A Functional Fibroblast Growth Factor-1 Immunoglobulin Fusion Protein*

Proteins of the fibroblast growth factor (FGF) family play diverse roles in embryonic development, angiogen-esis, and wound healing. The most well studied targets of FGF activity typically are cells of mesodermal and neuroectodermal origin; in addition, expression of FGF-1 (acidic FGF) is increased at several sites of chronic immunologic injury, and recent studies show that FGF-1 also may interact with cells of the immune system. In some human T cells, FGF-1 can induce signals necessary for production of interleukin-2, a key cyto-kine required for T cell proliferation. To better charac-terize the interaction of FGF-1 with FGF receptors on T cells, a fusion protein was constructed containing a portion of the constant region of human IgG1 (Fc) at the amino terminus of FGF-1. The Fc-FGF-1 fusion protein retained FGF function as determined by stimulation of tyrosine phosphorylation and DNA synthesis in NIH 3T3 cells. Binding of the intact fusion protein to FGF receptor 1 (FGFR1) on T cells was demonstrated by immunoprecipitation of the receptor bound to Fc-FGF-1 and by flow cytometry showing binding of fusion protein to T cells expressing FGFR1. This functional Fc-FGF-1 protein should prove useful in identifying FGFR-expressing cells. factors centrifugation, the supernatant containing Fc- FGF purified on heparin-Sepharose Cl-6B Biotech). After initial experiments demonstrated some proteolytic cleavage and release of free FGF-1 with storage at 4 the protein 70 In some experiments, the was further purified by elution from protein A-Sepharose with 0.2 M glycine-HCl 2.5) followed the addition of 1 M Tris pH or dialysis. Transfection of Jurkat T cells— A plasmid containing the human CD2 promoter and locus control region for expression in CD2 1 T cells, pM151 pM151. pCD2-FGFR1 transfect cells transfectants FGF-1, I pM151

Proteins of the fibroblast growth factor (FGF) family play diverse roles in embryonic development, angiogenesis, and wound healing. The most well studied targets of FGF activity typically are cells of mesodermal and neuroectodermal origin; in addition, expression of

FGF-1 (acidic FGF) is increased at several sites of chronic immunologic injury, and recent studies show that FGF-1 also may interact with cells of the immune system. In some human T cells, FGF-1 can induce signals necessary for production of interleukin-2, a key cytokine required for T cell proliferation. To better characterize the interaction of FGF-1 with FGF receptors on T cells, a fusion protein was constructed containing a portion of the constant region of human IgG1 (Fc) at the amino terminus of FGF-1. The Fc-FGF-1 fusion protein retained FGF function as determined by stimulation of tyrosine phosphorylation and DNA synthesis in NIH 3T3 cells. Binding of the intact fusion protein to FGF receptor 1 (FGFR1) on T cells was demonstrated by immunoprecipitation of the receptor bound to Fc-FGF-1 and by flow cytometry showing binding of fusion protein to T cells expressing FGFR1. This functional Fc-FGF-1 protein should prove useful in identifying FGFR-expressing cells.
Classical targets for stimulation by heparin binding growth factors FGF-1 and FGF-2 (basic FGF) 1 include cells of mesodermal and neuroectodermal origin such as endothelial cells, smooth muscle cells, fibroblasts, and myoblasts (reviewed in Ref. 1). FGF-1 and FGF-2 potently induce proliferation and migration of these cells as well as angiogenesis, events that are required during normal embryogenesis and tissue repair. Although expression of FGF-1 and FGF-2 at sites of injury is therefore beneficial, some evidence suggests these growth factors may also contribute to vascular pathology by promoting excessive intimal hyperplasia (2)(3)(4)(5). FGF-1 expression is increased at several sites of chronic immune injury, including the synovium in rheumatoid arthritis (6 -8), the myocardium in human hearts after transplantation (9 -11), and in transplanted kidneys undergoing chronic rejection (12,13). The pathologic lesions in these sites demonstrate cellular responses typical of FGF effects on mesenchymal cells that result in vascular intimal hyperplasia, increased extracellular matrix deposition, and neoangiogenesis. In addition, these sites are characterized by chronic infiltration of T lymphocytes, suggesting there may be interactions between the immune system and fibroblast growth factors as demonstrated by the finding that T cells can produce FGF-2 (14,15). Evidence that FGFs may have immunoregulatory effects on T cells was first provided in 1985 by Johnson and Torres (16), who showed that FGF, at physiologically relevant concentrations, could replace the requirement for IL-2 or helper cells in production of interferon-␥. Although FGF could activate intracellular signals necessary for interferon-␥ production, FGF alone could not stimulate proliferation of T cells. More recent studies show directly that some human T cells express receptors for FGF-1 and that the normally small subpopulation of FGF-responsive T cells is expanded in the peripheral blood of patients with rheumatoid arthritis and in patients after heart transplantation (8,17). These data suggest that T cells bearing FGF receptors can be stimulated and expanded in the FGF-enriched environment at sites of immune injury and subsequently migrate to the peripheral blood. As found in the earlier studies (16), FGF alone does not stimulate proliferation of T cells but together with engagement of the T cell antigen receptor induces production of IL-2 and proliferation (17). In T cells, FGF thus functions in a manner analogous to other well described "costimulators" (18,19) to activate a second signal transduction pathway necessary for T cell proliferation and effector function.
These findings suggest that FGF and FGF receptors in T cells may function quite differently than in cells in which FGF alone can stimulate proliferation, migration, and secretion of effector molecules such as plasminogen activator (20). Our efforts to investigate T cells that express FGF receptors and the mechanisms by which FGF signals in T cells have been hampered by the lack of reagents that can conveniently identify these cells and allow us to examine the fate of FGF and its receptors in T cells. To address this difficulty, the experiments reported here describe the production and characterization of a fusion protein that includes a portion of the constant region (Fc) of human IgG1 at the amino terminus of full-length FGF-1. In contrast to a previously reported fusion protein of FGF-1 with diphtheria toxin A chain, Fc-FGF-1 retains full FGF function, including the ability to induce DNA synthesis, tyrosine phosphorylation, and FGFR1 binding as well as Fc-mediated binding to protein A and anti-human IgG antibodies.

EXPERIMENTAL PROCEDURES
Production of Fc-FGF-1-Plasmid BS-2 mutFc was kindly provided by Dr. Melanie Spriggs, Immunex Corp. This plasmid includes the murine IL-7 leader sequence, a nonamer FLAG epitope, CH2 and CH3 domains of human IgG1 (amino acids 242-470 (21)), and a (Gly 4 Ser) 2 repeat to provide a flexible linker between the Fc domain and the amino terminus of the introduced protein (22). The Fc region of this construct includes several amino acid mutations introduced to diminish Fc receptor binding. SpeI and NotI sites were introduced at the 5Ј and 3Ј ends, respectively, of full-length FGF-1 cDNA by polymerase chain reaction, and the product was cloned into these sites downstream of CH3. The entire insert was released and cloned into pET23b, thereby introducing a T7 tag from the pET23b vector. The fusion protein produced in Escherichia coli could not be solubilized. Therefore, the T7 tag and IL-7 leader were removed to generate an insert containing the FLAG epitope, Fc region, and flexible linker flanked by a 5Ј NdeI site and 3Ј SpeI site. The insert and pET23b construct containing FGF downstream of the SpeI site were digested with NdeI and SpeI and ligated together. The insert encodes a protein comprised of an amino-terminal FLAG epitope and human IgG1 Fc fused to the amino terminus of FGF-1. Conditions for isopropyl-1-thio-b-D-galactopyranoside induction and lysis were modified to improve the yield of fusion protein in the soluble fraction of the bacterial lysate. Isopropyl-1-thio-b-D-galactopyranoside concentration was decreased to 0.4 mM, and bacteria were grown at 30°C for 5 h. The bacteria were harvested and resuspended in 80 ml of lysis buffer (50 mM Tris-HCl (pH 8.0), 150 mM NaCl, 10 mM EDTA, 10 mM glucose, 10 g/ml lysozyme, and protease inhibitor mixture (1 mM diisopropyl fluorophosphate, 5 g/ml aprotinin, 5 g/ml leupeptin, 0.5 g/ml pepstatin, 1 mM phenylmethylsulfonyl fluoride). After sonication and centrifugation, the supernatant containing Fc-FGF was purified on heparin-Sepharose Cl-6B (Amersham Pharmacia Biotech). After initial experiments demonstrated some proteolytic cleavage and release of free FGF-1 with storage at 4°C, the protein was stored in aliquots, frozen at Ϫ70°C. In some experiments, the protein was further purified by elution from protein A-Sepharose with 0.2 M glycine-HCl (pH 2.5) followed by the addition of 1 M Tris (pH 8.5) to pH 7.0 or dialysis.
Transfection of Jurkat T cells-A plasmid containing the human CD2 promoter and locus control region for expression in CD2 ϩ T cells, pM151 (23), was a gift from Dr. Dimitri Kioussis. FGFR1␤a1 cDNA (24) was provided by Dr. Wallace McKeehan. It was released with BamHI, blunt-ended, and ligated into the SmaI site of pM151. A plasmid containing FGFR1␤a1 in the correct orientation was designated pCD2-FGFR1 and used to transfect Jurkat T cells as described below. To produce Jurkat transfectants expressing FGF-1, a cDNA encoding fulllength FGF-1 (amino acids 1-154) in pMEXneo was excised, blunt ended, and ligated into the SmaI site of pM151 as above. A plasmid containing FGF-1 in the correct orientation was designated pCD2-FGF and used to transfect Jurkat T cells as follows. For both pCD2-FGFR1 and pCD2-FGF, co-transfection with pcDNA3.1 was performed to provide a selectable resistance gene. Transfections were performed by electroporation of 10 7 Jurkat cells in serum-free RPMI 1640 with 20 g of pCD2-FGF-1 or pCD2-FGFR1 plus 1 g of pcDNA3.1. Selection with G418 (1 mg/ml) was initiated 24 h later, and transfectants were maintained in medium containing G418. Cells transfected with pCD2-FGF-1 were screened by Western blot of cell lysates with rabbit anti-FGF-1 (Sigma). Cells transfected with pCD2-FGFR1 were screened for FGFR1 mRNA by reverse transcription-polymerase chain reaction of DNasetreated RNA using primers for FGFR1. Two lines were chosen, 7C-4-8, transfected with pCD2-FGF-1, designated Jur/FGF, and C2-14, transfected with pCD2-FGFR1, designated Jur/FGFR1.
Cell Culture, FGF Stimulation, and Western Blotting-NIH 3T3 cells were grown in DMEM with 10% FBS plus 20 g/ml gentamicin and 10 mM HEPES. For proliferation, NIH 3T3 cells were plated at 10 4 /well in 96-well tissue culture plates in medium with serum. After 24 h, medium was replaced with serum-free DMEM for 24 h. Recombinant human FGF-1 (1-154) (R&D Systems) or Fc-FGF in Dulbecco's modified Eagle's medium were added at the indicated concentrations with 10 units/ml heparin, and the plates were incubated at 37°C, 5% CO 2 for 24 h during which each well received 1 Ci of [ 3 H]thymidine for the last 18 h. [ 3 H]Thymidine incorporation was determined by liquid scintillation counting. Data are the mean Ϯ S.D. cpm of quadruplicate wells. For tyrosine phosphorylation, 5 ϫ 10 5 NIH 3T3 cells were plated in Dulbecco's modified Eagle's medium with 10% serum on 10-cm tissue culture dishes coated with 0.2% gelatin. After 24 h, the medium was replaced with serum-free medium. After 48 h, cells were stimulated with recombinant human FGF-1 (20 ng/ml) or Fc-FGF (60 ng/ml) plus 10 units/ml heparin in Dulbecco's modified Eagle's medium plus 0.1% BSA for 10 min at 37°C. The monolayers were rinsed 3ϫ with PBS, 0.1 mM sodium orthovanadate, following which the cells were scraped with 100 l of hot SDS sample buffer containing 10 mM dithiothreitol and 1 mM sodium orthovanadate. Whole cell lysates were run on 7% SDSpolyacrylamide gel electrophoresis, transferred to Immobilon, blocked with 5% nonfat dry milk in 10 mM Tris (pH 7.5), 100 mM NaCl, 0.1% Tween 20. Western blotting was performed with rabbit anti-phosphotyrosine (Transduction Laboratories) and horseradish peroxidase-conjugated goat anti-rabbit Ig and developed by ECL. Parental Jurkat T cells were grown in RPMI 1640, 10 mM glutamine, 10 mM HEPES, 10% fetal bovine serum, and 20 g/ml gentamicin. Jur/FGFR and Jur/FGF transfectants were grown in the same medium except for substitution of 1 mg/ml G418 for gentamicin.
Flow Cytometry with Fc-FGF-Jurkat T cells or transfectants (1-2 ϫ 10 6 cells) were washed and resuspended in Dulbecco's modified Eagle's medium, 0.1% BSA, 2.5 g/ml heparin. Fc-FGF was added (100 ng/ml), and the cells were incubated on ice for 1 h. Cells were collected by brief microcentrifuge spin and washed in the same medium and then in PBS with 250 g/ml heparin unless otherwise indicated and finally with PBS only. The cells were stained with FITC goat anti-human IgG (Fc) (1:80) or FITC (FabЈ) 2 goat anti-human IgG (Fc) (0.5 g) for 1 h on ice.
In some experiments, the cells were subsequently stained with phycoerythrin-anti-CD3.
FGF Receptor Cross-linking with Radiolabeled Ligands-FGF-1␣ (amino acids 21-154) was produced in E. coli DH5␣ (25). FGF-1␣ was labeled with Na 125 I using chloramine T and purified on heparin-Sepharose as described previously (26). Fc-FGF was labeled and purified in essentially the same manner except that the incubation time with chloramine T was decreased from 90to 75 s. For binding of 125 I-labeled FGF-1 or Fc-FGF, 5 ϫ 10 6 cells were incubated with 10 ng of FGF-1 or 20 -30 ng of Fc-FGF in PBS, 0.1%BSA, 2.5 g/ml heparin in a final volume of 250 l for 1 h on ice. The cells were collected by brief microcentrifuge spin, washed once in PBS, 0.1% BSA in 250 l, then in PBS containing 250 g/ml heparin, and finally in PBS. The cells were resuspended in PBS with varying concentrations of bis(sulfosuccinimidyl) suberate (BS 3 ; Pierce) as indicated for 20 min at room temperature. The cross-linking reaction was quenched by adding 50 mM Tris-HCl and 10 mM glycine (final concentration) for 10 min at room temperature. For direct analysis, cells were lysed in PBS, 1% Triton X-100, mixed with SDS sample buffer with 1 mM dithiothreitol, and run on SDS-polyacrylamide gels with 0.5 ϫ 10 6 cells/lane. Otherwise, the cells were lysed in PBS, 1% Triton X-100 with protease inhibitor mixture (Boehringer Mannheim) plus 1 mM sodium orthovanadate. For immunoprecipitation, lysates were incubated with protein A-Sepharose directly (15 l) or 1:500 final dilution of goat anti-human IgG (Dako) followed by the addition of protein A-Sepharose. For immunoprecipitation of FGFR1, 1 g of rabbit anti-FGFR1 (anti-flg, Santa Cruz Biotechnology) was used followed by binding to goat anti-rabbit IgG coupled to agarose (Sigma). Approximately 2-3 ϫ 10 6 cells were loaded/lane.
Other Reagents-Sypro-Orange fluorescent protein stain was from Molecular Probes (Eugene, OR), horseradish peroxidase-goat anti-rabbit Ig was from Southern Biotechnology, horseradish peroxidase-goat anti-mouse Ig was from Boehringer Mannheim, anti-FLAG monoclonal antibody was from Kodak, Na 125 I (17 Ci/mg) in 10 Ϫ5 M NaOH (pH 8 -11) was obtained from NEN Life Science Products.

Fc-FGF Binds Heparin and Is Recognized by Anti-FGF-1 and
Anti-human IgG-Purification of the bacterial lysate on heparin-Sepharose yielded a predominant band of approximately 50 kDa corresponding to the predicted molecular mass of the desired Fc-FGF-1 fusion protein (Fig. 1A). A portion of the fusion protein was additionally purified with protein A-Sepharose, and Western blotting was performed on the eluates with antibodies to FGF-1, human IgG, and the FLAG epitope (Fig.  1B). In each case, the 50-kDa protein stained with antibody, indicating the presence of the correct epitopes. The anti-FLAG antibody was least sensitive for detection of the fusion protein, presumably because of the presence of a single FLAG epitope and hence limited binding by the monoclonal antibody com-Fc-FGF-1 Fusion Protein pared with the polyclonal anti-FGF and anti-human IgG antibodies. The additional purification step on protein A-Sepharose appeared to offer no advantage and resulted in substantial loss of protein because of precipitation of the fusion protein after elution with 0.2 M glycine. Therefore, subsequent experiments were performed with Fc-FGF purified on heparin-Sepharose only. Preparations of Fc-FGF stored at 4°C for more than a week began to show cleavage to lower molecular mass proteins that reacted with anti-FGF by Western blotting (not shown). Fc-FGF stored at Ϫ20°C or Ϫ70°C was stable and showed no release of free FGF detectable by Western blotting when incubated with cells in culture medium overnight at 37°C (Fig. 1C).
Fc-FGF Stimulates Proliferation of NIH 3T3 Cells and Phosphorylation of p90 -To confirm that Fc-FGF was functional, serum-starved NIH 3T3 cells were stimulated with either Fc-FGF or recombinant human FGF-1, and DNA synthesis was measured by [ 3 H]thymidine incorporation. Fc-FGF stimulated DNA synthesis by NIH 3T3 cells in the presence of heparin; the concentration required was approximately 3-fold higher on a molar basis than that required for an equivalent response to recombinant human FGF-1 (Fig. 2). Little or no response was observed in the absence of heparin (data not shown).
After binding to FGF receptors, FGF-1 and FGF-2 rapidly stimulate tyrosine phosphorylation of several intracellular substrates, including a prominent membrane-associated protein of approximately 90 kDa (27,28) that may be an important link between FGFR and downstream effector Ras/mitogen-activated protein kinase signaling pathways. Goh et al. (29) identified a 90-kDa protein that is phosphorylated in response to FGF-2 and binds to the SH2 adapter protein GRB-2 (29). This protein is identical to 80K-H, a protein of unknown function. More recently, Kouhara et al. (30) identified a different, novel protein of 92-95 kDa, designated FRS2 (fibroblast growth factor receptor substrate 2), that is phosphorylated upon FGF-1 stimulation and also binds GRB-2 (30). To determine whether Fc-FGF similarly activates tyrosine phosphorylation via FGFR, Western blotting with anti-phosphotyrosine was performed on NIH 3T3 cells stimulated with FGF-1 or Fc-FGF. Both Fc-FGF and recombinant human FGF-1 induced tyrosine phosphorylation of a prominent 90-kDa protein and proteins between 83-86 kDa (Fig. 3). Taken together, the data indicate that Fc-FGF has functional FGF-1 and Ig properties. The following studies were performed to assure that the functional FGF activity of Fc-FGF resided in the intact fusion protein and did not result from cleavage of free FGF-1.

Fc-FGF Binds to FGFR1␤ on Jurkat T Cells-Previous stud-
ies demonstrated that a subset of human T cells and T cell clones can be stimulated to proliferate and produce IL-2 by the combination of FGF-1 and T cell antigen receptor engagement by anti-CD3 antibody (17). Similar results were observed with the human T cell tumor line Jurkat, which was found to express FGFR1 by reverse transcription-polymerase chain reaction and immunostaining with polyclonal antibody to FGFR1 ectodomain. Further experiments with Jurkat T cells have revealed that FGFR1 expression is highly variable in this cell line; the presence or absence of low or high levels of FGFR1 expression could not be reproducibly related to obvious differences or changes in culture conditions such as cell density,

Fc-FGF-1 Fusion Protein
concentration or source of serum, or duration of passage in vitro. To establish a more predictable system for further understanding of FGF-1 function and stimulation in T cells, Jurkat T cells were stably transfected with a plasmid directing expression of FGFR1␤ under the control of a T cell-specific promoter and locus control region. In addition, because T cells can produce FGF-2 (14, 15) and isoforms of FGF-2 regulate expression of FGFR1 (31, 32), Jurkat transfectants expressing FGF-1 (1-154) were produced using the same expression vector. Northern analysis demonstrated that Jurkat transfected with pCD2-FGFR1 (Jur/FGFR) expressed considerable amounts of mRNA for FGFR1␤ (Fig. 4). Nontransfected Jurkat and Jurkat transfected with pCD2-FGF (Jur/FGF) had little or no detectable endogenous mRNA for FGFR1. These cell lines were used first to examine binding of 125 I-labeled FGF-1␣ to demonstrate that the FGFR1 mRNA in Jurkat transfectants produced a functional receptor that binds FGF and second, to confirm that intact Fc-FGF binds to FGFR1.
Nontransfected and transfected Jurkat were incubated with 125 I-labeled FGF-1␣ followed by cross-linking with BS 3 , a water-soluble, membrane-impermeant cross-linker. The crosslinked cells were solubilized, and FGFR1 binding of labeled FGF-1␣ was analyzed by SDS-polyacrylamide gel electrophoresis of whole cell lysates and autoradiography (Fig. 5A). In Jur/FGFR, prominent bands were seen at approximately 140 kDa and Ͼ200 kDa, corresponding to the approximate molecular masses expected for FGF-1␣ cross-linked to FGFR1␤ monomers and dimers or higher order oligomers. Fainter bands of the same size were also seen in nontransfected Jurkat, consistent with the low level of FGFR␤ that was observed in these cells previously. Virtually no binding of 125 I-FGF-1␣ was seen with Jurkat transfected with FGF-1. When the same procedure was performed with 125 I-Fc-FGF, only high molecular mass bands were seen (Fig. 5A). These findings suggested several conclusions. First, the FGFR1 mRNA expressed in Jur/ FGFR encoded receptors capable of binding FGF as evident from the strong binding of FGF-1␣. Far fewer receptors were expressed by the parental nontransfected Jurkat and fewer yet by Jur/FGF. Second, the cross-linking conditions used appeared to favor formation of dimers or higher order oligomers with Fc-FGF compared with FGF-1␣ because no band was observed at the molecular mass expected for cross-linked monomers of Fc-FGF and FGFR1. This may not be surprising considering the long Fc domain presumably extending from Fc-FGF bound to the receptor and the observation that other Fc-fusion proteins spontaneously dimerize (33). Third, cross-linking of Fc-FGF to FGFR in these cells did not demonstrate the band at 140 kDa seen with cross-linking of FGF-1. This suggests that there was not substantial cleavage of labeled FGF-1 from Fc-FGF to produce monomers of labeled FGFbound FGFR1.
To determine whether monomeric cross-linking of Fc-FGF to FGFR1 could be identified, a titration of the cross-linker was performed (Fig. 5B) as described in the original characterization of BS 3 (34). In the absence of cross-linker, monomers and spontaneous dimers of labeled Fc-FGF were present. With increasing concentration of BS 3 , monomers and dimers of free Fc-FGF decreased and new higher molecular weight bands appeared at molecular weights appropriate for monomers of  3 . After washing, the soluble fraction of Triton lysates was mixed with SDS buffer and run on a 12-3% polyacrylamide gradient gel followed by autoradiography. C, co-precipitation of Fc-FGF with FGFR1. Cells were incubated with 125 I-labeled Fc-FGF for 1 h on ice followed by washing first in PBS with 250 g/ml heparin and then in PBS. The cells were either not crosslinked (left three lanes, no BS 3 ) or cross-linked with 0.5 mM BS 3 (right three lanes) followed by lysis in 1% Triton. The lysates were immunoprecipitated with rabbit anti-FGFR and goat anti-rabbit Ig-agarose, and the precipitates were eluted with SDS sample buffer and run on a 12-3% gradient SDS-polyacrylamide gel followed by autoradiography.

Fc-FGF-1 Fusion Protein
fusion protein compared with nontransfected Jurkat, consistent with the larger number of receptors expressed by the transfectant. As seen before with the whole cell lysates, Jur/FGF (3rd lane) showed the least binding of Fc-FGF to receptor. The results show that Fc-FGF is co-precipitated with FGFR1. These experiments were also performed with cross-linking before lysis and immunoprecipitation (Fig. 5C, 4th-6th lanes). In Jurkat (4th lane), bands corresponding to monomers of FGFR bound to labeled Fc-FGF monomers, and dimers are seen. In Jur/FGFR (5th lane), free Fc-FGF monomers and dimers are also present along with fusion protein cross-linked to FGFR, whereas little binding to Jur/FGF is seen (6th lane). These results confirm binding of Fc-FGF to FGFR1.
As an alternative approach to demonstrate binding of intact fusion protein to FGFR1, the fusion protein was immunoprecipitated via its Fc domain with protein A-Sepharose followed by blotting with antibody to FGFR1 to identify FGFR1 that co-precipitated with Fc-FGF (Fig. 6). Unlabeled Fc-FGF was allowed to bind at 4°C, the lysates were immunoprecipitated with protein A alone, and the eluted proteins were immunoblotted with anti-FGFR1 (Fig. 6A). In Jur/FGFR that had not been cross-linked (5th lane), prominent bands were seen at approximately 120 and 100 kDa. The 120-kDa band is at the molecular mass expected for FGFR1␤, demonstrating the coprecipitation of FGFR1␤ with Fc-FGF. The identity of the 100-kDa band is not known. Lower amounts of FGFR1 are seen for nontransfected Jurkat (4th lane) and Jur/FGF (6th lane), consistent with the results found in Fig. 5. Similar results were seen with binding of Fc-FGF at 37°C. (Fig. 6B, 1st lane), and the specificity of the immunoprecipitation is demonstrated by the absence of a FGFR1 band in the transfectant with no added fusion protein (Fig. 6B, 2nd lane). When the cells were crosslinked before lysis (Fig. 6A, 1st-3rd lanes), the Jur/FGFR transfectants showed additional bands at 170 -190 and Ͼ250 kDa, consistent with the molecular masses of cross-linked Fc-FGF/FGFR1␤ monomers and dimers (Fig. 6A, 2nd lane). In summary, the data show that immunoprecipitation of FGFR1 co-precipitates intact Fc-FGF and conversely, immunoprecipitation of the fusion protein via its Fc domain demonstrates that it is bound to FGFR1. Scatchard analysis demonstrated that Jur/FGFR expressed approximately 30,000 high affinity FGF receptors/cell compared with approximately 4,000 receptors on nontransfected Jurkat. The dissociation constant for FGF-1 was 0.18 nM compared with 2.5 nM for Fc-FGF (data not shown). These results are very similar to findings reported for a fusion protein of FGF-7 with Ig CH (33).
Flow Cytometry with Fc-FGF-A potential application of the fusion protein is identification of FGFR1-expressing cells in mixed populations; therefore Fc-FGF was tested for its utility in flow cytometry (Fig. 7). After incubation with Fc-FGF, Jurkat and Jur/FGFR were stained with FITC-anti-human IgG directly or first washed with 250 g/ml heparin to remove Fc-FGF from low affinity sites on the cell surface. Nontransfected Jurkat (Fig. 7A) showed significant binding of Fc-FGF detected by FITC-anti-human Ig that was substantially diminished by washing with heparin, indicating binding to a large number of low affinity sites and a small number of high affinity sites. There was considerably more binding of Fc-FGF to Jur/ FGFR than to nontransfected Jurkat, and this binding was only minimally diminished by washing with heparin, indicating binding to high affinity receptors on Jur/FGFR (Fig. 7B). The requirement for washing with 250 g/ml heparin to distinguish high affinity from low affinity binding did not interfere with analysis of other cell surface markers as shown for staining with phycoerythrin-anti-CD3 after staining for the fusion protein (Fig. 7, C-E). All of the Jurkat cell lines stained well with anti-CD3 following the heparin wash. The slightly greater than one log increase in fluorescence intensity in Jur/ FGFR versus Jurkat detected by flow cytometry correlates with FGFR expression found by immunoprecipitation and Scatchard analysis.

DISCUSSION
The experiments presented here show that a fusion protein comprised of human IgG1 Fc at the amino terminus of FGF-1 retains both FGF function and IgG constant region function, shown by stimulation of DNA synthesis and tyrosine phosphorylation in NIH 3T3 cells, binding to heparin and FGFR1, and binding to protein A and antibodies to human IgG. These Fc-FGF-1 Fusion Protein results differ in a potentially important way from a previously reported FGF-1 fusion protein, aFGF-dtA, a construct with diphtheria toxin A chain at the carboxyl terminus of FGF-1 (35,36). aFGF-dtA failed to induce cell proliferation or DNA synthesis in NIH 3T3 cells, although it did induce tyrosine phosphorylation, suggesting it bound to FGFR on NIH 3T3 cells (35). In contrast to native FGF-1, aFGF-dtA was found only in the cytoplasmic fraction and failed to translocate to the nucleus. Whether the fusion protein in the cytoplasmic fraction was internalized in the cytosol or remained bound to surface FGFR was not determined. In U2OS Dr1 cells that express diphtheria toxin receptors but are resistant to toxin action, combination of aFGF-dtA with dtB resulted in translocation to the nucleus via toxin receptors and induced DNA synthesis, but no tyrosine phosphorylation was induced because these cells lack FGF receptors (35). Further experiments in cells transfected with FGFR4 showed that this FGF receptor could translocate native FGF-1 to the nucleus, but not aFGF-dtA, despite high affinity binding of both native FGF-1 and the fusion protein. These data were interpreted to indicate that both FGFR-dependent tyrosine phosphorylation and nuclear translocation of FGF-1 must occur to induce cell proliferation (36). If this is the case, the experiments presented here suggest that Fc-FGF-1 does translocate to the nucleus as well as inducing tyrosine phosphorylation via FGF receptors and retains the functions of native FGF-1. The structural features that might be responsible for the distinct functional properties of the two fusion proteins are unclear at this time.
A concern with fusion proteins or epitope tags engineered at the amino terminus of FGF-1 is that proteolytic cleavage at the amino terminus is common during purification of native FGF-1, leading to the isolation of functional molecules lacking up to 20 amino-terminal amino acids (37)(38)(39). The addition of protease inhibitors during purification of FGF-1 from tissue markedly increased the yield of full-length FGF-1 (39). If similar cleavage events occur with an amino-terminal fusion protein, then functional FGF (e.g. 21-154) may be released from the fusion protein as a result of this proteolysis. We cannot exclude the possibility that some of the functional response of NIH 3T3 cells to Fc-FGF-1 results from such proteolysis, but it is unlikely that most of the activity of the fusion protein results from release of free FGF. The amount of fusion protein required to elicit DNA synthesis in NIH 3T3 cells was only moderately higher on a molar basis than for recombinant human FGF, and Western blotting of the fusion protein after incubation with cells did not reveal a substantial amount of degradation that would result in liberation of substantial free FGF (Fig. 1C). Immunoprecipitation of FGFR1 after binding to fusion protein, either with protein A, anti-human IgG, or anti-FGFR1, also demonstrates that the bound fusion protein is intact. Taken together, the results suggest that the functional activities identified are because of binding of intact fusion protein to cell surface FGFR1.
Other fusion proteins with members of the FGF family have been reported, including keratinocyte growth factor (FGF-7, KGF) with a mouse IgG1 (hinge, CH2, CH3) at the carboxyl terminus (33), recombinant FGF-2-saporin toxin (40), and fusions of FGF-1 with Pseudomonas exotoxin (41). The KGF-HFc fusion protein maintained both KGF and IgG Fc functions as evident from 1) induction of DNA synthesis in KGF-responsive cells, 2) specific binding to NIH 3T3 cells transfected with KGF receptor and not FGFR1 or FGFR2 shown by flow cytometry, 3) immunoprecipitation with anti-mouse IgG, and 4) immunohistochemical staining of KGF receptor in tissues. As seen with Fc-FGF-1, the KGF-HFc protein spontaneously formed dimers, functional assays required somewhat higher concentrations of the fusion protein, and there was a very similar 10-fold difference in receptor affinity for KGF versus KGF-HFc (K d 0.13 nM versus 1.4 nM). Based on these limited data with the FGF-7 fusion protein and Fc-FGF-1 described here, the addition of IgG Fc domains to FGF family members appears able to preserve function and receptor binding despite the relatively large size of the added Fc region compared with the FGF. The reported toxin conjugates of both FGF-1 (41) and FGF-2 (40) are at the carboxyl termini of the respective FGFs and bind to high affinity FGFR. In the FGF-2-saporin fusion protein, proteolytic processing at the carboxyl terminus of FGF-2 appears to be required for liberating the toxin in the cytoplasm after receptor-mediated internalization to obtain maximal toxicity. Thus, determining whether amino-terminal or carboxyl-terminal fusions with FGF family members are most suitable for the desired goal may be largely empiric.
Fc-FGF may have a number of useful applications for studies of FGF function and cells expressing FGFR. Flow cytometry demonstrated binding with nontransfected Jurkat, expressing as few as 4,000 high affinity FGFR/cell; however, the shift from control is very modest and at the limit of detection. Jur/FGFR transfectants expressing approximately 25-35,000 high affinity FGFR1 are easily identified. Taken together, the results suggest that cells expressing 8,000-10,000 high affinity receptors should be readily identifiable by flow cytometry with Fc-FGF. Experiments in which varying percentages of Jur/FGFR were mixed with human peripheral blood mononuclear cells revealed that as few as 3% Jur/FGFR could be easily identified, whereas addition of Jur/FGFR as 1% of the total cell population was essentially indistinguishable from background (data not shown). This reagent could thus be useful for sorting mixed populations to obtain cells expressing either high or low levels of receptor. Because the frequency of FGFR ϩ T cells in human peripheral blood ranges from 1/2000 to 1/40,000 CD3 ϩ T cells by functional assay (17), it is unlikely that flow cytometry with Fc-FGF will be useful in demonstrating these cells in unselected peripheral blood lymphocytes. Studies are in progress to determine whether the fusion protein can be used for immunohistology to identify FGFR ϩ cells in tissue sections that may be useful in studies of developmental biology, oncology, and angiogenesis.
A second potential application of Fc-FGF is in tracking the intracellular fate of FGF and its receptors. Preliminary studies using immunofluorescent localization show that the appearance of membrane-bound Fc-FGF differs when cells are exposed to the fusion protein at 37°C versus 4°C. Diffuse membrane staining, present after 4°C incubation, becomes highly focal with incubation at 37 C°, suggesting that oligomerization of the receptors can be directly observed by immunofluorescence (data not shown). If intact fusion protein remains bound by receptors after internalization, this approach may allow visualization of the nuclear localization of FGF and its receptors, and investigation of the nuclear factors with which they associate by immunoprecipitation via the Fc domain.