FAP52 Regulates Actin Organization via Binding to Filamin*

FAP52, a focal adhesion-associated phosphoprotein, is a member of a FAP52/PACSIN/syndapin family of proteins. They share a multidomain structure and are implicated in actin-based and endocytotic functions. We show, by using both native and recombinant proteins, that FAP52 selectively binds to the actin cross-linking protein filamin (ABP-280). This was based on an affinity purification followed by a sequence determination by mass spectrometry, co-immunoprecipitation, overlay binding, and surface plasmon resonance analysis. Binding studies with deletion mutants showed that the sites of the interaction map to the highly α-helical N-terminal part of FAP52 and to the C-terminal region of filamin, which also contains binding sites to some transmembrane signaling proteins. In immunofluorescence and immunoelectron microscopy of cultured fibroblasts, a different overall subcellular distribution was seen for filamin and FAP52 except for a stress fiber-focal adhesion junction where they showed a notable overlap. Overexpression of the full-length and mutant forms of FAP52 led to an extensive reorganization of actin and filamin in cultured fibroblasts. Thus, the results show that FAP52 interacts with filamin, and we propose that this interaction is important in linking and coordinating the events between focal adhesions and the actin cytoskeleton.

FAP52 is a recently described focal adhesion-associated protein which in immunofluorescence microscopy (IFM) 1 colocalizes with vinculin, paxillin, and talin (1). The latter are well known focal adhesion proteins that are also closely associated with the actin cytoskeleton (2). More recently, two homologues of FAP52, PASCIN2 and syndapin II, were identified (3,4). They all share a modular structure typified by a well conserved FER-CIP4 homology domain in the very N terminus, followed by a highly ␣-helical region, and a C-terminal Src homology 3 (SH3) domain. In the middle, there is a "linker" region that shows a high degree of sequence variation and does not conform to any specific domain signature. Thus far, no specific function is assigned to FAP52. PACSIN 2, its closest homologue, participates in the organization of the actin cytoskeleton and in the regulation of vesicular traffic (5). Syndapin II, on the other hand, along with other syndapin isoforms, is involved in endocytosis and actin dynamics (4). Both syndapin II and PACSIN 2 bind N-WASP, an important regulator of actin cytoskeleton organization (4,5). Thus, all the three members of the family share a property of being closely associated with the actin cytoskeleton-associated proteins or functions.
In this study, we set out to identify binding partners of FAP52. A single polypeptide band of 280 kDa was purified from cultured cells by applying co-purification and co-immunoprecipitation techniques and, was shown to be filamin (actinbinding protein, 280 kDa; ABP-280) by using electrospray mass spectrometric analysis of tryptic peptides. Interaction between FAP52 and filamin was further verified by using direct binding assays. In IFM and immunoelectron microscopy (IEM) of cultured cells, distinctly demarcated co-distribution of FAP52 and filamin was seen at sites where the actin stress fibers abut the focal adhesions. In cells overexpressing FAP52, a distinct reorganization of actin and filamin was seen. On the basis of these results we suggest that filamin serves as a link between FAP52 and the actin cytoskeleton and that this interaction is important in linking focal adhesions to actin cytoskeleton.

EXPERIMENTAL PROCEDURES
General Procedures-The standard solutions, buffers, and procedures for the purification and precipitation of DNA, restriction enzyme digestions, ligation reactions, SDS-PAGE runs, and stainings were as described in Sambrook et al. (6). DNA sequencing was carried out on an automated ABI Prism 377XL DNA Sequencer (PerkinElmer Life Sciences).
Materials-Escherichia coli strain BL21(DE3) cells, expression vectors pGEX-2T and pGEX-2TK, glutathione (GSH)-Sepharose 4B, thrombin, Rainbow High Molecular Weight Markers, and synthetic oligonucleotides were purchased from Amersham Biosciences; isopropyl-␤-D-thiogalactopyranoside was from MBI Fermentas; Site-directed Mutagenesis kit, Pfu polymerase, and T4 DNA ligase were from Stratagene; restriction enzymes were from MBI Fermentas and Promega Co., and chicken gizzard filamin was from Progen Biotechnik GmbH. For blot overlay (OL) assay, chicken gizzard filamin was purified according to the method of Feramisco and Burridge (7). HeLa cells were obtained from the American Type Culture Collection; Immu-Mount mounting medium was from Shandon, Inc.; FuGENE 6 transfection reagent, reduced GSH, protease inhibitors aprotinin, leupeptin, and phenylmethylsulfonyl fluoride were from Roche Molecular Biochemicals; ampicillin, bovine serum albumin (BSA), and benzamidine were from Sigma; and Triton X-100 was from Merck. Cell culture media and fetal calf serum were from HyClone Laboratories, Inc.; and nitrocellulose (NC) membranes were from Schleicher & Schuell.
Antibodies-Rabbit polyclonal antibody to FAP52, denoted Affi-K7, was produced and affinity-purified as described previously (1). Mouse monoclonal anti-filamin antibody was obtained from Chemicon International, Inc.; mouse monoclonal anti-GST and rabbit polyclonal an-ti-HA epitope antibodies were from Santa Cruz Biotechnology; rhodamine-phalloidin, Alexa Fluor 488 goat anti-rabbit IgG, and Alexa Fluor  546 goat anti-mouse IgG were from Molecular Probes; mouse monoclonal anti-paxillin antibody and rabbit anti-mouse IgG were from  Zymed Laboratories Inc.; and horseradish peroxidase (HRP)-conjugated  goat anti-rabbit IgG, HRP-conjugated goat anti-mouse IgG, goat antirabbit IgG-agarose, and goat anti-mouse IgG-agarose were purchased  from Sigma. For IEM, protein A-gold particles (diameter, 5 and 10 nm), prepared by Slot and Geuze (8), were used.
DNA Constructs and Protein Expression-All the cDNA constructs and the corresponding polypeptides were designated as "FAP52" or "Fil" for FAP52 and filamin, respectively, followed by the subscripted number of the first and the last amino acid (aa) residue expressed.
Full-length FAP52 and its fragments were produced as fusion proteins with glutathione S-transferase (GST) in BL21(DE3) cells by employing the expression vectors pGEX-2T and pGEX-2TK. The 1.35-kb cDNA encoding the full-length FAP52 was inserted into the BamHI/ EcoRI-cloning site in the vector as described previously (1). For the cloning and expression of the 3Ј-and 5Ј-truncated mutants of FAP52, PCRs with the oligonucleotide primers with add-on sequences for BamHI and EcoRI restriction sites and the full-length cDNA of FAP52 as template were utilized. The resulting cDNAs were cut with the matching restriction enzymes and ligated to BamHI/EcoRI-cut pGEX-2T or pGEX-2TK vectors.
For the production of the C-terminal fragments of filamin, pGEX-4T-1 vectors with the cDNAs encoding for the fragments Fil-(1524 -2283), Fil-(2283-2490), and Fil-(2495-2647) were employed (a kind gift from Dr. T. P. Stossel, Division of Hematology, Brigham and Women's Hospital, Boston (9)). Further C-terminal truncation mutants were generated from the construct encoding for the fragment Fil-(1524 -2283) by using a Site-directed Mutagenesis kit and oligonucleotide primers to generate stop codons to appropriate sites of the cDNAs.
GST fusion proteins were expressed in E. coli BL21(DE3) cells according to standard protocols, purified from the cell lysates on GSH-Sepharose 4B beads, and liberated from the beads by incubation with reduced GSH. For some experiments, FAP52 or its mutant form was released from its fusion partner by incubation with thrombin. For the control experiments, GST was expressed from the plasmid pGEX-2TK and purified as above.
For the expression in cultured animal cells, cDNA constructs encoding the full-length FAP52 or its fragments were produced by PCR with engineered restriction sites. They were cut with EcoRI and EcoRV and subcloned into the EcoRI/EcoRV-cloning site of pRK5 vector (a kind gift from Dr. Joseph Schlessinger, New York University, Medical Center, New York, NY). A hemagglutinin (HA) epitope tag was engineered to the C terminus of the constructs by primer design. The following constructs were generated: FAP52-HA (FAP52-(1-448) plus HA tag); FAP52 Nt -HA (FAP52-(1-293) plus HA tag); and FAP52 SH3 -HA (FAP52-(390 -448) plus HA tag).
Cell Cultures, Transfections, and Immunofluorescence Microscopy-Cultures of chicken embryo heart fibroblast (CEHF) were established, and the cells were grown as described previously (1). HeLa cells were grown according to the same protocol with the exception that Dulbecco's modified Eagle's medium was used. For transfections, the cells were grown on glass coverslips to a confluence of 50 -70%. Transfections were carried out by using a FuGENE 6 transfection reagent and following the manufacturer's instructions.
For IFM, CEHF cells grown on glass coverslips were fixed and stained essentially as described (1). In cases with staining of actin filaments, post-fixation was carried out with ethanol for 1 min. As primary antibodies, Affi-K7 (rabbit), rabbit polyclonal anti-HA, mouse monoclonal anti-filamin, or mouse monoclonal anti-paxillin were used at appropriate dilutions. Fluorochrome-conjugated secondary antibodies or phalloidin (Alexa Fluor 488 goat anti-rabbit IgG, Alexa Fluor 546 goat anti-mouse IgG or rhodamine-phalloidin) were applied as secondary antibodies. Viewing was under a Zeiss LSM510 laser scanning microscope equipped with a Zeiss Axiovert 110M inverted microscope (Carl Zeiss Microscopy). The images were processed by using an LSM three-dimensional software, version 5.2 (Carl Zeiss Microscopy).
Immunoelectron Microscopy-For IEM, CEHF-cells were grown close to confluence. After washing with Hanks' salt solution and fixation with 4% paraformaldehyde in phosphate-buffered saline (PBS) containing 2.5% sucrose for 1 h, they were harvested with a cell scraper and pelletted. The pellet was resuspended in 2.3 M sucrose and frozen in liquid nitrogen. Cryosections (thickness 80 nm) were cut by using a Leica Ultracut UCT microtome and placed on Butvar-coated nickel grids. They were washed with PBS and blocked with 5% BSA with 0.1% ColdWaterFishSkin gelatin in PBS for 10 min. For double labeling, the sections were first incubated with the monoclonal mouse anti-filamin antibody in 0.1% BSA in PBS (BPBS) for 60 min, followed by an incubation with the rabbit anti-mouse IgG in BPBS for 30 min, and with protein A-gold (diameter, 10 nm) in BPBS for 30 min. Between each of these steps, the sections were washed with PBS/glycine (6 times for 5 min). After this, the sections were fixed in 1% glutaraldehyde in PBS and blocked with 1% BSA in PBS/glycine for 30 min. They were then incubated with Affi-K7 followed by protein A-gold (diameter, 5 nm), both in BPBS. Washings were with PBS/glycine as above. The sections were then post-fixed in 2% glutaraldehyde for 5 min, washed with PBS and distilled water, counterstained with neutral 2% uranyl acetate for 5 min, and coated with 1.8% methylcellulose with 0.3% uranyl acetate. The sections were analyzed using a Philips 410 LS transmission electron microscope.
Preparation of Cell Lysates-Whole-cell lysates were prepared from the cells grown to confluence. They were first rinsed with PBS and then scraped into a lysis buffer (50 mM Hepes, pH 7.5, 150 mM NaCl, 1 mM EGTA, 10% glycerol, 1.5 mM MgCl 2 , 0.2 mM Na 3 VO 4 , and 1% Triton X-100) supplemented with protease inhibitors (0.1 M aprotinin, 0.1 mM leupeptin, 1 mM phenylmethylsulfonyl fluoride, and 5 mM benzamidine), and incubated in an ice bath for 20 min. The lysates were cleared by centrifugation at 12,000 ϫ g for 5 min at 4°C.
Immunoprecipitation-For immunoprecipitation (IP), Affi-K7 or anti-filamin antibodies were added to 500 l of the cell lysate (0.15 mg/ml). For control reactions, an equal volume of PBS was added instead of the antibody. Then 50 l of anti-rabbit IgG-agarose or anti-mouse IgGagarose, respectively, in PBS was added and incubated on a nutator for 16 h at 4°C. The beads were collected by centrifuging at 500 ϫ g for 2 min and then washing three times with PBS. The volume was adjusted to 60 l with PBS. After addition of the Laemmli's sample buffer (15 l) and boiling for 2 min, the eluates were run on a 7.5 or 10% SDS-PAGE and then analyzed by immunoblotting (IB).
Affinity Chromatography and Pull-down Experiments-Bacterially produced GST-FAP52 or GST was immobilized on GSH-Sepharose 4B beads in PBS. The beads (25 l) were incubated with the CEHF lysate (0.15 mg of protein in 500 l of lysis buffer) overnight at ϩ4°C on a nutator. They were then washed three times with PBS, and the volume was adjusted to 60 l with PBS. After addition of 15 l of a sample buffer and boiling, the proteins were separated on a 7.5% SDS-PAGE gel which was then stained with Coomassie Brilliant Blue (CBB).
For the pull-down experiments, 5 g of the GST fusions of fragments of FAP52 or filamin were immobilized on 20 l of GSH-Sepharose 4B beads in Tris-buffered saline (TBS). Twenty g of the chicken gizzard filamin or the bacterially produced FAP52, respectively, in 500 l of TBS was added, and the mixture was then incubated for 2 h at ϩ4°C on a nutator. The beads were washed three times with TBS and, after addition of 20 l of a sample buffer, subjected to SDS-PAGE separation, followed by IB with mouse anti-filamin or Affi-K7, respectively.
Immunoblotting-For IB, the electrophoretically separated proteins were transferred to NC membranes that were then blocked in 3% nonfat dried milk powder in TBS. Thereafter, the filters were incubated with Affi-K7 or mouse anti-filamin for 1 h at room temperature (RT). After an overnight washing with TBS at RT, the filters were incubated either with the HRP-conjugated goat anti-rabbit IgG or the HRP-conjugated goat anti-mouse IgG, respectively. After a 5-h washing with TBS, the blot was developed by the ECL method as described previously (1).
Blot Overlay Assay-For blot OL assays, bacterially produced FAP52 and GST or GST fusions of filamin mutants were separated on SDS-PAGE and transferred to NC membranes. The filters were blocked by incubating with 3% nonfat dried milk powder in TBS overnight at 4°C or for 2 h at RT, and then incubated with the purified filamin or recombinant FAP52 in 1% nonfat dried milk powder in TBS for overnight at 4°C. For the visualization of the protein bands, the overlays were washed with TBS for 3 h and then incubated with either the mouse anti-filamin antibodies or Affi-K7, respectively, as above, followed by an HRP-conjugated goat anti-mouse IgG or anti-rabbit IgG, respectively. The blots were developed by using the ECL detection system.
Electrospray Ionization Mass Spectrometry-For mass spectrometry, pieces corresponding to the protein bands of interest were excised from the CBB-stained polyacrylamide gels and processed as described by Shevchenko et al. (10). Briefly, the proteins were in gel-reduced, alkylated, and digested with trypsin at 37°C. The supernatant obtained was acidified with formic acid, loaded onto a Poros R2 (Perspective Biosystems) microcolumn, and desalted according to Gobom et al. (11). The peptides eluted with 40% methanol, 5% formic acid were introduced into a nanoelectrospray needle (Protona). Nanoelectrospray tandem mass spectrometry was performed on an API III mass spectrome-ter (PerkinElmer Life Sciences) equipped with a nanoelectrospray source, as described elsewhere (12).
Surface Plasmon Resonance Analysis-Surface plasmon resonance (SPR) measurements were performed on a BIACORE 3000 analyzer under a control of a BIACORE control software, version 3.1.1. (Biacore AB). For experiments with the GST fusions of FAP52, of Fil-(1524 -2283), or mutants thereof as a ligand, anti-GST antibody (Biacore AB) was immobilized on a carboxymethyl-coated CM5 sensor chip by employing the GST capture and the amine coupling kits (Biacore AB). Full-length GST-FAP52, or its mutants (0.15 mg/ml in 10 mM sodium acetate, pH 5.5), or GST-Fil-(1524 -2283), or its mutants (0.15 mg/ml in HBS-EP; 0.01 M Hepes, pH 7.4, 0.15 M NaCl, 3 mM EDTA, 0.005% (v/v) polysorbate 20; Biacore AB), were then injected onto the chip for 30 s with a flow rate of 30 l/min. The chip was then equilibrated with HBS-EP.
For binding assays, chicken gizzard filamin or recombinant FAP52, both in HBS-EP, were then passed over the chip at concentrations ranging from 0 to 800 nM at a flow rate of 30 l for 1 min for filamin or 3 min for FAP52. The binding of the analyte to the ligand was detected from the change in the sensorgram. To correct the sensorgram for a background binding and bulk refractive index changes, a different flow cell without GST fusion protein was used as a reference. As controls, GST-coated chips were used over which filamin or FAP52 was passed at a concentration of 100 nM. Kinetics was analyzed by BIAevaluation software, version 3.1.

Identification of Filamin as an FAP52-binding Protein-To
identify binding partners of FAP52, we carried out IP experiments from the lysates of cultured CEHF by using affinitypurified rabbit anti-FAP52 antibodies (Affi-K7). IP followed by an SDS-PAGE and silver staining revealed reproducibly a 280-kDa polypeptide band that was seen in the experimental (Fig.  1A, lane 2) but not in the control precipitations (Fig. 1A, lane 1). A band of a similar molecular weight was also detected in the affinity binding ("pull-down") experiments in which the cell lysate was passed over a column in which GST-FAP52 was coupled onto GSH-conjugated Sepharose beads (Fig. 1B, lane  2). No such band was seen in control experiments with GSTcoated beads (Fig. 1B, lane 1). In pull-down but not in IP experiments another prominent band of about 220 kDa was seen.
The 280-kDa band was excised from the CBB-stained gel and subjected to in-gel trypsin digestion. The extracted peptides then underwent analysis by nanoelectrospray tandem mass spectrometry. The partial primary structures of several peptides in the mass spectrum of the 280-kDa band were compared with the sequences in the data bases. In five cases there was a match with GST, in four cases with FAP52, in one case with myosin, and in one case with filamin (ABP-280 (13)). We also analyzed the 220-kDa band seen in Fig. 1B. Five of seven peptides showed a match with myosin heavy chain. We concluded that the myosin sequence found among the peptides derived from the 280-kDa band is actually originating from the more abundant material in the 220-kDa band.
Because the filamin-like sequence derived from the band that was not seen in the controls and because the partial sequence and the molecular weight of the band match with those of filamin, we decided to explore the possibility of an interaction between filamin and FAP52. The presence of FAP52, GST, and myosin sequences among the peptides derived from the 280-kDa band was deemed unauthentic and was most probably due to an abundance of these proteins in the type of experiments that were carried out.
Association between Filamin and FAP52 in Cells in Vivo-The association between filamin and FAP52 was further explored by IP of FAP52 from a lysate of CEHF cells and then analyzing the precipitate by IB for coprecipitation of filamin. In a reciprocal experiment with HA-tagged FAP52 (FAP52-HA)transfected HeLa cells, anti-filamin immunoprecipitate was analyzed for co-IP of FAP52-HA. As shown in Fig. 2A, filamin was present in the precipitate obtained by using Affi-K7 but not in the control precipitates ( Fig. 2A, lanes 2 and 1, respectively). On the other hand, FAP52-HA was seen in the antifilamin but not in the control immunoprecipitates (Fig. 2B,  lanes 2 and 1, respectively). These data show that FAP52 and filamin co-immunoprecipitate from the cell lysates and suggest that there could be a direct interaction between FAP52 and filamin.
Demonstration of a Direct Interaction between FAP52 and Filamin-To determine whether the observed co-IP of FAP52 and filamin is due to a direct interaction between them or is mediated by some other protein(s), we carried out direct binding assays. In OL assay, bacterially produced FAP52 or GST fusions of the C-terminal fragments of filamin were run in SDS-PAGE, transferred to an NC membrane, and then overlaid with purified filamin or FAP52, respectively, followed by an incubation with relevant antibodies and an ECL detection. As seen in Fig. 3A, filamin binds to FAP52 (Fig. 3A, lane 2) but not to GST that was used as a control (Fig. 3A, lane 1). In In a reciprocal experiment, binding of FAP52 to filter-anchored filamin was tested. For that purpose, bacterially produced C-terminal fragments of filamin in the fusion with GST were used. In OL, binding of FAP52 is seen to GST-Fil-(1524 -2283) (Fig. 3B, lane 1) but not to GST-Fil-(2283-2496) (lane 2) or to GST-Fil-(2495-2647) (lane 3). In Fig. 3B, lanes 4 -6, a parallel SDS-PAGE run of the same fusion proteins is shown. The polypeptides were transferred to NC filter and visualized by the anti-GST antibody and the ECL detection system. These data show that there is a direct interaction between FAP52 and filamin and that in filamin the interaction site resides in the fragment encompassing the residues 1524 -2283.
The interaction between filamin and FAP52 was further explored by SPR analysis. In these experiments, anti-GST antibody was immobilized on a chip and used to capture the GST-FAP52 fusion protein. The sensorgram (Fig. 3C) obtained upon injection of chicken gizzard filamin at concentrations ranging from 0 to 800 nM onto the GST-FAP52-coated chip shows a high affinity binding between filamin and FAP52. The association and dissociation rate constants k a and k d were determined to be 1.06 ϫ 10 5 M Ϫ1 s Ϫ1 and 1.02 ϫ 10 Ϫ3 s Ϫ1 , respectively, yielding a dissociation rate constant (K D ϭ k d /k a ) of 9.64 ϫ 10 Ϫ9 M. No binding was observed in the control experiments in which the GST-coated chip was overlaid with filamin ( Fig. 3C, inset).
Mapping of the Interaction Site in FAP52-Pull-down assay and SPR were used to map the filamin-binding region in FAP52. For that purpose, a large number of truncation mutants of FAP52 were expressed as GST fusions in bacteria, purified, and used in the assays.
For the pull-down assays, GST-FAP52 or its mutants were immobilized on GSH-Sepharose beads and incubated with chicken gizzard filamin. After collecting and washing the beads, they were run on SDS-PAGE. The electrophoretically separated proteins were then transferred to an NC membrane, followed by an incubation with the mouse anti-filamin and the ECL detection. Fig. 4A shows that anti-filamin reactive material could be harvested not only from the beads carrying the full-length GST-FAP52 (lane 5) but also those carrying the peptides 1-184, 1-219, and 1-359 (lanes 2-4). No binding of filamin was seen to beads carrying the more N-terminal pep-  6 -11).
For SPR analysis, the GST fusion of FAP52 or its mutants were immobilized onto a sensor chip, and chicken gizzard filamin was then passed over the chip at a concentration of 100 nM. Both the OL and the SPR results indicated, in total conformity, that the residues 146 -184 have an essential role in the binding of filamin. The capability of this region alone to bind filamin was tested with SPR, in an experiment in which filamin was passed over the GST-FAP52-(146 -184)-coated chip. No binding was observed (Fig. 4B, inset). Thus, it can be concluded that the region spanning aa 146 -184 is necessary but not sufficient for the binding of filamin. For the binding, a sequence N-terminal to the aa 145 is needed as indicated by binding with an N-terminal 1-184 peptide.
Mapping of the Interaction Site in Filamin-The FAP52binding site in filamin was determined by pull-down experiments and by SPR by employing sequentially truncated mutants of the C-terminal half of filamin in fusion with GST.
For the pull-down experiments, filamin mutants in fusion with GST were immobilized onto the GSH-Sepharose beads which were then incubated with the recombinant FAP52. Binding of FAP52 to the beads was analyzed by SDS-PAGE followed by a transfer to an NC filter, IB with Affi-K7, and ECL detection. Fig. 5A shows the binding of FAP52 to the filamin peptides  1524 -1858, 1524 -1956, 1524 -2064, 1524 -2132, 1524 -2172,

FIG. 3. Filamin binds FAP52 in OL assay and in SPR analysis.
A, FAP52, derived from the bacterially produced GST-FAP52 by a thrombin cleavage (lane 2), and, as a control, bacterially produced GST (lane 1), were run on 12% SDS-PAGE, transferred to NC filter, and then overlaid with the chicken gizzard filamin. Binding of filamin was probed with the anti-filamin antibodies and the ECL detection system. The positions of GST and FAP52 in a parallel, CBB-stained 12% SDS-PAGE gel are shown in the lanes 3 and 4, respectively. B, the bacterially produced truncation mutants of filamin in fusion with GST were run on 10% SDS-PAGE and transferred to an NC filter. The filter was then overlaid with the recombinant FAP52. The binding of FAP52 was probed by incubating with Affi-K7 and by ECL detection. The position of the same polypeptides (lanes 4 -6, respectively) in the filter is shown by parallel 10% SDS-PAGE and IB with the anti-GST antibody followed by ECL detection system. Migration of the molecular mass markers is shown on the left. C, SPR analysis. GST-FAP52 was immobilized on a CM5 sensor chip, and chicken gizzard filamin was passed over the chip in mobile phase at concentrations ranging from 0 to 800 nM. Inset, sensorgram of a control experiment with GST anchored to the chip. Arrows indicate the beginning and the end of filamin injection. and 1524 -2283 (lanes 3-8). There was no binding of FAP52 to the peptides 1524 -1780 and 1524 -1664 (lanes 1 and 2).
For SPR analysis, GST-Fil-(1524 -2283) or its truncated mutants were captured onto the sensor chip, and the recombinant FAP52 was passed over the chip. In Fig. 5B, binding of FAP52 to the truncated filamins is shown. A strong binding is seen to the peptides 1524 -1780, 1524 -1858, 1524 -1956, 1524 -2064, 1524 -2132, 1524 -2172, and 1524 -2283 and a weak binding to 1524 -1664. In Fig. 5C, SDS-PAGE analysis of the purified, bacterially produced GST fusions of the filamin fragments used in the assay is shown.
The results of the pull-down experiments suggest the residues 1780 -1858, corresponding to the repeat 16, as the binding region for FAP52. Closely similar binding data were obtained from the SPR analysis, except that Fil- (1524 -1780), which in the pull-down assay did not bind FAP52, also showed a strong binding. The different binding of Fil-(1524 -1780) in the two assays is probably not due to artifactual conformational changes because in both cases the filamin-GST chimera was anchored to the substrate via a linker (anti-GST antibody in SPR and GSH in pull-down assay). Rather, it may reflect the different sensitivities of the techniques. The lacking or low binding of Fil-(1524 -1664) in both assays clearly indicates, however, that the principal binding of FAP52 to filamin is in the region aa 1664 -1858 spanning the repeats 15-16.
Colocalization of FAP52 and Filamin in Cultured Fibroblasts-Biochemical studies described indicated a direct interaction between FAP52 and filamin. Thus, it was expected that there is at least a partial colocalization between FAP52 and filamin in cells in vivo. This was studied by using a confocal laser scanning IFM and IEM. We applied double immunostaining with Affi-K7 and the anti-filamin antibodies. Antibodies to focal adhesion protein paxillin and a filamentous actin-binding compound phalloidin also were used for comparative purposes.
Confocal IFM of well spread (Fig. 6A) and spreading CEHFs (Fig. 6B) showed that FAP52 is typically associated with the focal adhesions (verified by a double staining for paxillin; not shown) and filamin along the actin fibers (verified by visualizing actin fibers with phalloidin; not shown). At a closer scrutiny and in merged images, distinct overlaps of the staining patterns could be seen. One was in the areas where the filamindecorated filaments merge with the focal adhesions (Fig. 6A,  arrows). The overlap area seems to correspond to the more proximal part of the focal adhesion or a putative "juncture" region in which the fibers are linked to the focal adhesions. Another distinct overlap was seen in the spreading cells with a smooth-contoured forward-moving front region (Fig. 6B, arrow). In them, in addition to the focal adhesion structures, a distinct colocalization was seen in distinct spots that were located at regular intervals along the very edge of the cell (Fig.  6B, arrowheads).
Immunoelectron Microscopy- Fig. 7 is an electron micrograph of a cryosection of cultured CEHFs. The sections were double-stained with antibodies to FAP52 and filamin. The localization of the bound antibodies was revealed by secondary antibodies conjugated to small (inner diameter, 5 nm) and large (inner diameter, 10 nm) gold particles, respectively. The larger particles were seen most prominently along filamentous structures that could be discerned even though the overall contrast after uranyl staining was quite low (Fig. 7, arrows). The smaller particles, corresponding to FAP52, on the other hand, were only seen in the peripheral parts of the cells, and especially in elongated structures which, by their morphology and localization, represent focal adhesions (Fig. 7, arrowheads). In Fig. 7, B and C, which represent higher magnifications of the areas indicated by rectangles in Fig. 7A, a colocalization of FAP52 (smaller particles) and filamin (larger particles) was seen (circled areas). Interestingly, the gold particles were often linearly arranged, strongly suggesting that they are associated with an underlying filamentous structure (Fig. 7, B and C, bar-ended arrows).
Rearrangement of Actin and Filamin Cytoskeleton in Cells Overexpressing FAP52-Direct interaction between FAP52 and filamin, as shown by biochemical techniques, and a close colocalization between FAP52 and filamin, as shown by confocal microscopy and IEM, prompted us to study the changes in the distribution of filamin in cells overexpressing FAP52 or its specific regions/domains. Due to the actin cross-linking properties of filamin, the effect of the forced expression of FAP52 on the organization of actin cytoskeleton was also investigated. For these purposes, confocal microscopy of CEHFs transiently transfected either with HA-tagged FAP52 (FAP52-HA) or its truncated forms was used. Double staining with appropriate antibodies and phalloidin was used to analyze the distributions of FAP52-HA, filamin, and actin.
A typically elongated or polygonal fibroblastic morphology was seen in most naive (untransfected) CEHF cells (Fig. 8A,  arrows). Cells overexpressing HA-FAP52, on the other hand, displayed several types of abnormal morphologies. Some of them were overly elongated and showed several cell surface protrusions (Fig. 8A, arrowheads). More typical were cells with an outlook suggesting abnormal or arrested spreading (Fig.  8B). In some cells, prominent ruffling edge-type formations were seen (Fig. 8, C and D, arrows). A co-accumulation of actin and FAP52 was also seen in the ruffling edge formations (Fig.  8, C and D).
In most transfectants, a distinctly abnormal filamin distribution was seen; instead of a linear organization along actin fibers, numerous dot-like densities were seen especially close to the cell surface (Fig. 8B, red). Almost invariably, the dots also contained FAP52 (Fig. 8B, green). Instead of stress fibers, only thin and shorter filaments were seen running from one dot to another. This was also seen in double staining with phalloidin; actin was only seen as distinct dots that also contained FAP52 and as thin filaments interconnecting the dots as a meshwork (Fig. 8D).
We also carried out transfection experiments using the HAtagged N-terminal peptide of FAP52 (aa 1-293; FAP52 Nt -HA) encompassing the filamin-binding site. For the most part similar changes in actin and filamin organization were seen as with the overexpression of the full-length protein. A specific change seen with the N-terminal peptide was, however, the presence of transfectants with numerous and often very long filopodial extensions on the cell surface (Fig. 8, E and F). In more quantitative terms, the proportion of the cells with numerous filopodia was 30% in untransfected cells and 60 and 64% in FAP52-HA-and FAP52 Nt -HA-transfected cells, respectively, with cells with very long extensions (which were sometimes severed from the cell body; Fig. 8F, arrow) seen exclusively in FAP52 Nt -HA transfectants. No major changes were seen upon transfection with the constructs encoding the SH3 domain of FAP52 (not shown).
Due to the presence of FAP52 in the focal adhesions in naive cells, we also wanted to see what happens to focal adhesions in cells overexpressing FAP52-HA or FAP52 Nt -HA. Surprisingly, even in cells in which there were major distortions of the cell FIG. 6. FAP52 and filamin colocalize in cultured CEHFs. In double staining, FAP52 (green) shows a partial colocalization with filamin (red). FAP52 is in the elongated focal adhesions, whereas filamin is seen as filamentous structures. In the merged image, a distinct overlap of the staining is seen in the area corresponding to the centrally oriented part of the focal adhesions (arrows and arrowheads). Bars, 10 m. morphology and actin network, there were still distinct focal adhesions to be seen as revealed by a staining with antipaxillin. They were not normal, however. Instead of a radially oriented axis and regular spacing, they displayed a more haphazard orientation. In most cases they were also smaller and fewer in number than their counterparts in the naive cells (61 versus 40 focal adhesions per cell in nontransfected versus transfected cells; n ϭ 61 and 57, respectively; Fig. 8G).
Based on these studies, it is clear that the overexpression of the full-length FAP52 and its N-terminal, filamin-binding sitecontaining peptide brings about major derangements of the actin/filamin organization accompanied by unscheduled and exaggerated filopodia formation and minor derangements in the focal adhesion formation and architecture. DISCUSSION In this study, we provide evidence that FAP52 binds filamin and suggest that, via filamin, FAP52 is involved in the regulation of the actin cytoskeleton. Filamin, as a putative ligand for FAP52, was discovered and identified by using affinity purification and mass spectrometry. An in vivo association between FAP52 and filamin was verified by reciprocal co-IP studies, and a direct interaction between the proteins was demonstrated by using blot OL assays and SPR analysis. Also the results from the studies on the distribution of FAP52 and filamin in cultured cells are consistent with an interaction between the proteins. Confocal microscopy of the cells overexpressing FAP52 or its filamin binding fragments also showed that an aberrant expression of FAP52 leads to a gross derangement of the actin cytoskeleton organization. This is taken as evidence that FAP52, via its interaction with filamin, is involved in the regulation of the actin cytoskeleton organization.
Filamin emerged as a putative binding partner of FAP52 from a pull-down experiment employing GST-FAP52 and from co-IP experiments from the solute of cultured fibroblasts. Positive identification was based on a direct sequencing of the 280-kDa band that was repeatedly seen in the pull-down experiments. Occasionally, also a smaller band of about 220 kDa was seen which turned out to be heavy chain of myosin. In subsequent studies we were, however, unable to demonstrate a direct interaction between myosin and FAP52. This observation implies that FAP52 could be involved in a multimolecular complex also encompassing myosin.
Filamin is an actin cross-linking protein and is involved in the maintenance of the cytoskeletal architecture and in the adhesion and migration of cells (14). Most of the protein, which occurs as a dimer, is composed of 23 ϳ96-aa long repeats. They are flanked by an N-terminal actin-binding domain and a Cterminal homodimerization domain which both are important for the organization of the cortical actin network (14,15). They determine the capacity of filamin to direct the positioning of actin and to form orthogonal actin/filamin networks (13,16). Filamin is a multifunctional protein, and a great number of proteins binding to its repeat domains have been described. Most of them are either receptors or signaling proteins and are thought to be involved in synchronizing cell surface events with changes in the cytoskeletal architecture (14,17). Especially pertinent in regard to the present study is a binding of filamin to some integrins that are closely linked both to cell adhesion and actin organization. Thus, filamin binds to the cytoplasmic tail of integrin ␤ 1 (18), the domain that is responsible for the focal adhesion targeting (19) and cytoskeletal linkage of the molecule. It also binds to integrin ␤ 7 cytoplasmic domain (18). In platelets, glycoprotein Ib-IX (GPIb-IX), the receptor for von Willebrand factor, is constitutively associated with filamin (20). Filamin also binds to the cytoplasmic tail of ␤ 2 integrin close to the binding site of ␣-actinin, another actin cytoskeleton-associated protein (21).
In cultured fibroblasts, FAP52 is seen in focal adhesions where it colocalizes with paxillin, talin, and vinculin (1). It is a phosphoprotein, being phosphorylated on serine but not on threonine or tyrosine residues. Up until now, no "ligands" for FAP52 have been described. FAP52 has a distinctly multidomain structure. In the N-terminal part, encompassing the aa residues 1-145, there is a FER-CIP4 homology domain (22) FIG. 8. Double immunofluorescence microscopy shows an altered cellular morphology and filamin, actin and focal adhesion organization in cells overexpressing FAP52. A, some of the transfected cells (arrowheads) exhibited an elongated cell shape with cell surface protrusions. Expression of HA-tagged FAP52 (green) and filamin (red) was detected. Untransfected cells are shown by arrows. B, an abnormal dot-like staining for filamin (red) was seen in many of the cells overexpressing FAP52. The dots also contain FAP52 (green), as seen in the merged image. C, a ruffling edge-type staining (arrows) of some cells overexpressing FAP52 (green). In the central part of the cell, FAP52 forms a dot-like staining pattern. The dots are distributed along the actin fibers (red), as seen in the merged image. D, co-distribution of FAP52 (green) and actin (red) in the areas of the ruffling edge formation (arrows). E and F, cells overexpressing FAP52 Nt -HA (green) typically exhibit, at their edges, long filopodia that are sometimes severed from the cell body (F, arrow). G, the cells overexpressing FAP52 (green) exhibit smaller and more randomly localized focal adhesions (red) than naive cells. Bars, 10 m. and, encompassing aa residues 146 -280, a highly ␣-helical region. In the C terminus, there is an SH3 domain. In the middle part of the molecule, encompassing the aa residues 281-389, there is a flexible linker region that does not form any recognizable fold (1).
The putative binding site for filamin, as shown by the OL and SPR assays employing truncated FAP52, resides in the N-terminal half of the protein with the region spanning from aa 146 to 185 as its indispensable estate. Interestingly, in the secondary structure prediction analysis, this region shows a high degree of ␣-helicity and a propensity to a coiled-coil arrangement (1). In this respect it seems to differ from the other known filamin-binding peptides which, according to our own analysis (not shown), do not display any notable ␣-helicity. It is noteworthy, however, that there is no obvious similarity between the filamin-binding regions of the various known filamin ligands except for a cluster of charged residues which, according to helical wheel analysis, could serve as a platform to filamin binding in integrins and GpIb␣ (21). This variability, on the other hand, is quite expected in the light of the scattered nature of the binding sites in filamin.
In filamin, the binding site for FAP52 was localized to the region of aa 1524 -1858. This area corresponds to the repeats 14 -16 and the calpain-sensitive hinge region 1. Importantly, binding of furin (23), tumor necrosis factor receptor-associated factor 2 (17), dopamine D-receptor (24,25), and calcium-sensing receptor (26,27) have been localized to the same area. Among some other filamin-binding proteins, GpIb␣ binds to the repeats 17-19 (28,29), SEK-1 to the repeats 21-23 (30), and presenilin-1 to the repeats 22-24 (31). Thus, it is apparent that sufficient unique information resides in the repeats to create specific binding sites for a number of proteins. Binding of FAP52 close to the hinge region (between the repeats 15 and 16) could affect the flexibility of the filamin dimer and thus could have an effect on the overall organization of the filamin cross-linked actin network.
In cultured fibroblasts filamin is mostly seen associated with the stress fibers (32,33). FAP52, on the other hand, is associated with the focal adhesions. In double staining experiments they mostly showed a complementary pattern in that FAP52 was seen in the focal adhesions where actin filaments abut head on. At a close examination and in merged images it was obvious, however, that at the juncture of the focal adhesions and the microfilaments, there is an area where the distributions of FAP52 and filamin overlap. It was seen especially in the focal adhesions in well spread and stationary cells. On the other hand, there were also focal adhesions that did not show any distinct overlap with filamin staining. This suggests that FAP52-filamin association is only seen in structures representative of a certain subclass of focal adhesions.
There are several detailed studies on the localization of filamin in cultured cells at both the light and electron microscopic levels. Based on them, filamin has been shown to be associated with stress fibers and membrane ruffles (32,34), microspikes (32), and in a delicate actin-based subcortical net and polygonal actin filament nets (33). In stress fibers, filamin is distributed in a discontinuous manner ("spotty" or "patchy") at shorter or longer intervals, depending on the cell type (33,35,36).
The present results, based on a careful analysis of confocal micrographs, clearly show that FAP52 is not only a focal adhesion-associated protein but that it is also present in the juncture area connecting focal adhesions and stress fibers and even to varying lengths to the distal ends of stress fibers proper. This could be taken to implicate a "supportive" function in this hinge region to FAP52 and/or to the components to which it might bind. Given that there is a juncture of some sorts be-tween the focal adhesion and its abutting stress fiber, it is reasonable to speculate that there are overlapping elements that either guide the well orchestrated co-evolvement and/or reinforce the structure of these two complex architectural organizations. Interestingly, remarkably little is known about the molecular mechanisms of the link between these two structures, especially when compared with the detailed knowledge on the structure and composition of the focal adhesions and stress fibers in separation.
The close association with filamin strongly suggests a role for FAP52 in the actin organization. This was clearly demonstrated by the overexpression experiments in which major rearrangements of the actin cytoskeleton were seen. They include the extensive filopodia formation in response to an overexpression of the full-length FAP52 and an emergence of a deranged net of actin fibers instead of stress fibers in cells overexpressing the N-terminal domain of FAP52. In this respect, it is interesting that an overexpression of the Ras-related small GTPase RalA, which also binds filamin, elicits formation of filopodia on the surfaces of the Swiss 3T3 cells and also recruits filamin to the sites of filopod formation (9). These data along with the known actin-organizing properties of the homologs of FAP52 point to the role of FAP52, via its binding to filamin, as a major regulator of the actin cytoskeleton.