FAP52, a Novel, SH3 Domain-containing Focal Adhesion Protein*

Src-homology 3 (SH3) domain is a 60–70-amino acid motif present in a large variety of signal transduction and cytoskeletal proteins. We used reverse transcriptase-polymerase chain reaction with degenerate and specific primers and chicken brain mRNA to clone a cDNA that codes for a novel SH3 domain-containing protein. The sequence predicts a 448-amino acid polypeptide with a molecular mass of 51,971 daltons. In the amino terminus, it shows a very high propensity for α-helicity, suggesting coiled-coil and possibly a higher order oligomeric arrangement. In the carboxyl terminus, there is a unique SH3 sequence. In Northern blotting, a major 3.7-kilobase and a minor 7.2-kilobase transcript was detected in most chicken tissues. In immunofluorescence microscopy and immunoelectron microscopy on cultured chicken fibroblasts, the protein was localized to focal adhesions in which it showed a distinct codistribution with the focal adhesion proteins vinculin, talin, and paxillin. Phosphoamino acid analysis showed that in cultured chicken heart fibroblasts, the protein contains phosphoserine, but no phosphothreonine or phosphotyrosine, and that the phosphorylation is not dependent on fibronectin. We propose this protein the name FAP52, for Focal Adhesion Protein of52 kDa, and suggest that it forms part of the multimolecular complex constituting focal adhesion sites.

Focal adhesions (FAs) 1 are specialized membrane domains in cultured cells that mediate the attachment of cells to the growth substratum and extracellular matrix. They consist of pericellular and transmembrane structures connected to the actin-based cytoskeleton. The biochemical composition of FAs provides a large inventory of proteins, including structural proteins, such as cytoskeletal proteins vinculin, paxillin, and talin, integrins which provide the transmembrane linkage to the extracellular matrix, and several regulatory and signaling molecules such as proteases, protein kinases, and phosphatases (1,2).
According to the prevailing view based on the characterization of the individual components and intermolecular associations, FAs are sites of complex interactions, which not only structurally couple integrins to the actin cytoskeleton but are also "hot spots" of cell signaling (1,2). Accordingly, a diversity of signaling pathways have been shown to be activated by cell-matrix contacts on an FA-dependent manner (3). From a wealth of structural, biochemical, and genetic information, it has become apparent that the protein-protein interactions within FAs are based on conserved protein modules. These include, e.g. in the regulatory components of the FAs, the well known Src-homology 2 (SH2) and Src-homology 3 (SH3) domains that are important, e.g. in the substrate recognition of kinases and in the protein targeting, respectively (4).
SH3 domain is a 60 -70-amino acid-long protein motif, which occurs widely, often in conjunction with a SH2 domain, in proteins of the signal transduction pathways in which it is involved in mediating protein-protein interactions (5)(6)(7). It is also present in various cytoskeletal proteins, such as a spectrin and myosin I (8,9). The high degree of divergence of the known SH3 domains suggests that our current catalog of SH3-containing proteins is by no means exhaustive. In an effort to find new SH3-containing proteins, we have used polymerase chain reaction (PCR) and degenerate primers, designed on the basis of the conserved sequences in a repertoire of known SH3-containing proteins (10). Here, we report the identification, cDNA cloning, sequence analysis, subcellular distribution, and phosphoamino acid analysis of a novel SH3 domain-containing protein, which, based on the immunofluorescence studies, appears to be a focal adhesion-associated protein. Due to its subcellular localization and molecular mass, we propose it a name Focal Adhesion Protein of 52,000 daltons, FAP52.

EXPERIMENTAL PROCEDURES
General Procedures and Computer Programs-The solutions, buffers, and procedures for standard purification and precipitation of DNA, for restriction enzyme digestion, and ligation reactions were as described in Sambrook et al. (11). For sequence analysis by the dideoxynucleotide chain termination method of Sanger (12), the T7 sequencing kit (Pharmacia Biotech Inc.) was used. Synthetic oligonucleotides, obtained from the Oligonucleotide Core Facility of Biocenter Oulu, were synthesized on a 392 DNA synthesizer (Applied Biosystems). For sequence alignments, the program CLUSTAL W (13) was used. Secondary structure prediction was by PHD, an automatic mail server for secondary structure prediction (14 -17).
cDNA Cloning and Sequencing-Reverse transcription (RT) and PCR with degenerate oligonucleotide primers were used to amplify fragments of mRNAs encoding for SH3-containing proteins expressed in a developing chicken brain. For that purpose, total RNA was prepared from chicken embryonal (gestation age 16 days) brain tissue by the guanidine isothiocyanate method (18). RNA was reverse-transcribed using highly degenerate downstream primers designed on the basis of the moderately well conserved tryptophan-containing region (6) in the C termini of many SH3 proteins (in ␣-spectrin this corresponds to the amino acids residues 1003-1009; Ref. 9). The primer sequence was 5Ј-VWVMVCYTYCCACCARTC-3Ј (for the symbols, see Nomenclature Committee of the International Union of Biochemistry (19)), with a degeneracy of 864.
RT was carried out by using avian myeloblastosis virus reverse transcriptase (Boehringer Mannheim GmbH). For PCR, an aliquot of cDNA was amplified by using the downstream primers described above and upstream primers based on the consensus sequence of a conserved N-terminal portion of the SH3 domain (6) which in ␣-spectrin corresponds to the amino acid residues 972-978 (9). The primer sequence was 5Ј-KTSHKDGCDYTBTAYGAYTWY-3Ј, with a degeneracy of 20,736. After an initial melting step of 5 min at 95°C, 35 cycles were carried out (1 min, 95°C; 1 min, 45°C; 1 min, 72°C), followed by a 15-min final extension step in the following mixture: 10 mM Tris-HCl, pH 8.3, 50 mM KCl, 15 mM MgCl 2 , 200 M dNTPs, with 2.5 units of AmpliTaq DNA polymerase (Perkin-Elmer).
The PCR products were resolved in agarose gel electrophoresis by using 4% NuSieve GTG-agarose (FMC Bioproducts). The bands of the expected size of approximately 100 bp were cut out and the DNA purified by using a silicon-based matrix (Promega). The purified DNA was treated with Klenow Fragment of DNA polymerase I (Boehringer Mannheim GmbH) and subcloned into SmaI-cut pGEM7-vector (Promega). The construct was transfected into competent XL1-blue cells (Stratagene) by using the standard CaCl 2 precipitation. The sizes of the insert from several clones were determined, and several inserts corresponding to the expected size of about 100 bp were sequenced.
Several of the sequenced insert DNAs had a sequence comparable to a unique SH3 domain. The nucleotide sequence of one of them was extended by using PCR cloning from a chicken brain cDNA library (Chicken Brain 5Ј-Stretch cDNA in gt10, Clontech) and from our own chicken brain cDNA library (9). Specific sense (5Ј-TTAAAGCAGGG-GATGAGTTAACTAAAA-3Ј) and antisense (5Ј-TTTTAGTTAACTCA-TCCCCTGCTTTAA-3Ј) primers, corresponding to the middle portion of the clone obtained from the previous step, were designed. PCR cloning was carried out by using the template DNA and the above primers along with gt10 forward (5Ј-CTTTTGAGCAAGTTCAGCCTGGTTAAG-3Ј) and gt10 reverse (5Ј-GAGGTGGCTTATGAGTATTTCTTCCAGGGTA-3Ј) primers. Several bands were obtained. The gel-purified DNA was ligated into a T/A cloning vector (Invitrogen). Both strands were sequenced by using SP6 and T7 promoter primers (Promega). Three clones, two extending 3Ј and one 5Ј of the original sequence, were harvested. Their sequences were obtained by using specific primers (Fig. 1A).
For the PCR amplification of the 5Ј-end of the corresponding mRNA, the 5Ј-AmpliFINDER RACE kit (Clontech) was used. The oligonucleotide 5Ј-GCCAGTTTCTCTCTTTA-3Ј, corresponding to the 5Ј-sequence of the 5Ј clone obtained from the previous step, was used as a primer for the first strand synthesis. A nested primer 5Ј-TAATCCTCCATCATTT-GCTTATG-3Ј was used for PCR amplification. The longest clone obtained extended about 600 nucleotides from the priming site and overlapped the 5Ј clone obtained from the cDNA library screening by about 200 nucleotides. The authenticity of the sequence obtained by RACE was verified by carrying out RACE with different primers and also by independent RT-PCR cloning by using thermostable reverse transcriptase enzyme (GeneAmp thermostable rTth reverse transcriptase RNA PCR kit, Perkin-Elmer).
Northern Blot Analysis-Total RNA was prepared from the embryonic chicken brain, lung, intestine, gizzard, liver, cardiac muscle, skeletal muscle, skin, kidney, and eye by using guanidine isothiocyanate method (18). Poly(A) ϩ -rich RNA was then purified by affinity chromatography on an oligo(dT) column (Pharmacia). mRNA from cultured chicken embryo heart fibroblasts (CEHF) was purified by using Quick-Prep Micro mRNA purification kit (Pharmacia). Five g of total RNA (tissue samples) and 5 g of mRNA (CEHF), along with the RNA size markers (Promega), was then resolved in formaldehyde-agarose electrophoresis and transferred to Hybond-N nylon filter (Amersham Corp.). The cDNA clone obtained from the chicken brain cDNA library screening and corresponding to the nucleotides 478 -1475 in the final sequence (see Fig. 1B) was labeled with [ 32 P]dCTP to a specific activity of 1 ϫ 10 9 cpm/g using a hexanucleotide random priming kit (Boehringer Mannheim GmbH).
Cell Culture-CEHF cultures were established from the explants of the hearts of 16-day-old chicken embryos. They were grown in Eagle's essential medium (Life Technologies, Inc.), supplemented with 10% FCS (Life Technologies, Inc.), 2 mM glutamine, 1% nonessential amino acids, and antibiotics (penicillin, 100 units/ml; streptomycin sulfate, 100 g/ml; amphotericin B, 0.25 g/ml) at 37°C in 5% CO 2 in a humidified atmosphere. For some experiments, the cells were grown on coverslips coated with plasma fibronectin (10 g/ml), which was purified from an outdated human plasma obtained from the Finnish Red Gross Blood Transfusion Service (Helsinki, Finland) by using gelatin-Sepha-rose affinity chromatography (20). The cells were used for the experiments between the second and eighth passages.
Preparation of Antibodies-To generate antibodies, the cDNA corresponding to nucleotides 482-1471 of the full-length sequence (Fig. 1B) was amplified by using PCR and subcloned into a pET-5a prokaryotic expression vector (Novagen) and expressed in bacteria. The cells were suspended in a lysis buffer (140 mM NaCl, 2.7 mM KCl, 10 mM Na 2 HPO 4 , 1.8 mM KH 2 PO 4 , 1% Triton X-100) and lysed by sonication. Laemmli's sample buffer was added to the lysates, which were then cleared and resolved in 10% SDS-PAGE. After staining with Coomassie Blue, pieces of gel containing the expressed polypeptide band of 48 kDa, corresponding to the polypeptide expressed from the insert, was cut out and sonicated briefly in 50% methanol to remove the dye. Thereafter, the gel was washed and neutralized and the polypeptides electroeluted from the gel by using Bio-Rad model 422 electroeluter. The electroeluted protein was dialyzed against 50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 0.1% Triton X-100 and used for immunizations. The authenticity of the protein used for immunizations was verified by partial amino acid sequencing (Procise 494, Perkin-Elmer). One of the obtained antisera, denoted K7, was selected for further studies on the basis of its specificity as assessed by immunoblotting. It was further affinity-purified on CNBr-activated Sepharose 4B beads (Pharmacia) coated with the electroeluted 48-kDa polypeptide. The affinity-purified antiserum is denoted Affi-K7.
Immunoblotting and Immunoprecipitation-For immunoblotting, CEHFs were washed with PBS and then scraped in radioimmune precipitation buffer (158 mM NaCl, 5 mM EDTA, 0.1% SDS, 1% sodium deoxycholate, 0.2 mM sodium orthovanadate, 1% Triton X-100 and 10 mM Tris-HCl, pH 7.2) supplemented with the protease inhibitors NaF (50 mM), aprotinin (200 nM), leupeptin (1 M), phenylmethylsulfonyl fluoride (0.25 mM), and benzamidine (0.5 mM). For immunoblotting of tissues, small pieces of 18-day-old chicken embryos were freshly excised and frozen immediately. The frozen samples were sliced with a razor blade on ice, and the pieces were homogenized in an extraction buffer (1% SDS, 40 mM dithiothreitol, 5 mM EDTA, 7.5 mM sodium phosphate, pH 7.4) supplemented with protease inhibitors as above with a Dounce homogenizer. The extracts were boiled in Laemmli's sample buffer for 5 min and subjected to sonication. For immunoprecipitation of CEHFs, the cells were lysed with lysis buffer (50 mM Hepes, pH 7.5, 150 mM NaCl, 1 mM EGTA, 10% glyserol, 1.5 mM MgCl 2 , 0.2 mM sodium orthovanadate, and 1% Triton X-100) supplemented with protease inhibitors as above, and the extracts were clarified by centrifugation. The immune complexes with K7 antiserum, Affi-K7, and the preimmune serum were collected by incubation with protein A-Sepharose (Sigma). The protein concentrations in the supernatants were estimated using Bio-Rad protein assay reagent. Constant amounts of protein were separated by 10% SDS-PAGE and subjected to Western blotting. Thereafter, the electrophoretically separated polypeptides were transferred onto nitrocellulose filters (Schleicher & Schuell) as described by Towbin et al. (21). The filters were then incubated in TTBS buffer (50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 0.1% Tween 20) with Affi-K7, followed by peroxidase-conjugated anti-rabbit immunoglobulins (Biosys). The blots were developed with diaminobenzidine (0.5 mg/ml in TBS) with H 2 O 2 and 0.03% NiCl or with the ECL method (Amersham).
Immunofluorescence Microscopy-For immunofluorescence microscopy, the cells were grown on glass coverslips. They were briefly washed in Hanks' salt solution and then fixed with 4% formaldehyde in a cytoskeleton-stabilizing buffer (100 mM Pipes, pH 6.8, 5 mM EGTA, 2 mM MgCl 2 (22), with 0.2% Triton X-100) for 10 min at room temperature. After being washed in PBS, the cells were post-fixed in Ϫ20°C methanol for 5 min, incubated with 10% PBS-glycine (PBS with 20 mM glycine) for 30 min, and washed with PBS. They were then overlaid with Affi-K7 at 4°C for 30 min followed by Texas Red-conjugated anti-rabbit IgG (Jackson ImmunoResearch). For double labeling experiments, the primary antibodies to talin (mouse monoclonal; Developmental Studies Hybridoma Bank, Department of Biology, University of Iowa), to paxillin (mouse monoclonal: Zymed Laboratories Inc.), to vinculin (mouse monoclonal; Biohit-Locus, Inc.), and the secondary antibodies tetramethylrhodamine isothiocyanate-conjugated (Caltag Laboratories) or fluorescein isothiocyanate-conjugated (Dako A/S) anti-mouse IgG antibodies were used. The cells were viewed under Olympus BH2 fluorescence microscope equipped with appropriate filters.
Immunoelectron Microscopy-For ultrastructural localization studies, double labeling immunoelectron microscopy on whole mount cytoskeletal preparations of cultured cells was done by following the procedure described elsewhere (23). Briefly, CEHFs grown onto halfconfluence on gold grids were permeabilized and fixed by a treatment with 0.1% Triton X-100, 0.1% glutaraldehyde in cytoskeleton-stabiliz-ing buffer for 2 min, followed by an incubation with 8% paraformaldehyde in cytoskeleton-stabilizing buffer for 8 min. For the double immunolabeling, the cells were incubated first with the monoclonal antipaxillin antibody (diluted in 5% FCS in PBS-glycine) for 45 min and then with the rabbit anti-mouse IgG (Dako A/S) for 30 min, followed by a protein A-gold complex (size 10 nm; Janssen Life Sciences Products) for 30 min. The cells were then incubated with protein A (Pharmacia) in PBS (0.1 mg/ml) for 10 min. This was followed by an incubation with Affi-K7 (diluted in 5% FCS in PBS-glycine) for 45 min and then with a protein A conjugated to gold particles of 15 nm in diameter (a kind gift from Dr. Varpu Marjomä ki, University of Jyvä skylä , Jyvä skylä , Finland) for 30 min. After immunolabeling, the cells were first postfixed with 2.5% glutaraldehyde in phosphate buffer for 10 min and then negative-stained by, first, washing briefly in distilled water and then immersing in a drop of 1% aqueous uranyl acetate for 1 min. For a control staining, rabbit preimmune serum was substituted for Affi-K7 antibodies. The cells were examined in a Philips 410 LS transmission electron microscope.
Phosphoamino Acid Analysis-Phosphoamino acid analysis of attaching and well spread cells on both plain culture dishes and on culture dishes precoated with plasma fibronectin (10 g/ml) both in the presence and absence of FCS was carried out as follows: Subconfluent CEHFs were trypsinized and washed with phosphate-free buffer. Phosphoproteins were labeled by plating the cells and incubating in Eagle's essential phosphate-free medium (Flow Laboratories) with 4% phosphate-free FCS and [ 32 P]orthophosphate (370 MBq/ml; Amersham) at 1.5 mCi/ml for 6 h at 37°C. The cells were harvested and lysed by scraping in radioimmune precipitation buffer. Thereafter, FAP52 was immunoprecipitated by adding 100 l of Affi-K7 to 5 ml of the clarified lysate and incubating at 4°C for 6 h. Seventy l of a 50% slurry of anti-rabbit IgG-Sepharose (Sigma) was added to the mixture that was then incubated at 4°C for 10 h on a rocking platform. Immunoprecipitates on the Sepharose-beads were washed in radioimmune precipitation buffer. The proteins in the immunoprecipitates were resolved in 10% SDS-PAGE and transferred to polyvinylidene difluoride membrane (Immobilon, Millipore) as described by Boyle et al. (24). The phosphoproteins on the membrane were visualized by autoradiography. The band corresponding to FAP52 was cut out and subjected to acid hydrolysis (6 N HCl for 1 h at 110°C), whereafter the phosphoamino acids were separated in two dimensions on thin-layer cellulose plates as described (24).

Isolation and
Sequencing of cDNA Clones-The RT-PCR cloning strategy and the clones obtained are shown in Fig. 1A. Sequencing of the amplification products revealed several novel SH3-encoding cDNAs, as judged by the comparison of the deduced amino acid sequences against the data banks. One of them was selected for further cloning (striped boxes). Oligonucleotide primers based on this sequence and chicken brain cDNA libraries were used to extend the sequence to 5Ј-and 3Ј-directions. Two 3Ј-extending clones, which were colinear and which both presented a putative stop codon (arrows to the right in Fig. 1A), were obtained. One clone extended 1100 nucleotides upstream from the priming site but lacked a possible start site of translation (an arrow to the left in Fig. 1A). By using RACE and RT-PCR with thermostable reverse transcriptase, several colinear clones extending upstream and containing a putative start site for translation were obtained. One of them, a 600-bp cDNA obtained by RACE, was selected for further characterization (a thick bar in Fig. 1A).
The nucleotide and the predicted amino acid sequence of the longest open reading frame are shown in Fig. 1B The predicted amino acid sequence was analyzed for a similarity with other known sequences. An alignment with three sequences that showed the highest degree of similarity is shown in Fig. 1C. H74, a sequence deduced from a mouse brain cDNA for a protein with an unknown function, shows a 70% identity with the FAP52 sequence. EM13 and EG13 are proteins isolated from Echinococcus multilocularis and Echinococcus granulosus, respectively, on the basis of their immunogenicity (25). Their sequences show approximately 50% identity with FAP52. The highest similarities between the proteins are present in the C-terminal SH3 domains.
The EM13 and EG13 sequences in the data base differ from FAP52 sequence in having truncated N-and C termini. However, a closer scrutiny of the EG13 nucleotide sequence discloses an upstream open reading frame of 21 amino acids (amino acids 16 -36) that closely corresponds to the N-terminal sequence of FAP52. In the C termini of both EM13 and EG13, a frameshift of one nucleotide, reveals a stretch of 6 amino acids that corresponds to the C-terminal part of the SH3 domain of FAP52 and H74 and recovers the last SH3 ␤-strand (see below). On the basis of this comparison, we suggest that FAP52, H74, EM13, and EG13 are members of the same family of proteins.
Secondary structure prediction by using the multiple alignment and the network server PHD proposes a very high propensity for ␣-helicity in the N-terminal two-thirds of the proteins. There are six long ␣-helices, each about 40 amino acids long, and several shorter regions with a high degree of ␣-helicity (Fig. 1C). These are flanked by short linker regions with a predicted coil structure. The confidence scores above 9 indicate a Ͼ90% reliability of the prediction. The ␣-helices appear to be too long to fold with a globular structure. The regular arrangement of the hydrophobic amino acids with a heptad periodicity suggests a coiled-coil arrangement. However, in the heptads, not only the positions a and d but also other positions tend to be conserved, suggesting that the helices may form higher order oligomeric structures rather than simple coiled-coil dimers.
Following the ␣-helical N-terminal portion, there is a region of about 130 amino acids that shows the lowest degree of sequence similarity between the four proteins and no clear secondary structure. It probably represents a nonglobular linker region between the N-terminal ␣-helical domain and the C-terminal SH3 domain. In summary, the analysis of the sequence suggests a three-domain structure with an ␣-helical, rod-like N-terminal domain and a C-terminal SH3 domain interspersed by a nonglobular linker region.
Expression of FAP52 in Chicken Tissues-Northern blot analysis revealed two transcripts of sizes 3.7 and 7.2 kb. They were present in all the tissues studied: gizzard, liver, cardiac muscle, skeletel muscle, brain, lung, intestine, kidney, skin, and eye (Fig. 2, lanes b-k) and also in CEHF cells (Fig. 2, lane  l). The major transcript in all the tissues was of 3.7 kb.
Immunolocalization of FAP52-Affinity-purified antibodies to FAP52 (Affi-K7) were used to localize FAP52 in cultured CEHFs, both by using immunofluorescence and immunoelectron microscopy. The specificity of the antibodies was demonstrated by immunoblotting of the total lysates of CEHFs and various chicken tissues and by immunoprecipitation of FAP52 from total CEHF lysate by using Affi-K7 (Fig. 3). In blotting, only a single band of about 63 kDa was seen in CEHFs and in all the tissues tested (Fig. 3, lanes a-e). A band of a similar molecular mass along with a faster migrating band of 55 kDa corresponding to the heavy chain of immunoglobulins was also seen in immunoblot of a Affi-K7 immunoprecipitate of CEHFs (Fig. 3, lane g). No corresponding bands were detected in a immunoprecipitate of CEHF lysate (Fig. 3, lane f) and in immunoblotting of tissues (data not shown) when preimmune serum was substituted for Affi-K7, attesting to a specific recognition of a band of 63 kDa by the antibodies and a lack of cross-reacting species in the tissues and cells tested. Notably, in SDS-PAGE, FAP52 migrates considerably slower than is expected on the basis of its molecular mass of 51,971 Da.
In immunofluorescence microscopy of permeabilized cells with Affi-K7, a distinct decoration of elongated, plaque-like structures were seen (Fig. 4, a-c and g). They were located at the ventral surface and predominantly along the perimeter of the cell, as judged by a differential focusing. No such staining was seen in unpermeabilized cells, indicating that the immu- The major open reading frame begins at the position 224, which corresponds to the putative ATG start codon. Termination codon TGA is indicated by an asterisk at the position 1570. C, an alignment and a sequence comparison of FAP52 and related proteins and a secondary structure prediction of FAP52. The alignment was generated with the program Clustal W and the secondary structure prediction with PHD. The sequences are colored according to the following scheme: all glycines (G, orange) and prolines (P, yellow) are colored. Other coloring is by the recurring feature: hydrophobic residues are blue; tyrosine is light blue; asparagines, glutamines, serines, and threonines (N, Q, S, and T) are green; aspartate and glutamate (D and E) are purple; arginine and lysine (R and K) and fully conserved cysteine (C) are red; more than 50% (40% for a positive charge) occurrence of a property results in coloring. Columns that are left white show a poor conservation of a residue or a property. The single-letter code is used for amino acids. Sequences with a high propensity for ␣-helix are marked by H, and sequences with a propensity for ␤-sheet by E. Asterisks indicate stop codons. ∧ in EG13 and EM13 sequence marks the sites of a frameshift. The accession numbers of the sequences in the EBI data base are as follows: Em_EM13, M96565; Eg_EG13, M96564, and M-H74, X85124, C_FAP52, Z50798. noreactivity is associated with the cytoplasmic leaflet of the plasma membrane (not shown). In double immunostaining, an extensive overlap (arrows in Fig. 4) was seen between the staining patterns obtained with the antibodies to paxillin (Fig.  4d), to vinculin (Fig. 4e), or to talin (Fig. 4f), the principal components of the focal adhesions, and the staining pattern obtained with Affi-K7 (Fig. 4, a-c). The specificity of the staining reaction with Affi-K7 was verified by preabsorbing Affi-K7 with a lysate of the bacteria expressing or not expressing the corresponding immunogen; no distinct staining pattern was seen in the cells incubated with the antibodies preabsorbed with the lysate of bacteria producing the corresponding immunogen (Fig. 4i). On the other hand, no blocking of the staining reaction was seen in cells incubated with the antibodies preabsorbed with the lysate of uninduced bacteria (Fig. 4h).
The extensive co-occurrence of FAP52 with the principal focal adhesion protein paxillin could also be shown in immunoelectron microscopy in CEHFs (Fig. 5A, an arrow); in double labeling (Fig. 5B), paxillin (small gold particles, arrows) was found in clusters with FAP52 (large gold particles, arrowheads). These clusters were found along plaque-like or filament bundle structures that morphologically resemble focal adhesion sites. No such decoration was seen in control stainings with the preimmune serum (data not shown). Collectively, these results indicate that FAP52 is closely associated with the focal adhesions, the specific plasma membrane domains that are responsible for the anchorage of the cells to their growth substratum (26).
Phosphoamino Acid Analysis-The phosphorylation state of FAP52 in growing cultures of CEHFs was studied by immunoprecipitating FAP52 with Affi-K7 from a lysate of 32 P-labeled cells (Fig. 6A) and then subjecting it to a phosphoamino acid analysis (Fig. 6B). In cultures grown on fibronectin-coated dishes in the presence of the label for 6 h, FAP52 is present in a phosphorylated state as shown by immunoprecipitation and autoradiography in which a distinctly labeled band of 63 kDa could be seen (Fig. 6A, lane b). No band was seen in the control precipitation with the preimmune serum (Fig. 6A, lane c). Similar results were obtained on cells grown on nonprecoated dishes (data not shown). Two-dimensional thin-layer electrophoresis showed that FAP52 is exclusively phosphorylated on serine (s arrow) residues with no signs of phosphorylation on threonine (t arrow) or tyrosine (y arrow) (Fig. 6B). Neither could we show any phosphorylation to tyrosine by using immunoprecipitation with Affi-K7 followed by immunoblotting with anti-phosphotyrosine antibodies (data not shown).

DISCUSSION
In this study, our aim was to identify novel SH3 domaincontaining proteins. Our strategy was based on the fact that, despite their overall low degree of similarity, there are fairly well conserved regions in the N and C termini of the various SH3 domains (6). Primers for RT-PCR cloning were designed on the basis of the most highly conserved sequences, which in the N terminus encompass the residues 1-10 and in the C terminus the residues 35-40 (for a justification of the domain boundaries and of the numbering, see Ref. 6). These sites contribute to a smooth patch on the domain surface, which has been shown to correspond to the ligand binding site of the SH3 domain (27). It is interesting, however, that the amino acid residues in the ligand-binding site are not completely conserved, suggesting that SH3 domains recognize a family of related domains or proteins in different species and tissues. Thus, a strategy of using degenerate primers with a high degree of degeneracy can be expected to allow for identification of novel sequences with distinct ligand binding properties. This is an important consideration since, due to generally low binding affinities to their ligands (27)(28)(29), identification and purification of SH3 domain-containing proteins by using conventional protein affinity purification techniques has met with little success.
In this study, one of the clones obtained encodes for a 52-kDa protein with a SH3 domain in its C terminus. Antibodies were raised against a polypeptide that corresponds to the N-terminal and middle portions of the protein. In immunofluorescence and immunoelectron microscopy, the protein was found to be localized in focal adhesions as judged by its colocalization with the focal adhesion proteins vinculin, talin, and paxillin. Thus, the protein appears to be a novel SH3-containing protein that is associated with the focal adhesions. Hence, we propose it the name FAP52 for Focal Adhesion Protein of 52 kDa.
The 2947-bp composite cDNA, obtained in this study, contains a 1347-bp coding region flanked by an 223-bp 5Ј-noncoding region and an 1377-bp 3Ј-noncoding region. Northern blot analysis revealed a major mRNA species of 3.7 kb. Therefore, 0.7 kb of noncoding region remains unaccounted for. We believe that it includes additional 3Ј-noncoding sequences, since we have not been able to identify a polyadenylation signal in the 3Ј-noncoding sequence. We also saw a mRNA of 7.2 kb in the Northern blots. Whether it represents a larger isoform or a cross-hybridizing species remains to be studied.
SH3 domains are found especially in proteins involved in signal transduction pathways and in cytoskeletal proteins (for recent reviews, see Refs. 4 and 30). They are distinct protein modules that can be, e.g. crystallized independently of the rest of the protein (10). X-ray crystallography and NMR spectroscopy has shown that the domain encompasses a relatively flat surface, which forms the ligand-binding site flanked by two loops (27). The N and C termini of the domain are in close apposition, which makes the ligand-binding site bulge out from the surface of the rest of the protein.
In sequence comparison, the SH3 domain of FAP52 shows a close conformity with the consensus sequence completed from more than 70 distinct SH3 sequences (6). Unlike most other SH3 domains, which have a well conserved double tryptophan in the C-terminal part of the domain, however, SH3 domain of FAP52 has only one tryptophan followed by cysteine. In this respect, FAP52 shows a similarity with the SH3 domain of Abl oncoprotein. Moreover, unlike most other SH3 domain-containing proteins, FAP52 does not contain any other recognizable sequences characteristic of signal transduction cascade proteins, such as kinase, SH2, PTPase, Cdc25, or GAP domains (30). In this sense, it resembles ␣-spectrin, myosin I, ABP-1, SLA1 and BEM1, and p47 hox and p67 hox components of the neutrophil oxidase (6). Interestingly, most of these "SH3-only" domain proteins, such as the yeast proteins ABP-1, SLA-1, and BEM1, seem to have a function closely associated with the control of the cell morphology or to be involved in the organization of the actin cytoskeleton (31)(32)(33). Thus, the domain structure of FAP52 is well in harmony with its localization in focal adhesions, which represent principal plasma membrane anchorage sites of actin filaments in cultured cells (26,34). On the basis of the speculation concerning the domain structure of FAP52, it could be surmised that FAP52 serves as an "adaptor" with SH3 mediating the linkage to the plasma membrane and the helical N terminus interacting with actin or some other focal adhesion components.
Comparison of FAP52 to other sequences in the data bases showed a high degree of similarity to H74, a mouse protein of unknown function, and to EM13 and EG13, major antigenic proteins of E. multilocularis and E. granulosus, respectively. Apart from an association of EM13 with the surface membrane of E. multilocularis (25), nothing is known of the function of EM13 or EG13 that could be instructive of the function of FAP52. More recently, we noticed that, albeit a low overall similarity, FAP52 shows a striking similarity in domain structure to a Schizosaccharomyces pombe protein Cdc15 (35), a cell cycle-dependent protein that serves as a key component in the reorganization of F-actin during cytokinesis. As in FAP52, there is a coiled-coil region also in the N terminus of Cdc15, and an SH3 domain in the C terminus. Interestingly, studies with genetically defective S. pombe suggest that an interaction with a fission yeast profilin and myosin regulatory chain, both of which interact with actin, may be important for the actinassociated functions of cdc15 (35). In regard to the similarity in their domain structure and their association with actin-containing structures, we suggest that also FAP52 and Cdc15 may represent members of the same family of proteins.
Focal adhesions are specialized membrane domains at the ventral surface of cultured cells. They are best known for their role as "feet," which provide the attachment sites by which the cells anchor themselves to the growth substratum (36). At the cytoplasmic face of the plasma membrane, focal adhesions serve as attachment points of the actin filament bundles to the plasma membrane (34). Apart from this structural role, focal adhesions have recently emerged as major sites of signal transduction pathways stimulated by cell-matrix interactions (37)(38)(39)(40). Accordingly, in addition to the well characterized struc- tural components, such as ␣-actinin (41), vinculin (42), talin (43), and paxillin (34), also a growing number of proteins with enzymatic or as yet unidentified functions have been described in focal adhesions (44). These include v-src in src-transformed cells (45), protein kinase C (46), focal adhesion kinase (pp125 FAK (47,48)), FAK-related protein without a kinase domain (FRNK (49)), tenuin (50), the myristoylated, alanine-rich protein kinase C substrate of protein kinase C (MARCKS (51)), calpain II (52), zyxin (53), and cCRP (54). The role of focal adhesions as nodes in signal transduction pathways is highlighted in the src-transformed cells in which the focal adhesionassociated v-src oncoprotein pp60 v-src is responsible for the highly enhanced tyrosine phosphorylation of paxillin and tensin (37). Also talin, vinculin, and ␤ 1 -integrin subunit have been reported to be tyrosine-phosphorylated in src-transformed cells, supposedly reflecting the presence of activated Src in focal adhesions (55)(56)(57)(58). Interestingly, some of the interactions involved are dependent on the SH3 domain of pp60 src , as evidenced by the binding of a recombinant fusion protein containing the SH3 domain of pp60 src to paxillin (59).
Phosphoamino acid analysis showed that FAP52 is present in a phosphorylated state in growing cells. Only phosphorylation on serine, with no detectable phosphorylation on tyrosine or threonine, could be found in cells that were allowed to attach on either uncoated or fibronectin-precoated culture dishes for several hours. This is drastically different from the rapid phosphorylation on tyrosine of, e.g. FAK, which occurs in the early phase of the formation of focal adhesions brought about plating the cells onto fibronectin-coated dishes (48). This implies that, at least in its phosphorylated state, FAP52 plays a role in the later rather than the early phases of the emergence of focal adhesions.
In conclusion, we describe in this paper the identification, sequence analysis, and partial characterization of FAP52, a novel focal adhesion associated SH3 domain-containing protein. Based on its modular structure, SH3 domain in its C terminus and a highly ␣-helical N terminus with a predicted propensity to form coiled-coil structures, and on its phosphorylation on serine residues, we suggest that FAP52 is involved, in a modifiable manner, in the protein-protein interactions within focal adhesions. For the elucidation of its functional role, identification of its binding partner(s) is needed.