| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
* The Walter and Eliza Hall Institute of Medical Research, Victoria, Australia;
Bioinformatics Support Unit, Ben-Gurion University of the Negev, Beer-Shiva, Israel;
Department of Biological Regulation, The Weizmann Institute of Science, Rehovot, Israel; and
St. Vincent’s Institute, Fitzroy, Victoria, Australia
2Correspondence: St. Vincent Institute of Medical Research, 9 Princes St., Fitzroy, Victoria 3065, Australia. E-mail:obernard{at}svi.edu.au
SPECIFIC AIMS
THE ORIGINAL AIM of this study was to elucidate the mechanism that controls the concentration of LIM kinase 1 (LIMK1) in cells. Computer searches indicated that the kinase domain of LIMK1 contains Hsp90 recognition sequences similar to those found in ErbB2, an Hsp90 client protein. We, therefore, studied the effects of Hsp90 inhibition on the concentration of LIMK1 protein.
PRINCIPAL FINDINGS
1. LIMK1 stability is regulated by transphosphorylation
LIMK1 is a very stable protein with a half-life of 
20 h. In contrast, the half-life of two dominant-negative counterparts, which lack kinase activity (kinase dead) and therefore cannot be transphosphorylated, is only 4 h. More importantly, the half-life of endogenous LIMK1 proteins that coimmunoprecipitated with the overexpressed wild-type or kinase-dead LIMK1 was similar to that of the overexpressed proteins (20 and 6 h, respectively). While overexpressed LIMK1 can transphosphorylate endogenous LIMK1, kinase-dead LIMK1 cannot transphosphorylate the endogenous protein, which results in unphosphorylated endogenous LIMK1 with a shorter half-life of 6 h. These results are consistent with the possibility that phosphorylation protects LIMK1 from degradation.
2. Hsp90 protects LIMK from degradation
Computer searches of protein databases revealed that LIMK1 contains a short amino acid sequence within its kinase domain similar to that of the ErbB-2 sequence responsible for its binding to Hsp90 (Fig. 1
A). We, therefore, examined the effects of Hsp90 on the stability of these proteins. Treatment of 293T cells with the well-characterized antagonists of Hsp90, 17-AAG, and Radicicol resulted in a dramatic reduction in the concentration of LIMK1 and LIMK2, with only 10% of the protein remaining after 24 h incubation (Fig. 1B-D
). The decrease in LIMK levels was associated with reduced levels of p-cofilin, as incubation of cells with 17-AAG or Radicicol decreased the levels of endogenous p-cofilin, the main LIMK1 substrate, by 60 and 80%, respectively (Fig. 1E
). This finding indicated that, in the absence of active Hsp90, LIMK proteins undergo degradation.
|
3. Identification of the amino acid-mediating LIMK1-Hsp90 association
To identify the amino acid within the kinase domains of LIMK1 and LIMK2 that is responsible for the association with Hsp90, we have chosen the proline-to-glutamic acid mutation because, in the case of EGFR/HER2, the GVGSPYVS sequence (alphaC-beta4 loop) of ErbB-2/HER2 determines Hsp90 binding. Based on analogies to other kinases, we proposed that replacing the proline of LIMK with an acidic residue would inhibit Hsp90 binding. Indeed the P394E mutation affected the interaction between LIMK1 and Hsp90, which resulted in significant reduction in the amount of LIMK1-P394E that was associated with Hsp90 compared with that of the wild-type protein (Fig. 2
A). Moreover, addition of 17-AAG reduced Hsp90 association with wt-LIMK1 but had no further effect on its weak association with LIMK1-P394E or on its concentration in the cells (Fig. 2A-C
). In vitro kinase assays showed that the level of LIMK-P394E phosphorylation was reduced by 70% compared with that of wt-LIMK1. While 17-AAG inhibited LIMK1 phosphorylation by 60%, it had only a slight effect (10%) on the level of transphosphorylation of the P394E mutant. Expression of LIMK1-P394E in 293T cells resulted in 25% reduction in the level of endogenous phospho-cofilin in these cells, indicating that the mutation influenced the level of LIMK1 transphosphorylation and its activity. Covalent cross-linking experiments revealed that wt-LIMK1 could form homodomers while the mutant was not able to dimerize. LIMK1 homodimerization was dependent on the presence of Hsp90, as no dimers were formed on addition of 17-AAG. Comparison of the half-lives of F-LIMK1 and the P394E mutant demonstrated that the half-life of LIMK1 was 12 h, while that of LIMK1-P394E was only 3 h. The half-life of the P394E mutant was similar to the half-life of the kinase-dead LIMK1-D460E.
|
CONCLUSION AND SIGNIFICANCE
LIMK1 is a very stable protein with a half-life of 16–20 h. We have demonstrated previously that LIMK1 levels and its phosphorylation are markedly elevated in metastatic breast cancer cell lines compared with nonmetastatic and normal cells. The extended half-life of LIMK1 is not due to the stability of its mRNA because no significant differences were observed in the levels of LIMK1 mRNA in these cell lines. In addition, studies using microarrays also did not reveal significant differences in the concentration of LIMK1 mRNA between invasive and noninvasive breast cancer cell-lines. Together these findings suggest that the concentration of LIMK1 protein is regulated by posttranslational modification. In this study we confirmed this hypothesis and furthermore have provided data to support that LIMK1 dimer formation and transphosphorylation are responsible for the stability of LIMK1 protein.
The data obtained in this study identified two new members in addition to the long list of Hsp90 client proteins. Binding of Hsp90 to the LIMK proteins protects them from degradation, as after 24 h incubation with Hsp90 inhibitors only less than 10% of these proteins can be detected. This corresponds to a half-life of 4 h, which is similar to that of the kinase-dead LIMK1 and unphosphorylated endogenous LIMK1 proteins. These findings strongly suggest that Hsp90 promotes dimer formation and transphosphorylation of LIMK1. Indeed, we have shown here that Hsp90 binds to LIMK1, leading to the formation of LIMK1 homodimers. We have previously demonstrated that deletion mutants lacking the LIM and PDZ domains of LIMK1 have increased kinase activity. This was attributed to findings that the LIM and PDZ domains interact with the kinase domain and thereby inhibit its association and activation by ROCK and PAK. We propose that the binding of Hsp90 to the kinase domain of LIMK1 enables the opening up of the molecule and promotion of homodimers formation.
Although the recognition motif of LIMK is similar to that of ErbB-2, the role that Hsp90 binding plays is different between these two protein kinases. While binding of Hsp90 to the kinase domain of ErbB-2 restrains its catalytic activity and prevents heterodimer formation with other ErbB proteins, Hsp90 binding to the kinase domain of LIMK1 results in homodimer formation enabling LIMK1 transphosphorylation and its activation. Our cross-linking experiments with overexpressed LIMK1 demonstrated the formation of LIMK1 homodimers of 140 kDa only in the absence of 17-AAG. In contrast, LIMK1-P394E does not form homodimers in the absence or presence of the Hsp90 inhibitors. This suggests that the interaction between these two proteins depends on the Hsp90 binding motif in the kinase domain of LIMK.
Recently it was revealed that the interaction between Hsp90 and ErbB-2 is mediated by a loop in the N lobe of the kinase domain of ErbB-2. Mutations in this loop that disrupt the positively charged patch and the hydrophobic strip abolish the association between these two molecules. Indeed, modeling of the kinase domain of LIMK1 demonstrated the existence of a similar loop also in LIMK1 and other EGFR family members. Moreover, comparison of the sequence within this loop demonstrated that more than half the residues are identical or conservatively substituted between LIMK1 and ErbB-2. Interestingly, three amino acids conserved between LIMK and ErbB-2 are also shared between ErbB-2 and ErbB-1 and are not involved in the association between ErbB-2 and Hsp90. In contrast, mutation of P394, a residue conserved among LIMK1, ErbB-2, and ErbB-1 to the negatively charged glutamic acid, greatly reduced the association between LIMK1 and Hsp90. It remains to reveal whether changes in the charge or/and the backbone conformation, or the introduction of a longer side chain in this position of this region affect the interaction between these two molecules.
Given the fact that Hsp90 inhibitors are known to be potent anticancer drugs and are currently in clinical trials, this study postulates a novel anticancer mechanism of Hsp90 inhibitors via their perturbation of LIMK family function, molecules that have previously been implicated in tumor growth, invasion, and metastasis.
|
FOOTNOTES
1 Present address: The Baker Heart Institute, Melbourne, Victoria 3181, Australia. 
To read the full text of this article, go to http://www.fasebj.org/cgi/doi/10.1096/fj.05-5258fje
This article has been cited by other articles:
|
|
 |
|
  Y. Fukuyo, M. Inoue, T. Nakajima, R. Higashikubo, N. T. Horikoshi, C. Hunt, A. Usheva, M. L. Freeman, and N. Horikoshi Oxidative Stress Plays a Critical Role in Inactivating Mutant BRAF by Geldanamycin Derivatives Cancer Res., August 1, 2008; 68(15): 6324 - 6330. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 Y. Ding, T. Milosavljevic, and S. K. Alahari Nischarin Inhibits LIM Kinase To Regulate Cofilin Phosphorylation and Cell Invasion Mol. Cell. Biol., June 1, 2008; 28(11): 3742 - 3756. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. Amiri, F. Noei, T. Feroz, and J. M. Lee Geldanamycin Anisimycins Activate Rho and Stimulate Rho- and ROCK-Dependent Actin Stress Fiber Formation Mol. Cancer Res., September 1, 2007; 5(9): 933 - 942. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
