Regulation of apoptosis during neuronal differentiation by ceramide and b-series complex gangliosides.

Lipid analysis of gestational day E14.5 mouse brain revealed elevation of ceramide to a tissue concentration that induced apoptosis when added to the medium of neuroprogenitor cells grown in cell culture. Elevation of ceramide was coincident with the first appearance of b-series complex gangliosides (BCGs). Expression of BCGs by stable transfection of murine neuroblastoma (F-11) cells with sialyltransferase-II (ST2) resulted in a 70% reduction of ceramide-induced apoptosis. This was most likely due to an 80% reduced expression of prostate apoptosis response-4 (PAR-4). PAR-4 expression and apoptosis were restored by preincubation of ST2-transfected cells with N-butyl deoxinojirimycin (NB-DNJ) or PD98059, two inhibitors of ganglioside biosynthesis or p42/44 mitogen-activated protein (MAPK) kinase, respectively. In sections of day E14.5 mouse brain, the intermediate zone showed intensive staining for complex gangliosides, but only low staining for apoptosis (TUNEL) and PAR-4. Apoptosis and PAR-4 expression, however, were elevated in the ventricular zone which only weakly stained for complex gangliosides. Whole cell patch clamping revealed a 2-fold increased calcium influx in ST2-transfected cells, the blocking of which with nifedipine restored apoptosis to the level of untransfected cells. In serum-free culture, supplementation of the medium with IGF-1 was required to maintain MAPK phosphorylation and the anti-apoptotic effect of BCG expression. BCG-enhanced calcium influx and the presence of insulin-like growth factor-1 may thus activate a cell survival mechanism that selectively protects developing neurons against ceramide-induced apoptosis by up-regulation of MAPK and reduction of PAR-4 expression.

Ceramide and gangliosides are sphingolipids that are abundant in the plasma membrane of neuronal cells and are suggested to have a regulatory function in cellular differentiation and apoptosis (1,2). Ceramide is known to induce differentiation or apoptosis in a variety of different cell types whereas the physiological significance of gangliosides for these processes is still unclear (3,4). Recent studies with transgenic mice lacking neutral glucosphingolipids (glucosyltransferase knockout) or complex gangliosides (N-acetylgalactosaminyltransferase I knockout) indicate a high incidence of apoptosis in the nervous system during embryonal development or adulthood (5)(6)(7).
Most recently, the N-acetylgalactosaminyltransferase I/sialyltransferase II (GalNAcT 1 /ST2) double knockout mouse has been reported to show spontaneous death and audiogenic seizures (8). These results prompted us to investigate the role of gangliosides as potential anti-apoptotic effectors during early mouse brain development. Cell survival is crucial in this developmental period because about half of the proliferative neurons and glial cells die by apoptosis between gestational days E12 and E18 (9). This period is coincident with a switch from simple to complex gangliosides in brain tissue at days E14-E16 that has also been found during neuronal differentiation of teratocarcinoma-derived embryonic stem cells (2,10). The simple gangliosides GM3 and GD3 are converted to the b-series complex gangliosides (BCGs) of the type GD1b, GT1b, and GQ1b by up-regulation of ST2 and GalNAcT.
We have established an in vitro model for conversion of simple gangliosides to BCGs in neuronal cell culture by stable expression of ST2 in F-11A cells (11). F-11A cells are derived from murine neuroblastoma x rat dorsal root ganglion F-11 cells that were selected for a predominant expression of GM3 (Ͼ80% of total gangliosides). The ST2 cDNA was endowed with a Cterminal FLAG epitope and fused with green fluorescent protein (ST2-FLAG-GFP) to monitor the enzyme expression and subcellular localization by immunostaining and fluorescence microscopy. Recently, we have reported that F-11 cells undergo apoptosis upon incubation with novel ceramide analogs that have been synthesized by N-acylation of serinol (12). These ceramide analogs are highly soluble in water (up to 0.5 mM) and are not degraded by enzymes active in ceramide metabolism. We have used N-oleoyl serinol (S18) in addition to N-acetylsphingosine (C2-ceramide) to induce apoptosis in untransfected, GFPtransfected, or ST2-FLAG-GFP transfected F-11A cells. These cells were analyzed for the degree of apoptosis and signaling pathways known to participate in ceramide-induced cell death, in particular the PKC/PAR-4 and PI3K/MEK1 pathways (13)(14)(15)(16). The significance of BCG expression for a specific cell signaling event was verified by preincubation with N-butyl deoxinojirimycin (NB-DNJ), an inhibitor of glycolipid biosynthesis that does not induce apoptosis by itself (17). The results from the analysis of apoptosis and protein expression in neuroblastoma cells were compared with those obtained from preimplantation blastocystderived murine embryonic stem (ES-J1) cells which were used as in vitro model for neuronal differentiation. The biological significance of ceramide and BCGs for neuronal apoptosis was investigated by analysis of the lipid and regulatory protein expression in mouse brain at different stages of embryonal development.

Methods
Construction of the ST2-FLAG-GFP Vector and Stable Transfection of F-11A Cells-A cDNA encoding ST2 was amplified from a 15.5-dayold mouse embryo cDNA library as previously described (11). The ST2 cDNA was endowed with a NheI restriction site on its N terminus and with a FLAG-epitope sequence/SalI restriction site on its C terminus using a primer combination of sense 5Ј-taggtaccgatatcacaccgaggctgcgatgag-3Ј and antisense 5Ј-tagtcgacttgtcatcgtcgtccttgtaatc-3Ј for polymerase chain reaction amplification. The polymerase chain reaction product was ligated in-frame into the NheI/SalI restriction sites of the GFP vector. The sequence of the ST2-FLAG-GFP construct was verified by DNA sequencing. Murine neuroblastoma x rat dorsal root ganglion (F-11) cells were separated into single cells and the clones were analyzed for their ganglioside expression. A clone termed F-11A expressed mainly GM3 and was used for stable transfection of the ST2-FLAG-GFP vector following the LipofectAMINE procedure as described by the manufacturer (Life Technologies, Inc.). Clones transfected with the vector were isolated using fluorescence-activated cell sorting. Transfected cells were maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum and 500 g/ml geneticin to prevent loss of the vector. The analysis of enzyme activity followed a protocol that has been previously described (11).
Synthesis of S18 and Lipid Analysis-The novel ceramide analog N-oleoyl serinol (S18) was synthesized from serinol (2-amino-1,3-propanediol) and oleoyl chloride following a previously described procedure (12). Lipid preparation and analysis followed a standard protocol (17). Analysis of the acidic lipids was performed by HPTLC using the solvent system CHCl 3 , CH 3 OH, 0.2% CaCl 2 (50:45:10; v/v) whereas the analysis of ceramide was performed using the solvent system CH 3 OH/HOAc (9:1, v/v). Lipids were stained with 3% cupric acetate in 8% phosphoric acid and compared with standard lipids as described elsewhere (17,18). Gangliosides were specifically stained for by use of the resorcinol-HCl reagent.

Cultivation of Embryonic Stem Cells and Analysis of Embryonal Mouse Brain-Preimplantation blastocyst embryonic stem cells (ES-J1)
were grown on ␥-irradiated feeder fibroblasts and neuronal differentiation was induced by serum deprivation of embryoid bodies (EBs) as previously described (19). Protein was prepared from undifferentiated fibroblast-free ES-J1 cells, EBs, and differentiated neuroprogenitor cells at day 4 post-attachment of EBs on poly-L-ornithin/laminin-coated tissue culture dishes. Forebrains were prepared from mouse embryos extracted at days E12.5 to E18.5 by axial incision above the eye anlage followed by careful removal of the outer skin layers from the embryo head. The forebrains (5 to 20, depending on age) were taken up in 1 ml of deionized water and subjected to protein analysis and lipid extraction. For immunofluroescence microscopy, whole embryos were fixed by immersion in 10% formalin followed by gelatin embedding and saggital Vibratome sectioning at 30 m thickness. The embryo slices were first permeabilized for 5 min with 0.5% Triton X-100 and subjected to the TUNEL assay as described by the manufacturer (Oncogene). For identification of single cells, the slices were stained for 30 min with 1 g/ml Hoechst 33258 before immunostaining. The identification of complex gangliosides was performed by first incubating the embryo slices with 100 milliunits/ml Vibrio cholerae sialidase for 2 h at 37°C which resulted in the conversion of a-and b-series complex gangliosides into GM1. The desialylation reaction was followed by staining of GM1 with 1 g/ml Alexa 596-labeled cholera toxin subunit B (cholera toxin B) for another 2 h at 37°C.
Determination of Apoptosis-The degree of cell death was determined by fluorescent staining of living and dead cells by the use of the Live/Dead Viability/Cytoxicity assay with cells cultivated on 96-well tissue culture dishes following the protocol given by the manufacturer (Molecular Probes). A quantitative analysis of apoptosis was performed using a TUNEL-based fluorescein-FragEL assay (Oncogene) by counting of TUNEL-positive cells in 5 areas with an average cell number of 200 cells that were grown on coverslips. Apoptosis was verified by visualization of fragmented DNA that was separated by agarose gel electrophoresis following the procedure of the Suicide Tracker DNAladdering assay (Oncogene).
Whole-cell Patch Clamping-The recording of whole-cell calcium currents was performed as described elsewhere (20). Calcium influx was monitored using a bath solution composed of 20 mM BaCl 2 , 20 mM CsCl 2 , 110 mM tetraethylammonium chloride, 1.8 mM MgCl 2 , 15 mM glucose, and 0.1 M tetrodotoxin in 10 mM HEPES-buffer adjusted to pH 7.4 with CsOH. The pipette solution contained 120 mM N-methyl-Dglucamine, 20 mM tetraethylammonium chloride, 11 mM EGTA, 1 mM CaCl 2 , 4 mM MgATP, 0.1 mM Na 2 GTP, and 14 mM phosphocreatine in 10 mM HEPES buffer adjusted to pH 7.2 with methanesulfonic acid. A holding potential of Ϫ80 mV was employed with depolarization currents evoked at Ϫ90 to 50 mV and digitized at 180 s per point using low-pass filtering at 5 kHz (Ϫ3 dB). Analysis of the data was performed using a Power Macintosh 8600/200 computer supplied with PCI-16 Host Interface card. It was connected to an ITC-16 Data Acquisition Interface (Instrutech Corp, Port Washington, NY) utilizing Pulse Control 5.0 XOPs (obtained from Richard Bookman, Jack D. Herrington, and Kenneth R. Newton, University of Miami, FL) with Igor software (Waves-Metrics, Lake Oswego, OR).
Miscellaneous-The amount of protein was analyzed using a modification of the Folin phenol reagent (Lowry) assay as described (21). Protein precipitation was performed according to the Wessel and Flü gge method (22). SDS-PAGE was carried out using the Laemmli method followed by immunoblotting as described (23). Primary antibodies were used at a concentration of 1 g/ml for immunoblotting and 5 g/ml for immunofluorescence staining.

Ceramide and BCGs Are Up-regulated during Embryonal
Mouse Brain Development-Forebrains were removed from mouse embryos and analyzed for the expression of gangliosides and ceramide. Fig. 1A shows that the predominant gangliosides at day E12.5 were GM3 and GD3 (lane 2). BCGs were detectable at day E14.5 (lane 3), which was followed by the decline of GM3 and GD3 at day E16.5 (lane 4). The expression of GM1 and GD1a was initiated between days E16.5 (lane 4) and E18.5 (lane 5) and then intensified during post-natal development (P7, lane 6; adult, lane 7). The expression of ceramide was already found at day E12.5 (Fig. 1B, lane 1) and increased by about 40% at day E14.5 as determined by densitometry (lane 2). This was followed by a decline by more than 50% from day E14.5 to E18.5 (Fig. 1B, lanes 2-4). The ceramide composition resolved into three bands, the middle band of which was intensified during post-natal development (P7, lane 5; P21, lane 6; adult, lane 7). The structural composition of the different ceramide species has not been characterized yet, but the separation of these bands may be a result of the differences in the fatty acid chain length (17). A quantitative HPTLC with stearoylsphingosine as standard (Fig. 1B, lanes 8 -11) revealed that the ceramide level at peak time day E14.5 was about 0.6 Ϯ 0.1 g of ceramide/mg of cell protein which is equivalent to a tissue concentration of ϳ25 Ϯ 5 M (based on brain wet weight).
ST2-FLAG-GFP Transfected F11A Cells Express Predominantly BCGs-F-11A-cells were transfected with a cDNA construct composed of a 1341-base pair sequence encoding ST2 and a C-terminal linked FLAG-epitope followed by the sequence of epidermal growth factor protein. Several independent clones were isolated by fluorescence-activated cell sorting and were found to have a 4-fold higher specific ST2 activity than that of untransfected or GFP transfected control cells. As shown in Fig. 2A, expression of ST2-FLAG-GFP resulted in an altered cell morphology revealing bipolar process formation and completely obliterated clustering. Fig. 2B shows that ST2-FLAG-GFP was localized in the Golgi apparatus which was identical to the subcellular distribution found with wild-type ST2 (11). SDS-gel electrophoresis and immunoblotting of a protein preparation from transfected cells showed the expression of a 75-kDa protein corresponding to the molecular mass of ST2-FLAG-GFP (Fig. 2B).
Sphingolipids were isolated from transfected cells and analyzed by high-performance thin-layer chromatography (HPTLC). As shown in Fig. 2C, the neutral sphingolipid fraction of ST2-FLAG-GFP-transfected cells (lanes 7 and 10) revealed no difference in the expression of glycosphingolipids, sphingomyelin, or ceramide as compared with F-11A cells that were only transfected with GFP (lanes 6 and 9). Upon stable expression of ST2-FLAG-GFP the major gangliosides changed from GM3 found in GFP-transfected cells (lane 3) to GT1b and GQ1b (lane 4) which amounted to 50% of the total gangliosides. GD3, GD2, and GD1b were expressed in smaller amounts (40% of total gangliosides) whereas GM3 was only detectable in trace amounts. This result indicated that the expression of ST2-FLAG-GFP shifted the ganglioside composition from a-to bseries and enhanced the amount of complex gangliosides (bseries complex gangliosides or BCGs). As shown in Figs. 1A (lane 4) and 2C (lane 4), the composition of gangliosides in ST2-FLAG-GFP-transfected F-11A cells was almost identical to that in mouse brain at day E16.5 of embryonal development.
BCGs Inhibit Ceramide-induced Apoptosis-The effect of ganglioside expression on ceramide-induced apoptosis was investigated by overnight incubation of untransfected, GFP, or ST2-FLAG-GFP-transfected cells with 30 M N-acetylsphingosine (C2-ceramide) or 80 M of the novel ceramide analog S18. GFP-transfected cells did not reveal any difference in the degree of apoptosis as compared with untransfected cells. As shown in Table I, however, ST2-FLAG-GFP expressing cells revealed up to a 70% reduction in apoptosis depending on the concentration and type of the pro-apoptotic agent used for the incubation reaction. This result was verified by a DNA fragmentation assay that showed no appearance of 200-base pair fragments (DNA laddering) upon incubation of ST2-FLAG-GFP-transfected cells with S18 (Fig. 3A, lane 4) whereas GFPtransfected cells revealed intensive DNA laddering (Fig. 3A,  lane 2). It should be noted, however, that the degree of apoptosis was critically depending on cell density as previously reported (11). All of the experiments were performed at exactly 50% confluence, a cell density that previously revealed the highest degree of apoptosis induced with ceramide analogs in untransfected F-11A cells.
The involvement of gangliosides in the reduction of apoptosis was analyzed by preincubation of GFP or ST2-FLAG-GFPtransfected cells with 250 M of the glycolipid biosynthesis inhibitor NB-DNJ prior to the addition of S18 or C2-ceramide.
As shown in Fig. 2C (lane 5), ST2-FLAG-GFP-transfected cells were completely deprived of gangliosides. NB-DNJ did not affect the degree of ceramide-induced apoptosis in GFP-transfected control cells but restored that of ST2-FLAG-GFP-transfected cells to control level (Table I). This was verified by the appearance of DNA laddering that was induced by S18 after preincubation with NB-DNJ (Fig. 3A, lane 3).
PAR-4 Expression Is Reduced and bcl-2 Is Up-regulated by BCGs-The expression of the pro-apoptotic protein PAR-4 was analyzed by means of indirect immunofluorescence microscopy in combination with a TUNEL assay in untransfected F-11A cells upon incubation with S18. As shown in Fig. 3A (right panel), more than 80% of the TUNEL-positive cells were strongly stained for PAR-4 indicating that its expression is necessary for the induction of apoptosis. The number of strongly stained cells, however, exceeded that of TUNEL-positive cells. ST2-FLAG-GFP-transfected cells expressed about 80% less PAR-4 and showed no TUNEL staining (not shown).
The overall expression of PAR-4 was further analyzed by SDS-PAGE and immunoblotting of protein extracts from detergent-solubilized cells. Fig. 3B shows that the amount of immu-  I C2-ceramide or S18-induced cell death in GFP or ST2-FLAG-GFP transfected F-11A cells Cells were cultivated in 96-well dishes and incubated with 30 M N-acetylsphingosine (C2-cer) or 80 M N-oleoylserinol (S18) in serumsupplemented or serum-free medium for times indicated. Supplements included the addition of different BCGs (40 M), N2 or ITS-supplement (1:100), or single growth factors at a concentration of 100 ng/ml (NGF, IGF-1, LongR3-IGF-1 peptide analog, insulin) or 5 g/ml (insulin). The effect of different inhibitors was analyzed by preincubation for 5 h with 50 M PD98059 (MEK1-inhibitor) or for 72 h with 250 M NB-DNJ (glycolipid biosynthesis inhibitor). The number of living cells was quantified by the determination of calcein-esterase activity, and the number of dead cells was determined by diethidium bromide staining (Dead/Alive assay, Molecular Probes). All measurements were done with a sample size of n ϭ 6. The mean standard variation was 15% of the average value as given in the BCGs Enhance Growth Factor-dependent Activation of p42/44 MAPK and Down-regulate BAD-The effects of growth factors on cell proliferation and apoptosis were analyzed by incubation of GFP or ST2-FLAG-GFP-transfected F-11A cells with N2, ITS (insulin-transferrin-selenite), NGF, bFGF/FGF-2, insulin, or IGF-1 supplemented serum-free medium. Removal of serum resulted in an increased cell death in GFP or ST2-FLAG-GFP-transfected cells, even without the addition of C2ceramide or S18 to the medium (Table I). Cell death resulted from apoptosis as revealed by punctate Hoechst staining of the nuclei. Apoptosis, however, could be prevented by the addition of 0.5% FBS, N2, or ITS supplement to the cell culture medium. Among single growth factors, it was found that NGF (100 ng/ml), insulin (5 g/ml), IGF-1 (100 ng/ml), or its peptide analog LongR3-IGF-1 (100 ng/ml) could reduce cell death by at least 70% (Table I). Insulin, however, failed to significantly rescue the S18-incubated cells at a lower concentration (100 ng/ml) in the cell culture medium. The effect of bFGF/FGF-2 on the reduction of cell death (by about 30%) was lower than that of NGF, IGF-1, or insulin (at 5 g/ml).
The phosphorylation of the NGF receptor, tyrosine receptor kinase A (p490-trkA), and the expression of the IGF-1 receptor (IGF-1R) were analyzed by immunostaining on Western blots of protein prepared from GFP or ST2-FLAG-GFP-transfected cells. Fig. 3C shows that the trkA phosphorylation at serine 490 was specifically elevated in serum-containing or NGF-supplemented medium (lanes 2, 3, 6, and 7) as compared with serumfree medium (lanes 1 and 5) or medium supplemented with IGF-1 (lanes 4 and 8). In GFP-transfected cells, immunostaining of p490-trkA in the presence of serum or NGF (lanes 2 and 3) was more intense than in ST2-FLAG-GFP-transfected cells (lanes 6 and 7). The expression of IGF-1R appeared not to be significantly affected by transfection or the presence of single growth factors. Immunostaining of phospho-p42/44 MAPK (pMAPK) in GFP-transfected cells grown in serum-free medium showed none or only a low level of phosphorylation, irrespective of the presence of NGF or IGF-1 (lanes 1, 3, and 4). In ST2-FLAG-GFP-transfected cells, however, there was a clear signal of immunostained pMAPK in serum-free medium and in the presence of NGF (lanes 5 and 7). Immunostaining was strongest in the presence of serum or IGF-1 (lanes 6 and 8). This was not due to elevated protein expression, but to enhanced phosphorylation of MAPK as shown by immunostaining with an antibody against total MAPK. Preincubation of ST2-FLAG-GFP-transfected cells for 72 h with 250 M NB-DNJ resulted in a reduction by 80% of the IGF-1 induced phosphorylation of MAPK (lane 9).
The pro-apoptotic protein BAD was down-regulated by MAPKdependent phosphorylation at serine 112 in GFP or ST2-FLAG-GFP-transfected cells incubated with serum, NGF, or IGF-1 (lanes 2-4 and lanes 6 -8). Removal of serum resulted in reduction of BAD phosphorylation consistent with enhanced apoptosis (lanes 1 and 5). Immunostaining with a BAD-specific antibody revealed that the difference in phosphorylation was not due to altered protein expression. Only IGF-1 or IGF-1 plus NB-DNJ incubation showed reduced or enhanced BAD expression levels, respectively. Incubation with NB-DNJ decreased BAD phosphorylation in IGF-1-treated cells which is in agreement with a reduced activity of MAPK (lane 9). The MAPK-dependent down-regulation of BAD was in accordance with the effect of the PI3K-inhibitor wortmannin on ceramide-induced apoptosis. Simultaneous incubation of ST2-FLAG-GFP-transfected cells with 200 nM wortmannin and 80 M S18 restored apoptosis to the level of untransfected cells.
Calcium Influx Is Enhanced by BCG Expression and Modulates Apoptosis-Voltage-dependent calcium influx was monitored by whole cell patch clamping of GFP or ST2-FLAG-GFPtransfected cells. Fig. 4A shows a single cell recording with a holding potential of Ϫ80 mV. The calcium current was about 2-fold higher for ST2-FLAG-GFP-transfected cells with peak amplitude at 0 Ϯ 10 mV. As shown in Fig. 4B, preincubation of ST2-FLAG-GFP-transfected cells with 250 M NB-DNJ for 72 h reduced the calcium current to less than the level of GFPtransfected cells. The significance of calcium influx for apoptosis was analyzed by incubation of GFP or ST2-FLAG-GFPtransfected cells with nifedipine, an L-type specific calcium channel blocker. The addition of 10 M nifedipine increased S18-induced apoptosis in ST2-FLAG-GFP-transfected cells by 2-fold within 48 h of incubation. The rate or degree of S18induced apoptosis in GFP-transfected cells, however, remained unaffected by nifedipine.

PAR-4 and Apoptosis Are Down-regulated during Neuronal Differentiation of ES Cells-Preimplantation mouse embryonic stem cells (ES-J1
) were committed to neuronal differentiation by a serum-deprivation protocol. At three differentiation stages, apoptosis was induced by overnight incubation with 100 M S18 or 30 M C2-ceramide. Fig. 5A shows Hoechst staining of undifferentiated ES cells on feeder fibroblasts (day 2 after seeding) and TUNEL staining of embryoid bodies (EBs (day 8)). About 70% of the undifferentiated ES cells showed a punctate nuclear Hoechst staining that is typical for apoptotic cells. Apoptosis was less prominent in EBs and was confined to cell clusters on the outer layer. Early neuroprogenitor (NP) cells on day 1 post-plating of the EBs revealed about 50% apoptosis upon overnight incubation with 100 M S18 or 30 M C2-ceramide (not shown). Fig. 5B shows Hoechst, TUNEL, and PAR-4 staining in about 3,000 differentiating NP cells on day 4 post-plating of the EBs on coated tissue culture dishes. About 250 cells were TUNEL positive corresponding to 12% apoptosis induced by S18. 80% of the TUNEL positive cells stained intensively for PAR-4, consistent with the results found for F-11A cells. As with F11A cells, the number of NP cells revealing a strong PAR-4 signal (about 25% of all cells) exceeded that of apoptotic cells.
To investigate the correlation of apoptosis with the level of PAR-4 expression, protein was prepared from ES cells at three stages of differentiation and analyzed by means of SDS-PAGE and immunoblotting. As shown in Fig. 5C PAR-4 and BCG Expression Is Correlated with Apoptosis and Cell Survival in Embryonal Mouse Brain-As shown in Fig. 6A, immunoblot analysis of protein prepared from embryonal mouse brain revealed a grouping into three different modes for expression regulation (lanes 1-6): (i) gradual up-regulation of the differentiation markers GAP-43, MAP 2a/b, and synaptophysin; (ii) gradual down-regulation of growth factor receptors (IGF-1R␤ and p490-trkA; note, however, the up-regulation of the entire population of trkA/B/C receptors) and PAR-4; and (iii) peaking of bcl-2, caspase 3 (cleaved), and pMAPK on day E16.5 (lane 3). The expression of PKC appeared not to change significantly during mouse brain development. Fig. 6B shows a correlation of the protein expression to that of lipids, which was discussed in the preceding section. Fig. 7 shows a saggital vibratome section of embryonal mouse brain at day E14.5 that was stained for complex gangliosides (cholera toxin B staining upon incubation with V. cholerae sialidase). Complex gangliosides were predominantly expressed in the intermediate zone (iz) and the cortical plate (cp). A second, but smaller portion was stained in the dorsal cell margin surrounding the ventricle (v). There was almost no expression of complex gangliosides in the ventricular zone (vz), an embryonal cell layer that stained intensively for PAR-4 and showed the predominant TUNEL signal. DISCUSSION The coincidence of intensive apoptosis with a major change of the ganglioside composition in mouse brain at gestational day E14.5 prompted us to analyze the biological significance of sphingolipids for neuronal apoptosis during embryonal development. The expression of BCGs is concomitant with an elevation of endogenous ceramide, indicating that ceramide may induce apoptosis in a population of developing neuroprogenitor cells whereas another population may be protected by the expression of BCGs. This assumption is supported by the observation that in neuroprogenitor cells grown in cell culture, exogenously added ceramide induced apoptosis at a concentration equivalent to that in mouse brain tissue at the peak time of apoptosis. Furthermore, stable transfection of neuroblastoma F-11A cells with ST2-FLAG-GFP results in the predominant expression of BCGs concomitant with a reduction of apoptosis that is induced with ceramide or its analog S18. Reduction of apoptosis is correlated with a lower expression level of the pro-apoptotic PKC inhibitor protein PAR-4 and a higher level of the anti-apoptotic protein bcl-2. Apoptotic (TUNEL positive) F-11A control cells that were only transfected with GFP stain strongly for PAR-4, indicating that its expression is necessary for ceramide-induced apoptosis. The number of strongly stained cells, however, exceeds that of TUNEL positive cells suggesting that PAR-4 expression is necessary, but not sufficient for induction of apoptosis. This conclusion is corroborated by the observation that the same correlation between PAR-4 expression and TUNEL staining has also been found in ES cell-derived neuroprogenitor cells and embryonal mouse brain tissue.
Recently, several studies have demonstrated that ceramide promotes an inhibitory PKC⅐PAR-4 complex formation and that inhibition of PKC results in suppression of bcl-2 expres-sion (13)(14)(15). Bcl-2 is known to counteract ceramide-induced apoptosis by preventing mitochondrial cytochrome c-release and subsequent activation of caspases 3 and 9 (24,25). The reduced degree of apoptosis in ST2-FLAG-GFP-transfected cells may thus be attributable to a reduced expression of PAR-4, resulting in an up-regulation of anti-apoptotic bcl-2. Low PAR-4 and/or high bcl-2 expression may also protect the population of F-11A or neuroprogenitor cells that do not undergo apoptosis upon incubation with C2-ceramide or S18.
The level of PAR-4 expression has been reported to be downregulated by activation of p42/44 MAPK-kinase (MEK1) (16). This is consistent with the observation that in ST2-FLAG-GFPtransfected cells, preincubation with the MEK1 inhibitor PD98059 restores the level of PAR-4 expression and the degree of ceramide-mediated apoptosis to that of control cells. Restoration of PAR-4 expression and S18-induced apoptosis has also been observed upon preincubation of ST2-FLAG-GFP-transfected cells with NB-DNJ, an inhibitor of ganglioside biosynthesis. Thus, suppression of PAR-4 expression is most likely due to a BCG-mediated activation of MEK1.
The cultivation of ST2-FLAG-GFP-transfected cells in serum-free medium obliterates the anti-apoptotic effect of BCGs, suggesting that their activity is depending on the presence of growth factors. The complementation of serum-free medium with N2 or ITS supplement, insulin, or IGF-1 reduced ceramide or S18-induced apoptosis in ST2-FLAG-GFP-transfected cells, indicating a functional role of these growth factors for MEK1 activation. This effect is concomitant with enhanced phosphorylation of MAPK and is consistent with studies reporting MEK activation by IGF-1 (26). The significance of insulin for the reduction of apoptosis is unclear since it rescued S18incubated cells from apoptosis only at higher concentrations (Ͼ1 g/ml) in the cell culture medium. At these concentrations, insulin has been reported to stimulate both insulin and IGF-1 receptors (27). Hence, we conclude that the reduction of apoptosis rather relies on the effect of IGF-1 receptors than on insulin receptors. The significance of BCGs for this anti-apoptotic activity has been shown by the reduction of IGF-1 in- duced MAPK phosphorylation upon preincubation with NB-DNJ.
The NGF-receptor kinase trkA has been analyzed as another potential mediator of the anti-apoptotic activity of BCGs. Complex gangliosides have been reported to activate trkA, which results in a subsequent stimulation of MEK (28). ST2-FLAG-GFP-transfected cells, however, do not reveal a higher level of phosphorylated trkA (p490-trkA) and also show less immunostaining of pMAPK upon incubation with NGF than with IGF-1. Thus, the anti-apoptotic effect of BCGs is most likely due to activation of the IGF-1 to MEK1 signaling pathway.
IGF-1 is known to be endogenously expressed in embryonal brain tissue and plays a critical role for the regulation of axon formation (29,30). It is not known, however, how BCGs may interact with IGF-1 or its receptor. Most recently, it has been shown that cerebellar neurons from knockout mice lacking BCGs undergo apoptosis when exposed to elevated KCl concentrations (6). Apoptosis was found to result from impaired intracellular calcium regulation upon depolarization by KCl. Several other studies have shown that the IGF-1 receptor interacts with calcium L-channels (31,32). We have demonstrated that the calcium current in ST2-FLAG-GFP-transfected cells is 2-fold higher than that of GFP-transfected cells and can be reduced by preincubation with NB-DNJ. We thus suggest that in ST2-FLAG-GFP-transfected cells, MEK1 is synergistically activated by IGF-1 and BCG-enhanced calcium influx. F-11 cells are known to express dihydropyridine-sensitive L-type calcium channels (33). Blocking of L-type calcium channels with nifedipine increases S18-induced apoptosis only in ST2-FLAG-GFP-transfected cells. Accordingly, protection against ceramide-or S18-induced apoptosis by BCGs may critically rely on a regulation of calcium-influx by L-type channels. It should be noted that this mechanism is different from the one suggested for calcium dysregulation in cerebellar neurons of the GalNAcT/ST2 double knockout mouse, which has been attributed to enhanced efflux of calcium from the nucleus (6).
A synergistic activation of MEK by IGF-1 and calcium may proceed via up-regulation of protein kinase B/Akt-kinase by PI3K (phophatidylinositol 3-kinase). This has been demonstrated in various cell systems and is supported by the observation that the PI3K inhibitor wortmannin increases S18-induced apoptosis in ST2-FLAG-GFP-transfected cells (31,32). Fig. 8 shows how the protection against ceramide-induced apoptosis in these cells may emerge from four anti-apoptotic events: (i) activation of PI3K by BCG-enhanced calcium-influx and IGF-1 stimulation; (ii) suppression of PAR-4 expression by downstream signaling of p42/44 MAPK; (iii) up-regulation of bcl-2 expression induced by downregulation of PAR-4; and (iv) inactivation of the bcl-2 antagonist BAD by MAPK-dependent phosphorylation. Ceramide has been reported to affect two sites of this potential cell survival pathway: (i) inhibition of the PI3K/Akt-kinase step via ceramide-activated protein phosphatase or a yet unidentified ceramide target protein (34,35); and (ii) ceramide-promoted formation of an inhibitory PKC⅐PAR-4 complex (14,15). As shown in Fig. 8, this second target site of ceramide is of particular significance since it has been suggested that IGF-1 may stimulate a MAPK to PKC cross-talk via activation of PI3K and that PKC directly activates MEK (36 -38). In addition, it has been demonstrated that PKC can be activated or inhibited by ceramide depending on its concentration (12,37,39). A concentration dependent activity of ceramide would explain apparently contradictory studies reporting activation, inactivation, or no effect of ceramide on the MAPK pathway (40,41). As shown in Fig. 8, activation of PKC stimulates MAPK, whereas PAR-4 mediated inactivation downregulates this cell survival pathways the regulation of which may rely on the intracellular concentration of PAR-4 and ceramide.
The biological significance of BCGs and PAR-4 expression for ceramide-induced apoptosis during neuronal differentiation has been analyzed in day E14.5 embryonal mouse brain sections. TUNEL staining reveals that apoptosis is confined to a major area of the ventricular zone. This observation is consistent with previous reports on apoptosis in embryonal mouse brain (9). The majority of the ventricular zone shows expression of PAR-4, but stains only weakly for complex gangliosides. The lower level of TUNEL staining and PAR-4 expression but higher level of complex ganglioside staining in the intermediate zone indicates that these cells are protected against apoptosis. We assume that the complex gangliosides are mainly composed of BCGs since staining with cholera toxin B only occurs after sialidase digestion and HPTLC analysis shows only trace amounts of a-series complex gangliosides at day E14.5. BCG expression in the intermediate zone may thus result in the suppression of PAR-4 expression and protection against apoptosis. This model is corroborated by the results obtained from the studies with ST2-FLAG-GFP-transfected F-11A cells.
The analysis of protein expression reveals a simultaneous elevation of pMAPK, bcl-2, and caspase 3 at day E16.5. A simultaneous expression of pro-and anti-apoptotic proteins suggests that the embryonal brain tissue is composed of two cell populations: apoptotic cells expressing PAR-4, but not BCGs, and nonapoptotic cells expressing BCGs, but not PAR-4. This conclusion is in agreement with the hypothesis that BCGs activate the MEK/ERK1 signaling pathway resulting in downregulation of PAR-4 and up-regulation of bcl-2 (Fig. 8). En- FIG. 8. BCG/Ca 2؉ /IGF-1 dependent signaling pathway for apoptosis and cell survival. BCG-enhanced calcium influx and IGF-1 synergistically activate the PI3K to MEK1/MAPK pathway, resulting in down-regulation of pro-apoptotic PAR-4 and BAD (increased phosphorylation), and enhanced expression of anti-apoptotic bcl-2. This cell survival pathway can be inhibited by ceramide due to down-regulation of PI3K or MEK1. PAR-4 may mediate inactivation of PKC at proapoptotic concentrations of ceramide resulting in reduced MEK1/MAPK activity. hanced bcl-2 expression may then protect the BCG containing cell population against ceramide-induced apoptosis. Selective apoptosis of GM3 and survival of BCG containing neuronal cells would also explain the sudden switch from simple gangliosides to BCGs during embryonal mouse brain development. As shown in the Appendix, a mathematical modeling of selective apoptosis/cell survival is consistent with the ganglioside expression as determined by HPTLC analysis. It is reasonable to speculate that during migration to the intermediate zone, neuronal progenitor cells can acquire protection against apoptosis by expression of BCGs due to up-regulation of GalNAcT. Hence, the subventricular/ventricular zone may not only provide neuronal progenitor cells but also select for those cells that express predominantly BCGs.
In summary, we have provided evidence that the spatial and temporal distribution of ceramide and BCGs participates in a regulatory mechanism that is significant for selective apoptosis or cell survival during embryonal mouse brain development. Survival of BCG expressing neuronal cells is reasonable since several studies have suggested that complex gangliosides participate in synaptic functions, in particular for regulation of Ca 2ϩ -flux (42)(43)(44)(45). We will now investigate the molecular mechanisms for regulation of ceramide and BCG expression and further analyze the sphingolipid-regulated signaling pathways that are significant for neuronal development.