Nerve Growth Factor-induced p75-mediated Death of Cultured Hippocampal Neurons Is Age-dependent and Transduced through Ceramide Generated by Neutral Sphingomyelinase*

Binding of nerve growth factor (NGF) to the p75 neurotrophin receptor (p75) in cultured hippocampal neurons has been reported to cause seemingly contrasting effects, namely ceramide-dependent axonal outgrowth of freshly plated neurons, versus Jun kinase (Jnk)-dependent cell death in older neurons. We now show that the apoptotic effects of NGF in hippocampal neurons are observed only from the 2nd day of culture onward. This switch in the effect of NGF is correlated with an increase in p75 expression levels and increasing levels of ceramide generation as the cultures mature. NGF application to neuronal cultures from p75exonIII−/− mice had no effect on ceramide levels and did not affect neuronal viability. The neutral sphingomyelinase inhibitor, scyphostatin, inhibited NGF-induced ceramide generation and neuronal death, whereas hippocampal neurons cultured from acid sphingomyelinase−/− mice were as susceptible to NGF-induced death as wild type neurons. The acid ceramidase inhibitor, (1S,2R)-d-erythro-2-(N-myristoylamino)-1-phenyl-1-propanol, enhanced cell death, supporting a role for ceramide itself and not a downstream lipid metabolite. Finally, scyphostatin inhibited NGF-induced Jnk phosphorylation in hippocampal neurons. These data indicate an initiating role of ceramide generated by neutral sphingomyelinase in the diverse neuronal responses induced by binding of neurotrophins to p75.

The p75 neurotrophin receptor (p75) 1 is the shared receptor for all four mammalian neurotrophins (1) as well as other unrelated ligands (2)(3)(4)(5)(6). It is expressed in a wide range of neuronal and non-neuronal cells (7,8), with a corresponding diversity of roles attributed throughout development and in the adult (9 -11). In addition to enhancing responsiveness of cells co-expressing p75 and Trk receptors (12), p75 has in recent years been established as a signaling receptor in its own right (13,14). Independent signaling of p75 has been reported to modulate many aspects of neuronal physiology including sensory functions (15), axon outgrowth (16 -18), and survival or apoptotic effects of neurotrophins (19 -24). Typically, the diverse effects observed after p75 activation are explained by "cell context," which may involve the differential activation of a number of intracellular signaling pathways, including NFB translocation (25), Jun kinase phosphorylation (26), and ceramide generation (27).
Ceramide is a lipid second messenger implicated in diverse intracellular pathways, most prominently those regulating cell death in assorted cell types (28 -30). A plethora of studies have looked at the effects of exogenously added ceramide analogues on cultured cells, and in cultured neurons both outgrowth and survival/death effects have been reported (31-33). However, less attention has been paid to endogenous ceramide generation in neurons. Endogenous ceramide can be generated by hydrolysis of sphingomyelin (SM) or by de novo synthesis, processes that occur at different intracellular locations, and by different modes of regulation (29). Ceramide generated by SM hydrolysis can be produced by either neutral or acid sphingomyelinases (N-SMase or A-SMase, respectively) (30), which may be differentially distributed between the cell body and the axon (34).
A prominent example of an endogenous signaling system that generates ceramide in the nervous system is p75 (27,35). Although an early report (26) suggested a role for ceramide in p75-mediated cell death of oligodendrocytes, subsequent studies on neurotrophin-induced death of neurons have focused primarily on the role of a pathway involving Jun kinase, mixed lineage kinases, and p53 (13,20,21,36,37). Moreover, two recent studies on hippocampal neurons documented strikingly different effects for NGF in the same cell type and focused on two different second messengers downstream of p75 (17,21). In our study, we observed a ceramide-dependent enhancement of axonal elongation in cultured hippocampal neurons by NGF (17), whereas another study observed Jnk-dependent cell death of hippocampal neurons (21), albeit cultured under different conditions.
To resolve this apparent contradiction, we have now performed a systematic study to rigorously dissect the effects of NGF at different stages of neuronal development in culture and the role of ceramide signaling in these processes. We show that in addition to its ability to stimulate axonal outgrowth (17), ceramide generated via binding of NGF to p75 can induce cell death in hippocampal neurons only from the 2nd day of culture onward, which parallels an increase in p75 expression levels.
We further demonstrate a requirement for N-SMase activity in both neuronal cell death and Jnk phosphorylation and the lack of requirement for an A-SMase activity in these processes. Thus, in the same neuron, NGF signaling via p75 can play two different roles, both of which require ceramide as an upstream signaling component.
Animals-Rats (Wistar) and mice (C57BL6) were purchased from Harlan Animal Laboratories and were maintained in the animal facility of the Weizmann Institute. P75 exonIIIϪ/Ϫ mice (38) on a C57BL6 background were obtained from Jackson Laboratories, and a breeding colony was maintained in the animal facility of the Weizmann Institute. A-SMase Ϫ/Ϫ mice (39) were kindly provided by Dr. R. Kolesnick (Sloan-Kettering, New York), and a heterozygous breeding colony was maintained.
Hippocampal Cultures-Rat hippocampal neurons were cultured at low density as described previously (17,31,40), with modifications to allow culturing in defined medium. Briefly, the dissected hippocampi of embryonic day 18 rats were dissociated by trypsinization (0.25% w/v, for 15 min at 37°C). The tissue was washed in Mg 2ϩ /Ca 2ϩ -free Hanks' balanced salt solution (Invitrogen) and dissociated by repeated passage through a constricted Pasteur pipette. Cells were plated in minimal essential medium with 10% horse serum at a density of 25,000 cells per 13-mm glass coverslip that had been precoated with poly-L-lysine (1 mg/ml). After 2-4 h, coverslips were transferred into 24-well multidishes (Nunc), containing B27 supplemented Neurobasal medium (41), and cultures were maintained in this defined medium.
For experiments using scyphostatin, neurons were transferred into 24-well multidishes (Nunc) that contained a monolayer of astroglia, as described previously (31, 40). In this case, coverslips were placed with the neurons facing downwards, separated from the glia by paraffin "feet," and maintained in serum-free medium (minimal essential medium) at a density of 25,000 neurons per 13-mm coverslip, which included N2 supplements (40), ovalbumin (0.1%, w/v), and pyruvate (0.1 mM).
Neurons cultured at high density (230,000 cells per 24-mm glass coverslip in 100-mm Petri dishes), also maintained in Neurobasal medium, were used for biochemical analyses. Neurons were also cultured at low (25,000 neurons per 13-mm coverslip) and high density from embryonic day 17 p75 exonIIIϪ/Ϫ mice (38) and from A-SMase Ϫ/Ϫ mice (39). Mouse neurons were cultured exactly as described for rat neurons and were cultured both in the absence or presence of a glial co-culture. For A-SMase Ϫ/Ϫ mice, heterozygous males and females were mated, and the genotype of individual embryos was determined by both RT-PCR (39) and by analysis of A-SMase enzyme activity (17).
Analysis of Neuronal Cell Death-Live and dead cells were distinguished using 2 M calcein acetoxymethyl ester and 4 M ethidium homodimer-1, respectively, as detailed in the Live/Dead viability/cytotoxicity kit (Molecular Probes, OR). At least 300 -400 cells were counted per coverslip. Neurons were examined using a Plan 25ϫ objective of a Zeiss Axiovert 35 microscope. Apoptosis was measured using Hoechst 33342, and cells were examined using the 60ϫ objective of a Nikon Eclipse TE 300 microscope.
Ceramide Formation-Neurons were plated at high density and incubated with C6-NBD-SM (dissolved in ethanol). After various times of incubation, cells were removed from the coverslips by scraping with a rubber policeman into ice-cold distilled water and were lyophilized. C6-NBD lipids were extracted and analyzed as described (17). Two methods were used to quantify NBD fluorescence. In some cases, fluorescent spots were removed from the coverslips by scraping, lipids extracted, and fluorescence analyzed using a PerkinElmer Life Sciences LS-5B luminescence spectrometer. Alternatively, fluorescence was quantified using a Fluor-S TM -MultiImager (Bio-Rad), using Quantity One software for scanning and quantification. When significant amounts of C6-NBD-glucosylceramide were detected on the TLC plates, a correction was made to calculate total C6-NBD-ceramide formed (17).
RT-PCR-Reverse transcriptase-PCR was carried out using the Titan one-tube system from Roche Molecular Biochemicals, as per manufacturer's instructions. Band intensity was quantified using a Bio-Rad model GS-690 Imaging Densitometer and analyzed using Molecular Analyst software, version 2.1.2. Expression of p75, TrkA, and glyceraldehyde 3-phosphate dehydrogenase (GAPDH) levels in hippocampal neurons was analyzed using the following primers: for p75, GTCGTGGGCCTTGTGGCC and CTGTGAGTTCACACTGGGG; for TrkA, CGTTGATGCTGGCTTGTGC and GGAGAGATTCAGGTGACT-GA; and for glyceraldehyde-3-phosphate dehydrogenase (GAPDH) TTA-GCACCCCTGGCCAAGG and CTTACTCCTTGGAGGCCATG. Primers for genotyping of A-SMase Ϫ/Ϫ embryos were as follows: for neomycin, CTTGGGTGGAGAGGCTATTC and AGGTGAGATGACAGGAGATC); and for ASMase AGCCGTGTCCTCTTCCTTAC and CGAGACTGTTG-CCAGACATC.
Western Blotting-Neuronal proteins were extracted from 20 high density coverslips per culture in an ice-cold buffer composed of 50 mM ␤-glycerophosphate, 1.5 mM EGTA, 1.0 mM EDTA, 1 mM dithiothreitol, 1% Igepal detergent, 0.1 mM sodium orthovanadate, and proteinase inhibitors. Equal loading on SDS-PAGE gels was verified by cell number and parallel Western blots for total Jnk and ␣-tubulin. Western blots for p75 were performed using the 9651 polyclonal antiserum as previously described (17). Phospho-jnk was identified using the antidiphospho-Thr-183/Tyr-185-Jnk antibody from New England Biolabs according to the manufacturer's instructions. (17) that application of NGF to hippocampal neurons, immediately after plating, stimulated axonal outgrowth through activation of N-SMase and ceramide generation. However, another recent study (21) demonstrated that neurotrophins, including NGF, can induce apoptosis in 5-day-old cultured hippocampal neurons. We therefore examined the effects of increasing NGF concentrations on neuronal viability at different stages of culture. No NGF-induced cell death was observed in neurons exposed to NGF at concentrations up to 500 ng/ml immediately after plating (Fig. 1A). In contrast, increasing concentrations of NGF did affect neuronal viability when added to the neuronal cultures on day 1 or on day 4, with 50 -60% neuronal cell death at the highest NGF concentrations (Fig. 1A). We confirmed that this cell death was due to apoptosis by using Hoechst 33342, which specifically labels apoptotic nuclei (Fig. 1B). A 1.6-fold increase in nuclei displaying chromatin condensation was detected in neurons treated with NGF on day 1 in culture.

NGF-induced Ceramide Generation and Neuronal Cell Death Increase with Time in Culture and Are Correlated with Increased p75 Expression-We demonstrated recently
We demonstrated previously (17) that during the first 24 h in culture, hippocampal neurons express low levels of p75. To determine whether the increased ability of NGF to kill hippocampal neurons on day 1 versus day 0 in culture could be due to changes in neurotrophin receptor expression, p75 and TrkA levels were analyzed by RT-PCR on days 0, 1, and 4 in culture. P75 transcript levels increased ϳ4-fold between days 0 and 1 and remained constant for the next 3 days (Fig. 2A). A TrkA transcript was not detected at any of these stages in culture. Increased levels of p75 protein were also detected by Western blotting (Fig. 2B), with an accumulative increase of ϳ2-fold from day 0 to day 4. We next measured ceramide formation upon binding of NGF to the p75 receptor. There was an essentially linear correlation between NGF concentration and ceramide formation when analyzed on 1-day-old neurons (Fig. 3A). Moreover, ϳ70% more ceramide was generated by application of 100 ng/ml NGF to day 1 neurons as compared with neurons immediately after plating, and ϳ125% more ceramide on day 4 neurons (Fig. 3B). Note that the amount of ceramide generated after NGF application on day 0 is not significantly higher than the basal levels in control cells on day 1 (Fig. 3B).

p75-induced Neuronal Cell Death via Neutral Sphingomyelinase
NGF Does Not Kill Neurons or Elevate Ceramide Levels in p75 exonIIIϪ/Ϫ Mice-To prove definitively a connection between binding of NGF to p75, ceramide generation, and neuronal cell death, we analyzed the effects of NGF on neurons cultured from p75 exonIIIϪ/Ϫ mice (38). RT-PCR analysis confirmed that hippocampal neurons cultured from p75 exonIIIϪ/Ϫ mice contained no full-length p75 and also no TrkA during the entire period of culture. Application of 500 ng/ml NGF had no effect on neuronal viability in p75 exonIIIϪ/Ϫ mice, whereas wild type mouse hippocampal neurons were killed (Fig. 4A). In contrast, C6-ceramide (5 M) killed both wild type and p75 exonIIIϪ/Ϫ neurons (Fig. 4A). Moreover, ceramide levels increased by ϳ2fold upon treatment of neurons from wild type mice with 500 ng/ml NGF, whereas there was no increase in ceramide levels in p75 exonIIIϪ/Ϫ mice (Fig. 4B). Thus, p75 is required for ceramide generation upon NGF application to hippocampal neurons.
Ceramide Generated via N-SMase Activation Is Required for Neuronal Cell Death-We next examined the mechanism of ceramide generation upon binding of NGF to p75. To distinguish between activation of N-SMase or A-SMase, neurons were incubated with scyphostatin (42, 43), which has an IC 50 for N-SMase 50-fold lower than for A-SMase (44). Moreover, scyphostatin had no effect on human A-SMase activity when measured in vitro at either pH 4.7 or pH 6.4, even up to concentrations as high as 50 M (not shown), at which complete inhibition of N-SMase occurs (17).
Due to a marked toxicity of scyphostatin to neurons cultured in Neurobasal medium without glia, experiments with this inhibitor were carried out at concentrations of up to 1 M on hippocampal neurons co-cultured with glia. Under these conditions scyphostatin was not toxic and inhibited ceramide generation upon NGF application (Fig. 5B). Moreover, preincubation with scyphostatin inhibited NGF-induced neuronal cell death (Fig. 5A), supporting the involvement of an N-SMase in this process.
We next examined the possible role of A-SMase in NGFinduced death of hippocampal neurons using A-SMase Ϫ/Ϫ mice (39). A-SMase Ϫ/Ϫ hippocampal neurons were equally susceptible to NGF-induced death as neurons from their wild type littermates (Fig. 5, C and D). Neurons were also tested for viability in the presence of glutamate (45), and as described previously (46), A-SMase Ϫ/Ϫ neurons were partially resistant to glutamate-induced death (data not shown). Thus, in contrast to the excitotoxic stress-activated pathway of ceramide generation, which is mediated via A-SMase (46), NGF-induced death of hippocampal neurons is mediated by N-SMase.
Ceramide Itself, and Not a Downstream Lipid Metabolite, Signals NGF-induced Neuronal Death-Ceramide can be degraded by ceramidases to downstream metabolites such as sphingosine 1-phosphate, which has been shown to act as an extracellular initiator of apoptosis (47, 48). We therefore used the ceramidase inhibitor, D-e-MAPP (49), to determine whether ceramide itself or a downstream metabolite(s) is responsible for neuronal death. In the absence of NGF, increasing concentrations of D-e-MAPP resulted in an increase in both ceramide levels and in neuronal cell death, confirming that elevating ceramide elicits neuronal cell death (Fig. 6A). Importantly, D-e-MAPP enhanced the ability of NGF to kill neurons (Fig.  6A). Finally, the combination of NGF and D-e-MAPP generated additively higher levels of ceramide than either NGF or D-e-MAPP by themselves (Fig. 6B). Taken together, these results support the involvement of ceramide itself, and not a down-   (20,21) have implicated Jun kinase activation in neuronal cell death induced by neurotrophin binding to p75. To determine whether ceramide generation and Jun kinase phosphorylation are part of the same cascade or represent distinct signaling pathways, we analyzed Jnk activation in hippocampal neurons upon NGF application. Hippocampal neurons from 4-day-old cultures were incubated with 500 ng/ml NGF for different times, revealing a 2-fold increase in phospho-Jnk levels from 90 min onward (Fig. 7A). Preincubation with scyphostatin before NGF application inhibited the NGF-induced increase in phospho-Jnk (Fig. 7B). Thus ceramide generation by N-SMase is required for NGF-induced Jnk phosphorylation in hippocampal neurons. DISCUSSION In our current and previous studies, we demonstrate that cultured hippocampal neurons undergo a switch in their response to NGF after the first 24 h in culture, from acceleration of the rate of axonal outgrowth (17) to apoptotic cell death (this study). Both responses depend on the p75 neurotrophin receptor (Figs. 2 and 4 and Ref. 17), and cell death is correlated with increased expression of p75. Intriguingly, both effects also re-quire ceramide generation via N-SMase (Figs. 3 and 5 and Ref. 17). Ceramide itself (and not a downstream lipid metabolite) is required for the death signal (Fig. 6), and finally, ceramide generation is upstream of Jun kinase phosphorylation (Fig. 7) in the signaling cascade initiated by NGF binding to p75 in hippocampal neurons.
Correlating p75 Levels, Ceramide Generation, and the Extent of Neuronal Cell Death-When p75 is expressed at low levels, namely during the initial 24 h after plating neurons, ceramide generation is sufficient to enhance axonal outgrowth (17) but not neuronal death. In contrast, when p75 levels increase, i.e. after 24 -48 h in culture, ceramide generation increases significantly and is sufficient to induce neuronal death. We demonstrated previously that exogenously added ceramide can have dual effects in hippocampal neurons, either stimulating neuronal growth or inducing death (31), and we now demonstrate that endogenous ceramide levels can be regulated by changes in the level of the p75 receptor, which presumably activates N-SMase, although the mechanism of activation/interaction is not known (see below). Interestingly, a recent report (50) demonstrated that p75-mediated Akt activation occurs at p75 levels much lower than those required to elicit apoptosis in the same cells. Changes in expression of an intermediary linking ceramide to the Jnk pathway could also account for the strikingly different response of day 0 versus older neurons. Alternatively, the differential response might be attributed solely to the in- Another possibility is that ceramide generation, probably at the higher levels required to induce apoptosis, modulates the membrane environment, perhaps due to its propensity to laterally segregate within the plane of the lipid bilayer and due to its smaller packing volume than SM (29). Indeed, exogenously added (51, 52) and endogenously generated ceramide (53) affect endocytosis and vesiculation in model membranes and cells, and chronically applied ceramide inhibits NGF internalization in sympathetic neurons (34). Because ceramide generation occurs in a highly compartmentalized manner (29, 54), increases as low as 2-fold (such as that observed between days 1 and 2, Fig. 3) could represent a much higher localized increase in the membrane compartments containing p75. Whether the rate of p75 internalization is affected by ceramide is not known; however, it is conceivable that such an increase might affect p75 internalization or retrograde trafficking. In this context it is interesting to note that blocking internalization of the p55 tumor necrosis factor receptor selectively inhibits apoptotic signaling (55), while not affecting other p55-dependent pathways.
N-SMase Versus A-SMase in Apoptotic Signaling in Neurons-Many recent studies (56) examining the role of endogenous ceramide in apoptotic cell death have focused on ceramide generated from A-SMase, most notably for members of the NGF/tumor necrosis factor receptor family. Our current findings implicating a neuronal N-SMase in apoptotic signaling of p75 suggest a difference between p75 signaling and that of other NGF/tumor necrosis factor receptor family members. Indeed the receptor domain and interactors linking the p55 tumor necrosis factor receptor to N-SMase have been shown to be distinct from the domain and the interactors that signal apoptosis (57). However, some evidence has been presented for the involvement of an A-SMase in p75 signaling in PC12 cells (56,58,59), in which p75 signaling was localized to caveolar domains, and was regulated by phosphoinositide 3-kinase in the same structures (56). These findings may reflect differences between p75 signaling in primary neurons versus cycling cells such as PC12. Because N-SMases are less well characterized than A-SMases, the mode and mechanism of interaction between p75 and N-SMase cannot be determined until a genuine N-SMase involved in ceramide signaling pathways is identified and sequenced (43, 60,61).
Cross-talk between Ceramide and Other Signaling Effectors Downstream of p75-It was demonstrated recently that all four neurotrophins elicit the death of hippocampal neurons by binding to p75 (21) and that this effect is mediated via Jun kinase. Interestingly, dual roles have been suggested for Jun kinase, in development and stress responses, with different Jun kinase pools serving different functions (62); likewise, different roles have been suggested for Jun kinase in ceramide signaling (63). Our data show that ceramide generation is necessary for both NGF-induced neuronal cell death and Jnk activation, suggest- ing that ceramide somehow regulates or modulates one of the interactors or kinases upstream of Jnk. The current state of knowledge regarding downstream targets of ceramide (64), or activation pathways to Jnk (65), does not allow the delineation of a precise mechanism by which ceramide might activate Jnk. An intriguing option might involve modulation of the accessibility of p75 and interactant proteins to Jnk via scaffolding proteins such as the Jun-interacting proteins (66). Mediators that could provide an initiating link from p75 to Jnk have not yet been determined, although a growing list of p75 interactants has emerged in recent years (reviewed in Ref. 13, see also Refs. 67 and 68). It will obviously be of interest to examine the effects of manipulating ceramide levels on the interaction of these diverse proteins with p75 and on the downstream cascades thus activated.
Finally, it should be noted that an understanding of the in vivo physiological significance of the NGF-induced effects described herein awaits further research. Very recent publications from the Dechant group (69) on the more drastic phenotype of the p75 exonIVϪ/Ϫ mouse and from Hempstead and colleagues (70) on high affinity interactions of pro-neurotrophins with p75 will contribute to the design of future experiments on the in vivo role of the p75-N-SMase pathway in neuronal apoptosis. 22