INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Cell migration relies both on the protrusion of motile organelles such as lamellipodia and on the adhesion of cells to the extracellular matrix via specialized sites termed focal adhesions, which link the extracellular substrate to the actin cytoskeleton (reviewed in Horwitz and Parsons, 1999).

Zyxin is found at focal adhesions, cell-cell contacts, and along stress fibers, where it is suggested to play a central role in the regulation of actin dynamics. In addition, several observations indicate a similar role for zyxin at the tips of lamellipodial protrusions. First, the amino terminus of zyxin harbors polyproline-rich stretches, providing binding sites for the SH3 domain of the guanine nucleotide exchange factor for Rho GTPases, Vav (Hobert et al., 1996), and for the EVH1 domain of the Ena/vasodilator-stimulated phosphoprotein (VASP) family proteins VASP and Mena (Gertler et al., 1996; Niebuhr et al., 1997; Drees et al., 2000). Ena/VASP proteins are recruited to the distal edge of lamellipodia in amounts that directly correlate with protrusion rates (Rottner et al., 1999b). Second, ectopical expression of a zyxin-mutant harboring the CAAX membrane-targeting motif causes the induction of actin-rich surface projections (Golsteyn et al., 1997), an effect that is less prominent with a zyxin-CAAX-mutant lacking functional Ena/VASP-binding sites (Drees et al., 2000). Third, microinjection of a zyxin-derived peptide, which blocks the interaction of zyxin with -actinin, causes the retraction of the cell edge and perturbs cell migration and spreading (Drees et al., 1999). These findings have lead to the proposal that zyxin might serve as a linker, recruiting proteins that contribute to the regulation of actin polymerization at the plasma membrane in a spatially and temporally regulated manner (Goldsteyn et al., 1997; Beckerle, 1998; Jay, 2000; Holt and Koffer, 2001).

Given that VASP is sharply localized to the tips of protruding lamellipodia and filopodia (Rottner et al.,1999b), it would be expected from the above-mentioned observations that zyxin shares the same localization. With the use of green fluorescent protein (GFP)-tagged constructs of zyxin as well as a new antibody probe, we found that this was not the case. Our findings, presented here, also provide novel insights into the differential dynamics of zyxin, vinculin, paxillin, and VASP during focal adhesion assembly/disassembly.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Generation of Antibodies and Western Blotting

Purified zyxin from human platelet extracts was kindly provided by Dr. M. Reinhard (Medizinische Universitätsklinik, Würzburg, Germany) and used to produce monoclonal antibodies in mice as described (Niebuhr et al., 1998). Hybridoma supernatants were screened by immunofluorescence microscopy on HeLa cells and in parallel by Western blot analysis by using total extracts of HeLa cells. By subclass analysis, the monoclonal antibodies, designated 164D4 and 184A3, were identified as IgG1. Spot synthesis was performed according to Frank (1992) with an Abimed ASP 222 automated SPOT robot. Mapping of the epitopes of the monoclonal antibodies was performed as described (Niebuhr et al., 1998).

Total SDS extracts of HeLa cells were separated on 10% SDS gels and electrophoretically transferred to polyvinylidene difluoride membranes. After incubation with the appropriate antibodies, signals were visualized with the use of an enhanced chemiluminescence detection kit (Amersham Pharmacia Biotech AB, Uppsala, Sweden).

Enhanced Green Fluorescent Protein (EGFP)-Constructs

The cDNAs of human zyxin (Macalma et al., 1996) and human paxillin (Salgia et al., 1995) were kindly provided by Dr. D. von der Ahe (Kerckhoff Klinik, Max-Planck Institute, Bad Nauheim, Germany) and Dr. R. Salgia (Harvard Medical School, Boston, MA), respectively. The full-length human zyxin sequence was amplified by polymerase chain reaction with primers containing the restriction sites BamHI/EcoRI and cloned into the pEGFP-N1 vector (CLONTECH, Palo Alto, CA). The construct was verified by DNA sequencing. The EGFP-VASP (Carl et al., 1999) and EGFP-paxillin constructs were kindly provided by Dr. U.D. Carl and Marcus Geese, respectively (Gesellschaft für Biotechnologische Forschung, Braunschweig, Germany).

Cells

All reagents were purchased from Invitrogen (Carlsbad, CA) unless stated otherwise. HeLa cells (ATCC CCL-2) and rat embryo fibroblasts (REFs) were maintained in Dulbecco's modified Eagle's medium with 10% fetal calf serum and 1 mM glutamine at 37°C in the presence of 5% CO₂. Mouse melanoma cells B16-F1 (ATCC CRL-6323) were grown as described above except with 10% fetal calf serum from PAA Laboratories (Linz, Austria). Goldfish fin fibroblasts (CAR, ATCC CCL-71) were maintained in basal Eagle's medium with Hanks' balanced salt solution, 1 mM glutamine, 1 mM nonessential amino acids, and 15% fetal bovine serum (Hyclone Laboratories, Logan, UT) at 25°C without CO₂.

Immunolabelings

Indirect immunofluorescence was performed essentially as described previously (Herzog et al., 1994) with minor modifications. Cells were routinely replated onto acid-washed glass coverslips coated with 50 µg/ml fibronectin (Roche Molecular Biochemicals, Mannheim, Germany).

HeLa cells (Figure 1, D and E) were fixed with a mixture of 3% formaldehyde and 0.3% Triton X-100 in cytoskeleton buffer [10 mM 2-(N-morpholino)ethanesulfonic acid, 150 mM NaCl, 5 mM EGTA, 5 mM glucose, 5 mM MgCl₂, pH 6.1] for 15 min and stained with monoclonal antibodies 164D4 or 184A3 followed by secondary Alexa 546-conjugated goat anti-mouse IgG antibodies (Molecular Probes, Leiden, The Netherlands).

View larger version (60K): [in this window] [in a new window]   Figure 1.   Properties of the generated mAb against zyxin. (A) Schematic representation of the amino acid sequence of zyxin. The four EVH1 binding sites (PP 1-4), nuclear export signal (NES), three LIM domains, and position of the epitopes of both zyxin-specific mAb are indicated. (B and C) Western blot analysis of the zyxin-specific monoclonal antibodies. HeLa cell extracts were separated by SDS-polyacrylamide gel electrophoresis and blotted onto polyvinylidene difluoride membranes. Parallel blots were stained with the antibodies 164D4 (B) and 184A3 (C). Molecular weight markers are indicated on the left. (D and E) Indirect immunofluorescence images of HeLa cells with mAb 164D4 (D) and mAb 184A3 (E) as primary antibodies. Bars, 10 µm.

For the double-labeled fluorescence images in Figure 2, B16-F1 cells stably expressing EGFP-VASP were fixed with a mixture of formaldehyde (4%) and 0.1% Triton X-100 in phosphate-buffered saline, pH 7.0) for 1 min followed by formaldehyde (4%) in phosphate-buffered saline for 20 min. Monoclonal anti-zyxin 164D4, anti-paxillin 349 (Transduction Laboratories, Lexington, KY), anti-vinculin hVIN-1 (Sigma-Aldrich, Taufkirchen, Germany), or monoclonal anti-Mena antibodies (clone 49C2B12; Wehland, unpublished data) were mixed with polyclonal antibodies against GFP (CLONTECH) to enhance the GFP-VASP signal. The secondary reagent was a mixture of Cy3-conjugated goat anti-mouse IgG and fluorescein isothiocyanate-conjugated goat anti-rabbit antibodies (both Jackson Immunoresearch, West Grove, PA). For the -actinin labelings in Figure 7, cells were fixed on the microscope stage with methanol for 5 min (Mies et al., 1998) and incubated with anti--actinin IgM BM.75.2 (Sigma-Aldrich) followed by Cy3-conjugated goat anti-mouse IgM (Jackson Immunoresearch). Pictures of cells fixed and stained on the microscope were taken in cytoskeleton buffer containing 100 mM dithioerythritol (DTE) to avoid photobleaching.

View larger version (146K): [in this window] [in a new window]   Figure 2.   Antibodies against zyxin, vinculin, or paxillin do not counterstain lamellipodial tips containing GFP-VASP. B16-F1 cells transfected with EGFP-VASP (A-D) were plated on fibronectin, fixed, and stained with antibodies to zyxin (164D4) (A'), vinculin (B'), paxillin (C'), and Mena (D'). Note that zyxin, vinculin, and paxillin colocalize with VASP at focal adhesions but not at the tips of lamellipodia (A and A', arrows), whereas Mena colocalizes with VASP also at these sites. Bars, 5 µm. The bar in C' is valid for C and C', and the bar in A' is valid for all panels except C and C'.

Transfections and Microinjection

B16-F1 cells were transiently transfected with the EGFP-zyxin construct with the use of LipofectAMINE (Invitrogen) as described previously (Ballestrem et al., 1998). At 12-24 h after transfection, B16-F1 cells were replated in Ham's F-12 medium (Invitrogen) containing 10% fetal calf serum onto acid-washed coverslips coated with 25 µg/ml laminin (Sigma-Aldrich) or 50 µg/ml fibronectin.

CAR fibroblasts were transfected with EGFP-VASP or EGFP-paxillin constructs with the use of the Superfect Transfection reagent according to the manufacturer's instructions (QIAGEN, Hilden, Germany). CAR cells were used for microscopy up to 3 d after transient transfections. Stable CAR cell lines (Kaverina et al., 1999) expressing the EGFP-zyxin construct were kindly provided by Dr. I. Kaverina (Austrian Academy of Sciences, Salzburg, Austria). B16-F1, REF, or CAR cells were replated onto 50 µg/ml fibronectin before microinjections.

Injections were performed with sterile Femtotips I (Eppendorf, Hamburg, Germany) with the use of Leitz (Leitz, Austria) or Narishige model M0188NE micromanipulators (Narishige, Tokyo, Japan) with a pressure supply from an Eppendorf microinjector 5242 (Eppendorf). Cells were injected with the back-pressure mode (set to 20-80 hPa) to give a continuous outflow from the needle.

Proteins for Microinjection and Drugs

The fluorescent derivative of turkey gizzard vinculin (5-TAMRA-vinculin) was prepared as described (Rottner et al., 1999a). Recombinant L61Rac was expressed as a glutathione S-transferase fusion protein in Escherichia coli and purified as described (Ridley and Hall, 1992). Hybridoma supernatant containing anti-zyxin antibodies 164D4 was purified with the use of protein G-Sepharose (Sigma-Aldrich) according to manufacturer's instructions. Purified antibodies were coupled with the use of the Alexa Fluor 488 protein labeling kit (Molecular Probes). After separation of antibodies and excess dye with the use of PD10 columns (Amersham Pharmacia Biotech AB), coupled antibodies were dialyzed into 2 mM Tris, 50 mM KCl, pH 7.0, before microinjection. The purity of proteins was confirmed by Coomassie-stained SDS-polyacrylamide gels and protein concentrations were determined with the use of the Bradford assay (Bio-Rad Laboratories, Munich, Germany). To visualize endogenous zyxin in live cells, Alexa 488-coupled antibodies were microinjected at 1 mg/ml.

To visualize vinculin and zyxin, paxillin, or VASP simultaneously in living cells, 5-TAMRA-vinculin was microinjected at concentrations of 0.5-1 mg/ml into cells expressing EGFP-zyxin, EGFP-paxillin, or EGFP-VASP.

To study the differential dynamics of different adhesion proteins during focal adhesion dissociation in live cells, we developed an assay to transiently but globally mimic the sequence of events taking place during this process. This was achieved by microinjection of CAR cells with mixtures of L61Rac and Y27632 (kindly provided by Yoshitomi Pharmaceutical Industries), an inhibitor of the Rho-associated protein kinase p160ROCK (Uehata et al., 1997). Injection of cells with Y27632 (2.5 mM) alone caused the rapid displacement of EGFP-zyxin, indicative of a down-regulation of the Rho pathway (Ishizaki et al., 1997); however, the cells subsequently retracted their edges, hindering longer term analysis of focal adhesion sites. The addition of L61Rac to the injection mixture further enhanced the dissociation of focal adhesions, presumably due to the antagonistic activities of the Rac and Rho pathways (Hirose et al., 1998; Rottner et al., 1999a). At the same time, the inclusion of Rac facilitated cell spreading and thus allowed analysis of the fate and reformation of focal adhesions throughout the cell. This injection protocol provided a highly reproducible method for inducing rapid disassembly of focal adhesions in live cells. Y27632 was dissolved in microinjection buffer (50 mM Tris pH 7.5, 150 mM NaCl, 5 mM MgCl₂, 1 mM DTE) at a concentration of 5 mM and mixed 1:1 with L61Rac (2 mg/ml).

Video Microscopy

B16-F1 and REF cells were observed and microinjected in an open heating chamber (Warner Instruments, Hamden, CT) maintained at 37°C on an inverted microscope (Axiovert 135TV; Zeiss, Jena, Germany) equipped for epifluorescence and phase contrast microscopy. CAR cells were observed and injected at room temperature. For microinjections, 40× objectives (numerical aperture [NA] 0.6 LD Achroplan or NA 1.3 Plan Neofluar; Zeiss) and for video microscopy a 100× objective (NA 1.4 Plan-Apochromat; Zeiss) were used in combination with or without 1.6 optovar intermediate magnification. The microscope was additionally equipped with shutters (Optilas, Puchheim, Germany) in the transmitted and epifluorescence light paths controlled by a homemade interface. A computer-driven filter-wheel (Technical Video, Woods Hole, MA) facilitated separate recordings of video sequences in phase contrast and/or different fluorescence channels. Excitation filters for red and green fluorescence in the filter wheel were used in combination with dichroic beamsplitters and emission filters (Chroma, Brattleboro, VT). Data were acquired with a back-illuminated, cooled charge-coupled device camera (Princeton Research Instruments, Trenton, NJ) driven by IPLab software (Scanalytics, Fairfax, VA), and processed with the use of IPLab, Scion Image 1.62 (Scion, Frederick, MD) and Adobe Photoshop 5.0.2 or 5.5 (Adobe Systems, Mountain View, CA) software.

Quantitation of Focal Adhesion Intensities

Focal adhesions were marked with the use of the segmentation tool in the IPLab software, taking advantage of the fact that adhesions could be identified as structures with intensities above a given threshold level. At least 300 focal adhesions were analyzed for each of the six data groups: zyxin, vinculin, and VASP, each before and 3 min after microinjections with Y27632/Rac. The intensities of focal adhesions were subtracted from the average background intensity in the cytoplasm of each cell (measured at as many points as the number of adhesions). The intensity measurements before treatment were normalized to 100 to pool the data derived from four (3 in the case of VASP) independent cells for each experimental group. The after-treatment data were transformed to percentage of the mean intensities before treatment. Statistical analysis was performed with the use of Microsoft Excel 98 and SigmaStat 2.0, and the graph in Figure 6 was created with Sigma Plot 4.0. The after-treatment data sets were each compared with the before-treatment sets and also with each other with the use of a Mann-Whitney Rank Sum test, from which statistically significant differences could be confirmed (p < 0001).

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Characterization of Monoclonal Antibodies against Human Zyxin

Two monoclonal antibodies (mAbs) against human zyxin were generated (164D4 and 184A3), and their epitopes were mapped on an array of overlapping synthetic peptides (Frank, 1992; Niebuhr et al., 1998) covering the entire human zyxin sequence (Macalma et al., 1996). This assay enabled the identification of the linear epitopes (Figure 1A; amino acids numbered according to Macalma et al., 1996) recognized by mAb 164D4 (352-TLKEVE-357) and by mAb 184A3 (87-PLAGD-81). The specificity of the monoclonal zyxin antibodies for these epitopes was verified by competitive inhibition with the respective soluble, synthetic peptides). The epitope of the antibody 164D4 maps to the nuclear export signal of zyxin (Nix and Beckerle, 1997), whereas the epitope of 184A3 is localized within the first of the EVH1 domain binding motifs of human zyxin, as defined in Niebuhr et al. (1997) (Figure 1A). On Western blots derived from total cell extracts from HeLa cells, both antibodies labeled a single band at ~84 kDa (Figure 1, B and C). In the same cell type, these antibodies predominantly stained focal adhesions (Figure 1, D and E), and in fibroblasts also stress fibers in a periodic manner (Figure 3A). This is in agreement with previous immunofluorescence studies with the use of rabbit antisera increased against a peptide derived from human zyxin (Macalma et al., 1996) and against porcine p83/zyxin (Reinhard et al., 1995). Because 164D4 recognizes a highly conserved zyxin epitope, it labels zyxin in cell lines from several species, including mouse, rat, cow, and pig, whereas mAb 184A3 is more species restricted, both in immunofluorescence microscopy and on Western blots.

View video and larger version (140K): [in this window]   Figure 3.   Microinjected mAb 164D4 and ectopically expressed zyxin do not target to lamellipodial tips. (A-D) Representative example of a REF cell microinjected with fluorescently coupled mAb 164D4. (B-D) Video sequence showing an enlargement of the area boxed in the overview in A demonstrates that the fluorescently coupled, microinjected mAb 164D4 does not target to the tip of the protruding lamellipodium (B' and C', arrowheads), although it incorporates into newly formed focal adhesions (C and D, arrows; also see video supplement). (E and F) B16-F1 mouse melanoma cells were transiently transfected with EGFP-zyxin or EGFP-VASP to directly compare the dynamics of the respective constructs in these cells moving on laminin. Ectopically expressed EGFP-zyxin (E) is not concentrated at lamellipodial tips, which is sharply contrasted by the dynamics of EGFP-VASP (F) (also see video supplement). Bars, 10 µm (A and F, also valid for E) and 5 µm (D'). Video Figure 3 B-D: Dynamics of anti-zyxin mAb 164D4 microinjected into REF cell. Whereas the Alexa 488-coupled mAbs incorporate into stress fibers and focal adhesions, there is no recruitment of antibodies to the tips of the protruding lamellipodium (compare phase contrast). Video Figure 3 E-F: B16-F1 cells expressing EGFP-zyxin compared with EGFP-VASP moving on laminin. Both constructs are targeted to focal adhesions. In addition, the motility of these cells induced on this substrate is accompanied by extensive recruitment of EGFP-VASP to the tips of protruding and ruffling lamellipodia. In contrast, EGFP-zyxin is not concentrated at these sites, although there is a slight increase in fluorescence intensity in the entire lamellipodium.

Zyxin Is not Recruited to Tips of Protruding Lamellipodia

Because zyxin binds to VASP and Mena (Reinhard et al., 1995; Gertler et al., 1996; Drees et al., 2000) and VASP localizes to the tips of protruding lamellipodia (Rottner et al., 1999b), we next tried to determine whether zyxin and Ena/VASP proteins colocalized at these sites, as proposed previously (Beckerle, 1998).

B16-F1 cells expressing EGFP-VASP were plated on fibronectin and counterstained with our monoclonal anti-zyxin antibody 164D4 (Figure 2, A and A'). Whereas EGFP-VASP localized to the tip of the protruding lamellipodium and to focal adhesions, zyxin only localized to the latter structures (Figure 2, A and A'). A comparison of the localizations of vinculin, paxillin, and Mena in EGFP-VASP-expressing B16-F1 cells is shown in Figure 2, B'-D'. The Mena label matched the distribution of EGFP-VASP entirely (Figure 2, D and D'), indicating that the localization of endogenous proteins at the lamellipodium tip is not affected by fixation. On the other hand, vinculin and paxillin are only found in focal adhesions (Figure 2, B and B', and C and C', respectively). In summary, neither zyxin, vinculin, nor paxillin colocalized with VASP and Mena at the tips of lamellipodia.

To corroborate the lack of zyxin from lamellipodial tips, we followed the dynamics of this protein in living cells. This was performed in one set of experiments by ectopically expressing EGFP-zyxin, and in another set by microinjecting fluorescently labeled mAb 164D4.

Figure 3A shows a REF cell microinjected with mAb 164D4 coupled to Alexa 488, demonstrating the intense labeling of zyxin in focal adhesions and along stress fibers. Microinjection of this antibody did not affect zyxin dynamics, because adhesion patterns and turnover were not altered for at least 12 h after injection. Figure 3, B-D, shows the dynamics of the fluorescently labeled antibody during extension of the cell periphery (also see video supplement). Although the antibody labels newly formed adhesion sites during this video sequence, there is no concentration of the antibody at the tip of the protruding lamellipodium, as marked by arrowheads in the phase contrast images (Figure 3, C' and D').

On laminin, B16-F1 mouse melanoma cells are highly motile and express broad lamellipodia (Ballestrem et al., 1998), providing a useful system to analyze the dynamics of cytoskeletal proteins during cell motility and the protrusion of lamellipodia (Rottner et al., 1999b). Here, we have compared the dynamics of EGFP-zyxin with EGFP-VASP (also see video supplement). Figure 3, E and F, show video frames of B16 cells expressing GFP-tagged zyxin and VASP, respectively. As shown previously, GFP-VASP appears as a sharp line at the edge of protruding lamellipodia (Figure 3F), whereas EGFP-zyxin is incorporated into focal adhesions and stress fibers but not into lamellipodial tips (Figure 3E). The general, diffuse labeling of the lamellipodium in Figure 3E was only marginally more intense than observed with GFP alone.

Comparison of Zyxin and Vinculin Dynamics

To investigate the dynamics of zyxin in more detail, we followed focal adhesion formation in EGFP-zyxin-expressing B16-F1 cells that were previously injected with fluorescently labeled turkey gizzard vinculin (5-TAMRA-vinculin). Comparison of the video sequences showed that both zyxin and vinculin simultaneously incorporated into newly formed adhesion sites at the cell front (Figure 4, A and B, arrows).

View video and larger version (127K): [in this window]   Figure 4.   Comparing zyxin with vinculin reveals differential dynamics during adhesion disassembly. B16-F1 cell transiently transfected with EGFP-zyxin and microinjected with rhodamine-labeled vinculin. The video sequence shows the dynamics of zyxin and vinculin in A and B, respectively. Time points of the selected frames are given in minutes. Note that zyxin and vinculin simultaneously incorporate into newly formed adhesion sites (A and B, arrows). In contrast, zyxin clearly dissociates from dissolving focal adhesions earlier than vinculin (A and B, arrowheads). The arrowheads in the top half of the images point toward two adhesion sites, which disappear consecutively in the zyxin panel. By 39', both sites have disappeared completely, whereas the vinculin picture still shows a prominent adhesion site (compare top arrowheads in A and B, respectively (also see video supplement). The second pair of arrowheads points toward another example showing that as soon as the adhesion site starts sliding (by 11'), zyxin is entirely diminished (also see video supplement). Bar, 10 µm. Video Figure 4: Dynamics of EGFP-zyxin and rhodamine-labeled vinculin in B16-F1 cells on fibronectin. Vinculin and zyxin incorporate simultaneously into newly formed adhesions at the cell front (arrows). Conversely, zyxin dissociates from focal adhesions much earlier than vinculin (follow arrowheads). Focal adhesion disassembly is characterized in these cells by the detachment and sliding of vinculin-containing sites (arrowheads), which are devoid of zyxin.

However, closer inspection of overall focal adhesion dynamics revealed that zyxin dissociated from dissolving adhesions earlier than vinculin, as seen in the video sequence in Figure 4, A and B, arrowheads. To analyze this process in more detail, we used the Rho kinase inhibitor Y27632 (Uehata et al., 1997) to promote focal adhesion disassembly (Rottner et al., 1999a). We established that the injection of goldfish fibroblasts (CAR) with a mixture of Y27632 and constitutively active L61Rac caused the rapid but reversible delocalization of EGFP-zyxin from focal adhesions (compare Figure 5, A and A'; see MATERIALS AND METHODS and video supplement). With the use of this approach we were able to compare the dynamics of different focal adhesion proteins during focal adhesion disassembly.

View video and larger version (143K): [in this window]   Figure 5.   Comparison of various adhesion components during experimentally induced focal adhesion disruption confirms early dissociation of zyxin. Goldfish fibroblasts (CAR) expressing EGFP-zyxin (A) were microinjected with a mixture of constitutively active Rac (L61Rac) and the Rho kinase inhibitor Y27632. A and A' show the same cell directly before (A) or after microinjection (A'). Before injection, EGFP-zyxin localizes to focal adhesions and stress fibers (A) but is already dislocalized 3 min after microinjection of the L61Rac/Y27632 mixture (A'). (C, E, and G) Examples of CAR fibroblast expressing EGFP-zyxin (C), EGFP-paxillin (E), or EGFP-VASP (G) that have additionally been microinjected first with rhodamine vinculin (B, D, and F) and subsequently with the L61Rac/Y27632 mixture. B, C, and D, E as well as F, G represent image pairs of the same cells 3 min after injection with the secondary mixture, a time point after Rac/Y27632 injection at which zyxin was entirely dislocalized in all experiments performed. In contrast to EGFP-zyxin (C), 5-TAMRA-vinculin (B) still localizes to focal adhesions 3 min after injection, whereas EGFP-paxillin (E) and vinculin (D) behave virtually identically. The intensity of EGFP-VASP (G) in focal adhesions is significantly reduced but not entirely diminished like EGFP-zyxin (C); in contrast, EGFP-VASP is strongly recruited to the tips of the lamellipodia induced by this treatment (G). The cells from all panels except the pair in D, E are provided as supplemental videos. Bar, 10 µm. Video Figure 5 A-A': In an experimental setup to initiate the sequence of events leading to focal adhesion disassembly, EGFP-zyxin-expressing CAR fibroblasts are injected with a mixture of constitutively active Rac and the Rho kinase inhibitor Y27632. Under these conditions, zyxin is rapidly released from focal adhesions. The time point of injection is indicated and can be followed in phase contrast. Video Figure 5 B-C: EGFP-zyxin-expressing CAR cells already injected with fluorescent vinculin are injected again with the mixture of L61Rac/Y27632. Whereas zyxin dislocates from focal adhesions very rapidly, the distribution of vinculin remains almost unaffected. This experiment confirms the differential dynamics of vinculin and zyxin during adhesion disassembly observed in untreated cells (as shown in the supplemental video for Figure 4). Video Figure 5 F-G: EGFP-VASP-expressing CAR fibroblasts microinjected with rhodamine-labeled vinculin are injected again with the Rac/Y27632 mixture. In contrast with zyxin, this treatment causes a partial loss of VASP from focal adhesions, whereas the distribution of vinculin is again almost unchanged. Note that EGFP-VASP recruitment to the tips of lamellipodia is strongly enhanced by this treatment.

CAR fibroblasts expressing EGFP-zyxin were first injected with 5-TAMRA-vinculin to record the dynamics of zyxin and vinculin during secondary injections with the Rac/Y27632 mixture. Figure 5, B and C, show the distribution of vinculin and zyxin in the same cell 3 min after the secondary injection, respectively. Before secondary injections, both vinculin and zyxin were strongly incorporated into focal adhesions; (see supplementary video). However, after injections with Rac/Y27632, EGFP-zyxin was rapidly dislocated from focal adhesions (Figure 5C), whereas most of the fluorescent vinculin analog remained in these sites (Figure 5B). To exclude the possibility that the observed effects were due to different labeling methodologies for the two proteins, we performed analogous experiments with cells expressing EGFP-paxillin instead of EGFP-zyxin. In control cells, the EGFP-paxillin construct localized to focal adhesions, as expected from antibody labelings and from previous work (Ludin and Matus, 1998). As shown in Figure 5, 3 min after Rac/Y27632 injection, the distributions of 5-TAMRA-vinculin and EGFP-paxillin were virtually identical (Figure 5, D and E, respectively). EGFP-zyxin-expressing cells that were not injected with 5-TAMRA-vinculin were also injected with Rac/Y27632 and then fixed and stained for paxillin. In line with the findings for vinculin, the focal adhesions had retained paxillin but not zyxin.

Quantitation of the intensities of zyxin and vinculin in focal adhesions of CAR cells immediately before and 3 min after injection of Rac/Y27632 revealed that the treatment reduced the average intensity of zyxin to almost background levels (5%), whereas the average intensity of vinculin fell only 22% below the level before injection (Figure 6). Taken together, these data demonstrate for the first time significant differences between zyxin and vinculin or paxillin with respect to their dynamics during focal adhesion disassembly.

View larger version (13K): [in this window] [in a new window]   Figure 6.   Quantitation of the dislocation of different focal adhesion components in response to injections of Y27632/Rac. Average intensities of focal adhesions were analyzed immediately before () and 3 min after microinjection of Y27632/L61Rac () for the different focal adhesions components indicated. Each bar corresponds to the arithmetic mean with SEM.

VASP Is Recruited to Focal Adhesions not Only by Zyxin

Zyxin and vinculin harbor binding motifs for the EVH1 domain of Ena/VASP proteins, suggesting that both are required for the recruitment of Ena/VASP proteins to focal adhesions (Gertler et al., 1996). To test this hypothesis, EGFP-VASP-expressing CAR fibroblasts were injected with 5-TAMRA-vinculin followed by secondary injections with the Rac/Y27632 mixture. This treatment caused a partial dislocation of EGFP-VASP from focal adhesions (Figure 5G; also see supplementary video), whereas the localization of vinculin (Figure 5F; supplementary video) was again almost unaffected. In response to the injection of Rac/Y27632, the behavior of VASP differed from that of both zyxin and vinculin, in that ~47% was dislocated from focal adhesions (Figure 6). These data demonstrate that the recruitment of VASP to focal adhesions is not exclusively mediated by zyxin.

Interestingly, this treatment further illustrated the differential dynamics of zyxin and VASP in lamellipodia. For EGFP-VASP, the partial loss from focal adhesions resulted in a dramatic relocalization to the tips of the induced lamellipodia (Figure 5G), whereas for EGFP-zyxin, no such relocalization was observed, despite active lamellipodial protrusion (Figure 5C).

Displacement of Zyxin from Focal Adhesions Is Accompanied by Depletion of -Actinin

Because subcellular targeting of zyxin was suggested to be mediated by its interaction with -actinin (Reinhard et al., 1999), we tested whether experimentally induced dislocation of zyxin by Rac/Y27632 injections was accompanied by dislocation of -actinin. EGFP-zyxin-expressing CAR fibroblasts were injected with the Rac/Y27632 mixture as before, fixed, and stained with antibodies against -actinin. Figure 7A shows EGFP-zyxin incorporated into focal adhesions, which is again lost upon injection with Rac/Y27632 (Figure 7A'). At the time of almost complete dislocation of zyxin, the levels of -actinin, both in focal adhesions and stress fibers, were markedly reduced, compared with noninjected control cells (compare Figure 7, A" and B). We conclude that the initiation of focal adhesion disassembly coincides with the early displacement of both zyxin and -actinin.

View video and larger version (86K): [in this window]   Figure 7.   Comparison of zyxin and -actinin distribution after experimentally induced zyxin displacement. Goldfish fibroblasts (CAR) expressing EGFP-zyxin were microinjected with a mixture of constitutively active Rac (L61Rac) and the Rho kinase inhibitor Y27632. (A and A') EGFP-zyxin distribution directly before and after microinjection of the mixture, respectively. Cells were fixed right upon virtually complete displacement of EGFP-zyxin. A' corresponds to the last videoframe before fixation and staining of a representative cell with anti--actinin antibodies (A"). Note that at the time of zyxin-displacement, -actinin was grossly displaced from stress fibers and focal adhesions compared with noninjected neighboring cells (B). Bar, 10 µm.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Focal adhesions and lamellipodial tips of fibroblasts are the two sites at which microinjected, fluorescently labeled actin is first incorporated, marking them as centers of actin polymerization (reviewed in Small et al., 1998). Here, the functional similarity stops. Focal adhesions anchor actin filament bundles via transmembrane linkages to the extracellular matrix, whereas lamellipodia are protrusive structures whose tips are highly mobile. We can thus expect that the molecular makeup of the two sites of actin filament generation are correspondingly different. Indeed, the only cytoskeletal components so far found in both sites are Ena/VASP proteins (Rottner et al., 1999b) and profilin (Geese et al., 2000), supporting a general role of these proteins in the dynamic processes of actin reorganization.

Recently, it has been proposed that zyxin may serve as a molecular scaffold, recruiting proteins capable of promoting site-specific actin assembly in lamellipodia (Beckerle, 1998; Drees et al., 1999, 2000). This conclusion is based on immunofluorescence localization studies (Reinhard et al., 1995, Drees et al., 1999) and on data derived from experimentally induced redistribution of zyxin within cells (Golsteyn et al., 1997; Drees et al., 1999, 2000). The retraction of the cell edge after zyxin dislocation was concluded to be caused by the disassembly of molecular complexes regulating actin assembly close to the membrane (Drees et al., 1999). However, this effect may equally well be explained by the disruption of peripheral focal adhesions; thus, myosin II-based contractility is required for the maintenance of focal adhesions (Chrzanowska-Wodnicka and Burridge, 1996) and focal complexes (Rottner et al., 1999a), and the local application of contraction inhibitors was shown to be sufficient to effect the retraction of the cell edge (Kaverina et al., 2000). Interestingly, artificial targeting of zyxin to the membrane by a CAAX motif causes the loss of stress fibers and the induction of F-actin-rich cell "surface projections" (Golsteyn et al., 1997), which are capable of recruiting Ena/VASP proteins (Drees et al., 2000). But this observation does not prove that zyxin is responsible for the recruitment of Ena/VASP proteins to normal actin-mediated cellular protrusions. Rather, the disassembly of stress fibers upon experimentally induced mislocalization of zyxin indicates that zyxin is involved in focal adhesion and stress fiber maintenance.

Our findings clearly demonstrate that in contrast to VASP, zyxin is not present at the protruding tips of lamellipodia, making zyxin an unlikely player in the process of lamellipodial protrusion. We have also noted that the treatment of EGFP-VASP-expressing cells with the Rho kinase inhibitor causes a partial loss of VASP from focal adhesions and a notable incorporation into the tips of the newly protruding lamellipodia. In contrast, the same treatment did not induce a dynamic relocalization of EGFP-zyxin to lamellipodial tips. However, we occasionally observed a weak localization of zyxin throughout the entire width of lamellipodia, for instance, in B16 cells moving on laminin or in fibroblasts spreading on fibronectin (Rottner, unpublished data). This is consistent with the presence of its interaction partner -actinin (Crawford et al., 1992) in lamellipodia (Schulze et al., 1989). Interestingly, zyxin can also be detected at low levels along the actin tail of motile Listeria monocytogenes as well as around intracellullar nonmotile bacteria (Frischknecht et al., 1999; Krause and Wehland, unpublished results), where -actinin is also found (Dabiri et al., 1990; Temm-Grove et al., 1994; Sechi et al., 1997). In contrast, Ena/VASP proteins (Chakraborty et al., 1995) and profilin (Geese et al., 2000) are recruited to the surfaces of Listeria where actin monomer insertion occurs, in a situation analogous to lamellipodial tips. These observations further support the view that lamellipodia and the tails of intracellular Listeria share similarities with respect to their molecular composition and function (Machesky, 1997). The recruitment of Ena/VASP proteins to the bacterial surface is mediated by FPPPP motifs present in ActA, mimicking a mechanism of positioning of Ena/VASP proteins within the cell by cellular analogs such as vinculin, zyxin, or Fyb/SLAP. The latter protein colocalizes with, and links Ena/VASP proteins to, WASP and the Arp2/3 complex at the interface of T cells and antigen-presenting cells (Krause et al., 2000). However, Fyb/SLAP is restricted to the hematopoetic system and both vinculin and zyxin do not localize at the tips of lamellipodia. Moreover, the cytoplasmic expression of ActA fragments harboring the FPPPP motifs does not interfere with recruitment of Ena/VASP proteins to these sites (Bear et al., 2000). Therefore, at least in nonhematopoetic cells, the targeting of Ena/VASP proteins to the tips of dynamic protrusions such as lamellipodia and filopodia is mediated by an FPPPP-independent mechanism, the nature of which is still controversial (Bear et al., 2000; Nakagawa et al., 2001).

The detailed analysis of zyxin dynamics compared with another prominent focal adhesion protein, vinculin, revealed an intriguing difference. Meigs and Wang (1986) were the first to compare the dynamics of two focal adhesion components in the same cell in response to stimulation by the phorbolester (TPA). They showed that -actinin was removed before vinculin from focal adhesions and that vinculin persisted until substrate dissociation of these sites, as judged by interference reflection microscopy. Here, we confirm and extend these findings with the use of an assay designed to specifically dissociate focal adhesions. By inhibiting the Rho pathway with the Rho kinase inhibitor Y27632 (Uehata et al., 1997), we show that vinculin and VASP display different dissociation dynamics from zyxin. As an actin filament cross-linker, -actinin contributes to the maintenance of stress fibers and focal adhesions and presumably performs this role synergistically with myosin and other components. Such a role of -actinin is further supported by the observation that its overexpression results in the formation of more stable attachment sites, whereas a general reduction of -actinin synthesis is associated with an increase in cell motility (Glück and Ben-Ze'ev, 1994). One possible partner mediating this postulated role of -actinin may be zyxin. In agreement with this, we could demonstrate that zyxin dissociates much earlier from dissolving focal adhesions than vinculin and that experimentally induced zyxin dislocation coincides with the displacement of -actinin, confirming the differential dislocation of -actinin and vinculin from focal adhesions (Meigs and Wang, 1986). Therefore, the maintenance and integrity of focal adhesions may be influenced by both -actinin and zyxin. Interestingly, VASP was more tenaciously bound to adhesion sites than zyxin, indicating that VASP recruitment to adhesion sites is not solely dependent on zyxin. Vinculin is one likely candidate responsible for this residual VASP recruitment, because it harbors an FPPPP motif (Brindle et al., 1996; Gertler et al., 1996; Reinhard et al., 1996), but there may be additional focal adhesion proteins involved in this process, such as palladin (Parast and Otey, 2000).

In conclusion, we demonstrate that zyxin does not target Ena/VASP proteins to the tips of lamellipodia. Furthermore, the experimental strategy of simultaneously visualizing the dynamics of two different cytoskeletal proteins within the same cell enabled us to dissect early molecular events upon disintegration of focal adhesions. Our results suggest a role for zyxin in the regulation of focal adhesions and reopen the search for molecules that target Ena/VASP proteins to the tips of lamellipodia and filopodia.