Acidic and Basic Fibroblast Growth Factors Stimulate Tyrosine Kinase Activity

We assessed the ability of acidic and basic fibroblast growth factor (FGF) to stimulate tyrosine kinase activity in intact cells. Immunoblot with polyclonal antiphosphotyrosine antibodies detected a 90-kDa phosphotyrosine-bearing protein in lysates of Swiss 3T3 cells exposed to pituitary-derived FGF, recombinant acidic FGF, or recombinant basic FGF, but not from unstimulated cells or cells exposed to epidermal growth factor or platelet-derived growth factor. Phosphotyrosine and its analogue phenyl phosphate, but not phosphoserine, phosphothreonine, or tyrosine itself, blocked recognition of the 90-kDa protein by antiphosphotyrosine antiserum. A monoclonal antiphosphotyrosine antibody also recognized the 90-kDa protein and was used to partially purify the protein by immunoaffinity chromatography. Phosphoamino acid analysis of the 90 kDa band revealed that it contained 20% phosphotyrosine, 35% phosphothreonine, and 46% phosphoserine. Tyrosine phosphorylation of the 90-kDa protein was detectable within 30 s and reached a plateau within 10 min of FGF addition. The addition of suramin, which blocks the interaction of FGF with its receptor, caused rapid disappearance of the 90 kDa band. Cell fractionation experiments were consistent with the 90-kDa protein being membrane-associated, but cross-linking studies revealed that the FGF receptor had an M. between 145 and 210 kDa in Swiss 3T3 cells, distinct from the 90-kDa major substrate for tyrosine phosphorylation. These data demonstrate that both acidic and basic FGF activate a tyrosine kinase in vivo leading to phosphorylation of a unique 90-kDa substrate, and they suggest that protein modification by phosphorylation at tyrosine is involved in eliciting the mitogenic effect of FGF.

insulin (7), and insulin-like growth factor 1 (8) all stimulate the tyrosine kinase activity of their respective receptors. Moreover, tyrosine kinase activity has been correlated strongly with the proliferative effects of a number of oncogene products (9,10). Recently, Huang and Huang (11) reported that brain-derived growth factor, an acidic FGF-like molecule, stimulated the phosphorylation of tyrosine residues in a 135-kDa protein in membrane preparations derived from Swiss 3T3 cells. We now demonstrate the stimulation of tyrosine phosphorylation by both acidic and basic FGF in intact cells and show that the major in vivo substrate of this reaction is a 90-kDa protein.
EXPERIMENTAL PROCEDURES Swiss 3T3 cells were obtained from the American Type Culture Collection and grown in Dulbecco's modified Eagle's medium with 10% fetal bovine serum. Stock cultures were passaged every 3-4 days. For experiments, cultures were used 4 days after plating in 35-mm dishes at a 1:3 split ratio.
For antiphosphotyrosine blots, cultures in 35-mm dishes were incubated in duplicate with the indicated agonists for 10 min at 37 "C. The medium was then quickly removed and the cells lysed in LB (100 mM Tris-HCI (pH 8.0), 30 mM sodium pyrophosphate, 50 mM NaF, 100 PM sodium orthovanadate, 5 mM EDTA, 5 mM [ethylenebis(oxyethyIenenitrilo)]tetraacetic acid, 1 mM PMSF, 1% SDS, and 100 mM dithiothreitol). The lysate was boiled immediately for 5 min, sonicated, then subjected to SDS-PAGE on a 7% acrylamide gel. The separated proteins were transferred electrophoretically to nitrocellulose paper in 200 mM glycine, 20 mM Tris base, and 20% methanol. After transfer, the nitrocellulose sheets were soaked for 1 h at room temperature in TBS (150 mM NaCI, 50 mM Tris-HCl (pH 7.4), 5% bovine serum albumin, and 0.05% Tween 20), incubated at 4 "C overnight in the same buffer containing antiphosphotyrosine antiserum, then incubated for 1 h at room temperature with horseradish peroxidase-conjugated goat anti-rabbit IgG. Blots were washed with three changes of TBS after each incubation. Horseradish peroxidaselabeled proteins were localized by incubation with hydrogen peroxide and horseradish peroxidase-color development reagent (Bio-Rad) containing 4-chloro-1-napthol. The antiphosphotyrosine antiserum used in these studies was a gift from J. Y. J. Wang, University of California, San Diego, and has been describedpreviously (12). Affinity purification of this antiserum over phosphotyramine-Sepharose was performed as described previously (4, 13).
Monoclonal antiphosphotyrosine antibodies were purified from ascites by adsorption to a phosphotyramine column and immunospecific elution with phenyl phosphate (4, 13). The purified antibodies were coupled to cyanogen bromide activated Sepharose at 15 mg of protein/ml packed beads (4). Affinity purification of phosphotyrosine-bearing proteins using the antiphosphotyrosine beads was performed as described previously (4,13).
To examine 32P incorporation into phosphoproteins, cultures in 35-mm dishes were labeled for 3 h at 37 "C with 1 mCi of o r t h~[~* P ] phosphate/700 pl of phosphate-free Dulbecco's modified Eagle's medium per dish, stimulated for 10 min with agonist, then lysed in LB with 1% Triton X-100 and 0.2% Nonidet P-40 and without SDS or dithiothreitol. Three replicate dishes were pooled for each point. The pooled lysates were adsorbed to antiphosphotyrosine beads which were washed and immunospecifically eluted with 40 mM phenyl phosphate as described (4,13). Eluates were analyzed by SDS-PAGE and autoradiography.
For phosphoamino acid analysis, cultures were labeled with or-th~[~*P]phosphate as described above. Lysates from six replicate dishes were immunoaffinity-purified using antiphosphotyrosine beads and subjected to SDS-PAGE and autoradiography. The region of the gel corresponding to the 90 kDa band was excised and treated as described previously. Phosphoamino acids were separated by onedimensional thin layer electrophoresis in water/acetic acid/pyridine (945:50:5) and detected by autoradiography.
Synthetic genes encoding bovine acidic or basic FGFs (14,15) were expressed in Escherichia coli using the tac-1 promoter system (16). Details of this and other expression systems for recombinant FGFs will be described elsewhere? In some experiments, bovine acidic FGF containing a deletion of lysine 9 was employed; identical results were obtained with either the wild-type or mutant forms. Recombinant growth factors were purified in a single step using heparin-Sepharose chromatography (15).
For ligand-binding studies, acidic FGF was iodinated using iodogen (Pierce Chemical Co.) to a specific activity of 3-4 pCi/pmol. Free "' I was separated from iodinated FGF by chromatography on Sephadex G-25. Analysis by SDS, 15% PAGE and autoradiography revealed a single band at 16 kDa. Confluent monolayers of Swiss 3T3 cells in 35-mm wells were incubated with 1251-acidic FGF (50-150 PM) at 4 "C in binding buffer (Dulbecco's minimal Eagle's medium containing 20 mM Hepes (pH 7.4) and 0.05% bovine serum albumin (Pentex)). After 4 h, monolayers were washed with cold binding buffer (3 X 2 ml, 10 min/wash), then solubilized and 1251-quantitated by y counter. Nonspecific binding determined in the presence of 15 nM acidic FGF was always less than 25% of total binding. For "'I-FGF cross-linking studies, two protocols were employed. In one, Swiss 3T3 cells were incubated with '=I-FGF a t 4 "C and washed as described above, except that the wash buffer did not contain bovine serum albumin. Cells were then exposed to 0.1 mM disuccinimidyl suberate in 150 mM NaCI, 10 mM Hepes (pH 7.35), 1 m1/35-mm well, for 16 min a t 20 "C. The reaction was quenched by the addition of 0.1 volume of 200 mM glycine, 10 mM Tris-HC1,2 mM EDTA (pH 7.3). After 1 min, the wells were drained and lysis buffer added. Lysates were analyzed by SDS-PAGE and autoradiography. In a second protocol designed to minimize processing time and mimic the conditions used for detecting the 90-kDa protein by immunoblot, cells were incubated with T -F G F (0.36 nM) for 10 min at 37 "C, then washed as quickly as possible, and exposed to cross-linker for 5 min at 37 "C. The reaction was quenched and lysates processed as described above.

RESULTS
Polyclonul Antiphosphotyrosine Antiserum Recognizes a 90-kDa Protein in FGF-stimulated Cultures-To detect phosphorylation of proteins at tyrosine in vivo, lysates of confluent Swiss 3T3 cultures were analyzed by immunoblot using antiphosphotyrosine antiserum. In previous studies, this antiserum was used to detect protein tyrosine phosphorylation stimulated by platelet-derived growth factor, epidermal growth factor, and insulin (12,17).3 In each of these cases, the major substrate recognized by antiphosphotyrosine antiserum was the respective receptor that was autophosphorylated when activated by the binding of ligand. Antiphosphotyrosine antiserum recognized a 90 kDa band in lysates from cells exposed to highly purified recombinant acidic FGF (Fig.  1, lane 1 ), but not in lysates from unstimulated cultures (Fig.   1, lane 2). To enhance specificity, the antiphosphotyrosine antiserum was affinity-purified by adsorption to phosphotyramine coupled to Sepharose beads followed by immunospecific elution with the phosphotyrosine analogue, phenyl phosphate. Analysis of lysates by immunoblot with affinity-purified antiphosphotyrosine antibody revealed a 90 kDa band in samples from FGF-stimulated cultures (Fig. 1, lane 3), but not in samples from unstimulated cultures (Fig. 1, lane 4   The specificity of the antiserum was confirmed by demonstrating that phosphotyrosine and its analogue phenyl phosphate, but not phosphoserine, phosphothreonine, or tyrosine itself, blocked recognition of the 90 kDa FGF-stimulated band, as well as recognition of the 180 kDa PDGF-receptor band (data not shown). The latter is known to be recognized on the basis of autophosphorylation of tyrosine (3,4,12).
Monoclonal Antiphosphotyrosine Antibodies Recognize a 90-kDa Protein in FGF-stimulated Cultures-To confirm that FGF stimulates tyrosine phosphorylation in vivo, an independent method of identifying tyrosine-phosphorylated proteins was taken. In these experiments, monoclonal antiphosphotyrosine antibody was immobilized on agarose beads (APT beads (4,13)) and used as an affinity matrix to isolate phosphotyrosine-bearing proteins from lysates of ~r t h o [~~P ] p h o sphate-labeled cultures. As shown in Fig. 2, a 32P-labeled 90-kDa phosphotyrosine-bearing protein was detected in cultures stimulated by FGF (lane 2), but not in unstimulated cultures (lane 1 ).
Phosphoamino Acid Analysis Demonstrates Phosphotyro- phosphate for 3 h at 37 "C, then incubated 10 additional min with diluent ( l a n e I ) or FGF ( l a n e 2). Lysates were immunopurified using monoclonal antiphosphotyrosine antibody coupled to Sepharose beads as described under "Experimental Procedures." The phenyl phosphate eluates were subjected to polyacrylamide gel electrophoresis and autoradiography. The migration of molecular weight standards is indicated at the right margin of this autoradiogram. sine in the 90-kDa Protein-The presence of phosphotyrosine in the 90-kDa protein was confirmed by phosphoamino acid analysis. Swiss 3T3 cells were labeled with o r t h~[~~P ] p h o sphate, then stimulated for 10 min at 37 "C with FGF. The 90-kDa protein was purified partially by immunoaffinity chromatography and SDS-PAGE as described above. The 90 kDa band was excised from the gel and its phosphoamino acid composition was determined as described under "Experimental Procedures." As shown in Fig. 3, 20% of "P incorporated into phosphoamino acids in the 90-kDa protein was in phosphotyrosine.

Fibroblast Growth Factor Activates Tyrosine Kinase in Vivo
Thus both polyclonal and monoclonal antiphosphotyrosine antibodies recognized a 90-kDa protein in FGF-stimulated cultures using Western blot analysis and immunoaffinity chromatography, respectively. Recognition of the 90-kDa protein by these antibodies was blocked by phosphotyrosine and its analogues, but not by other phosphoamino acids or by tyrosine itself. FGF stimulated the incorporation of "P into a 90-kDa protein that could be purified by affinity chromatography using immobilized monoclonal antiphosphotyrosine antibody; phosphoamino acid analysis of this protein confirmed the presence of phosphotyrosine. Taken together, these data show that FGF activates a tyrosine kinase in intact cells. recombinant basic FGF and partially purified pituitary-derived FGF to stimulate tyrosine phosphorylation in whole cells (Fig. 4). All forms of FGF tested elicited tyrosine phosphorylation of the 90-kDa protein. The fact that FGF produced in a bacterial expression system and FGF purified from pituitary both elicited tyrosine phosphorylation makes it ex- tremely likely that this effect is due to FGF itself and not to a minor contaminant of either preparation, as these preparations are unlikely to share the same contaminants. The high potency of FGF in eliciting 90-kDa protein phosphorylation was consistent with a receptor-mediated event (Fig. 5). The observation that both acidic and basic FGF stimulated the same intracellular event, phosphorylation of the 90-kDa protein on tyrosine residues, is consistent with binding and crosslinking data suggesting that both forms of FGF share a common cell surface receptor (20).

Fibroblast Growth Factor Activates Tyrosine Kinase in
Maintenance of Levels of the 90-kDa Phosphoprotein Requires Continued Occupancy of the FGF Receptor-Phosphorylation of the 90-kDa protein on tyrosine was detectable within 30 s and reached a plateau within 10 min of FGF addition. Levels of this phosphoprotein remained at this plateau for at least 4 h in the continued presence of FGF (data not shown), consistent with a balance between rapid phosphorylation and dephosphorylation or degradation, or with persistence of a stable phosphorylated protein. In order to distinguish between these two possibilities, the polyanion suramin was utilized to dissociate FGF from its receptor (Table I and Fig. 6), thereby turning off FGF-induced tyrosine phosphorylation so that the decay of phosphorylated substrates could be followed. Swiss 3T3 cells were incubated for 10 min at 37 "C in the presence of FGF in order to achieve  11) or EGF (100 ng/ml, lane 12), then lysed and processed for antiphosphotyrosine immunoblot as described under "Experimental Procedures." The migration of molecular weight standards is indicated at the righthand margin (see legend to Fig. 1).

Suramin displaces bound FGF from its receptor
Whole cell ligand binding studies with FGF were performed as described under "Experimental Procedures." Confluent Swiss 3T3 cultures were incubated with '*'I-FGF for 5 h at 4 "C in the absence ("total binding") or presence ("Nonspecific binding") of excess unlabeled FGF. The incubation was performed at 4 'C to prevent internalization of ligand. Cells were then washed and immediately solubilized for counting ("Control") or further incubated 20 min at 37 "C in the presence ("Plus suramin") or absence ("Minus suramin") of 1 mM suramin, then solubilized for counting. The values shown are the mean counts/min/well, n = 3. Standard deviations were less than 10% of their respective means. phosphorylation of the 90-kDa protein, then washed, and further incubated in medium containing either FGF or 1 mM suramin. Cells were then lysed and processed for antiphosphotyrosine immunoblot (Fig. 6). Note that the 90-kDa phosphoprotein is undetectable within 10 min of FGF removal by suramin treatment ( l a n e 6). This effect of suramin was reversible in that removal of suramin and restimulation with FGF again caused 90-kDa phosphorylation (data not shown). These data suggest that that activity of FGF-stimulated tyrosine kinase is dependent upon the continuous occupancy of the FGF receptor, and that the level of 90-kDa protein observed in these studies represents a balance between rapid phosphorylation and dephosphorylation or proteolysis.  1 and 2 ) or washed and further incubated at 37 "C in the presence of 10 ng/ml FGF (FGF, lanes 3-5) or 1 mM suramin  (Suramin, lanes 6-8) before lysis. The duration of the second incubation was 10 (lanes 3 and 6 ) , 30 (lanes 4 and 7), or 60 min (lanes 5 and 8). Lysates were analyzed by antiphosphotyrosine immunoblot as described under "Experimental Procedures." The 90 kDa band is indicated by the arrow at the right-hand margin.
receptor and a major substrate of the kinase is the receptor itself. To determine the relationship of the 90-kDa phosphoprotein ( Figs. 1 and 2) to the FGF receptor, lz5I-FGF was cross-linked to its receptor under the conditions used to detect the 90-kDa phosphoprotein by immunoblot. In preliminary studies, saturable and specific binding of 1251-a~idi~ FGF to intact Swiss 3T3 cells was demonstrated. Scatchard analysis revealed one species of binding sites with a K d of approximately 290 PM and a receptor density of about 20,000 sites/ cell (data not shown). Cross-linking of Iz5I-FGF to its receptor in intact cells at 4 "C using the homobifunctional crosslinking agent disuccinimidyl suberate revealed a high affinity binding site with M, of between 145 and 210 kDa (Fig. 7). Unlabeled acidic FGF displaced half of the label at a concentration of approximately 0.36 nM. Several minor bands of lower M, including bands at approximately 92,85,41, and 37 kDa were noted in cross-linking studies on overexposed autoradiograms, but more than 95% of the cross-linked label migrated between 145 and 210 kDa. These results are similar to those previously reported for this cell type (19). Crosslinking carried out under conditions identical to those employed to stimulate tyrosine phosphorylation of the 90-kDa substrate again revealed the M, of the major cross-linked species to be 145-210 kDa (Fig. 7). Thus, the apparent size of the FGF receptor in Swiss 3T3 cells is considerably higher than that of the 90-kDa major substrate of FGF-stimulated tyrosine kinase. It should be noted that neither the apparent size of the FGF receptor by cross-linking nor the size of the 90-kDa kinase substrate changed when gels were run under nonreducing conditions, suggesting against the possibility that the FGF receptor contains a disulfide-linked dimer of the 90-kDa substrate. Other substrates for the FGF-stimulated tyrosine kinase were noted. A faint band at 140 kDa (perhaps representing receptor (11)) and a 40 kDa band ( Fig.  1) were sometimes seen on antiphosphotyrosine blots of FGFstimulated cultures. The relative intensity of these signals was always less than one-tenth that of the 90 kDa band. Thus over 90% of the substrate for FGF-stimulated tyrosine kinase migrates at 90 kDa under the same conditions that labeled FGF cross-linked to its receptor migrates at 145-210 kDa.
Temperature Dependence of FGF-induced Tyrosine Kinase Actiuity-The temperature dependence of FGF-induced tyrosine kinase is also distinct from that of PDGF and EGF. Swiss 3T3 cultures were incubated for 10 min at 37 "C or for up to 16 h at 4 "C in the presence or absence of PDGF or FGF, then processed for antiphosphotyrosine immunoblot. As reported previously, PDGF induced phosphorylation of its 180-kDa receptor at tyrosine at both 37 and 4 "C (17); EGF behaved similarly. By contrast, FGF elicited tyrosine phosphorylation of the 90-kDa protein only at 37 "C (data not shown).
The cross-linking and temperature dependence data suggest, but do not prove, that the 90-kDa protein is not a subunit or fragment of the FGF receptor but rather a distinct molecule. Definitive identification of these molecules awaits their purification.
Subcellular Fractionation Experiments Are Consistent with a Membrane Association of the 90-kDa Substrate-In order to localize the 90-kDa substrate to membrane or cytosol, subcellular fractionation was performed. Swiss 3T3 cultures were incubated for 10 min at 37 "C in the presence of absence of FGF or PDGF. Cells were harvested by scraping in LB containing 1 mM dithiothreitol and no detergent, disrupted by sonication, and immediately centrifuged at 4 "C for 20 min at 400,000 X g. Proteins in the supernatant were recovered as described (21); pellet (nuclear and membrane) and superna-tant (cytosol) fractions were analyzed by antiphosphotyrosine blot. The 90-kDa phosphoprotein was detected only in pellets derived from FGF-stimulated cultures, and the 180-kDa autophosphorylated PDGF receptor used as a control in these experiments was detected only in pellets derived from PDGFstimulated cultures. Neither the 90-nor 180-kDa proteins were detected in supernatants. The short half-life of the 90-kDa phosphoprotein in broken cell preparations precluded more detailed cell fractionation studies. These data suggest that the 90-kDa protein is associated with cell membranes.

DISCUSSION
These data show that both acidic and basic FGF stimulate tyrosine kinase activity in intact cells. It is likely that FGFstimulated tyrosine phosphorylation plays a role in transducing the mitogenic response to FGF given the association of tyrosine kinase activity with growth factor action and oncogene products. A possible role for the 90-kDa substrate of FGF-stimulated tyrosine kinase in signal transduction is suggested by the rapid phosphorylation of this protein upon FGF stimulation and its rapid disappearance upon FGF removal. The 90-kDa phosphotyrosine-bearing protein was not detected in cultures stimulated by PDGF, EGF, insulin, or bombesin ( Fig. 5 and data not shown). Thus stimulation of 90-kDa protein tyrosine phosphorylation may provide a specific assay for FGF-like activity.
The ability of FGF to activate a tyrosine kinase in vivo makes it like other mesenchymal cell growth factors such as EGF, PDGF, insulin, and insulin-like growth factor 1 (3-8) in this regard. However, the tyrosine kinase activity stimulated by EGF, PDGF, insulin, and insulin-like growth factor 1 is intrinsic to their respective receptors, and a major substrate of the activity in each case is the receptor itself (3-8, 18). The apparent molecular weight of the FGF receptor by cross-linking studies was between 145 and 210 kDa; yet under the same conditions, the molecular weight of the major substrate for FGF-stimulated tyrosine kinase was 90 kDa. Moreover, in contrast to the PDGF, EGF, and insulin receptors, the 90-kDa protein is not phosphorylated at 4 "C. It is thus possible that the 90-kDa protein is a specific substrate of FGF-stimulated tyrosine kinase rather than a fragment or subunit of the FGF receptor. Indeed, precedent exists for ligand-stimulated tyrosine phosphorylation of a polypeptide noncovalently associated with the murine T cell antigen receptor (22). Unambiguous definition ofthe relationship of 90-kDa FGF-stimulated tyrosine kinase substrate to the FGF receptor awaits the purification of these molecules.