Vimentin intermediate filament reorganization by Cdc42: Involvement of PAK and p70 S6 kinase

Vimentin intermediate filament reorganization byCdc42: Involvement of PAK and p70 S6 kinase

Wing Chan1)a, Robert Kozmab, Yoshihiro Yasuic, Masaki Inagakic, Thomas Leunga, Ed Mansera, Louis Lima,ba Glaxo-IMCB Group, Institute of Molecular and Cell Biology, Singapore/Singaporeb Institute of Neurology, London/UKc Division of Biochemistry, Aichi Cancer Center Research Institute, Nagoya/Japan

Received May 27, 2002Received in revised version July 25, 2002Accepted August 1, 2002

Cdc42 ± vimentin ± MRCK ± PAK ± p70 S6K

Rho family GTPases play a major role in actin cytoskeletonreorganization. Recent studies have shown that the activationof Rho family GTPases also induces collapse of the vimentinintermediate filament (IF) network in fibroblasts. Here, wereport that Cdc42V12 induces the reorganization of vimentin IFsin Hela cells, and such reorganization is independent of actinand microtubule status. We analyzed the involvement of threeserine/threonine kinase effectors,MRCK, PAKand p70 S6K inthe Cdc42-induced vimentin reorganization. Surprisingly, theROK-related MRCK is not involved in this IF reorganization.We detected phosphorylation of vimentin Ser72, a site phos-phorylated by PAK, after Cdc42 activation. PAK inhibitionpartially blocked Cdc42-induced vimentin IF collapse suggest-ing the involvement of other effectors. We report that p70 S6kinase (S6K)1 participates in this IF rearrangement since theinhibitor rapamycin or a dominant inhibitory S6K could reducethe Cdc42V12 or bradykinin-induced vimentin collapse. Further,inhibition of PAK and S6K in combination very effectivelyprevents Cdc42-induced vimentin IF collapse. Conversely, onlyin combination activePAKand S6K could induce a vimentin IFrearrangement that mimics the Cdc42 effect. Thus, Cdc42-induced vimentin reorganization involves PAK and, in a novelcytoskeletal role, p70 S6K.

Abbreviations. IFs Intermediate filaments. ± MRCK Myotonic dystrophykinase-related Cdc42-binding kinase. ± MTs Microtubules. ± PAK p21-Activated kinase. ± S6K p70 S6 kinase.

Introduction

Microfilaments, microtubules and intermediate filaments (IFs)form three major cytoskeletal networks in most eukaryoticcells. Although each network executes individual tasks, thethree networks intercommunicate through specific bindingproteins (Fuchs and Yang, 1999) to carry out many integratedcellular processes such as adhesion, migration and division. IFscontribute substantially to maintaining cell shape and mechan-ical properties of the cytoplasm in higher eukaryotic cells.Although IFs are relatively stable structures in vivo comparedwith the other two cytoskeletal components, they too undergodynamic changes during cell growth, polarization and differ-entiation. IF reorganization is regulated by phosphorylation ina site-specific manner (reviewed by Inagaki et al. (1996)),through protein kinases such as PKA and PKC (Evans, 1988;Inagaki et al., 1987), CaMKII (Tokui et al., 1990), cdc2 kinase(Chou et al., 1990, 1991; Kusubata et al., 1992) and RhoAeffector, ROK� (Goto et al., 1998; Kosako et al., 1999; Sinet al., 1998).

Rho family GTPases play a major role in actin cytoskeletonreorganization. In mammalian cells, RhoA promotes theformation of stress fibers and focal adhesion through itsdownstream effector ROK� (Amano et al., 1997; Ishizakiet al., 1997; Leung et al., 1996). Rac1 promotes membraneruffles and lamellipodia while Cdc42 induces formation offilopodia and microspikes (Kozma et al., 1995; Nobes and Hall,1995). The kinase effectors of Cdc42 and Rac1, such as PAK andMRCK have been shown to mediate some of these morpholog-ical effects (Chen et al., 1999; Leung et al., 1998; Manser et al.,1997; Sells et al., 1997). However, despite their extensive rolesin actin filament organization, little is known on the effects ofthese GTPases in the intermediate filament reorganization.Previously protein kinase N, which is activated by RhoA, wasshown to phosphorylate neurofilaments (Mukai et al., 1996).ROK is reported to phosphorylate vimentin and glial fibrillaryacidic protein (Goto et al., 1998; Kosako et al., 1997). Phos-phorylation of vimentin by ROK� causes the disassembly and

0171-9335/02/81/12-692 $15.00/0

1) Dr. Wing Chan, Glaxo-IMCB Group, Institute of Molecular and CellBiology, 30 Medical Drive, Singapore 117609/Singapore, e-mail:[email protected], Fax: �6567740742.

EJCB692 European Journal of Cell Biology 81, 692 ± 701 (2002, December) ¥ ¹ Urban& Fischer Verlag ¥ Jenahttp://www.urbanfischer.de/journals/ejcb

the collapse of the vimentin network in Swiss 3T3 fibroblastsand HeLa cells (Sin et al., 1998). Such phosphorylation ofvimentin at one specific site (Ser71) only occurs at the cleavagefurrow during cytokinesis (Goto et al., 1998).

PDGF and bradykinin treatment leads to a marked reorga-nization of vimentin in cells. Cdc42 and Rac1 have recentlybeen found to be involved in this process. The collapse of IFsinto dense perinuclear bundles was mimicked by the over-expression of constitutively active mutant Rac1V12 or Cdc42V12,suggesting Rac1 and Cdc42 are important regulators of bothvimentin and actin filaments (Meriane et al., 2000; Valgeirs-dottir et al., 1998). Tyrosine phosphorylation events weresuggested to be involved in the Cdc42- and Rac1-inducedvimentin reorganization. However, no tyrosine phosphoryla-tion of vimentin was observed nor is it clear how tyrosinekinases might indirectly affect vimentin IFs (Meriane et al.,2000).

In this report, we analyze the involvement of three down-stream effectors of Cdc42, MRCK, PAK and p70S6K, in theCdc42-induced vimentin collapse in Hela cells. Our resultsindicate that MRCK does not mediate the Cdc42-inducedeffect on vimentin IFs. Instead, vimentin Ser72, a site phos-phorylated by PAK, becomes phosphorylated after Cdc42activation. However inhibiting PAK only partially blocks theCdc42-induced vimentin collapse. Non-CRIB Cdc42-bindingeffector S6K was found to participate in this process. InhibitingS6K by rapamycin and a dominant inhibitory S6KT389A/�C

mutant more efficiently protects cells against Cdc42- orbradykinin-induced vimentin collapse. In combination, wefind that simultaneous inhibition of both PAK and S6Kprevents any Cdc42-induced vimentin IF rearrangements.

Materials and methods

Cell cultureHeLa cells were maintained in minimal essential medium (MEM)supplemented with 10% fetal bovine serum. COS-7 and Swiss 3T3fibroblast cells were maintained in Dulbecco×s modified Eagle×smedium (DME) supplemented with 10% fetal bovine serum.

MicroinjectionHeLa cells were cultured on glass coverslips for 24 ± 48 h, andsubconfluent cells were microinjected with various plasmid constructsencoding proteins tagged with HA at a concentration of 50 ng/�l usingan Eppendorf micromanipulator system. In co-injection experiments, a4 ± 5-fold excess of inhibitor plasmids (e.g. PAK inhibitory fragment,MRCK triple mutant or p70 S6K dominant inhibitory mutant) wereused.

Immunofluorescence microscopyTwo to four h after injection, cells were fixed with 100% cold methanoland kept at �20 �C for 6 min (for vimentin staining) or fixed with 4%paraformaldehyde for 10 minutes (for tubulin staining). For stainingwith Ser72 (anti-rabbit) or Ser6 (anti-mouse) phospho-vimentin anti-bodies, cells were fixed with 3.7% formaldehyde at 4 �C for 10 minutesfollowed by methanol at �20 �C for 10 minutes. After washing andrehydration with PBS, cells were incubated with rabbit antibodiesagainst HA (Zymed Laboratories; CA, USA) and mouse antibodiesagainst vimentin (Sigma; St. Louis, MO), phospho-vimentin or tubulin(Santa Cruz; CA, USA) overnight at 4 �C. After washing, the cells wereincubated with rhodamine-conjugated anti-rabbit antibody and FITC-conjugated anti-mouse antibody (Boehringer Mannheim; Germany).Labeled cells were analyzed using a conventional fluorescence micro-scope (model Axioplan2; Carl Zeiss, Inc., Germany) with 40� 1.3 oil

immersion objective or with a confocal microscope (BioRad, radiance2000; CA, USA).

Transfection and immunoprecipitationSubconfluent COS-7 cells were transfected with 5 �g FLAG-taggedplasmid DNA encoding various kinases (ROK-kinase domain or S6K)using Lipofectamine (Bethesda Research Laboratories,). Cells wereharvested 18 ± 24 h after transfection and lysed with 500 �l immunopre-cipitation (IP) buffer as described previously (Leung et al., 1996). Thecell extracts were centrifuged (20000g) for 20 min at 4 �C. Forimmunoprecipitation, 250 �l extracts were incubated with 20 �l proteinA beads containing immobilized anti-FLAG M2 (Kodak/IBI; NY, USA)for 2 h at 4 �C. After washing four times with 400 �l IP buffer with 0.1%Triton X-100, the beads were used in kinase assays.

Protein kinase assay andWestern blot analysisKinase assays were carried out for 20 min at 30 �C in a 50-�l reactionmixture containing kinase buffer (50 mM HEPES, pH 7.3, 50 M KCl,10 mM �-glycerophosphate, 5 mM NaF, 10 mM MgCl2, 2 mM MnCl2,1 mM dithiothreitol and 0.05% Triton X-100), 10 �g of kinase substrates(GST-vimentin, GST-MLC, GST-PIX441 or GST-S6 peptide fusionprotein), 10 �M ATP, and 10 �Ci [�-33P]ATP (Amersham; UK). Thereaction was stopped by adding 50 �l of 2�SDS sample buffer. Afterboiling for 10 min, half of the sample was run on a 10% SDS-PAGE gel.Separated samples were subsequently transferred to PVDF membranes(NEN; Boston, USA), stained with Coomassie blue and exposed toautoradiography. The PVDF blot was then probed with mouse anti-FLAG antibody (1 :1000, Sigma) to check the amount of proteins thatwas immunoprecipitated. Specific immunoreactivity was detected usingthe ECL kit (Amersham).

PCR cloning of vimentin and p70 S6K frommouse brain cDNA and generation of p70 S6KmutantsThe 1400-bp full-length mouse vimentin was obtained from mouse braincDNA (Clontech, CA, USA) by PCR using primers 5�-GGATC-CATGTCTACCAGGTCTGTGTCC-3� and 5�-GCGGCCGCTTATT-CAAGGTCATCGTGATG-3�. The 1508-bp full-length mouse p70 S6Kwas obtained from the same cDNA library by PCR with a pair of primers(5�-GATCCATGGCAGGAGTGTTTGACATA-3� and 5�-CGG-CCGCTCATAGATTCATACGCAGGTG-3�). Vimentin was sub-cloned into pGEX4T1 (Pharmacia; New Jersey, USA) whereasp70S6K was subcloned into the pXJ40-HA (or pXJ40-FLAG) mamma-lian expression vectors (Manser et al., 1997). The truncated (amino acids1 ± 393) S6KT389A/�C mutant was generated by PCR using the QuickChange site-directed mutagenesis kit (Stratagene; CA, USA) with anoligo containing T389A (5�-GTCTTTCTGGGTTTTGCATAT-GTGGCTCCAT-3�).

Bacterial expression and purification of GST-vimentin and GST-S6K substrate peptide fusionproteinsGST-vimentin fusion protein was expressed in E. coli BL21. Thebacterial pellet (from 100 ml) was resuspended in 5 ml lysis buffer(PBS/50 mM Tris-HCl, pH 8.0/1 mM EDTA/25% sucrose). Lysozymewas added to 1 mg/ml and incubated for 30 min at 4 �C in the presence ofprotease inhibitors (complete cocktail, Roche). Additionally, 10 mMMgCl2 and DNase I (40 �g/ml) was added to the lysates and theincubation was continued for 15 min at 4 �C. Two volumes of lysis bufferwere added and the suspension was passed through a 20-gauge needlethree times and centrifuged at 5000 rpm for 12 min. The inclusion bodieswhich contained the GST-vimentin fusion proteins were washed threetimes with lysis buffer and then dissolved in buffer containing 8 M urea,0.1 M NaH2PO4 and 0.01 M Tris, pH 8.0, and dialyzed against PBS. Thefinal suspension was clarified by centrifugation (38000 rpm for 30 min)resulting in soluble GST-vimentin used in in vitro kinase assays.

For expression and purification of GST-S6K substrate peptide,an oligonucleotide (5�-CAAAGCGACGACGACTATCATCGC-

693Cdc42 and vimentin IF collapseEJCB

TGCGTGCT-3�) encoding the S6K substrate peptide sequence(AKRRRLSSLRA) was synthesized and subcloned into the Bam HIand Not I sites of the GST fusion protein expression vector pGEX-4T-1(Pharmacia; New Jersey, USA). GST-S6 peptide fusion proteins wereexpressed in E. coli strain BL21 and purified by glutathione-Sepharoseaffinity chromatography (Pharmacia).

Results

Overexpression of constitutively active Cdc42induces vimentin IF collapse in Hela cellsWe have previously shown that expression of RhoA and activeROK� causes the collapse of filamentous vimentin in Hela cells(Sin et al., 1998). Conversely, RhoA and ROK� promote theformation of actin-containing stress fibers and focal adhesions

(Leung et al., 1995, 1996). This effect of RhoA is antagonizedby Cdc42 which promotes disassembly of these structures(Kozma et al., 1995). Recently, Cdc42 and Rac1 were shown toinduce vimentin IF reorganization in rat and human fibroblasts(Meriane et al., 2000). As shown in Figure 1 this reorganizationcan also be observed in epithelial cells within hours ofexpression of active Cdc42V12 suggesting this is an early eventin cytoskeletal rearrangement. Since Cdc42V12-expressing cellslose their internal stress fibers (Fig. 1A, panels c and d), it isunlikely that the vimentin reorganization is a result of theeffects on actin. In Cdc42V12-injected cells the vimentin networkhad rearranged into a compact ring-like structure around theperinuclear region (Fig. 1A, panels a and b). Approximately10% in uninjected HeLa cells showed this phenotype whichmay relate to their cell cycle state (Fig. 1B). Although similareffects were seen with GFP-Cdc42V12 in HeLa cells, we foundGFP alone caused vimentin rearrangement after microinjec-tion (data not shown). Therefore, this tag was not used infurther experiments. Unlike the previous study (Meriane et al.,2000) we found that the effector mutants Cdc42V12/C40 andCdc42V12/A37 both exhibited a decreased ability to causevimentin rearrangement (Fig. 1B), suggesting that more thanone effector protein is involved in the vimentin rearrangement.The Cdc42V12/C40 does not directly bind PAK-like effectors(Lamarche et al., 1996).

Since vimentin IF collapse has been linked to the depoly-merization of microtubules (Fuchs and Weber, 1994), weexamined the microtubule (MT) network in the Cdc42V12-expressing cells. MTs and the microtubule-organizing centerremain intact in these cells, although microtubule filamentsbecame re-orientated after Cdc42 expression (Fig. 2a). Athigher magnification the MT filaments were seen as intact byconfocal microscopy compared with nocodazole-treated cells(Fig. 2b ± d). Taken together, our results suggested that acti-vated Cdc42 caused vimentin IF to rearrange under theconditions of net stress fiber disassembly, and that this is not aresult of MT instability.

MRCK does not mediate Cdc42-inducedvimentin IF rearrangementIF reorganization is regulated by serine/threonine phosphor-ylation at specific sites in the head domain (reviewed in(Inagaki et al., 1996)). Although tyrosine phosphorylationevents were suggested to be involved in the Cdc42- and Rac1-induced vimentin reorganization on the basis of the inhibitorgenistein, no direct tyrosine phosphorylation of vimentin wasobserved (Meriane et al., 2000). The tyrosine kinase Cdc42effector ACK1 did not induce vimentin IF reorganization (datanot shown).

In search for kinase(s) that might be involved in the Cdc42-induced vimentin IF collapse, we first tested MRCK, a kinaseeffector of Cdc42 related to ROK. Both ROK and MRCK,which share 50% sequence identity in the catalytic domain,were shown to act on similar substrates such as MLC-2 (Leunget al., 1998). Microinjection of a plasmid encoding full-lengthMRCK into HeLa cells resulted in vimentin IF collapse in 70%of the injected cells (Fig. 3A, panels a and b, and 3B). This wasinhibited upon co-injection of a plasmid encoding the MRCK-TM which is both kinase-dead and p21-binding defective(Fig. 3A, panels c and d, and 3B), because this prevents kinasetransphosphorylation (Tan et al., 2001b). Injection of MRCK-TM alone did not cause any effect on vimentin IFs (Fig. 3A,panels e and f). While this MRCK-TM blocks Cdc42-mediated

Fig. 1. Cdc42V12 induces collapse of vimentin IFs. (A) HeLa cellscultured on glass coverslips for 24 ± 48 h were microinjected with aplasmid encoding the full-length Cdc42V12 together with 1 mg/ml rabbitIgG protein (a and c). After 2 h, cells were fixed with either methanolor 4% paraformaldehyde and stained with anti-vimentin antibody (b)or TRITC-conjugated phalloidin (d). The injected cells (indicated byarrows) were visualized with rhodamine-conjugated anti-rabbit IgG.Bar� 10 �m. (B) Scoring percentages of injected cells in (A) anduninjected cells showing condensed vimentin IFs. Cdc42 effectormutants (Cdc42V12/A37 and Cdc42V12/C40) were also microinjected intoHela cells, and cells with reorganized vimentin structure were scoredand calculated. The data represent the mean values (�SE) by scoring atleast 50 cells for injected and uninjected populations.

694 W. Chan, R. Kozma et al. EJCB

induction of focal adhesion and filopodia (Leung et al., 1998),there was no effect on Cdc42-induced vimentin IF rearrange-ment. When plasmids encoding MRCK-TM and Cdc42V12 wereco-injected into HeLa cells, �90% of the injected cellsexhibited collapsed vimentin IFs (Fig. 3A, panels g and h, and3B). A different MRCK dominant inhibitory construct,MRCK-CPC which contains the cysteine-rich/PH C-terminaldomain (Chen et al., 1999), co-injected with the Cdc42V12

plasmid also failed to prevent the Cdc42-induced collapse ofvimentin IFs (data not shown).

ROK and MRCK kinase were then compared for their abilityto phosphorylate vimentin. Flag-tagged MRCK and ROKkinase domains were overexpressed in Cos-7 cells, immuno-precipitated and used in in vitro kinase assays using GST-vimentin as substrate and GST-MLC as positive control. Bothkinases phosphorylated MLC to the same extent (Fig. 3C,lanes 2 and 3). However, although ROK phosphorylatedvimentin (lane 5), we could not detect such phosphorylationby MRCK-KIN (lane 6). Thus, although the MRCK kinasedomain shares 50% sequence similarity with the ROK kinasedomain, it cannot phosphorylate vimentin directly. Vimentinreorganization upon overexpression of full-length MRCK inHela cells is partly due to the fact that MRCK is active even inthe absence of Cdc42 (Leung et al., 1998). Another possibility isthat other kinase(s) downstream of MRCK are involved. Takentogether, we conclude that MRCK is not an effector of Cdc42that directly mediates vimentin IF collapse.

PAK is implicated in vimentin phosphorylationdownstream of Cdc42Since MRCK is not directly involved in the Cdc42-inducedvimentin collapse, we next investigated whether the Cdc42effector, PAK, plays a role similar to ROK� in mediatingRhoA-induced collapse of vimentin IFs. PAK is activated byCdc42 and Rac, and involved in the dissolution of stress fibers(Manser et al., 1997; Sells et al., 1997). PAK has also beenshown to phosphorylate vimentin in vitro (Goto et al., 2002). Toexamine whether PAK could play such a role in Cdc42-inducedvimentin IF rearrangement in vivo, we used the phospho-specific anti-vimentin antibodies to investigate intracellularvimentin phosphorylation downstream of Cdc42 in interphasecells. When Cdc42V12 and �PAK were coexpressed in HeLacells, strong Ser72 phosphorylation was detected (Fig.4a, b)similar to the effect of PAKL107F (Goto et al., 2002). Cdc42V12-expressing cells had detectable levels of vimentin phospho-Ser72 (Fig. 4c, d) whereas phospho-Ser6, a site phosphorylatedby PKC was not detectably modified under the same condition(Fig. 4e, f). Neither could we detect increased vimentinphospho-Ser50 and Ser38 in Cdc42V12-expressing cells. Thediffuse staining pattern by anti-phospho-Ser72 antibody mayreflect the fact that the phospho-vimentin occurs in non-assembled forms (Goto et al., 2002). While PAK is clearly

� Fig. 2. Vimentin IF rearrangements induced by Cdc42 do notcorrelate with microtubule status. Hela cells were fixed and labeledwith anti-� tubulin antibody 3 h after injection with a plasmid encodingHA-tagged Cdc42V12 (a and c). Images were taken at low (a) and highmagnification (b ± d). The control shows the tubulin structure ofuninjected cells (b). For nocodazole treatment (d), cells were incubatedwith 10 �M nocodazole for 30 minutes and then fixed and stained. Cellmicroinjected with Cdc42V12 (white arrow); microtubule-organizingcenter (black arrowhead). Bar� 10 �m.


implicated in the vimentin rearrangement, PAK-KID, aninhibitor of� and �PAKs (Zhao et al., 1998), did not effectivelyprevent vimentin collapse (see Figure 7B) prompting us toconsider the involvement of other downstream kinases.

p70 S6 kinase participates in Cdc42-inducedvimentin IF rearrangementThe Cdc42 effector mutant data suggested a non-CRIB Cdc42-binding effector (i.e. not ACK, PAKor MRCK) may participatein the process. The involvement of p70 S6K downstream ofCdc42 and Rac1 has been clearly shown, although a directconnection between the kinase and GTPases is not clearlyestablished (Chou and Blenis, 1996). p70 S6K is effectivelyinhibited by rapamycin, but does not affect the closely relatedp90 S6 kinase (Chung et al., 1992). We found that rapamycintreatment (30 nM) prior to Cdc42V12 plasmid injection de-

creased the observed vimentin IF collapse to �35% of cells(data not shown), which was more effective than the PAKinhibitor (as illustrated in Fig. 5A, panels a and b). Rapamycindoes not have any effect on vimentin IF structures in uninjectedcells (data not shown). However, we could not discount sideeffects of the drug unrelated to S6K.

Cdc42V12-induced vimentin IF collapse in fibroblasts was alsopartially inhibited by rapamycin (data not shown). We con-firmed the involvement of S6K by using a dominant inhibitoryS6KT389A/�C mutant that lacks the major phospho-activation site(T389) and the essential regulatory S-404 phosphorylation sitesin the linker region (S-404) (Dennis et al., 1996; Pearson et al.,1995). This truncated mutant has been tested by in vitro kinaseassays and is completely inactive (data not shown). Bradykininis an upstream activator of Cdc42 in Swiss 3T3 cells (Kozmaet al., 1995; Lamarch-Vane and Hall, 1998; Martinez-Quiles

Fig. 3. MRCK is not involved in Cdc42-induced vimentin reorganiza-tion. (A) Full-length MRCK induces vimentin IF collapse. HeLa cellswere microinjected with a plasmid encoding full-length HA-taggedMRCK (MRCK-FL) (a and b) or co-injected with dominant inhibitoryMRCK�triple mutant (MRCK-TM) plasmid DNA (c and d). MRCK-TM was also microinjected into cells alone (e and f). Finally, HeLa cellswere co-injected with plasmids encoding HA-tagged Cdc42V12 andMRCK-TM (g and h). Arrows indicate injected cells. Bar� 10 �m. (B)Percentages of injected cells showing condensed vimentin IF in eachinjection experiment shown in (A) were scored and calculated. The

data represent the mean values (�SE) by scoring at least 90 cells and25 cells for the co-injection and single injection experiments, respec-tively. (C) MRCK� does not phosphorylate vimentin in vitro. COS-7cells were transfected with Flag-tagged expression plasmids as shownon top. The Flag-tagged MRCK or ROK kinase domains (MRCK-KINor ROK-KIN, respectively) were immunoprecipitated and incubatedwith either GST-vimentin (lanes 4 ± 6) or GST-MLC as positive control(lanes 1 ± 3). The kinase assay was performed as described in ™Materialsand methods∫. � , present; � , absent.


et al., 2001) and causes similar vimentin rearrangements(Fig. 5B). Swiss 3T3 cells expressing the truncated S6KT389A/�C

mutant were relatively resistant to bradykinin treatment(Fig. 5C).

Immunoprecipitated S6K was tested for its ability tophosphorylate GST-vimentin directly in vitro using the GST-S6 substrate peptide fusion protein as a positive control. TheS6K immunoprecipitated from cell lysates phosphorylated theGST-S6 substrate peptide fusion protein (Fig. 6, lane 3), butphosphorylation of GST-vimentin was hardly detectable com-pared to the ROK� catalytic domain (ROK-KIN) control(Fig. 6, lane 5).

PAK and S6K cooperate in Cdc42-inducedvimentin IF rearrangementConstitutively active PAKL107F expression causes extensiveretraction of Hela cells (Manser et al., 1997), which did notallow for examination of specific vimentin condensation. Wetherefore used another active PAK mutant, PAKDE, which hasacidic substitutions in the kinase ×activation loop× that mimicautophosphorylation (Manser et al., 1997). This mutant couldinduce partial vimentin IF reorganization in Hela cells underconditions in which the cells remained spread (Fig. 7A, panels a

and b). S6K expression alone in cells induced mild vimentin IFrearrangement (data not shown). However when both activePAKDE and S6K were coexpressed in HeLa cells, vimentin IFswere rearranged into a compact ring-like structure very similarto that induced by active Cdc42 (Fig. 7A, panels c and d).Although PAK phosphorylates vimentin in vitro and in vivo(Yasui et al., 2001), inhibition of the kinase did not effectivelyprevent the Cdc42-induced vimentin collapse (Fig. 7B). Whenco-expressed with Cdc42V12 in HeLa cells, the S6KT389A/�C

mutant also reduced the Cdc42-induced vimentin IF phenotype(Fig. 7B). However, when both PAK kinase inhibitory domain(KID) and S6K inhibitory mutant were expressed withCdc42V12, �80% of the double mutant-injected HeLa cellsretained the extended vimentin IF cytoskeleton (Fig. 7B). Thuswhile Cdc42-induced IF rearrangements were relatively un-affected by PAK-KID (50%,) or S6K mutant (45%) expression(Fig. 7B), in combination they potently blocked the Cdc42effect. Taken together, our results suggest vimentin reorgani-zation downstream of Cdc42, occurs through serine/threoninekinases, involving direct phosphorylation of vimentin Ser72 byPAK (Goto et al., 2002) as well as other uncharacterized sites.

Discussion

Coordination of actin, intermediate filamentsand microtubulesThe Rho-family GTPases are implicated in regulating actin-based structures which contribute to cellular morphology,changes of which underlie processes of adhesion, motility,differentiation and division. Rho and Rac are also known toregulate the reorganization of intermediate filaments (Pater-son et al., 1990; Valgeirsdottir et al., 1998) which participate incertain of these morphological events. In this study we showthat active Cdc42 can cause dramatic changes to the vimentin IFnetwork in Hela cells, involving their reorganization into densestructures in the perinuclear region. Taken with the similareffects in fibroblasts (Meriane et al., 2000), we suggest that RhoGTPases may play a general role in coordinating intermediateand microfilament systems. While activated Rho and ROKcause vimentin IF reorganization with concomitant actin stressfiber formation (Sin et al., 1998) the situation in Cdc42V12-expressing cells is quite different (cf. Fig. 1). This effect ofCdc42 is similar to that of Rac1, which induces loss of both actinstress fibers and peripheral vimentin IFs (Valgeirsdottir et al.,1998). Previous reports also link actin microfilaments and IFdynamics in areas of the cell such as membrane cortex andadhesion plaques (Bershasky et al., 1987; Green et al., 1986).Thus stress fiber integrity is not essential for IF reorganizationsince the latter can occur regardless of whether actin stressfibers are largely intact or disassembled (Fig. 1). Similarly,microfilament disassembly by cytochalasin D treatment doesnot affect the IF network (Meriane et al., 2000).

Type III vimentin IF organization depends on the micro-tubule network. Previous studies have shown that microtubule-disrupting drugs, such as vinblastine and nocodazole, causevimentin IFs to reorganize into coiled aggregates closelyassociated with the nucleus, clearly indicating that the integrityof the microtubule network affects the vimentin IForganization(as reviewed in (Goldman et al., 1996; Herrmann and Aebi,2000)). Although growth factor-induced activation of Rac/Cdc42 leads to PAK1-mediated phosphorylation of stathmin, a

Fig. 4. PAK mediates Cdc42-induced vimentin IF rearrangement.PAK phosphorylates vimentin Ser72 downstream of Cdc42 in vivo.After microinjection with plasmids encoding the HA-tagged Cdc42V12

(c ± f) or co-injection with HA-tagged PAK-WT (a, b), cells were fixedand stained with anti-HA and phospho- and site-specific anti-vimentinantibody Ser72 (b, d). Another phospho-specific anti-vimentin anti-body against Ser6 was used as control (f). Arrows indicate injected cells.Asterisks indicate uninjected neighboring cells. Bar� 10 �m.


microtubule-destabilizing protein (Daub et al., 2001), we didnot observe any major change in the microtubule content ofCdc42V12-expressing cells (Fig. 2). Cdc42 has also been shown toparticipate in microtubule-organizing center reorientation inserum-starved wounded monolayer of 3T3 fibroblasts, but suchreorientation is independent of Cdc42-induced changes in actinand MT stability (Palazzo et al., 2001). Thus reorganization ofvimentin IFs by active Cdc42 does not result from changes inthe integrity of the microtubule network. A separate study onthe differential sensitivities of stable MTs and vimentin IFs tookadaic acid similarly suggested these two filament systemswere independently regulated by phosphorylation (Vilaltaet al., 1998).

Downstream effectors of Cdc42 and IFreorganizationIt is well established that organization of IFs is modulatedthrough their phosphorylation state at serine and threonineresidues (Inagaki et al., 1996). A recent report suggested agenistein-sensitive tyrosine kinase is involved in Cdc42 orRac1-induced vimentin collapse (Meriane et al., 2000). Sincethis does not involve tyrosine phosphorylation of vimentin itselfthe significance of this observation is unclear and may resultfrom non-specific effects of the drug. Cdc42-associated kinase(ACK) is the only tyrosine kinase described so far to be an

effector of Cdc42, but over-expression of active ACK1 in HeLacells does not induce vimentin rearrangement (data not shown).

We have tested Cdc42 ×effector domain× mutants Cdc42V12/A37

and Cdc42V12/C40 which bind to different targets (the lattercannot bind PAK) (Lamarche et al., 1996), but in our handsboth mutants showed reduced capacity to induce vimentin IFcollapse (Fig. 1). Differences between our results and thoseobtained by Meriane et al., (2000) in this regard may relate toour observation that GFP, which was used as a tag in theseexperiments affects the vimentin IF network. It is likely thatmultiple effectors, including non-CRIB Cdc42-binding pro-teins participate in the process. In searching for the other kinaseeffectors that might mediate Cdc42-induced vimentin IFcollapse, we tested three effectors downstream of Cdc42:MRCK, PAK and p70 S6K. We have used two criteria toestablish a role for these in the Cdc42-induced vimentinrearrangement: firstly whether a dominant inhibitory mutantcould block the Cdc42 effect, and secondly whether the activekinase could phosphorylate vimentin in vitro and/or in vivo. Wehave found that sequential microinjection experiments (i.e.prior to expression of inhibitor protein) gave similar results toco-injection experiments.

MRCK being a ROK-like kinase effector of Cdc42 (Leunget al., 1998) represented a good candidate to directly affectvimentin IFs. Both MRCK and ROK kinases are active even in

Fig. 5. The p70 S6 kinase is involved in Cdc42-induced vimentin IFcollapse. (A) Rapamycin can inhibit Cdc42-induced vimentin IFcollapse in Hela cells. HeLa cells were incubated with rapamycin (finalconcentration 30 nM) before they were microinjected with a plasmidencoding HA-tagged Cdc42V12 and rabbit IgG. (B) Bradykinin inducesvimentin IF rearrangement. Swiss 3T3 cells grown in culture mediumwere stimulated with (a) and without (b) 100 ng/ml bradykinin for1 hour. Bar� 10 �m. (C) A phosphorylation-deficient mutant of S6Kpartially blocks bradykinin (BK)-induced vimentin IF rearrangement.Swiss 3T3 cells were microinjected with S6KT389A/�C and incubated fortwo hours. The cells were then treated with bradykinin as described in(B). Cells with condensed vimentin IFs were scored and presented aspercentage of total microinjected cells.


the absence of their partner GTPase (Leung et al., 1996, 1998)and act on similar substrates including the myosin-bindingsubunit of PP1� and MLC (Leung et al., 1998; Tan et al., 2001a).We observe that full-length MRCK (Fig. 3A, panels a and b)causes vimentin IF reorganization when overexpressed in Helacells. These results, however, do not distinguish whetherMRCK acts as a true Cdc42 effector or merely mimics theaction of the related kinase ROK, since MRCK (like ROK) isactive even in the absence of its GTPase partner Cdc42 (Leunget al., 1998). We rule out MRCK as the mediator of IFrearrangement even though introduction of exogenousMRCK can cause vimentin IF collapse: firstly the in vitrokinase assay indicates that MRCK cannot phosphorylatevimentin directly and secondly the dominant inhibitoryMRCK-TM failed to block the Cdc42-mediated rearrangementin the vimentin network (Fig. 3). Thus, vimentin represents thefirst ROK substrate that is not also acted upon by MRCK.

PAK has been suggested as responsible for cell-cycledependent phosphorylation of vimentin in its head domain atserine 72 (Yasui et al., 2001). Such phosphorylation is in partresponsible for dissolution of the filaments to allow post-mitotic separation of cells. A follow-up study (Goto et al., 2002)has shown that PAK can induce phosphorylation of Ser25,Ser38, Ser50, Ser65, and Ser72 on vimentin in vitro. Using site-and phosphorylation state-specific antibodies, it is shown thatectopic expression of constitutively active PAK in COS-7 cellsinduced similar vimentin phosphorylation (Goto et al., 2002).We show here that PAK is involved downstream in Cdc42-induced vimentin IF reorganization by the following experi-ments: (1) active PAKDE causes vimentin IF to rearrange whenoverexpressed in vivo; (2) Ser72 phosphorylated vimentinlevels increase when Cdc42 is activated in vivo. Additionally,PAK phosphorylates vimentin in vivo and in vitro on other sites(Goto et al., 2002) but these may be difficult to detect in thecontext of physiological levels of PAK.

That more than one Cdc42 effector regulates vimentin isconsistent with the PAK inhibitor (PAK-KID) causing only apartial block of Cdc42-induced vimentin IF rearrangements.This PAK fragment acts on both �- and �-PAK (Zhao et al.,1998). Crystal structures of a complex between this N-terminusinhibitory region and the C-terminus kinase domain indicatethat this fragment can act on all forms of the kinase PAK (�, �and � isoforms) (Lei et al., 2000), though it is not clear if Cdc42effectors PAK4, 5 and 6 are similarly affected. The similarity inPAK4 action on actin cytoskeleton (Abo et al., 1998) suggests itmight also act on vimentin and may provide an alternate routefor the involvement of Cdc42 in vimentin IF reorganization.This needs to be addressed.

We have found that rapamycin, a S6K inhibitor, significantlyreduces the Cdc42-induced vimentin IF collapse. Conversely,overexpression of S6K causes changes to vimentin IFs, althoughthe lack of vimentin phosphorylation in vitro suggests this is not

Fig. 6. S6K barely phosphorylates vimentin in vitro. COS 7 cells weretransfected with plasmid constructs encoding Flag-tagged S6K (lanes 2,3 and 4) or Flag-tagged ROK-KIN (lane 5). The Flag-tagged proteinswere immunoprecipitated and incubated with various GST-taggedsubstrates as indicated. The in vitro kinase assay was performed asdescribed in ™Materials and methods∫.

Fig. 7. PAK and S6K synergistically induce vimentin IF rearrange-ment. (A) Active PAK induces partial vimentin IF rearrangement. Aplasmid encoding an HA-tagged constitutively active PAKDE mutantwas microinjected or co-injected with an S6K-encoding plasmid intoHeLa cells. Four hours later, cells were fixed and stained with anti-vimentin antibodies and visualized as described previously. Bar�10 �m. (B) PAK inhibitor and truncated S6K mutant effectively blockthe Cdc42 effects on vimentin IFs. HeLa cells were co-injected withplasmids encoding HA-tagged Cdc42V12 together with plasmids encod-ing an HA-tagged PAK-KID and an inhibitory S6KT389A/�C mutant.Cells with condensed vimentin IF were scored and presented aspercentage of total microinjected cells. At least 50 injected cells werescored for each of the co-injection experiments. Bar� 10 �m.


direct. We could not detect a change in the phosphorylationstatus of vimentin at Ser6, Ser38, and Ser72 upon over-expression of S6K in HeLa cells (data not shown). Bothrapamycin and a dominant inhibitory S6KT389A/�C partiallyblocked vimentin rearrangement due to active Cdc42. S6Kitself requires phosphorylation on multiple sites to be fullyactive (Pullen and Thomas, 1997) involving participation ofCdc42 and Rac1 (Chou and Blenis, 1996). However p70 S6K hasno Cdc42/Rac-interacting/binding (CRIB) domain and has notbeen shown to bind to Cdc42 directly. A recent report suggeststhat activation of p70 S6K by Rac1 is mediated through MLK3which has a CRIB domain and shows binding to Cdc42(Burbelo et al., 1995; Lambert et al., 2002). Thus MLK3 mightmediate the involvement of S6K in vimentin reorganizationdownstream of Cdc42.

We cannot discount the fact that other activators of p70 S6Ksuch as mTOR are possibly involved in the vimentin IFreorganization. mTOR has been shown to phosphorylate p70S6K directly in vivo and in vitro (Brown et al., 1995; Saitohet al., 2002). So far mTOR has not been shown to act down-stream of Cdc42 in mammals. In S. cerevisiae, however, mTORacts upstream of RHO1 and RHO2 in regulating the actincytoskeleton (Schmidt et al., 1997). It is therefore possible thatmTOR participates in crosstalk between Cdc42 and Rhopathways.

Since S6K effects on vimentin are relatively weak, we suspectthat S6K acts indirectly on the vimentin IF network. Our datashows that when PAK and S6K were expressed an additiveeffect was observed resulting in a vimentin phenotype closelyresembling cells expressing active Cdc42 (Fig. 7). Significantly,inhibiting both kinases efficiently blocks the Cdc42 effects.

In conclusion, we have shown that Cdc42 can induce HeLavimentin IFs to undergo reorganization into bundles aroundthe nucleus. MRCK does not mediate this Cdc42-inducedeffect, but rather our data implicates PAK and S6K which incombination account for all the effects of Cdc42 on vimentinIFs. This is a novel morphological role for S6K, downstream ofCdc42, in addition to its well-established role in translation andtranscription.

Acknowledgements. We thank Ivan Tan, Xiang-Qun Chen, Cheng-GeeKoh, Zhuo-Shen Zhao, Jing-Ming Dong for providing reagents, andDr. Bor-Luen Tang for helpful comments on the manuscript. This workwas supported by the Glaxo Singapore Research Fund.

References

Abo, A., Qu, J., Cammarano, M. S., Dan, C., Fritsch, A., Baud, V.,Belisle, B., Minden, A. (1998): PAK4, a novel effector for Cdc42Hs, isimplicated in the reorganization of the actin cytoskeleton and in theformation of filopodia. EMBO J. 17, 6527 ± 6540.

Amano, M., Chihara, K., Kimura, K., Fukata, Y., Nakamura, N.,Matsuura, Y., Kaibuchi, K. (1997): Formation of actin stress fibersand focal adhesions enhanced by Rho-kinase. Science 275, 1308 ±1311.

Bershasky, A. D., Tint, I. S., Svitkina, T. M. (1987): Association ofintermediate filaments with vinculin-containing adhesion plaques offibroblasts. Cell Motil. Cytoskeleton 8, 274 ± 283.

Brown, E., Beal, P., Keith, C., Chen, J., Shin, T., Schreiber, S. (1995):Control of p70 S6 kinase by kinase activity of FRAP in vivo. Nature377, 441 ± 446.

Burbelo, P., Drechsel, D., Hall, A. (1995): A conserved binding motifdefines numerous candidate target proteins for both Cdc42 and RacGTPases. J. Biol. Chem. 270, 29071 ± 29074.

Chen, X. Q., Tan, I., Leung, T., Lim, L. (1999): The myotonic dystrophykinase-related Cdc42-binding kinase is involved in the regulation ofneurite outgrowth in PC12 cells. J. Biol. Chem. 274, 19901 ± 19905.

Chou, M. M., Blenis, J. (1996): The 70 kDa S6 kinase complexes with andis activated by the Rho family G proteins Cdc42 and Rac1. Cell 85,573 ± 583.

Chou, Y. H., Bischoff, J. R., Beach, D., Goldman, R. D. (1990): Inter-mediate filament reorganization during mitosis is mediated byp34cdc2 phosphorylation of vimentin. Cell 62, 1063 ± 1071.

Chou, Y. H., Ngai, K. L., Goldman, R. (1991): The regulation ofintermediate filament reorganization in mitosis. p34cdc2 phosphor-ylates vimentin at a unique N-terminal site. J. Biol. Chem. 266, 7325 ±7328.

Chung, J., Kuo, C. J., Crabtree, G. R., Blenis, J. (1992): Rapamycin-FKBP specifically blocks growth-dependent activation of andsignaling by the 70 kd S6 protein kinases. Cell 69, 1227 ± 1236.

Daub, H., Gevaert, K., Vandekerckhove, J., Sobel, A., Hall, A. (2001):Rac/Cdc42 and p65PAK regulate the microtubule-destabilizingprotein through phosphorylation at serine 16. J. Biol. Chem. 276,1677 ± 1680.

Dennis, P., Pullen, N., Kozma, S., Thomas, G. (1996): The principalrapamycin-sensitive p70(s6k) phosphorylation sites, T-229 and T-389are differentially regulated by rapamycin-insensitive kinase kinases.Mol. Cell. Biol. 16, 6242 ± 6251.

Evans, R. M. (1988): Cyclic AMP-dependent protein kinase-inducedvimentin filament disassembly involves modification of the N-terminal domain of intermediate filament subunits. FEBS Lett. 234,73 ± 78.

Fuchs, E., Weber, K. (1994): Intermediate filaments: structure, dynam-ics, function and disease. Annu. Rev. Biochem. 63, 345 ± 382.

Fuchs, E., Yang, Y. (1999): Crossroads on cytoskeletal highways. Cell 98,547 ± 550.

Goldman, R., Khuon, S., Chou, Y. H., Opal, P., Steinert, P. M. (1996):The function of intermediate filaments in cell shape and cytoskeletalintegrity. J. Cell Biol. 134, 971 ± 983.

Goto, H., Kosako, H., Tanabe, K., Yanagida, M., Sakurai, M., Amano,M., Kaibuchi, K., Inagaki, M. (1998): Phosphorylation of vimentin byRho-associated kinase at a unique amino-terminal site that isspecifically phosphorylated during cytokinesis. J. Biol. Chem. 273,11728 ± 11736.

Goto, H., Tanabe, H., Manser, E., Lim, L., Yasui, Y., Inagaki, M. (2002):Phosphorylation and reorganization of vimentin by p21-activatedkinase (PAK). Genes Cells 7, 91 ± 97.

Green, K. J., Talian, J. C., Goldman, R. D. (1986): Relationship betweenintermediate filaments and microfilaments in cultured fibroblasts:evidence for common foci during cell spreading. Cell Motil.Cytoskeleton 6, 406 ± 418.

Herrmann, H., Aebi, U. (2000): Intermediate filaments and theirassociates: multi-talented structural elements specifying cytoarchi-tecture and cytodynamics. Curr. Opin. Cell Biol. 12, 70 ± 90.

Inagaki, M., Nishi, Y., Nishizawa, K., Matsuyama, M., Sato, C. (1987):Site-specific phosphorylation induces disassembly of vimentin fila-ments in vitro. Nature 328, 649 ± 652.

Inagaki, N., Tsujimura, K., Tanaka, J., Sekimata, M., Kamei, Y., Inagaki,M. (1996): Visualization of protein kinase activities in single cells byantibodies against phosphorylated vimentin and GFAP. Neurochem.Res. 21, 795 ± 800.

Ishizaki, T., Naito, M., Fujisawa, K., Maekawa, M., Watanabe, N., Saito,Y., Narumiya, S. (1997): p160ROCK, a Rho-associated coiled-coilforming protein kinase, works downstream of Rho and induces focaladhesions. FEBS Lett. 404, 118 ± 124.

Kosako, H., Amano, M., Yanagida, M., Tanabe, K., Nishi, Y., Kaibuchi,K., Inagaki, M. (1997): Phosphorylation of glial fibrillary acidicprotein at the same sites by cleavage furrow kinase and Rho-associated kinase. J. Biol. Chem. 272, 10333 ± 10336.

Kosako, H., Goto, H., Yanagida, M., Matsuzawa, K., Fujita, M., Tomono,Y., Okigaki, T., Odai, H., Kaibuchi, K., Inagaki, M. (1999): Specificaccumulation of Rho-associated kinase at the cleavage furrow duringcytokinesis: cleavage furrow-specific phosphorylation of intermedi-ate filaments. Oncogene 18, 2783 ± 2788.


Kozma, R., Ahmed, S., Best, A., Lim, L. (1995): The Ras-relatedproteins Cdc42Hs and Bradykinin promote formation of peripheralactin microspikes and filopodia in Swiss 3T3 fibroblasts. Mol. Cell.Biol. 15, 1942 ± 1952.

Kusubata, M., Tokui, T., Matsuoka, Y., Okumura, E., Tachibana, K.,Hisanaga, S., Kishimoto, T., Yasuda, H., Kamijo, M., Ohba, Y., et al.(1992): p13suc1 suppresses the catalytic function of p34cdc2 kinasefor intermediate filament proteins, in vitro. J. Biol. Chem. 267,20937 ± 20942.

Lamarche, N., Tapon, N., Stowers, L., Burbelo, P. D., Aspenstrom, P.,Bridges, T., Chant, J., Hall, A. (1996): Rac and Cdc42 induce actinpolymerization and G1 cell cycle progression independently ofp65PAK and the JNK/SAPK MAP kinase cascade. Cell 87, 519 ± 529.

Lamarche-Vane, N., Hall, A. (1998): CdGAP, a novel proline-richGTPase-activating protein for Cdc42 and Rac. J. Biol. Chem. 30,29172 ± 29177.

Lambert, J., Karnoub, A., Graves, L., Campbell, S., Der, C. (2002): Roleof MLK3-mediated activation of p70 S6 kinase in Rac1 transforma-tion. J. Biol. Chem. 277, 4770 ± 4777.

Lei, M., Lu, W., Meng, W., Parrini, M., Eck, M., Mayer, B., Harrison, S.(2000): Structure of PAK1 in an autoinhibited conformation reveals amultistage activation switch. Cell 102, 387 ± 397.

Leung, T., Chen, X. Q., Manser, E., Lim, L. (1996): The p160 RhoA-binding kinase ROK alpha is a member of a kinase family and isinvolved in the reorganization of the cytoskeleton. Mol. Cell. Biol. 16,5313 ± 5327.

Leung, T., Chen, X. Q., Tan, I., Manser, E., Lim, L. (1998): Myotonicdystrophy kinase-related Cdc42-binding kinase acts as a Cdc42effector in promoting cytoskeletal reorganization. Mol. Cell. Biol. 18,130 ± 140.

Leung, T., Manser, E., Tan, L., Lim, L. (1995): A novel serine/threoninekinase binding the Ras-related RhoAGTPase which translocates thekinase to peripheral membranes. J. Biol. Chem. 270, 29051 ± 29054.

Manser, E., Huang, H. Y., Loo, T. H., Chen, X. Q., Dong, J. M., Leung,T., Lim, L. (1997): Expression of constitutively active �-PAK revealseffects of the kinase on actin and focal complexes. Mol. Cell. Biol. 17,1129 ± 1143.

Martinez-Quiles, N., Rohatgi, R., Anton, I., Medina,M., Saville, S., Miki,H., Yamaguchi, H., Takenawa, T., Hartwig, J., Geha, R., Ramesh, N.(2001): WIP regulates N-WASP-mediated action polymerization andfilopodium formation. Nat. Cell Biol. 3, 484 ± 491.

Meriane, M., Mary, S., Comunale, F., Vignal, E., Gauthier-Rouviere, C.(2000): Cdc42Hs and Rac1 GTPases induce the collapse of thevimentin intermediate filament network. J. Biol. Chem. 275, 33046 ±33052.

Mukai, H., Toshimori, M., Shibata, H., Kitagawa, M., Shimakawa, M.,Miyahara, M., Sunakawa, H., Ono, Y. (1996): PKN associates andphosphorylates the head-rod domain of neurofilaments. J. Biol.Chem. 271, 9816 ± 9822.

Nobes, C. D., Hall, A. (1995): Rho, rac and cdc42 GTPases: regulators ofactin structures, cell adhesion and motility. Biochem. Soc. Trans. 23,456 ± 459.

Palazzo, A., Joseph, H., Chen, Y., Dujardin, D., Alberts, A., Pfister, K.,Valle, R., Gundersen, G. (2001): Cdc42, dynein, and dynactin regulateMTOC reorientation independent of Rho-regulated microtubulestabilization. Curr. Biol. 11, 1536 ± 1541.

Paterson, H. F., Self, A. J., Garrett, M. D., Just, I., Aktories, K., Hall, A.(1990): Microinjection of recombinant p21rho induces rapid changesin cell morphology. J. Cell Biol. 111, 1001 ± 1007.

Pearson, R., Dennis, P., Han, J., Williamson, N., Kozma, S., Wettenhall,R., Thomas, G. (1995): The principal target of rapamycin-inducedp70s6k inactivation is a novel phosphorylation site within a conservedhydrophobic domain. EMBO J. 14, 5279 ± 5287.

Pullen, N., Thomas, G. (1997): The modular phosphorylation andactivation of p70s6k. FEBS Lett. 410, 78 ± 82.

Saitoh, M., Pullen, N., Brennan, P., Cartrell, D., Dennis, P., Thomas, G.(2002): Regulation of an activated S6 kinase 1 variant reveals a novelmammalian taget of rapamycin phosphorylation site. J. Biol. Chem.277, 20104 ± 20112.

Schmidt, A., Bickle, M., Beck, T., Hall, M. (1997): The yeastphosphatydylinositol kinase homolog TOR2 activates RHO1 andRHO2 via the exchange factor ROM2. Cell 88, 531 ± 542.

Sells, M. A., Knaus, U. G., Bagrodia, S., Ambrose, D. M., Bokoch, G. M.,Chernoff, J. (1997): Human p21-activated kinase (Pak1) regulatesactin organization in mammalian cells. Curr. Biol. 7, 202 ± 210.

Sin, W. C., Chen, X. Q., Leung, T., Lim, L. (1998): RhoA-binding kinasealpha translocation is facilitated by the collapse of the vimentinintermediate filament network. Mol. Cell. Biol. 18, 6325 ± 6339.

Tan, I., Ng, C., Lim, L., Leung, T. (2001a): Phosphorylation of a novelmyosin binding subunit of protein phosphatase 1 reveals a conservedmechanism in the regulation of actin cytoskeleton. J. Biol. Chem. 276,21209 ± 21216.

Tan, I., Seow, K. T., Lim, L., Leung, T. (2001b): Intermolecular andintramolecular interactions regulate catalytic activity of myotonicdystrophy kinase-related Cdc42-binding kinase�. Mol. Cell. Biol. 21,2767 ± 2778.

Tokui, T., Yamauchi, T., Yano, T., Nishi, Y., Kusagawa, M., Yatani, R.,Inagaki, M. (1990): Ca2(�)-calmodulin-dependent protein kinase IIphosphorylates various types of non-epithelial intermediate filamentproteins. Biochem. Biophys. Res. Commun. 169, 896 ± 904.

Valgeirsdottir, S., Claesson-Welsh, L., Bongcam-Rudloff, E., Hellman,U., Westermark, B., Heldin, C. H. (1998): PDGF induces reorganiza-tion of vimentin filaments. J. Cell Sci. 111, 1973 ± 1980.

Vilalta, P. M., Zhang, L., Hamm-Alvarez, S. F. (1998): A novel taxol-induced vimentin phosphorylation and stabilization revealed bystudies on stable microtubules and vimentin intermediate filaments.J. Cell Sci. 111, 1841 ± 1852.

Yasui, Y., Goto, H., Matsui, S., Manser, E., Lim, L., Nagata, K.-i.,Inagaki, M. (2001): Protein kinases required for segregation ofvimentin filaments in mitotic process. Oncogene 20, 2868 ± 2876.

Zhao, Z. S., Manser, E., Chen, X. Q., Chong, C., Leung, T., Lim, L.(1998): A conserved negative regulatory region in �PAK: Inhibitionof PAK kinases reveals their morphological roles downstream ofCdc42 and Rac1. Mol. Cell. Biol. 18, 2153 ± 2163.


Documents

Vimentin intermediate filament reorganization by Cdc42: Involvement of PAK and p70 S6 kinase