The nuclear trafficking of extracellular fibroblast growth factor (FGF)-1 correlates with the perinuclear association of the FGF receptor-1alpha isoforms but not the FGF receptor-1beta isoforms.

The alternatively spliced fibroblast growth factor receptor (FGFR)-1 isoforms, FGFR-1alpha and FGFR-1beta, are characterized by the presence of either three or two Ig-like loops in the extracellular domain and are differentially expressed during embryonic development and tumor progression. We have previously shown that in cells irreversibly committed to DNA synthesis by FGF-1, approximately 15% of cell surface FGFR-1 traffics to a perinuclear locale as a structurally intact and functional tyrosine kinase (Prudovsky, I., Savion, N., Zhan, X., Friesel, R., Xu, J., Hou, J., McKeehan, W. L., and Maciag, T. (1994) J. Biol. Chem. 269, 31720-31724). In order to define the structural requirement for association of FGFR-1 with the nucleus, the expression and trafficking of FGFR-1 in FGFR-1alpha and FGFR-1beta L6 myoblast transfectants was studied. Although FGFR-1alpha was expressed as p145 and p125 forms, FGFR-1beta was expressed as p120 and p100 forms in the L6 myoblast transfectants. Tunicamycin and N-glyconase experiments suggest that these forms of FGFR-1alpha and FGFR-1beta are the result of differential glycosylation. However, only the p145 form of FGFR-1alpha and the p120 form of FGFR-1beta were able to bind FGF-1 and activate tyrosine phosphorylation. Pulse-chase analysis of FGFR-1 biosynthesis suggests that the p125 and p100 proteins are the precursor forms of p145 FGFR-1alpha and p120 FGFR-1beta, respectively. Because ligand-chase analysis demonstrated that FGFR-1beta L6 myoblast transfectants exhibited a reduced efficiency of nuclear translocation of exogenous FGF-1 when compared with FGFR-1alpha transfectants, the intracellular trafficking of the FGFR-1alpha and FGFR-1beta isoforms was studied using an in vitro kinase assay to amplify immunoprecipitated FGFR-1. Indeed, the appearance of the FGFR-1alpha but not FGFR-1beta isoform in the nuclear fraction of L6 myoblast transfectants suggests that the distal Ig-like loop in FGFR-1alpha mediates the differential nuclear association of FGFR-1alpha as a structurally intact and functional tyrosine kinase. Further, the FGFR-1beta L6 myoblast transfectants but not the FGFR-1alpha myoblast transfectants exhibited a pronounced morphologic change in response to exogenous FGF-1. Because this phenotype change involves the induction of a rounded cellular shape, it is possible that the FGFR-1alpha and FGFR-1beta may ultimately exhibit differential trafficking to adhesion sites.

The alternatively spliced fibroblast growth factor receptor (FGFR)-1 isoforms, FGFR-1␣ and FGFR-1␤, are characterized by the presence of either three or two Ig-like loops in the extracellular domain and are differentially expressed during embryonic development and tumor progression. We have previously shown that in cells irreversibly committed to DNA synthesis by FGF-1, approximately 15% of cell surface FGFR-1 traffics to a perinuclear locale as a structurally intact and functional tyrosine kinase (Prudovsky, I., Savion, N., Zhan, X., Friesel, R., Xu, J., Hou, J., McKeehan, W. L., and Maciag, T. (1994) J. Biol. Chem. 269, 31720 -31724). In order to define the structural requirement for association of FGFR-1 with the nucleus, the expression and trafficking of FGFR-1 in FGFR-1␣ and FGFR-1␤ L6 myoblast transfectants was studied. Although FGFR-1␣ was expressed as p145 and p125 forms, FGFR-1␤ was expressed as p120 and p100 forms in the L6 myoblast transfectants. Tunicamycin and N-glyconase experiments suggest that these forms of FGFR-1␣ and FGFR-1␤ are the result of differential glycosylation. However, only the p145 form of FGFR-1␣ and the p120 form of FGFR-1␤ were able to bind FGF-1 and activate tyrosine phosphorylation. Pulse-chase analysis of FGFR-1 biosynthesis suggests that the p125 and p100 proteins are the precursor forms of p145 FGFR-1␣ and p120 FGFR-1␤, respectively. Because ligand-chase analysis demonstrated that FGFR-1␤ L6 myoblast transfectants exhibited a reduced efficiency of nuclear translocation of exogenous FGF-1 when compared with FGFR-1␣ transfectants, the intracellular trafficking of the FGFR-1␣ and FGFR-1␤ isoforms was studied using an in vitro kinase assay to amplify immunoprecipitated FGFR-1. Indeed, the appearance of the FGFR-1␣ but not FGFR-1␤ isoform in the nuclear fraction of L6 myoblast transfectants suggests that the distal Ig-like loop in FGFR-1␣ mediates the differential nuclear association of FGFR-1␣ as a structurally intact and functional tyrosine kinase. Further, the FGFR-1␤ L6 myoblast transfectants but not the FGFR-1␣ myoblast transfectants exhibited a pronounced morphologic change in response to exogenous FGF-1. Because this phenotype change involves the induction of a rounded cellular shape, it is possible that the FGFR-1␣ and FGFR-1␤ may ultimately exhibit differential trafficking to adhesion sites.
Members of the fibroblast growth factor (FGF) 1 gene family encode nine proteins that play important roles in embryogenesis, wound repair, angiogenesis, tumor growth, and other biologic and pathologic processes (1)(2)(3). The biological activities of the FGFs, including the signal peptide-less prototypes, FGF-1 (acidic) and FGF-2 (basic), are mediated through high affinity cell surface-associated FGF receptors (FGFRs). Presently there are four FGFR gene family members encoding proteins that contain a ligand-activated intrinsic tyrosine kinase domain (4). The ability of the FGF prototypes to initiate DNA synthesis requires the continual exposure of the target cell to exogenous FGF during the entire G 1 phase of the cell cycle in vitro (5,6). Throughout this period, receptor-mediated internalization of FGF-1 results in its partition between the nucleus and cytosol (5,7) including the perinuclear trafficking of FGFR-1 (8).
In addition to the exogenous signaling pathway, there also appears to be an endogenous pathway for mediating the intracellular traffic of FGF-1, FGF-2, and FGF-3 (9, 10). Indeed, sequences responsible for nuclear localization have been characterized in FGF-1 (7,11,12), FGF-2 (13-15) and FGF-3 (16). Interestingly, a nucleolar localization sequence has recently been characterized in FGF-3 as an endogenous protein (17) and the intranuclear trafficking of FGF-1, 2 FGF-2 (18), and FGF-3 (17) as endogenous intracellular proteins have been correlated with a decrease in proliferation potential in vitro. In addition, the nuclear trafficking of endogenous FGF-2 (19 -22) and FGF-3 (23) but not FGF-1 appears to be further complicated by the presence of alternative 5Ј-CUG translational start sites. In contrast, however, the biological significance of the nuclear trafficking properties of exogenous FGF is not clear, although the internalization of exogenous FGFs is receptor-dependent and correlates with an increased proliferative potential in vitro (1-3, 5, 9, 10). Indeed, the forced secretion of the FGF prototypes yields either a prominent transformed phenotype in vitro (24) or exaggerated hyperplasia in vivo (25,26), and many of the signal peptide-containing FGF gene family members have been characterized as oncogenes (2,3). It therefore appears that there may be distinct intracellular trafficking pathways for endogenous and exogenous FGFs that lead to different cellular phenotypes, which are dependent upon the ability of the ligand to interact with its receptor. The expression of the FGFR genes as translation products is regulated by alternative splicing (27)(28)(29)(30)(31), and the two isoforms of FGFR-1 serve as an example. FGFR-1␣ and FGFR-1␤ have been characterized as containing three and two Ig-like structures in the amino-terminal ectodomain of their translation products (29 -31). The differential expression of the FGFR-1␣ and FGFR-1␤ isoforms may be biologically significant because it has recently been shown that malignant astrocytomas express both forms of FGFR-1, whereas normal fetal and adult brain express only FGFR-1␣ (32). Further, similar FGFR-2␣ and ␤ isoforms are differentially expressed during amphibian development (33), and it has been demonstrated that the FGFR-1␣ and FGFR-1␤ isoforms exhibit different affinities for ligand and heparin (34,35). We have shown that FGFR-1 is translocated to a perinuclear locale as a structurally intact and functional tyrosine kinase during the G 1 transition period in NIH 3T3 cells in vitro (8). Because ligand-induced FGFR-1 trafficking may be a component of the FGF-1 signaling pathway, we compared the intracellular trafficking of the FGFR-1␣ and FGFR-1␤ isoforms using stable L6 myoblast transfectants and report that unlike the perinuclear association of FGFR-1␣, FGFR-1␤ does not traffic to a perinuclear locale.

EXPERIMENTAL PROCEDURES
Cell Culture and Transfection-L6 myoblasts (a gift from Dr. Lewis Williams, University of California-San Francisco) were grown in DMEM supplemented with either 10% (v/v) bovine serum or 10% (v/v) fetal bovine serum and were density arrested in DMEM containing 0.5% (v/v) fetal bovine serum for 18 h. Quiescent cells were stimulated with 40 ng/ml of recombinant human FGF-1 expressed and purified as described previously (11) in the presence of heparin (10 g/ml). L6 myoblasts were transfected with expression constructs encoding FGFR-1␣ and FGFR-1␤. The cDNA encoding human FGFR-1 with the two Ig-like disulfide loops, pZip-FGFR-1 (29), was digested with BamHI and inserted into the pMEXneo plasmid under the control of the murine sarcoma virus LTR (36). The cDNA encoding the human three Ig-like disulfide loop form of FGFR-1 in a pBluescript SK plasmid was digested with BamHI and inserted into the pMEXneo plasmid. Cell transfection was performed using calcium phosphate according to the recommended protocol from Stratagene.
Antibodies-Rabbit antiserum recognizing a carboxyl-terminal sequence in FGFR-1 was prepared against a synthetic peptide based on residues 764 -776 of the deduced amino acid sequence of Xenopus FGFR-1 (FR-1) as described (37). The FGFR-1␣-specific monoclonal antibody M2F12 was raised against the bacterially expressed extracellular domain of the 50-kDa FGFR-1 three Ig-like disulfide loop antigen with a recognition domain on the first Ig loop (residues 50 -79) as described (38). Monoclonal anti-vinculin antibodies (Sigma, V-4505) were used for immunofluorescence microscopy.
Cell Fractionation-Cells were lysed in 1 ml of 20 mM Tris, pH 7.5, containing 300 mM sucrose, 60 mM KCl, 15 mM NaCl, 0.5 mM EDTA, 0.5% (v/v) Triton X-100, 1 mM phenylmethylsulfonyl fluoride, 2 g/ml aprotinin, 10 g/ml leupeptin, and 1 mM sodium vanadate (nuclear preparation buffer), and the cytosolic fraction and nuclear pellet were prepared as described (8). The purity of nuclear preparations was determined by the presence of less than 2% acid phosphatase activity in the nuclear fraction of the total cell lysate activity. Transmission electron microscopic analysis also demonstrated the absence of cytoplasmic membranes and organelles in the nuclear fraction. The nuclei were solubilized in 0.25 ml of nuclear preparation buffer containing 0.4% (w/v) SDS, 1 mM MgCl 2 , and 10 units DNase 1 (Promega) for 3 min at 4°C followed by the addition of 0.75 ml of nuclear preparation buffer. The nuclear extract was incubated at 37°C for 5 min and at room temperature for an additional 10 min and centrifuged at 14,000 ϫ g for 10 min. In experiments with the antibody M2F12, the nuclei were solubilized by ultrasonication (10 pulses, 0.5 s each) in 1 ml of nuclear preparation buffer because SDS prevents the binding of M2F12 to FGFR-1␣.
Immunoprecipitation and in Vitro Kinase Assay-Aliquots of the nuclear and cytosolic fractions (0.5 ml each) were immunoprecipitated with the appropriate anti-FGFR-1 antibody as described previously (5). The various FGFR-1⅐antibody complexes were precipitated using Protein A-Sepharose beads and washed, and the in vitro kinase reaction was initiated by the incubation of the precipitates with 10 Ci of [␥-32 P]ATP as described previously (8).
Cross-linking of [ 125 I]FGF-1 and Immunoblot Analysis-[ 125 I]FGF-1 was prepared as described previously (8). Transfected L6 cells in 100-mm diameter dishes were incubated for 20 min at 22°C with 4 ml of binding buffer containing 10 ng of [ 125 I]FGF-1/ml in the presence or the absence of 1 g of unlabeled FGF-1/ml as described (39). Cells were transferred to 37°C for 5 min and then rapidly washed at 37°C in phosphate-buffered saline and disuccinimidyl suberate added to a final concentration of 0.3 mM. Dishes were incubated for an additional 15 min at 37°C and then rapidly washed as described (39). Cell lysates were analyzed by 7.5% (w/v) SDS-PAGE and prepared for autoradiography.
Cell lysates from L6 myoblast FGFR-1 transfectants were resolved by 7.5% (w/v) SDS-PAGE and transferred to a nitrocellulose filter. The filters were processed with either the affinity-purified anti-FGFR-1 carboxyl-terminal antibody (FR-1) or a monoclonal antiphosphotyrosine antibody (Upstate Biotechnology) as described previously (5).
Pulse-Chase and Ligand-Chase Analysis-Transfected L6 myoblasts were washed with phosphate-buffered saline and incubated in methionine-free DMEM for 20 min, after which [ 35 S]methionine was added for 30 min followed by the addition of an excess of cold methionine (in DMEM) for the specified times. Cells were lysed and immunoprecipitation performed as described above.
Ligand-chase analysis was used to assess the kinetics of [ 125 I]FGF-1 nuclear trafficking and was performed as described previously (5,12). Briefly, serum-starved (48 h in DMEM containing 10 g/ml of insulin and 10 g/ml of transferrin) FGFR-1␣ and FGFR-1␤ L6 myoblast transfectants were exposed to 10 ng/ml of cold FGF-1 and 10 units/ml of heparin for various periods of time; the cells were washed with 10 g/ml of heparin and exposed to [ 125 I]FGF-1 (5 ng/ml) for 2 h after which cytosol; and nuclear fractions were obtained. The intracellular content of [ 125 I]FGF-1 was analyzed by 15% (w/v) SDS-PAGE, and cells maintained at 4°C served as a negative control (5).
Immunofluorescence Microscopy-Serum-starved and FGF-1-stimulated FGFR-1␣ and FGFR-1␤ L6 myoblast transfectants were grown on coverslips, fixed for 3 min in acetone (Ϫ80°C), air-dried, incubated for 1 h in phosphate-buffered saline containing 1% (w/v) bovine serum albumin and 0.1% (v/v) Triton X-100 (Buffer A), and incubated with anti-vinculin monoclonal AB (Sigma) for 1 h in Buffer A. The coverslips were washed three times with phosphate-buffered saline, incubated for 1 h with fluorescein-conjugated goat-anti-mouse Ig, washed three times with phosphate-buffered saline, and mounted in 50% (v/v) glycerol on microscope slides. Microscopy was performed with an Olympus fluorescent microscope.

The Expression of FGFR-1␣ and FGFR-1␤ Exhibit Multiple Molecular Weight Forms, but Only the High M r Forms Are Able
to Bind Exogenous FGF-1-Stable L6 myoblast FGFR-1␣ and FGFR-1␤ transfectants were obtained, and all exhibited a significant increase in DNA synthesis in response to either 10% (v/v) fetal bovine serum or 10 ng/ml of FGF-1 in vitro (data not shown). The L6 myoblast FGFR-1␣ and FGFR-1␤ transfectants were also unable to form myotubes at high cell density, a characteristic of the parental cell line (40). Immunoblot analysis of lysates from the L6 myoblast FGFR-1␣ transfectants (Fig. 1A) using a FGFR-1 antibody directed against the carboxyl terminus of the protein (FR-1) revealed the presence of two polypeptides, p145 and p125, that were not detected in lysates derived from the nontransfected L6 myoblast controls. Immunoblot analysis of lysates from L6 myoblast FGFR-1␤ transfectants under similar conditions also revealed the presence of two polypeptides, p120 and p100 (Fig. 1A). Similar data were also obtained using NIH 3T3 cell transfectants (data not shown). Because the FR-1 antibody recognizes a common carboxyl-terminal domain between FGFR-1␣ and FGFR-1␤, the multiple M r forms of FGFR-1 expressed in L6 myoblast FGFR-1␣ and FGFR-1␤ transfectants are likely to represent different amino-terminal ectodomain FGFR-1 isoforms.
In order to determine whether the various forms of the FGFR-1␣ and FGFR-1␤ translation products are able to associate with extracellular FGF-1, covalent [ 125 I]FGF-1 cross-linking experiments were performed. We utilized L6 myoblast transfectants because the endogenous FGFR population is quite low in contrast with the FGFR population present on the surface of other cell lines such as the NIH 3T3 cell (5,39). Although attempts to covalently cross-link [ 125 I]FGF-1 to nontransfected L6 myoblasts were not successful, multiple covalently cross-linked bands were readily visible in both FGFR-1␣ and FGFR-1␤ L6 myoblast transfectants (Fig. 1B). The FGFR-1␣ L6 myoblast transfectants exhibited a major band with an apparent M r between 160 -165 kDa, which correlates with a competitive ligand-receptor complex between FGF-1 and the p145 form of FGFR-1␣. Interestingly, we were not able to detect a cross-linked band that would correlate with a complex between [ 125 I]FGF-1 and the p125 form of FGFR-1␣. In contrast, competitive [ 125 I]FGF-1 covalent cross-linking analysis of the FGFR-1␤ L6 myoblast transfectants revealed the presence of two bands, a major band with an apparent M r between 135 and 140 kDa and a minor band with a M r of approximately 100 kDa (Fig. 1B). Although it is likely that the high M r band represents a covalent complex between [ 125 I]FGF-1 and p120, we were unable to detect a band that would correlate with a covalent complex between [ 125 I]FGF-1 and p100 (approximate M r of 120 kDa). Although we do not know the identity of the p100 band, it may represent a degradation product of the high M r FGFR-1␤⅐[ 125 I]FGF-1 complex. These data suggest that the high M r forms of FGFR-1␣ (p145) and FGFR-1␤ (p120) but not the low M r forms of FGFR-1␣ (p125) and FGFR-1␤ (p100) are able to associate with exogenous FGF-1.
FGF-1 Is Able to Induce the Tyrosine Phosphorylation of the High M r Forms of FGFR-1␣ and FGFR-1␤ in L6 Myoblast Transfectants-Because the data obtained from competitive [ 125 I]FGF-1⅐FGFR-1 covalent cross-linking analysis suggested that only the high M r forms of FGFR-1␣ and FGFR-1␤ are able to bind FGF-1, we were curious whether this discriminatory pattern would also be evident in the ability of FGF-1 to induce FGFR-1 autophosphorylation of tyrosine residues. Therefore, FGFR-1␣ and FGFR-1␤ L6 myoblasts were treated with exogenous FGF-1, and the cells were subjected to immunoprecipitation with anti-FGFR-1 antiserum followed by immunoblot analysis using an anti-phosphotyrosine antibody. As shown in Fig. 2, we were only able to detect the presence of p145 and p120 as phosphotyrosine-containing proteins in the FGFR-1␣ and FGFR-1␤ L6 myoblast transfectants, respectively. Thus, our failure to detect p125 and p100 bands as phosphotyrosinecontaining proteins in the FGFR-1␣ and FGFR-1␤ L6 myoblast transfectants is consistent with the prior observation that only the high M r forms of FGFR-1␣ and FGFR-1␤ are able to associate with exogenous FGF-1.
The Low M r Forms of FGFR-1␣ and FGFR-1␤ Are Precursors of the High M r Forms of These Receptors-Because the low M r form of FGFR-1␣ (p125) and FGFR-1␤ (p100) are unable to respond to exogenous FGF-1 and autophosphorylate on tyrosine residues, we examined the possibility that there existed a product-precursor relationship between the high and low M r forms of FGFR-1␣ and FGFR-1␤. In order to address this issue, pulse-chase analysis of [ 35 S]methionine-labeled FGFR-1␣ and FGFR-1␤ L6 myoblasts was performed. Using FGFR-1 immunoprecipitation of the FGFR-1␣ L6 myoblast transfectants, we observed that after a 15-min incubation with [ 35 S]methionine, the radiolabel was present only in the p125 of FGFR-1␣ (Fig. 3). After 1 h, the p145 form of FGFR-1␣ was detected, and after 165 min, the label was present exclusively in the p145 form of FGFR-1␣. The use of this strategy with FGFR-1␤-transfected L6 myoblasts revealed a similar relationship between the early appearance of the p100 form of FGFR-1␤ and a significant increase in the presence of the p120 form of FGFR-1␤ very late in the chase (Fig. 3). These data suggest that the p125 form is the precursor of the p145 product of FGFR-1␣, and similarly, Serum-starved FGFR-1␣ and FGFR-1␤ L6 myoblast transfectants were stimulated with FGF-1 (10 ng/ml) and heparin (10 g/ml) for 1 and 6 h. Equal amounts of the FGFR-1␣ and FGFR-1␤ transfected lysates from either serum-starved or FGF-1-stimulated cells were precipitated with the FR-1 antibody described in the legend to Fig. 1. The precipitates were resolved by 7.5% (w/v) SDS-PAGE, and the tyrosine phosphorylation was analyzed by immunoblot analysis using a monoclonal antiphosphotyrosine antibody. Positions of the p145 and p120 bands are indicated with arrows. cont, control. the p100 form is the precursor of the p120 product of FGFR-1␤.
The High and Low M r Forms of FGFR-1␣ and FGFR-1␤ May Undergo Differential N-Glycosylation in Vitro-Because the high M r forms of FGFR-1␣ and FGFR-1␤ are derived from their corresponding low M r forms and FGFR-1 does contain N-glycosylation sites (41), we questioned whether the high M r forms of FGFR-1␣ (p145) and FGFR-1␤ (p120) are post-translational modifications of the low M r forms of FGFR-1␣ (p125) and FGFR-1␤ (p100). Pretreatment of [ 35 S]methionine-labeled FGFR-1␣ and FGFR-1␤ L6 myoblast transfectants with tunicamycin followed by immunoprecipitation with FGFR-1 antiserum revealed that FGFR-1␣ and FGFR-1␤ are synthesized as proteins with apparent M r of 110 and 95 kDa, respectively (Fig.  4). In addition, we did not observe high M r forms of FGFR-1␣ and FGFR-1␤ under these conditions. Similarly, using an in vitro kinase assay, it was possible to demonstrate that treatment of the FGFR-1␣ and FGFR-1␤ immunoprecipitates with N-glyconase yielded a significant decrease in M r in which the p145 and p125 form of FGFR-1␣ were converted to p110 and the p120 and p100 forms of FGFR-1␤ were converted to p95 (data not shown). These data suggest that the p145 and p125 forms of FGFR-1␣ and the p120 and p100 forms of FGFR-1␤ are synthesized as a result of post-translational N-glycosylation of the p110 and p95 forms of FGFR-1␣ and FGFR-1␤, respectively.

FGFR-1␤ but Not FGFR-1␣-transfected Cells Exhibit Morphological Changes upon the Stimulation with FGF-1-We examined the FGFR-1␣ and FGFR-1␤ L6 myoblast transfectants
for FGF-1-dependent morphological changes. As shown in Fig.  5 (A and B), the treatment of the FGFR-1␣ L6 myoblast transfectants with FGF-1 did not result in a change in monolayer phenotype. In contrast, similar treatment of the FGFR-1␤ L6 myoblast transfectants exhibited a pronounced morphologic change as manifested by the less flattened monolayer appearance and the presence of well rounded yet anchored cells (Fig.  5, C and D). This phenotype was exhibited in at least 50% of the FGFR-1␤ L6 myoblast transfectant population. Immunofluorescence analysis using anti-vinculin antibodies revealed a significant decrease in the level of focal adhesion sites in the FGFR-1␤ (Fig. 5, E and F) but not in the FGFR-1␣ (data not shown) L6 myoblast transfectants.
FGFR-1␤ but Not FGFR-1␣ Exhibits a Reduced Efficiency of Nuclear Translocation of Exogenous FGF-1-It is well described that the exogenous FGF prototypes traffic to a nuclear locale in a receptor-dependent manner during the entire G 1 transition period (5,6,12,42). In an attempt to determine the contribution of the FGFR-1␣ and FGFR-1␤ isoforms as potential mediators of this trafficking event, FGFR-1␣ and FGFR-1␤ L6 myoblast transfectants were examined for their ability to partition exogenous [ 125 I]FGF-1 between cytosol and nuclear fractions during the immediate-early (2 h) and mid-to-late (6 h) G 1 phase of the cell cycle. In these experiments, the transfected L6 myoblasts were pretreated at 37°C with cold FGF-1 for the periods of time described in Fig. 6, washed with heparin, and treated with [ 125 I]FGF-1 for 2 h, after which the cells were lysed and cytosol and nuclear fractions prepared. As shown in Fig. 6, FGFR-1␣ L6 myoblast transfectants exhibited a significant level of nuclear-associated [ 125 I]FGF-1 after 2 and 6 h of exposure to cold FGF-1. In contrast, FGFR-1␤ L6 myoblast transfectants exhibited a reduction in the level of [ 125 I]FGF-1 associated with the nuclear fraction (Fig. 6). These data suggest that the isoforms of FGFR-1␤ expressed in the FGFR-1␤ L6 myoblast transfectants are unable to efficiently translocate exogenous FGF-1 from the cell surface to the nucleus during the immediate-early and mid-to-late G 1 phase of the cell cycle. These data also imply that the p145 FGFR-1␣ isoform may be responsible for the receptor-dependent trafficking of exogenous FGF-1 from the cell surface to the nucleus.  FGFR-1␣ and FGFR-1␤ L6 myoblast transfectants were density arrested, pretreated with tunicamycin (tun, 10 g/ml) for 1 h, and further incubated with tunicamycin and [ 35 S]methionine for 3 h. The cells were harvested, and cell lysates were immunoprecipitated with either the FR-1 antibody described in the legend to Fig. 1 or preimmune serum (pi). The immunoprecipitants were resolved by 7.5% (w/v) SDS-PAGE analysis, and the positions of the p145, p125, p120, p110, p100, and p95 proteins in the autoradiogram are indicated with arrows. Nucleus in L6 Myoblast Transfectants-We have previously reported the nuclear association of FGFR-1␣ in FGFR-1␣transfected and control NIH 3T3 cell and L6 myoblasts (8) and confirmed the presence of the p125 and p145 FGFR-1␣ isoforms in the nuclear fraction of control L6 myoblasts using both the nonspecific FR-1 antibody and the highly specific monoclonal antibody M2F12, which recognized the first Ig-like loop of FGFR-1␣ (data not shown). Because the FGFR-1␤ L6 myoblast transfectants exhibited a reduced efficiency of exogenous FGF-1 nuclear trafficking when compared with FGFR-1␣ transfectants, we sought to determine whether FGFR-1␤ exhibited a reduction in the perinuclear trafficking. Thus, the perinuclear association of FGFR-1␤ was studied in FGFR-1␤ L6 myoblast transfectants using the nondiscriminatory anti-FGFR-1 antibody FR-1 for immunoprecipitation followed by the amplification in an in vitro kinase reaction and resolution of the phosphorylated proteins by radioautography. As shown in Fig. 7, the p120 and p100 translation products of the FGFR-1␤ gene were present in the cytosolic fraction and absent in the nuclear fraction. However, the presence of endogenous p125 and p145 isoforms of FGFR-1␣ was detected both in cytosolic and nuclear fractions. These data suggest that the absence of the FGFR-1␤ isoforms in the nuclear fraction may be the result of a reduction in the efficiency of the FGFR-1␤ to associate with the nuclear fraction and argue that the first Ig-like loop may contain the structural information requisite for perinuclear trafficking. We also examined the importance of glycosylation for the nuclear association of FGFR-1␣ in FIG. 5

. Morphological and immunofluorescence analysis of FGFR-1␣ and FGFR-1␤ L6 transfectants upon stimulation with FGF-1.
Serum-starved FGFR-1␣ and FGFR-1␤ L6 cell transfectants were stimulated for 24 h without (A and C) and with FGF-1 (10 ng/ml) and heparin (10 g/ml) (B and D). Cells were photographed using the phase contrast (200ϫ). Immunofluorescence analysis (600ϫ) of FGFR-1␤ L6 myoblast transfectants stimulated (F) or not stimulated (E) with FGF-1 and heparin was performed as described previously (8)  FGFR-1␣ L6 myoblast transfectants. As shown in Fig. 8, the appearance of the p145 and p125 FGFR-1␣ isoforms in the nuclear fraction may require N-glycosylation because pretreatment of the cells with tunicamycin significantly reduces the level of the p145 and p125 FGFR-1␣ isoforms in the nuclear fraction. It should be noted that although the presence of the p110 precursor of the p125 and p145 FGFR-1␣ isoforms was readily detected in the cytosol fraction of tunicamycin-treated cells, the nonglycosylated p110 FGFR-1␣ precursor was not present in the nuclear fraction (Fig. 8). DISCUSSION The nuclear trafficking properties of the FGFs are well described and may involve the function of two pathways. The endogenous pathway of nuclear traffic appears to be limited to the high M r forms of FGF-2 (18 -22) and FGF-3 (16, 17, 23), whose expression is regulated by alternative translational control at upstream CUG start sites. FGF-1 is excluded from this pathway because the open-reading frame encoding FGF-1 is flanked by termination codons (42), and the identification and characterization of a cytosol-retention sequence in FGF-1 is consistent with its presence in the cytosol as an endogenous translation product. 2 Although the function of the endogenous pathway of FGF nuclear trafficking appears to be associated with the repression of cell growth, the exogenous pathway of FGF nuclear trafficking is a receptor-dependent process (5,(12)(13)(14) and has been associated with the stimulation of cell growth. However, the potential function of FGF trafficking to the nucleus by the exogenous pathway as a repressor of cell proliferation has not been eliminated. Indeed, the requirement for the continual presence of extracellular FGF-1 (5) and FGF-2 (6) during the entire G 1 transition period for the initiation of maximal DNA synthesis and the remarkable long-term (months) stability of nuclear-associated FGF-1 following cell proliferation in vivo (43) imply that the exogenous pathway of FGF nuclear trafficking may be both permissive for cell proliferation in the short term yet may ultimately be repressive for cell proliferation in the long term.
Because FGFR isoform expression is a result of alternating gene splicing (27)(28)(29)(30)(31), it has been difficult to anticipate their individual contribution to the exogenous FGF nuclear trafficking pathway. Indeed, our data suggest that the FGFR-1␣ and FGFR-1␤ isoforms do exhibit differential biochemical properties that are both permissive and restrictive of their involvement in the exogenous FGF trafficking pathway. Analysis by pulse-chase, N-glyconase treatment, and tunicamycin interference argue that the high and low M r FGFR-1␣ and FGFR-1␤ isoforms are the products of common precursor forms; the p145 and p125 FGFR-1␣ isoforms are ultimately derived from a common p110 precursor, whereas the p120 and p100 FGFR-1␤ isoforms are derived from a p95 precursor. The p125 form of FGFR-1␣ and the p145 FGFR-1␣ product are both the result of N-glycosylation, and it appears that the p125 FGFR-1␣ isoform is the precursor for the mature p145 FGFR-1␣ product. Likewise, the p120 FGFR-1␤ and p100 FGFR-1␤ isoforms are also FIG. 6. Ligand-chase analysis of [ 125 I]FGF-1 intracellular traffic in FGFR-1␣ and FGFR-1␤ L6 myoblast transfectants. Serumstarved untransfected L6 myoblast FGFR-1␣ and FGFR-1␤ transfectants were stimulated with nonradiolabeled FGF-1 (40 ng/ml) and heparin (10 g/ml) for 2 or 6 h at 37°C, after which the transfectants were washed with heparin (10 g/ml) and exposed to [ 125 I]FGF-1 for an additional 2 h at 37°C (8). The transfectants were harvested, lysed, and fractionated into cytosol (cyt) and nuclear (nuc) fractions, and the fate of [ 125 I]FGF-1 was analyzed by 15% (w/v) SDS-PAGE. The ratio of cytosol to nuclear lysate in the sample load was 1 to 50. Serum-starved transfectants not exposed to nonradiolabeled FGF-1 at 37°C but exposed to [ 125 I]FGF-1 for 2 h (0h), and cells exposed to [ 125 I]FGF-1 at 4°C for 2 h (4°) served as controls for the 2-and 6-h ligand chase.
FIG. 7. The appearance of FGFR-1 tyrosine-kinase activity in the cytosol and nuclear fraction of FGFR-1␤ L6 cell transfectants. Serum-starved FGFR-1␤ L6 cell transfectants were stimulated for 10 h with 40 ng/ml FGF-1 in the presence of 10 g/ml heparin. The cells were lysed and fractionated into cytosolic and nuclear fractions as described under "Experimental Procedures." In this experiment, the nuclear fraction was solubilized by DNase and SDS treatment as described previously (8), and the cytosol (cyt) and nuclear (nuc) fractions were immunoprecipitated with either preimmune serum (p) or the FR-1 antibody (i) described in the legend to Fig. 1. The in vitro kinase reaction was initiated as described under "Experimental Procedures," and the reaction products (5 l of cytosol and 40 l of nuclear lysate) were resolved by 7.5% (w/v) SDS-PAGE analysis. The positions of the relevant bands are indicated with arrows. FIG. 8. FGFR-1 isoform tyrosine kinase activity in FGFR-1␣ L6 myoblast transfectants treated with tunicamycin. Serum-starved FGFR-1␣ L6 myoblast transfectants were treated with FGF-1 (40 ng/ ml) and heparin (10 g/ml) for 10 h in the absence or the presence of tunicamycin (10 g/ml). Cells were harvested, lysed, and fractionated into cytosol and nuclear fractions as described under "Experimental Procedures." The nuclear fraction was solubilized by DNase and SDS treatment as described previously (8), and the cytosol (cyt) and nuclear (nuc) fractions were immunoprecipitated with the FR-1 antibody described in the legend to Fig. 1. The in vitro kinase reaction was initiated as described under "Experimental Procedures," and the reaction products (5 l of cytosol and 40 l of nuclear lysate) were resolved by 7.5% (w/v) SDS-PAGE analysis. The positions of the relevant bands are indicated with arrows. the result of N-glycosylation of the common p95 FGFR-1␤ precursor, and the p100 FGFR-1␤ isoform is the precursor for the mature p120 FGFR-1␤ product. Although the mature p145 FGFR-1␣ isoform appears to be involved in the regulation of FGF-1 nuclear trafficking, the mature p120 FGFR-1␤ isoform does not. However, both forms of the mature p145 FGFR-1␣ and p120 FGFR-1␤ are able to recognize exogenous FGF-1 and are functional as FGF-1-induced tyrosine kinases. Thus it appears that the amino-terminal Ig-like loop is involved in the sorting of FGFR-1␣ but not FGFR-1␤ to a perinuclear locale during the G 1 phase of the cell cycle. This observation is consistent with the recent report that the p110 FGFR-1 isoform is present in breast epithelial cells in the cytosolic compartment (44). Likewise, the requirement for N-glycosylation within the amino-terminal FGFR-1␣ Ig loop for perinuclear trafficking is also consistent with this conclusion. However, the presence of a ligand-induced and functional FGFR-1 tyrosine kinase are not sufficient to ensure the function of the FGF-1 nuclear trafficking pathway.
The perinuclear traffic of the p125 FGFR-1␣ isoform in L6 myoblasts is also interesting because FGF-1 neither binds the p125 FGFR-1␣ isoform nor activates its intrinsic tyrosine kinase, even though the enzymatic activity of the p125 FGFR-1␣ isoform is functional in an in vitro kinase reaction. Although we do not understand the mechanism of the p125 FGFR-1␣ isoform perinuclear transport, it is possible that this isoform may be competent to enter the trafficking pathway by dimerization with the ligand-binding competent p145 FGFR-1␣ isoform.
We chose to study FGFR-1 isoform traffic in the L6 myoblast because they are not responsive to exogenous FGF and are considered to be FGFR-deficient (45). Their classification as a FGFR-deficient cell line is based upon their inability to bind radiolabeled FGF (45), and we have also not been able to detect a [ 125 I]FGF-1⅐FGFR complex using competitive covalent radioligand cross-linking methods. However, Hawker and Granger, using covalent ligand-receptor cross-linking and immunoblot analysis (46), have recently reported low levels of FGFR in the L6 cells. We also have been able to demonstrate by FGFR-1 immunoprecipitation and enzymatic amplification in an in vitro kinase assay the presence of the p125 FGFR-1␣ and p145 FGFR-1␣ isoforms in untransfected L6 myoblasts (8). Interestingly, these FGFR-1␣ isoforms appear to traffic independent of the presence of exogenous FGF-1. Although these data are consistent with the results obtained from FGFR-1␣ and FGFR-1␤ L6 myoblast transfectants, they do suggest that the L6 myoblast may be able to express low levels of an extracellular member of the FGF gene family. Because the rat DNA sequences for the majority of the FGF gene family are not available, it has not been possible to utilize either reverse transcriptase-polymerase chain reaction or RNase protection analysis to assess this possibility at least at the mRNA level.
The observation that the p120 FGFR-1␤ and p100 FGFR-1␤ isoforms do not traffic to a perinuclear locale argues that differential biological properties associated with the FGFR-1␣ and FGFR-1␤ isoforms (32,33) should be considered relative to their differential intracellular trafficking properties. Indeed, this may be particularly informative with regard to differences between established cell lines and diploid cell strains. Likewise, the intracellular trafficking patterns of other members of the FGFR gene family where equivalent alternatively spliced FGFR isoforms have been characterized (27,33) may also likely provide structural correlates to biological patterns of differential isoform expression (44). Because the differential trafficking of the FGFR-1␣ isoforms occurs during the G 1 transition period as a structurally intact and functional tyrosine kinase, it is possible that FGFR-1␣ isoforms traffic from the plasma mem-brane via a novel mechanism. The perinuclear trafficking of the FGFR-1␣ isoforms is also interesting because we have described a biphasic interaction between FGFR-1 and Src during the immediate-early and mid-to-late G 1 phase of the cell cycle (47). We have also reported that FGF-1 is able to regulate the phosphorylation of the Src substrate, cortactin, during these time periods (48) as well as induce a biphasic association between Src and cortactin (47). Interestingly, (i) cortactin is a F-actin-binding protein (49), (ii) Src (50) and related tyrosine kinases (51,52) are also associated with adhesion sites and traffic to a perinuclear locale during the G 1 transition period, and (iii) the cytoskeleton is linked to adhesion sites (53). In addition, FGFR-1␣ has been reported to exhibit associative properties with the glycosylaminoglycan, heparin (34). Because (i) the members of the FGF gene family are well documented as heparin/heparan sulfate proteoglycan-binding growth factors (1-3), (ii) the heparan sulfate proteoglycans are localized at adhesion sites (54), (iii) many of the adhesion macromolecules contain heparin-binding domains (55,56), and (iv) the low affinity heparan sulfate proteoglycans appear to be essential to propagation of a FGF-induced mitogenic signal (57)(58)(59)(60), it is possible that these sites may contain regulatory features for biologically astute FGF-FGFR interactions. The observations that (i) the heparin-binding domain in FGFR-1␣ has also been recognized for a homology with cell adhesion molecules and (ii) antibodies against this domain block biological signal responses specific for these cell adhesion molecules (61) are also consistent with this premise.
Unlike FGFR-1␣ transfectants, L6 myoblasts transfected with FGFR-1␤ acquired a FGF-1-dependent morphology resembling highly malignant tumor cells including a well rounded shape and a decrease in the level of focal adhesion sites. This observation is consistent with the report that the induction of FGFR-1␤ and not FGFR-1␣ expression may be involved in the regulation of the malignant potential of astrocytes (32). However, the mechanisms responsible for this activity of FGFR-1␤ (but not FGFR-1␣) is unclear, and two alternative pathways are suggested. First, there may be a differential association of FGFR-1␣ and FGFR-1␤ with focal adhesion sites, and these differences may involve the ability of the FGFR-1 isoforms to associate with cell matrix and/or cell adhesion molecules. Alternatively, the nuclear trafficking of FGF-1 provided by FGFR-1␣ but not by FGFR-1␤ could potentially downregulate the morphological effects that are triggered by exogenous FGF-1. Indeed, it has been suggested that the intranuclear trafficking of endogenous FGF-1, 2 FGF-2 (18), and FGF-3 (17) not only decreases cell proliferation but also induces a more flattened monolayer phenotype. Thus, it is possible that the intranuclear trafficking of exogenous FGF-1 provided by FGFR-1␣ may play a feedback role that regulates the mitogenic activity of FGF-1 and thus limit the development of a malignant phenotype. The absence of this feedback by FGF-1-induced signaling of FGFR-1␤ may result in an attenuation of the metastatic potential.
We have also noted that the first Ig-like loop in the FGFR-1␣ isoforms does not contain a sequence with structural similarity to known nuclear localization signals (62), and this is consistent with our prior observation that the translocation of FGFR-1␣ is restricted to a perinuclear locale in NIH 3T3 cells (8). Thus, the role of the first Ig-like loop in the FGFR-1␣ isoforms may be limited to the trafficking of the appropriate exogenous FGF ligand to a perinuclear locale that would enable the exogenous FGF ligand to utilize its structural nuclear localization signal for nuclear (5,42) and perhaps nucleolar translocation (13,14). Indeed, the reduced efficiency of exogenous FGF-1 nuclear trafficking exhibited by point mutants in the FGF-1 nuclear localization sequence (12) are consistent with this suggestion.