Identification of p90, a prominent tyrosine-phosphorylated protein in fibroblast growth factor-stimulated cells, as 80K-H.

Tyrosine phosphorylation of cellular proteins occurs rapidly upon treatment of fibroblasts with acidic or basic fibroblast growth factors (aFGF, bFGF), suggesting a role for protein phosphorylation in the FGF signaling pathway. Stimulation of Swiss 3T3 cells and MRC-5 fibroblasts with bFGF results in the tyrosine phosphorylation of several proteins, of which the most prominent has been designated as p90. The phosphorylation of p90 is observed within 30 s of treating the cells with FGF but not with other growth factors. Microsequencing of p90 resolved on two-dimensional polyacrylamide gel electrophoresis indicated an N-terminal amino acid sequence which corresponded to a protein previously named as 80K-H. Polyclonal antibodies raised against the predicted C terminus of 80K-H recognized p90 on all Western blots. p90 was found to bind specifically to GRB-2-glutathione S-transferase fusion protein and to be immunoreactive with 80K-H antibody. In addition, anti-phosphotyrosine antibodies immunoprecipitated 80K-H from cell lysates of FGF-stimulated but not from control fibroblasts. The biological function of 80K-H is yet unknown. However, from this study and a previous observation of the obligatory dependence of p90 phosphorylation on FGF receptor occupation, it appears that 80K-H is involved in FGF signaling.

Tyrosine phosphorylation of cellular proteins occurs rapidly upon treatment of fibroblasts with acidic or basic fibroblast growth factors (aFGF, bFGF), suggesting a role for protein phosphorylation in the FGF signaling pathway. Stimulation of Swiss 3T3 cells and MRC-5 fibroblasts with bFGF results in the tyrosine phosphorylation of several proteins, of which the most prominent has been designated as p90. The phosphorylation of p90 is observed within 30 s of treating the cells with FGF but not with other growth factors. Microsequencing of p90 resolved on two-dimensional polyacrylamide gel electrophoresis indicated an N-terminal amino acid sequence which corresponded to a protein previously named as 80K-H. Polyclonal antibodies raised against the predicted C terminus of 80K-H recognized p90 on all Western blots. p90 was found to bind specifically to GRB-2-glutathione S-transferase fusion protein and to be immunoreactive with 80K-H antibody. In addition, anti-phosphotyrosine antibodies immunoprecipitated 80K-H from cell lysates of FGF-stimulated but not from control fibroblasts. The biological function of 80K-H is yet unknown. However, from this study and a previous observation of the obligatory dependence of p90 phosphorylation on FGF receptor occupation, it appears that 80K-H is involved in FGF signaling.
Fibroblast growth factors (FGFs) 1 comprise a family of structurally related heparin-binding polypeptides of which the best characterized members are acidic FGF (aFGF) and basic FGF (bFGF). These polypeptides are important regulators of differentiation and embryogenesis; they support the survival of neuronal cells and are extremely potent inducers of DNA synthesis in ectoderm-or mesoderm-derived cell types, including endothelial cells and smooth muscle cells (1).
FGFs elicit cellular responses by binding to and activating high affinity FGF receptor tyrosine kinases (2,3). In agreement with the general scheme for receptor tyrosine kinases, ligand binding induces dimerization of FGF receptors, followed by activation of the intrinsic kinase activity and autophosphorylation of the receptor molecules (2). Two autophosphorylation sites in FGFR-1 have been identified. One, Tyr-653 (4), is located in the kinase domain and could have a regulatory role. The other, Tyr-766, has been shown to mediate the direct binding of phospholipase C-␥ (5).
The signal transduction pathways following activation of FGF receptors are thought to be primarily mediated by the tyrosine phosphorylation of key substrates, as is the case with other receptor tyrosine kinases. In addition to phospholipase C-␥, several other substrates have been identified, including SHC (6), ERK-2, ERK-1 (7,8), cortactin (9,10), and Src (9,11). A number of groups have demonstrated the prominent phosphorylation of a 90-kDa protein by both aFGF and bFGF (7,(12)(13)(14)(15)(16). It was recently demonstrated that p90 phosphorylation was induced in cells transfected with both the FGFR-1 and the keratinocyte growth factor receptor, but not the FGFR-4 (11). The identification of p90 would thus be relevant to our understanding of the complexity of FGF signaling.
Recent investigations have shown that receptor tyrosine kinases link to downstream signaling components, like ERKs, via adaptor proteins that have no apparent enzymatic function but trigger off signaling cascades by assembling proteins into reactive complexes. Protein-protein interactions occur via binding of modular domains on the adaptor protein, such as Src homology 2 (SH2), Src homology 3 (SH3), pleckstrin homology, and phosphotyrosine-binding domains, to specific sequence motifs on the target proteins (17,18). One such adaptor protein, GRB-2, essentially consists of two SH3 domains flanking a solitary SH2 domain. GRB-2 has been shown to bind to autophosphorylated receptors such as those of EGF, PDGF, CSF-1, and insulin, as well as to tyrosine-phosphorylated, non-receptor proteins such as the insulin receptor substrate (IRS). Recently, GRB-2-GST fusion proteins were also used to identify novel tyrosine-phosphorylated proteins that bind to GRB-2 (19 -22).
We now report the identification of p90 by a combination of several techniques, which include partial amino acid sequencing of the protein, Western blotting with different antibodies, and GRB-2 binding studies. The collective evidence points to the identity of p90 as 80K-H, a protein whose function has not been characterized. Our studies suggest the involvement of 80K-H in FGF signaling.
Cells and Cell Stimulation-Human MRC-5 fibroblasts (American Tissue Culture Collection, Rockville, MD) were grown as described previously (23). One hour before stimulation with the various agonists, the cells were washed twice with serum-free Eagle's minimal essential medium and then incubated with the same medium until the cells were lysed. Swiss 3T3 cells (American Tissue Culture Collection) were maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum (HyClone Laboratories, Logan, UT), 2 mM glutamine, 10 mM HEPES (pH 7.3), and 100 units/ml penicillin/streptomycin. When the cells were 80 -90% confluent, the medium was aspirated, the cells washed and maintained for another 24 h in serum-free Dulbecco's modified Eagle's medium. Various agonists were added to the quiescent cells prior to the medium being aspirated, the cells rapidly washed in cold PBS, and lysed for subsequent analysis.
Subcellular Fractionation-After stimulation, cells in 15-cm dishes were washed with Buffer A (25 mM Tris-HCl, pH 7.5, 250 mM sucrose, 2.5 mM magnesium acetate, 2 mM dithiothreitol, 10 mM benzamidine, 10 mM sodium fluoride). The plates were well drained of Buffer A and replaced with 1 ml of homogenization buffer (Buffer A containing 5 mM EGTA, 5 mM EDTA plus protease inhibitors 250 g/ml leupeptin, 150 g/ml aprotinin, and 1 mM phenylmethylsulfonyl fluoride). The cells were scraped and homogenized with 30 strokes of a Dounce homogenizer with a tight-fitting pestle. The homogenate was centrifuged for 10 min at 90,000 rpm in a Beckman TLA-100 rotor at 4°C. The supernatant (cytosolic extract) was added with 500 l of 5 ϫ Laemmli buffer and boiled for 10 min. The pellet (membrane) was extracted for 20 min on ice with 1 ml of Buffer B (25 mM Tris-HCl, pH 7.5, 5 mM dithiothreitol, 10 mM benzamidine, 10 mM sodium orthovanadate, 5 mM EDTA, and 1% Nonidet P-40 plus other protease inhibitors used in Buffer A). The detergent-soluble proteins were collected by centrifugation at room temperature at 12,000 rpm. The supernatant containing the detergentsoluble fraction (membrane extract) was added with 500 l of 5 ϫ Laemmli buffer and boiled as above.
Resolution of the p90 Protein on Two-dimensional PAGE-Two 15-cm plates of MRC-5 cells were prepared as above for stimulation with bFGF for 10 min. After washing with cold PBS, each plate of cells was lysed in 500 l of the lysis buffer used by Coughlin et al. (13). 200 l of 5 ϫ Laemmli buffer was added to each aliquot, and after boiling for 10 min the entire lysate was loaded onto each side of a one-dimensional PAGE minigel (Bio-Rad). Prestained molecular size markers (Bio-Rad) were used to track the running of the gel in order to excise a section of separated proteins between 85 and 100 kDa. The excised gel sections were stained in 0.1% bromphenol blue and cut into small cubes using a clean scalpel blade. Proteins were eluted from the gel pieces in a Little Blue Tank electrophoretic elution tank (ISCO, Lincoln, NE) with elution buffer (12.5 mM Tris base, 95 mM glycine, 0.05%, w/v, SDS). The electro-eluate was dialyzed against 1% Nonidet P-40 in PBS before the proteins were precipitated by adding two volumes of acetone and incubating on ice for 30 min. The precipitated protein was collected by maximal centrifugation in a bench-top microcentrifuge (Eppendorf) for 15 min at 4°C. The protein pellet was solubilized in two-dimensional PAGE sample loading buffer as described previously (24) and loaded onto the first dimension of the Millipore Investigator two-dimensional electrophoresis system (Millipore, Bedford, MA). The running conditions were essentially as described previously (23), except that the first dimensional isoelectric focusing tubes were reduced to 7.5 cm in length and the second dimension was run on a minigel apparatus (Bio-Rad), followed by electroblotting in a miniblot apparatus (Bio-Rad) onto PVDF membranes. Isoelectric point markers (Bio-Rad) were used for calibrating the first-dimensional isoelectric focusing. Prestained molecular weight markers (Sigma) for the second dimension were prepared in 1% agarose and cast into capillary tubes. The membrane was stained with Amido Black, washed in deionized water and wrapped in plastic sheet for photocopying. The membrane was then probed with PY20H antibody and the location of tyrosine-phosphorylated proteins visualized by ECL (Amersham). The location of any Amido Black-stained spots coinciding with the p90 phosphotyrosine signal was noted.
Microsequencing of p90 from PVDF Membrane-A duplicate experiment was performed with the intention of microsequencing the p90 protein obtained from the membrane stained with Amido Black. The putative p90 spot on the stained membrane was excised and subjected to direct N-terminal microsequencing. The excised membrane was loaded into a blot cartridge of a protein sequencer (Procise, Applied Biosystems, Foster City, CA) and run using cycles recommended by the manufacturer. The amino acid sequence obtained was searched against protein data bases using the BLAST facility (NCBI). The remaining membrane was probed with PY20H antibody to confirm the removal of the p90 phosphotyrosine signal.
Preparation and Purification of Polyclonal 80K-H Antibodies-Peptides corresponding to the 14 N-terminal amino acids (VEVKRPRGVS-LTNH) and the 14 C-terminal amino acids (PPPEAPTEDDHDEL) of the predicted human 80K-H protein sequence were synthesized, conjugated to keyhole limpet hemocyanin, and injected into rabbits by Neosystems Laboratoire (Strasbourg, France). Two C-terminal antibodies designated 80K-H-1029 and 80K-H-1030, and one N-terminal antibody designated 80K-H-1031, recognized a protein on one-and two-dimensional immunoblots that corresponded to the tyrosine-phosphorylated protein p90. These polyclonal antibodies were purified on peptide affinity columns prepared by coupling the synthetic peptide to CNBr-activated Sepharose 4B (Pharmacia, Uppsala, Sweden) according to the manufacturer's recommendations. The column, containing 1 ml of coupled gel, was equilibrated with five gel volumes of PBS. 0.5 ml of the antibody-containing serum was diluted with 0.5 ml of PBS and clarified through a 0.22-m filter. The diluted serum was mixed with the equilibrated gel and incubated for 1 h at 4°C. Unbound proteins were collected in the flow-through. The column was then washed with 5 ml of PBST (PBS containing 0.1% Tween-20 and 0.5 M NaCl). Bound proteins were eluted in five fractions of 2 ml of IgG elution buffer (Pierce). The fractions were neutralized immediately with 1 M Tris-HCl to pH 7.5 and assayed by immunoblotting cell lysates known to contain the 80K-H protein.
Immunoprecipitation and Western Blotting-Immunoprecipitations using the 80K-H antibody were performed as follows; 80 l of lysate (containing membrane proteins from one 15-cm plate of MRC-5 cells) was added to 400 l of affinity-purified 80K-H antibody and 480 l of 2 ϫ immunoprecipitation buffer (2% Triton X-100, 300 mM NaCl, 20 mM Tris, pH 7.4, 2 mM EDTA, 2 mM EGTA, 0.4 mM sodium orthovanadate, 0.4 mM PMSF, 1% Nonidet P-40). The mixture was incubated for 1 h at 4°C. 50 l of anti-rabbit IgG, conjugated to agarose beads were then added and the incubation continued for another 1 h. The beads were then washed three times in 1 ϫ immunoprecipitation buffer before boiling with 30 l of Laemmli buffer. The supernatant from the boiled beads was then analyzed by SDS-PAGE. Immunoblot analyses have been described previously (27). The commercial antibodies were used according to manufacturers' recommendations.
Binding Experiments with GRB-2 and Its SH2 and SH3 Domains-Cells were serum-starved as described above before stimulation and lysis. The cells were treated with bFGF (10 ng/ml) or with PDGF (50 ng/ml) or with diluent only. After washing with cold PBS, the cells were lysed with 500 l/plate of an Nonidet P-40 buffer (0.5% (v/v) Nonidet P-40, 10 mM HEPES, pH 7.9, 30 mM Na 4 P 2 O 7 , 5 mM EDTA, 5 mM EGTA, 1 mM dithiothreitol, 1 mM PMSF, 50 mM sodium fluoride, 100 M sodium orthovanadate, and 10 mM genistein). The lysates were spun in a bench-top centrifuge at full speed for 10 min, and the resultant pellet was discarded. The protein content of the lysate was assayed with a BCA protein assay kit (Pierce). Lysate protein (2 mg/ml) was used for each experiment. Lysate (500 l) was added to 500 l of GRB-2 association buffer (2% (v/v) Triton X-100, 30 mM NaCl, 20 mM Tris-HCl, pH 7.4, 2 mM EDTA, 2 mM EGTA, 0.4 mM sodium orthovanadate, 1 mM PMSF, 1% (v/v) Nonidet P-40) and 20 l of GRB-2-GST protein fusion beads were added and incubated for 2 h at 4°C. The beads were washed three times with 1 ml of 1 ϫ GRB-2 association buffer before the bound proteins were released by boiling with 30 l of 2 ϫ Laemmli buffer for 5 min. The supernatant was subjected to one-or two-dimensional PAGE and the separated proteins visualized by incubating with the appropriate antibodies and ECL. Controls consisted of GST linked to agarose beads or agarose beads alone.

RESULTS
A Protein Designated as p90 Is Rapidly and Uniquely Tyrosine-phosphorylated following FGF Stimulation-Swiss mouse 3T3 or human MRC-5 fibroblasts were used to study the time course of protein tyrosine phosphorylation induced by bFGF. Cells were lysed after stimulation, followed by resolu-p90 Is 80K-H tion of the cellular proteins on SDS-polyacrylamide gels, electroblotting onto PVDF membranes, and probing with antiphosphotyrosine antibodies (PY20H). Our observations were similar to previous reports (13), where p90 tyrosine phosphorylation is induced as early as 30 s after addition of bFGF and the response is detectable at doses as low as 0.5 ng/ml (data not shown). The level of phosphorylation peaked between 5 and 10 min and dropped slightly by 30 min. However, others have previously shown that p90 phosphorylation is sustained for at least 4 h and appears to be directly dependent on the continuous occupancy of the FGF receptor (13).
We then investigated whether this p90 response can be elicited by other growth factors and agonists. Different ligands were added to either Swiss 3T3 or MRC-5 cells, and the phosphotyrosine profiles of total lysates were compared to that obtained with bFGF (Fig. 1, lane 3). Neither EGF, when added to MRC-5 cells (lane 2), nor PDGF, when added to Swiss 3T3 cells (lane 8), was able to induce the phosphorylation of p90 although these ligands induced tyrosine phosphorylation of more proteins than bFGF in the respective cells. A similar lack of p90 phosphorylation was observed when hydrogen peroxide (a putative activator of tyrosine kinases and/or inhibitor of tyrosine phosphatases) was added to MRC-5 cells (lane 4). IL-6, which causes tyrosine phosphorylation through the recruitment of Janus kinases to its receptor components, was unable to induce phosphorylation of p90 in Swiss 3T3 cells (lane 6). Other ligands which are known to stimulate tyrosine phosphorylation and/or gene induction in these fibroblasts were also unable to induce the tyrosine phosphorylation of p90. These included both human and murine homologues of ␣/␤ and␥ interferons, tumor necrosis factor-␣, IL-1, CSF-1, and transforming growth factor-␤ (data not shown).
Tyrosine-phosphorylated p90 Is Located in the Membrane-The subcellular location of the tyrosine-phosphorylated p90 protein was ascertained by fractionating the cell into cytosolic and membrane extracts after treatment with bFGF. These fractions, along with total cell lysates from stimulated and unstimulated cells, were resolved by SDS-PAGE and transferred onto a PVDF membrane. The tyrosine-phosphorylated proteins were visualized by ECL after probing with anti-phosphotyrosine antibody (PY20H). Tyrosine phosphorylation of p90 occurs only in the membrane fraction of bFGF-stimulated cells (Fig. 2, lane 4)  The identification and characterization of p90, which is so far the only protein to undergo tyrosine phosphorylation uniquely in response to FGF, may be relevant to understanding the specificity of FGF signaling.
Two-dimensional Electrophoretic Resolution and Identification of p90 -For the partial purification of p90, MRC-5 cells were stimulated with bFGF for 10 min and the cell lysates resolved by two-dimensional PAGE (see "Experimental Procedures"). A prominent phosphotyrosine spot appeared on the blot (Fig. 3A) at a molecular mass of 90 kDa and an acidic pI of 4.5, which coincided with a group of three closely spaced spots stained by Amido Black. The corresponding protein spots from a parallel experiment, that does not involve PY20H probing, were excised and used for direct N-terminal sequencing. One unambiguous amino acid sequence was obtained (VEVKRPRGVSLTNHHFYD) which matched exactly with the first eighteen residues from the truncated N terminus of a human protein designated as 80K-H (25), as shown in Fig. 3B.
Rabbits were immunized with synthetic peptides derived from the experimental N terminus and the deduced C terminus of the protein. Two antibodies (80K-H-1029 and 80K-H-1030) raised against the 14-residue C-terminal peptide and one (80K-H-1031) raised against the 14-residue N-terminal peptide recognized a protein that co-migrates on two-dimensional PAGE gel with the heavily tyrosine-phosphorylated p90 protein in FGF-stimulated cells (Fig. 3C). Control Western blots using the respective preimmune sera did not detect any spots at 90 kDa (data not shown).
Immunoprecipitation studies carried out with crude or purified 80K-H antibodies were not successful. However, anti-phosphotyrosine immunoprecipitation clearly indicated that 80K-H is tyrosine-phosphorylated only in the bFGF-stimulated but not control cells (Fig. 4). MRC-5 cells were treated with bFGF for 10 min and separated into cytosol and membrane fractions. The membrane pellet was solubilized in SDS-containing buffer and incubated with 4G10 antibody conjugated to agarose. In Fig. 4, 80K-H is detected at 90 kDa in lane 4, which corresponds to tyrosine phosphoproteins derived from bFGF-stimulated cells. No 80K-H is detectable in immunoprecipitates obtained from unstimulated cells (lane 3). Although the amount of lysates used for immunoprecipitation in lanes 3 and 4 is about 20-fold higher than the loading in lanes 1 and 2, the signal for 80K-H is stronger in the latter lanes. This implies that the tyrosine-phosphorylated 80K-H represents only a small fraction of the total cellular 80K-H. p90 Is 80K-H these proteins. Since our efforts with direct 80K-H immunoprecipitation were unsuccessful, we set out to explore other affinity methods of isolating p90 to facilitate further proof of its identity as 80K-H. We speculated that p90, being phosphorylated on tyrosine residues, might bind GRB-2 and therefore screened a GRB-2-GST fusion protein for this purpose. A GRB-2-GST fusion protein, conjugated to agarose beads, was incubated with native lysates of Swiss 3T3 cells that were treated with optimal doses of bFGF or PDGF. After incubation, the associated proteins were separated, transferred to a PVDF membrane, incubated with anti-phosphotyrosine antibodies, and revealed by ECL. Treatment with bFGF led to four prominent tyrosine-phosphorylated proteins being associated with GRB-2 (Fig. 5A, lane 2). These proteins migrated at masses of 60, 72, 90, and 115 kDa, three of which were in common with the binding profile obtained from cells stimulated with PDGF (Fig. 5A, lane 3). The 90-kDa protein appeared to be unique to lysates derived from bFGF-stimulated cells. No proteins were found to associate nonspecifically with the GST-agarose matrix (lanes 4 -6).
The 60-and 72-kDa proteins were identified as SHC and PTP1D, respectively, by stripping the membranes and reprobing with the appropriate antibodies (data not shown). These proteins have already been reported to bind directly to GRB-2 (29,30). The 90-kDa protein was investigated in experiments described below, while the 115-kDa protein remains unidentified.
We decided to compare the two-dimensional electrophoretic behavior of the GRB-2-associating p90 with that of p90 in the total lysate to verify that they were identical proteins. Native cell lysates from bFGF-stimulated Swiss 3T3 cells were incubated with GRB-2-GST fusion protein conjugated to agarose beads. The complexed proteins were eluted from the beads with a two-dimensional PAGE loading buffer (32) and then separated by two-dimensional PAGE. The resulting autoradiograph (Fig. 5B) shows four major protein groups, consistent with the four main tyrosine-phosphorylated proteins observed in onedimensional SDS-PAGE (Fig. 5A, lane 2). In particular, the 90-kDa spot (Fig. 5B) co-migrates with the p90 spot in total cell lysates (Fig. 3A), possessing a similar pI and the same oblique inclination. The membrane was stripped and reprobed with 80K-H antibody which, upon visualization by ECL, gives a result similar to that shown in Fig. 3C.
80K-H Binds Mainly to the SH2 Domain of GRB-2-To evaluate the relative contribution of the three domains of GRB-2 in binding the four FGF-stimulated tyrosine-phosphorylated proteins, particularly p90/80K-H, three separate GST fusion proteins that incorporated either the N-terminal SH3, the middle SH2, or the C-terminal SH3 domain were employed in binding experiments similar to those described above. The resulting autoradiograph from one such experiment is shown in Fig. 6A.  The apparent amounts of tyrosine-phosphorylated SHC, PTP1D, and p90 are enhanced when compared to the equivalent amounts of each protein in the whole cell lysate. Both SHC and PTP1D have been shown to bind to GRB-2 protein via specific interactions between phosphotyrosine-containing motifs and the SH2 domain of GRB-2 (29,30). It is likely that p90 interacts with the GRB-2 SH2 domain in a similar manner. Lanes 5 and 7 show that neither the Nnor C-terminal SH3 domains associate with any phosphotyrosine proteins in the control cell lysates. Lanes 6 and 8 are the equivalent lanes obtained from binding experiments with lysates from FGF-stimulated cells. Only lane 8 contains a significant amount of tyrosine-phosphorylated proteins. Notably the 115-kDa protein is present in this lane, which implies that it binds to the C-terminal SH3 domain of GRB-2 in a specific manner. As SH3 domains interact with proline-rich motifs, the small but detectable amount of p90 protein present in lane 8 suggests a proline-rich sequence in p90. The absence of tyrosine-phosphorylated proteins in lanes 9 and 10 demonstrates the absence of any nonspecific interactions between phosphotyrosine proteins and the GST-agarose matrix.
Further proof of the identity of p90 as 80K-H and an estimation of the extent of its tyrosine phosphorylation was obtained by examining the p90 that bound to GRB-2(SH2)-GST fusion protein. In Fig. 6B, lanes 1 and 2 contain the proteins in cytosol and membrane fractions of FGF-treated MRC-5 cells, respectively. Aliquots of these lysates were incubated with GRB-2(SH2)/GST proteins, eluted and separated in lanes 3 (cytosol) and 4 (membrane). It is apparent that the SH2 domain of GRB-2 concentrates tyrosine-phosphorylated p90 from mem-branes derived from bFGF-stimulated cells (upper panel). The blot was subsequently stripped and reprobed with 80K-H antibody and visualized by ECL. From the immunoblot shown in Fig. 6B (lower panel), substantial amounts of 80K-H protein are found in the cytosol and membrane fractions but only a small amount of 80K-H in the membrane fraction binds to the SH2 domain of GRB-2. This is consistent with our above conclusion, based on immunoprecipitation with anti-phosphotyrosine antibody, that the tyrosine-phosphorylated 80K-H represents only a small fraction of the total cellular 80K-H.
In summary, the identification of p90 as 80K-H was initially obtained from microsequencing of a protein resolved on twodimensional PAGE. Antibodies raised against 80K-H consistently recognize p90 on all one-dimensional-and two-dimensional Western blots. The use of anti-phosphotyrosine immunoprecipitation and GRB-2 association studies provided proof that 80K-H is tyrosine-phosphorylated and binds to the SH2 domain of GRB-2 in vitro. DISCUSSION The involvement of receptor tyrosine kinases in cytokine and growth factor signaling has been reported extensively. Many of these studies have focused on the receptors for EGF and PDGF, FIG. 5. p90 binds to GRB-2-GST fusion protein in vitro. A, quiescent Swiss 3T3 cells were treated with diluent, bFGF (10 ng/ml), or PDGF (50 ng/ml) for 10 min prior to the lysates being incubated with GRB-2-GST fusion protein conjugated to agarose beads (lanes 1-3) or to GST-agarose beads alone (lanes 4 -6). The bound proteins were prepared for immunoblotting as described under "Experimental Procedures" and visualized by ECL. Lanes 1 (control), 2 (bFGF), and 3 (PDGF) are eluted from GRB-2-GST fusion protein and lanes 4 (control), 5 (bFGF), and 6 (PDGF) are eluted from GST-agarose beads. B, Swiss 3T3 cells were stimulated with bFGF (10 ng/ml for 10 min) and the lysate allowed to interact with GRB-2-GST fusion protein conjugated to agarose. The complexed proteins were eluted from the beads and separated by two-dimensional PAGE. The membrane was incubated with PY20H antibody and visualized by ECL. The arrow indicates the position of the 90-kDa protein that binds to GRB-2.  1 and 3) and membrane (lanes 2 and 4) extracts and separated on one-dimensional PAGE gels. The protein loading in the cytosolic fraction is 5 times higher than the membrane fraction. Aliquots of the two extracts were also incubated with GRB-2(SH2)/GST protein and subsequently eluted and separated on the same gel (lanes 3 and 4). The separated proteins were blotted onto a PVDF membrane, which was probed with PY20H antibody and visualized by ECL (upper panel). The same membrane was subsequently stripped and reprobed with 80K-H antibody (lower panel). The arrows indicate the position of p90/80K-H.
p90 Is 80K-H which were the first to be sequenced from this family. Relatively little is known about the activation and downstream signaling of the FGF receptor subfamily. Previous work had identified a few proteins that were tyrosine-phosphorylated in the early stages of FGF-induced signal transduction. These proteins included Src, cortactin, SHC, ERK-1, and ERK-2 (7-9, 11, 31-33). One prominent protein that remained unidentified was designated p90. The interest in p90 was heightened by its apparent specificity to the FGF signaling system, in contrast to the common activation of Src, SHC, and ERKs by numerous different growth factors.
In EGF-and PDGF-stimulated cells, both autophosphorylated EGFR and PDGFR are, respectively, the dominant tyrosine phosphoproteins seen in Western blots from whole cell lysates. The phosphorylated tyrosine residues on these receptors serve to recruit specific proteins containing complementary SH2 domains. The protein complexes thus formed are thought to be responsible for initiating various signaling pathways into the interior of the cell. We and others have observed that FGF receptors were not noticeably tyrosine-phosphorylated in response to FGF stimulation of various cells. Similarly, the insulin receptor shows a low level of tyrosine phosphorylation upon binding with insulin (34). In the latter case, however, a protein known as IRS becomes heavily tyrosine-phosphorylated after associating with the activated insulin receptor, thereby presenting multiple docking sites for SH2containing signaling proteins (34). Thus, in terms of the low level of receptor autophosphorylation, the insulin and FGF receptors show similar features following ligand activation. It is therefore possible that the FGF-specific tyrosine phosphorylation of p90 may serve a role similar to that seen with IRS in the insulin system where it could trigger signal propagation involving the formation of specific protein complexes.
Several proteins that are devoid of intrinsic catalytic activity can facilitate the assembly of these signaling protein complexes. These "adaptor" proteins include GRB-2, Crk, and Nck, which consist almost entirely of SH2 and SH3 domains. A number of tyrosine-phosphorylated proteins have been shown to be capable of binding to GRB-2 in vivo (17,18). They include, among a growing list, both the insulin receptor and IRS. In our in vitro assays, the FGF-specific p90 protein was found to associate with GRB-2, in addition to PTP1D, SHC, and an unidentified 115-kDa protein. Both PTP1D and SHC have well characterized SH2 domains and have been shown previously to bind directly to GRB-2 (29,30). Beyond our verification that PTP1D and SHC bind only to the SH2 domain of GRB-2, we have demonstrated that p90 also binds mainly to the SH2 domain. Since such associations were specific, they would provide a means for enriching tyrosine-phosphorylated p90 protein from cell lysates and aid in its identification. The 115-kDa protein, and to a lesser extent p90, were further shown to bind to the C-terminal SH3 domain of GRB-2, possibly via their respective proline-rich motifs.
The p90 protein was resolved by two-dimensional PAGE and subjected to microsequencing. An 18-residue N-terminal sequence was obtained, which matched exactly with the first 18 residues from the truncated N terminus of a human protein previously designated as 80K-H (25). Polyclonal antibodies were raised in rabbits against the predicted C terminus (14 residues) and observed N terminus (14 residues) of 80K-H. Several of these antibodies recognized a protein that co-migrated with p90 on Western blots derived from one-and twodimensional PAGE. Additional evidence for the identity of p90 comes from anti-phosphotyrosine immunoprecipitation, in vitro GRB-2-GST association experiments, and reversed-phase high performance liquid chromatography (data not shown) where p90 is being recognized by the 80K-H antibody.
The 80K-H protein was isolated several years ago during an effort to identify a ubiquitous 80 -87 kDa protein that is a strong substrate of protein kinase C (35). The main target protein of this work turned out to be the MARCKS protein (36), also designated 80K-L (L for light), which is now a well characterized PKC substrate. The 80K-H protein (H for heavy) is only a weak PKC substrate in vitro and appears not to be a PKC substrate in vivo (35). Aside from its subsequent cloning (25), the 80K-H protein was not characterized in greater detail. Its amino acid sequence deduced from cDNA contains several noteworthy features. Analysis performed on the Swiss-Prot data base revealed four putative protein kinase C phosphorylation sites, a possible calcium binding EF-hand domain, and a prominent glutamic acid repeat around the middle of the protein, which is seen in some proteins including tropomyosin, prothymosin, neurofilament triple L protein, myc-transforming protein, and adenovirus 5Ј terminal protein. There is a Cterminal HDEL sequence, which is a possible endoplasmic reticulum retention signal, although functional HDEL sequences are usually not found in mammalian cells (37). There are also at least three proline-rich motifs, PXXP, which may account for the low level binding of 80K-H to the SH3 domain of GRB-2. A consensus motif pYXNX has been reported in various proteins such as PTP1D, SHC, EGFR, and focal adhesion kinase, where the SH2 domain of GRB-2 binds (17). While the tyrosine-phosphorylated 80K-H binds the SH2 domain of GRB-2, no such sequence was found in the former protein.
Further characterization of the 80K-H/GRB-2 interaction may help to resolve this discrepancy.
The deduced amino acid sequence of 80K-H predicts a size of 59 kDa, which is much smaller than the 90-kDa migration in SDS-polyacrylamide gels. The predicted isoelectric point is around 4.4, which corresponds well with the experimental value. The size discrepancy is not without precedence since highly acidic proteins have been known to migrate anomalously upon SDS-PAGE. For example, the bovine MARCKS protein, with a pI similar to that of 80K-H, migrates at 80 -87 kDa although its predicted molecular mass is only 32 kDa (38). The 80K-H sequence contains 15 tyrosine residues, which are evenly distributed in the protein. There is, however, a rather polar distribution of cysteine residues toward both ends of the protein, suggesting that the native structure for 80K-H can contain multiple intra-and/or intermolecular disulfide bridges. Molecular modeling of 80K-H was not successful, as the protein does not contain enough sequence homology to known structural domains in existing data bases. We have further shown that while this protein is present in both the cytosol and membrane, the tyrosine-phosphorylated species is found in the membrane and can bind the SH2 domain of GRB-2 in vitro. Based on densitometric scanning, we estimated the proportion of tyrosine-phosphorylated 80K-H to be less than 1% of total cellular 80K-H.
The physiological function of 80K-H is yet unknown. We now report a possible role for 80K-H in FGF signaling. The biochemical characteristics of this novel p90/80K-H protein; its expression in different cells, tissues, and organisms; and its potential function in FGF-induced cell signaling and development will be the subject of further study.