Fibroblast Growth Factor Receptor 3 Induces Gene Expression Primarily through Ras-independent Signal Transduction Pathways*

Fibroblast growth factor receptors (FGFR) are widely expressed in many tissues and cell types, and the temporal expression of these receptors and their ligands play important roles in the control of development. There are four FGFR family members, FGFR-1–4, and understanding the ability of these receptors to transduce signals is central to understanding how they function in controlling differentiation and development. We have utilized signal transduction by FGF-1 in PC12 cells to compare the ability of FGFR-1 and FGFR-3 to elicit the neuronal phenotype. In PC12 cells FGFR-1 is much more potent in the induction of neurite outgrowth than FGFR-3. This correlated with the ability of FGFR-1 to induce robust and sustained activation of the Ras-dependent mitogen-activated protein kinase pathways. In contrast, FGFR-3 could not induce strong sustained Ras-dependent signals. In this study, we analyzed the ability of FGFR-3 to induce the expression of sodium channels, peripherin, and Thy-1 in PC12 cells because all three of these proteins are known to be induced via Ras-independent pathways. We determined that FGFR-3 was capable of inducing several Ras-independent gene expression pathways important to the neuronal phenotype to a level equivalent of that induced by FGFR-1. Thus, FGFR-3 elicits phenotypic changes primarily though activation of Ras-independent pathways in the absence of robust Ras-dependent signals.

Fibroblast growth factor receptors (FGFR) are widely expressed in many tissues and cell types, and the temporal expression of these receptors and their ligands play important roles in the control of development. There are four FGFR family members, FGFR-1-4, and understanding the ability of these receptors to transduce signals is central to understanding how they function in controlling differentiation and development. We have utilized signal transduction by FGF-1 in PC12 cells to compare the ability of FGFR-1 and FGFR-3 to elicit the neuronal phenotype. In PC12 cells FGFR-1 is much more potent in the induction of neurite outgrowth than FGFR-3. This correlated with the ability of FGFR-1 to induce robust and sustained activation of the Ras-dependent mitogen-activated protein kinase pathways. In contrast, FGFR-3 could not induce strong sustained Rasdependent signals. In this study, we analyzed the ability of FGFR-3 to induce the expression of sodium channels, peripherin, and Thy-1 in PC12 cells because all three of these proteins are known to be induced via Ras-independent pathways. We determined that FGFR-3 was capable of inducing several Ras-independent gene expression pathways important to the neuronal phenotype to a level equivalent of that induced by FGFR-1. Thus, FGFR-3 elicits phenotypic changes primarily though activation of Ras-independent pathways in the absence of robust Ras-dependent signals.
The fibroblast growth factors (FGFs) 1 play roles in development, angiogenesis, wound healing, and tumorigenesis (reviewed in Ref. 1). FGF actions are mediated by activation of FGF receptor (FGFR) tyrosine kinases. FGFRs are a gene family of four members, termed FGFR-1-4. These receptors are widely expressed in many tissues and different cell types, and the temporal expression of the receptors and their ligands is regulated during development (reviewed in Ref. 2). Analysis of naturally occurring mutations in these receptors has indicated that they control the differentiation of specific cell types during development. Point mutations within the genes encoding human FGFR-1, -2, or -3 cause different syndromes that involve bone development (reviewed in Refs. 3 and 4) and some of these syndromes (Aperts and thanatophoric dysplasia) may also manifest effects in the central nervous system. Point mutations in FGFR-3 that cause activation of its tyrosine kinase activity have been shown to be responsible for the commonest form of dwarfism in humans (4 -7). Recently FGFR-3 has also been implicated in multiple myeloma, where its abnormal overexpression because of a chromosomal translocation has been detected in ϳ25% of cases (8,9). However, it is not clear what role this expression contributes to the phenotype of this disease. Studies analyzing the consequences of null mutations in FGFRs in mice also implicated these receptors as playing a role in development. The knockout of either FGFR-1 or FGFR-2 (10,11) in mice resulted in embryonic lethality, whereas that of FGFR-3 was nonlethal. The FGFR-3-deficient mice developed an overgrowth of the long bones and abnormal curvature of the spine and tail (12,13) and were deaf (12).
The FGFRs are very similar in structure. In particular, their tyrosine kinase domains are highly conserved, and overlapping subsets of ligands induce their activation. Regulation appears to take place at two different levels. Temporal control of the expression of both ligands and receptors is an important mechanism for regulating signal transduction during development. In addition, the receptors seem to have differing signaling capabilities. Studies have indicated that FGFR-1 is much better at producing mitogenic signals than either FGFR-3 or FGFR-4 when assayed in BaF3 cells (14 -16). We have demonstrated that there is also a difference between FGFR-1 and -3 in their abilities to induce neurite outgrowth in PC12 cells when activated by FGF-1 (17,18). FGFR-3 can barely induce neurite outgrowth, whereas activation of FGFR-1 induces rapid and robust neurite outgrowth. In the BaF3 and PC12 cell systems it appears that sustained signals that lead to the Ras-dependent activation of the extracellular regulated kinases are necessary for the biological phenotypes observed.
The above results imply that FGFR-3 is not able to induce strong sustained signals. However, in vivo it obviously plays key roles in development. There are two potential nonexclusive hypotheses that could explain these observations. The threshold for signaling in these developmental tissues may be low, and sustained signals may not be mandatory. Alternatively FGFR-3 may induce physiologically relevant and strong signals using signaling pathways that are distinct from the Ras-extracellular regulated kinase pathway. In this study we have chosen to examine this latter possibility. The nature of FGFR-3 signaling and the downstream targets are still not well understood and are the subject of intense investigation. To establish the signaling capabilities of FGFR-3 and to determine the nature of FGFR-3 induced signals, we took advantage of the PC12 cell system in which either NGF or FGF-1 can induce the elaboration of a neuronal phenotype. Underlying the growth factor-induced acquisition of neuronal phenotype are a variety of signaling pathways and gene expression changes that lead to the specific neuronal traits. Prominent neuronal traits are mediated by both Ras-dependent and Ras-independent signaling pathways. For example, gene expression events leading to morphological differentiation (neurite outgrowth) are predominantly mediated via a Ras-dependent mitogen-activated protein kinase pathway, whereas the establishment of a sodium based action potential is mediated by the expression of voltagedependent sodium channel genes in a Ras-independent manner. Normal PC12 cells express both FGFR-1 and FGFR-3, and both of these receptors can be activated by FGF-1. This makes it difficult to distinguish pathways activated by FGFR-1 from those activated by FGFR-3. However, a variant PC12 cell line, termed fnr-PC12, exists that has lost the expression of functional FGFR-1 but has retained functional FGFR-3 (17,18). The use of these cells allowed us to assay the ability of FGFR-3 to activate several distinct signal transduction pathways that are important for the development of the neuronal phenotype. In this paper we report that although FGFR-3 is not capable of inducing sustained activation of Ras-dependent pathways, it is as capable of inducing the activation of Ras-independent pathways to levels equivalent to those seen with the activation of FGFR-1.

MATERIALS AND METHODS
Cell Culture-PC12, 17N-2 PC12, fnr-PC12 cells, and fnr-PC12-derived transfectant lines have been described previously (17,18). Cells were grown on tissue culture dishes in Dulbecco's modified Eagle's medium supplemented with 10% donor horse serum, 5% fetal bovine serum, and 1% penicillin/streptomycin in an atmosphere of 10% CO 2 at 37°C. Recombinant human NGF was used at a final concentration of 50 -100 ng/ml. FGF-1 (a kind gift of Dr. M. Jaye) was added at the final concentration of 50 -100 ng/ml together with heparin (50 g/ml). The Src family member tyrosine kinase inhibitor PP2 was purchased from Calbiochem.
Western Blot Analysis-Cells were lysed in lysis buffer (20 mM Tris-HCl, pH 7.6, 150 mM NaCl, 50 mM NaF, 1 mM EDTA, 1 mM Na 3 VO 4 , 1 mM phenylmethylsulfonyl fluoride, and 1% Nonidet P-40). To dissolve Nonidet P-40 -insoluble peripherin protein, cells were lysed in RIPA buffer (20 mM Tris, pH 7.6, 135 mM NaCl, 2 mM EDTA, 0.1% SDS, 0.5% sodium deoxycholate, 1% Triton X-100, 1 mM Na 3 VO 4, and 1 mM phenylmethylsulfonyl fluoride). The lysates were clarified by centrifugation. The protein concentration was determined using Coomassie Plus protein assay reagent (Pierce). Aliquots of supernatant (100 g) of each sample were subjected to SDS-polyacrylamide gel electrophoresis on a 7.5% polyacrylamide gel and transferred onto a nitrocellulose membrane using a Bio-Rad Trans Blot according to the manufacturer's instructions. After incubation for 1 h with blocking solution (5% nonfat dry milk in PBS), the protein blots were probed with primary antibody for 2 h at room temperature or overnight at 4°C. For specific detection of Thy-1 and peripherin proteins, culture medium of hybridoma clones producing anti-Thy-1 antibody (a kind gift of Dr. J Trimmer) and anti-intermediate filament protein antibody (19) were used after dilution in blocking solution. The antibody against phospho-Akt (New England Biolabs) was used according to the manufacturer's instructions. The blots were probed with anti-mouse IgG horseradish peroxidaseconjugated antibody (Amersham Pharmacia Biotech) for 1 h at room temperature. The blots were treated for 1 min using ECL kit (Amersham Pharmacia Biotech) and exposed to x-ray film (Kodak).
Northern Blot Analysis-Isolation of total cellular RNA from PC12 cells or transfectant lines and Northern blot analysis were carried out as described previously (20). The concentration of RNA was determined by measuring the optical density at 260 nm. RNA samples (10 g/lane) were electrophoresed through agarose gels containing 2.2 M formaldehyde, 40 mM MOPS, pH 7, 10 mM sodium acetate, and 1 mM EDTA and then electrophorectically transferred to nylon membranes (Duralon-UV; Stratagene) at 200 mAmp at 4°C overnight. The blots were crosslinked using UV cross-linker (Stratagene). Blots were incubated in prehybridization buffer (5ϫ SSC, 10ϫ Denhardt's solution, 50 mM sodium phosphate, pH 6.7, 50% formamide, 0.5% SDS, and 0.5 mg of denatured salmon sperm DNA/ml (31) for at least 2 h at 68°C. The prehybridization solution was completely removed and replaced with hybridization buffer (5ϫ SSC, 1ϫ Denhardt's solution, 20 mM sodium phosphate, 50% formamide, 0.5% SDS, and 0.1 mg of denatured salmon sperm DNA/ml) supplemented with 5 ϫ 10 6 cpm/ml [␣-32 P]UTP-labeled antisense RNA probes. Hybridization was carried out overnight at 68°C. RNA probes were synthesized using the following linearized cDNA clones as templates: RB211 (21) encoding a conserved sodium channel coding region and pIB15 (22) encoding cyclophilin subcloned into pSP65; pG7TR1, encoding transin (23) subcloned into pGEM7Zf (ϩ); pPeripherin, encoding a partial sequence of rat peripherin (nucleic acid 1159 -1503), which was isolated by polymerase chain reaction from a PC12 dT/random cDNA library and subcloned into pBluescript II SK(ϩ) plasmid; pST4, encoding Thy-1 (24) subcloned into pSP65. All riboprobes were generated according to the manufacturer's instruction using SP6 polymerase (Promega) except for peripherin probes, which was synthesized using T7 polymerase. The blots were washed twice in 2ϫ SSC-0.1% SDS at 68°C and twice in 0.2% SSC, 0.1% SDS for 20 min at 68°C. Levels of radiolabeled probe bound to the blot were determined by PhosphorImager (Molecular Dynamics) analysis, and all values were normalized to the level corresponding to cyclophilin mRNA. Immunocytochemistry-Cells were plated on coverslips coated overnight with 25 g/ml poly-L-lysine (Sigma) and 10 g/ml laminin (Collaborative Biomedical Products), grown for 24 h, and treated with 50 ng/ml of FGF-1 for an additional 3 days. Cultures were fixed in 3.7% formaldehyde and 0.12 M sucrose in PBS for 10 min and then rinsed with PBS, permeabilized in methanol at Ϫ20°C for 15 min, and then blocked with 15% goat serum in PBS for 1 h at room temperature. Cells were incubated with affinity-purified anti-sodium channel protein antibody (25) that was diluted in blocking solution overnight at room temperature. After rinsing twice with blocking solution, cells were incubated with goat anti-rabbit antibody conjugated with Alexa 568 (Molecular Probes, Inc.) diluted in blocking solution for 1 h at room temperature. Cultures were rinsed twice with PBS and mounted using Vectashield mounting medium (Vector Laboratories). Confocal images were obtained using a Zeis LSM510 laser scanning confocal microscope.
Sodium Current Recordings-Whole cell recordings of PC12 sodium current were made by means of an Axopatch 200A amplifier (Axon Instruments Inc., Burlingame, CA). The recording solution contained 140 mM NaMES, 0.2 mM MgCl 2 , 0.2 mM CaCl 2 , 10 mM NaHEPES, pH 7.2. The pipette solution contained 140 mm CsMES, 10 mM CsEGTA, 10 mM CsHEPES, pH 7.2. To record sodium currents the cells were held at Ϫ100mV and stepped to positive potentials for 20 ms. PC12 cell sodium currents were digitized at 50 kHz and analyzed off-line using HEKA Pulse & Pulse Fit software (Instrutech, Great Neck, NY). Capacitive transients were compensated using a combination of manual compensation on the amplifier and further processing using either a P/4 or P/10 leak subtraction protocol.
PC12 Survival Assay-Cells were plated on 24-well plates precoated with rat tail collagen (Collaborative Research). Prior to plating the cells were washed five times with serum-free medium and then resuspended at a concentration of 2 ϫ 10 5 /ml in serum-free medium. 0.5 ml of cells were plated into each well and were appropriate FGF-1 (50 ng/ml) or NGF (100 ng/ml) was added, cultures were done in triplicate. The cultures were refed with factor containing medium every 2 days, After 8 days of treatment, the cell both in the medium and attached to the plate were lysed in 100 l of 0.1ϫ PBS, 0.5% Triton X-100, 2 mM MgCl 2 , 0.5% ethylhexadecyldimethylammonium bromide, 0.28% glacial acetic acid, 2.8 mM NaCl, and bromphenol blue. The nuclei were counted with a hemocytometer, but only nuclei with a distinct membrane and nucleoli representative of live cells were counted.

RESULTS
In this study we wanted to compare the abilities of FGFR3 to activate various signal transduction pathways in PC12 cells. We were particularly interested in comparing pathways that had been shown previously to be either activated in a Ras-de-pendent or Ras-independent manner. These studies were done using NGF and a PC12 cell line, 17N-2 PC12 cells, that constitutively expresses a dominant-interfering mutant of Ras (20). NGF was shown to activate the expression of transin in a Ras-dependent manner, whereas the type II sodium channel and peripherin were activated in Ras-independent manner. Therefore, to confirm that FGF-1 activated the expression of these genes in a similar way, we used the 17N-2 PC12 cells to determine what effect this dominant-interfering mutant of Ras has on the activation of these genes. Fig. 1A shows a Northern blot analysis that demonstrates that whereas the sodium channel gene expression is still induced by addition of FGF-1, the levels of transin are unchanged. The induction of the sodium channel in these cells by FGF-1 was reproducible even though it was low (Fig. 1A). Sodium channel induction in these cells by NGF was also found previously to be low (20). The induced expression of the peripherin protein was also not affected by the N17-ras mutant (Fig. 1B). This indicates that just like NGF, FGF-1 induces transin via a Ras-dependent pathway and peripherin and the sodium channel via Ras-independent pathways. FGF-1 induction of Thy-1 expression was also found to be Ras-independent and to require the activation of a Src-dependent pathway in a similar manner to NGF activation (data not shown and Fig. 6) Having determined that FGF-1 activation proceeds through both Ras-dependent and Ras-independent pathways we now wanted to determine the abilities of FGFR-1 and FGFR-3 to activate these pathways. To assay for the ability of FGFR-3 to activate different signaling pathways, we compared signals induced by FGF-1 in three different cell lines. Fnr-PC12 cells are the parental cell line that does not express FGFR-1 but does express low levels of FGFR-3. FGFR-3b cells are fnr-PC12 cells that have been transfected with FGFR-3, and FGFR-31b cells are fnr-PC12 cells transfected with a chimeric FGF receptor, composed of the extracellular domain of FGFR-3 fused to the cytoplasmic domain of FGFR-1. The use of this chimeric receptor allowed us to eliminate differences in signaling that could be attributed to FGF-1 binding because both of the transfected receptors have the same FGF-binding domains. We have shown previously that these two transfected cell lines overexpress equivalent amounts of receptors (17). The use of these cells eliminates the possibility that any differences in signaling can be attributed to receptor number. By assaying for FGF-1induced signals in the fnr-PC12 cells, we will be able to identify signaling pathways that FGFR-3 can activate. Then, by comparing FGF-1 induced signals in the two transfected cell lines, we can compare the efficiency with which FGFR-1 and FGFR-3 can activate these pathways.
We chose to look primarily at the induction of gene expression because our previous analysis had revealed that the initiation of signaling by FGFR-1 and -3 was similar (17,18). However, the downstream consequences that require robust gene expression were dramatically different. To initiate these studies, we chose to look at the induction of the gene transin by FGF-1. Transin is induced via a Ras-dependent signaling pathway (20). Therefore, its induction would serve as a control to establish that using induction of a Ras-dependent gene l would also reveal the differing signaling capabilities of FGFR-1 and -3. Fig. 2 shows analysis of the induction of transin mRNA using Northern blot analysis. In the fnr-PC12 cells there was a barely detectable induction of transin mRNA by FGF-1. In contrast, treatment of the fnr-PC12 cells with NGF did induce transin mRNA expression (data not shown), indicating that the signaling pathway to transin mRNA induction is intact in these cells. In the FGFR-3b cell line that overexpresses FGFR-3, transin mRNA induction was now detected after 72 h of treatment with FGF-1. However, analysis of the FGFR-31b cell line revealed that this receptor could induce much higher levels of transin mRNA after 72 h (Fig. 2). Quantitation of the amount of mRNA present under the different conditions demonstrated that the level of induction of transin mRNA at the 72 h time point was ϳ15-fold higher in the cells that expressed the FGFR-31b chimeric receptor in comparison with the FGFR-3b cells. As a loading control the levels of cyclophilin mRNA was also measured, and this demonstrated that similar amounts of RNA were loaded in each lane (Fig. 2). Together with the lack of induction seen in the fnr-PC12 cells (Fig. 3), these results indicate that FGFR-3 is much less efficient than FGFR-1 at inducing this Ras-dependent pathway. These observations confirm our previous results and validate this approach to show potential differences between signaling by FGFR-1 and -3.
The ability of FGFR-3 to activate pathways that are known to be Ras-independent was examined next. The activation of the Type II sodium channel gene has been shown to occur through a pathway that is independent of Ras. Fig. 3 shows the

FIG. 1. FGF-1 induced signal transduction in PC12 cells expressing a dominant-interfering mutant of Ras.
A, 17N-2 PC12 cells were treated with 50 ng/ml FGF-1 for 48 h. Total cellular RNA (10 g) was electrophoresed through 0.8% agarose gels and transferred onto a nylon membrane. The blot was hybridized with a probe specific for type II sodium channel (top panel) or a probe specific for transin (middle panel) and with a probe specific for the internal control cyclophilin (bottom panel). B, 17N-2 PC12 cell lysates treated for 0, 24, or 72 h with FGF-1 were electrophoresed through a 7.5% SDS-polyacrylamide gel and transferred onto a nitrocellulose filter membrane. Blots were probed with anti-intermediate filament protein antibody and exposed to x-ray film as described under "Experimental Procedures. "   FIG. 2. Transin induction mediated by FGFR-1 and FGFR-3 receptors. Cells from the Fnr-PC12, FGFR-31b, or FGFR-3b cell lines were treated with 50 ng/ml of FGF-1 for 72 h. Total cellular RNA (10 g) was electrophoresed through 0.8% agarose gels and transferred onto a nylon membrane. The blot was hybridized with a probe specific for transin (upper panel) and with a probe specific for the internal control (CON) cyclophilin (lower panel).
induction of mRNA encoding the type II sodium channel as measured by Northern blot analysis. Analysis of the induction of the Type II sodium channel in the fnr-PC12 cells after 60 h of treatment by FGF-1 demonstrates that FGF-1 could induce the increased expression of the type II sodium channel. As a control we also measured transin induction after 60 h of treatment with FGF-1 (Fig. 3). The data indicated that the endogenous FGFR-3 in the fnr-PC12 cells can induce type II sodium channel, albeit weakly. We compared the abilities of the overexpressed receptors to induced type II sodium channel mRNA. As can be seen in Fig. 3, FGFR-3 is as efficient as FGFR-1 in inducing the increased expression of type II sodium channel mRNA, whereas there is a major difference between these receptors in their abilities to induce the Ras-dependent transin mRNA.
We also measured the induction of functional channel expression by these receptors and the expression of the channels within the cells. Fig. 4A documents that the increase in mRNA levels of the type II sodium channel correlates well with an increase in channel protein levels. Immunofluorescence localization studies revealed a staining pattern indicative of localization of the channel proteins to the surface of the differentiated PC12 cells. To determine that the immunofluorescence corresponded to functional sodium channels, whole cell patch clamp recordings were performed. Recordings from FGFtreated FGFR-3b cells indicated large inward sodium currents when the cells were depolarized to positive membrane potentials (Fig. 4B). In recordings from eight FGF-treated FGFR-3b cells, all exhibited inward current, the overall average corresponding to 357pA. By contrast, in recordings from 11 control cells, only 4 exhibited inward sodium current. All sodium current in treated and nontreated cells was inhibited by addition of 1 M tetrodotoxin, an inhibitor of voltage-dependent sodium channels.
We next looked at the induction of the protein peripherin. The expression of this protein is controlled through another Ras-independent pathway that in this case involves phospholipase C␥ (PLC␥) activation (26). 2 Analysis of mRNA levels in the fnr-PC12 cells by Northern blot demonstrated that both FGF-1 and NGF induced peripherin to similar levels (data not shown). Comparison of the induction of peripherin in the cells overexpressing the two FGFRs showed that in both cases there was induction of mRNA to similar levels (Fig. 5A). Quantitation of the levels of induction indicated that FGFR-3b activation gave rise to a 5-fold increase in mRNA levels after 72 h, whereas the FGFR31b receptor induction was 4-fold. This indicates that FGFR-3 can activate this pathway as efficiently as FGFR-1. We also looked at the induction of the expression of peripherin at the protein level by Western blotting. Fig. 5B shows that FGF-1 can induce peripherin in the fnr-PC12 cells and that in the cells overexpressing the FGFR-3 receptor the induction of peripherin levels is increased significantly. This 2 D.-Y. Choi and S. Halegoua, manuscript in preparation.

FIG. 3. Type II sodium channel gene induction mediated by FGFR-1 and FGFR-3 receptors.
Cells from the Fnr-PC12, FGFR-31b, and FGFR-3b cell lines were treated with 50 ng/ml of FGF-1 for 60 h. Total cellular RNA (10 g) was isolated and electrophoresed through 0.8% agarose gels and transferred onto nylon membrane. Blot was hybridized with a probe specific for sodium channel mRNAs (pRB211) and with a probe specific for transin (upper panel) or with a probe specific for the internal control (CON) cyclophilin (lower panel).

FIG. 4. Induced expression of functional sodium channels by the FGFR-3 receptor.
A, FGFR-31b cells were treated with 50 ng/ml of FGF for 3 days. Cells were fixed and stained for sodium channel protein using affinity-purified pan-sodium channel protein antibody, and a projection image was acquired as described under "Experimental Procedures." B, whole cell voltage clamp recordings from a single FGF treated (top traces) and a nontreated FGFR-3b cell (bottom traces). The cell membrane potential was stepped from Ϫ100 mV to levels between Ϫ40 and Ϫ10 mV in 10 mV depolarizing increments for 20 ms. A family of inward current traces is observed for FGF treated cells as compared with the nontreated cells, which show no inward sodium current. The membrane voltage is shown below the two current sets.
indicates that FGFR-1 and FGFR-3 are equally efficient at the induction of this PLC␥-dependent pathway.
The Thy-1 protein is a cell surface glycoprotein whose expression is induced by FGF-1 treatment of PC12 cells via another distinct Ras-independent pathway that involves a Srcdependent branch point (Ref. 20 and below). Analysis of the induction of mRNA encoding for Thy-1 in the fnrPC12 cells revealed that at the 72-h time point FGF-1 could induce a 3-fold increase in Thy-1 mRNA levels (Fig. 6A), and this was a similar level to that seen with NGF (data not shown). A time course of the induction of the expression of Thy-1 mRNA in the FGFR-31b and FGFR-3b overexpressing fnrPC12 cells revealed that both of these receptors induced equivalent levels of Thy-1 mRNA (Fig. 6A). By 72 h FGFR31b had induced a 7-fold increase and FGFR3b had induced a 10-fold increase. Analysis of Thy-1 protein levels in the fnrPC12 cells and the FGFR-3b overexpressing cells also demonstrated that FGF-1 could induce Thy-1 protein expression and that this was greatly increased in the cells expressing more FGFR-3b (Fig. 6B). These data demonstrate that FGFR-3b can induce Thy-1 expression with similar efficiencies to FGFR-31b. To demonstrate that the induction of Thy-1 by FGFR-3b involves a Src family member we used the inhibitor, PP2. This class of inhibitor preferentially inhibits Src family member tyrosine kinases (27) and at the concentration used has very little effect on FGFR-3b kinase activation (data not shown). Fnr-Pc12 cells expressing FGFR-3b were activated by FGF-1 either in the presence or absence of PP2 and the induction of Thy-1 monitored after 48 h by Western blotting. As can be seen in Fig. 6C, in comparison with no treatment or 10 min treatment Thy-1 protein levels after 48 h incubation were increased by the addition of FGF-1 (lane 3). However, if the cells were treated with PP2 and stimulated with FGF-1, there was no induction of Thy-1 (lane 5). Thus, treatment of the cells with PP2 completely blocks the induction of Thy-1, demonstrating that the induction is via a Src family member-dependent pathway. These data are in agreement with earlier studies that indicated that Thy-1 induction by NGF was via a Src-dependent branch point (7).
Finally, we tested the ability of FGFR-3 to inhibit cell death following withdrawal of serum. As shown in Fig. 7A, withdrawal of serum from fnr-PC12 cells leads to cell death. Stimulation of the cells by NGF allows cell survival. When the cells were stimulated by FGF-1 there was some increase in cell survival, but it was not as efficient as NGF. This indicates that FGFR-3 can activate cell survival pathways but not as efficiently as the NGF receptor. To compare the abilities of FGFR-1 and FGFR-3 cytoplasmic domains to allow cell survival, we compare the fnr-PC12 cells overexpressing these two receptors. As shown in Fig. 7A these two receptors are equally able to induce cell survival when overexpressed. Similar results were also seen when cell survival was measured using annex-inV staining (data not shown). Although it is not clear which pathways are important for regulating this effect, the activation of protein kinase B (also known as Akt), via the PI-3 kinase pathway has been shown to inhibit apoptosis in similar systems. Activation of Akt requires phosphorylation, and there are antibodies available that are directed against the phosphoryl- Cell lysates were electrophoresed through a 7.5% SDS-polyacrylamide gel and transferred onto a nitrocellulose filter membrane. Blots were probed with anti-intermediate filament protein antibody and exposed to x-ray film as described under "Experimental Procedures." CON, control.

FIG. 6. Thy-1 gene induction mediated by FGFR-1 and FGFR-3 receptors.
A, Fnr-PC12, FGFR-31b, and FGFR-3b cells were treated with 50 ng/ml of FGF-1 for the indicated times. Total cellular RNA (10 g) was electrophoresed through a 1% agarose gel and transferred onto a nylon membrane. The blot was hybridized with an antisense RNA probe specific for Thy 1.2 (upper panel) or with a probe specific for the internal control cyclophilin (lower panel). B, Fnr-PC12 and FGFR-3b cells were treated with 50 ng/ml of FGF-1 for 72 h. Cell lysates were electrophoresed through a 7.5% SDS-polyacrylamide gel and transferred onto a nitrocellulose filter membrane. The blot was probed with anti-Thy 1 antibody and exposed to x-ray film as described under "Experimental Procedures." C, FGFR-31b expressing fnr-PC12 cells were either untreated (lane 1) or treated with 100 ng/ml of FGF-1 for either 10 min (lane 2) or 48 h (lanes 3 and 5). The kinase inhibitor PP2 5 g/ml was added to the cells for 48 h either alone (lane 4) or together with FGF-1 (lane 5). Western blotting for Thy-1 was performed as above in B. CON, control. ated and activated Akt. Therefore, we treated fnr-PC12 cells expressing FGFR-1 and FGFR-3 with FGF-1 and performed a Western blot analysis using the antibodies that detected activated Akt. As can be seen in Fig. 7B, FGFR-1 and FGFR-3 induced identical levels of activation of Akt (compare lanes 3 and 6). This may explain how these two receptors can both inhibit cell death in PC12 cells. DISCUSSION The cytoplasmic signaling domains of FGF receptors are highly conserved; yet data are accumulating that they do not all signal equivalently. A comparison of the ability of FGFR-1 and FGFR-3 to induce proliferation of the lymphoid cell line BaF3 or neurite outgrowth in PC12 cells indicated that only FGFR-1 was able to mediate these things efficiently (14 -18). In these cell systems it appears that sustained and robust signals that lead to the Ras-dependent activation of the extracellular regulated kinases are necessary for the biological phenotype analyzed. This implied that FGFR-3 was not able to induce strong sustained signals. However, in vivo it obviously plays key roles in normal development and in several diseases. This indicated that either the threshold for signaling in these developmental tissues is low so that sustained signals are not mandatory or that FGFR-3 can induce physiologically relevant and strong signals using other signaling pathways. To determine whether the latter possibility was correct, we took advantage of the PC12 cell system in which the elaboration of a neuronal phenotype can be induced by FGF-1. This growth factor mediates changes in gene expression that underlie the induction of various neuronal phenotypic changes through both Ras-dependent and Ras-independent pathways. In this report we demonstrate that the FGFR-3 can in fact induce Ras-independent signaling pathways as efficiently as FGFR-1, and this induction leads to important neuronal traits. Thus FGFR-3 is capable of strong physiologically important signaling via Rasindependent pathways.
Previous analysis of activation of Ras-dependent pathways in PC12 cells indicated that FGFR-3 was unable to induce efficient gene expression through the Ras-dependent pathways (17,18). Analysis of the ability of FGFR-3 to induce the Rasdependent gene transin in this report confirms that FGFR-3 is very poor at inducing Ras-dependent genes. However, as we demonstrate in this report, FGFR-3 was clearly capable of inducing all the Ras-independent pathways that we analyzed equally as well as FGFR-1. These included the induction of the expression of the protein peripherin. The expression of this gene is induced through a pathway that requires PLC␥. The activation of PLC␥ is accomplished via interaction with a tyrosine autophosphorylation site within the carboxyl terminus of the FGFRs. This site and its surrounding amino acid sequence are conserved between FGFR-1 and FGFR-3; therefore, it is perhaps not so surprising that both the receptors can induce the expression of genes via PLC␥-dependent pathways. Recent experiments using chimeric receptors that could be activated by platelet-derived growth factor also demonstrated that PLC␥ was equally well phosphorylated by both FGFR-1 and -3 (28). These data agree with ours that the difference in signaling capabilities between these two receptors cannot simply be explained by differences in kinase activities.
The two other genes we analyzed for induction by FGFR-3, namely the type II sodium channel and Thy-1, are known to be induced via Ras-independent pathways. We found that FGFR-3 was as good as FGFR-1 in its ability to induce these two genes. The activation of cyclic AMP-dependent pathways has been implicated in sodium channel induction. Similarly, Thy-1 is activated in a Ras-independent manner by NGF, and this involves a novel branch point off a Src-dependent pathway. Our studies using the Src family member kinase inhibitor PP2 indicate that FGF-1 activates Thy-1 expression via a Src-dependent branch point. We also compared the abilities of FGFR-1 and FGFR-3 to inhibit cell death and to activate the protein tyrosine kinase Akt. Again we found that FGFR-3 was equally efficient as FGFR-1 in inhibiting cell death and activation Akt. Therefore, in clear contrast to its inability to activate Ras-dependent pathways, it is clear that FGFR-3 can activate all of the Ras-independent pathways we analyzed equally as well as FGFR-1.
In our analysis we did not identify an FGFR-3-specific signal transduction pathway, in that FGFR-1 was found to also be capable of activating all of the pathways we analyzed. Recently, FGFR-3 has been shown to be able to activate the protein Stat-1, and this is a candidate pathway that may be specific to FGFR-3 among the FGFR family (29,30). This activation appeared to be cell type-specific because it was only clearly shown in chondrocytes. PC12 cells express high levels of Stat-1; however, we and others were unable to demonstrate activation of Stat-1 by either FGFR-1 or FGFR-3 in the overexpressing cells (30). 3 1 and 4) or were incubated in serum-free medium for 24 h either in the absence (lanes 2 and 5) or the presence of 100 ng/ml FGF-1 (lanes 3 and 6). Cell lysates were analyzed by Western blotting for the presence of activated Akt using phospho-Akt specific antibodies. may involve cell type-specific adaptor proteins that are absent from PC12 cells. Thus, signaling through FGFR-3, in addition to occurring primarily via Ras-independent pathways, may also involve the use of specific proteins to expand the repertoire of the signaling cascades. The unique ability to signal primarily through Ras-independent pathways may explain why activating mutations in FGFR-3 only have penetrance in certain tissues. This may involve the activation of specific pathways that are not easily duplicated by other receptors and may also involve additional cellular proteins that are only expressed in the tissues affected. The unique signaling ability may also explain why only FGFR-3 of the four family members is overexpressed in multiple myeloma cells.
In summary, we have demonstrated that FGFR-3 can induce the equivalent activation of various Ras-independent pathways to FGFR-1. In contrast, as shown previously FGFR-3 could not induce sustained robust activation of Ras-dependent pathways. This indicates that FGFR-3 induces signal transduction primarily through strong activation of Ras-independent pathways in the absence of a robust Ras-dependent signal.