Cdc42Hs and Rac1 GTPases Induce the Collapse of the Vimentin Intermediate Filament Network*

In this study we show that expression of active Cdc42Hs and Rac1 GTPases, two Rho family members, leads to the reorganization of the vimentin intermediate filament (IF) network, showing a perinuclear collapse. Cdc42Hs displays a stronger effect than Rac1 as 90% versus 75% of GTPase-expressing cells show vimentin collapse. Similar vimentin IF modifications were observed when endogenous Cdc42Hs was activated by bradykinin treatment, endogenous Rac1 by platelet-derived growth factor/epidermal growth factor, or both endogenous proteins upon expression of active RhoG. This reorganization of the vimentin IF network is not associated with any significant increase in soluble vimentin. Using effector loop mutants of Cdc42Hs and Rac1, we show that the vimentin collapse is mostly independent of CRIB (Cdc42Hs or Rac-interacting binding)-mediated pathways such as JNK or PAK activation but is associated with actin reorganization. This does not result from F-actin depolymerization, because cytochalasin D treatment or Scar-WA expression have merely no effect on vimentin organization. Finally, we show that genistein treatment of Cdc42 and Rac1-expressing cells strongly reduces vimentin collapse, whereas staurosporin, wortmannin, LY-294002,R p-cAMP, or RII, the regulatory subunit of protein kinase A, remain ineffective. Moreover, we detected an increase in cellular tyrosine phosphorylation content after Cdc42Hs and Rac1 expression without modification of the vimentin phosphorylation status. These data indicate that Cdc42Hs and Rac1 GTPases control vimentin IF organization involving tyrosine phosphorylation events.

Although the role of Rho GTPases on actin cytoskeleton organization has been extensively studied, little is known on their effects on one of the other major component of the cytoskeleton of eucaryotic cells, the intermediate filaments (IFs). IFs consist of a heterogeneous tissue-specific family of proteins, which are prevalent in the perinuclear region and extend radially through the cytoplasm, eventually forming close associations with the cell surface, concentrated in regions containing desmosomes (cadherin-mediated cell-to-cell junctions), hemidesmosomes (integrin-mediated adhesive junctions), and other types of adhesion sites (18,19). Cytoskeletal IF also interact with other cytoskeletal elements such as microtubules and microfilaments (20 -23). Interaction with plasma membrane and other cytoskeletal elements involve a number of IF-associated proteins that are essential for maintaining the integrity of IF network (24). Recently, it has been shown that, at least in actively growing cells, IF are dynamic structures. IF phosphorylation appears to be one of the most predominant biochemical events in coordinating intracellular organization of the IF network (25). Cytoplasmic IF disassembled when phos-phorylated by protein kinase A, protein kinase C, calcium calmodulin kinase II (CaMKII), and Cdc2 kinases. Interestingly, protein kinase N (PKN), a protein kinase activated by Rho, associates and phosphorylates a subunit of neuron-specific intermediate filament, NFL (26) and ROK␣ (RhoA-binding kinase ␣) phosphorylates glial fibrillary acidic protein (GFAP) (27) and vimentin (28).
In the present study we analyzed the possible role of two Rho GTPases, Rac1 and Cdc42Hs, in regulating the vimentin IF organization. We show that expression of active Rac1 and Cdc42Hs (V12 or L61 mutants) modified vimentin IF distribution: the normally well-spread distribution of IF became dramatically reorganized around the nucleus. Furthermore, Cdc42Hs and Rac1-dependent vimentin collapse was both different and more pronounced than RhoA-dependent vimentin reorganization. This vimentin collapse was observed following activation of endogenous Cdc42Hs by bradykinin and Rac1 by PDGF/EGF or after active RhoG expression. Because the IF network was closely associated with the actin microfilaments, the dynamics of which was highly regulated by Rho GTPases, we expressed two effector loop mutants of Rac1 and Cdc42Hs that had a differential effect on F-actin organization. Interestingly, the Y40C mutants of Cdc42Hs and Rac1, which still induced F-actin rearrangements, promoted the collapse of IF as efficiently as did the V12 or L61 mutants. The F37A mutants of Cdc42Hs and Rac1, which had lost their ability to induce filopodia and membrane ruffling, respectively, no longer induced the collapse of IF. In addition, by using various drugs known to have kinase inhibitory activity, we show that vimentin IF reorganization involved tyrosine phosphorylation events.
Immunocytochemistry-18 h after transfection, cells were fixed for 5 min in 3.7% formalin (in PBS) followed by a 2-min permeabilization in 0.1% Triton X-100 (in PBS) and incubation in PBS containing 0.1% bovine serum albumin. Expression of GFP-tagged proteins was visualized directly. Cells were stained for vimentin distribution using a mouse monoclonal anti-vimentin (Sigma, France) (1:200 dilution), followed by incubation with affinity-purified tetramethyl-rhodamin-5 (and 6) isothiocyanate-conjugated goat anti-mouse antibody (Cappel-ICN) (1:40 dilution). Cells were stained for F-actin using coumarin phenyl isothiocyanate-conjugated phalloidin (Sigma, France) and for phosphotyrosine epitopes using the 4G10 monoclonal antibody. Expression of Myc epitope-tagged proteins was visualized after a 60-min incubation with 9E10 anti-Myc monoclonal antibody (gift from D. Mathieu, Montpellier, France) (one-half dilution in PBS/bovine serum albumin), followed by incubation with affinity-purified fluorescein isothiocyanate-conjugated goat anti-mouse antibody (Cappel, ICN). Cells were washed in PBS, mounted in Mowviol (Aldrich, Milwaukee, WI), and observed using a DMR B Leica microscope using a 40ϫ (NA 1.00) or 63ϫ (NA 1.32) planapochromatic lens. Images obtained were captured with a Micro-Max 1300 Y/HS (B/W) cooled (Ϫ10°C) charge-coupled device camera as 16-bit images, and using a MetaMorph (v.4.11) control program (RS-Princeton Instruments) run by a PC-compatible microcomputer. Images were saved in TIFF format (16 bit) and subsequently adapted as TIFF 8-bit format after they were opened with Adobe Photoshop for processing and mounting with Adobe Illustrator.
Extraction of Soluble and Filamentous Cellular Proteins Using High Salt and Triton X-100 -Two distinct methods were used to evaluate the amount of soluble vimentin in cell extracts from control REF-52 cells or REF-52 cells transfected with GFP-Cdc42HsV12 or GFP-Rac1V12. After washing with ice-cold PBS, cells were scraped, collected, and centrifuged at 10,000 rpm. Pellets were resuspended in either a lysis buffer containing 1% Nonidet P-40, 10% glycerol, 20 mM Hepes, pH 7.5, 150 mM NaCl, 1 mM phenylmethylsulfonyl fluoride, 20 mM NaF, a protease inhibitor mixture (Sigma), 100 M Na 3 VO 4 as described previously by Valgeirsdottir et al. (29), or 1% Triton, 50 mM MES, 600 mM KCl, 10 mM MgCl 2 . For the second method, the first soluble pool obtained underwent three successive extractions of 10 min in ice-cold lysis buffer (30). The soluble and insoluble fractions obtained with both methods were loaded onto 10% polyacrylamide gels and then transferred onto nitrocellulose. Initial lysates were normalized for protein content (BCA, Sigma). Membranes were saturated in 8% milk in Tris-HCl, pH 7.5, containing 0.1% Tween and subsequently incubated with a mouse monoclonal antibody directed against vimentin (clone V9, Sigma) (1/ 1000 dilution) followed by peroxidase-conjugated anti-mouse antibody (Amersham Pharmacia Biotech) (1/2000 dilution). After extensive washing, membranes were incubated with chemiluminescence reagent (ECL, PerkinElmer Life Sciences) and analyzed with a PhosphorImager (Molecular Dynamics).
Detection of Tyrosine-phosphorylated Proteins-Total cell lysates from untransfected or cells transfected with either empty pEGFPN1 vector (MOCK), Cdc42HsV12, or Rac1V12 were obtained by the addition of 1% boiling SDS, 10 mM Tris-HCl, pH 7.4. After scraping, samples (30 g of protein) were loaded onto a 10% polyacrylamide gel and then transferred onto nitrocellulose. Membranes were treated as described above and incubated with an anti-phosphotyrosine antibody (4G10, 1/200 dilution) followed with peroxidase-conjugated anti-mouse antibody (Amersham Pharmacia Biotech) (dilution 1/2000). After extensive washing, membranes were incubated with chemiluminescence reagent (ECL, PerkinElmer Life Sciences) and analyzed with a PhosphorImager (Molecular Dynamics).
Detection of Tyrosine-phosphorylated Vimentin-Untransfected cells or cells transfected with either empty pEGFPN1 vector (MOCK), Cdc42HsV12, or Rac1V12 were lysed for 20 min in ice-cold modified radioimmune precipitation buffer (1% Triton X-100, 10 mM sodium pyrophosphate, 0.1% SDS, 1% deoxycholate, 10% glycerol, 20 mM Tris-HCl, pH 7.4, 150 mM NaCl, 2.5 mM EDTA) supplemented with 20 mM ␤-glycerophosphate, 1 mM phenylmethylsulfonyl fluoride, 20 mM NaF, 100 M Na 3 VO 4 . Extracts were immunoprecipitated using a mouse monoclonal anti-vimentin antibody (V9, 1/100 dilution), separated by a 10% polyacrylamide gel, and then transferred onto nitrocellulose. Membranes were probed with the 4G10 anti-phosphotyrosine antibody as described above. In addition, the activated RhoA mutant, RhoAV14, was also expressed. 18 h later, cells were fixed and immunostained with antibodies directed against vimentin and for filamentous actin (F-actin) using rhodamine-conjugated phalloidin (Fig. 1A). Expression of each of these GTPases induced the formation of specific actin-driven structures. Cdc42Hs (Fig. 1A, panel a) induced filopodia (panel b), Rac1 (panel e) induced lamellipodia/ruffles (panel f), and RhoA (panel i) induced stress fibers (panel j) as previously reported (4,12). Interestingly, expression of these GTPases also led to a marked change in the organization of the vimentin network (panels c and g). Indeed, although in control nontransfected cells vimentin showed a well-spread distribution from the perinuclear region to the cell periphery (see control nontrans-fected cells in panels c, g, and k), in Cdc42Hs-or Rac1-expressing cells, vimentin accumulated at the perinuclear area. Vimentin IF organization was also extensively modified after active RhoA expression (panel k), which is in total agreement with a previous report (32). Corresponding Normarski images were shown to precisely localize cell margins (panels d, h, and l). Similar effects of Cdc42Hs, Rac1, and RhoA on vimentin distribution were also observed in Hs-68 human fibroblasts (Fig. 1B). More than 90% of Cdc42Hs-expressing cells and around 75% of Rac1-expressing cells showed extensive IF reorganization, with the same efficiency in REF-52 and Hs-68 cells. In contrast, less than 50% of RhoA-expressing cells presented a modified vimentin IF network. For both cell types, no significant modification in vimentin IF distribution (less than 5%) was detected upon transfection with pEGFPC1.

Expression of Active Cdc42Hs and Rac1 GTPases Leads to the Reorganization of the Vimentin IF Network-To
Vimentin IF Collapse following Activation of Endogenous Cdc42Hs and Rac1-Bradykinin (Bdk) has been shown to activate Cdc42Hs and induce filopodia formation, whereas platelet-derived growth factor (PDGF) and epidermal growth factor (EGF) are two mitogens that activate Rac1 and produce lamellipodia (4, 10, 12). To activate endogenous Rac1 and Cdc42Hs proteins, we thus stimulated REF-52 cells with Bdk or PDGF/ EGF for different periods of time. Cells were then fixed and immunostained for F-actin to control endogenous Cdc42Hs and Rac1 activation and for vimentin IF distribution (Fig. 2). In control unstimulated cells, both actin microfilaments and vimentin IF formed a well-spread network from the perinuclear region to the cell periphery (panels a and b). 15 min after Bdk addition, thin F-actin-rich filopodial extensions were detected at the edges of the cells (panel c). Concomitantly, extensive Because previous work reported that IF reorganization could be associated with changes in vimentin solubility (34), we analyzed the amount of vimentin extractable using either a mild Nonidet P-40-containing lysis buffer (29) or high salt and Triton X-100-containing lysis buffer (30) after expression of active Cdc42Hs and Rac1 GTPases. The amount of vimentin present in a soluble form extractable with a mild Nonidet P-40-containing buffer was not modified after expression of these GTPases (Fig. 2B) as was the case when using the high salt and Triton X-100-containing lysis buffer (data not shown).
Taken together, these data show that expression of active Cdc42Hs and Rac1 or activation of endogenous Cdc42Hs and Rac1 all led to vimentin IF perinuclear reorganization without modification of the solubility of vimentin.
Vimentin Reorganization Is Obtained Mainly with Y40 Mutants of Cdc42Hs and Rac1-We next investigated the pathways controlled by Rac1 and Cdc42Hs responsible for vimentin IF collapse. We used effector loop mutants of GTPases previously shown to differentially bind and activate downstream effectors (35-37) (Fig. 3). The Y40C mutants of Cdc42Hs and Rac1 had lost their ability to interact with CRIB (Cdc42Hs or Rac-interacting binding) motif-containing proteins and did not activate PAK-1 and JNK activity, but they still induced cortical F-actin polymerization, filopodia, and membrane ruffling, respectively. Conversely, the F37A mutants of Cdc42Hs and Rac1 still bound the CRIB motif-containing proteins, activating PAK and JNK, but were less efficient for inducing filopodia or membrane ruffling. Cells expressing the Y40C mutants of Cdc42Hs and Rac1 showed vimentin IF reorganization comparable to the one observed in active Rho GTPases-expressing cells (V12 or L61 mutants) (compare Fig. 1B with Fig. 3). Under the same conditions, expression of F37A mutants of Cdc42Hs and Rac1 did not significantly affect vimentin IF distribution, because only 10 -20% of expressing cells showed weak vimentin reorganization.
Interaction of IF with microfilaments is thought to regulate IF organization in vivo (23). Rho GTPases of the Rac/Cdc subgroup are well-known key regulators of actin microfilaments (1), suggesting that the vimentin IF collapse we observed might result from Rac1-or Cdc42-dependent F-actin reorganization. Two subpopulations of actin structures were affected by Rac1 and Cdc42Hs expression: submembranous cortical actin, which was extensively modified to produce ruffles/lamellipodia and microvilli/filopodia, and stress fibers, which were mostly depolymerized. To test the existence of a relationship between stress fibers, depolymerization, and vimentin reorganization, REF-52 cells were treated with cytochalasin D, an F-actindepolymerizing drug, fixed, and stained for F-actin and vimentin IF distribution (Fig. 4A). 30 min after cytochalasin D addition, the level of F-actin staining became barely detectable (panel b), whereas the vimentin IF network remained unaffected (panel c). We next transfected REF-52 cells with Scar-WA, a mutant form of the Arp2/3-interacting protein Scar, which prevents assembly of F-actin structures (38). As for cytochalasin-treated cells, expression of Myc-tagged Scar-WA  Fig. 4B (panel d) is a cell expressing only GFP Rac1V12 showing a collapsed vimentin. These data show that, although an overall F-actin depolymerization did not affect vimentin organization, inhibition of Cdc42Hs-and Rac1-dependent F-actin modification impaired Cdc42Hs and Rac1-induced vimentin redistribution.
Tyrosine Phosphorylation Inhibition Prevents Vimentin IF Collapse Induced by Cdc42Hs and Rac1-Phosphorylation has been shown to be a major regulatory pathway coordinating intracellular organization of the IF network, so we used various drugs having known kinase inhibitory activity to address the involvement of phosphorylation events in vimentin IF collapse in Cdc42Hs-or Rac1-expressing cells (Table I). First, we analyzed the effects of inhibiting protein kinase A activity, a protein that induced both the collapse of vimentin IF and F-actin reorganization (30) by two ways: microinjection of the RII regulatory subunit of protein kinase A in GFP-Cdc42HsV12-or GFP-Rac1V12-expressing cells or addition of a cAMP antagonist R p -cAMP to GFP-Cdc42HsV12-or GFP-Rac1V12-expressing cells. Second, we treated GFP-Cdc42HsV12-and GFP-Rac1V12-expressing cells with the PI3Kinhibitors LY-294002 and wortmannin (39). Third, we used staurosporin, an inhibitor of various serine/threonine kinases. In all cases, no modification of the vimentin IF reorganization induced by Cdc42Hs and Rac1 was observed. Finally, we treated GFP-Cdc42HsV12-and GFP-Rac1V12-expressing cells with the tyrosine kinase inhibitor genistein for 15 to 120 min (40). As shown in Fig. 5, cells treated with this compound displayed a less pronounced reorganization of the vimentin IF network upon Cdc42Hs and Rac1 expression (panels a/b and c/e) as compared with Fig. 1. This inhibitory effect on vimentin IF reorganization was not correlated with F-actin modification, because Cdc42Hs-and Rac1-expressing cells showed filopodia and lamellipodia, respectively, as well as reduction of stress fibers content (panels c and f).
These data show that genistein-dependent tyrosine kinase inhibition impaired the vimentin collapse induced by Cdc42Hs and Rac1.
Increase of Tyrosine Phosphorylation after Cdc42Hs and Rac1 Expression-To further study whether Cdc42Hs or Rac1 expression leads to increased tyrosine phosphorylation, GFP-Cdc42HsV12-or GFP-Rac1V12-expressing cells were stained with 4G10 anti-phosphotyrosine antibody (Fig. 6A). As expected, focal adhesions are modified onto focal contacts by Cdc42Hs and Rac1 expression. In addition, Cdc42HsV12-expressing cells (panel a) and Rac1V12-expressing cells (panel c) showed increased phosphotyrosine staining compared with nonexpressing cells (panels b and d). This was observed in 50% of Cdc42HsV12-expressing cells and in 40% of Rac1V12-expressing cells. Additionally, cell extracts from control cells, MOCK-transfected cells, or Cdc42HsV12-or Rac1V12-expressing cells were separated by gel electrophoresis, transferred to nitrocellulose, and immunoblotted with 4G10 anti-phosphotyrosine antibody (Fig. 6B). Although in control cells a high steady-state level of tyrosine-phosphorylated proteins was observed, three additional tyrosine-phosphorylated proteins were detected in cells extracts from Cdc42Hs-and Rac1-expressing cells (marked with arrows). We finally analyzed whether vimentin might be directly phosphorylated in Cdc42Hs-and Rac1V12-expressing cells. Vimentin was immunoprecipitated from control untransfected REF-52 cells, empty pEGFPN1 vector (MOCK), or Cdc42Hs-or Rac1-expressing REF-52 cells, and tyrosine-phosphorylated vimentin was analyzed by immunoblotting with the 4G10 antiphosphotyrosine antibody (Fig. 6C). No marked changes in  Taken together these data show that, although increased tyrosine phosphorylation was observed after Cdc42Hs and Rac1 expression, no direct modification in vimentin phosphorylation was detected.

DISCUSSION
In this study, we showed that expression of constitutively active Cdc42Hs and Rac1 led to reorganization of the vimentin IF network. Activation of endogenous Cdc42Hs and Rac1 by bradykinin or PDGF/EGF treatment, respectively, or active RhoG expression, which activates both Cdc42Hs and Rac1 (13), also induced vimentin IF collapse. No modification on vimentin solubility was detected. By using effector loop mutants of Cdc42Hs and Rac1 we found that mainly the Y40C mutants led to the vimentin IF collapse. Cdc42Hs and Rac1 expression induced an increased tyrosine phosphorylation and addition of tyrosine kinase inhibitors led to a strong diminution of Cdc42Hs-and Rac1-induced vimentin IF collapse. Thus we suggest that Cdc42Hs and Rac1 expression leads to specific tyrosine kinase activation which in turn induces vimentin IF collapse.
Work during the past decade has established important roles of Rho GTPases in regulating the organization of the actin cytoskeleton. Using mammalian fibroblasts as model systems but also leukocytes and neuronal cells, Cdc42Hs and Rac1 have been shown to trigger the formation of filopodia and lamellipodia, respectively, whereas RhoA triggers the assembly of focal contacts and stress fibers (4). In this study we showed that Cdc42Hs and Rac1 elicit also the modification of another major cytoskeleton component, the vimentin IF network, which instead of being well-spread and distributed from the perinuclear region to the cell periphery becomes collapsed all around the nucleus. This collapse has been observed under three different experimental conditions: 1) expression of constitutive active Cdc42Hs and Rac1, 2) activation of endogenous Cdc42Hs by bradykinin (12) or Rac1 by PDGF/EGF (10), and 3) expression of active RhoG, which activates both Cdc42Hs and Rac1 (13). It has previously been reported that PDGF treatment of porcine aortic endothelial cells provokes a marked vimentin IF collapse as does expression of constitutively active Rac1 (29). Constitutively active RhoAV14-injected cells also show a collapsed IF network into irregular thick bundles within the cytoplasm (32), which differs from the Cdc42Hs-and Rac1-dependent IF collapse described above. Interestingly, blocking the RhoA signaling pathway had no effect on Cdc42Hs-and Rac1-induced vimentin collapse, suggesting that RhoA and Cdc42HS/Rac1 act through independent pathways (data not shown).
Because the three components of the cytoskeleton, namely F-actin microfilaments, microtubules, and intermediate filaments, have been shown to be connected physically (23), the reorganization of the vimentin IF after Cdc42Hs or Rac1 expression could result from a deregulated interaction between IF and the actin cytoskeleton, because both Cdc42Hs and Rac1 induce local and peripheral actin polymerization (ruffles/lamellipodia and filopodia, respectively) and a reduction in stress fibers. This is consistent with our data showing that mainly the Y40C Cdc42Hs and Rac1 mutants, which are still able to modify the actin cytoskeleton, elicit the vimentin collapse. Microfilament disassembly upon cytochalasin D treatment or Scar-WA expression did not affect the IF network, suggesting that the IF collapse is unlikely to result from the disassembly of the stress fibers. We are unable to test only the consequence of local cortical actin polymerization on vimentin IF organization. However, in the absence of cortical F-actin polymerization upon cytochalasin D treatment or Scar-WA expression, Cdc42Hs and Rac1 do not affect vimentin reorganization, again suggesting a close correlation between actin and IF cytoskeleton. A recent study of fibroblasts expressing a chimeric GFPvimentin reveals that both the typical IF ends and short filamentous structures termed "vimentin squiggles" are frequently detected at the edge of the cell according to a pattern similar to focal adhesions (41). Because the expression of activated Cdc42Hs or Rac1 proteins leads to a redistribution of focal adhesions into focal contacts (4), this might therefore modify the well-spread vimentin IF distribution. Interestingly, the use of GFP-tagged vimentin showed that the collapse of IF bundles did not necessarily involve reassembly near the nucleus but rather the network being pushed back into the perinuclear region (42).
If organization has been reported to be mainly regulated by phosphorylation (25). We have thus examined whether Cdc42Hs-and Rac1-dependent IF vimentin collapse might be mediated by protein kinase activation. We show that, although inhibition of PKA, PI3K, and staurosporin-serine/threonine kinases did not impair vimentin collapse, a genistein-sensitive protein-tyrosine kinase is involved in the pathway leading to this IF reorganization. Among the Cdc42Hs and Rac1 effectors described so far, only one protein-tyrosine kinase family has been described (43). These Cdc42Hs-associated kinases (ACK1 and -2) might only account for Cdc42Hs-induced vimentin reorganization, because Rac1 does not bind to these proteins. Whatever the protein-tyrosine kinase activated by Cdc42Hs and Rac1, this protein is not responsible for a direct vimentin phosphorylation. In this respect, the Cdc42Hs-and Rac1-dependent vimentin reorganization again differs from the RhoAdependent collapse, because ROK or protein kinase N (PKN), two RhoA effectors, directly phosphorylate vimentin (26,28,44). In addition, Rho kinase inhibition did not modify Cdc42Hsand Rac1-dependent vimentin collapse, discounting any Rho kinase involvement in this process (data not shown). Various targets for such protein-tyrosine kinases might be proposed, such as IF-associated proteins (45,46) or several components of focal adhesions, including vinculin, talin, tensin, and paxillin. Tyrosine kinases such as pp125FAK or p60v-src have been found associated with focal adhesions (47).
Although the absence of a clear effect of vimentin knockout mice does not help in the understanding of vimentin IF function (48), one can propose that Cdc/Rac-induced vimentin collapse contributes to the mechanisms of cell movement. Indeed, the pseudopod of crawling cells is in general devoid of filamentous IF (49), and Cdc42Hs and Rac1 GTPase are known to induce cell motility (4). An attractive function for this collapse might also be the liberation of vimentin IF-associated proteins, which participate in Cdc42Hs-and Rac1-dependent pathways. In this line, ROK␣ has been shown to be associated with the vimentin IF network and to translocate to the cell periphery upon vimentin IF collapse (28). Interestingly, the protein kinase Src, also associated with vimentin IF (50), is translocated at the cell periphery after PDGF treatment or Rac1 expression (51). Further studies on the determination of target protein kinases may help to define and elucidate the functional significance of IF collapse in living cells.