Changes in the lipid turnover, composition, and organization, as sphingolipid-enriched membrane domains, in rat cerebellar granule cells developing in vitro.

In the present paper, we report on the properties of sphingolipid-enriched domains of rat cerebellar granule cells in culture at different stages of neuronal development. The major lipid components of these domains were glycerophospholipids and cholesterol. Glycerophospholipids were 45-75% and cholesterol 15-45% of total lipids of the domains. This corresponded to 5-17% of total cell glycerophospholipids and 15-45% of total cell cholesterol. Phosphatidylcholine, mainly dipalmitoylphosphatidylcholine, was 66-85% of all the glycerophospholipids associated with these domains. Consequently, the palmitoyl residue was significantly enriched in the domains. The surface occupied by these structures increased during development. 40-70% of cell sphingolipids segregated in sphingolipid-enriched membrane domains, with the maximum ganglioside density in fully differentiated neurons. A high content of ceramide was found in the domains of aging neurons. Then, the sphingolipid/glycerophospholipid molar ratio was more than doubled during the initial stage of development, whereas the cholesterol/glycerophospholipid molar ratio gradually decreased during in vitro differentiation. Phosphorylated phosphoinositides, which were scant in the domains of undifferentiated cells, dramatically increased during differentiation and aging in culture. Proteins were minor components of the domains (0.1-2.8% of all domain components). Phosphotyrosine-containing proteins were selectively recovered in the sphingolipid-enriched domain. Among these, Src family protein-tyrosine kinases, known to participate to the process of neuronal differentiation, were associated with the sphingolipid-enriched domains in a way specific for the type of kinase and for the developmental stage of the cell. Proteins belonging to other signaling pathways, such as phosphoinositide 3-kinase and its downstream target, Akt, were not associated with the domains.

Neuronal development involves many different spatially and temporally regulated events (reviewed in Refs. [1][2][3][4], including cell migration (which implies changes in the adhesion properties of the neuronal cell and in its interactions with the changing local environment and other cells), and deep morphological and biochemical changes leading to the processes of neuritogenesis and synaptogenesis, and to the complex spatial organization which is typical of the nervous tissue. In all these developmental events, cell surface properties, determined by membrane components, are of primary importance. Lipids are responsible for the physicochemical properties of the membrane itself, largely determining its asymmetry, fluidity and plasticity features, and its organization in domains. Moreover, many membrane lipids play more specific functional roles, and are in many different ways directly involved in the machinery devoted to the control of information in the neuronal cell. Inositol phospholipids (5), sphingolipids (6,7), and plasmalogens (8) are precursors of bioactive molecules. Glycosphingolipids, asymmetrically located in the outer leaflet of the membrane bilayer, are of primary importance as receptors and cell surface antigens (9). A particular function of lipids at the cell surface is linked to their ability to interact with membrane proteins, such as receptors or ion channels (6,10,11), and to modulate the functional activity of the protein itself. Some membrane lipids undergo segregation, creating a lipid phase distinct from the bulk phase of the membrane. Recent findings indicate that a particular set of lipids and proteins are segregated together within the plasma membrane, forming functional units that are involved in signal transduction processes (12,13). Spontaneous segregation of membrane sphingolipids seems to be the driving force of these structures, hence termed sphingolipid-enriched membrane domains. A functional role for these domains is particularly well documented in neural cells (reviewed in Ref. 14), where the specific interaction of sphingolipids with protein kinases of the Src family is particularly relevant. The nonreceptor protein-tyrosine kinases belonging to this family seem to be involved in signaling pathways that regulate the function of postmitotic, terminally differentiated cells. They regulate multiple cell functions, including cytoskeletal organization, cell-substrate, and cell-cell adhesion, and cell-cell communication (15,16). In the nervous system, many lines of evidence indicate that they are involved in the process of neuronal differentiation, suggesting a tight link between Src family kinases and neurite extension and guidance. In the developing cerebellum, the expression of c-Src coincides with the onset of neuronal differentiation (17). The expression and activation of c-Src and Lyn are increased during differentiation (18 -22), and postmitotic neurons from the central nervous system express high levels of structurally distinct forms of c-Src and Fyn (15,23). c-Src and Fyn are required in the response to different neural cell adhesion molecules in cerebellar neurons (24,25), participating in the control of neurite extension. c-Src, Fyn, and Yes are enriched in growth cones of neurons and neuroblastoma cells as a complex with other proteins (26). c-Src and Fyn also have a role in the mature synapse, related to the induction of long-term potentiation (27,28). A close association between these proteins and sphingolipid-enriched domains has been demonstrated in many cell types, including neurons (29 -32). Moreover, several GPI 1 -anchored proteins, which are usually associated with the sphingolipid-enriched domains, have been implicated in signal transduction mediated by Src family tyrosine kinases. In the present paper, we analyze the changes in lipid composition, turnover, and membrane organization in cerebellar neurons at different stages of development in vitro. The data we obtained clearly indicate a strict correlation between the functional status of the neuronal cell and the level of lipid and protein organization within sphingolipid-enriched membrane domains.

EXPERIMENTAL PROCEDURES
Materials-Commercial chemicals were the purest available, common solvents were distilled before use, and water was doubly distilled in a glass apparatus. Trypsin, crystalline bovine serum albumin, and all the reagents for cell culture and the cholesterol assay kit were from Sigma Chemical Co., except for basal modified Eagle's medium and fetal calf serum, which were purchased from Flow Laboratories. Antiphosphotyrosine mouse monoclonal IgG 2b antibody, anti-c-Src goat polyclonal IgG (N- 16), anti-c-Src (SRC 2), anti-Lyn, anti-Fyn, anti-Akt1/2, anti-PI3K (p110) rabbit polyclonal IgG antibodies, horseradish peroxidase-conjugated secondary antibodies, and Protein A/G PLUSagarose were from Santa Cruz Biotechnology. Sphingosine was prepared from cerebroside (33). Standard sphingolipids and glycerophospholipids were extracted from rat brain, purified, and characterized (34). [1-3 H]Sphingosine was prepared by specific chemical oxidation of the primary hydroxyl group of sphingosine followed by reduction with sodium boro[ 3 H]hydride (35) (radiochemical purity over 98%; specific radioactivity 2 Ci/mmol). [  Cell Culture-Granule cells, obtained from the cerebella of 8-day-old Harlan Sprague-Dawley rats, were prepared and cultured as described previously (36 -38). The cells were plated in 100-mm dishes at a density of 9 ϫ 10 6 cells/dish and cultured with 10 ml of basal modified Eagle's medium containing 10% fetal calf serum for up to 17 days. After the 8th day in culture, the medium was supplemented with 1 mg/ml glucose. Cell viability was assessed by the Trypan blue exclusion method. Primary cultures from 8-day-old rat cerebellum are greatly enriched in neurons. Contamination by non-neuronal cells (mainly astrocytes), as determined by immunological and morphological criteria, is below 5% (37,38). The replication of non-neuronal cells was prevented by adding 10 M cytosine arabinoside to the culture medium. In these cultures, about 90% of the cells are immature excitatory granule cells. The cultures also contain about 5% ␥-aminobutyric acid-ergic, inhibitory interneurons. Rat cerebellar granule cells in culture immediately start to differentiate and spontaneously undergo a developmental pattern that resembles that of cerebellar neurons in vivo, but in a condensed time scale, reaching a mature state after 8 days. The experiments were performed at three different stages of neuronal development in culture. The 2nd day in culture corresponded to the initial stage of neuronal differentiation and neuritogenesis. At this stage, the cells already emit short neurite-like processes (38). The 8th day in culture corresponded to morphologically and biochemically fully differentiated neurons; the cells are mostly grouped in large aggregates, connected by a complex net of fasciculated fibers. The progressive formation of axo-axonic synapses was observed. From the biochemical point of view, granule cells at this stage of in vitro development are characterized by the expression of voltage-dependent sodium channels, neurotransmitter receptors, neuronal surface sialoglycoproteins. Moreover, they acquire the ability to synthesize and release glutamate under depolarizing conditions (36,39). The 17th day in culture corresponded to a late stage of neuronal development, characterized by the onset of age-induced apoptotic death of cerebellar granule cells (40). Thus, the primary cultures are a valuable model for the study of neuronal development in vitro. Typical protein content at these times was 240, 700, and 770 g of protein/dish.
Analysis of Endogenous Lipids-Cells (4 -7 ϫ 10 7 ) at the 2nd, 8th, and 17th day in culture were harvested and lyophilized. Lipids were extracted with chloroform/methanol 2:1 (v/v), and the total lipid extract was subjected to a two-phase partitioning as previously described (41), resulting in the separation of an aqueous phase containing gangliosides and in an organic phase containing all other lipids. The ganglioside content was determined in the aqueous phase as lipid-bound sialic acid by the method of Svennerholm (42). The phospholipid content was determined in the organic phases as phosphate after perchloric acid digestion by the method of Bartlett (43). Gangliosides from the aqueous phases and phospholipids, cholesterol, and ceramide from the organic phases were separated by HPTLC as described below. Identification of lipids after separation was assessed by comigration with standard lipids and confirmed by their susceptibility to enzymatic and chemical treatments as previously described (44). After chromatographic separation, compounds were chemically detected and their amounts were determined by densitometry as described below.
The mass lipid composition of the sphingolipid-enriched membrane fractions was calculated on the basis of the endogenous lipid content in the cell homogenate and the percent distribution of the radioactivity associated with each lipid species in the fraction, as described below.

Treatments of Cell Cultures with [ 3 H]Sphingosine or [ 35 S]Methionine or [ 32 P]
Orthophosphate-Cells at the day of preparation (0 day in culture), at the 6th and at the 15th day in culture, were incubated in the presence of 3 ϫ 10 Ϫ8 M [1-3 H]sphingosine (5 ml/dish) in cell-conditioned medium for a 2-h pulse followed by a 48-h chase. Under these conditions, free radioactive sphingosine was hardly detectable in the cells, and all sphingolipids and phosphatidylethanolamine (obtained by recycling of radioactive ethanolamine formed in the catabolism of [1-3 H]sphingosine) were metabolically radiolabeled (30,44). Cells at the 1st, 7th, and 16th day in culture were preincubated in methionine-free medium for 2 h and subsequently incubated in the presence of 25 Ci/ml L-[ 35 S]methionine (5 ml/dish) for 20 h, to achieve steady-state radiolabeling of proteins (44 -46). Cells at the 2nd, 8th, and 17th day in culture were incubated in the presence of 50 Ci/ml carrier-free [ 32 P]orthophosphate (5 ml/dish) in phosphate-free culture medium for 4 h (44,47). These experimental conditions allowed introduction of radioactivity into phosphoproteins and glycerophospholipids, these latter including the very minor ones, such as the phosphorylated phosphoinositides (47).
with chloroform/methanol 2:1 (v/v) (41). The total lipid extract was analyzed by HPTLC as described below, followed by radioactivity imaging for quantification of radioactivity. Identification of lipids separated by HPTLC was accomplished by comigration with standard lipids and confirmed by susceptibility of compounds to following enzymatic and chemical treatments (44). A sample of the lipid extract was treated at 37°C for 2 h, in 50 l of water, in the presence of 1 milliunit of Vibrio cholerae sialidase, to yield GM1. Sphingomyelin, phosphatidylethanolamine, and phosphatidylcholine were purified according to the HPTLC-blotting technique previously reported (48): they were separated by HPTLC, identified by spraying with primulin, blotted to PVDF membrane where the corresponding bands were cut and subjected to elution. Sphingomyelin was treated at 37°C overnight in 30 l of 100 mM Tris-HCl, pH 7.4, 0.5 mM MgCl 2 , 0.05% sodium deoxycholate, in the presence of 11 milliunits of Bacillus cereus sphingomyelinase, to yield ceramide; phosphatidylethanolamine was characterized following its degradation under alkaline conditions. The enzymatic or chemical reaction mixtures were separated by HPTLC, and the reaction products were identified by chromatographic comparison with standard lipids.
Ganglioside and glycerophospholipid species were quantified after separation on HPTLC followed by specific detection with a p-dimethylaminobenzaldehyde reagent (50), or a molybdate reagent (51), respectively. The relative amounts of each ganglioside or of each phospholipid were determined by densitometry, and their mass content was calculated on the basis of the percent distribution and total ganglioside and phospholipid content, determined as described above.
Cholesterol and ceramide were separated by monodimensional HPTLC using the solvent systems hexane/diethylether/acetic acid, 80: 20:1, and hexane/chloroform/acetone/acetic acid, 10:35:10:1, respectively. Cholesterol and ceramide were quantified after separation on HPTLC followed by visualization with 15% concentrated sulfuric acid in 1-butanol. The quantities of cholesterol and ceramide were determined by densitometry and comparison with 0.1-2 g of standard compounds using the Molecular Analyst program (Bio-Rad Laboratories). In the case of cholesterol, the S.D. of the procedure was Ϯ20% in the range of 0.1-0.5 g, and Ϯ10% in the range of 1-2 g. These values were similar to those already reported for analytical methods based on HPTLC separation and charring of cholesterol (52). In preliminary experiments, we obtained similar results on the cell cholesterol content using the TLC-sulfuric acid procedure and the cholesterol oxidase assay. The data on the cellular ceramide content obtained by the TLC-sulfuric acid procedure were within 5% of those in the literature, obtained by the diacylglycerol kinase method (53).
Gas-Liquid Chromatography and Mass Spectrometry-The fatty acid composition of the total lipids extracted from cell lysates and from the sphingolipid-enriched sucrose gradient fractions prepared from cerebellar granule cells at the 8th day in culture was determined by GC and GC-MS. Samples of the lipid extracts were dissolved in 0.5 ml of methanol/5 M HCl, 9:1 (v/v), and maintained at 75°for 16 h. Fatty acid methyl esters were extracted three times with 1.5 ml of n-hexane. The hexane solution was dried, and the residue was dissolved in 50 l of dichloromethane. Fatty acid methyl esters were separated on a 25-m SPB-5 capillary column using a Dani-3865 gas chromatograph, equipped with a program temperature vaporization injector. Analyses were carried out using a total injection program, at 5°C/min, from 180 to 240°C. Mass spectrometry characterization of the fatty acid methyl esters was carried out on an hp-5890 GC-MS equipped with a 30-m HP-5 capillary column. ESI-MS of PC samples, obtained by preparative TLC from cell lysates and sphingolipid-enriched fraction lipid extracts, was carried out on a ThermoQuest Finningan LCQdeca mass spectrometer equipped with an electrospray ion source and a Xcalibur data system. Samples were dissolved in methanol at a concentration of 2-3 ng/l and introduced into the electrospray needle by mechanical infusion through a microsyringe at a flow rate of 3 l/min. Ionization was performed under the following conditions: capillary voltage, 4 keV; sheath gas flow, 50 arbitrary units; capillary temperature, 260°C.
Protein Analysis-Cell lysates and sucrose gradient fractions obtained after labeling cerebellar granule cells with [ 32 P]orthophosphate or [ 35 S]methionine were analyzed to determine protein content and pattern. 35 S-Proteins from the sphingolipid-enriched membrane fractions were analyzed by SDS-PAGE and by two-dimensional electrophoresis (isoelectric focusing on a pH gradient from 3.5 to 10 in the presence of 9.5 M urea and 1% Nonidet P-40, followed by 10% acrylamide SDS-PAGE for the second dimension) (30,54). After separation, proteins were transferred to PVDF membranes, and radioactive proteins were detected by autoradiography.
The presence of c-Src, Lyn, Fyn, Akt, and PI3K was assessed by immunoblotting with commercially available specific antibodies, followed by reaction with secondary horseradish peroxidase-conjugated antibody and enhanced chemiluminescence detection (Pierce Supersignal).
Analysis of phosphotyrosine-containing proteins was by immunoprecipitation (44). Aliquots of the fractions obtained from cells labeled with [ 32 P]orthophosphate (containing ϳ15-30 g of protein) were diluted 10-fold in immunoprecipitation (IP) buffer (50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 2 mM NaF, 1 mM EDTA, 1 mM EGTA, 1 mM Na 3 VO 4 , 1 mM phenylmethylsulfonyl fluoride, 75 milliunits/ml aprotinin, 1% Triton X-100). After preclearing for nonspecific binding, 2 g/ml mouse antiphosphotyrosine monoclonal IgG 2b or 2 g/ml normal mouse IgG (as negative control) was added to the supernatants. The mixtures were placed on a rotary stirrer overnight at 4°C. Immunoprecipitates were recovered using protein A/G-Sepharose beads, washed three times with IP buffer, recovered by centrifugation (270 ϫ g, 2 min), suspended in 50 l of SDS-sample buffer, heated to 95°C for 3 min, and centrifuged (1000 ϫ g, 2 min). The radioactivity associated with immunoprecipitates was determined by liquid scintillation counting. The total amount of immunoprecipitable radioactivity was calculated for each fraction and for the cell lysate. Data were expressed for each fraction as the percentage of total immunoprecipitable radioactivity present in the amount of cell lysate loaded on gradient. Radioactivity associated with negative controls never exceeded 5% of radioactivity found in immunoprecipitates.
Other Analytical Methods-The radioactivity associated with cells, with cell fractions, with lipids, and with delipidized pellets was determined by liquid scintillation counting. Digital autoradiography of the HPTLC plates and of the PVDF membranes was performed with a Beta-Imager 2000 instrument (Biospace, Paris) using an acquisition time of about 24 h. The radioactivity associated with individual lipids and proteins was determined with the specific ␤-Vision software provided by Biospace. Autoradiography of 32 P-and 35 S-labeled proteins was carried out using Kodak Biomax MR and MS films. Protein content was determined according to Lowry (55) using bovine serum albumin as reference standard.

Protein and Lipid Compositions of Rat Cerebellar Granule
Cells Developing in Vitro-The protein and lipid composition of cerebellar granule cells at different stages of neuronal development in vitro are reported, as absolute and relative values, in Tables I and II. An overall increase in cell protein and lipid content was observed during the development in culture. As expected, glycerophospholipids, which represent the bulk membrane lipids, comprised the highest lipid content of these cells at all investigated stages of development, 79 -82% of total cell lipids. The glycerophospholipid content significantly increased within the days in culture. We observed a 2.7-fold increase from the 2nd to the 8th day in culture, and a further 1.4-fold increase from the 8th to the 17th day in culture. Within the different glycerophospholipids, PC and PE were the most abundant at all investigated stages of development (Table II), being about 50 and 20% of total glycerophospholipids, respectively. The glycerophospholipid patterns (Table II) were very similar during development in culture, with two major exceptions, the plasmalogen species of PE and PC. In undifferentiated cells, PPC was relatively abundant, representing 11.7% of total glycerophospholipids; its relative amount dramatically dropped to 1-3% at later stages of development. Relative PPE content increased from 3.7% at the 2nd day in culture to 12.6% at the 8th day in culture (corresponding to a 9-fold increase in the absolute mass content, unique among all phospholipids) and remained relatively high at the latest investigated time (6.9%).
Cholesterol was the second major lipid in granule cells at the three stages of culture (9 -18% of total cell lipids, Table I). Its absolute cellular content increased about 1.7-fold from the 2nd to the 8th day in culture, and remained almost unchanged later on during development. Hence, the cholesterol/glycerophospholipids molar ratio markedly decreased during in vitro development of cerebellar neurons.
Sphingolipids were minor components of undifferentiated cells, being 0.26%, 1.68%, and 0.48% of total cell components in the case of ceramide, sphingomyelin, and gangliosides, respectively, as shown in Table I. Their content markedly increased from the 2nd to the 8th day in culture, as previously reported in the case of gangliosides. This increase was particularly high (about 11-fold, by mass), from 0.074 nmol/10 6 cells (0.48% of total cell lipids) at the 2nd day in culture to 0.79 nmol/10 6 cells (1.99% of total cell lipids) at the 8th day in culture, and it was not paralleled by the increase in glycerophospholipids and cholesterol (Table I). Thus, it resulted in an increase in the ganglioside/glycerophospholipid and ganglioside/cholesterol molar ratios. The increases in sphingomyelin and ceramide (Tables I) were also marked, but less dramatic, from 0.260 nmol/10 6 cells (1.68% of total cell lipids) and 0.040 nmol/10 6 cells (0.26% of total cell lipids) at the 2nd day in culture, to 1.00 nmol/10 6 cells (2.52% of total cell lipids) and 0.22 nmol/10 6 cells (0.55% of total cell lipids) at the 8th day in culture, respectively. Comparing the different sphingolipids at the 8th and 17th day in culture, we found that sphingomyelin content remained unchanged, ganglioside content increased slightly, and ceramide content increased to a greater extent. Only in the case of ceramide was the molar ratio of sphingolipid to glycerophospholipid not decreased.
In agreement with previous reports (56,57), some remarkable differences were observed in the relative amounts of the different gangliosides during the time in culture (Table I).
Palmitic acid was the main fatty acid detected in the total complex lipid mixture, (40%), followed by oleic acid (34%), and stearic acid (22%). These data are in good agreement with the fatty acid composition of PC and PE in differentiated cells, previously reported in the literature (58). Fig. 1 shows the patterns of radioactive lipids extracted from cerebellar granule cells at different stages of development in culture after HPTLC separation. Data on the incorporation of radioactivity into different lipid classes are reported in Table I. The incorporation of radioactivity into sphingolipids was very similar at all investigated days in cul-  35 S labeling (as cpm). Sphingolipid radioactivity is from 3 H labeling (as dpm). Glycerophospholipid radioactivity is from 32 P labeling (as cpm). Protein molar content was calculated on the average of the protein molecular mass, as determined by SDS-PAGE of steady-state 35 S-labeled proteins followed by autoradiography and densitometrical analysis and must be considered an approximate value. Percent distribution is referred to the mass content. ture (Table I). Keeping in mind the large differences in their endogenous content, this fact indicates that the sphingolipid turnover is dramatically different in granule cells at different stages of development. In particular, it is much more rapid in undifferentiated cells than later on during development. In fully differentiated cells, the turnover was significantly reduced and remained almost unchanged at the 17th day in culture. This is particularly evident in the case of gangliosides: ganglioside-associated radioactivity was almost identical, whereas their endogenous content increased about 11-fold from the 2nd to the 8th day in culture. Among various sphingolipid classes, the incorporation of radioactivity was not proportional to their mass content even at the same day in culture. The radioactivity ratio between gangliosides and sphingomyelin was 12-, 3-, and 4-fold higher than the mass ratio at the 2nd, 8th, and 17th day in culture, respectively, indicating that ganglioside turnover is always more rapid than sphingomyelin turnover in these cells. This difference is remarkably high at the early stage of development. The radioactivity ratio between gangliosides and ceramide was also always higher than the mass ratio (4-fold at the 2nd and 8th day in culture, and 1.3-fold at the 17th), indicating a faster turnover for gangliosides than for ceramide. However, this difference is very small in cells at the 17th day in culture. Remarkably, these data indicate that both molar content and turnover of ceramide are significantly increased in cells at a late stage of development. This is in good agreement with a role of ceramide in cell senescence, as previously described in the case of fibroblasts (59). Among the different gangliosides, the distribution of radioactivity in each species closely reflects the endogenous patterns at all the investigated days in culture, with the exception of GM3 and GD3 gangliosides in cells at the 2nd day in culture (Table II). The amount of radioactivity incorporated in these gangliosides represent 13.9 and 30.1% of total ganglioside radioactivity. The mass content of GM3 was extremely low under these experimental conditions, and GD3 was 15.1% of total cell gangliosides, indicating that the turnover of these species is surprisingly rapid. We are not able to estimate the turnover of neutral glycosphingolipids, because data about their endogenous content in these cells are not available. No significant differences were observed in the in-corporation of radioactivity into GlcCer and LacCer at different days in culture (about 2 and 5% of total sphingolipid radioactivity, respectively).

Lipid Metabolism in Rat Cerebellar Granule Cells Developing in Vitro-
The incorporation of radioactivity into glycerophospholipids, by metabolic labeling with [ 32 P]orthophosphate, is reported in Table I. The specific radioactivity of glycerophospholipids was 1.5-fold higher and 3.7-fold lower at the 2nd and 17th day in culture, respectively, than at the 8th. This indicates that the overall turnover of this lipid class significantly differs in these cells at different stages of development in culture, being slightly higher in undifferentiated cells and markedly lower in cells at the late stage of development, as compared with fully differentiated cells. Remarkably, dramatic differences were observed among the different glycerophospholipids (Table II). In the case of PE, PC, and PS (the main cell phospholipids), the incorporation of radioactivity, expressed as a percentage of the total [ 32 P]phospholipid radioactivity, was quite constant during the development in culture, and always lower than the percent mass content, indicating that their relative turnovers are similar during this process and relatively slow. On the other hand, the specific radioactivity of PPC was 3-and 86-fold higher at the 8th day in culture that at the 2nd and 17th, respectively, indicating very relevant differences in its turnover. Similar, but less dramatic differences were observed for phosphatidylinositol. A very rapid turnover was also observed for phosphatidylinositol plasmalogen and phosphorylated phosphoinositides, PIP and PIP 2 . In fact, these lipids incorporated large amounts of radioactivity, despite their very low molar content, that made difficult or impossible their chemical detection. In this case it is difficult to compare their relative turnover at different stages of development, but it is worth noting that the incorporation of radioactivity in these lipids is significantly higher in fully differentiated granule cells than in undifferentiated or senescent neurons. Fig. 2 shows the distribution of different lipids or lipid classes in SEMF prepared from rat cerebellar granule cells at different stages of development in culture. The fraction of cellular glycerophospholipids recovered in SEMF was relatively small, varying from 4.84% of total glycerophospholipids at the 2nd day in culture to 17.3% at the 17th. Nevertheless, at all the different stages of development in culture we investigated, glycerophospholipids were the main lipid component of SEMF, representing 44.2-73.0% of total SEMF lipids. Because glycerophospholipids account for the bulk of membrane lipids, these data suggest that the total surface area occupied by the sphingolipid-enriched domains gradually increases during development in culture. A relevant (25-40%) amount of cellular cholesterol was associated with SEMF; cholesterol was very abundant in SEMF in the first stage of culture, about 45% of total SEMF components, then gradually decreased to 15% at 17th day in culture. As clearly shown, all sphingolipids (namely ceramide, sphingomyelin, gangliosides, and neutral glycosphingolipids) are largely associated with SEMF, regardless from the stage of development (40 -70% of the radioactivity associated with each sphingolipid in the cell homogenate). Thus, these structures are more enriched in sphingolipids than in cholesterol. The incorporation of radioactivity into different lipid classes and the lipid composition of SEMF from cerebellar granule cells at the 2nd, 8th, and 17th days in culture are reported in Table III. The data on the cholesterol content were derived from direct chemical analysis. The data on the sphingolipid and glycerophospholipid contents were calculated from the radioactivity distribution in SEMF after labeling cells with [1-3 H]sphingosine or [ 32 P]orthophosphate, respectively, and from the total lipid composition of these cells, as reported in Table I. These data clearly indicate that the enrichments (relative to the homogenates) of SEMF in sphingolipids and cholesterol with respect to glycerophospholipids linearly decreased with the number of days in culture. However, the percent lipid composition of SEMF during development in culture varied in a complex way, as indicated by the data in Table III. In particular, the relative ganglioside and sphingomyelin content was higher in fully differentiated cells, ceramide increased after the initial stage and then remained almost unchanged, while cholesterol gradually decreased along the days in culture.

Lipid Composition of Sphingolipid-enriched Membrane Fractions from Rat Cerebellar Granule Cells Developing in Vitro-
At all the days in culture, the distributions of different gangliosides within SEMF were very similar to the patterns observed in the homogenates ( Fig. 1 and Table IV), with a major exception, related to the content of GD1a in SEMF at the 2nd day in culture. The pattern of radioactive gangliosides observed at the 2nd day in culture was relatively changeable, compared with those at other stages of development. This variability between different experiments can be easily explained keeping in mind the very rapid variations of ganglioside content and patterns at this early stage of differentiation (56,60) and the high variability of the starting biological material (the age of commercially available newborn rats is determined with a precision of 12 h, a lapse of time not negligible at this stage of neonatal development). Despite this variability, GD1a relative content with respect to total ganglioside at the 2nd day in culture was always significantly higher in SEMF (an increase of 20 -40%) than in the cell homogenate (Fig. 1). Fig. 3 shows the distribution of each glycerophospholipid in SEMF from rat cerebellar granule cells at different stages of development in culture with respect to the cell homogenate.
Very relevant differences were observed among different glycerophospholipids and different days in culture. At all the stages of development, the fraction of PC recovered in SEMF was always much higher than that of any other glycerophospholipid, thus making PC by far the most abundant lipid in SEMF (93 and 75% of total SEMF glycerophospholipids at the 2nd and 17th day in culture, respectively, Table IV). In the case of other glycerophospholipids, the fraction associated with SEMF greatly varied among the different days in culture, thus resulting in very different patterns (Table IV). In particular, the relative amount of PE in SEMF gradually increased during differentiation from 2.9% to 26.4%; PPE was particularly high in cells at the latest stage of development; the amount of radioactivity associated with phosphorylated phosphoinositides dramatically increased (especially for phosphatidylinositol 4,5-bisphosphate) during differentiation.
Palmitic acid was the main fatty acid of complex lipids from SEMF. Palmitic acid was significantly enriched in SEMF respect to the cell lysate. In fact, its relative content was 1.4-fold higher in SEMF than in the cell lysate. In parallel, we observed a slight reduction in the relative amount of unsaturated fatty acids in SEMF. SEMF enrichment in palmitic acid was mainly due to dipalmitoyl-PC, that represented the main glycerophospholipid specie in this fraction. PC isolated from the total cell homogenate and from SEMF was analyzed by ESI-MS according with previous methods (61). The MS spectrum of PC contained the molecular ions at m/z 756, 782, 810 corresponding to the dipalmitoyl-, the 1-palmitoyl-2-oleyl-, and 1-stearoyl-2oleyl-species. Fig. 4 clearly shows the large increase in the relative abundance of the molecular ion corresponding to dipalmitoyl-PC, at m/z 756, in SEMF.

Protein Compositions of Sphingolipid-enriched Membrane Fractions from Rat Cerebellar Granule Cells Developing in
Vitro-Previous evidence indicated that proteins associated with sphingolipids-enriched membrane domains are selected and are potentially involved in signal transduction events (12,30,44). Thus, we analyzed the protein composition of SEMF prepared from cerebellar granule cells at different stages of development in culture. The protein content of granule cells increased during culture, as expected, from 26.2 to 85.6 g/10 6 cells. To compare these data to those for lipids, we calculated that at the 2nd, 8th, and 17th day in culture the approximate molar protein content was 0.44, 1.25 and 1.43 Ϯ 20% nmol/10 6 cells, respectively. The SEMF always contained a very minor portion of cell proteins, but this portion dramatically increased from the 2nd to the 8th day in culture (0.3 Ϯ 0.1%, 1.6 Ϯ 0.2%, and 2.3 Ϯ 0.3%, of the total cell protein at the 2nd, 8th, and 17th day in culture, respectively) (Fig. 5, left panel). Fig. 6 shows the two-dimensional 35 S-protein patterns of SEMF prepared from granule cells at different days in culture. The protein pattern in differentiated cells comprised a wide variety of species, encompassing the whole molecular mass and isoelectric point range. On the other hand, the protein patterns of cells  35 S labeling (as cpm). Sphingolipid radioactivity is from 3 H labeling (as dpm). Glycerophospholipid radioactivity is from 32 P labeling (as cpm). The mass composition of the sphingolipid-enriched membrane fraction was calculated on the basis of the endogenous lipid and protein content in the homogenate (reported in Table I) and the distribution of radioactivity in each species within the gradient. at 2nd and 17th DIC were much simpler and particularly lacking in high molecular mass proteins. A low molecular mass protein (about 20 kDa) was particularly abundant in SEMF from cells at the latest stage of development. Fig. 5 (right panel) shows the distribution of phosphotyrosine-containing proteins in SEMF, as determined by immunoprecipitation with anti-phosphotyrosine antibody. The portion of phosphotyrosine-containing proteins associated with SEMF was relatively high and markedly increased along the development in culture (7.7%, 22.3%, and 32.1% of total cell phosphotyrosine-containing proteins at 2nd, 8th, and 17th day in culture, respectively).
Among proteins belonging to the SEMF, three members of the Src family were identified by immunoblotting with specific antibodies. Fig. 7 shows the distribution patterns of Src-family tyrosine kinases c-Src, Fyn, and Lyn in SEMF and cell homogenates from cerebellar granule cells at different stages of development in culture. From the relative intensities of the immunoblotting signals, we calculated that the amount of c-Src associated with SEMF was about 20% of total c-Src present in these cells, indicating a very high enrichment of c-Src in this fraction. The amount of c-Src recovered in SEMF was not significantly changed during the development. Fyn was also highly enriched in SEMF, but the amount of Fyn associated with SEMF increased from 15% of total Fyn at the 2nd day in TABLE IV Ganglioside and glycerophospholipid species composition of the sphingolipid-enriched membrane fraction from rat cerebellar granule cells during development in culture Sphingolipid radioactivity is from 3 H labeling (as dpm). Glycerophospholipid radioactivity is from 32 P labeling (as cpm). The mass composition of the sphingolipid-enriched membrane fraction was calculated on the basis of the endogenous lipid content in the homogenate (reported in Tables  I and II) and the distribution of radioactivity in each species within the gradient.  culture, to about 45% at the 8th and 17th day in culture. Lyn was clearly observed in SEMF from cells at the 8th and 17th day in culture, although the amount of Lyn associated with SEMF was relatively low (about 5% of the total cell content), but it could not be detected in SEMF prepared from cells at 2nd day in culture (even loading large amounts of sample on the gel and prolonging the exposure with the ECL reagent).
Given the high levels of phosphorylated phosphatidylinositols in SEMF at some stages of development in culture, and the property of these cells to remain vital in culture for long time, we analyzed the distribution patterns of PI3K and of the serine-threonine kinase Akt (a downstream target of PI3K signaling known to be involved in the control of neuronal survival), in SEMF and cell homogenates from cerebellar granule cells at different stages of development in culture. As shown in Fig. 7, both proteins were expressed in our cells but could not be detected in the SEMF under our experimental conditions. DISCUSSION Membrane lipids have been reported to change to some extent during the development, maturation, or aging of the nervous system, or during the onset of other physiological or pathological changes in neuronal functions. In our study, the changes in the composition and turnover of each membrane lipid class and individual lipid during the development of cultured neurons were related to changes in the organization of sphingolipid-enriched membrane domains. As a model, we used primary cultures of rat cerebellar granule cells, which represent a very homogeneous population of neurons. This highly simplified model obviously lacks the ability to provide any information about the interactions between neurons and glia. However, these cells spontaneously undergo morphological and biochemical changes similar to those occurring during the different stages of in vivo neuronal development, thus representing a valuable model for the study of neuronal differentiation. We performed our experiments at three different stages of in vitro development: the 2nd day in culture, corresponding to the initial stage of neuronal differentiation and neuritogenesis, the 8th day in culture, corresponding to morphologically and biochemically fully differentiated neurons, and the 17th day in culture, corresponding to a late stage of neuronal development, followed by the eventual age-induced apoptotic cell death. The content of all membrane lipid classes (glycerophospholipids, cholesterol, and sphingolipids) greatly increased during the progression from undifferentiated to fully differentiated granule cells, indicating an overall increase in the surface and complexity of neuronal plasma membrane, corresponding to the extension of a very elaborate neurite network. This event is also characterized by a great increase in the cell protein content. This increase is higher in the case of sphingolipids, particularly gangliosides, than of other lipids, in agreement with previously reported data (56,57,60). Thus, the surface density of gangliosides in the membrane of fully differentiated neurons is much higher than in undifferentiated cells. The lipid turn- over also varied during this phase of development, being higher in undifferentiated cells both for phospholipids and sphingolipids. The latter is in agreement with the previous finding that, in these cells, the activity of 3-ketosphinganine synthase markedly decreases during differentiation (62). In particular, ganglioside turnover at the 2nd day in culture was about 11-fold faster than at the 8th day in culture, whereas in the case of phospholipids this difference was 1.5-fold. The turnover of gangliosides was always more rapid than sphingomyelin and ceramide turnover, and this difference was the highest in undifferentiated cells (12-and 4-fold, for sphingomyelin and ceramide, respectively). In undifferentiated cells, the turnover of some ganglioside species, namely GM3 and GD3, was surprisingly high, as indicated by the high amount of radioactivity associated with these species compared with their mass content. Moving toward a later stage of development, resembling that of aging neurons, the density of gangliosides in the cell membrane showed little or no change, as indicated by the constant mass ratios between gangliosides and glycerophospholipids or gangliosides and cholesterol.
The turnover of glycerophospholipids was significantly lower in aging cerebellar neurons (this fact probably reflects a slower overall renewal of the plasma membrane), whereas the turnover of gangliosides was similar to that in fully differentiated cells.
We focused our attention on the organization of sphingolipidenriched domains, which are emerging as a compartment within the plasma membrane, where most cell gangliosides are concentrated and segregated together with selected proteins and other lipids. The results we obtained are summarized in Tables III and IV. Our data indicate that a membrane compartment, characterized by a very high enrichment in sphingolipids and cholesterol and for this reason named sphingolipid-enriched domain, always exists in cerebellar granule neurons, regardless of the stage of development. The bulk structure of these domains is largely determined by PC, and, within the different PC species, dipalmitoyl-PC was much more abundant than in the total cell membrane. Thus, the lipid core of the SEMF results highly enriched in palmitic acid respect to the cell membrane.
The surface occupied by these structures seems to increase during development in culture, in a way that is not simply proportional to the overall cell membrane surface, as indicated by their relative glycerophospholipid content. Moreover, a particularly high ganglioside density (as indicated by the ganglioside/glycerophospholipid mass ratio) within the domains seems to be always required, independent of the developmental stage. In fact, in undifferentiated neurons, where the ganglioside content is lower and the neurons are rapidly synthesizing these lipids, we observed in the sphingolipid-enriched domain the highest enrichment in gangliosides with respect to the cell homogenate, an 8-fold increase in the ganglioside/glycerophospholipid molar ratio. A similar behavior was also observed for sphingomyelin. The SEMF from undifferentiated neurons is also specifically characterized by the particular enrichment in GD1a ganglioside. The highest ganglioside/glycerophospholipid mass ratio was measured in sphingolipid-enriched domains from fully differentiated neurons, which are also characterized by a relatively high protein content (6-fold higher than in undifferentiated cells) and a complex protein composition. Despite the very low protein mass content, SEMF prepared from neurons at all stages of in vitro development we investigated contained a significant fraction of cell phosphotyrosine-containing proteins, that increased during differentiation. This evidence indicates that only a relatively restricted number of highly selected protein molecules are segregated together in a lipid environment. The data discussed above indicate that this association is dependent on the developmental status of the neuron and could be of primary importance in modulating it. In aging neurons, sphingolipid-enriched domains structure is also characterized by a high sphingolipid density and protein and phosphotyrosine-containing protein enrichment. However, in these domains the ganglioside density is decreased respect to fully differentiated neurons, and the protein pattern seems to be less complex. It is noteworthy that, in granule cells at the latest stage of development, the increase in ceramide levels is higher than those of other sphingolipids. Cerebellar granule cells after the 17th day in culture undergo massive age-induced death by an apoptotic process. Because ceramide has been widely implicated as a mediator of cell senescence (59) and programmed cell death (reviewed in Ref. 7), a ceramide-mediated pathway could be more relevant in SEMF from aging neurons. Thus, sphingolipid-enriched domains could function as "glycosignaling domains" during neuronal differentiation and neuritogenesis and as "ceramidesignaling domains" during neuronal aging and age-induced apoptosis.
If the dynamic association of a specific set of proteins and lipids within sphingolipid-enriched domains has a functional meaning related to neuronal development, it is very important to determine which kind of signal is transduced in these units. We focused our attention on Src family protein-tyrosine kinases. A specific association of Src-family proteins with gangliosides has been reported in many cellular types, including nervous system, and has been implicated in the process of neuronal differentiation. In neuroblastoma Neuro2a cells, c-Src and its regulatory protein, Csk, are organized in sphingolipid-enriched membrane domains, and ganglioside-induced differentiation is mediated by the modulation of c-Src activity within sphingolipid-enriched membrane domains, followed by mitogen-activated protein kinase activation (29). In sphingolipid-enriched domains from fully differentiated rat cerebellar granule cells, Src, Lyn, and Csk are associated with other proteins and the sphingolipid component of the domain (30). In particular, the neural cell adhesion molecule TAG-1 is interacting with Lyn, and treatment of cultured cerebellar neurons with antibodies against GD3 ganglioside or TAG-1 induce Lyn activation, with consequent tyrosine phosphorylation of an 80-kDa protein (31,32). Moreover, in these cells, the artificially induced depletion of glycosphingolipids at the cell surface is able to disrupt Lyn-mediated TAG-1 signaling and to interfere with the association of TAG-1 with the domain (32).
These data strongly support the hypothesis that sphingolipids are essential components of sphingolipid-enriched membrane domains as signaling units. The question is, if any naturally occurring changes in the lipid composition of the domain, such as those we observed during in vitro development of cerebellar neurons, will lead to a change in the functional status of signaling proteins in the domain itself? Our data clearly show that, in cultured cerebellar granule cells at different stages of development, the association of different Srcfamily proteins with the sphingolipid-enriched domain follows different patterns. The amount of c-Src associated with SEMF was very high at all the stages of development and did not significantly change. Fyn and Lyn association with SEMF markedly increased during neuronal development, but the amount of Lyn recovered in the sphingolipid-enriched fraction was much lower than that of c-Src and Fyn (in the case of undifferentiated cells, no Lyn was detected in SEMF). Thus, the specific association of signaling proteins with the domain seems to be dependent on its structure, as determined by its lipid composition. The translocation of these proteins from or into the domain, consequent to organizational changes, could be a mechanism of regulation. In Neuro2a neuroblastoma cells, the activation of c-Src in the sphingolipid-enriched domain, that mediates ganglioside-induced neuritogenesis, is accompanied by the removal of Csk from the domain (29). Because Csk is an inhibitory regulatory kinase of c-Src, this is a good example of regulation mediated by an alteration of sphingolipidenriched domain structure. Thus, it will be very interesting to investigate whether the activity of these proteins within sphingolipid-enriched domains actually changes during in vitro development and whether the transient association with other protein molecules could be responsible for this change.
More specific differences in the levels, turnover, and organization of single lipid species were observed. Some of them are intriguing, even if still difficult to interpret. For example, a single ganglioside, GD1a, is significantly more enriched than others in sphingolipid-enriched domains from undifferentiated cells. This could indicate a specific, yet unknown role for GD1a. PPE is particularly enriched in sphingolipid-enriched domains from aging neurons, and several lines of evidence indicate that the receptor-mediated breakdown of plasmalogens is a relevant event in neural membranes during neurodegeneration (8). The turnover of phosphorylated phosphoinositides is particularly intense in sphingolipid-enriched domains from fully differentiated neurons. Thus, lipid-mediated signaling pathways other than sphingolipid signaling could play a role within these domains. This will not be surprising, because our data clearly indicate that nonsphingoid lipids also undergo a selection process leading to the segregation within domains. Nevertheless, neither the phosphoinositide 3-kinase nor the serine-threonine kinase Akt, which is known to be one of the phosphorylated phosphoinositide signaling downstream targets, were detected in the sphingolipidenriched domains.