experiments was complicated by increased expression of the related talin2, a consequence of talin1 loss10. On page 1062 in this issue, Zhang and colleagues11 suppress talin2 expression (using short interfering RNA) in talin1-null cells, thus generating cells that were effectively talin-deficient. Molecular and pharmacological intervention, coupled with relevant imaging techniques, has enabled the authors to identify three temporally distinct phases during cell spreading that vary in their requirement for talin(s) (Fig. 1).
Figure 1: The effects of talin depletion on distinct phases of cell spreading are depicted.
(a) In cells lacking either talin1 or talin2, there is no activation of 
1 integrin, yet Src-dependent early spreading extensions are still evident. However, these do not persist, and the cells round up within 15–20 min after plating. (b) Re-expressing the isolated talin1 head, which contains the major 1-integrin-interacting FERM domain, promotes the activation of 1 integrin at the membrane. Although this leads to persistent spreading edges, the maturation of adhesion and progression to the fully elongated, contractile morphology are blocked. (c) Re-expression of full-length talin slows the rearward actin flow and restores the correct positioning of the actin and microtubule cytoskeletal networks, intracellular vesicles and peripheral phosphorylated myosin light chain. This is linked to adhesion maturation and the elongated fully spread morphology depicted. Restoration of signalling to FAK requires the tension generated between matrix-bound integrins and the actomyosin cytoskeleton, which is a function of the whole talin molecule. The image shows a fully spread cell (containing endogenous talin1 and talin2), with tensile actin filaments (revealed by phalloidin staining in red) tethered into focal adhesions containing FAK (revealed by anti-FAK staining in green; image provided courtesy of Emma Sandilands).
A striking and surprising observation made by the authors was that the initial cell-spreading reaction, including primary adhesion and early cell-edge extension, is unaffected by a loss of talin. This implies that during the initial stages of spreading, actin rearrangements induced by integrin engagement are not mediated by talin, and these processes do not seem to require a detectable contribution from activated 1 integrins. Zhang and colleagues found that early activation of Src-family kinases is not compromised by talin deficiency, showing that talin is dispensable for integrin-induced signalling to Src and for the Src-mediated events during the initial phase of cell spreading. Much previous work has described a key role for Src-family kinases in outside-in signalling that regulates integrin-induced actin reorganization12. However, Zhang et al. noted that, without talin, the further reorganization of actin filaments and formation of focal adhesions is severely disrupted. Consequently, the ventral membrane is poorly attached to the substrate (as elegantly demonstrated by a loss of total internal reflection fluorescence signal from green fluorescent protein (GFP) behind early edge protrusions), and cells without talin round up within 15–20 min of plating.
Interesting questions are raised by these data, particularly regarding the function of integrins during the initial talin-independent phase of cell spreading. Although the authors found a requirement for integrins during the initial phase of cell spreading by using blocking peptides, it seems that this can occur without talin-driven integrin activation. There are no observable activated 1 integrin heterodimers at the spreading membrane. One possible interpretation of this curiosity is that the initial spreading reaction is mediated by heterodimers that do not contain 1 integrin chains, and that inside-out activation of these, at least during the initial events of spreading, is independent of talin. However, studies have indicated that 3-containing integrins are also activated by talin6, 13. It would be interesting to examine the talin dependence of the activation of 
v (and other) integrins and to compare the distribution with that of 51, using this fascinating system for monitoring the initial Src-dependent events during talin-independent early edge protrusion.
Talin consists of an amino-terminal 'head' domain linked to a carboxy-terminal 'rod' region, and these are organized as antiparallel dimers. The FERM (four-point-one, ezrin, radixin, moesin) domain within the talin head forms the primary molecular contact with the -integrin cytoplasmic tail, although there may also be some contribution from the rod domain8, 14. There is now a wealth of structural and other data describing the fine details of this protein–protein interaction and how this leads to inside-out activation of the ligand-binding capacity of integrins9, 14. The actin-binding site of talin is within the carboxy-terminal rod domain, thus providing a molecular link between integrins and actin filaments8, 14. By expressing the isolated head domain in talin-deficient cells, Zhang et al. studied the consequences of talin-driven integrin activation in the absence of any talin-mediated link between integrins and actin filaments. They found that re-expression of the talin1 head was sufficient to promote widespread activation of 1 integrins at the spreading edge, and this enabled cells to remain spread for up to 90 min. However, absence of the talin rod domain caused cells to retain their intermediate 'fried-egg' shape, and they did not convert to a fully elongated/contractile morphology (Fig. 1). Integrin-induced phosphorylation of focal adhesion kinase (FAK) was also impaired.
Focal adhesion formation is normally associated with slower actin-rearward flow, dependent on myosin ll-mediated contractility, behind leading membrane extensions15. By visualizing GFP–-actinin in real time with various microscopy approaches, they found that talin deficiency (or re-expression of only the talin head domain) promotes persistent rapid rearward flow of actin that is dependent on myosin ll. Moreover, the myosin inhibitor blebbistatin also suppressed the phosphorylation of FAK-Tyr 397, suggesting that the talin-dependent formation of focal adhesions provides the 'anchoring' sites for mechanical force, and traction, that leads to full FAK activation. In the absence of talin, the links are weak and traction is insufficient. In addition to slowing down actin-rearward flow, inhibition of myosin ll restores the gradual peripheral movement of vesicles and also of microtubules, which are retained in the perinuclear region of the cytoplasm as a result of the rapid rearward flow of peripheral actin. Phosphorylated myosin light chain, which normally lies along peripheral actin arcs in spreading cells, is retained within the perinuclear region of talin-deficient cells. Taken together, the results make a compelling case that a talin–actin linkage is responsible for slowing the rearward actin flow during focal adhesion assembly, and hence it is likely that the role of talin is to couple integrin–ECM links to actomyosin contractility required for adhesion consolidation and subsequent maturation. Moreover, these contractility-associated events are necessary for signalling from integrins to FAK and thus for the cell–ECM adhesion to communicate with integrin-dependent pathways that promote cell proliferation and suppress apoptosis.
Use of the doubly talin-deficient cells by Zhang and colleagues thus allowed greater dissection of the temporal stages of spreading and has exploited talin dependence to define several clearly separable cellular events (Fig. 1). We now also have a fuller understanding of the importance of different talin functions. Previous work had implicated two functionally distinct actin networks that drive protrusion in migrating cells, one at the very leading edge and the other involving actomyosin contractility16 and a hierarchical transmission of actin movement through focal adhesions17. Presumably the leading-edge lamellipodia that were only poorly coupled to the contractile cytoskeleton in the earlier study16 were Src-dependent. These studies had proposed the existence of a 'molecular clutch' that links the contractile actin cytoskeleton with integrin-bound matrix proteins, so controlling actin and adhesion dynamics, and localized contractility, in motile cells16, 17. Now Zhang et al. propose that talin is a good candidate for such a clutch. Indeed, talin fulfils the required criteria of binding to both matrix-bound integrins and the contractile actin cytoskeleton. Moreover, it provides the mechanical link that enables force to be generated on ECM, through integrins, while promoting the maturation of assembling adhesive structures, and enabling tension-induced signalling into the cell interior.
