BMP-2 augments FGF-induced differentiation of PC12 cells through upregulation of FGF receptor-1 expression

Polypeptide growth factors are involved in numerous developmental processes in the nervous system, including axon growth and neuronal survival (Baird, 1994; Bikfalvi et al., 1997; Chiu, 1989). These factors include a family of fibroblast growth factors (FGFs), comprising over 20 structurally related proteins. FGF family members are expressed in specific spatio-temporal patterns and play important roles in neural development, as well as in the development of other tissues (Emoto et al., 1989; Ozawa et al., 1996; Wilcox and Unnerstall, 1991). Like nerve growth factor (NGF), FGF-1 (acidic FGF) and FGF-2 (basic FGF) trigger the differentiation of PC12 cells into sympathetic neuron-like cells that are characterized by electrical excitability, expression of neuron-specific genes and neurite outgrowth (Dichter et al., 1977; Greenberg et al., 1985). The activity of FGFs is initially mediated by their binding to FGF receptor (FGFR) tyrosine kinases, which are encoded by a family of four genes: FGFR-1, FGFR-2, FGFR-3 and FGFR-4 (Jaye et al., 1992; Johnson and Williams, 1993; Partane et al., 1992). Expression patterns of FGFRs in the nervous systems suggest that the effects of FGFs differ, depending on cell type, throughout development. For instance, FGF induces neurite outgrowth in PC12 cells via FGFR-1, but not via the other FGFRs expressed in these cells (Lin et al., 1996). Activated FGFR-1 appears to recruit FRS2 (Kouhara et al., 1997), src, ras, raf and extracellular signal-related kinase (ERK), all of which are involved in neurite induction; in particular, activation of ERK1/2 has been shown to be necessary for induction of PC12 differentiation by neurotrophic factors (Cowley et al., 1994; Kouhara et al., 1997; Marshall, 1995; Traverse et al., 1994).

Bone morphogenetic proteins (BMPs) are members of the TGF-β superfamily, which mediate multiple biological processes, including neuronal development. BMPs transduce their signals into the nucleus by binding to two different serine/threonine kinase receptors (type I and type II). Upon activation, BMP receptors recruit and phosphorylate several smad transcription factors (smad1, smad5 or smad8), which then translocate into the nucleus to regulate gene expression (Derynck et al., 1998; Heldin et al., 1997). BMP ligands and receptors are expressed throughout neuronal development and act on lineage-restricted, embryonic central nervous system progenitor cells to promote regionally restricted neuronal survival, differentiation and apoptosis (Mehler et al., 1997). Iwasaki et al. reported that BMP-2 induces neuronal differentiation of PC12 cells (Iwasaki et al., 1996) and that, in contrast to FGF-induced differentiation, the process is mediated by activation of p38/MAPK (mitogen-activated protein kinase) but not ERK1/2 (Iwasaki et al., 1999).

In vivo, multiple growth factors act synergistically or antagonistically toward their respective target cells. Such joint regulation can also be demonstrated in vitro; for instance, BMP-2 induced expression of TrkC receptor and its ligand, neurotrophin (NT)-3, in sympathetic neurons derived from rat superior cervical ganglia, thereby enhancing NT-3-evoked survival responses (Kobayashi et al., 1998). In the present study, we examine possible cooperation between BMP-2 and FGF in the regulation of PC12 neural differentiation. We show that although BMP-2-induced neurite outgrowth in PC12 cells was not as strong as was previously reported (Iwasaki et al., 1996; Iwasaki et al., 1999), BMP-2 induced FGFR-1 expression, thereby augmenting FGF-induced neurite outgrowth.

Rat pheochromocytoma PC12 cells were obtained from the Riken Gene Bank (Tsukuba, Japan). Recombinant human BMP-2 was kindly provided by Yamanouchi Pharmaceutical Co., Ltd (Tsukuba, Japan). The pGL3ti(SBE)4-luciferase plasmid (Jonk et al., 1998) was a generous gift from Dr W. Kruijer (Groningen Biomolecular Sciences and Biotechnology Institute, Netherlands). ¹²⁵I-labeled FGF-2 was purchased from New England Nuclear (Boston, MA, USA). Bovine FGF-1, FGF-2 and NGF were from R&D Systems (Minneapolis, MN, USA). Heparin was from Sigma (St Louis, MO, USA). Anti-smad1 antibody was from Santa Cruz Biotechnology (Santa Cruz, CA, USA). Anti-phospho-smad1 antibody was from Upstate Biotechnology (Lake Placid, NY, USA). Anti-FGFR-1 antibody was from Santa Cruz Biotechnology (Santa Cruz, CA, USA). Anti-ERK-1/2 and anti-phospho-ERK1/2 antibodies were from Promega (Madison, WI, USA) and NEB (Beverly, MA, USA), respectively. WGA-agarose was from Seikagaku Co., Ltd (Osaka, Japan). SU5402, a specific FGFR–1 inhibitor, was from Calbiochem (Darmstadt, Germany).

PC12 cells were maintained in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% fetal calf serum and 5% horse serum. To assess their differentiation, the cells were plated to a density of 5×10³cells/well on collagen type IV-coated, 24-well culture plates (Becton Dickinson), and maintained for 24 hours, after which they were starved for 24 hours in serum-free medium (DMEM containing 2 mg/ml bovine serum albumin, 1 μg/ml insulin, 2 μg/ml transferrin, 30 nM Na₂SeO₃, 20 nM progesterone, and 10 mM Hepes, pH 7.4). The cultures were then initiated (day 0) in the presence of a single polypeptide growth factor for various times. For some experiments, such as in Figs 2 and 6, the starved cells were cultured in the serum-free medium with or without BMP-2. After 48 hours (day 2), the cells were refed with the same medium with or without BMP-2 and further cultured. After 24 hours (day 3), FGF or NGF was added to the medium and the cells were further cultured for various times. For the BMP-2 plus samples, BMP-2 was present throughout the culture period (days 0-5). Then, differentiation of the cells was examined under a phase-contrast microscope. Randomly selected fields, containing approximately 100 cells each, were photographed and the numbers of undifferentiated and differentiated cells counted. The experimental criterion for distinguishing differentiated from undifferentiated cells was neurite outgrowth: cells having neurites greater in length than two cell-body diameters were considered differentiated. Data are expressed as the means ± s.d. of four independent counts. To inhibit FGF-induced FGFR autophosphorylation, in some experiments, cells were pretreated for 5 minutes with SU5402 (Mohammadi et al., 1997), after which FGF was added for the indicated times.

For transient transfection of pGL3ti(SBE)4-luciferase (Jonk et al., 1998) and beta-galactosidase expression plasmid, cells were plated to a subconfluent density on 24-well culture plates. The next day, cells were transiently transfected with these plasmids using Lipofectamine Plus Reagent (Life Technologies) according to the manufacturer’s instructions. The cells were stimulated with 50 ng/ml BMP-2 for 48 hours and lysed, and luciferase activity was measured using a Luciferase Assay System (Promega). As an internal control, beta-galactosidase activity was measured using Beta-Galactosidase Enzyme Assay System (Promega) to normalize luciferase activity. Data are expressed as the means + s.d. of three independent determinations.

Semi-quantitative reverse transcriptase-polymerase chain reaction (RT-PCR) was essentially as described in our earlier reports (Ozawa et al., 1996; Ozawa et al., 1997). Briefly, total RNA was isolated using Isogen (Nippon Gene), a reagent containing phenol and guanidine thiocyanate, after which a 1 μg sample of total RNA was reverse-transcribed using Superscript II (Life Technologies) according to the respective manufacturer’s instructions. To evaluate the level of gene expression, cloned cDNAs having the desired sequences (1×10¹-1×10⁵ copies with tenfold serial increments for FGFR2, FGFR3 and FGFR4; 1×10³-1×10⁷ copies for FGFR1 and glyceraldehyde phosphate dehydrogenase, GAPDH) were used as quantitative controls. As a template, 5% of the reverse transcript (0.05 μg of total RNA) was amplified in 25 μl of reaction buffer, using AmpliTaq Gold (PE Applied Biosystems). For amplification of the respective FGFRs, oligomers were selected from the coding exons to avoid amplification of genomic DNA. The specific primers used for amplification were: FGFR1 (sense; 5′-ttc tgg gct gtg ctg gtc ac-3′, antisense; 5′-gcg aac ctt gta gcc tcc aa-3′); FGFR2 (sense; 5′-ttc atc tgc ctg gtc ttg gt-3′, antisense; 5′-aat aag gct cca gtg ctg gtt tc-3′); FGFR3 (sense; 5′-cta gtg ttc tgc gtg gcg gt-3′, antisense; 5′-ttc tta tcc att cgc tcc gg-3′); FGFR4 (sense; 5′-ctg ttg agc atc ttt cag gg-3′, antisense; 5′-cgt gga agg cct gtc cat cc-3′); and GAPDH (sense; 5′-ttc att gac ctc aac tac atg-3′, antisense; 5′-gtg gca gtg atg gca tgg ac-3′). For amplification of smad7, specific primers were: sense; 5′-cca tct tca tca agt ccg cca c-3′, antisense; 5′-cta ccg gct gtt gaa gat gac c-3′. In a preliminary experiment, the expression levels of the targeted mRNAs in PC12 cells were assessed, and conditions limiting amplification of the product to within the linear exponential range were determined (Table 1). Following PCR, samples of the product were run on 1.0% agarose gels, after which the resultant DNAs were denatured with 0.5 N NaOH and transferred onto Hybond N+ membranes (Amersham-Phermacia). The filters were then hybridized with the corresponding DNA probe labeled with digoxygenin-labeled dUTP using the random priming method (DIG-High Prime, Roche). The protocols for hybridization, post-hybridization, washing and detection were described previously (Ozawa et al., 1996; Ozawa et al., 1997). The intensity of each band on the image was measured, and the data were processed using NIH Image version 1.61 image processing software.

PC12 cells were plated to a density of 1.2×10⁶ cells per 60 mm collagen type IV-coated culture dish, cultured and starved for 24 hours each as described above, and then treated with BMP-2 for 3 days. Thereafter, the cells were washed twice with 5 ml of binding buffer (DMEM containing 10 mM Hepes, pH 7.3 and 1% BSA) and incubated for 90 minutes at room temperature in 2 ml of the binding buffer containing 15 ng/ml 125I-labeled FGF-2 and 5 μg/ml heparin, with or without a 200-fold excess of cold FGF-2. The cells were then washed twice with the 5 ml of binding buffer and incubated in binding buffer containing 1 mM of the cross-linker, disuccinimidyl suberate (Sigma) for 30 minutes, after which the crosslinking reaction was stopped by adding 0.2 ml of reaction-stop buffer (50 mM Tris-HCl, pH 7.3, 0.2 M glycine, 22 mM EDTA, 1 μg/ml aprotinin and 10 μM E-64). The cells were then lysed on ice with 0.5 ml of lysis buffer (50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1% NP-40), the resultant lysate centrifuged at 12,000 g for 30 minutes, and the supernatant precleared by adding 10 μg of protein A+G Sepharose beads (Calbiochem) and gently agitating on a rotator for 1 hour at 4°C. The precleared lysate was incubated overnight at 4°C with 5 μg of WGA-agarose or anti-FGFR1 polyclonal antibody, after which the immunocomplex was precipitated by addition of 10 μg of protein A+G Sepharose beads. The precipitate was washed 3 times with 500 μl of the lysis buffer and boiled in 3× Laemmli’s sample buffer before electrophoresis (7% SDS-PAGE). Autoradiography was then carried out using a Fuji BAS-2000 radioimage analyzer.

PC12 cells were plated to a density of 6×10⁵ cells/well in collagen type IV-coated 6-well plates, cultured and starved for 24 hours each (see above), and treated with BMP-2 for 3 days. The cells were then washed twice with 0.5 ml of binding buffer and incubated for 90 minutes at room temperature in 1 ml binding buffer containing 0.625, 1.25, 2.5, 5.0 or 10 ng/ml 125I-labeled FGF-2 plus 5 μg/ml heparin, with or without a 200-fold excess of cold FGF-2. They were then washed four times with 0.5 ml binding buffer followed by one wash with 0.5 ml binding buffer containing 5 μg/ml heparin, and incubated for 30 minutes in the binding buffer containing 1 mM of disuccinimidyl surerate. These cells were then lysed in 1 ml of 0.1 N NaOH for 30 minutes at room temperature, and the radioactivity in samples of lysate was determined using a gamma-counter (Beckman Coulter). Each data point represents the mean of duplicate determinations and was analyzed according to the method of Scatchard.

Growth factor-stimulated PC12 cells were collected and washed twice with 1 ml of ice-cold phosphate-buffered saline, and then lysed for 10 minutes on ice with 100 μl of lysis buffer and centrifuged. The protein concentration in the supernatant was determined using a Bio-Rad Protein Assay. Equal amounts of protein (40 μg for smad1 and phospho-smad1, 10 μg for ERK1/2 and phospho-ERK1/2) were separated by SDS-PAGE, electrophoretically transferred to nitrocellulose transfer membranes (Protran; Schleicher & Schuell), and probed with each antibody. Immunoreactive bands were visualized by enhanced chemiluminescence (ECL; Amersham-Pharmacia). For ERK1/2 phosphorylation assay, the intensity of each phospho-ERK1/2 signal was measured and normalized by the intensity of the corresponding ERK1/2 signal.

It has been reported that BMP-2 induces neurite outgrowth in PC12 cells (Iwasaki et al., 1996; Iwasaki et al., 1999). We first examined the activity using representative PC12 cells obtained from Riken Gene Bank, Japan (RCB0009). When cells bearing neurites over two cell bodies in length were counted, it was found that less than 10% of the cells differentiated after 4 days in the presence of BMP-2 (Fig. 1A). To examine whether the low response to BMP-2 was clonal-specific, we further examined three representative PC12 cells obtained from independent cell depositories (CRL-1721 of ATCC USA, PC12 cells of Dainippon Pharmaceutical Co., Japan; and a kind gift from Dr Simona Raffioni at the University of California, Irvine, USA). All these cells have been studied in many preceding studies. We found that all these PC12 cells differentiated very poorly by BMP-2 stimulation alone (data not shown), similar to the results shown in Fig. 1A. To further examine the possibility of clonal specificity, we performed single-cell cloning and isolated 12 independent clones from the PC12 cells we used in this study. None of the clones differentiated in response to BMP-2 (data not shown). Thus, our observation that differentiation of PC12 cells is not induced by BMP-2 is not specific to this particular clone of PC12 cells. The PC12 cells we used in Fig. 1A were used in all the following experiments.

That cells exposed to BMP-2 failed to differentiate was not indicative of a nonspecific inability to differentiate. When exposed to 50 ng/ml NGF, 60% of the cells were induced to differentiate within 4 days (Fig. 1C), whereas exposure to the same concentration of FGF-1 or FGF-2 induced approximately 40% of the cells to differentiate (Fig. 1B).

To confirm that intracellular BMP signaling was activated in the cells upon their treatment with BMP-2, the phosphorylation status of the smad1, activity of the smad binding element (SBE) in JunB promoter, and upregulation of smad7 mRNA levels, were examined. The smad1 is an intracellular mediator of BMP-2 signaling. As shown in Fig. 1D, treatment of the cells with BMP-2 for 1 hour induced phosphorylation of smad1. JunB gene is known to be transcribed by BMP-2 signaling. We used pGL3Ti (SBE)4-luciferase reporter plasmid, which contains four SBEs from JunB promoter that was successfully used to detect signaling induced by BMP-2, TGF-p and activin (Jonk et al., 1998). As shown in Fig. 1E, treatment of the cells with BMP-2 clearly enhanced the SBE activity. Finally, smad7 has been also described as a BMP-2-inducible gene (Ishisaki et al., 1999; Souchelnytskyi et al., 1998). We found that the smad7 mRNA level was indeed upregulated by stimulation of the cells with BMP-2 for 30 minutes and 60 minutes (Fig. 1F). All the results indicate that BMP-2 signaling was induced in the BMP-2-treated PC12 cells we used in this study.

Thus, the weak potential of BMP-2 to differentiate PC12 cells (Fig. 1A) was neither due to a nonspecific inability to differentiate nor to an absence of BMP-2 signaling in the cells.

To determine the extent to which BMP-2 affects the neurotrophic activity of FGF or NGF, PC12 cells were first exposed or not to 50 ng/ml BMP-2 for 3 days, then to selected concentrations of NGF or FGF-1. BMP-2 was present or absent in culture throughout the assay period. We found that cells pretreated with BMP-2 differentiated more vigorously in response to FGF-1 than cells not exposed to BMP-2 (Fig. 2A,E). This effect was particularly apparent at subthreshold concentrations of FGF-1 that, by themselves, did not induce significant differentiation (Fig. 2A,E). For example, when cells were exposed to 1 ng/ml or 5 ng/ml FGF-1 alone, little neurite outgrowth was induced (4% and 9% of cells differentiated by day 4, respectively; Fig. 2C). However, the synergistic effect of 50 ng/ml BMP-2 on FGF-1-induced neurite outgrowth was apparent within 1 day of the addition of FGF-1 (16% and 27% of cells differentiated, respectively; Fig. 2C,E). Similar results were obtained with the same concentrations of FGF-2, which alone induced differentiation of only 7% and 13% of cells by day 4, respectively, but induced differentiation of 24% and 30% of cells pretreated with BMP-2 (Figs 2D).

By using PC12 cells obtained from an additional three independent cell depositories, we reproducibly observed that their FGF-induced differentiation was augmentated by pretreatment with BMP-2, similar to the results shown in Fig. 2A,E (data not shown).

On the other hand, cell differentiation induced by NGF was not augmented by BMP-2 (Fig. 2B).

To investigate the mechanism by which BMP-2 augments FGF-induced differentiation of PC12 cells, we examined the expression of FGFR mRNA using the semi-quantitative RT-PCR/Southern blot method we established earlier (Ozawa et al., 1996; Ozawa et al., 1997). It was found that FGFR-1 mRNA was expressed to a greater degree (2.39×10⁵ copies/μg total RNA) in untreated PC12 cells (day 0) than FGFR-2, FGFR-3 or FGFR-4, which were expressed at 1.05×10⁴, 1.13×10² and 8.97×10¹ copies/μg total RNA, respectively (Fig. 3). Furthermore, BMP-2 markedly upregulated FGFR-1 gene transcription in a time- and concentration-dependent manner (Fig. 3B,C). While expression of FGFR-3 mRNA also increased, its absolute copy number was over 2,000 times less than that of FGFR-1 (Fig. 3B); expression FGFR-2 and FGFR-4 mRNA was unaffected by BMP-2 (Fig. 3).

We further confirmed that in all the additional PC12 cell stocks we examined, FGFR-1 gene expression was induced by BMP-2 stimulation (data not shown).

Also examined were the expression of mRNAs for NGF and NT receptors TrkA, TrkB and TrkC, which were found to be unaffected by BMP-2 (data not shown).

Binding experiments were carried out to confirm the BMP-2-induced upregulation of FGFR-1 at the level of the protein. PC12 cells were incubated with ¹²⁵I-labeled FGF-2, after which the ligand-receptor complex was cross-linked and isolated by cell lysis, precipitated using WGA-agarose beads (Imamura et al., 1988), and resolved by SDS-PAGE. Fig. 4A shows the 180-kDa receptor-ligand complex in samples from cells incubated with ¹²⁵I-FGF-2 in the absence of unlabeled FGF-2; BMP-2 treatment increased the signal for the receptor-ligand complex. Without WGA-agarose absorption, distinct signals were not obtained from untreated cells, due to the low level of expression (data not shown).

To examine further which type of FGFR was responsible for the increased receptor-ligand signal, the cross-linked material was immunoprecipitated using specific antibodies against the respective C-termini of the four FGFR types. We found that the ¹²⁵I-FGF-2-receptor complex immunoprecipitated with anti-FGFR-1 antibody, migrating to 180 kDa, and that pretreating the cells with BMP-2 increased levels of the complex (Fig. 4B). It appears from the molecular mass of the complex that the single band obtained with these cells represents the three immunoglobulin (IgG)-like domain isoform of FGFR-1 reported previously (Maher, 1999). Thus, like its mRNA, levels of FGFR-1 protein are upregulated by treating PC12 cells with BMP-2. ¹²⁵I-FGF-2-receptor complexes were not precipitated by antibodies against any of the other FGFR types (data not shown).

A Scatchard binding analysis revealed that the numbers of FGF-2 binding sites on PC12 cells were 5985 and 8152 sites/cell before and after BMP-2 treatment, respectively (Fig. 5). There was no significant change in the K_d (99.0 and 96.2 pM, respectively).

Earlier studies showed that sustained activation of ERK1/2 was important for neurite outgrowth in PC12 cells, and that receptor number determined the responsiveness of the cells to growth factors (Cowley et al., 1994; Marshall, 1995; Traverse et al., 1994). We found that BMP-2 alone did not induce phosphorylation of ERK1/2 in PC12 cells (data not shown). To ascertain whether the increased FGFR-1 expression leads to augmented phosphorylation of ERK1/2, we compared FGF-induced ERK1/2 phosphorylation in PC12 cells pretreated with BMP-2 and those not. We found that under either condition, ERK1/2 phosphorylation was strongly but transiently induced following FGF stimulation, with the level of phosphorylation reaching a maximum within 5-10 minutes and then declining to lower sustained levels (Fig. 6). However, both the transient peak and the sustained level of ERK1/2 phosphorylation were higher in BMP-2-treated cells, indicating that FGF signaling via ERK1/2 was augmented by BMP-2.

To further confirm that the enhanced FGF-induced differentiation of PC12 cells pretreated with BMP-2 was due to the augmentation of the FGFR-1 signaling cascade, the effects of an inhibitor of FGFR signaling was examined. When PC12 cells were pretreated with BMP-2 and then exposed to SU5402, a specific inhibitor of tyrosine phosphorylation of FGFR-1 (Mohammadi et al., 1997), the effect of BMP-2 on FGF-induced cell differentiation was completely suppressed (Fig. 7).

Our findings indicate that exposure to BMP-2 increases PC12 cell sensitivity to FGF-induced neurite formation; indeed following pretreatment with BMP-2, PC12 cells responded to FGF at concentrations that were not effective in the absence of BMP-2. As FGFR-1 was the most abundant FGFR on PC12 cells, and as expression of its mRNA and protein were both significantly enhanced by BMP-2 treatment, it is conceivable that augmented FGFR-1 expression is the mechanism responsible for the synergistic effect of BMP-2 on FGF-induced differentiation. The importance of FGFR-1 activity is further supported by the observation that the FGFR-1 inhibitor SU5402 abolished the induction of differentiation by subthreshold concentrations of FGF.

Although some PC12 cell clones express FGFR-1, FGFR-3 and FGFR-4 (Lin et al., 1996; Raffioni et al., 1999), induction of neurite outgrowth is mediated exclusively by FGFR-1 (Lin et al., 1996). This conclusion was confirmed by studies carried out using a PC12 clone expressing only FGFR-1 (Maher, 1999), or using fur-PC12 mutant cells, which lack functional FGFRs and are thus unresponsive to FGF; fur-PC12 mutants were differentiated by FGF following transfection with FGFR-1, confirming that FGFR-1 is essential for transduction of FGF activity leading to induction of neurites (Lin et al., 1996). The present findings are in agreement with these earlier observations with respect to both the expression levels of each FGFR family member and to the importance of FGFR-1 in FGF-induced neurite induction.

To examine if enhanced expression of FGFR-1 augments responsiveness of the PC12 cells to FGF, we transfected PC12 cells with FGFR-1 expression vector (a generous gift from Dr D. Ornitz, Washington University Medical School) and obtained stable transfectants. The FGFR-1 expression level in the transfectants was approximately twice of that in parental PC12 cells. Using these transfectants, it was found that FGF-1 alone at 1 ng/ml induced differentiation of approximately 20% of the cells (data not shown), while in parental PC12 cells the same extent of differentiation was observed only when FGF was given at much higher doses (Fig. 2A). In addition, in the FGFR-1 transfectants, augmentation of FGF-1 activity by BMP-2 was observed only when FGF-1 was used at doses lower than 5 ng/ml (data not shown). These results suggest that in the FGFR-1 transfectants, FGF signaling can be fully activated by FGF-1 at doses over 5 ng/ml. Thus, elevated expression level of FGFR-1 in the transfectants has provided the cells higher responsiveness to FGF-1 at its low concentrations. These results support the notion that enhanced expression of FGFR-1 in PC12 cells by BMP-2 as shown in Figs 3, 4 and 5 is responsible for the augmentation of FGF-induced differentiation of PC12 cells.

Peak FGF-induced ERK1/2 phosphorylation was enhanced in BMP-2-pretreated PC12 cells, as was the sustained level (Fig. 6). Evidence indicates that the duration of ERK1/2 activation is critical for induction of TRK-mediated events in PC12 cells (Marshall, 1995); more specifically, that sustained activation of ERK1/2 is necessary for efficient neurite outgrowth (Cowley et al., 1994; Marshall, 1995; Traverse et al., 1994); and that overexpression of TRKs enhances ligand-induced differentiation of PC12 cells (Traverse et al., 1994). It thus appears that BMP-2-induced upregulation of FGFR-1 expression, in turn, results in augmented FGF-evoked ERK1/2 MAPK activation, leading to enhanced neurite outgrowth in PC12 cells.

We do not exclude the possibility, however, that other mechanisms may also contribute to the synergistic effect of BMP-2 on FGF-induced differentiation. BMP-2-related intracellular signaling molecules such as smads, MKK3/6 and p38/MAPK (Iwasaki et al., 1999) might directly or indirectly modulate the activity of FGF-signaling intermediates such as FRS2, Grb2, Ras and Raf. This possibility needs to be addressed in future studies.

Earlier reports from one laboratory indicate that BMP-2 alone was able to induce differentiation of one PC12 clone (Iwasaki et al., 1996; Iwasaki et al., 1999). Although we extensively examined representative PC12 cells obtained from four cell depositories, none of them differentiated in response to BMP-2 alone, while all the cells differentiated when treated with FGF (data not shown). Furthermore, in all these cells, BMP-2 augmented FGF-induced differentiation and FGFR-1 expression (data not shown). Thus, our observation that differentiation is strongly induced by the combination of BMP-2 and FGF but not by BMP-2 alone is not specific to particular clone of PC12 cells. It is possible that the BMP-2-sensitive PC12 clone that was used in the earlier reports may have enhanced signaling ability compared to the representative PC12 cells.

Cell fates are regulated by complex signaling networks. Interestingly, FGF and BMP act antagonistically in many cases and synergistically in others. For example, FGF and BMP have opposing functions in the morphogenesis of the branching of the lung bud (Weaver et al., 2000), in limb outgrowth (Niswander and Martin, 1993), and in neural fates specified in utero (Wilson et al., 2000). By contrast, both FGF and BMP induce angiogenesis in the chick chorioallantoic membrane, and are synergistically deployed in binary combination in order to accentuate angiogenesis (Ramoshebi and Ripamonti, 2000). The mechanism underlying their joint regulation remains unknown, however. Our demonstration of FGFR-1 upregulation by BMP-2 in PC12 cells is thus the first evidence of the molecular mechanism by which these growth factors act in concert.

The mechanism by which BMP-2 augments FGFR-1 gene expression is a subject of particular interest to us. One possibility is suggested by the fact that the FGFR-1 gene contains Sp1- and AP-1-binding sequences in its promoter region (Saito et al., 1992) as does the smad7 gene (Brodin et al., 2000). The smad7 gene expression is upregulated by BMP-2 (Fig. 1E; Ishisaki et al., 1999; Souchelnytskyi et al., 2000) after interaction of receptor-regulated smads with an Sp1 family transcription factor (Pardali et al., 2000) and its cooperation with AP-1 (Brodin et al., 2000). Thus, expression of the FGFR-1 gene may be induced following the same regulatory pathway. Indeed, it was recently shown that the FGFR-3 promoter is regulated by Sp1 family transcription factors (McEwen and Ornitz, 1998).

We thank Drs Kazunori Nakamura at National Institute of Advanced Industrial Science and Technology and at Tsukuba University for supporting H. H. in his undergraduate study, and Tatsuji Nishihara at the Department of Oral Microbiology, Kyushu Dental College for his generous gift of BMP-2 protein and for discussion. We thank Dr W. Kruijer at the Groningen Biomolecular Sciences and Biotechnology Institute, Netherlands, and Dr C.-H. Heldin at the Ludwig Institute for Cancer Research, Sweden, for the generous gift of pGL3ti(SBE)4-luciferase plasmid.