Close encounters of the LIM-kinase


During cell movement, new actin structures are formed, under the orders of the small GTPase Rac. One of Rac’s orders is to inhibit the depolymerization of existing actin filaments. A chain of events from Rac to the inhibition of actin depolymerization has now been forged.


Acell that has been ordered to move faces a formidable task. It must define and establish a forward polarity, mobilize and completely reorganize its cytoskeleton, adjust the strength of its connections with the extracellular matrix, and fire up myosin motor proteins to power the journey. One of the early steps in this process, the formation of a lamellipodium (a thin, actin-rich network that stretches out from the cell body) at the ‘leading edge’ of the motile cell, is accompanied by a localized pulse of actin polymerization. It has been clear for some time that this process is governed by Rac, a member of the Ras superfamily of small GTPases, but the mechanisms by which Rac’s orders are carried out remain obscure.

On page 253 of this issue, Edwards et al.1 establish a direct link between two protein kinases, p21-activated kinase (Pak) and LIM-domain-containing kinase (LIM-kinase), both of which were known previously to be regulated by Rac and to affect actin dynamics. In doing so, Edwards et al. bring us one step closer to solving the puzzle of how Rac induces changes in cell shape and motility.

In its activated state, Rac is a veritable powerhouse. It has pronounced effects on the actin cytoskeleton, and also on gene transcription, cell-cycle progression and the generation of superoxide anions2. Expression of activated Rac induces the formation of lamellipodia and the dissolution of pre-existing actin stress fibres, long actin- and myosin-based cables that are linked to the plasma membrane and which, when contracting, can exert tension. In motile fibroblasts, a single lamellipodium extends in the direction of movement, and expression of either dominant-interfering forms of Rac or proteins that inhibit its activity block both lamellipodia formation and cell locomotion. These and a plethora of related experiments have firmly established the importance of Rac in regulating actin-based motility in a wide variety of eukaryotic cells.

How does Rac do it? At least three major events that affect actin structure and disposition can be traced back to Rac. These are the uncapping of actin filaments, the de novo synthesis of new actin polymers, and the control of depolymerization and severing rates of pre-existing actin filaments. The first of these events, the uncapping of actin filaments, is thought to be related to the Rac-stimulated generation of phosphatidylinositol-4,5-bisphosphate3. This phospholipid catalyses the removal of capping proteins that block the barbed (growing) ends of actin filaments, thus freeing up these ends for elongation by addition of actin monomers.

Rac also induces the synthesis of new filaments, perhaps through the recruitment and activation of the actin-polymerizing Arp2/3 complex. How this occurs is a bit of a mystery, although, by analogy to the related GTPase Cdc42, Rac may accomplish this task by recruiting the Arp2/3 complex into nascent lamellipodia through a protein known as WAVE or SCAR4. Finally, Rac regulates the length and stability of actin filaments by inhibiting the activity of the actin-depolymerizing protein cofilin. This protein has two activities: it increases actin depolymerization from the pointed (slowly growing) filament end and also cuts pre-existing actin filaments5. When cofilin is inactivated, there is a net increase in filamentous actin, most likely resulting from the loss of cofilin’s depolymerizing activity (Fig. 1).

Figure 1: Regulation of actin by the GTPase Rac.

Activated Rac stimulates the formation of lamellipodia at the leading edge of motile cells by altering at least three signalling pathways that affect actin dynamics. Rac recruits the enzyme phosphatidylinositol-4-phosphate-5-OH kinase (PtdIns(4)P-5-OH kinase), leading to the formation of phosphatidylinositol-4,5-bisphosphate (PtdIns(4,5)P2), which aids in the uncapping of pre-existing actin filaments. Rac also recruits and activates the Arp2/3 complex, perhaps through a WAVE/SCAR protein. The Arp 2/3 complex can act to initiate formation of new actin polymers. The coloured boxes show a third pathway, addressed in ref. 1. Rac binds and activates the protein kinase Pak, which in turn binds to, phosphorylates and activates LIM-kinase (LIMK), which then phosphorylates (circled ‘P’) cofilin, inhibiting its ability to depolymerize actin. It is likely that LIMK is also regulated by other kinases and phosphatases, and that its activity is rapidly modulated during cell movement. Pak also downregulates myosin-light-chain kinase (MLCK), perhaps contributing to the loss of stress fibres observed in cells expressing activated Rac.

It is the link from Rac to cofilin that is addressed by Edwards et al.1. Cofilin is inactivated by phosphorylation of a serine residue at its extreme amino terminus, and last year two groups showed that LIM-kinase can catalyse this reaction6,7. In addition, the activity of LIM-kinase is modestly increased by activated Rac, and dominant-interfering LIM-kinase mutants block lamellipodia formation by Rac, but not filopodia formation by Cdc42 (ref. 7). Thus, LIM-kinase appears to act downstream of Rac, but not Cdc42, to regulate cofilin and hence actin. Intriguingly, neither group could show that LIM-kinase associates directly with Rac, so it seemed probable that one or more extra proteins were interposed between Rac and LIM-kinase. That assumption is now confirmed by Edwards et al., who show that LIM-kinase is phosphorylated by Pak, a kinase that directly binds to, and is activated by, both Rac and Cdc42 (ref. 8).

Pak has a chequered history as a GTPase effector for regulating the actin cytoskeleton. In both budding and fission yeast, Pak homologues clearly play a part in controlling cell polarity and actin organization, perhaps in part by phosphorylating (and thereby activating) the motor protein myosin. In mammalian cells, however, the situation is more complex and controversial. Although several groups have shown that overexpression of Pak leads to a variety of changes in actin-filament organization, mutants of Rac that fail to bind or activate Pak are, nevertheless, perfectly competent at generating lamellipodia. This seeming conundrum may be more apparent than real, as Pak may be activated by GTPases even in the absence of direct binding, and certain newly described isoforms of Pak, such as Pak4 (ref. 9), can bind effectively to the very GTPase mutants that were used to exclude the other members of the Pak family as cytoskeletal effectors.

Pak has several different effects on the actin cytoskeleton. High-level expression of activated Pak mutants induces near-total loss of stress fibres and cell shrinkage, whereas more modest expression of such mutants induces formation of polarized lamellipodia and is associated with increased cell motility and persistence of movement10,11,12. Interestingly, it also has some effects that are independent of its kinase activity.

Until now, only three direct substrates of Pak had been described, namely Raf13, Mek14 and myosin-light-chain kinase (MLCK)15. Phosphorylation of Raf and Mek is thought to be important for strengthening the association of these proteins with each other and with Ras, whereas phosphorylation of MLCK inhibits its activity and may be important for Pak’s effects on reducing stress fibres (Fig. 1). However, the effects of Pak on myosin activity may depend on cell type and/or expression levels12,15 and, in any case, the formation of lamellipodia cannot be fully explained by Pak-induced reductions in MLCK activity (or phosphorylation of Raf and Mek).

It is in this context that the identification of LIM-kinase as a target for Pak represents a potential breakthrough. First, Edwards et al.1 show that Pak can markedly activate LIM-kinase by phosphorylating it at threonine 508, a residue within its activation loop. Second, in cells, Rac-activated Pak associates with LIM-kinase, and inhibition of endogenous Pak (by overexpression of a Pak fragment that binds to and inhibits the full-length protein) blocks both activation of LIM-kinase and its cytoskeletal effects. Finally, kinase-defective LIM-kinase blocks many of the cytoskeletal effects of Rac and Pak, and also, interestingly, some of those of Cdc42. The authors thus propose the existence of a Rac→Pak→LIM-kinase→cofilin axis, the functional consequence of which would be to slow down the rate of actin-filament depolymerization when Rac is active (Fig. 1).

This model raises several questions. First, as mentioned earlier, cofilin does more than just depolymerize actin; it can also sever pre-existing actin filaments. It is not clear that blocking this severing activity would be in Rac’s best interests, if the goal of this GTPase is to reorganize actin filaments into the meshwork structures required for lamellipodia and, presumably, for cell movement. It is also important to remember that, in many cases, stimulation of cell motility increases, rather than inhibits, total cofilin activity. Either there are spatial or temporal issues involved (for example, Rac inactivates cofilin in one place or time while other regulators activate it elsewhere), or cofilin’s depolymerizing activity is more important than its severing activity. It is also possible that cofilin must cycle rapidly between active and inactive states during periods of dynamic actin reorganization.

A second question relates to the relevance of this model to Rac’s cousins Cdc42 and TC10. Like Rac, these GTPases are potent Pak activators, but they have distinct effects on cytoskeletal architecture. It is somewhat surprising that Cdc42 barely activates LIM-kinase in transfection experiments, and even this small effect may be indirect7. In addition, blockade of LIM-kinase activity does not affect Cdc42’s ability to induce filopodia or microspikes1,7, although it does affect the peripheral reorganization of actin by both Cdc42 and Rac. Clearly there are some elements missing from the model, perhaps relating to further regulatory proteins that impinge on LIM-kinase and/or cofilin. That this is the case may also be inferred from the finding that the effects of Pak blockade on LIM-kinase activity are only partial, implying that other kinases/phosphatases may also target this protein. An effector for Rho, Rho-associated kinase (Rock), may fit the bill, as Maekawa et al. recently reported that Rock, like Pak, phosphorylates and activates LIM-kinase16. Thus LIM-kinase can be positively regulated by both Rac and Rho, through their effectors Pak and Rock, even though these protein pairs are generally thought to act in opposition to one another, as, for example, in their effects on MLCK activity and stress-fibre formation.

The Pak–LIM-kinase connection may also be relevant to our understanding of certain human neurological disorders. Pak3 and LIMK1 are expressed primarily in neuronal tissues, and mutations of the genes encoding these proteins are associated with cognitive defects in man. A mutation encoding a kinase-deficient form of Pak3 is found in an X-linked mental retardation syndrome17, while hemizygosity of the limk1 gene is associated with Williams syndrome, a disorder characterized by specific visuospatial abnormalities. Could both of these defects be somehow related to abnormal cofilin regulation? Does a similar defect affecting actin-filament polymerization underlie the phenotype produced by Pak mutations in flies, in which a neuronal structure known as the mushroom body fails to develop properly18? Edwards et al.’s findings lead one to suspect that this may be so, but the creation and analysis of additional animal models, and a great many more biochemical experiments, will be required before we know if this is true. But for now, although the totality of the cell-movement machine is not yet in view, we understand the logic of its design a little more clearly.


  1. 1

    Edwards, D. C, Sanders, L. C., Bokoch, G. M. & Gill, G. N. Nature Cell Biol. 1, 253–259 (1999).

    CAS  Article  Google Scholar 

  2. 2

    Mackay, D. J. G. & Hall, A. J. Biol. Chem. 273, 20685–20688 (1998).

    CAS  Article  Google Scholar 

  3. 3

    Ren, X. D. & Schwartz, M. A. Curr. Opin. Genet. Dev. 8, 63–67 (1998).

    CAS  Article  Google Scholar 

  4. 4

    Miki, H., Suetsugu, S. & Takenawa, T. EMBO J. 17, 6932–6941 (1998).

    CAS  Article  Google Scholar 

  5. 5

    Bamburg, J. R., McGough, A. & Ohno, S. Trends Cell Biol. 9, 364–370 (1999).

    CAS  Article  Google Scholar 

  6. 6

    Arber, S. et al. Nature 393, 805–809 (1998).

    CAS  Article  Google Scholar 

  7. 7

    Yang, N. et al. Nature 393, 809–812 (1998).

    CAS  Article  Google Scholar 

  8. 8

    Manser, E. et al. Nature 367, 40–46 (1994).

    CAS  Article  Google Scholar 

  9. 9

    Abo, A. et al. EMBO J. 17, 6527–6540 (1998).

    CAS  Article  Google Scholar 

  10. 10

    Sells, M. A. et al. Curr. Biol. 7, 202–210 (1997).

    CAS  Article  Google Scholar 

  11. 11

    Manser, E. et al. Mol. Cell. Biol. 17, 1129–1143 (1997).

    CAS  Article  Google Scholar 

  12. 12

    Sells, M. A., Boyd, J. T. & Chernoff, J. J. Cell Biol. 145, 837–849 (1999).

    CAS  Article  Google Scholar 

  13. 13

    King, A. J. et al. Nature 396, 180–183 (1998).

    CAS  Article  Google Scholar 

  14. 14

    Frost, J. A. et al. EMBO J. 16, 6426–6438 (1997).

    CAS  Article  Google Scholar 

  15. 15

    Sanders, L. C., Matsumura, F., Bokoch, G. M. & de Lanerolle, P. Science 283, 2083–2085 (1999).

    CAS  Article  Google Scholar 

  16. 16

    Maekawa, M. et al. Science 285 895–898 (1999).

    CAS  Article  Google Scholar 

  17. 17

    Allen, K. M. et al. Nature Genet. 20, 25–30 (1998).

    CAS  Article  Google Scholar 

  18. 18

    Melzig, J. et al. Curr. Biol. 8, 1223–1226 (1998).

    CAS  Article  Google Scholar 

Download references

Author information



Rights and permissions

Reprints and Permissions

About this article

Cite this article

Chernoff, J. Close encounters of the LIM-kinase. Nat Cell Biol 1, E115–E117 (1999).

Download citation

Further reading


Nature Briefing

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing