Microtubule-Actin Cross-talk at Focal Adhesions

See allHide authors and affiliations

Science's STKE  02 Jul 2002:
Vol. 2002, Issue 139, pp. pe31
DOI: 10.1126/stke.2002.139.pe31


Focal adhesions are dynamic structures in which traction forces are exerted against the substratum during cell migration and are sites for the organization of signaling complexes. Palazzo and Gundersen discuss how focal adhesions may also be the site of cross-talk between the actin-based and microtubule-based cytoskeletons. Microtubules appear to deliver factors that can regulate the formation and dissolution of focal adhesions, whereas focal adhesions contribute to microtubule localization and stability.

For many cell functions, the actin and microtubule cytoskeletons must be coordinated to generate a polarized cell response. One well-studied case is cell migration. The actin cytoskeleton provides the propulsive and contractile forces necessary to extend the cell front and move the cell body forward, respectively, whereas the microtubule cytoskeleton appears to regulate and coordinate these actin-based activities in many migrating cells. Three recent studies have identified candidate molecular mechanisms through which the microtubule cytoskeleton cross-talks with the actin cytoskeleton (1-3). These mechanisms appear to be directed at sites, known as focal adhesions, where the actin cytoskeleton is connected to the extracellular matrix through transmembrane integrin receptors and associated proteins. Focal adhesions are sites where traction forces are exerted against the substratum during cell migration; they also act as signaling centers to regulate the cell cycle and other processes. The recent studies point to a new role for focal adhesions as sites where the actin and microtubule cytoskeletons cross-talk.

The molecular mechanisms by which microtubules contribute to cell migration have been puzzling. Early experiments with colchicine and other microtubule-depolymerizing drugs demonstrated that an intact microtubule array is essential for directed migration of fibroblasts, endothelial cells, and other crawling cells (4, 5). In cells treated with microtubule inhibitors, actin-based lamellipodia and membrane ruffles became delocalized from the leading edge to the entire cell periphery [Fig. 1 and Movie 1], leading to the idea that microtubules limit the activity of the actin cytoskeleton to specific locations in the cell. It is not necessary to depolymerize microtubules to detect this microtubule role in regulating actin activity and cell migration. Low concentrations of microtubule drugs, which decrease microtubule dynamics without perturbing microtubule length or organization, inhibit lamellipodia formation and fibroblast migration (6, 7). Similarly, microinjection of antibodies that inhibit kinesin motor activity, but do not substantially perturb microtubules, inhibit cell protrusion and migration (8). These studies indicate that microtubule dynamics and motor activity contribute to regulating the actin cytoskeleton.

Fig. 1.

Microtubules are necessary for cell migration. NIH 3T3 fibroblasts are shown migrating into an in vitro wound. Approximately 4 hours after wounding, 10 μM nocodazole was added (+noc; panels 4 through 7) to break down microtubules. After another 4 hours, nocodazole was washed out to allow the regrowth of microtubules (-noc; panels 8 through 10). The nocodazole treatment can be seen to inhibit cell migration into the wound [compare cell movement before nocodazole treatment (panels 1 through 3) to cell movement during nocodazole treatment (panels 4 through 7)] without decreasing actin-based protrusive activity at the front of the cell, although the protrusive activity becomes delocalized. Nocodazole also blocks cells from completing mitosis, as indicated by the accumulation of rounded refractile cells. Both cell migration and cell division resume when the nocodazole is removed from the medium [for migration, compare cell movement during nocodazole treatment (panels 4 through 7) to cell movement after nocodazole washout (panels 8 through 10); for cell division, compare cells at the end of nocodazole treatment (panel 7) to cells after extended time in nocodazole free medium (panel 10)]. [See time-lapse video microscopy (Movie 1)]

Relaxing with Microtubules

In a series of studies, Small's group used direct imaging of living cells expressing fluorescently labeled tubulin to mark microtubules and fluorescently labeled vinculin or zyxin to mark focal adhesions, in order to explore the relationship between these structures. Dynamic microtubules were observed to grow toward focal adhesions, probably guided there by actin stress fibers (9). This "targeting" of microtubules to focal adhesions was temporally correlated with the dissolution or turnover of the targeted focal adhesion at the cell periphery (10). In one particularly compelling movie, a single focal adhesion in a closely spaced group of four was repeatedly targeted by multiple microtubules and then subsequently turned over, while neighboring focal adhesions persisted [figure 7B, Kaverina et al. (10)]. Kaverina et al. also found that such dissolution of peripheral focal adhesions led to cell edge retraction (10). In migrating cells, such a microtubule-dependent turnover of focal adhesion may be important to limit adherence in the front of the cell and to dissolve focal adhesions in the rear of the cell. Indeed, turnover of focal adhesions in the tail of migrating cells is dependent on microtubules (11).

How does microtubule targeting to a focal adhesion result in focal adhesion turnover? One mechanism, proposed by Small et al., is that microtubules deliver a "relaxing factor" that stimulates the turnover of targeted focal adhesions (10). Alternatively, microtubules may transport factors away from focal adhesions, stimulating focal adhesion disassembly. In their latest study, Small's group tested these possibilities by microinjecting function-blocking antibodies to the motor proteins dynein and conventional kinesin (type 1) into cells expressing fusions between green fluorescent protein (GFP) and vinculin or zyxin (1). They found that microinjection of an antibody against dynein did not affect the steady-state size or number of focal adhesion; however, microinjection of an antibody against kinesin caused enlargement of focal adhesions and reduced their numbers. The latter results phenocopy those observed in cells treated with microtubule-depolymerizing drugs, suggesting that kinesin participates in the microtubule-targeted turnover of focal adhesions. In support of this interpretation, they found similar enlargement of focal adhesions by expressing a kinesin mutant that formed rigor complexes on microtubules. They also found that kinesin inhibition did not interfere with microtubule dynamics or targeting to focal adhesions. These results are consistent with earlier observations that cells microinjected with inhibitory antibody to kinesin have reduced protrusion and morphologically resemble cells treated with nocodazole (8). Taken together, the studies from Small's group suggest that the plus-end-directed motor kinesin uses microtubules to locally deliver a relaxing factor to dissolve focal adhesions (Fig. 2A).

Fig. 2.

Possible mechanisms for microtubule-regulated cross-talk to focal adhesions. (A) Microtubules stimulate focal adhesion turnover by three potential mechanisms: (i) by providing a track for the delivery of a relaxing factor by conventional kinesin, (ii) by sequestering and thus inhibiting GEF-H1, and (iii) by providing a track for the kinesin Kif3 (also known as KapA) to deliver APC, which can then locally activate the Rac GEF ASEF. (B) The pathways by which these factors regulate focal adhesion turnover. It is unclear how the putative relaxing factor delivered by conventional kinesin affects focal adhesions. In contrast, ASEF would increase Rac-GTP levels, and this could affect focal adhesions by multiple mechanisms, including deceased Rho-GTP levels and increased activation of Pak kinases, which ultimately result in the inhibition of the myosin activity that is required to maintain focal adhesions. Analogously, local GEF-H1 inactivation could lead to decreased Rho-GTP levels and inhibition of myosin-based contraction.

Rho-ing with Microtubules

There may be other mechanisms by which microtubules affect focal adhesions. Focal adhesion formation is stimulated by active Rho bound to guanosine triphosphate (GTP), and nocodazole treatment increases the amount of Rho-GTP in cells (12). Thus, the enlargement of focal adhesions in the presence of nocodazole may not reflect decreased turnover of focal adhesions due to the lack of microtubule targeting per se, but increased formation of focal adhesions due to elevated Rho-GTP levels. Perhaps inhibition of kinesin, which phenocopies the nocodazole result, also leads to increased Rho-GTP levels? In support of this idea, kinesin-inhibited cells appear to have increased cell contractility [see movie 8 in the supplemental material of (1)], a hallmark of increased Rho activity. A single-cell assay for Rac activation has been developed (13), and when such an assay becomes available for Rho, it will be possible to determine whether inhibition of kinesin leads to Rho activation.

Given that nocodazole-induced breakdown of microtubules leads to global Rho activation, does it follow that in the presence of microtubules, Rho may be inactivated? Indeed, could the turnover of focal adhesions as the result of microtubule targeting be due to local inactivation of Rho? Perhaps, but any inactivation of Rho by microtubules would have to be localized to account for the highly selective turnover of focal adhesions observed in the earlier studies.

To date, factors that might serve to inactivate Rho, such as Rho-GTP-activating proteins or Rho-guanosine dissociation inhibitors, have not been localized on microtubules. Instead, a number of GTP exchange factors (GEFs), such as p190RhoGEF, Vav2, and Lfc (also known as GEF-H1), which activate Rho, have been localized on microtubules in vivo (2, 14-16). The Bokoch laboratory found that GEF-H1 binding to microtubules inhibits its Rho-GTP exchange activity. Forms of GEF-H1 that did not bind to microtubules were constitutively active in vivo, and morphological changes induced by nocodazole could be inhibited by dominant negative forms of GEF-H1, including a form that does not act by simply sequestering Rho (2). Krendel et al. were not able to show direct regulation of GEF-H1 by microtubules in vitro, suggesting that another factor may be involved. Nonetheless, these results provide the first evidence that the activity of a Rho GEF is regulated by association with microtubules.

Can GEF-H1 be another factor involved in regulating focal adhesion turnover? Might it even be the microtubule-dependent relaxing factor hypothesized by Small's group? If so, microtubule targeting to focal adhesions would have to result in the inhibition of GEF-H1, perhaps by sequestering and inhibiting GEF-H1 in the vicinity of the focal adhesion and thus locally decreasing Rho-GTP (Fig. 2, A and B). This possibility seems unlikely, given the highly localized turnover of focal adhesions by microtubule targeting. Also, kinesin data reported by Small's laboratory indicate that microtubules are delivering factors rather than taking factors away.

One way to reconcile these two studies is if kinesin were providing an activity that released active GEF-H1 from sites in the membrane, near focal adhesions. According to this idea, the factor delivered by kinesin would dislodge GEF-H1 adjacent to the targeted focal adhesion, so that GEF-H1 could bind and be inactivated by microtubules in the vicinity. To test this idea, it would be interesting to determine whether kinesin inhibition leads to a redistribution of GEF-H1 from microtubules to membrane or focal adhesion locations.

An alternative possibility is that there is no relationship between the putative relaxing factor and GEF-H1. Kinesin-based delivery of the relaxing factor along microtubules and inactivation of GEF-H1 by association with microtubules may be separate ways in which microtubules are used to regulate focal adhesions, and these could be employed in different parts of the cell. For example, the relaxing factor may function at the leading edge of the cell to selectively trim down the number of focal adhesions formed in a Rho-rich environment, whereas inactivation of GEF-H1 could function in the tail, where release of focal adhesions can be more indiscriminant.

We have described the association of GEF-H1 with microtubules as if it would be involved solely in the inactivation of GEF activity. It is also possible that the association of GEF-H1 with microtubules represents a way to deliver GEF-H1, and in doing so, to activate it and locally increase Rho levels. From what is currently known, this would depend on the ratio of microtubule polymerization to depolymerization, so if an area of the cell had net microtubule depolymerization, GEF-H1 would be released and Rho-GTP could be activated. At the leading edge where focal adhesions are forming, microtubules usually show net polymerization (to keep up with the extending leading edge), suggesting that GEF-H1 would tend to be inactivated near the front of the cell.

Rac-ing with Microtubules

Another signaling molecule regulated by microtubule dynamics is the small GTPase Rac. When microtubules are repolymerized after the microtubule-depolymerizing drug nocodazole has been washed out, Rac-GTP levels increase, suggesting that microtubule polymerization activates Rac (17). The increased levels of Rac-GTP may result in focal adhesion disassembly by decreasing myosin contractility, which is necessary for focal adhesion assembly (18). Active Rac can reduce myosin contractility through three mechanisms (Fig. 2B): (i) by activating the Rac effector Pak1, which phosphorylates myosin light-chain kinase to inhibit it (19); (ii) by activating a Pak-like kinase to phosphorylate the heavy chain of myosin (20); and (iii) indirectly, by decreasing Rho-GTP levels (21). How might microtubules affect Rac-GTP levels? Waterman-Storer et al. suggest that Rac-GTP, which binds tubulin but not microtubules (22), may be released upon tubulin polymerization (17). Although Rac-GTP may be released during microtubule polymerization, this model does not explain why there is a increase in total Rac-GTP levels, which is what is presumably being measured by the Rac-GTP pull-down assays. Indeed, the interaction of Rac with tubulin does not appear to stimulate or inhibit Rac-GTP formation.

An alternative possibility comes from studies of the adenomatous polyposis coli (APC) protein by Akiyama's group (3, 23). Their initial study showed that APC binds to the Rac GEF ASEF, and in doing so, stimulates its Rac exchange activity (23). In a second study, APC was shown to interact with the Kap3 subunit of the heterotrimeric kinesin Kif3 (3). APC bound to the heterotrimeric form of Kif3, and a dominant negative version of Kap3 prevented APC from accumulating in clusters near the cell periphery. Thus, it is very likely that Kif3 is the motor responsible for the movements of APC-containing particles on microtubules that were found in an earlier study with GFP-APC (24). Given these new results, an intriguing possibility is that microtubules deliver APC to activate ASEF at or near focal adhesions (Fig. 2A). The local activation of Rac would function to decrease focal adhesions through the mechanisms discussed above. It is important to note that APC is unlikely to be Small's relaxing factor, because Small used inhibitors of conventional kinesin that do not affect the activity of Kif3. Hence, Rac activation by APC through ASEF may represent yet another way in which microtubules cross-talk with the actin cytoskeleton at focal adhesions.

Focal Adhesions Actin' on Microtubules

Until now, we have been discussing how microtubules affect focal adhesions, but what about the reverse? As mentioned above, microtubules are targeted to focal adhesions, probably guided by actin stress fibers. Once targeted, microtubules are transiently stabilized for brief periods (~15 s) (9). These microtubules should not be confused with long-lived stable microtubules that have long half-lives (hours), do not end in focal adhesions (25), and are produced in response to Rho-GTP (26) and the Rho effector mDia (27). It turns out that focal adhesions also locally signal the production of these long-lived stable microtubules through the activation of focal adhesion kinase (FAK) (28). Active forms of mDia, but not active Rho, can stimulate stable microtubule production in the absence of FAK activity, indicating that FAK may be required for Rho-GTP to activate mDia.


The studies described above provide the first molecular clues about how microtubules may regulate the actin cytoskeleton. They emphasize the role of motor proteins in delivering localized signals that regulate the actin cytoskeleton through their effects on focal adhesions and Rho and Rac GTPases. It is not yet clear whether the relaxing factor hypothesized by Small's group works through Rho and Rac regulation, but its characteristics seem to suggest that it is distinct from GEF-H1or APC-ASEF. These studies also add microtubules to the growing list of ways in which Rho-family GEFs can be activated and controlled. The general paradigm that seems to be emerging is that microtubule growth inhibits contraction and promotes actin polymerization and, thus, cell spreading, whereas microtubule depolymerization activates contraction. These effects may be localized to a narrow region near the microtubule plus end. Future studies should reveal whether the different microtubule-based mechanisms for regulating Rho, Rac, and focal adhesions are segregated to different domains within cells.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
  28. 28.
  29. 29.
View Abstract

Navigate This Article