Address correspondence and reprint requests to M. K. Stachowiak, Department of Pathology and Anatomical Sciences, State University of New York, 317 Farber Hall, Buffalo, NY 14214–3000, USA. E-mail: email@example.comThese authors contributed equally to this article.
The integrative nuclear FGFR1 signaling (INFS) pathway functions in association with cellular growth, differentiation, and regulation of gene expression, and is activated by diverse extracellular signals. Here we show that stimulation of angiotensin II (AII) receptors, depolarization, or activation protein kinase C (PKC) or adenylate cyclase all lead to nuclear accumulation of fibroblast growth factor 2 (FGF-2) and FGFR1, association of FGFR1 with splicing factor-rich domains, and activation of the tyrosine hydroxylase (TH) gene promoter in bovine adrenal medullary cells (BAMC). The up-regulation of endogenous TH protein or a transfected TH promoter-luciferase construct by AII, veratridine, or PMA (but not by forskolin) is abolished by transfection with a dominant negative FGFR1TK-mutant which localizes to the nucleus and plasma membrane, but not by extracellularly acting FGFR1 antagonists suramin and inositolhexakisphosphate (IP6). Mechanism of TH gene activation by FGF-2 and FGFR1 was further investigated in BAMC and human TE671 cultures. TH promoter was activated by co-transfected HMW FGF-2 (which is exclusively nuclear) but not by cytoplasmic FGF-1 or extracellular FGFs. Promoter transactivation by HMWFGF-2 was accompanied by an up-regulation of FGFR1 specifically in the cell nucleus and was prevented FGFR1(TK-) but not by IP6 or suramin. The TH promoter was also transactivated by co-transfected wild-type FGFR1, which localizes to both to the nucleus and the plasma membrane, and by an exclusively nuclear, soluble FGFR1(SP-/NLS) mutant with an inserted nuclear localization signal. Activation of the TH promoter by nuclear FGFR1 and FGF-2 was mediated through the cAMP-responsive element (CRE) and was associated with induction of CREB- and CBP/P-300-containing CRE complexes. We propose a new model for gene regulation in which nuclear FGFR1 acts as a mediator of CRE transactivation by AII, cell depolarization, and PKC.
sodium dodecyl sulfate polyacrylamide gel electrophoresis
The modulation of gene activity enables the cells of the nervous system to meet environmental challenges and to maintain homeostasis (Bailey and Kandel 1993). One such gene, whose activity is regulated by environmental signals, encodes tyrosine hydroxylase (TH), the rate-limiting enzyme in biosynthesis of catecholamines (CA; reviewed in Stachowiak and Goc 1992; Kumer and Vrana 1996).
Up-regulation of TH mRNA in BAMC by forskolin is not affected by cycloheximide. In contrast, cycloheximide prevents TH mRNA up-regulation by depolarization, or AII receptors, implicating signaling through inducible or high turnover rate proteins (Goc et al. 1992; Stachowiak et al. 1994b). Numerous studies have suggested that AII, PKC, or acetylcholine may exert their biological effects by up-regulating the endogenous fibroblast growth factor 2 (FGF-2; reviewed in Peng et al. 2001). We have shown that carbachol, nicotine, AII, or agents that directly activate their intracellular signaling pathways all stimulate FGF-2 expression in cultured BAMC (Stachowiak et al. 1994b, 1997b). Induction of 18 kDa FGF-2 and of the N-terminally extended, high molecular weight (HMW; 21–24 kDa) isoforms of FGF-2, all of which lack signal peptides, does not lead to the detectable presence of FGF-2 in the extracellular medium of BAMC. Instead, a robust accumulation of newly synthesized FGF-2 in the cell nucleus is observed. Induction of nuclear FGF-2 was observed in a variety of cells and may constitute a common response to cell stimulation (reviewed in Stachowiak et al. 1997b; Peng et al. 2001).
In BAMC the number of the high-affinity FGF-2-binding sites in the nucleus is over 10-fold greater than on the cell surface. BAMC express only FGF receptor-1 (FGFR1) that accounts for the high-affinity FGF-2-binding sites in both locations (Stachowiak et al. 1996a, 1997b). Treatments that induce the translocation of endogenous FGF-2 to the nucleus also increase the intranuclear accumulation of activated (phosphorylated) FGFR1 and this accumulation is accompanied by an overall increase in FGF-2-inducible tyrosine kinase activity (Stachowiak et al. 1996a,b, 1997a). FGF-2 and FGFR1 associate with the nuclear matrix (Stachowiak et al. 1996a,b) and are thus strategically positioned in the nucleus to be directly involved in gene regulation (Berezney 1980). We recently showed that nuclear FGFR1 can transactivate the FGF-2 gene (Peng et al. 2001). In the present study we report that AII receptors, cell depolarization, and a downstream, PKC-dependent signaling pathway activate the TH gene promoter through this novel signal transducing mechanism.
Materials and methods
The TH-Luc containing −425/+25 bp fragment of the bovine TH promoter, and its promoter mutants, were described by Kim et al. (1996). pcDNA3.1-FGFR1 expressing the IIIC form of FGFR1 was described previously (Stachowiak et al. 1997a). In FGFR1(SP-), amino acids (3–19) containing the hydrophobic signal peptide required for protein insertion into the ER membrane were deleted. In FGFR1(SP-/NLS) the 3–19 sequence was replaced with the nuclear localization signal (NLS) from the SV40 large T antigen [PKKKRKV (Dingwall and Laskey 1991)]. FGFR1(TK-) and FGFR1(TK-)(SP-/NLS), which lack the tyrosine kinase domain, were generated by deleting the remainder of the FGFR1 sequence beginning 21 bp downstream from the transmembrane domain. FGFR1-EGFP (enhanced green fluorescent protein) was constructed by an in-frame fusion of FGFR1 cDNA (lacking the 36 C-terminal amino acids and the stop codon) with the N-terminus of EGFP in pEGFP. The deletion of these C-terminal amino acids had no effect on the subcellular localization of non-tagged FGFR1 or its transactivating functions. All mutations were verified by DNA sequencing.
Cells and promoter assays
BAMC were purified and maintained in serum-free Dulbecco's minimal essential media (DMEM)/F12 supplemented with 0.25% bovine serum albumin (BSA) as described in (Stachowiak et al. 1990b, 1994b, 1996a). The TE671 cells were cultured as in Kim et al. (1998). Cells were transfected using calcium phosphate or lipofectin (Stachowiak et al. 1994b). Drugs were added 48 h after the transfection. Luciferase activity was expressed in numbers of light units per pg of transfected intracellular luciferase cDNA (quantitative dot-blot hybridization) or per µg of cellular protein (Stachowiak et al. 1994b). The results obtained with and without this direct DNA normalization or using indirect normalization to transfected control CMV-Luc or CMV-EGFP were essentially the same.
Quantitative microscopic analysis of CMV-EGFP expression
Thirty-five-millimeter dishes were transfected with 2 µg of pCMV-EGFP and with 2 µg of FGFR1, FGFR1(TK-) or control pcDNA3.1 plasmid. The images of live green fluorescent cells were acquired 48 h later with a XILLIX Microimager cooled CCD camera on a Nikon FXA fluorescent microscope. At that time cells were 60–80% confluent. All images gathered were in the linear range of the camera and were analyzed using the ONCOR image program. Cells in contact with edge of the dish were excluded. The mean cell intensity (as a measure of cell expression) was calculated by dividing the integral intensity in the entire dish by the number of fluorescent cells.
Immunocytochemistry, confocal and electron microscopy
Cells were fixed and stained using a FGFR1 polyclonal C-term Ab (Hanneken et al. 1995), a TH polyclonal Ab (Chemicon International, Temecula, CA, USA), or a FGF-2 mAb (Upstate Biotechnology, Lake Placid, NY, USA), Y12 mAb as described previously (Stachowiak et al. 1994b, 1996a; Wei et al. 1999). Fluorescent staining of antigen–antibody immune complexes was performed using CY3-conjugated secondary antibodies (for C-term FGFR1 Ab and TH Ab) or Alexa 488-conjugated goat anti-mouse IgG (Molecular Probes, Eugene, OR, USA; for FGF-2 mAb or Y12 mAb). Confocal optical sections of fixed immuno-stained cells were acquired using a Bio-Rad MRC 1024 confocal microscope (Bio-Rad Laboratories, Hercules, CA, USA) with a 15 mW Krypton/Argon laser, operating on a Nikon Optiphot upright microscope and an oil immersion 60 × 1.4NA objective, as previously (Peng et al. 2001). The Cy3 and Alexa dyes were excited using the 568 nm and 488 nm laser lines and images were recorded using collection filters HQ598/40 and 522DF32, respectively. Native fluorescence of the EGFP was imaged using the settings for the Alexa 488 dye. Optical slices were acquired at 0.5 µm steps and the X,Y resolution was 0.23 µm (the average nuclear diameter was 3–5 µm). Co-localization analysis of double labeled cells was performed using Bio-Rad's Lasersharp V3.0. Possibility of bleed-through was excluded by acquiring the images in sequential mode. Co-localization analysis produced coefficients (proportion of co-localizing pixels in each component of a dual-color image), and images of white pixels (co-localized) superimposed over the original double-colored image (Fig. 1c, panel XI; Fig. 1d). Single nuclei of randomly selected cells were used to␣generate the fluorescence intensity plots (Fig. 1d) with NIH's ImageJ V1.23q software, for both red and green dyes and also for images of co-localized pixels only obtained with lasersharp. Particle analysis (NIH IMAGEJ vl.23q; http://rsb.info.nih.gov/nih-image/) was used to assess the presence of nuclear speckles on representative nuclear sections; this involved segmenting pixels with intensity values higher than 10x the average noncellular background, and counting groups of adjacent pixels (‘particles’) larger than 1.0 µm2.
Immunoelectron microscopy was performed using a monoclonal FGF-2 Ab (UBI, Lake Placid, NY, USA) and a colloidal gold-conjugated secondary IgG (Stachowiak et al. 1996a). The specificity of FGFR1 staining was demonstrated in Stachowiak et al. (1996a,b, 1997a): (i) the staining was not observed when the primary antibody was omitted or replaced with preimmune serum; (ii) similar staining was observed using two different antibodies (C-term FGFR1 Ab and mAb6) and different staining techniques; (iii) the same staining technique detected FGFR1 in the nuclei and the cytoplasmic enzyme, tyrosine hydroxylase, in the cytoplasm; (iv) neutralization of the C-term FGFR1 Ab by an excess of the C-terminal FGFR1 peptide reduced cytoplasmic FGFR1 staining and eliminated nuclear FGFR1 staining (Stachowiak et al. 1996a,b, 1997a). The presence of, and changes in, the levels of nuclear FGFR1 immunoreactivity were confirmed by western blot analysis of FGFR1 in subcellular fractions, by detecting an increase in FGFR1-IR in cells transfected with FGFR1 expressing plasmids, and by detecting the nuclear fluorescence of EGFP in cells expressing the FGFR1-EGFP fusion protein. The specificity of the FGF-2 staining was shown previously by Joy et al. (1997) and Stachowiak et al. (1994b, 1997b).
Western blot analyses
The nuclear fraction and the extranuclear fraction (cytoplasmic fraction including plasma membrane) were isolated and characterized as described in Stachowiak et al. (1996a,b). The isolated nuclei contained approximately 90% of the total TCA-precipitable DNA (Stachowiak et al. 1996a) and nuclear FGFR1 co-purified with typical nuclear proteins such as CREB and cCBP/p300 and with histones (not shown). Analysis of the isolated nuclei by phase-contrast microscopy showed no contamination with cytoplasmic membranes and organelle. The nuclei contained less then 5% of the total cellular activity of 5′nucleotidase (plasma membrane marker), and less than 2% of the total activity of acid phosphatase (lysosomal marker; Stachowiak et al. 1996a,b, 1997a). The purity of the nuclear preparation was also demonstrated by the absence of FGF-1 in the nuclear fraction (Fig. 4). Equal amounts of nuclear and non-nuclear proteins were separated by sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS–PAGE) and transferred to nitrocellulose. Equal protein loading was confirmed by staining the proteins with Coomassie blue. The nitrocellulose membranes were probed with anti-FGFR1 mAb6 (Hanneken et al. 1995), FGF-2 mAb, FGF-1 mAb or antiphospho CREB Ab (Upstate Biotechnology, Lake Placid, NY, USA). Immune complexes were visualized using chemiluminescence. In some experiments, the nuclear fraction was immunoprecipitated with a FGFR1 C-term Ab (Santa Cruz Biotechnology, Santa Cruz, CA, USA) or control IgG, as described␣in Stachowiak et al. (1996a,b). Immunoprecipitates were subjected to western blot analysis with FGFR1 mAb6 or a rabbit antiphosphotyrosine Ab prepared as described (Pasquale et al. 1988).
Electrophoretic mobility shift assay
Nuclear extracts were prepared and electrophoretic mobility shift assay (EMSA) was performed as in Moffett et al. (1996, 1998), Peng et al. (2001). Control antibodies or antibodies against FGFR1 (mAb6, or C-term Ab; Hanneken et al. 1995), CBP/p-300, or an antibody against the DNA-binding domain of CREB (Santa Cruz Biotechnology) were added to some reactions and the incubations continued for an additional 8 h at 4°C. Extended incubation with control antibodies had no effect on the formation of protein–DNA complexes. DNA–protein complexes were resolved on non-denaturing polyacrylamide gels (gel concentration varied between 4.5 and 5.5% in different experiments to allow for optimal separation of different protein–DNA complexes).
Stimuli that activate the TH gene increase the nuclear accumulation of FGF-2 and FGFR1 and induce an association of FGFR1 with splicing factor-rich extranucleolar domains
Western blot analysis of total cellular extracts prepared from unstimulated cultures of BAMC in serum-free medium showed predominantly 18 kDa FGF-2 (Fig. 1a). However, upon stimulation with a stable analog of AII (sar1-AII), a marked induction of HMW FGF-2 and an up-regulation of 18 kDa FGF-2 were observed. The NLS-containing HMW FGF-2 isoforms are localized to the nuclei in many different cell types (reviewed in Stachowiak et al. 1997b; Szebeney and Fallon 1999). In agreement with this, immunocytochemical analysis showed nuclear increases in FGF-2-immunoreactivity (IR) in AII-stimulated BAMC (Fig. 1c). The increases were rapid (within 1 h) and were observed in BAMC treated with sar1-AII, as well as with the depolarizing agent veratridine, or PMA (but not the inactive 4-α phorbol didecanoate; Stachowiak et al. 1994b). During the subsequent hour, the FGF-2-IR decreased (Fig. 1c) but remained elevated relative to control cells.
The increases in nuclear FGF-2 were accompanied by a more gradual nuclear accumulation of FGFR1 as determined by staining BAMC with an affinity-purified polyclonal C-term FGFR1 Ab (Hanneken et al. 1995; Fig. 1c). An increase in nuclear FGFR1-IR was observed within 1 h of sar1-AII addition. In the subsequent hour, the staining for FGFR1 became stronger than that for FGF-2. Co-treatment with sar1-AII and veratridine for 2 h caused a greater nuclear accumulation of FGFR1-IR than the treatment with each factor alone. PMA also increased nuclear FGFR1-IR. Treatment of BAMC with forskolin had the same effect on FGF-2 and FGFR1 as sar1-AII (shown in previous studies; Stachowiak et al. 1994b, 1996a). FGFR1-IR was abolished by pre-incubating the C-term FGFR1 Ab with its cognate peptide and similar nuclear staining was observed with a monoclonal FGFR1 (mAb6; Stachowiak et al. 1996a). The increase in nuclear FGFR1 in BAMC treated with sar1-AII for 1.5 h cells was confirmed by western blot analysis (Fig. 1b) with mAb6, which is specific for the N-terminal portion of FGFR1 (Hanneken et al. 1995). In the extranuclear fraction, mAb6 detected three bands of approximately 100, 115, and 130 kDa FGFR1, which represent different degrees of glycosylation of the plasma membrane-associated receptor (Stachowiak et al. 1996a, 1997a,b). In the nuclear fraction, the antibody detected three bands of 95, 115 and 130 kDa which also represent different glycosylated forms of full-length FGFR1 (Stachowiak et al. 1996a, 1997a,b). The intranuclear FGFR1-IR in stimulated BAMC had a characteristically patchy distribution. In contrast, FGF-2-IR was more diffuse, although occasional concentrations of FGF-2IR speckles within the FGFR1-positive speckles were observed. Also, unlike FGFR1-IR, FGF-2IR was not excluded from the nucleoli. Approximately 14–20% of the nuclear FGFR1 speckles overlapped with those of FGF-2 (Fig. 1c, panels IX-XI).
As a first approach towards identifying the functions of nuclear FGFR1, we examined the nature of the FGFR1-containing extranucleolar speckles. One type of extranucleolar nuclear speckle can be decorated with an antibody raised against the spliceosome assembly factor, SC-35 (Fu and Maniatis 1990; Spector 1993; Blencowe et al. 1994) or with mAbY12 which recognizes common core proteins of snRNPs involved in RNA processing (Lerner et al. 1981; Wei et al. 1999). Recent studies have established that these nuclear speckles are sites for RNA Pol II-mediated transcription as well as co-transcriptional, pre-mRNA processing (reviewed in Wei et al. 1999). Hence we examined whether nuclear FGFR1 is specifically localized within these transcriptionally active extranucleolar domains. In unstimulated BAMC, Y12-IR and FGFR1-IR pixels of variable intensity were distributed throughout the nucleus with the exception of the nucleolus (Fig. 1d). After 1 h of sar1-AII treatment, several large speckles which stained intensely with both mAbY12 and a FGFR1 Ab emerged (Fig. 1d). Their size and number were consistent with previous descriptions of Y12/SC35 speckles using different types of cells (20–50 speckles/nucleus; 0.5–3.0 µm; Spector 1993; Smith et al. 1999; Wei et al. 1999). Merging the Y12 and FGFR1 images and co-localization analysis confirmed that the FGFR1 and Y-12 speckles coincided (Fig. 1d). Since the resolution in the z plane of the confocal microscope is 0.5 µm and is lower than that in the x/y-planes (0.23 µm), FGFR1 could actually be located just above or below the Y12 speckles thereby giving the appearance of co-localization in a particular optical section. To investigate this possibility, we analyzed three-dimensional stacks of FGFR1- and Y12-stained nuclei. Several examples of FGFR1 sites in the interior of Y12 speckles were observed (not shown).
The emergence and co-localization of FGFR1 and Y12 speckle-like domains in sar1-AII-treated cells is illustrated by fluorescence intensity plots (Fig. 1d). Eighty-five to 98% of FGFR-1-IR pixels overlapped with Y12-IR pixels, indicating that nearly all FGFR1 is associated with the Y12-containing speckles. In contrast, in control cells less than 10% of FGFR1 and Y12 pixels were co-localized and the pixels did not assemble into speckle-like domains (Fig. 1d). Also, the high level of FGFR1 and Y12 co-localization in sar1-AII-treated cells was not random as shown by the significantly lower coincidence (14–20%) for FGFR1-IR and FGF-2-IR, even though FGF-2–IR was more abundant than Y12-IR (Fig. 1c).
The fine subcellular localization of FGF-2 in stimulated BAMC was examined by immunocytochemistry electron microscopy. No FGF-2-IR was detected in secretory vesicles or associated with the plasma membrane and only a small amount was seen in the cytosol (Fig. 1e). In contrast, very strong staining for FGF-2 was found at the nuclear membrane and in the nuclear interior. No staining was observed when the primary antibodies were omitted or replaced with pre-immune serum (Fig. 1e).
FGFR1 signaling is necessary for the activation of the TH␣gene promoter by AII, depolarization and PKC
The rapid nuclear accumulation of FGF-2 and FGFR1 in BAMC in response to treatment with AII (Fig. 1) coincides with the transcriptional activation of the TH gene (Goc et al. 1992). Hence, we investigated whether the up-regulation of FGF-2 and FGFR1 in the nucleus could play a role in the activation of the TH gene by AII and other stimuli. Transcriptional activation of the TH gene was monitored by transfecting BAMC with the TH-Luc plasmid which contains a −425/+25 bp promoter fragment of the bovine TH gene linked to luciferase cDNA (Goc and Stachowiak 1994; Kim et al. 1996). Regulation of either bovine- (−425/+25 bp; Stachowiak et al. 1990a, 1994a; Goc and Stachowiak 1994; Kim et al. 1996) or rat TH promoter- (−219/+27 bp; Fung et al. 1992; Lewis et al. 1987; Hiremagular et al. 1993) reporter constructs by a variety of stimuli parallels that of the endogenous TH gene, TH mRNA, and TH protein (Lewis et al. 1987; Stachowiak et al. 1990a,b,c; Craviso et al. 1992; Goc et al. 1992; Kilbourne et al. 1992; Goc and Stachowiak 1994). To determine the role of FGFR1 in the transcriptional activation of the TH gene by AII and other stimuli we used a dominant negative FGFR1 mutant with a deleted tyrosine kinase domain, FGFR1(TK-), that prevents FGFR signaling. FGFR1(TK-) forms non-phosphorylated, inactive dimers specifically with FGFR (Ueno et al. 1992; Li et al. 1994) and, in the case of BAMC, with FGFR1, the only type of FGFR expressed by these cells (Stachowiak et al. 1996a). BAMC were co-transfected with FGFR1(TK-) or empty vector along with the TH-Luc reporter plasmid. The empty vector had no effect on TH promoter activation, and, as in earlier studies (Stachowiak et al. 1994a; Goc and Stachowiak 1994; Kim et al. 1996), sar1-AII and veratridine increased promoter activity two- to three-fold (Fig. 2a). Similar to the activation of the endogenous TH gene (Stachowiak et al. 1990c), stimulation of the TH promoter with maximally effective concentrations of sar1-AII and veratridine was additive. Also, PMA increased luciferase expression three-fold and forskolin four- to six-fold (Fig. 2a). Background luciferase activity expressed from the promoter-less pGL2Basic was not affected by any of these stimuli, thus excluding a post-transcriptional modification of luciferase activity (Goc and Stachowiak 1994; Stachowiak et al. 1994a,b; Kim et al. 1996). Co-transfection of TH-Luc with FGFR1(TK-) completely prevented the elevation in luciferase expression induced by treatment with sar1-AII or veratridine (Fig. 2a). This inhibition was not overcome by combined stimulation with both sar1-AII and veratridine. The three-fold increase in promoter activity induced by PMA was also entirely prevented by FGFR1(TK-) (Fig. 2a), indicating that FGFR1 acts downstream of PKC activation. In contrast, FGFR(TK-) did not prevent the stimulation of TH-Luc by forskolin. Thus, FGFR1(TK-) does not cause a generalized inhibition of transcriptional activation and its inhibition of TH gene stimulation is specific for the pathways utilized by AII, veratridine and PKC, but not cAMP. These findings are in agreement with earlier reports that the activation of the TH gene by AII and veratridine is blocked by cycloheximide, while the activation by forskolin is not (Stachowiak et al. 1990c, 1994a). Co-treatment with sar1-AII and forskolin had significantly greater effect on TH promoter activity than with forskolin alone (Fig. 2a; p > 0.05, Student's t-test) similar to the earlier shown changes in the TH mRNA levels (Stachowiak et al. 1990c). Interestingly, the synergistic stimulation of TH-Luc by forskolin and AII was not significantly affected by FGFR1(TK-) (Fig. 2a). This result is consistent with data showing that AII alone has only a small effect on the cAMP content of BAMC but, together with forskolin, it can synergistically enhance cAMP levels (Boarder et al. 1988). Thus, AII appears to augment the␣stimulation of the TH gene by cAMP in a FGFR1-independent manner. FGFR1(TK-) reduced basal TH promoter activity by 20–40% in unstimulated BAMC (Fig. 2a). In contrast to these results with FGFR1(TK-), wild-type FGFR1 increased luciferase expression three-fold in unstimulated BAMC maintained in serum-free medium and to a smaller extent in BAMC treated with sar1-AII and veratridine (Fig. 2d). In FGFR1(TK-) both the TK and C-terminal domains have been deleted. However, FGFR1 lacking the 36 C-terminal amino acids transactivated the TH gene promoter three- to five-fold, similar to full-length FGFR1 (not shown). Thus, unlike FGFR1(TK-), the wild-type, full-length receptor or mutants that retain the TK domain act as TH promoter transactivators.
In different experimental paradigms in which TH promoter activity has been studied along with the TH mRNA and TH protein levels, changes in promoter activity consistently accompanied changes in protein and mRNA content. In the present study, the effects of FGFR1(TK-) on the expression of the co-transfected TH promoter-luciferase construct were essentially the same as on the endogenous TH gene. BAMC were co-transfected with pcDNA3.1-FGFR1(TK-) or control vector and pCMV-EGFP expressing green fluorescent protein and 2 days later the cells were treated with sar1-AII and veratridine or kept in control, serum-free medium for an additional 48 h and then stained with an antibody against TH protein and a CY3-conjugated secondary IgG (Fig. 2b). Approximately 5–10% of the cells were transfected as indicated by EGFP fluorescence. TH immunoreactivity was localized to the cytoplasm and was absent from the nuclei. In BAMC transfected with the control vector, treatment with sar1-AII and veratridine enhanced TH staining. In contrast, transfection with FGFR1(TK-) markedly reduced the intensity of TH staining in control as well as in sar1-AII and veratridine-treated cells, although it had no effect on the expression of EGFP. Furthermore, in BAMC that were not transfected [i.e. cells lacking EGFP fluorescence in dishes co-transfected with pCMV-EGFP and FGFR1(TK-)], the increase in TH staining induced by AII and veratridine was not affected.
Intracrine activation of TH gene promoter via FGFR1
In contrast to sar1-AII, veratridine, or PMA (Fig. 1b,c), treatment of BAMC with FGF-2 peptide did not affect the total cellular content of FGFR1 or induce the nuclear accumulation of the receptor (Stachowiak et al. 1996a). In the present study, incubation of BAMC with exogenous 18 kDa FGF-2 produced only a small (approximately 40%) elevation in luciferase activity from TH-Luc (Fig. 2c), similar to the elevation of TH mRNA (Puchacz et al. 1993). This stimulation by exogenous FGF-2 was completely blocked by inositolhexakisphosphate (IP6), an extracellularly acting FGFR antagonist (Sherman et al. 1993; Morrison et al. 1994). In contrast, sar1-AII-induced increases in TH-Luc expression in BAMC were not prevented by IP6 (Fig. 2c). Thus, AII stimulation of the TH promoter is unlikely to be due to an interaction of an extracellular ligand with cell surface FGFR1. These results are consistent with the absence of detectable FGF-2 outside stimulated BAMC (Stachowiak et al. 1994b) or associated with the cell surface or secretory vesicles (Fig. 1e).
In TE671 cells, transfected FGF-2 and FGFR1 are expressed in a similar manner as endogenous FGF-2 and FGFR1 in stimulated BAMC
So far our experiments have shown that FGFR1, perhaps in concert with FGF-2, mediates the activation of the TH gene promoter by AII, depolarization, or PKC in BAMC. The lack of inhibition by IP6 indicates an intracellular site of action for FGFR1. Given the nuclear localization of endogenous FGFR1 (Stachowiak et al. 1996a; Peng et al. 2001), these findings raise an intriguing question as to whether FGFR1 may directly transactivate the TH gene in the nucleus. We used medulloblastoma TE671 cells to further elucidate the molecular mechanisms through which FGFR1 and FGF-2 can activate the TH gene. These cells express lower levels of endogenous FGF-2 and FGFR1 (Peng et al. 2001; this study) and no detectable FGFR2-4 (our unpublished observations) and can be transfected at a markedly higher frequency than the primary BAMC cultures.
A western blot of TE671 cells transfected with pCMV-FGF-2 which expresses all (18, 21, 22 and 24 kDa) FGF-2 isoforms showed an accumulation of 18 kDa FGF-2 predominantly in the nuclear fraction and to a lesser extent in the extranuclear material. The HMWFGF-2 isoforms were localized exclusively to the cell nucleus (Fig. 3a). The same localization of HMWFGF-2 was observed in TE671 cells transfected with pCMV-HMWFGF-2 which expresses only the HMW (21–24 kDa) FGF-2 isoforms (Joy et al. 1997). No 18 kDa FGF-2 was detected in these cells (data not shown). Consistent with the nuclear accumulation of all FGF-2 isoforms as shown by western blot analysis (Fig. 3a), TE671 cells transfected with pCMV-FGF-2, with pCMV-HMWFGF-2, or with pCMV-18 expressing only 18 kDa FGF-2, showed increased levels of intranuclear FGF-2-IR (Fig. 3b). In contrast to FGF-2, in cells transfected with a construct that expresses FGF-1, the growth factor was detected only in the extranuclear fraction (Fig. 3c).
Forty-eight hours after transfection of TE671 cells with pcDNA3.1-FGFR1, we observed the presence of FGFR1 glycosylation isoforms of ∼135 and 120 kDa and a small amount of non-glycosylated 95 kDa FGFR1 in the nucleus and both the 135 kDa and 120 kDa FGFR1 in the extranuclear fraction (Fig. 4a). In TE671 cells transfected with a chimeric construct expressing a fused FGFR1-EGFP protein, the nuclear and cytoplasmic receptor bands showed approximately a 30-kDa shift upward in molecular weight, consistent with the molecular weight of the added EGFP (Fig. 4a). Cultures transfected with FGFR1(TK-) showed the presence of the truncated receptor in both the extranuclear and nuclear fractions (Fig. 4a). The nuclear localization of transfected FGFR1(TK-) was also observed by immunocytochemistry (not shown) and reported by other investigators (Saiki et al. 1999). The localization of recombinant FGFR1 in the nucleus was confirmed with immunocytochemistry and confocal microscopy. Transfection of TE671 cells with pcDNA3.1-FGFR1 resulted in the perinuclear and intranuclear accumulation of FGFR1-IR (Fig. 4b). The intranuclear presence of FGFR1 was confirmed in live TE671 cells by analyzing the EGFP fluorescence of cells transfected with FGFR1-EGFP under a confocal microscope (Fig. 4c). While FGFR1-EGFP was localized to both the nucleus and cytoplasm, approximately 80 kDa tubulin-EGFP (used as control) localized exclusively to the cytoplasm. The same results were obtained in BAMC transfected with FGFR1-EGFP (our unpublished observations). Overexpression of HMWFGF-2 increases FGFR1 mRNA and FGFR1 protein levels in a pancreatic acinar cell line (Estival et al. 1996). Consistent with this data, many cells in pCMV-HMWFGF-transfected, but not in control TE671 cultures, displayed increases in FGFR1-IR at the periphery of the nucleus and in the nuclear interior (Fig. 4d). The HMWFGF-2-induced accumulation of endogenous FGFR1 specifically in the cell nucleus was confirmed by western blot analysis (Fig. 4d). Furthermore, western blot analyses with a FGF-2 Ab showed that transfection of TE671 cells with pcDNA3.1-FGFR1 increases the expression of nuclear HMWFGF-2 (Peng et al.␣2001). We confirmed this earlier observation using immunocytochemistry and confocal microscopy which demonstrated an accumulation of FGF-2-IR in the nucleus of FGFR1-transfected TE671 cells (Fig. 3b). These results suggest that nuclear FGF-2 and FGFR1 act in concert to activate the TH gene promoter.
Activation of the TH gene promoter by nuclear FGF-2 and FGFR1
We next examined the effects of recombinant FGF-2 and FGFR1 on TH promoter activity. TE671 cells were co-transfected with TH-Luc and pCMV-FGF-2 or pCMV-HMWFGF-2. FGF-2 or HMWFGF-2 increased TH-Luc gene activity three- to four-fold when compared to cells transfected with control pCMV-βgal (Fig. 5a). In contrast, transfection of the cells with a FGF-1-expressing plasmid did not significantly increase TH promoter activity (Fig. 5a). Thus, the TH gene promoter is activated by nuclear FGF-2 (Fig. 3a) but not by non-nuclear FGF-1 (Fig. 3a), even though both FGFs can bind equally well to FGFR1 (Mikei et al. 1992; Zimmer et al. 1993). Transfection of TE671 cells with pcDNA3.1-FGFR1 also increased TH-Luc expression several-fold (Fig. 5b). A specific, two- to three-fold stimulation of the TH promoter by FGFR1 or FGF-2 was also seen when the increase in luciferase activity was compared in the same experiment between cells co-transfected with TH-Luc and either FGFR1 or FGF-2 and cells co-transfected with control CMV-Luc and either FGFR1 or FGF-2 (data not shown). Background luciferase activity expressed from the promoter-less pGL2Basic plasmid was not affected by transfected FGFR1 or FGF-2 (not shown) and mutation of the TH promoter CRE prevented stimulation by FGFR1 (see Fig. 7a). Finally, in additional control experiments, the number of cells expressing co-transfected pCMV-EGFP, as well as the level of EGFP expression, was not significantly affected by transfection with either FGFR1 or FGFR1(TK-) when compared to cells transfected with empty vector, thus excluding non-specific toxic effects of the FGFR1 plasmids. The mean integrated fluorescence intensities/cell (see Materials and methods) were 11 323 ± 1107 (pcDNA3.1), 9829 ± 809 (FGFR1), and 11 701 ± 915 (FGFR1(TK-)␣(n = 4).
The cross-regulation of FGF-2 by FGFR1 (Peng et al. 2001; Fig. 3b) and FGFR1 by FGF-2 (Fig. 4d) suggests that nuclear FGF-2 and FGFR1 may act in concert and that HMWFGF-2 may stimulate the TH promoter by activating nuclear FGFR1. To test this idea, we examined the effect of FGFR1(TK-) on the transactivation of the TH promoter by HMWFGF-2. As shown in Fig. 5(b), the transactivation of TH-Luc by HMWFGF-2 is inhibited by FGFR1(TK-) as is TH promoter activation by transfected FGFR1 (Fig. 5b). The activation of FGFR1 by FGF-2 results in receptor phosphorylation. Increased phosphorylation of endogenous nuclear FGFR1 in response to stimulation of BAMC was shown previously (Stachowiak et al. 1996a,b) and in TE671 cells transfected nuclear FGFR1 was also found to be phosphorylated (data not shown). To determine if FGFR1 can transactivate the TH promoter by acting specifically in the cell nucleus we used two additional receptor mutants. In FGFR1(SP-) the hydrophobic leader sequence (signal peptide, SP) was deleted. Transfected FGFR1(SP-) was detected as a distinct 95 kDa band consistent with the size of the non-glycosylated receptor (Stachowiak et al. 1997b). Although FGFR1(SP-) was found in both the extranuclear and nuclear fractions, the level of nuclear receptor was three- to five-fold lower (in different experiments) than that of cytoplasmic receptor (Fig. 5c). In the second mutant, FGFR1(SP-/NLS), the signal peptide was replaced with the NLS of the SV40 large T antigen. As an additional control we created a derivative of FGFR1(SP-/NLS), FGFR1(TK-)(SP-/NLS), with a deleted TK domain. Western blot analysis of TE671 transfected with FGFR1(SP-/NLS) or FGFR1(TK-)(SP-/NLS) confirmed the exclusively nuclear localization of these␣mutant receptors (Fig. 5c). Neither FGFR1(SP-) nor FGFR1(SP-/NLS) became biotinylated when TE671 cells were treated with NHS-sulfobiotin (data not shown), confirming that they were not inserted into the plasma membrane.
Next, we co-transfected TE671 cells with TH-Luc and FGFR1(SP-/NLS) and compared its effects with those of wild-type FGFR1 and soluble predominantly cytoplasmic FGFR1(SP-). FGFR1(SP-/NLS) transactivated the TH promoter twice as a effectively as wild-type FGFR1 (Fig. 5c), whereas FGFR1(SP-) was much less effective than the wild-type receptor. In contrast, FGFR1(TK-)(SP-/NLS) did not stimulate TH promoter activity. Thus, the specific accumulation of active FGFR1 in the cell nucleus is sufficient to activate transcription from the TH gene promoter and the transactivating efficiency of soluble FGFR1 correlates with its nuclear accumulation (Fig. 5C).
Nuclear FGF-2 and FGFR1 activate the TH gene promoter in an intracrine manner
Unlike transfected HMWFGF-2, exogenous FGF-2 had no effect on TH-Luc expression in TE671 cells (Fig. 6a). Since transfection of FGFR1 alone increased TH promoter activity (Fig. 5b,c), we next examined whether transfected FGFR1 could make the TH gene promoter responsive to stimulation by extracellular FGF-2. However, in cells transfected with FGFR1, no additional stimulation by FGF-2 was observed (Fig. 6a), even though the promoter transactivation by FGFR1 used in this experiment was less then the maximal eight-fold stimulation observed in other experiments (not shown). Furthermore, the increase in TH-Luc expression in TE671 cells transfected with FGFR1 and incubated with exogenous FGF-2 was not reduced by IP6 or by suramin, another agent that blocks the binding of FGFs to their cell surface receptors (Sherman et al. 1993; Morrison et al. 1994; Dai and Peng 1995; Estival et al. 1996). Suramin also had no␣effect on the stimulation of TH-Luc by HMWFGF-2 (not␣shown). In addition to blocking the interaction between cell surface FGFR and extracelllular ligands, suramin also prevents ligand-induced receptor internalization (Moscateli 1994; Estival et al. 1996). Thus, the lack of suramin inhibition of␣TH-Luc transactivation by FGFR1 or HMWFGF-2 indicates that the intracellular FGFR1 that mediates this transactivation was not derived from the cell surface. Thus,␣the activation of the TH gene by transfected HMWFGF-2 and the transactivation by FGFR1 is unlikely to be caused by extracellular FGFs interacting with surface receptors.
To provide further evidence for an intracellular origin for nuclear FGFR1 in TE671 cells, cell surface proteins were labeled with membrane-insoluble NHS-sulfobiotin (Vector Laboratories, Burlingame, CA, USA) for 30 min at 4°C and then either solubilized immediately or returned to the culture medium for 4 h. Biotin-tagged proteins were precipitated with avidin–agarose and the precipitates were analyzed for FGFR1 (Fig. 6b). In cells transfected with FGFR1 or FGFR1(TK-), large amounts of biotinylated receptors were found in the extranuclear fraction but not in the nuclear fraction, even at the 4 h time point. Since biotinylation of surface proteins does not hinder the internalization of cell surface FGFR1 (Maher 1996), these data further corroborate our conclusion that full-length and truncated nuclear FGFR1 do not come from the cell surface and are consistent with the results of the IP6 and suramin experiments (Figs 2c and 6a). Furthermore, these results clearly demonstrate the lack of contamination of the nuclear fraction with plasma membrane FGFR1.
Nuclear FGFR1 transactivates the TH gene promoter via the CRE
As discussed in the Introduction, regulation of the TH promoter depends primarily on both a CREB-binding cAMP-responsive element (CRE) and an activating protein-1 (AP1) site. Mutation of the CRE completely prevented TH-Luc transactivation by FGFR1 (Fig. 7a) and by HMWFGF-2 (not shown). In contrast, mutation of the AP1 site had no effect on promoter transactivation by FGFR1 (Fig. 6a). These results were consistent with the earlier report of Osaka and Sabban (1997) that, in PC12 cells, FGF-2 activates the rat TH gene promoter via the CRE but the AP1 site. Given these results, we used EMSA to examine whether FGFR1 could regulate protein binding to the CRE-containing region of the TH gene promoter (Fig. 7b–g).
Nuclear extracts from TE671 cells transfected with control plasmids (Fig. 7b; pBK-CMV, lanes 1 and 2, or pcDNA3.1, lane 6) showed weak binding to the TH CRE. In contrast, cells transfected with nuclear HMWFGF-2 (Fig. 7b, lanes 3 and 4), nuclear/cytoplasmic FGFR1 (lane 5) or exclusively nuclear FGFR1(SP-/NLS) (Fig. 7b, lane 7) exhibited a marked induction of high, ‘H’, and an up-regulation of low, ‘L’, mobility DNA-protein complexes (Fig. 7b). Exogenous FGF-2 peptide had little or no effect on protein binding to the TH CRE (not shown). Binding of nuclear proteins from HMWFGF-2- (Fig. 7c) or FGFR1-transfected cells (not shown) to the TH CRE probe was competed out by unlabeled probe (Fig. 7c, lane 2) or by a consensus CRE oligonucleotide (Fig. 7c, lane 3) and was abolished by a mutation of the CRE sequence in the labeled − 60/+1 bp EMSA probe (not shown). Thus, both CRE-mediated promoter transactivation and CRE protein binding can be specifically induced by nuclear FGFR1 and HMWFGF-2.
Co-transfection of FGFR1(TK-) markedly depleted the HMWFGF-2-induced ‘L’ bands (Fig. 7d; compare lanes 4 and 5 to lanes 1 and 2) and partially depleted the ‘H’ bands. Thus, stimulation of TH CRE protein binding by nuclear HMWFGF-2 requires FGFR1. Stimulation of TH CRE protein binding by FGFR1 was also inhibited by co-transfected FGFR1(TK-) (not shown). The up-regulation of TH CRE binding by nuclear FGFR1(SP-/NLS) and HMWFGF-2 show that FGFR1 and ligand can regulate protein binding to the TH promoter directly in the cell nucleus. To confirm this result, the DNA–protein binding reactions were incubated with an antibody directed against the extracellular domain of FGFR, mAb6, whose specificity was established in a number of independent studies (Hanneken et al. 1995; Maher 1996; Stachowiak et al. 1997a; Peng et al. 2001; see also Fig. 4a). Treatment of a DNA–protein complex with a specific antibody can result in either an upward shift of a band, in cases where the antibody does not interfere with protein binding to the DNA (a tertiary DNA–protein–Ab complex may form), or in the deletion of a band, when the reversible DNA–protein or protein–protein interaction is disrupted. mAb6 recognizes the region of FGFR1 that contains the ligand-binding site. Addition of mAb6 to the EMSA reactions almost completely abolished the formation of the ‘L’ bands induced by HMWFGF-2 (Fig. 7e, lane 2) or by transfected FGFR1 (Fig. 7f, lane 4), and partially reduced the ‘H’ bands, similar to the effects of co-transfected FGFR1(TK-) (Fig. 7d, lanes 4 and 5). Several control antibodies (three of which are shown in Fig. 7d, lane 1 and Fig. 7f, lanes 2 and 3) were tested and none had any effect on the protein–DNA complexes. In addition, a polyclonal FGFR1 antibody that recognizes the C-terminus of the receptor had only a minimal reducing effect on the slowest migrating ‘H’ band (Fig. 7f, lane 5), although this effect was slightly enhanced at a higher concentration of the antibody (not shown). Thus, it appears that the binding of an antibody specific to the extracellular domain of FGFR1 prevents the formation of some of the DNA–protein complexes. The changes in DNA–protein binding induced by transfection of FGFR1 and FGFR1(TK-) were specific for the CRE as FGFR1 or FGFR1(TK-) had no reproducible effect on protein binding to the AP1 site (Fig. 7h). These results further support our hypothesis that nuclear FGFR1 controls protein binding to the CRE region of the TH promoter. Since incubation of nuclear extracts with mAb6 FGFR1 did not yield detectable supershifted bands, it is not clear whether FGFR1 itself associates with the DNA-bound protein complexes or only facilitates their formation.
The proteins that bind to the CRE of several different genes include CREB and the CREB-associated co-factors CBP/p300 (Shaywitz and Greenberg 1999). Binding of CREB to the TH CRE was previously shown (Huang et al. 1991; Kilbourne et al. 1992; Kim et al. 1996). To confirm this, we treated DNA–protein binding reactions with an antibody against the DNA-binding domain of CREB or an antibody against the CREB-binding domain of CBP/p300 (Fig. 7f). CREB is known to bind constitutively to the CRE and, upon cell stimulation, it undergoes phosphorylation which facilitates its interaction with CBP/p300 and the formation multiple protein complexes. A CREB Ab (but not a CBP/P300 Ab) depleted the upper ‘H’ band (Fig. 7f, lane 7) indicating that this band contains CREB but not CBP/p300. The lower ‘H’ band, which contains CBP/p300 (Fig. 7f, lane 6), was also partially depleted by the CREB Ab, consistent with the regulation of CBP/p300 promoter binding by activated CREB (Shaywitz and Greenberg 1999). The slowest ‘L’ band was also displaced by the CREB Ab as well as by the CBP/p300 Ab. Thus, both CREB and CBP/p300 are present in these complexes. The second largest ‘L’ complex was not affected by the CREB or CBP/p300 Abs and thus may contain protein(s) other than CREB or its co-factors. Neutralization of FGFR1 by FGFR1(TK-) or by mAb6 reduced the formation of the slowest CREB-CBP/300-containing ‘L’ complex as well as the ‘L’ complex which lacks CREB-CBP/300. However, mAb6 had little or no effect on the smaller protein–DNA complexes found in the upper and lower ‘H’ bands which represent CREB and CBP/300 protein–DNA complexes, respectively. This suggests that FGFR1 function is essential for the formation of large, multiple protein CRE complexes that include CREB and its co-factors CBP/300, as well as other protein(s), but not for the smaller CREB- or CBP/p300-containing complexes.
The presence of phosphorylated, active CREB facilitates the formation of multiple protein complexes with the CRE (Shaywitz and Greenberg 1999). To determine if nuclear FGFR1 can up-regulate active (phosphorylated) CREB, we performed western blot analyses with an antibody that specifically recognizes phospho-133Ser-CREB. Transfection of TE671 cells with FGFR1 increased the content of phosphorylated CREB (Fig. 7g). A stronger increase, comparable to the increase induced in control cells by PMA, was observed in cells transfected with FGFR1(SP-/NLS).
This study describes a new mechanism involving FGF-2 and its receptor, FGFR1, through which AII, depolarization, and PKC signals are integrated to activate the TH gene via a common CRE transactivator. This new mechanism is demonstrated by: (i) the induction of nuclear FGF-2 and FGFR1 and an up-regulation of TH protein and gene promoter in BAMC by AII, veratridine and PMA in BAMC; (ii) the inhibition of TH protein and gene promoter up-regulation by a dominant negative FGFR1(TK-) mutant; and (iii) the transactivation of the TH gene promoter by ectopically expressed FGF-2 and FGFR1.
The FGF-2-mediated stimulation of the TH gene by AII, depolarization or PKC or by transfected FGF-2 appears to occur in an intracrine fashion: (i) no FGF-2 was detected in the medium conditioned by control or stimulated BAMC (Stachowiak et al. 1994b); (ii) no FGF-2 was detected between cells that might suggest deposition in extracellular matrix; (iii) the increases in the FGF-2 content were observed after the cells were washed to remove extracellular matrix (Stachowiak et al. 1994b); (iv) no FGF-2-IR was associated with the plasma membrane or secretory vesicles (Fig. 1e); (v) the increase in FGF-2 expression was detected predominantly inside the cell nucleus (Fig. 1); (vi) in BAMC, exogenous FGF-2 had a minimal effect on TH promoter activity; (vii) in TE671 cells, in which transfected FGF-2 increased TH promoter activity several-fold, exogenous FGF-2 had no effect on TH promoter activity even after transfection of recombinant FGFR1; (viii) IP6, which in BAMC prevented TH promoter activation by exogenous FGF-2, did not inhibit the activation by transfected HMWFGF-2 or by AII; and (ix) suramin did not inhibit the activation of the TH promoter by HMW-FGF-2. Thus, even if a small amount of FGF-2 is released from BAMC, it does not appear to play a role in the activation of the TH gene. These findings are in an agreement with earlier experiments utilizing IP6 which showed that PMA-induced astrocyte differentiation (Sherman et al. 1993) and the proliferation of glioma cells (Joy et al. 1997) are mediated by intracellularly acting FGF-2.
Bouche et al. (1987), Nakanishi et al. (1992) and Arese et al. (1999) showed that the biological effects of FGF-2 can be produced directly in the nucleus. The transactivation of the TH gene promoter by nuclear HMWFGF but not cytosolic FGF-1 shown in this study further demonstrates the nuclear action of FGF-2. Localization of transfected HMWFGF-2 exclusively in the cell nucleus and 18 kDa FGF-2 in the nucleus as well as in the cytoplasm (Fig. 3a,c) is consistent with earlier observations (reviewed in Szebeneyi and Fallon 1999). The nuclear HMWFGF-2 isoforms are especially abundant in the brain (Stopa et al. 1990; Coffin et al. 1995). In rat AM, HMWFGF-2 were essentially the only FGF-2 isoforms detected and were up-regulated during stress (our unpublished observations). The nuclear presence of FGF-2 was observed in nicotine-stimulated rat substantia nigra neurons (Belluardo et al. 1998), in BMP-7-stimulated sympathetic neurons (Horbinski et al. 2002), in rat hippocampal (Woodward et al. 1992), trigeminal and cerebellar neurons (Matsuda et al. 1992a, 1992b), and in human (Joy et al. 1997) and rat brain (Clarke et al. 2001) astrocytes. Thus, the mode of FGF-2 action in these tissues in vivo is likely to be predominantly nuclear. Unlike FGF-1 or FGF-2, many other members of the FGF family have leader sequences and are effectively released from cells. Thus, the fact that exogenous FGF-2 or FGF-1 can induce some cellular responses in vitro and in vivo may not be surprising since they can activate surface FGFRs which may normally interact with secreted FGFs (Szebeneyi and Fallon 1999).
Stimuli that increase the expression and nuclear accumulation of FGF-2 also increase the nuclear accumulation of FGFR1 (Fig. 1b; Stachowiak et al. 1996a,b, 1997a; Clarke et al. 2001). Furthermore, ectopically expressed FGF-2 up-regulates endogenous nuclear FGFR1 (Fig. 5b,c). The co-accumulation of FGF-2 and FGFR1 in the nuclei of BAMC is associated with phosphorylation of nuclear FGFR1 and up-regulation of its kinase activity (Stachowiak et al. 1996a). The ability of FGFR1(TK-) to block transactivation of the TH gene promoter by HMWFGF-2, shown in the present study, indicates that nuclear FGF-2 transactivates the TH promoter via FGFR1. Thus, FGF-2 and FGFR1 may constitute a nuclear signaling module that controls gene expression. However, the limited co-localization of FGF-2 and FGFR1 in the nucleus (Fig. 1b) suggests that their interaction may occur only briefly and involve a small population of nuclear FGF-2 at any given time. It is also possible that factors other than FGF-2 can activate nuclear FGFR1.
Studies in this and other laboratories have provided strong evidence for the localization of FGFR1 in the nuclei of BAMC, astrocytes, glioma cells (Stachowiak et al. 1996a,b; 1997a,b), neurons (Gonzales et al. 1995; Klimaschewski et al. 1999; Horbinski et al. 2002), fibroblasts (Maher 1996; Steger et al. 1998; Reilly and Maher 2001), and retinal cells (Guillonneau et al. 1998). The nuclear accumulation of endogenous or transfected FGFR1 was shown using western and far western assays with antibodies that recognize different FGFR1 epitopes (Stachowiak et al. 1996a,b; 1997a; see also Figs 1, 4 and 6). The purity of the nuclear fraction is described in Materials and methods. The lack of contamination of nuclear fractions with surface FGFR1 was also shown by treating cells with NHS-sulfobiotin. Biotinylated FGFR1 was only detected in the extranuclear fraction (Fig. 6b). Furthermore, no traces of cytosolic FGF-1 were found in the nuclear fraction of TE671 overexpressing FGF-1 (Fig. 3c) or BAMC expressing endogenous growth factor (not shown). Hence, the presence of FGFR1 in the isolated nuclei was not artifactual. We have also demonstrated the localization of FGFR1 within the nuclear interior using immunocytochemistry with confocal or electron microscopy and antibodies that recognize the C- or N-terminal portions of FGFR1 (Stachowiak et al. 1996a,b; 1997a; the present study). Analysis of confocal images of native EGFP fluorescence showed that, unlike tubulin-EGFP (control), a FGFR1-EGFP fusion protein enters the cell nucleus in transfected cells (Fig. 4c).
In BAMC, the nuclear accumulation of FGFR1 can be induced by heterologous stimuli such as the activation of AII or acetylcholine receptors, adenylate cyclase and PKC (Stachowiak et al. 1996a; present study), but not by incubation with 18 kDa FGF-2 (Stachowiak et al. 1996a). Thus, the FGFR1 molecules that enter the nucleus are unlikely to represent surface receptors internalized due to their interaction with extracellular FGFs. The biotinylation experiment (Fig. 6b) suggests that nuclear FGFR1 was not derived from the cell surface. However, it is also evident that nuclear FGFR1 is processed at least partially through the endoplasmic reticulum-Golgi, as indicated by its glycosylation. Therefore, we hypothesize that the association of FGFR1 with the reticular membranes may not be stable, and that the receptor is released into the cytosol before the endoplasmic vesicles fuse with the plasma membrane. This is consistent with our recent observations that replacement of the transmembrane domain with the membrane spanning segment of an unrelated protein hinders the intranuclear accumulation of FGFR1 (our unpublished observations). Wild-type FGFR1 lacks a typical NLS. Its candidate nuclear chaperones include the 21–24 kDa FGF-2 isoforms, which contain a functional NLS (Szebeneyi and Fallon 1999). In addition, FGFR1 interacts with importin β, a protein known to be involved in the translocation of proteins into the nucleus via nuclear pores (Reilly and Maher 2001). Thus, our findings allow us to dissociate two functions of FGFR1: (i) paracrine or autocrine signaling by the plasma membrane receptor (which might be continued following receptor internalization into the cytoplasm) and (ii) intracrine nuclear signaling by a separate pool of the receptor that is not derived from the plasma membrane but has a direct gene transactivating function. Indeed, the FGFR1(SP-/NLS) mutant that does not insert into cell membranes but is driven exclusively into the cell nucleus by an inserted NLS proved to be a markedly more effective TH gene transactivator than its predominantly cytoplasmic FGFR1(SP-) counterpart (Fig. 5c). Given this result and the inability of the extracellularly acting FGFR antagonists to block TH promoter activation by wild-type FGFR1, we conclude that FGFR1 acts as an intracrine, nuclear activator of TH gene transcription.
AII, cell depolarization, and calcium regulate the TH gene promoter activity through the CRE. Therefore, the finding that FGFR1 also transactivates the TH promoter via the CRE is consistent with the function of FGFR1 as a mediator of the actions of these agents. The nuclear localization of FGFR1 suggests that FGFR1 may directly transactivate the CRE in the cell nucleus. Indeed, the stimulation of surface FGFR1 with exogenous FGF-2 had little or no effect on protein binding to the CRE or on TH promoter activity. In contrast, nuclear HMWFGF-2 activated both the TH promoter and CRE binding and its effects were blocked by FGFR1(TK-).
CREB and CBP/p300 are well known transactivators of the CRE sequence common to many different genes (Shaywitz and Greenberg 1999). CBP and the closely related p300 bind to DNA as well as to activated (phosphorylated) CREB and other transcriptional factors, bridge these factors with the RNA polymerase II complex, and stimulate transcription. We show that the in vitro formation of large, multiprotein–CRE complexes containing CREB, CBP/p300 and other, as yet to be identified protein(s), is regulated by nuclear FGFR1. This was demonstrated by the stimulation of protein binding to the TH CRE by nuclear HMWFGF-2 and nuclear FGFR1(SP-/NLS) which correlated with the up-regulation of TH promoter activity. The observation that neutralization of nuclear FGFR1 by mAb6 reduced protein binding to the CRE in a manner similar to the co-transfection of FGFR1(TK-) further indicated that nuclear FGFR1 controls the assembly of CRE-associated protein complexes. This mechanism may involve the up-regulation by nuclear FGFR1 of phosphorylated CREB (Fig. 7g).
FGFR1 is not needed for the cAMP-dependent activation of the TH promoter but is essential for the activation by AII, veratridine or PMA which are known to act through changes in intracellular Ca+2 and/or PKC activity. Stimulation of the TH promoter by Ca+2 and PKC activators also involves CREB and CBP/p300, although serine/threonine kinases different from PKA may be involved as well (see Introduction). One such kinase, pp90 RSK-1 (Nakajima et al. 1996), interacts directly with the tyrosine kinase domain of nuclear FGFR1 (Y. Hu et al., manuscript in preparation). Thus, whether FGFR1 is involved in the transduction of a CRE-targeting signal may depend on its ability to interact with the kinase that phosphorylates CREB and/or its co-factors.
Using electron microscopy we observed that in stimulated BAMC FGFR1 localizes specifically in the granules associated with the nuclear matrix (Stachowiak et al. 1996a). Such granules, termed interchromatinic granules, are the presumed EM equivalent of the extranucleolar speckles identified with the antispliceosome antibodies, SC-35 and Y12 (Berezney 1980; Lerner et al. 1981; Fu and Maniatis 1990; Wei et al. 1999). The SC-35/Y12 compartment consists of 0.5–3 µm domains (speckles) enriched in numerous splicing factors, poly A RNA, a subset of poly(A) polymerase, RNA helicase, and a hyperphosphorylated form of RNA Pol II (reviewed in Wei et al. 1999). The speckles are enriched several-fold in active transcription sites compared to non-speckled extranucleolar regions with over 90% of the SC35/Y12 speckles exhibiting moderate to high levels of transcriptional activity (Wei et al. 1999). The SC35/Y12 speckles have also been implicated as sites for the dynamic recruitment of splicing factors at transcription sites. The present study shows that in stimulated BAMC, nuclear FGFR1 is distributed in a pattern which overlaps three-dimensionally with the Y12 speckles. The accumulation of nuclear FGFR1 within sites of RNA transcription and processing suggests that nuclear FGFR1 may control the expression of multiple genes. Consistent with this model is the recently described activation of another gene, FGF-2, by nuclear FGFR1 (Peng et al. 2001). This activation is mediated by a FGF-2 gene promoter element that is distinct from the TH CRE and does not compete for CRE-binding factors (Moffett et al. 1998). Thus, by being induced by a variety of heterologous signals, the INFS may constitute a common integrative pathway through which hormones and neurotransmitters, second messengers, and direct cell–cell interactions execute control over multigene programs for cellular adaptations, growth, and differentiation.
HP, JM, XF and EKS contributed similarly to this paper. We thank Dr Wade Sigurdson (Confocal Microscope and 3D Imaging Facility, School of Medicine and Biomedical Sciences, SUNY at Buffalo, NY, USA) for the assistance with confocal microscopy and Claudia Prada for the help with preparation of the manuscript. pExNeo and the plasmid expressing FGF-1 were the kind gift of Dr Toru Imamura. This study was supported by grants from the National Science Foundation (IBN-9728923), National Institutes of Health (HL-49376), and Parkinson's Disease Foundation (to M.K.S). P.A.M. was supported by the NIH (GM54604). Correspondence should be addressed to M.K.S.