Ceramide inhibits axonal growth and nerve growth factor uptake without compromising the viability of sympathetic neurons.

Ceramide inhibits axonal growth of cultured rat sympathetic neurons when the ceramide content of distal axons, but not cell bodies, is increased (Posse de Chaves, E. I., Bussiere, M. Vance, D. E., Campenot, R. B., and Vance, J.E. (1997) J. Biol. Chem. 272, 3028-3035). We now report that inhibition of growth does not result from cell death since although ceramide is a known apoptotic agent, C(6)-ceramide given to the neurons for 24 h did not cause cell death but instead protected the neurons from death induced by deprivation of nerve growth factor (NGF). We also find that a pool of ceramide generated from sphingomyelin in distal axons, but not cell bodies, inhibits axonal growth. Analysis of endogenous sphingomyelinase activities demonstrated that distal axons are rich in neutral sphingomyelinase activity but contain almost no acidic sphingomyelinase, which is concentrated in cell bodies/proximal axons. Together, these observations are consistent with the idea that generation of ceramide from sphingomyelin by a neutral sphingomyelinase in axons inhibits axonal growth. Furthermore, we demonstrate that treatment of distal axons with ceramide inhibits the uptake of NGF and low density lipoproteins by distal axons by approximately 70 and 40%, respectively, suggesting that the inhibition of axonal growth by ceramide might be due, at least in part, to impaired endocytosis of NGF. However, inhibition of endocytosis of NGF by ceramide could not be ascribed to decreased phosphorylation of TrkA.

Ceramide has been widely recognized as a lipid second messenger in many cell types, and ceramide can be generated from a variety of sources by the action of the enzyme ceramide synthase, from glycosphingolipid catabolism, or from the hydrolysis of sphingomyelin via one of several sphingomyelinases that have been identified (1)(2)(3)(4)(5)(6). These sphingomyelinases dif-fer in their subcellular location, pH dependence, and cation requirements. The best studied is the lysosomal sphingomyelinase that has an acidic pH optimum (7). Other sphingomyelinases have neutral pH optima, distinct requirements for divalent cations, and have been localized to the plasma membrane (8,9). The acidic (lysosomal) sphingomyelinase appears, at least in some cell types, to be responsible for generating a pool of ceramide that induces apoptosis (10 -12).
Ceramide is also a regulator of axonal growth in neurons. We have shown previously that axonal extension is inhibited when distal axons of rat sympathetic neurons are incubated with either the ceramide analog, C 6 -ceramide, 1 or with the glucosylceramide synthase inhibitor, PPMP. In contrast, when cell bodies are subjected to the same treatment, axonal growth is not impaired (36). Since distinct pools of ceramide can mediate different biological responses (37,38), we have now generated a pool of ceramide from cell surface sphingomyelin by treatment of distal axons with exogenously added bacterial sphingomyelinase, and we have shown that this pool of ceramide similarly inhibits axonal growth. The endogenous sphingomyelinase activity in distal axons is almost entirely a neutral sphingomyelinase, whereas the acidic sphingomyelinase is restricted to cell bodies/proximal axons. These observations suggest that a neutral sphingomyelinase in distal axons might generate the pool of ceramide that inhibits axonal growth. We have also confirmed that the inhibitory effect of ceramide on axonal elongation occurs independently of cell death and, moreover, that ceramide protects sympathetic neurons against apoptosis induced by NGF withdrawal. In addition, we show that ceramide inhibits the uptake of NGF by a mechanism that does not involve reduced phosphorylation of TrkA. Since axonal elongation of sympathetic neurons requires NGF, inhibition of axonal growth by ceramide might, at least in part, result from decreased endocytosis of NGF.

EXPERIMENTAL PROCEDURES
Materials-Trypsin was purchased from Calbiochem, and L15-CO 2 culture medium was from Life Technologies, Inc. Adult rat serum was prepared from blood supplied by Lab Animal Services, University of Alberta. Rat tail collagen was isolated as described (39). Nerve growth factor (2.5 S) was provided by Alomone Laboratories Ltd. (Jerusalem, Israel). The Teflon dividers were from Tyler Research Instruments (Edmonton, Alberta, Canada). C 6  Culture of Neurons-Sympathetic neurons were isolated from superior cervical ganglia of newborn Harlan Sprague-Dawley rats by treatment of the ganglia with trypsin 0.1% (w/v) and mechanical dissociation, as previously described (40,41). The basal culture medium was L15-CO 2 supplemented with the prescribed additives and 6% methylcellulose. Non-neuronal cells were eliminated by supplying 10 M cytosine arabinoside during the first 6 -7 days of culture. Neurons were cultured in two types of culture dishes, either 24-well plates at a density of 2 ganglia/well (mass cultures) or 3-compartment culture dishes, constructed as described previously (40,42). Briefly, rat tail collagen was air-dried onto 35-mm culture dishes. Parallel scratches were made in the collagen substratum, and a Teflon divider was attached to the collagen surface with silicone grease. The neuronal cell suspension (2-3 ϫ 10 3 neurons/dish) was plated in the center compartment of each dish. The axons extended at a rate of ϳ1 mm/day into adjacent side compartments that contained 100 ng/ml NGF. Typically, neurons were cultured for 9 -13 days prior to the start of an experiment.
Measurement of Axonal Extension-Neurons were plated in the center compartment of compartmented dishes. Distal axons were mechanically removed from left and right compartments with a jet of sterile water delivered with a syringe through a 22-gauge needle in a process termed "axotomy." The water was aspirated and the wash repeated twice, after which fresh medium was added. This procedure effectively removes all visible traces of axons from the side compartments. Axonal extension was measured on individual tracks using an inverted microscope fitted with a digitizer that tracks stage movements to an accuracy of Ϯ5 m. Axonal extension was defined as the distance from the grease barrier to the furthest extending axon (40).
Assessment of Neuronal Viability-Neurons were cultured in 24-well dishes for 12 days and then incubated with basal medium containing the indicated concentrations of C 6 -ceramide or C 6 -dihydroceramide in dimethyl sulfoxide. Control cultures were given equivalent amounts of dimethyl sulfoxide without the ceramide analog; the maximum final concentration of dimethyl sulfoxide to which the neurons were exposed was 0.5% (v:v). After 24 h, mitochondrial function was assessed using the Cell Titer 96 kit (Promega Corp., Madison, WI).
Hydrolysis of Sphingomyelin with Exogenously Added Bacterial Sphingomyelinase-Neurons were cultured in 24-well dishes for 9 days and then incubated with [ 3 H]choline (10 Ci/ml) for 3 days to allow steady-state labeling of choline-containing lipids. At that time, neurons were treated with 500 milliunits/ml bacterial sphingomyelinase from Staphylococcus aureus for the indicated times. Cellular material was collected in methanol:water, 1:1 (v/v), and then chloroform was added to give a final ratio of chloroform:methanol:water of 2:1:1 (v/v). The lipids were extracted according to the procedure of Folch et al. (43) and separated by thin layer chromatography in the solvent system chloroform:methanol:acetic acid:formic acid:water, 70:30:12:4:2 (v/v). Bands corresponding to authentic phosphatidylcholine and sphingomyelin were scraped from the plates and radioactivity was measured. Aliquots of the organic phase were used to determine the amount of lipid phosphorus (44).
Transport of C 6 -NBD-ceramide-C 6 -NBD-ceramide (20 M) was sup-plied to either the center or side compartments of 13-day-old, centerplated compartmented cultures of neurons. After 24 h, the medium was aspirated, and the neurons were rinsed several times with ice-cold, Tris-buffered saline (20 mM Tris-HCl (pH 7.4), 150 mM NaCl). Cellular material from the center and side compartments was harvested separately, and lipids were extracted. The amounts of lipid phosphorus and fluorescence were determined in aliquots from the organic phase. Fluorescence was quantitated using a Hitachi F-2000 fluorimeter (excitation wavelength 464 nm, emission wavelength 532 nm). Measurement of Neuronal Sphingomyelinase Activity-Neutral and acidic sphingomyelinase activities were measured essentially as described previously (1,45,46). Neurons were washed extensively with ice-cold, Tris-buffered saline (20 mM Tris-HCl (pH 7.4), 150 mM NaCl) and collected in ice-cold buffer containing 10 mM Tris-HCl (pH 7.4), 0.1% Triton X-100, and protease inhibitors (0.1 M aprotinin, 1 M leupeptin, 1 M pepstatin and 0.1 mM phenylmethylsulfonyl fluoride). The cell extract was sonicated and then centrifuged for 5 min at 800 ϫ g. An aliquot (25 l) of sonicate containing 20 -30 g of protein was typically used for the enzymatic assays. The substrate for each assay consisted of 60 pmol of [methyl-14 C]sphingomyelin (45 mCi/mmol) and 900 pmol of unlabeled sphingomyelin in 25 l of the following buffers. For determination of neutral sphingomyelinase activity, the buffer contained 0.2 M Tris-HCl (pH 7.5), 0.1% Triton X-100 and 10 mM MgCl 2 ; the buffer used for measurement of acidic sphingomyelinase activity contained 0.2 M sodium acetate (pH 5.0) and 0.1% Triton X-100. The enzymatic reaction was initiated by addition of substrate, and the samples were incubated at 37°C for 30 min. The reaction was terminated by addition of CHCl 3 :CH 3 OH:H 2 O (2:1:1) (43). Radioactivity was measured in the aqueous phase that contained released [ 14 C]phosphocholine.
Radioiodination of Nerve Growth Factor-NGF was radioiodinated by the lactoperoxidase method (47,48). The following components were mixed at room temperature for 1 h: 3-5 l (1.5 mCi) of Na 125 I, 10 g of NGF (1 g/ml), 37 l of 0.5 M potassium phosphate (pH 7.4), 10 l of lactoperoxidase (33 g/ml), and 10 l of 0.003% H 2 O 2 . After 30 min, an additional 10 l of H 2 O 2 were added. The reaction was terminated by addition of 5 l of ␤-mercaptoethanol and 415 l of bovine serum albumin (1 mg/ml) in potassium phosphate buffer (pH 7.4). 125 I-NGF was precipitated by addition of an equal volume of 20% trichloroacetic acid. Unbound iodine was removed by passage through a Sephadex G-25M gel filtration column. Typical specific radioactivities of radioiodinated NGF were 200 -300 cpm/pg protein. The 125 I-NGF was used within 2 weeks of its preparation. Molar concentrations of NGF were based on a molecular weight of 26,000.
Radiolabeling of Low Density Lipoproteins-Human LDL were isolated from plasma by sequential ultracentrifugation (49) and then radiolabeled with iodine monochloride (50) at 4°C (pH 10) which resulted in almost no labeling of lipid. Unbound Na 125 I was removed by dialysis against saline/EDTA at 4°C for 3-4 days. The specific radioactivity of 125 I-labeled LDL was determined after precipitation of the protein with 10% trichloroacetic acid.
Uptake and Retrograde Transport of NGF-Sympathetic neurons were plated in the center compartment of 3-compartment dishes and cultured for 7 days. Distal axons were removed and then allowed to regenerate for 4 -5 days after which they were incubated in medium containing one of 10 M C 6 -ceramide, 10 M C 6 -dihydroceramide, or no ceramide analog. After 24 h, 125 I-NGF (50 ng/ml) was added to distal axons in medium containing the same ceramide analogs, and the neurons were incubated for an additional 18 h during which time 125 I-NGF was retrogradely transported. Medium bathing cell bodies/proximal axons was collected first, and then cell bodies/proximal axons were harvested by addition of 10% sodium deoxycholate. Radioactivity was measured in both cellular material and culture medium from the cell body-containing compartment which together represented retrogradely transported 125 I-NGF. For determination of the amount of 125 I-NGF associated with distal axons, the axons were washed twice with phosphate-buffered saline at 4°C and then harvested by treatment with 10% deoxycholate. Radioactivity was measured, and the protein content was determined by the BCA assay (Pierce). The amount of retrogradely transported 125 I-NGF was calculated as radioactivity in cell bodies/ proximal axons combined with radioactivity in the medium bathing cell bodies/proximal axons. As in previous studies (48), background diffusion of 125 I-NGF under the Teflon divider was negligible. Nonspecific binding and transport of 125 I-NGF were assessed by performing the same procedures with cultures in which distal axons were incubated with 125 I-NGF in the presence of a 100-fold higher concentration of unlabeled NGF. The specific transport and specific axon association of 125 I-NGF were calculated by subtracting nonspecific values from the total amount of 125 I bound and transported. Typically, the nonspecific transport of 125 I-NGF into the center compartment accounted for less than 10% of the total transported, whereas the nonspecific association of 125 I-NGF with distal axons accounted for 45-55% of associated radioactivity (48).
Analysis of LDL Uptake-Neurons were plated in the center compartment of compartmented dishes and cultured for 7 days. The neurons were then axotomized and allowed to regenerate for 5 days, after which one of 10 M C 6 -ceramide, 10 M C 6 -dihydroceramide, or no ceramide analog was incubated with distal axons for 24 h. Subsequently, 125 I-LDL (100 g of cholesterol/ml) was added to distal axons for 18 h in the presence or absence of the same ceramide analog. The neurons were washed, and cellular material from center and side compartments was harvested separately and analyzed as described for uptake/transport of NGF.
Phosphorylation of TrkA-Neurons were plated in the center compartment of 3-compartmented dishes. After 7 days, the cells were axotomized, and axons were allowed to regenerate for 5 days. The incubation was continued for an additional 24 h either in the absence of ceramide or in the presence of C 6 -ceramide (10 M) or C 6 -dihydroceramide (10 M) in distal axons. During the first 18 h of this incubation, NGF (100 ng/ml) was present, whereas for the remaining 6 h the axons were incubated with medium lacking NGF so that the level of NGFinduced Trk phosphorylation was reduced to a basal level. Next, culture medium containing 100 ng/ml NGF was added for 15 min to induce TrkA phosphorylation. The extent of TrkA phosphorylation in distal axons was investigated by immunoblotting of proteins in lysates of distal axons.
Immunoblotting-Sympathetic neurons that had been cultured in 24-well plates were washed with ice-cold Tris-buffered saline containing 1 mM sodium orthovanadate and 10 mM sodium fluoride. Cellular material pooled from 4 dishes was harvested into sample buffer containing 0.1% SDS and then boiled for 5 min. The proteins were electrophoresed on 8% polyacrylamide gels containing 0.1% SDS and then transferred to polyvinylidene difluoride membranes in Tris:glycine buffer for 1 h at 4°C. The membranes were incubated for 1 h in blocking buffer (5% horse serum and 0.1% Tween 20 in Tris-buffered saline), and the proteins were immunoblotted using the following antibodies at the indicated dilutions in blocking buffer: monoclonal anti-phosphotyrosine clone 4G10 (1:5000, Upstate Biotechnology Inc., Lake Placid, NY); rabbit polyclonal anti-human TrkA C-14 (1:1000, Santa Cruz Biotechnology, Santa Cruz, CA); monoclonal anti-rat ␤-tubulin (Tub 2.1, Sigma). Immunoreactivity was detected by enhanced chemiluminescence. In some cases, as indicated, antibody was stripped from the membrane using Restore Western blot Stripping Buffer (Pierce) according to manufacturer's directions, and the blot was re-probed with a different antibody.

Ceramide Does Not Induce Neuronal Death but Protects from
Cell Death Caused by NGF Withdrawal-We have reported previously that elevation of ceramide within distal axons of rat sympathetic neurons by two independent approaches inhibits axonal growth (36). Since ceramide has been implicated as an inducer of apoptotic cell death in numerous cell types (9,(12)(13)(14)(15)(16)(17)(18)(19), including neurons and neuron-derived cells (20, 25, 26, 29 -31, 33), we assessed the viability of rat sympathetic neurons treated with cell-permeable C 6 -ceramide using the Cell Titer 96 assay that measures mitochondrial function. Sympathetic neurons were cultured for 12 days in 24-well plates in the presence of 100 ng/ml NGF and then incubated for 24 h with various concentrations (0 -50 M) of either C 6 -ceramide or the inactive analog, C 6 -dihydroceramide. Fig. 1 demonstrates that neither ceramide analog, at any of the concentrations used, decreased neuronal viability.
Not only did C 6 -ceramide not induce cell death, but this lipid actually protected against cell death induced by withdrawal of NGF. For the experiment depicted in Fig. 2, rat sympathetic neurons were cultured in 24-well plates in the presence of 100 ng/ml NGF. After 12 days, NGF was depleted from half of the cultures by providing medium that lacked NGF and contained anti-NGF antibodies (24 nM). At the same time, the neurons were incubated with or without C 6 -ceramide (20 M). As controls, other neurons were cultured throughout the experiment in the presence of 100 ng/ml NGF. After 1 and 2 days, mitochondrial function/cell viability was assessed by the Cell Titer 96 assay. As shown in Fig. 1, in the presence of NGF, C 6ceramide did not diminish neuronal survival. In the absence of ceramide, depletion of NGF reduced cell viability by ϳ45% after 1 day and by 60% after 2 days, compared with that of cells grown in the presence of NGF (Fig. 2). In contrast, when the neurons were incubated with medium that had been depleted of NGF, but contained ceramide, cell viability was the same as that of neurons cultured in the presence of NGF. Thus, ceramide did not cause death of sympathetic neurons but, instead, protected them from cell death induced by NGF withdrawal.
Ceramide Generated from Sphingomyelin Inhibits Axonal Elongation-Previously, we increased the intracellular ceramide content of sympathetic neurons by two methods as follows: (i) with threo-PPMP, an inhibitor of glucosylceramide synthase, and (ii) by incubation with C 6 -ceramide (36). Since one source of ceramide that is thought to be involved in regulating a variety of cellular processes is that generated from sphingomyelin by the action of a sphingomyelinase present at the plasma membrane, we incubated neurons with bacterial sphingomyelinase as a method of increasing the intracellular concentration of ceramide derived from sphingomyelin located at the cell surface. This approach also allowed us to investigate the possibility that the pool of ceramide generated from exogenously added sphingomyelinase (37,38,51) was distinct, in terms of its ability to regulate axonal growth, from the ceramide pools generated by treatment of the neurons with C 6ceramide or PPMP (36).
First, we established conditions under which bacterial sphingomyelinase efficiently hydrolyzed cellular sphingomyelin. Rat sympathetic neurons in 24-well plates were incubated with [methyl-3 H]choline for 3 days to radiolabel sphingomyelin and phosphatidylcholine. As shown in Table I, when the labeled neurons were treated with bacterial sphingomyelinase for 0, 2, and 4 h, the amount of radiolabeled sphingomyelin (dpm/nmol total phospholipid) decreased with increasing incubation time, whereas the amount of radiolabeled phosphatidylcholine remained essentially unchanged. These results demonstrate that the bacterial sphingomyelinase selectively hydrolyzed sphingomyelin without significantly degrading phosphatidylcholine. We next determined whether or not treatment of the neurons with bacterial sphingomyelinase inhibited axonal growth. Neurons were plated in the center compartment of 3-compartmented culture dishes and maintained for 10 days. The neurons were axotomized and then allowed to regenerate for 3 days, and distal axon extension was measured. Subsequently, some cultures were incubated for 4 h with bacterial sphingomyelinase provided to either distal axons alone or cell bodies/ proximal axons alone. The incubation was continued with medium lacking sphingomyelinase for an additional 18 h. As controls, other cultures were incubated for the entire 22 h without sphingomyelinase but in the presence or absence of C 6 -ceramide in either distal axon-containing compartments or cell body-containing compartments. Fig. 3 shows that in untreated cultures, or in cultures in which cell bodies/proximal axons had been incubated with C 6 -ceramide, the axons extended by ϳ0.8 mm during the 22-h incubation. However, incubation of distal axons with sphingomyelinase severely impaired axonal extension (Ͻ0.1 mm). The reduction in axonal growth was not due to a disruption of membrane integrity since when cell bodies/proximal axons were incubated with bacterial sphingomyelinase under the same conditions axonal growth was not inhibited (Fig. 3). As a control, and in agreement with our previous observations (36), when distal axons were treated with C 6 -ceramide, axonal extension was less than 0.1 mm (Fig.  3). The presence of C 6 -ceramide in all three compartments simultaneously, or incubation of all three compartments with bacterial sphingomyelinase, yielded the same results as when distal axons alone were incubated with C 6 -ceramide or bacterial sphingomyelinase, respectively (data not shown). We also incubated some neuron cultures with C 6 -dihydroceramide as a negative control to demonstrate the specificity of the effect of C 6 -ceramide (36); C 6 -dihydroceramide failed to inhibit axonal growth when added to either cell bodies or distal axons (data not shown). These experiments indicate that ceramide generated in distal axons by hydrolysis of cell surface sphingomyelin inhibits axonal growth. Moreover, the data support our previous conclusion (36) that an increased concentration of ceramide locally in distal axons, but not in cell bodies, inhibits axonal growth.
Anterograde Axonal Transport of Ceramide-We next investigated why an accumulation of ceramide in cell bodies did not inhibit axonal growth, whereas growth was inhibited when ceramide accumulated in distal axons. We hypothesized that the amount of ceramide reaching distal axons from cell bodies during the experiment did not result in an inhibitory concentration of ceramide being attained in distal axons. Therefore, we examined the transport of a fluorescent ceramide analog, C 6 -NBD-ceramide, between cell bodies/proximal axons and distal axons. We have shown previously (36) that C 6 -NBD-ceramide and C 6 -ceramide exert equivalent inhibitory effects on axonal growth. Compartmented cultures of 13-day-old sympathetic neurons were incubated with 10 M C 6 -NBD-ceramide for 16 h in either the distal axon-or cell body-containing compartment. Cellular material was separately harvested from the cell body-and distal axon-containing compartments, and the amount of lipid-associated fluorescence was determined. In  3. Bacterial sphingomyelinase treatment of distal axons inhibits axonal growth. Sympathetic neurons were plated in the center compartment of 3-compartment dishes, cultured for 10 days, and then axotomized and allowed to regenerate for 3 days. At that time, axon length was measured, and the neurons were treated for 4 h with bacterial sphingomyelinase (SMase) (500 milliunits/ml) in either distal axons (AX) or cell bodies (CB)/proximal axons (PAx). The sphingomyelinase-containing medium was removed, and the neurons were incubated for an additional 18 h in the absence of sphingomyelinase. Axonal extension was measured in three cultures (a total of 45-48 tracks) for each treatment. Other cultures were incubated either with no additions (Control) or with 10 M C 6 -ceramide (Cer) in distal axons or cell bodies/ proximal axons for 22 h after which axonal extension was measured. The experiment was repeated three times with similar results. Data are means Ϯ S.E. Statistically significantly differences from control cultures (p Յ 0.001) are indicated by * and were evaluated by the Student's t test.
cultures that were given C 6 -NBD-ceramide to the distal axoncontaining compartment alone, the amount of fluorescence in distal axons was 8.53 fluorescence units/mol total lipid phosphorus. In contrast, when C 6 -NBD-ceramide was similarly given to the cell body-containing compartment, the amount of fluorescence associated with distal axons was only 0.44 fluorescence units/mol total lipid phosphorus. These data demonstrate that 16 h after the fluorescent lipid had been added to the cell body-containing compartment, the concentration of C 6 -NBD-ceramide in distal axons was only 5% that resulting from the direct addition of C 6 -NBD-ceramide to distal axons. Therefore, the amount of ceramide transported from cell bodies to distal axons appears not to be sufficient to result in an inhibitory concentration of ceramide being attained in distal axons during the course of the experiment.
Distribution of Acidic and Neutral Sphingomyelinase Activities-We next investigated which class of sphingomyelinase in distal axons was likely to be responsible for generating the pool of ceramide that inhibits axonal growth. Several sphingomyelinases, with distinct pH optima, and activated by different agonists, have been implicated in sphingomyelin hydrolysis and ceramide generation (1)(2)(3)(4)(5)(6). Consequently, we investigated the distribution of neutral, magnesium-dependent sphingomyelinase activity and acidic sphingomyelinase activity in compartmented neuron cultures. Cellular material from the center compartment (cell bodies/proximal axons) and side compartments (distal axons alone) was separately harvested, and acidic and neutral sphingomyelinase activities were measured at pH 5.0 and pH 7.5, respectively. Fig. 4 shows that distal axons contain a neutral sphingomyelinase, the activity of which is negligible in the absence of magnesium (data not shown). In contrast, distal axons contain almost no acidic sphingomyelinase activity. The acidic sphingomyelinase activity, which is thought to be primarily lysosomal (7), was almost exclusively localized to cell bodies/proximal axons, in agreement with the finding that lysosomes are largely restricted to cell bodies (52)(53)(54). These data are consistent with the idea that a neutral, magnesium-dependent sphingomyelinase activity in distal axons is responsible for generating the pool of ceramide that inhibits axonal growth.
Ceramide Impairs the Uptake of NGF and LDL by Distal Axons-We next examined the possibility that the mechanism by which ceramide decreases axonal growth of sympathetic neurons is by inhibition of the uptake and/or retrograde transport of NGF by distal axons. Neurons were plated in the center compartment of 3-compartmented dishes. After 11 days, the cells were axotomized, and axons were allowed to regenerate for 4 -5 days. The incubation was continued for a further 24 h either in the absence of ceramide or in the presence of 10 M C 6 -ceramide or 10 M C 6 -dihydroceramide in distal axons. Under these conditions, C 6 -ceramide, but not C 6 -dihydroceramide, inhibits axonal growth by ϳ70% (data not shown and see Ref. 36). 125 I-NGF (50 ng/ml) was then added in the same media to distal axons and the incubation continued for an additional 18 h. The amount of 125 I present in cellular material from the distal axon-and the cell body-containing compartment, as well as in medium from the latter compartment, was determined. Fig. 5A shows that C 6 -ceramide reduced the amount of 125 I associated with distal axons by ϳ70% and that in cell bodies/ proximal axons by ϳ80% (summation of 125 I dpm/mg protein in cellular material and culture medium), whereas C 6 -dihydroceramide had no significant effect. Because radiolabeled NGF had been added to distal axons alone, any 125 I present in cell bodies/ proximal axons must have been derived from 125 I-NGF that had been taken up by distal axons and retrogradely transported into cell bodies/proximal axons; small molecules in the culture media are not transported across the barriers separating the compartments (40). Radiolabel present in medium harvested from the center compartment was derived from 125 I-NGF that had been endocytosed by distal axons, transported into cell bodies, and degraded therein (55). Since ceramide decreased the amount of 125 I-NGF present in distal axons and cell bodies/proximal axons to approximately the same extent, it appears that ceramide inhibits the uptake of NGF by distal axons without causing an additional inhibition of retrograde transport of NGF.
We next determined whether the inhibitory effect of C 6ceramide on NGF trafficking was specific to NGF or whether C 6 -ceramide also inhibited another receptor-mediated endocytic process, the uptake and retrograde transport of 125 I-LDL. LDL bind to specific LDL receptors on the cell surface and are internalized by receptor-mediated endocytosis (56). We have Compartmented cultures of rat sympathetic neurons were grown for 7 days and then axotomized, and the axons were allowed to regenerate for 4 -5 days. C 6 -ceramide (Ceramide, 10 M), C 6 -dihydroceramide (DHceramide, 10 M) or no ceramide analog was then added to distal axons for 24 h. Next, 125 I-NGF (50 ng/ml) (A) or 125 I-LDL (100 g cholesterol/ml) (B) was added to distal axons, and the neurons were incubated for 18 h in medium containing or lacking the ceramide analogs as before. Cellular material and culture medium were separately harvested from the center compartment (CB). CB radioactivity consists of 125 I in cellular material plus culture medium from the center compartment, and is a measure of 125 I-labeled protein internalized by and retrogradely transported into the center compartment from distal axons. Distal axons (Ax) were extensively washed; cellular material was collected, and radioactivity was analyzed. Values are means Ϯ S.D. of four determinations for each treatment. Statistically significant differences compared with values for neurons incubated without ceramide analogs are indicated by * and were evaluated by the Student's t test (p Յ 0.05). The experiment was repeated twice with similar results.
shown previously (49) that both distal axons and cell bodies/ proximal axons of rat sympathetic neurons express LDL receptors and internalize and retrogradely transport LDL. Fig. 5B shows that incubation of distal axons with C 6 -ceramide significantly decreased the amount of 125 I-LDL associated with both distal axons and cell bodies/proximal axons, whereas C 6 -dihydroceramide had no effect.
Thus, C 6 -ceramide interfered with the ability of axons to take up both NGF and LDL, two ligands known to be internalized by receptor-mediated endocytosis. The lack of a parallel effect by C 6 -dihydroceramide indicates that the inhibition is specific for ceramide. These experiments suggest that inhibition of axonal growth by ceramide might be due, at least in part, to a decreased endocytic uptake of NGF by distal axons.
Exposure of Distal Axons to C 6 -Ceramide Does Not Inhibit NGF-Induced Phosphorylation of TrkA-Neuronal survival and axonal growth are mediated by binding of NGF to its principal receptor, TrkA. This ligand-receptor association induces homodimerization and autophosphorylation of TrkA, leading to activation of downstream signaling pathways. Short term (5 min) exposure of rat pheochromocytoma (PC12) cells to C 2 -ceramide has been shown to inhibit NGF-induced tyrosine phosphorylation of TrkA (57). We, therefore, determined whether or not the inhibitory effect of ceramide on axonal growth could have been due to a reduction in NGF-induced TrkA phosphorylation. Neurons were plated in the center compartment of 3-compartmented dishes and incubated for 24 h either in the absence of ceramide or in the presence of C 6ceramide (10 M) or C 6 -dihydroceramide (10 M) in distal axons. The extent of TrkA phosphorylation in distal axons was investigated by immunoblotting of proteins in lysates of distal axons using anti-phosphotyrosine (Fig. 6, upper panel) and anti-TrkA (Fig. 6, middle panel) antibodies. As shown in Fig. 6 (upper panel), the amount of the phosphorylated 140-kDa protein (the same molecular mass as TrkA) was similar in vehicletreated (Ctl) and dihydroceramide-treated (diH-Cer) axons. In contrast, the amount of the phosphorylated 140-kDa protein was greatly increased in the presence of C 6 -ceramide (Cer). To confirm that the phosphorylated 140-kDa protein was TrkA, the anti-phosphotyrosine antibody was stripped from the membrane and the proteins were re-probed with anti-TrkA antibodies. The data shown in Fig. 6 (middle panel) provide evidence that the 140-kDa protein in the anti-phosphotyrosine immuno-blot is indeed TrkA. Immunoblotting of the proteins with anti-␤-tubulin antibodies (Fig. 6, bottom panel) suggested that the amount of protein loaded on to the gel from ceramide-treated neurons was marginally greater than that from the control, or dihydroceramide-treated, neurons. A similarly larger amount of immunoreactive TrkA was present in ceramide-treated, compared with control and dihydroceramide-treated, axons (Fig. 6,  middle panel). However, the markedly greater amount of phospho-TrkA detected in the ceramide-treated, compared with the control, axons could be only partially accounted for by the increased amount of TrkA in that sample. These observations demonstrate, therefore, that inhibition of axonal growth by ceramide is not the result of decreased tyrosine phosphorylation of TrkA. DISCUSSION The experiments reported herein extend our previous observations (36) showing that an increased ceramide content of distal axons, but not cell bodies, inhibits axonal elongation of rat sympathetic neurons. We now demonstrate that under conditions for which ceramide reduces axonal elongation, cell viability is maintained, indicating that ceramide independently regulates neuronal survival and axonal growth. The present studies begin to elucidate the mechanism by which axonal growth is inhibited by ceramide.
Ceramide Generated from Sphingomyelin in Distal Axons Inhibits Axonal Growth-Previously (36), we increased the ceramide content of sympathetic neurons, either by inhibiting glucosylceramide synthesis with PPMP or by incubating the neurons with the short chain, cell-permeable ceramide analog, C 6 -ceramide. Based on observations of others (37,38,51) that a diverse group of agonists can induce an accumulation of ceramide, we considered the possibility that increasing the ceramide content of neurons by different methods might generate distinct intracellular pools of ceramide, not all of which would inhibit axonal growth. Therefore, we incubated the neurons with bacterial sphingomyelinase so that sphingomyelin located in the outer leaflet of the plasma membrane (58,59) would be hydrolyzed to ceramide. Our data show that each of the three methods by which we increased the ceramide content of distal axons, including treatment with exogenous sphingomyelinase, inhibits axonal growth.
As an indication of whether a metabolite of ceramide, rather than ceramide per se, was responsible for the inhibition of axonal growth, we examined previously (36) the metabolites of C 6 -NBD-ceramide in cultured sympathetic neurons and found that after 24 h of incubation with C 6 -NBD-ceramide, 90.2% of the cell-associated fluorescence was recovered in ceramide. Of the remaining fluorescence, 2.3% was in ceramide phosphate, 2.0% in sphingomyelin, and small amounts were distributed among glucosylceramide, other glycosphingolipids, and hexanoate derived from the hydrolysis of ceramide. Thus, after 24 h, greater than 90% of the C 6 -NBD-ceramide had not been metabolized. Although these data do not exclude the possibility that one of these quantitatively minor metabolites of ceramide, rather than ceramide per se, is responsible for inhibition of axonal growth and protection against apoptosis, the most likely conclusion is that ceramide is the active molecule.
Distal Axons Contain a Neutral, but Not an Acidic, Sphingomyelinase Activity-Ceramide is generated when sphingomyelin is hydrolyzed in response to extracellular agonists (20,29). Several distinct sphingomyelinases have been implicated in the production of ceramide for intracellular signaling events (8). Our experiments in sympathetic neurons show that distal axons contain a magnesium-dependent, neutral sphingomyelinase activity but that acidic sphingomyelinase activity is almost undetectable in distal axons (Fig. 4). In contrast, the FIG. 6. C 6 -ceramide does not decrease TrkA phosphorylation in distal axons. Distal axons were incubated without a ceramide analog (Ctl), with 10 M C 6 -ceramide (Cer), or with 10 M C 6 -dihydroceramide (diH-Cer) for 18 h in medium containing 100 ng/ml NGF, followed by 6 h in medium lacking NGF with or without the same ceramide analogs. Next, NGF was added to the distal axon-containing compartments (final concentration 100 ng/ml) for 15 min to induce TrkA phosphorylation. Cell lysates from distal axons (4 dishes combined) were collected, and proteins were separated by polyacrylamide gel electrophoresis. The phosphorylation state of TrkA was assessed by immunoblotting with anti-phosphotyrosine antibodies (upper panel). Antibodies were stripped from the membrane and the proteins were re-probed with anti-TrkA antibodies to confirm that the 140-kDa protein was TrkA (middle panel). As a loading control, the same membrane was re-probed with anti-␤-tubulin antibodies (lower panel). Results are representative of two experiments performed under similar conditions. acidic sphingomyelinase activity is localized to the compartment containing cell bodies/proximal axons indicating that the acid sphingomyelinase is localized to cell bodies. These data suggest that the pool of ceramide that inhibits axonal growth is generated by a neutral sphingomyelinase activity located in distal axons.
The identity of the physiological agonist responsible for activation of the sphingomyelinase in distal axons is not known. Since neurotrophins can activate sphingomyelinase (20,24,29), these agents are strong candidates as inducers of the generation of ceramide that inhibits axonal growth. Neurotrophins bind to two types of cell surface receptors: Trk, a tyrosine kinase receptor, and p75 NTR . Binding of NGF to Trk initiates signaling cascades that promote axonal survival and axonal growth. All neurotrophins bind to p75 NTR which is widely distributed throughout the nervous system. Ceramide has been implicated as a mediator of some of the biological actions of p75 NTR . The involvement of p75 NTR in ceramide-mediated processes in sympathetic neurons is complicated, however, since these neurons express both p75 NTR and TrkA, and activation of TrkA by NGF inhibits the p75 NTR -dependent hydrolysis of sphingomyelin (20,23,24). We were unable to test directly a role for NGF in stimulating sphingomyelinase activity in sympathetic neurons since these cells require NGF for survival. However, it is conceivable that when a neurotrophin other than NGF binds to p75 NTR , sphingomyelin hydrolysis would be stimulated, the ceramide content increased, and axonal growth inhibited.
In addition to the pool of ceramide that originates from sphingomyelin hydrolysis, ceramide that accumulates upon inhibition of glucosylceramide synthase also inhibits axonal growth (36). Therefore, a scenario in which an agonist activates ceramide synthase (60) or inhibits ceramidase (34) must also be considered.
Ceramide Inhibits NGF Uptake but Not TrkA Phosphorylation-In Chinese hamster ovary cells, ceramide inhibits vesicular transport processes, including the endocytosis of LDL (61,62). We now show that the amount of NGF and LDL taken up by distal axons is markedly reduced when the ceramide content of distal axons increases, supporting the idea that ceramide interferes with the endocytic uptake of NGF. Our data demonstrate that when 125 I-NGF (or 125 I-LDL) is added to distal axons, C 6 -ceramide reduces to a similar extent the amount of 125 I-NGF (or 125 I-LDL) associated with distal axons and cell bodies/proximal axons. These findings are consistent with the idea that ceramide inhibits the endocytic uptake of NGF, rather than the retrograde transport of NGF from distal axons to cell bodies.
Since most biological functions of NGF are mediated by NGF binding to, and subsequently activating, the tyrosine kinase receptor TrkA, we asked whether or not inhibition of axonal growth by ceramide could be attributed to a reduced tyrosine phosphorylation of TrkA. This scenario would not have been unprecedented, since short term ceramide treatment has been shown previously (57) to block NGF-induced TrkA phosphorylation in PC12 cells. However, under conditions for which ceramide inhibited axonal growth, TrkA phosphorylation was not decreased. In contrast, 24 h of exposure of distal axons to C 6 -ceramide significantly increased TrkA phosphorylation. The mechanism by which ceramide augments TrkA phosphorylation is not clear. However, similar observations were made in PC12 cells treated with C 2 -ceramide for 8 h in both the presence and absence of NGF (63). Thus, the observed inhibition of axonal growth by ceramide is clearly not due to a reduction in TrkA phosphorylation.
The mechanism by which ceramide inhibits endocytosis (61, 62) has not been established. However, even at very low, perhaps physiological, concentrations (1-10 mol % of membrane lipids), ceramide can induce profound alterations in physical properties of membranes (64 -66) which might explain why a membrane-related process, such as receptor-mediated endocytosis, would be disrupted by ceramide.
In conclusion, we report that in rat sympathetic neurons the production of ceramide locally in distal axons upon treatment of distal axons with bacterial sphingomyelinase inhibits axonal elongation. The finding that distal axons contain a neutral, but not an acidic, sphingomyelinase activity is consistent with the idea that a neutral sphingomyelinase in distal axons generates the pool of ceramide that inhibits axonal growth. Our data also indicate that the impairment of axonal growth by ceramide is not mediated by a decreased tyrosine phosphorylation of TrkA but might be due, at least in part, to ceramide inhibiting the endocytic uptake of NGF.