Fibroblast Growth Factor Receptor-1-mediated Endothelial Cell Proliferation Is Dependent on the Src Homology (SH) 2/SH3 Domain-containing Adaptor Protein Crk*

Stimulation of fibroblast growth factor receptor-1 (FGFR-1) expressed on endothelial cells leads to cellular migration and proliferation. We have examined the role of the Src homology (SH) 2/SH3 domain-containing adaptor protein Crk in these processes. Transient tyrosine phosphorylation of Crk in fibroblast growth factor-2-stimulated endothelial cells was dependent on the juxtamembrane tyrosine residue 463 in FGFR-1, and a Crk SH2 domain precipitated FGFR-1 via phosphorylated Tyr-463, indicating direct complex formation between Crk and FGFR-1. Furthermore, Crk SH2 and SH3 domains formed ligand-independent complexes with Shc, C3G, and the Crk-associated substrate (Cas). Tyrosine phosphorylation of C3G and Cas increased as a consequence of growth factor treatment. We examined the role of Crk in FGFR-1-mediated cellular responses by use of cells expressing chimeric platelet-derived growth factor receptor-α/FGFR-1 (αR/FR) wild type and mutant Y463F receptors. The kinase activity of αR/FR Y463F was intact, but both Crk and the adaptor FRS-2 were no longer tyrosine-phosphorylated in the mutant cells. Both wild type and mutant receptor cells migrated efficiently, whereas cells expressing the mutant αR/FR Y463F failed to proliferate and Erk2 and Jun kinase activities were suppressed in these cells. In wild type αR/FR cells transiently expressing an SH2 domain mutant of Crk, Erk and Jun kinase activities as well as DNA synthesis were attenuated. Our data indicate that Crk participates in signaling complexes downstream of FGFR-1, which propagate mitogenic signals.

Stimulation of fibroblast growth factor receptor-1 (FGFR-1) expressed on endothelial cells leads to cellular migration and proliferation. We have examined the role of the Src homology (SH) 2/SH3 domain-containing adaptor protein Crk in these processes. Transient tyrosine phosphorylation of Crk in fibroblast growth factor-2-stimulated endothelial cells was dependent on the juxtamembrane tyrosine residue 463 in FGFR-1, and a Crk SH2 domain precipitated FGFR-1 via phosphorylated Tyr-463, indicating direct complex formation between Crk and FGFR-1. Furthermore, Crk SH2 and SH3 domains formed ligand-independent complexes with Shc, C3G, and the Crk-associated substrate (Cas). Tyrosine phosphorylation of C3G and Cas increased as a consequence of growth factor treatment. We examined the role of Crk in FGFR-1-mediated cellular responses by use of cells expressing chimeric platelet-derived growth factor receptor-␣/FGFR-1 (␣R/FR) wild type and mutant Y463F receptors. The kinase activity of ␣R/FR Y463F was intact, but both Crk and the adaptor FRS-2 were no longer tyrosine-phosphorylated in the mutant cells. Both wild type and mutant receptor cells migrated efficiently, whereas cells expressing the mutant ␣R/FR Y463F failed to proliferate and Erk2 and Jun kinase activities were suppressed in these cells. In wild type ␣R/FR cells transiently expressing an SH2 domain mutant of Crk, Erk and Jun kinase activities as well as DNA synthesis were attenuated. Our data indicate that Crk participates in signaling complexes downstream of FGFR-1, which propagate mitogenic signals.
Fibroblast growth factors (FGFs) 1 constitute a growing family of heparin-binding polypeptides, presently including 18 members (1)(2)(3)(4)(5). Their structural relatedness ranges from about 50% identity between the prototype members FGF-1 (acidic FGF) and FGF-2 (basic FGF) to about 20% between other members of the family. FGFs are mitogenic for a wide variety of cells in tissue culture and have been implicated in regulation of differentiation, cell motility, and transformation. FGFs have also been shown to be essential in normal physiological processes in vivo; these include embryonic and fetal development, neovascularization, and wound healing (6 -8).
FGFs induce their biological responses by binding to high affinity FGF receptors, which constitute a family of four (FGFR-1 to FGFR-4) structurally related transmembrane tyrosine kinases (9). The receptors contain two or three extracellular immunoglobulin-like loops, a characteristic stretch of acidic amino acids between the first and second loop, a single transmembrane region, and an intracellular kinase domain split by a 14 -17-amino acid-long kinase insert. Alternative splicing generates a multitude of structural variants that differ in ectodomain regions known to be involved in ligand binding (10). Binding of FGF together with heparin or heparan sulfates to the receptor induces receptor dimerization, leading to kinase activation and autophosphorylation of the receptor. Autophosphorylated tyrosine residues and adjacent amino acids provide specific binding sites for intracellular signal transduction proteins containing Src homology (SH) 2 domains. SH2 domaincontaining proteins are either enzymes or adaptor proteins, which may couple to enzymatic activities. Upon binding of SH2 domain proteins to the receptor, intrinsic or associated enzymatic activities transduce signals further in signaling cascades, eventually giving rise to a cellular response. The SH2 domain proteins are often equipped with other structurally conserved motifs such as the SH3 motif, which mediates constitutive binding to proline-rich stretches (11,12).
FGFR-1 contains at least seven autophosphorylation sites (13); thus far, only two of these sites, Tyr-653 and Tyr-766, have been shown to be important for receptor function. Tyr-653 is located in the kinase domain and appears to be involved in regulation of kinase activity. Tyr-766 is located in the C-terminal tail, and phosphorylation at this site allows binding of phospholipase C␥ (PLC␥) (14,15). Activation of PLC␥ appears not to be required for FGF-induced proliferation, at least in stable cell lines (15,16), and FGFR-1-mediated migration is independent of PLC␥ (17). Other SH2 domain proteins of the adaptor type, such as Shc and the FGF receptor substrate 2 (FRS-2) become tyrosine-phosphorylated via FGFR-1, without stable complex-formation with the receptor (14,18).
In this work we have studied the interaction between FGFR-1 and Crk II. v-Crk was originally identified as an oncoprotein of a chicken retrovirus, CT10 (19). Subsequently, the corresponding cellular proto-oncogene was isolated (20,21) and shown to exist in two splice variants, Crk I (28 kDa) and Crk II (42 kDa). Crk II consists of an SH2 domain followed by two SH3 domains, while Crk I lacks the second SH3 domain. In addition, another closely related gene, Crk L, has been identified (22). Crk is tyrosine-phosphorylated in platelet-derived growth factor (PDGF)-stimulated cells, but without apparent consequences for PDGF-induced cellular responses (23). Recent data indicate a role for Crk in nerve growth factor-stimulated neuronal cells (24). We show that tyrosine phosphorylation of Crk in FGF-2-stimulated cells is dependent on the juxtamembrane tyrosine residue 463 in FGFR-1 and that Crk is critical for FGFR-1-induced cell proliferation.

EXPERIMENTAL PROCEDURES
FGFR cDNA Constructions-cDNAs for FGFR-1 (25), PDGFR-␣ (26), and PDGFR-␤ (27) were subcloned into the pAlter vector TM (Promega Corp.), and site-directed mutagenesis was performed using the Altered Sites in vitro mutagenesis system (Promega Corp.). A schematic outline of the different receptor constructs used in this study is shown in Fig. 1. The chimeric receptor PDGFR-␣/FGFR-1 (denoted ␣R/FR) was constructed by cleaving the FGFR-1 and PDGFR-␣ cDNAs with HindIII and SalI followed by ligation of the fragment corresponding to the extracellular part of PDGFR-␣ to that corresponding to the intracellular part of FGFR-1 (17). Using the mutagenesis system described, point mutations that changed Tyr-766 or Tyr-463 to phenylalanine residues, or created stop codon including cleavage site for XbaI at position 2323-2329 in the intracellular part from FGFR-1, were introduced into the cDNA for ␣R/FR. The wild type and mutated cDNAs were inserted into the eukaryotic expression vector pcDNA3 (Invitrogen). We also used chimeric receptors in which the juxtamembrane domain or the kinase insert domain in FGFR-1 were replaced with the corresponding parts from the PDGFR-␤, to create FR-1/PR␤JM or FR-1/PR␤Ki. The construction of FR-1/PR␤Ki has been described before (28). FR-1/PR␤JM was constructed by point mutations creating cleavage site for HindIII and NruI at positions 1195-1200 and 1425-1430 of the FGFR-1 and at positions 1861-1866 and 1974 -1979 of the PDGFR-␤, which were introduced into the respective insert with oligonucleotides prepared using an Amersham Pharmacia Biotech Gene Assemble Plus synthesizer. All mutations and constructs were confirmed by nucleotide sequencing. FR-1/PR␤JM was then constructed by cleaving the FGFR-1 and PDGFR-␤ cDNAs with HindIII and NruI followed by ligation of the fragment corresponding to the juxtamembrane domain of PDGFR-␤ into the position of the FGFR-1 endogenous HindIII-NruI fragment. The wild type and the mutated cDNAs were inserted into the eukaryotic expression vector pcDNA1/neo (Invitrogen). All mutations were confirmed by nucleotide sequencing.
Antisera-Polyclonal antibodies against Crk II, C3G, and SH-PTP2 were purchased from Santa Cruz Biotechnology, Inc. A mouse monoclonal antibody specific for phosphotyrosine (4G10) was from Upstate Biotechnology, and an antibody against Shc was purchased from Transduction Laboratories. Anti-HA antibody was purchased from Roche Molecular Biochemicals, and phosphospecific MAPK antibody was from New England Biolabs, Inc. The rabbit antiserum against phospholipase C␥ and the rabbit antiserum against Erk-2 were kind gifts from Dr. Lars Rönnstrand, Ludwig Institute for Cancer Research, Uppsala, Sweden. The rabbit antiserum against FGFR-1 has been described before (28), and a rabbit antiserum specifically reacting with FRS-2 was raised against a peptide corresponding to the C-terminal part of FRS-2.
Transient Transfection-PAE cells expressing the wild type chimeric receptor ␣R/FR were cultured in Ham's F-12 medium supplemented with 10% FCS to 30% confluence. Transfections were done by using SuperFect (Qiagen). For Erk 2 kinase assay, the cells were cultured in T-25 flasks and transfected with 2 g each of cDNAs encoding HA-Erk and wild type Crk or the Crk SH2 domain mutant in the pCAGGS vector. For Jun kinase assay, the cells were cultured in T-25 flasks and transfected with 2 g each of cDNAs encoding HA-Jun kinase in the pSR␣ vector and wild type Crk or the Crk SH2 domain mutant. For analysis of labeling index, cells were seeded out on glass placed in 60-mm dishes and transfected with wild type Crk or Crk SH2 domain mutant cDNA using the amount of cDNA needed to get the same amount of Crk expressed in all cells. In all experiments, transfection with only the vector was used as a control. The original Crk II cDNA was kindly provided by Dr. Michiyuki Matsuda (Department of Pathology, National Institute of Infectious Diseases, Tokyo, Japan), the Crk cDNA expressing Crk II SH2 domain mutant was a kind gift from Dr. Kristiina Vuori (Burnham Institute, La Jolla, CA), HA-Erk 2 was from Dr. Ivan Dikic (Ludwig Institute for Cancer Research, Uppsala, Sweden), and HA-Jun kinase was provided by Dr. Pä r Gerwins (Department of Medical Biochemistry and Microbiology, Uppsala University, Uppsala, Sweden).
Immunoprecipitation and Immunoblotting-Cells in 75-cm 2 flasks were serum starved over night in Ham's F-12 supplemented with 1% FCS, followed by stimulation with PDGF-BB (100 ng ml Ϫ1 ) or FGF-2 (100 ng ml Ϫ1 ) for 7 min or for different time periods, as indicated, at 37°C. The monolayers were rinsed with ice-cold phosphate-buffered saline (PBS) containing 100 M Na 3 VO 4 and lysed for 10 min on ice in 1 ml of Nonidet P-40 lysis buffer (20 mM Hepes, pH 7.5, 150 mM NaCl, 1% Nonidet P-40, 10% glycerol, 300 M Na 3 VO 4 , 1% aprotinin, 1 mM phenylmethylsulfonyl fluoride, 1 mM dithiothreitol (DTT)). Lysates were clarified at 10.000 ϫ g for 15 min at 4°C, and the supernatants were incubated with antibodies for 1 h at 4°C, followed by a final incubation for 45 min with immobilized protein A (Immunosorb; EC Diagnostics, Uppsala, Sweden). The precipitates were washed three times in Nonidet P-40 lysis buffer and twice in PBS containing 100 M Na 3 VO 4 . Sample buffer (0.2 M Tris-HCl, pH 8.8, 0.5 M sucrose, 5 mM EDTA, 4% sodium dodecyl sulfate, 0.01% bromphenol blue, and 2% ␤-mercaptoethanol) was added, and the samples were boiled for 4 min at 95°C before SDS-polyacrylamide gel electrophoresis in 10% gels. For immunoblotting, proteins were electrophoretically transferred onto nitrocellulose membranes (Hybond-C extra, Amersham Pharmacia Biotech). The membranes were blocked in 0.2% Tween 20 in PBS containing 5% bovine serum albumin (BSA). Primary antibody was diluted in PBS containing 0.05% Tween 20 and 3% BSA and incubated with membranes for 1 h, followed by washing in PBS. Appropriate secondary antibody was diluted as above and incubated with the membranes for another 1 h. After careful washing in PBS, immunoreactive proteins were visualized by a chemiluminescence detection system based on a protocol described earlier (29). Before reprobing the filters, they were stripped in 62.5 mM Tris-HCl, pH 6.7, 2% SDS, and 100 mM ␤-mercaptoethanol at 50°C for 30 min.
In Vitro Association of GST Fusion Proteins-The SH2 domain of Grb2 was expressed as a part of a GST fusion protein (a kind gift from Dr. J Schlessinger, New York University Medical Center, New York, NY), and used as described earlier (14). The SH2 domain of Crk II and the SH2-SH3 domains of CrkII were also expressed as GST fusion proteins and were kindly provided by Dr. A. Sorokin (Dept. of Medicine and Cardiovascular Research Center, Medical College of Wisconsin, Milwaukee, WI). For association experiments, transfected PAE cells were cultured in 75-cm 2 flasks and serum starved over night in Ham's F-12 supplemented with 1% FCS, followed by treatment or not with growth factors for 7 min at 37°C, and then lysed in Nonidet P-40 lysis buffer. Clarified lysates were incubated with purified immobilized GST fusion protein (Grb2 SH2, Crk II SH2 or Crk II SH2 SH3) on glutathione-Sepharose 4B (Amersham Pharmacia Biotech) end-over-end for 2 h at 4°C. Samples were washed three times in Nonidet P-40 lysis buffer and twice in PBS containing 100 M Na 3 VO 4 and analyzed by SDSpolyacrylamide gel electrophoresis and immunoblotting, as described above.
Peptide Synthesis-The following synthetic peptide, phosphorylated at its tyrosine residue (indicated as pY), was used in this study: pY463, GVSEpYELPEDPRWELPR-COOH. The corresponding nonphosphorylated peptide was also synthesized and used as a control. Peptides were synthesized using Fmoc (N-(9-fluorenyl)methoxycarbonyl) chemistry, as described by Mori et al. (30).
Immobilization of Peptides on Agarose Beads-Immobilization of synthetic peptides on agarose beads (Affi-Gel-15, Bio-Rad) was performed according to the company's instructions. Briefly, 2 mg of the peptide was dissolved in 50 mM Hepes, pH 7.2, and mixed with 1 ml 1:1 slurry of prewashed Affi-Gel-15 agarose, and incubated end-over-end for 1 h at room temperature. To block the remaining active esters on the agarose, the beads were incubated in 100 mM ethanolamine HCl, pH 8, for 1 h at room temperature, followed by washing in Tris/HCl-buffered saline, pH 7.4, containing 1 mM DTT.
Protein Interaction Experiments Using Affi-Gel-immobilized Peptides-Cells cultured in 75-cm 2 flasks were washed in ice-cold PBS containing 100 M Na 3 VO 4 and lysed in 1 ml of ice-cold RIPA buffer (20 mM Tris-HCl, pH 7.4, 150 mM NaCl, 1% Nonidet P-40, 0.5% sodium deoxycholate, 0.1% SDS, 5 mM EDTA, 1 mM phenylmethylsulfonyl fluoride, 1% aprotinin, and 200 M Na 3 VO 4 ). Cell lysates were clarified at 10,000 ϫ g for 10 min at 4°C, and the samples were then incubated with phosphorylated or nonphosphorylated peptides immobilized on Affi-Gel-15, in the presence or absence of free competing peptide, endover-end for 1 h at 4°C. The agarose beads were washed twice in RIPA buffer, three times in RIPA buffer supplemented with 500 mM NaCl (high salt RIPA), and once with RIPA buffer. Bound proteins were eluted by boiling in sample buffer and analyzed by SDS-polyacrylamide gel electrophoresis and immunoblotting, as described above.
Chemotaxis Assay-The assay was performed in a modified Boyden chamber as described earlier (31) using micropore nitrocellulose filters (8 m thick, 8 m pore) coated with type-1 collagen solution at 100 g ml Ϫ1 (Vitrogen 100, Collagen Corp.). Cells were trypsinized and resuspended at a concentration of 5.5 ϫ 10 5 cells ml Ϫ1 in serum-free Ham's F-12 medium containing 0.25% BSA. The cell suspension was placed in the upper chamber, and the serum-free medium containing 0.25% BSA and 10 ng ml Ϫ1 or 50 ng ml Ϫ1 of PDGF-BB, was placed below the filter in the lower chamber. As a positive control, medium containing 10% FCS was added to the lower chamber. After 6 h at 37°C, the medium was removed and the cells sticking to the filter were fixed in pure methanol and stained with Giemsa stain. The cells that had migrated through the filter were counted. All samples were analyzed in triplicate at four separate occasions.
Cell Proliferation Assay-Cells in Ham's F-12 supplemented with 10% FCS were seeded (2 ϫ 10 4 cells/well) into 24-well dishes. After 2 h the medium was changed to starvation medium (Ham's F-12 containing 0.2% FCS) and the incubation continued for an additional 24 h. The medium was changed again at day 2 and day 4 (starvation medium), and at the same time PDGF-BB or FGF-2 at different concentrations (0, 1, 10, 20, and 100 ng ml Ϫ1 ) were added. As a control, cells were cultured in Ham's F-12 medium supplemented with 10% FCS. Cell numbers were scored after 5 days. All experiments were performed in triplicate for every concentration of PDGF-BB, and at least two independent cell clones for the chimeric wild type and the mutated (Y463F) receptors were analyzed.
Erk-2 Kinase Assay-After treatment of cells with 100 ng ml Ϫ1 PDGF-BB for 7 min at 37°C, cells were rinsed once with ice-cold PBS containing 100 M Na 3 VO 4 and lysed in lysis buffer (20 mM Hepes, pH 8.0, 1% Triton X-100, 0.5% deoxycholic acid, 10 mM EGTA, 5 mM MgCl 2 , 20 g/ml leupeptin, 1% aprotinin, 1 mM phenylmethylsulfonyl fluoride, 20 mM Na 4 P 2 O 7 , 50 mM NaF, 100 M Na 3 VO 4 , and 1 mM DTT). Clarified supernatants were incubated with Erk-2 antiserum, raised against a C-terminal MAP 2 kinase peptide (EETARFQPGYRS), end-over-end for 1.5 h at 4°C. Immobilized protein A (Immunosorb) was added, and the samples were mixed at 4°C for 30 min. The immune complexes were washed three times in lysis buffer and twice in kinase buffer (20 mM Hepes, pH 8.0, 20 mM MgCl 2 , 2 mM MnCl 2 , 1 mM DTT) and then incubated for 15 min at 30°C in 40 l of kinase buffer containing 10 g of myelin basic protein (MBP, Sigma) and 5 Ci of [␥-32 P]ATP (Amersham Pharmacia Biotech). The kinase reaction was terminated by addition of 40 l of sample buffer and boiling for 4 min. Samples were analyzed by SDS-PAGE in a 15% SDS-polyacrylamide gel. After fixation in methanol/acetic acid, the gel was dried and analyzed by autoradiography.
Erk 2 kinase activity, in transiently transfected cells expressing HA-Erk 2, wild type Crk, or Crk SH2 domain mutant, was also measured by immunoprecipitation and immunoblotting as described above. Erk 2 was immunoprecipitated by using HA antibodies, and immunoblotting was performed using phosphospecific MAPK antibody (New England Biolabs, Inc.).
JNK Assay-A solid phase assay was used, where c-Jun-(1-79) was expressed as a part of a GST fusion protein and coupled to glutathione-Sepharose 4B. The experiment was performed as described by Gerwins et al. (32). Briefly, chimeric PAE cells were stimulated with 100 ng ml Ϫ1 PDGF-BB or left untreated for 7 min at 37°C and then lysed in ice-cold Nonidet P-40 lysis buffer. Clarified lysates were incubated with immobilized GST-c-Jun-(1-79)-Sepharose 4B end-over-end 1 h at 4°C. The samples were washed twice in Nonidet P-40 lysis buffer and once in kinase buffer (20 mM Hepes, pH 7.5, 0.05% Triton X-100, 2 mM MnCl2, 10 mM MgCl 2 , and 1 mM DTT). The beads were resuspended in kinase buffer supplemented with 10 Ci of [␥-32 P]ATP (Amersham Pharmacia Biotech) per sample. The reactions were performed at 30°C for 20 min and were terminated by addition of Laemmli sample buffer. The samples were boiled for 4 min, and phosphorylated proteins were analyzed on SDS-PAGE and visualized by autoradiography.
Jun kinase activity, in transiently transfected cells expressing HA-Jun kinase, wild type Crk, or Crk SH2 domain mutant, was measured by immunoprecipitation and a kinase reaction as above. Jun kinase was immunoprecipitated by using HA antibodies, and GST Jun-(1-79) was added as a substrate in the kinase reaction.
Labeling Index-PAE cells expressing wild type ␣R/FR were cultured on coverslips and transiently transfected with wild type Crk and Crk SH2 domain mutant as described above. The cells were then starved overnight in Ham's F-12 supplemented with 0.25% BSA and labeled with 1 Ci/ml [ 3 H]thymidine for the last 2 h during culture. The cells were washed, fixed in paraformaldehyde, and covered with autoradiography emulsion (Eastman Kodak Corp.). After 1 week of exposure, the film was developed and unlabeled cells were stained with Mayer hematoxylin (Histolab Products AB, Gothenburg, Sweden). The cells were counted, and the results show percentage of labeled nuclei for three different experiments.

Tyr(P)-463 in the Juxtamembrane Region of FGFR-1 Is Involved in Mitogenic
Signaling-Chimeric PDGFR-␣/FGFR-1 (␣R/FR) wild type and mutant proteins were ectopically expressed in PAE cells, in order to study FGFR-1 signal transduction without interference of endogenous FGF receptors. The PAE cells express low levels of endogenous FGF receptors, but lack expression of PDGF receptors (28,33). We have previously shown that PAE cells expressing the ␣R/FR wild type protein migrate and proliferate efficiently in response to PDGF-BB (17). Using this model, we addressed the role of the FGFR-1 juxtamembrane tyrosine phosphorylation site Tyr-463 in signal transduction. The abilities of PAE cells expressing the wild type ␣R/FR and two independent clones of PAE cells expressing the mutant ␣R/FR Y463F to proliferate in response to growth factor (Fig. 2) were analyzed. The number of wild type ␣R/FR cells increased dose-dependently from 100% (control, serumstarved cells) to 250% for cells treated with 20 or 100 ng/ml PDGF-BB. A similar response was seen for PAE cells expressing the mutant ␣R/FR Y766F, in which the PLC␥ binding site is removed. PAE cells expressing intact FGFR-1 also increased in number to 250% of control in response to FGF-2 treatment. In contrast, untransfected PAE cells and two different clones of the ␣R/FR Y463F cells failed to increase in cell number at any of the concentrations of PDGF-BB used. All cell types responded similarly to treatment with 10% FCS (data not shown). Thus, loss of Tyr-463 interfered with the capacity of the FGFR-1 intracellular domain to mediate signals for proliferation.
The abilities of PAE cells expressing the wild type ␣R/FR and cells expressing the mutant ␣R/FR Y463F to migrate in a mini-Boyden chamber were examined. Cells were seeded on one side of a collagen-coated 8-mm-thick nitrocellulose filter, and the growth factor was suspended in serum-free medium on the other side of the filter. The number of cells that migrated to the other side of the filter during a 6-h incubation was measured. Fig. 3 shows that cells expressing the ␣R/FR wild type and Y463F mutant migrated with similar efficiencies in this assay, allowing the conclusion that phosphorylation at Tyr-463 is not required for FGFR-1-mediated migration. This is in agreement with our previous data showing that FGFR-1-mediated migration is dependent on a 15-amino acid residue stretch in the C-terminal tail of the receptor (17).
Tyr(P)-463 Is Required for Tyrosine Phosphorylation of FRS-2 and Crk-The adaptor protein FRS-2 has been reported to interact with the FGFR-1 juxtamembrane domain in a phosphotyrosine-independent manner, via a phosphotyrosine binding-like domain in FRS-2. We tested whether the loss of proliferative capacity of PAE cells expressing the ␣R/FR Y463F mutant to PDGF-BB could be due to decreased FRS-2 tyrosine phosphorylation, and thereby reduced Grb2 binding. Fig. 4A shows that the extent of FRS-2 tyrosine phosphorylation indeed was reduced in cells expressing ␣R/FR Y463F. Fig. 4B shows that this was not due to a general impairment of FGFR-1 kinase function since PDGF-BB-induced kinase activity of wild type ␣R/FR and mutant ␣R/FR Y463F were similar; furthermore, tyrosine phosphorylation of PLC␥, which is known to bind to Tyr(P)-766 in the FGFR-1 C-terminal tail, was induced to the same extent by activation of wild type and mutant chimeric ␣R/FR (data not shown). In addition, the migration capacity of the Y463F mutant was intact (Fig. 3). Fig. 4C shows that immunoprecipitation of FRS-2 did not allow detection of co-precipitation of FGFR-1, which is in agreement with previous reports (14). In contrast, immunoprecipitation with Crk II antiserum brought down a tyrosine-phosphorylated 150-kDa component after FGF-2 stimulation, which is likely to represent FGFR-1. FGF-induced co-precipitation of FRS-2 in the Crk immunoprecipitate could not be detected (data not shown). The sensitivity of the available FGFR-1 antiserum did not allow confirmation that the 150-kDa Crk-associated molecule corresponds to FGFR-1.
The sequence surrounding Tyr-463 (Y-E-L-P) conforms with the reported sequence for binding of the Crk SH2 domain (Y(P)-D-H-P). We used Affi-Gel-immobilized unphosphorylated and phosphorylated Tyr-463-containing synthetic peptides, which were incubated with PAE cell lysates to test whether Crk could bind to this region of FGFR-1. To test for specificity, free unphosphorylated or phosphorylated Tyr-463 peptides were mixed with the lysates before incubation with the peptidecoupled Affi-Gel matrix. As seen in Fig. 5A, Crk was retained by the phosphorylated Tyr(P)-463 matrix, but not by the unphosphorylated immobilized Tyr-463 peptides. Addition of free Tyr(P)-463 peptide competed out Crk binding, whereas free unphosphorylated Tyr-463 peptide failed to affect binding of Crk to the immobilized phosphorylated Tyr(P)-463 matrix. To ensure that Crk binding was not dependent on an intermediary component, we incubated a Crk SH2 domain fusion protein (see below) with immobilized phosphorylated and unphosphorylated Tyr-463 peptides. The Crk SH2 fusion protein was retained by the phosphorylated peptide only (data not shown).
To show that Crk is tyrosine-phosphorylated as a consequence of activation of FGFR-1 in intact cells, PAE cells expressing FGFR-1 were with treated for 7 min with FGF-2 (Fig.  5B), which induced a 6-fold increase in Crk tyrosine phosphorylation. FGF-2 stimulation of primary BCE cells expressing endogenous FGFR-1 led to a 3-fold induction of Crk tyrosine phosphorylation (Fig. 5B, lanes 1 and 2). To confirm the struc-tural requirement for Crk tyrosine phosphorylation, we examined Crk phosphorylation in PAE cells expressing a series of ␣R/FR mutants (see Fig. 1 for schematic outline of mutants).

FIG. 4. Tyr(P)-463 is required for tyrosine phosphorylation of FRS-2.
A, PAE cells overexpressing FGFR-1 or the chimeric receptors ␣R/FR wt and ␣R/FR Y463F were incubated in the presence (ϩ) or absence (Ϫ) of FGF-2 (100 ng ml Ϫ1 ) or PDGF-BB (100 ng ml Ϫ1 ) for 7 min at 37°C, lysed, and clarified. The samples were immunoprecipitated (Ip) with antibodies against FRS-2, subjected to SDS-PAGE, transferred to a nitrocellulose filter, and immunoblotted (Ib) with phosphotyrosine antibodies (4G10). B, cells expressing ␣R/FR wt and ␣R/FR Y463F were analyzed for induction of kinase activity using an in vitro kinase assay. Cells were treated (ϩ) or left untreated (Ϫ) with PDGF-BB (100 ng ml Ϫ1 ) for 7 min at 37°C, lysed and immunoprecipitated with antibodies against FGFR-1, and subjected to an in vitro kinase assay. Samples were analyzed by SDS-PAGE and autoradiography. C, cells expressing the wild type FGFR-1 were stimulated (ϩ) or not (Ϫ) with FGF-2 and processed by immunoprecipitation (Ip) with antibodies against FRS-2, FGFR-1, or Crk II. Samples were analyzed by immunoblotting (Ib) using anti-phosphotyrosine antibodies. Arrow indicates migration rate of FGFR-1.

FIG. 5. Tyrosine phosphorylation of Crk is dependent on tyrosine residue 463 in FGFR-1.
A. PAE cell lysates were incubated with nonphosphorylated or phosphorylated peptides immobilized on Affi-Gel-15, in the presence or absence of free competing peptide. The proteins that bound to the beads were eluted by sample buffer, subjected to SDS-PAGE, transferred to a nitrocellulose filter, and immunoblotted with Crk II antibodies. B, primary BCE cells and PAE cells expressing wild type FGFR-1, FR-1/PR-␤JM, and FR-1/PR-␤Ki were incubated in the presence (ϩ) or absence (Ϫ) of FGF-2 (100 ng ml Ϫ1 ) for 7 min at 37°C, lysed, and immunoprecipitated (Ip) with Crk II antibodies. Samples were subjected to SDS-PAGE, transferred to nitrocellulose filter, and immunoblotted (Ib) with phosphotyrosine antibodies (4G10). C, PAE cells expressing wild type FGFR-1 were incubated in the absence (Ϫ) or presence (ϩ) of FGF-2 (100 ng ml Ϫ1 ) at 37°C for different time periods as indicated, lysed, and immunoprecipitated (Ip) with Crk II antibodies. Samples were separated by SDS-PAGE and immunoblotted (Ib) with phosphotyrosine antibodies (upper panel) and Crk II antibodies (lower panel). D, untransfected PAE cells and PAE cells expressing the chimeric receptors ␣R/FR wt, ␣R/FR Y463F, and ␣R/FR Y766F were incubated in the presence (ϩ) or absence (Ϫ) of PDGF-BB (100 ng ml Ϫ1 ) for 7 min at 37°C, lysed, and immunoprecipitated (Ip) with Crk II antibodies. Samples were separated by SDS-PAGE, transferred to nitrocellulose filter, and immunoblotted (Ib) with phosphotyrosine antibodies (4G10). Cells were used that expressed FGFR-1 variants, in which the endogenous juxtamembrane domain (FR-1/PR-␤JM), or the kinase insert (FR-1/PR-␤Ki) has been replaced with the corresponding domains from the PDGFR-␤. As seen in Fig. 5B, cells expressing FR-1/PR-␤Ki still responded to FGF-2 treatment whereas in cells expressing FR-1/PR-␤JM, FGF-2 stimulation failed to induce an increase in Crk phosphorylation. Fig. 5C  (upper panel) shows that the kinetics of Crk tyrosine phosphorylation in FGF-2-stimulated PAE cells was very rapid, with a marked increase occurring already after 1 min of stimulation, indicating that Crk phosphorylation was mediated directly by the FGFR-1. After 30 min of stimulation, the level of Crk tyrosine phosphorylation was back to basal. During this time period, the levels of Crk protein remained unchanged (Fig. 5C,  lower panel).
To ensure that endogenous FGF receptors expressed in the PAE cells did not interfere in our analyses, we turned to PAE cells expressing the chimeric ␣R/FR and mutants of this construct. Fig. 5D shows that cells expressing the ␣R/FR wild type protein mediated FGFR-1-dependent Crk tyrosine phosphorylation in response to PDGF-BB stimulation. In PDGF-BB-stimulated cells expressing ␣R/FR Y463F, no detectable increase in Crk phosphorylation was observed. In contrast, in cells expressing the ␣R/FR Y766F mutant, which lacks the binding site for PLC␥, PDGF-BB-induced Crk tyrosine phosphorylation was similar to that in cells expressing the wild type protein.
Crk SH2 and SH3 Domains Mediate Formation of Multiprotein Complexes-Crk is known to couple to a wide spectrum of signal transduction cascades (34). To analyze potential Crk interactions in our cell model, GST fusion proteins covering Crk SH2 or Crk SH2-SH3 domains were coupled to glutathione-Sepharose 4B, and incubated with lysates of PDGF-BBstimulated and unstimulated PAE cells expressing the ␣R/FR wild type protein. A number of components were retained on the Crk SH2 and SH2-SH3 matrixes, as visualized by SDS-PAGE and immunoblotting using phosphotyrosine antibodies (Fig. 6). Immobilized Crk SH2 domain fusion protein bound the 46-kDa isoform of Shc in a ligand-independent manner. The Crk SH2-SH3 domain fusion protein retained the 54-kDa Shc isoform, as well as the nucleotide exchange factor C3G, as confirmed by blotting with specific antibodies (data not shown), both of which were tyrosine-phosphorylated at basal conditions. The Crk-associated substrate (Cas) also bound to the Crk SH2-SH3 domain. Tyrosine phosphorylation of both C3G and Cas were increased after ligand stimulation. In some experiments, the Crk SH2-SH3 domain fusion protein also retained the tyrosine phosphatase SHP-2, which was phosphorylated at increased levels after growth factor treatment. Binding of the nucleotide exchange factor Sos was not detected, in accordance with previous studies where such interactions have been difficult to identify. The adaptor FRS-2 was also not detected in these analyses (data not shown). Fig. 6C shows that the Crk SH2 domain pulled down a 150-kDa component from PAE cells expressing FGFR-1 or the ␣R/FR chimeric receptor after appropriate growth factor treatment. This molecule was not precipitated by the Crk SH2 domain fusion protein from cells expressing the ␣R/FR Y463F cells, indicating that the 150-kDa component indeed corresponds to the receptor. Direct confirmation by immunoblotting with FGFR-1 antiserum was not possible using available reagents. GST alone failed to bring down tyrosine-phosphorylated material (Fig. 6C).
Loss of Erk and Jun Kinase Activation in the Y463F Mutant Cells-Previous reports have demonstrated SH3-domain-dependent association between Crk and nucleotide exchange factors such as Sos (coupling to the MAP kinase cascade) and C3G (coupling to the Jun kinase cascade) (35)(36)(37)(38), the latter of which was associated with the Crk SH2-SH3 domain fusion protein and tyrosine-phosphorylated in response to activation of the ␣R/FR in our cell model (see above). We examined the effect of the Y463F mutation in FGFR-1 on activation of the MAPK kinase Erk2 and of the Jun kinase in PAE cells expressing wild type ␣R/FR or mutant ␣R/FR Y463F. Erk2 was immunoprecipitated and incubated in the presence of [␥-32 P]ATP and MBP. As seen in Fig. 7A, the level of MBP phosphorylation as FIG. 6. Crk interacts with other proteins and exists in multiprotein complexes. PAE cells expressing wild type chimeric receptor were treated with (ϩ) or without (Ϫ) growth factor for 7 min at 37°C, lysed, and clarified. Lysates were incubated with purified immobilized GST fusion protein (Crk SH2 or Crk SH2-SH3) on gluthathione-Sepharose-4B end-over-end for 2 h at 4°C. Samples were then analyzed by SDS-PAGE, transferred to nitrocellulose filter, and immunoblotted (Ib) with phosphotyrosine antibodies (4G10). To identify the different phosphoproteins, immunoblotting was performed with antibodies against Shc, SHP-2, C3G, and Cas (data not shown). Panel A shows a long and panel B a short exposure of the same blot. Panel C shows pull-down using immobilized GST alone or the Crk SH2 domain, incubated with cell lysates from unstimulated (Ϫ) or growth factor-stimulated (ϩ) cells expressing wild type FGFR-1, ␣R/FR, or ␣R/FR Y463F proteins. Samples were analyzed by immunoblotting (Ib). Arrow indicates migration rate of receptors. a result of Erk2 activation increased 3-fold in cells expressing the wild type receptor; in contrast, Erk2 activity was not induced in cells expressing the Y463F mutant. Jun kinase activity was measured by complex-formation of Jun kinase with immobilized c-Jun fusion protein. Phosphorylation of c-Jun was measured by incubation of the beads with [␥-32 P]ATP. As seen in Fig. 7B, stimulation of the wild type receptor cells led to an increase in Jun kinase activity, which was attenuated in the ␣R/FR Y463F cells. The overall higher level of Jun kinase activity in the wild type chimeric cells could be due to a basal stimulating activity, which was lost by the Y463F mutation in the mutant chimeric cells.
Transient Overexpression of Wild Type and Mutant Crk Interferes with FGF-2-induced increase in MAP kinase and Jun kinase activities-Since the ␣R/FR Y463F cells showed diminished tyrosine phosphorylation of FRS-2 as well as Crk, we wished to further define the role of Crk in regulation of MAP and Jun kinase activation. We therefore analyzed Erk2 and Jun kinase activities in ␣R/FR cells transiently overexpressing wild type Crk and a Crk SH2 domain mutant, in combination with HA-tagged Erk2 or HA-tagged Jun kinase. Overexpression of wild type Crk decreased the level of activated Erk2, analyzed by immunoprecipitation using HA antibodies and immunoblotting with antibodies specifically reactive with tyrosine-phosphorylated, activated Erk2 (Fig. 8A). Expression of the Crk SH2 domain mutant protein completely suppressed stimulation of Erk2 activity. The basal level of Jun kinase activity was markedly increased in the Crk-transfected cells. Growth factor stimulation led to an additional induction of Jun kinase activity in the wild type Crk transfected cells, but Jun kinase activity was not induced by growth factor treatment in cells transfected with the Crk SH2 domain mutant (Fig. 8B) Fig. 9 shows that overexpression of the wild type Crk abrogated growth factor-stimulated induction of labeling index, despite that Erk activation was only partially suppressed in these cells (Fig. 8A). In cells overexpressing the Crk SH2 domain mutant, treatment of cells with the growth factor also failed to stimulate DNA synthesis, in agreement with the reduced Erk kinase activity shown in Fig.  8A). These results indicate that overexpression of wild type Crk, as well the SH2 domain mutant of Crk, leads to a dominant-negative situation. In conclusion, these data show that overexpression of wild type Crk or a Crk SH2 domain mutant interfered with FGFR-1-induced mitogenesis. DISCUSSION We show in this paper that Crk is tyrosine-phosphorylated by FGFR-1 and that stable complex formation between the Crk SH2 domain and FGFR-1 is dependent on phosphorylated Tyr-463 in the receptor juxtamembrane domain. Thus far, FGFR-1 has been shown to associate in stable complexes only with PLC␥, whereas other signal transduction molecules such as Src, Shb, and FRS-2 are tyrosine-phosphorylated without stable complex formation with the activated dimerized receptors (39).
We used PAE cells expressing chimeric PDGFR-␣/FGFR-1 wild type and mutant proteins for these studies. PAE cells have been demonstrated to faithfully reproduce signal transduction events identified in primary cells (40). In order to avoid interference of endogenous FGF receptors in these cells, we have employed chimeric receptors and we show in this paper (Figs. 2 and 3), and have shown previously that signals for migration and proliferation are efficiently transduced via the chimeric receptor (17). In contrast, the mutant ␣R/FR Y463F fails to mediate proliferative signals. The amino acid sequence sur- teins, such as Nck (Y-D-E-P), Abl (Y-E-N-P) and SHP-2 (Y-I/V-X-P) (41). We were interested in the possibility that the adaptor molecule Nck is a substrate for FGFR-1; if this is the case, Nck may compete with Crk for interaction with the same site on FGFR-1. Nck is a widely expressed adaptor molecule, containing one SH2 and three SH3 domains (42), which causes transformation of fibroblasts and tumor formation in nude mice (43,44). Nck has been reported to participate in FGFR-1 signal transduction in mesoderm induction during Xenopus development (45). We failed to detect FGFR-1-dependent tyrosine phosphorylation of Nck in PAE cells (data not shown). Furthermore, overexpression of wild type Nck or an SH2 domain mutant of Nck still allowed increased DNA synthesis in response to growth factor treatment, although basal labeling index was increased (data not shown). Thus, our data do not show a role for Nck in FGFR-1-mediated proliferation in the PAE cells.
Interactions between Crk and two guanine nucleotide exchange factors, C3G and Sos, have been identified (46). It is well established that the Grb2/Sos complex mediates activation of Ras, which couples to a cascade of serine/threonine kinases (Raf, mitogen-activated protein kinase/extracellular signalregulated kinase kinase, and Erk1/Erk2). We show decreased Erk2 activation in cells expressing the Y463F mutant receptor, and in cells overexpressing a Crk SH2 domain mutant. The recently characterized FGFR-1 substrate FRS-2 (18) presents four binding sites for Grb2 and is therefore an important mediator of Ras activation. FRS-2 was not appreciably tyrosinephosphorylated in cells expressing mutated Y463F receptors, and we infer that the reduction in Erk2 activation and proliferative capacity of the ␣R/FR Y463F cells is due to both loss of Crk and FRS-2 signal transduction. FRS-2 has been shown to associate with the unstimulated FGFR-1 via its juxtamembrane domain in a phosphotyrosine-independent manner, using the yeast two-hybrid screen. Our data indicate that removal of Tyr-463 in FGFR-1 leads to a loss of FRS-2 adaptor function.
We have previously reported that the FGFR-1 Y463F mutant expressed in L6 myoblasts mediates intact activation of Raf (14), and Mohammadi et al. (13) have shown that an FGFR-1 mutant lacking four tyrosine phosphorylation sites, including Tyr-463, mediates increased incorporation of [ 3 H]thymidine similar to the wild type FGFR-1. However, PAE cells expressing the ␣R/FR Y463F mutant receptor failed to proliferate in response to growth factor. By employing chimeric receptors, we have ensured that exogenous or endogenous FGF-2 sources do not affect the outcome of the experiment. Most cells in tissue culture express FGF receptors and produce FGF. Furthermore, transient overexpression of wild type Crk, as well as the Crk SH2 domain mutant to similar protein levels (data not shown), obliterated growth factor-induced DNA synthesis in the ␣R/FR PAE cells (Fig. 9). The effect of the Crk SH2 domain mutant is likely to depend on saturation of downstream signal transduction components, thereby inhibiting endogenous normal wild type Crk function. The effect of overexpression of wild type Crk could in part be due to displacement of FRS-2, or to abortion of downstream signaling by saturating binding to signal transduction proteins, such as Sos and C3G. Dependent on the relative expression levels of receptors and signal transduction molecules, overexpression of wild type versions of signal transduction molecules may suppress downstream signal transduction as described previously (47).
The nucleotide exchange factor C3G is structurally related to Sos within the catalytic domain and Sos as well as C3G contains multiple proline-rich domains which interact with the Crk SH3 domain (36). Whereas Sos regulates Ras activity, C3G has been reported to activate Rap1/smgp21/Krev-1 (48), a Rasrelated GTPase, which counteracts the effects of Ras in transformation (49). Rap1 appears to transduce signals that regulate the kinetics of Erk 1/2 activation, possibly in a cell typeand stimulus-dependent manner (50,51).The Crk/C3G complex has also been shown to activate Jun kinase (37), by a Ras-independent mechanism (38). The Jun kinase is classically activated by stress stimulation, such as UV irradiation, hyperosmolarity and inflammatory cytokines (52) via a pathway involving the recently identified MEKK1-4 (32, 53) and the downstream MKK4 and MKK7 (54,55). This pathway has been shown to transduce signals for apoptosis (52) although Jun kinase appears to function also in other cellular responses and may also protect against apoptosis. Crk has been shown to promote apoptosis in Xenopus eggs, and immunodepletion of Crk inhibited apoptosis (56). We show that obstruction of Crk signal transduction downstream of the FGFR-1 attenuated activation of Jun kinase, in agreement with the report by Tanaka et al. (37). Thus, it is possible that FGFR-1 transduces both positive and negative signals and that the final read-out is dependent on the balance between these signals. The contribution of Jun kinase to FGFR-1-dependent cellular responses remain to be identified.
Crk has recently been shown to be a substrate for the PDGF ␣and ␤-receptors (23), although without apparent consequence for PDGF-induced biological responses. Tanaka and co-workers used the pheochromocytoma cell line PC12 to analyze the role of Crk in neuronal differentiation. Microinjection of Crk induced neurite formation, which was blocked by point mutation in either of the Crk SH2 or SH3 domains (57). Moreover, data recently reported by York et al. (24) indicated Rap1 in neuronal cell differentiation. One may infer from these studies that Crk is involved in a multitude of cellular responses, probably by virtue of its participation in signal transduction pathways gated via Ras and Ras-related proteins. We will focus our further studies on the role of signal transduction via Ras and Ras-like proteins, initiated by adaptors such as FRS-2 and Crk, in endothelial cell differentiation (58). 2