Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

Nitric oxide (NO) is an important intracellular and intercellular mediator involved in the modulation of many physiological processes in different tissues, including blood flow regulation, platelet aggregation, smooth muscle relaxation, apoptosis, central and peripheral neurotransmission, and different neuroendocrine responses (Moncada and Higgs, 1993; Nathan and Xie, 1994).

NO is synthesized by a family of three distinctive isoforms of nitric-oxide synthase (NOS), named after the tissues in which they were originally described. Neuronal and endothelial NOS [nNOS (or NOS I) and eNOS (or NOS III), respectively] are Ca²⁺/calmodulin-dependent enzymes constitutively expressed not only in neuronal and endothelial cells but also in muscle cells, fibroblasts, and various epithelial cells (Nathan and Xie, 1994). They release NO for short periods of time in response to receptor stimulation (Moncada et al., 1991). Inducible NOS [iNOS (or NOS II)] is mainly expressed in macrophages and other blood cells, astrocytes, microglia, and endothelial cells and is transcriptionally activated, in a Ca²⁺/calmodulin-independent manner, by cytokines and other pro-inflammatory agents (Moncada et al., 1991). Conversely, the activation of the constitutive forms of NOS (cNOS) is almost always strictly Ca²⁺/calmodulin-dependent; an increase in intracellular Ca²⁺ concentration ([Ca²⁺]_i) induces the formation of active Ca²⁺-calmodulin complexes that bind with high-affinity specific domains of these enzymes (Moncada et al., 1991; Nathan and Xie, 1994). Beside changes in [Ca²⁺]_i , which represent the crucial cofactor for the activation of cNOS, after receptor-dependent stimulation, different protein kinases and phosphatases may also regulate the activity of these enzymes (Garcia-Cardena et al., 1996; Corson et al., 1996; Fleming and Busse, 1999).

In their inactive configuration, cNOS are bound to specific plasma membrane proteins, from which they are released upon their activation. In particular, nNOS is kept localized to cell membrane compartments by its interaction with the 95-kDa postsynaptic density protein in neurons and to -syntrophin in muscle cells (Brenman et al., 1996), whereas, in resting cells, eNOS is localized, through myristoylation and palmitoylation, to particular structures of the plasma membrane called caveolae (Feron et al., 1996; Fleming and Busse, 1999). Caveolae are membrane micro-compartments involved in the regulation of endocytosis, Ca²⁺ homeostasis and clustering of GPI-anchored proteins. More recently, caveolae have also been proposed to represent specialized sites for the interaction among signal transduction effectors (Anderson, 1998). Caveolae have a distinct lipid composition because they are highly enriched in sphingolipids, among them ceramides, a class of sphingolipids that act as intracellular second messengers (Anderson, 1998). Ceramides are formed by acylation of sphingosine, induced by the enzyme ceramide synthase, or formed by hydrolysis of sphingomyelin operated by a family of sphingomyelinases (Kolesnick and Fuks, 1995; Hannun, 1996). As far as protein composition, caveolae are characterized by the presence of a family of three integral membrane proteins named caveolin 1, 2, and 3 (Anderson, 1998). Caveolin 1 is surely the most important protein involved in the three-dimensional structure and function of caveolae. In resting cells, caveolin 1 binds the inactive form of eNOS, preventing the catalytic activity of the enzyme and keeping it localized in the caveolae (Garcia-Cardena et al., 1996; Fleming and Busse, 1999). Upon cellular activation (i.e., increase in Ca²⁺ concentration), eNOS is released from caveolin-1 and translocates to the cytosol, where it dimerizes, binds its substrate L-arginine, and releases NO and L-citrulline. The cycle is completed by the eventual return of eNOS to the cell membrane as inactive enzyme (Fleming and Busse, 1999).

Recently, a novel Ca²⁺/calmodulin-independent mechanism for eNOS activation has been described in mammalian endothelial cells (Igarashi et al., 1999). Exogenously administrated ceramide can activate eNOS in a Ca²⁺-independent way. Thus, it was proposed that compounds able to activate the synthesis of ceramide might be regulators of eNOS activity.

Basic fibroblast growth factor (bFGF) is a powerful mitogen for most cell types, including Chinese hamster ovary fibroblasts (CHO-K1), and MAP kinase cascade has been demonstrated to represent a major transduction mechanism for this growth factor-induced cell proliferation. CHO-K1 cell duplication dramatically decreases if this enzymatic cascade is blocked, inhibiting MEK activity (Florio et al., 1999a). Through the regulation of endothelial cell proliferation and migration, bFGF represents also one of the major angiogenic factors produced by different types of tumors including most of human glioblastomas (Schmidt et al., 1999). Angiogenesis is now regarded as one of the most important tumor-mediated events responsible for tumor growth and dissemination (Folkman, 1995). NO production represents one of the main intracellular mechanisms responsible for tumor-dependent neoangiogenesis (Ziche et al., 1997). It has been demonstrated that NO, among its numerous intracellular functions, can stimulate cell proliferation in endothelial (Ziche et al., 1997) but also in CHO-K1 cells (Cordelier et al., 1997), mainly through the activation of the cGMP/protein kinase G (PKG) pathway (Cordelier et al., 1997). Although the role of NO in the activity of many angiogenic factors, such as vascular endothelial growth factor, is well defined (Ziche et al., 1997), the capability of bFGF to induce this intracellular second messenger is still controversial.

The aim of this study was to evaluate whether bFGF is able to activate the eNOS and to study the cellular and molecular mechanisms that mediate such an effect. Moreover, we studied the correlation between NO production induced by bFGF and its proliferative activity in CHO-K1 cells.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Materials

The following reagents were purchased as indicated: N-(1-naphthyl)ethylene diamide, sulfanilamide, and sodium nitroprusside (Sigma, St. Louis, MO); tricyclo-decan-9-yl xanthate (D609), Ly-83583, 8-bromo-cGMP, BAPTA/AM, KT 5823, S-methyl-isothiourea sulfate, L-N⁵-(1-iminoethyl)-ornithine dihydrochloride (L-NIO), cholecystokinin (CCK), N-NO₂-L-arginine (NNA), and N^G-nitro-L-arginine methyl ester hydrochloride (L-NAME) (Calbiochem, San Diego, CA); N-acetyl D-erythro-sphingosine, dihydro-N-acetyl D-erythro-sphingosine, N,N-dimethyl-sphingosine, fumonisin B1, and U 73122 (Alexis, Läufelfingen, Switzerland); [methyl-³H]thymidine and L-[³H]arginine (Amersham Biosciences, Piscataway, NJ); vanadate (ICN, Costa Mesa, CA); and sodium pyrophosphate (Merck, Whitehouse Station, NJ).

Antibodies

Anti-eNOS, -nNOS, -iNOS, and anti-caveolin 1 were from Santa Cruz Biotechnology (Santa Cruz, CA); anti-ERK1/2, phospho-ERK1/2, anti-phospho(Ser¹¹⁷⁹)-eNOS, anti-AKT, and phospho(Ser⁴⁷³)-AKT were from New England Biolabs (Beverley, MA)

Cell Culture

CHO-K1 were cultured under sterile conditions in Ham's F-12 medium (Invitrogen, Carlsbad ,CA) supplemented with 10% fetal calf serum (Invitrogen). Primary cultures of human skin fibroblasts from healthy subjects were obtained as described previously (Thellung et al., 1999). Fibroblasts from patients with Niemann-Pick disease were obtained from the "Laboratorio di Diagnosi PrePostnatale Malattie Metaboliche" (Istituto G. Gaslini, Genova Italy) using specimens from the collection "Cell lines and DNA bank from patients affected by genetic diseases". Three independent preparations were used. Both fibroblasts cultures were performed in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum (Invitrogen). Pertussis toxin (180 ng/ml) treatment was performed in serum-free medium for 18 h before the experimental treatments.

Determination of NO Production

Determination of Nitrite Accumulation Based on the Griess Reaction. NO, produced from the conversion of L-arginine to L-citrulline, is mainly oxidized to nitrite, which is an indicator of NO synthesis.

Determination of the L-[³H]Citrulline Accumulation. The conversion of L-arginine in L-citrulline was monitored by measuring the production of L-[³H]citrulline after incubation of the extract cytosolic fraction with L-[³H]arginine, using the Stratagene kit (Stratagene, La Jolla, CA), following the manufacturer's instructions. Briefly, cells were mechanically homogenized in a buffer containing 24 mM Tris-HCl, pH 7.4, 1 mM EDTA, and 1 mM EGTA, the particulate fraction was pelleted (14,000g, 5 min), and the cytosolic fraction was collected and incubated (10 mg/ml) in a reaction buffer (25 mM Tris HCl, pH 7.4, 1 µM flavin adenine dinucleotide, 1 µM flavin adenine mononucleotide, and 3 µM tetrahydrobiopterin) to which 1.2 mM NAPDH, 0.25 µCi of L-[³H]arginine, and 750 µM CaCl₂ have been added. The reaction was carried out for 60 min and then stopped with 400 µl of 50 mM HEPES, pH 5.5, and 5 mM EDTA. The formed [³H]citrulline, derived from the NO production, was then recovered by chromatography and measured in a -counter.

[³H]Thymidine Incorporation Assay

DNA synthesis was quantified using the [³H]thymidine incorporation assay. Cells were seeded into 24-multiwell plates at a density of 5 × 10⁵/well. After 24 h, cells were serum-deprived for 24 h and then treated with the test substances for 16 h. In the last 4 h of treatment, cells were pulsed with 1 µCi/ml of [³H]thymidine. Cells were collected by trypsin treatment for 5 min at 37°C and then filtered under vacuum through glass-fiber filters (GF/A; Whatman, Clifton, NJ). Unincorporated labeled nucleotides were removed by sequential 10 and 5% trichloroacetic acid (Sigma) and 95% ethanol washes. The filters containing the trichloroacetic acid-insoluble fraction were measured for radioactivity in a scintillation counter.

Immunofluorescence

The subcellular localization of eNOS and caveolin 1 was evaluated by indirect immunofluorescence experiments. Cells, plated on glass cover slips, were fixed in 4% paraformaldehyde for 15 min. After three washes in PBS, cells were permeabilized with 0.1% Triton X-100 for 5 min, washed again with PBS, and treated with PBS-glycine (1 M) for 10 min. Cells were stained with the primary antibody in PBS/0.1% FBS for 1 h, washed again three times with PBS, and then incubated with anti-rabbit fluorescein isothiocyanate-conjugated or anti-rabbit rhodamine-conjugated secondary antibody (Jackson ImmunoResearch Laboratories, West Grove, PA) (1:200) for 20 min. After washing in PBS and deionized H₂O, coverslips were mounted using Moviol (Calbiochem) and analyzed on a confocal microscope (argon laser excitation at 468 and 580 nm; 60× objective, MRC 1000; Bio-Rad, Hercules, CA) with a 0.4-µm step in z-plane acquisition. Each experiment was performed at least three times in duplicate.

Immunoprecipitation and Western Blot

CHO-K1 cells were plated in 10-cm Petri dishes until they reached 75% confluence. Then they were treated for the specified times with the test reagents. After two washes in PBS, cells were harvested in ice-cold lysis buffer containing 100 nM Tris-HCl, pH 8, 150 mM NaCl, 1% Nonidet P40, 0.4 mM EDTA, 10 mM NaF, 2 mM vanadate, 10 mM sodium pyrophosphate, 0.1 mg/ml phenylmethylsulfonyl fluoride, and a protease inhibitor mix (Complete; Roche Diagnostics, Mannheim, Germany). Soluble proteins (500 µg/ml) were incubated for 2 h at 4°C with anti-caveolin 1 antibody (1 µg/mg of protein) with continuous rotation. The samples were then incubated for 1 h at 4°C, with continuous rotation, with 50 µl of Sepharose-protein A (Sigma) to pellet the immunocomplexes. After three washes in the same lysis buffer, the immunoprecipitated proteins were resolved through a 10 or 15% SDS-polyacrylamide gel, transferred to a polyvinylidene difluoride membrane (Bio-Rad), and probed with the primary antibody. Immunoreactive proteins were visualized by the enhanced chemiluminescence immunodetection system (Amersham Biosciences).

Intracellular Calcium Measurement

Cells were plated on 25-mm glass coverslips and transferred to 35-mm Petri dishes. After 24 h, cells were serum-starved for a further 24 h. On the day of the experiment, cells were washed for 10 min with a balanced salt solution (HEPES 10 mM, pH 7.4, 150 mM NaCl, 5.5 mM KCl; 1.5 mM CaCl₂, 1.2 mM MgSO₄, and 10 mM glucose). Then cells were loaded with 4 µM Fura-2 penta-acetoxymethyl ester (Calbiochem) for 20 min at room temperature. Fluorescence measurements were performed as reported previously (Florio et al., 1999b). Briefly, Fura-2 fluorescence was imaged with an inverted Nikon diaphot microscope using a Nikon 40×/1.3 numerical aperture Fluor DL objective lens. Fluorescence (ratio 340/380) was then evaluated and converted in [Ca²⁺]_i using the Quanticell apparatus (VisiTech, Sunderland, UK). For the calibration of fluorescence signals, we used cells loaded with Fura-2; R_max and R_min are ratios at saturating and zero [Ca²⁺]_i, respectively, and were obtained by perfusing the cells with a salt solution containing 10 mM CaCl₂, 2.5 µM digitonin, and 2 µM ionomycin and subsequently with a Ca²⁺-free salt solution containing 10 mM EGTA. The values of obtained R_max and R_min were used to calculate the [Ca²⁺]_i using the Quanticell software, according to the equation of Grynkiewicz et al. (1985).

cGMP Detection

cGMP levels were measured in cell extract from confluent cells seeded in six multiwell plates and treated for the indicated times with the test substances in the presence of 500 µM IBMX as phosphodiesterase inhibitor. Cells were then extracted for 24 h at 4°C with 99% ethanol/1% HCl (v/v), and the supernatant was collected for the quantification of the cGMP by high-performance liquid chromatography (HPLC). Before the analysis, the samples were centrifuged and filtered through a nylon-66 filter, 0.2 µm (Rainin Instrument, Woburn, MA). The clear filtrate obtained was used directly for HPLC assay, as described previously (Spoto et al., 1991), or stored at 80°C until the assay.

Chromatographic Apparatus. The HPLC system (Beckman Coulter, Fullerton, CA) consisted of a two 110A pumps, a variable wavelength spectrophotometer (Spectroflow 783; Kratos Analytical) measuring at 254 nm; and an autosampler Promis (Spark Holland, Emmen, the Netherlands).

Chromatographic Conditions. The column used was a 5-µm Li-Chrospher 100CH 18/2 (250 × 4 mm) (Merck). The mobile phase employed for the separation of nucleotides consisted in 200 mM ammonium acetate, pH 6.0, with 2% acetonitrile (v/v). The flow rate was 1 ml/min; the detection was performed at 254 nm. Peak identities were confirmed by coelution with standards. Quantitative measurements were carried out by comparison using standard solutions of known concentrations.

Ceramide Measurement

After appropriate treatments, cells were washed in PBS and then harvested by scraping in 3 volumes of lysis buffer (0.5 M sucrose, 20 mM Tris HCl, pH 6.8, 150 mM NaCl, 3 mg/ml leupeptin, 0.5 mM phenylmethylsulfonyl fluoride, and 0.5 mM dithiothreitol) and normalized for protein content using the the Bradford method (Bradford, 1976). After sonication, total lipid content was extracted for 1 h in chloroform/methanol (2:1, v/v) containing butylated hydroxyanisole as antioxidant (Folch et al., 1957). After a brief centrifugation, the lower phase was collected, dried under nitrogen stream, and stored at 80°C until the assay. Samples were then processed for basic hydrolysis to remove other lipid species, dissolved in dimethyl sulfoxide, and then analyzed by high-performance liquid chromatography-electrospray ionization-mass spectrometry, as reported previously (Kalhorn and Zager, 1999) with modifications. We used a Hewlett Packard 1090 series II liquid chromatography system directly coupled with a Hewlett Packard 5989A "Engine" single-quadrupole mass spectrometer. An Alltech Adsorbosphere XL C8 300A 5µ column (250 × 4.6 mm) was used and an isocratic mobile phase of 20 mM methanol/acetic acid (90:10) at a flow rate of 0.8 ml/min was applied. The mass range was set to 100 to 800 atomic mass units, and the signal was optimized to observe the monosodium adduct of the molecular ion in the positive ion mode. The capillary exit voltage was adjusted to optimize the expected m/z ratio.

Statistics

Experiments involving NO and cGMP measurements or cell proliferation were performed in quadruplicate. All the experiments were repeated at least three times. Statistical analysis was performed by means of one-way analysis of variance; p  0.05 was considered statistically significant.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

bFGF Effects on NO Production in CHO-K1 Cells. bFGF treatment (30 ng/ml) of CHO-K1 cells caused time-dependent increases in production of both NO (measured using the Griess reagent) and cGMP (assessed by HPLC experiments) that were detectable starting after 1 h of stimulation and then followed by a constant increase for longer treatments, with a maximum after 24 h (data not shown). To maximize the cell response to bFGF, all the subsequent experiments were performed after 24 h of treatment. In Fig. 1A, the effect of different bFGF concentrations on NO production is shown. bFGF, dose dependently induced NO production starting at the concentration of 5 ng/ml. The increase in NO, in turn, activated a soluble guanylyl cyclase to stimulate cGMP production, the intracellular concentration of which was indeed significantly increased (by about 300%) (Fig. 1B). Similar results were also obtained by measuring the NO production as conversion of L-[³H]arginine in L-[³H]citrulline (basal, 19,734 ± 425 cpm/well; bFGF, 33,446 ± 759 cpm/well; p < 0.01). It was reported previously that at least two isoforms of NOS are expressed in CHO-K1 cells (i.e., nNOS and eNOS) (Cordelier et al., 1999). We confirmed these data by reverse transcription-polymerase chain reaction (data not shown) and Western blot analysis, which identified both cNOS isoforms in these cells (Fig. 2A). Conversely, iNOS was detected under neither basal nor bFGF-stimulated conditions (Fig. 2A). These data were also confirmed using a pharmacological approach. S-Methyl-isothiourea, a rather selective iNOS inhibitor (Szabo et al., 1994), reduced bFGF-dependent NO production only slightly, whereas the L-NIO compound that affects eNOS with higher affinity than the other NOS isoforms (Rees et al., 1990) reduced the NO production to a level similar to that obtained with the powerful but nonspecific NOS inhibitors L-NAME and NNA (Rees et al., 1990) (Fig. 2B). These results suggest that bFGF may induce NO synthesis through the activation of eNOS.

View larger version (13K): [in this window] [in a new window]   Fig. 1.   A, dose-response of bFGF on NO production evaluated with the Griess reaction. Cells were treated for 24 h with bFGF (1-100 ng/ml). Data are expressed as nitrate/nitrite accumulation measured as optical density (absorbance at 570 nm). B, bFGF (30 ng/ml, for 24 h) induced cGMP accumulation, measured in HPLC experiments, in the presence of the the phosphodiesterase inhibitor IBMX. **, p < 0.01 versus respective basal values

View larger version (21K): [in this window] [in a new window]   Fig. 2.   A, Western blot analysis showing the expression of eNOS and nNOS in CHO-K1 cells. The inducible isoform of the enzyme is not detectable in these cells under either basal or bFGF-stimulated conditions. C6 glioma cells after LPS stimulation (10 µg/ml, 6 h) were used as a positive control for iNOS expression. The apparent molecular masses of the bands shown are: 140 kDa for eNOS, 160 kDa for nNOS, and 130 kDa for iNOS. B, the effect of nitric oxide synthase inhibitors, selective for different isoforms of the enzyme, on bFGF-induced NO production. Cells were stimulated with bFGF (30 ng/ml, 24 h) in the absence or presence of S-methyl isothiourea sulfate (SMT; a selective inhibitor of iNOS), L-NIO (which affects eNOS with higher affinity), and the nonspecific NOS inhibitors NNA and L-NAME. Data are expressed as nitrate/nitrite accumulation measured as optical density (absorbance at 570 nm). **, p < 0.01 versus the basal value; °°, p < 0.01 versus bFGF-stimulated value

bFGF Activation of eNOS Is Ca²⁺-Independent. Both cNOS isoforms have been reported to be activated in a Ca²⁺-dependent manner. Thus we tested whether, in CHO-K1 cells, bFGF-induced NO production was the result of an increase in [Ca²⁺]_i. We pretreated the cells with increasing concentrations (3-100 µM) of the cell-permeable Ca²⁺ chelator BAPTA-AM and tested, under these experimental conditions, the effects of bFGF on NO production. However, bFGF (30 ng/ml), in the presence of the highest BAPTA-AM concentration, was still able to induce NO synthesis to levels comparable with those observed in control cells (Fig. 3A). As an internal control, we evaluated the effects of CCK, a peptide known to activate the NOS in a Ca²⁺-dependent manner through the activation of the CCK-A receptor subtypes, which are expressed in the CHO-K1 cells (Cordelier et al., 1999). CCK (1 µM) induced a significant increase in NO production that was completely blocked by the pretreatment with BAPTA-AM (Fig. 3A). Moreover, in microfluorometric experiments, we demonstrated that bFGF treatment does not modify [Ca²⁺]_i even at the high concentrations (100 ng/ml) that occur when treating the cells with CCK (1 µM) (Fig. 3B). To identify the possible intracellular mechanisms involved in bFGF effects, we tried to inhibit the NO production induced by this growth factor by blocking different signal transduction pathways known to be activated by FGF receptors. However, all the compounds that we tested [the inhibitors of MEK, phosphatidyl inositol-3 kinase, phospholipase C (PLC), and protein kinase C: PD98059, wortmannin, U73122, and staurosporine, respectively] did not affect the NO production caused by 30 ng/ml bFGF (data not shown).

View larger version (16K): [in this window] [in a new window]   Fig. 3.   A, effect of increasing concentrations of the cell permeable intracellular Ca2+ chelator, BAPTA-AM, on basal, CCK-, and bFGF-induced NO production. Data are expressed as nitrate/nitrite accumulation measured as optical density (absorbance at 570 nm). B, effects of bFGF and CCK on the intracellular calcium concentration in CHO-K1 cells, measured using the fluorescent probe Fura 2, in microfluorometry. At least 15 cells per field were analyzed, and the data are expressed as mean values.

bFGF Activation of eNOS Is Mediated by a Sphingomyelinase-Dependent Ceramide Production. Recently, a Ca²⁺-independent regulation of NO production by eNOS was identified that is mediated by exogenously administered ceramide in endothelial cells (Igarashi et al., 1999).

View larger version (17K): [in this window] [in a new window]   Fig. 4.   Effect of increasing concentrations (10-40 µM, 3 h) of the cell-permeable C2:0 ceramide and its inactive immediate precursor dihydroceramide on NO synthesis. Data are expressed as nitrate/nitrite accumulation measured as optical density (absorbance at 570 nm). **, p < 0.01 versus the basal value

View larger version (42K): [in this window] [in a new window]   Fig. 5.   A, Western blot analysis of the expression of eNOS in primary cultures of human skin fibroblasts derived from patients with Niemann-Pick disease (NP) or healthy subjects (norm), in comparison with the expression in CHO-K1 cells. All the fibroblast cultures analyzed express a significant amount of eNOS protein. The apparent molecular mass of the bands shown is 140 kDa. B, effect of inhibitors of ceramide production D609 (20 µM), desipramine (50 µM), or fumonisin B1 (40 µM), on bFGF (30 ng/ml)-induced NO synthesis in CHO-K1 cells and primary cultures of human skin fibroblasts derived from patients with Niemann-Pick disease (NP) or healthy subjects (norm). Data are expressed as a percentage of the basal nitrate/nitrite accumulation measured as optical density (absorbance at 570 nm). **, p < 0.01 versus respective basal value; °°, p < 0.01 versus respective bFGF-stimulated values. C, effect of D609 (20 µM) of bFGF (30 ng/ml)-induced cGMP synthesis in CHO-K1 cells, measured in HPLC experiments, in the presence of the phosphodiesterase inhibitor IBMX. *, p < 0.05 versus bFGF value. D, effect of bFGF (30 ng/ml), C8:0 ceramide (10 and 20 µM) and dihydroceramide (DH-cer) (20 µM) on fibroblasts derived from patients with Niemann-Pick disease (NP). Data are expressed as a percentage of the basal nitrate/nitrite accumulation measured as optical density (absorbance at 570 nm). **, p < 0.01 versus basal value.

View larger version (21K): [in this window] [in a new window]   Fig. 6.   Extracted ion chromatograms of molecular ion (monosodium adducts) from ceramide standards (A), control cells (B), and 30 ng/ml bFGF-treated cells (C). Ceramide standards used in A are: a, C6:0 = m/z 420.6; b, C8:0 = m/z 448.6; c, C14:0 = m/z 532.6; d, C16:0 = m/z 560.8.

bFGF Induces a Cytosolic Translocation of eNOS in a Ceramide-Dependent Manner. To delve deeper into the characterization of the mechanisms by which bFGF regulates the production of NO, we analyzed the effect of bFGF treatment on the intracellular localization of eNOS. Indeed, the inactive form of this enzyme is bound to the membrane in caveolae but, upon activation, it translocates to the cytosol (Fleming and Busse, 1999). By means of indirect immunofluorescence and confocal microscopy analysis, we found that, in basal conditions, most of the eNOS was localized in the membrane with a characteristic distribution in patches, probably corresponding to the caveolae (Fig. 7, top center). Indeed, using an antibody directed against caveolin 1, the major protein constituting the caveolae, we obtained a similar localization pattern of the fluorescence signal (Fig. 7, top left). After merging the fluorescence signals, a clear colocalization of the two proteins was observed (Fig. 7, top right). In bFGF-stimulated cells, eNOS localization showed a much more diffuse signal, mainly in the cytosol and around the nucleus, indicating that the enzyme was released from the caveolae, and translocated to the cytosol in the active form (Fig. 7, bottom center). Indeed, in the bFGF-stimulated conditions, the colocalization between eNOS and caveolin 1 was no longer observed (Fig. 7, bottom right). Conversely, no significant changes in nNOS intracellular localization were observed (data not shown). The blockade of the sphingomyelinase activity obtained using the compound D609 (20 µM) caused a significant decrease in the bFGF-induced translocation of eNOS, showing an immunofluorescence image very similar to that of the control cells (data not shown), confirming that the eNOS activation by bFGF was dependent on the ceramide production. A further analysis of the role of eNOS in the NO production induced by bFGF was performed by evaluating in Western blots the level of eNOS after immunoprecipitation with antibodies directed against caveolin 1, as reported previously (Feron et al., 1996). bFGF and C2:0 ceramide treatments significantly reduced the amount of eNOS bound to caveolin 1 (Fig. 8). Moreover, the pretreatment with D609 caused a significant reduction in the bFGF effect (Fig. 8). As a control, we demonstrated that in all the samples, the same amount of caveolin 1 was immunoprecipitated (Fig. 8). Moreover, by comparing on Western blots and in densitometric analysis the amount of eNOS immunoprecipitated from 100 µg of total cell lysate using the anticaveolin 1 antibody and the amount of eNOS present in 100 µg of CHO-K1 lysate, we found that more than 70% of the total eNOS present in untreated cells was immunoprecipitated bound to caveolin 1(data not shown). These data confirm that the changes we observed after bFGF and ceramide treatments represent a quantitatively important protein-protein dissociation.

View larger version (67K): [in this window] [in a new window]   Fig. 7.   Representative immunofluorescence pictures of eNOS localization in CHO-K1 cells using the specific antibodies directed against caveolin 1 (red) or eNOS (green). The merge of the two signals is also reported. Top, immunofluorescence signals in control cells; eNOS is almost completely anchored to the plasma membrane, showing a characteristic disposition in "patches". A fluorescence imagine comparable with what was observed using the eNOS antibody is also observed in control cells incubated with the anti-caveolin 1 antibody. The yellow color in the merge panel indicates the colocalization of the two proteins. Bottom, immunofluorescence signals in bFGF-treated cells. Membrane localization of eNOS is completely lost in cells treated for 1 h with bFGF (30 ng/ml), and the enzyme is spread in the cytosol, mainly in the perinuclear region. The absence of yellow in the merge panel indicates that dissociation between eNOS and caveolin 1 occurred. All the images were acquired using identical brightness parameters and contrast settings.

View larger version (22K): [in this window] [in a new window]   Fig. 8.   Colocalization of eNOS and caveolin 1 in CHO-K1 cells. Cells were plated in 10-mm dishes and left untreated (lane 1) or treated for 1 h with bFGF (30 ng/ml) (lane 2), bFGF + D609 (20 µM) (lane 3), or C2:0 ceramide (20 µM) (lane 4). Equal amounts of proteins from total cell lysate (300 µg) were immunoprecipitated (IPT) with anti-caveolin 1 antibody. Western blot (WB) was performed using either anti-eNOS (top lanes) or anti-caveolin 1 (bottom lanes) antibodies, in two parallel gels. In the graph is reported the ratio of the densitometric analysis of the above depicted gels, expressed as 100% of the control cells. The experiments have been repeated three times with superimposable results.

bFGF Activation of eNOS via Ceramide Production Does Not Involve Sphingosine 1-Phosphate Production. Besides its direct effects, ceramide may act through its metabolite sphingosine 1-phosphate (S1P), a biologically active sphingolipid that has been implicated in intra- and intercellular signaling via the activation of the EDG-1, EDG-3, and EDG-5 G-protein coupled receptors (Spiegel and Milstien, 2000). In particular, it was recently reported that the EDG-1 receptor activation by S1P may induce NO production via the Akt-dependent phosphorylation of eNOS (Ser¹¹⁷⁹) (Igarashi et al., 2001). Thus, we evaluated whether ceramide, produced after bFGF treatment, may activate eNOS through its metabolism to S1P. We measured the increase in NO production induced by bFGF in the presence of the sphingosine kinase inhibitor N,N-dimethyl-sphingosine (DMS) (Edsall et al., 1998) to block the generation of S1P or, after pretreatment with pertussis toxin (PTX), to uncouple the EDG receptors from the G protein. However, neither treatment modified the bFGF effect (Fig. 9, A and B). Then we tested whether the phosphorylation of eNOS at Ser¹¹⁷⁹ by Akt was also relevant for the bFGF-dependent NO production. Western blot analysis using phospho-specific antibodies for both Akt and eNOS showed that bFGF treatment induced a slight and short-lasting phosphorylation of Akt (present after 2 min of treatment and completely abolished after 15 min) (Fig. 10A), whereas no effects were identified on eNOS phosphorylation after either 2 (data not shown) or 15 min of treatment (Fig. 10B), thus excluding the involvement of this pathway from the bFGF effects in CHO-K1 cells. Conversely, as internal control, we demonstrated that, in agreement with recent studies (Montagnani et al., 2001), insulin caused a more robust and long-lasting (still present after 15 min of treatment) activation (Ser⁴⁷³ phosphorylation) of Akt and eNOS phosphorylation (Fig. 10, A and B).

View larger version (47K): [in this window] [in a new window]   Fig. 9.   Effect of the sphingosine kinase inhibitor DMS (A) and pertussis toxin (PTX) on bFGF-induced NO production. A, cells were treated with bFGF (30 ng/ml, 24 h) in the absence or presence of DMS; NO synthesis was evaluated by measuring the conversion of L-[3H]arginine in L-[3H]citrulline. B, cells were pretreated with PTX (200 ng/ml, 18 h), treated with bFGF (30 ng/ml, 24 h), and NO synthesis was evaluated by measuring the conversion of L-[3H]arginine in L-[3H]citrulline. Data are expressed as percentage of basal activity. **, p < 0.01 versus respective basal value

View larger version (22K): [in this window] [in a new window]   Fig. 10.   A, effects of bFGF (30 ng/ml, 2 and 15 min) and insulin (170 nM, 2 and 15 min) on Akt activation, evaluated as Ser473-phosphorylation in Western blot, using a phosphospecific antibody. Equal protein loading was demonstrated by probing the membrane with an antibody directed against the total Akt protein. B, effects of bFGF (30 ng/ml, 15 min) and insulin (170 nM, 15 min) on Ser1179-phosphorylation of eNOS in Western blot, using a phosphospecific antibody. Equal protein loading was demonstrated by probing the membrane with an antibody directed against the total eNOS protein.

Role of NO Produced after bFGF Treatment in CHO-K1 Cell Proliferation. We investigated a possible physiological role for this novel signaling pathway activated by bFGF, involving the synthesis of ceramide, NO, and cGMP.

View larger version (14K): [in this window] [in a new window]   Fig. 11.   A, effect of the NO donor sodium nitroprusside (SNP; 1 µM) and 8-bromo-cGMP (1 µM) on CHO-K1 cell proliferation measured using the [3H]thymidine incorporation assay. SNP effects were measured in the absence or presence of the guanylyl cyclase inhibitor Ly-83583 (LY; 10 nM) and the PKG inhibitor KT 5823 (KT; 100 nM). Cells (50,000/well) were serum-starved for 24 h and then treated for 18 h. In the last 4 h of treatment, cells were pulsed with 1 µCi/ml of [3H]thymidine. Data are expressed as cpm/well. *, p < 0.05; **, p < 0.01 versus basal value. B, effect of increasing concentrations (20-60 µM) of C2:0 and C8:0 ceramide on CHO-K1 cell proliferation, measured using the [3H]thymidine incorporation assay. C, effect of the NOS, guanylyl cyclase, and PKG inhibitors NNA, Ly-83583 (Ly), and KT 5823 (KT) on the C8:0-induced CHO-K1 cell proliferation, measured using the [3H]thymidine incorporation assay. In B and C, cells (50,000/well) were serum-starved for 24 h and then received the appropriate treatments for 2 h. Then, the medium was removed and replaced with regular serum-free medium for another 16 h. In the last 4 h, cells were pulsed with 1 µCi/ml of [3H]thymidine. In the experiments depicted in B, the inhibitors were added 10 min before ceramide and then readministered after the removal of ceramide from the culture medium. Data are expressed as cpm/50,000 cells. **, p < 0.01 versus respective basal value; °°, p < 0.01 versus respective ceramide value.

View larger version (23K): [in this window] [in a new window]   Fig. 12.   Activation of ERK1/2 in CHO-K1, measured using a phospho-specific ERK1/2 antibody (top). Equal protein loading was demonstrated by probing the membrane with an antibody directed against the total ERK1/2 proteins (bottom). Lane 1, untreated control cells; lane 2, bFGF (30 ng/ml, 10 min); lane 3, bFGF+NNA (20 µM, 10 min); lane 4, SNP (1 µM, 10 min); lane 5, 8-bromo-cGMP (1 µM, 10 min).

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

In this article, we demonstrate that bFGF is a powerful eNOS activator in CHO-K1 cells via a completely novel intracellular pathway, never described previously for the FGF receptors. Indeed, bFGF caused the activation of eNOS through the regulation of sphingomyelinase activity and the generation of different ceramide species. The role of ceramide in the eNOS activity was also proposed in endothelial cells after bradykinin stimulation (Igarashi et al., 1999). However, in that study, the agonist stimulation was able to induce NO production through the regulation of both the classic Ca²⁺-dependent and the novel ceramide-related pathways. In CHO-K1 cells bFGF regulates eNOS activity in a completely Ca²⁺-independent manner. Moreover, this pathway is also independent from other known intracellular pathways previously related to the FGF receptors (PLC, phosphatidyl inositol-3 kinase/Akt, and MAP kinase). Conversely, using pharmacological and molecular approaches, we demonstrated that NO synthesis induced by bFGF is dependent on ceramide production. Indeed, two inhibitors of ceramide generation, [i.e., D609 (an inhibitor of PC-PLC) and desipramine (a more specific inhibitor of acidic sphingomyelinase)] completely reversed the effects of bFGF. Moreover, in fibroblasts derived from patients with Niemann-Pick disease, who are genetically deficient for sphingomyelinase, bFGF is completely ineffective; bypassing the genetic defect by administering exogenous ceramide allowed NO production to be restored. Activation of eNOS by bFGF seems to occur inside localized membrane compartments called caveolae. Caveolae are specialized membrane structures reported to represent privileged signal transduction structures (Anderson, 1998). All the components of the metabolic pathway described in this study are highly concentrated in caveolae: caveolae are extremely rich in sphingolipids, from which ceramide is produced (Anderson, 1998), and in acidic sphingomyelinase (Testi, 1996), which seems to participate in the bFGF signaling to generate ceramide. Moreover, in its inactive form, eNOS is selectively sequestered in caveolae through the binding to the main caveolae structural protein, caveolin 1 (Fleming and Busse, 1999). Here, we show that, in resting CHO-K1 cells, eNOS is highly bound to caveolin 1 and that this interaction is disrupted by bFGF treatment in a ceramide-dependent manner. How ceramide is able to favor the translocation and activation of eNOS remains to be determined. It was reported that ceramide is able to activate both serine kinases and phosphatases (Hannun, 1996). In our experimental model, the ceramide-activated protein phosphatase (Dobrowsky and Hannun, 1992) does not seem to be involved because the pretreatment with okadaic acid did not modify the bFGF-dependent NO production (S. Arena, T. Florio and G. Schettini, unpublished results). More recently, it was reported that, in endothelial cells, eNOS activity may be also induced by the ceramide metabolite S1P (Igarashi et al., 2001). S1P is generated through the metabolism of ceramide in sphingosine and its subsequent phosphorylation by specific sphingosine kinases (Spiegel and Merrill, 1996). In turn, newly generated S1P was reported to cause the activation of the EDG subfamily of G-protein-coupled receptors, causing an Akt-dependent phosphorylation of eNOS at Ser¹¹⁷⁹ to stimulate NO production (Igarashi et al., 2001). However, in our cells, despite the significant activation of ceramide synthesis, bFGF does not seem to induce the production of S1P and the consequent activation of Akt and phosphorylation of eNOS. Indeed, the blockade of S1P production, inhibiting the activity of the sphingosine kinase, the inhibition of the EDG receptors by PTX pretreatment, or the blockade of the phosphatidyl inositol-3 kinase/Akt pathway by treatment with wortmannin did not reverse the bFGF-dependent NO production. Thus, in our experimental model, eNOS activation by bFGF seems to be directly regulated by the ceramide formation, without the involvement of S1P synthesis and the activation of the EDG receptors. A similar regulation was also reported for the NO synthesis induced by the B2 bradykinin receptors. Indeed, in endothelial cells, bradykinin activates eNOS through ceramide production (Igarashi et al., 1999) without activation of the S1P intracellular pathway, because, unlike S1P, bradykinin does not cause the activation of Akt and the phosphorylation of eNOS (Igarashi et al., 2001). Using the CHO-K1 cell line, another pathway involved in the NO generation through the CCK-A receptor was also described previously (Cordelier et al., 1999). In particular, it was reported that CCK was able to activate nNOS through tyrosine dephosphorylation of the enzyme mediated by the tyrosine phosphatase SHP2 (Cordelier et al., 1999). In our experiments, we confirmed the presence of both eNOS and nNOS isoforms in CHO-K1 cells, as also observed by the previous report (Cordelier et al., 1999). However, we demonstrated that the activation of a different family of membrane receptors (i.e., tyrosine kinase versus G-protein-coupled) is able to regulate the NOS activity through a completely distinct mechanism. Interestingly, the pretreatment with vanadate, which completely reverses the CCK-induced NO production (Cordelier et al., 1999), does not interfere with the eNOS activation by bFGF (S. Arena, T. Florio, and G. Schettini, unpublished observation). Thus, in the same cell type, the activation of two different classes of receptors causes the activation of different NOS isoforms through completely independent pathways. However, the final physiological effect of the NO, and the subsequent cGMP production induced by both bFGF and CCK, is the proliferation of CHO-K1 cells (current study; Cordelier et al., 1997). Indeed, although often regarded as degenerative agents, both NO and ceramide may be, depending on the intracellular levels and the time of exposure, important proliferative agents (Olivera et al., 1992; Cordelier et al., 1997). Different effectors, however, are involved in these effects; the NO/cGMP synthesis induced by CCK was reported to directly regulate the ERK1/2 pathway to induce cell proliferation (Cordelier et al., 1997) whereas, in our experiments, NO and cGMP do not modify ERK1/2 activation induced by bFGF. The relationship between NO and ERK activation is still very controversial also in other cell systems. For example, in vascular smooth muscle cells, NO and cGMP inhibit ERK1/2 phosphorylation/activation (Mitani et al., 2000). However, it is still unclear how, in CHO-K1 cells, NO generated in different ways (through eNOS by bFGF or through nNOS by CCK) may result in a diverse regulation of ERK activity, although in both cases it causes a proliferative effect.

In our experiments, NO/cGMP generation contributes to only a small percentage of the proliferative effects of bFGF, which, conversely, are at least 70% dependent on the ERK1/2 activation. This small effect allows us to hypothesize that in more specialized cells, growth factor-dependent NO production may be involved in more differentiated functions.

A recent report showed that in rat astrocytes, proliferation was associated with a decrease in ceramide production (Riboni et al., 2001). This observation, taken together with the growing bulk of articles showing either antiproliferative and apoptotic or proliferative effects of ceramide and NO in different cell systems, indicates that a careful evaluation of cell-specific responses to ceramide generation is mandatory.

In conclusion, we describe a completely novel intracellular pathway by which growth factor receptors, in particular FGF receptors, induce NO and cGMP synthesis. We report that, in CHO-K1 cells, FGF receptor activation is able to regulate the synthesis of ceramide that causes activation of the eNOS. Moreover, the NO/cGMP pathway activation causes an ERK1/2-independent proliferative effect.