INTRODUCTION

The neurotrophins are a well conserved family of proteins that play critical roles in the maintenance and development of the nervous system (1-7). Their cellular effects are mediated by two distinct classes of cell surface receptors. Highly related transmembrane receptor tyrosine kinases, termed Trk receptors, recognize the neurotrophins with a relatively high degree of binding specificity, with TrkA preferentially binding NGF, TrkB preferring BDNF¹ and neurotrophin-4/5, and TrkC interacting only with neurotrophin-3 (reviewed in Refs. 8 and 9).

The other class of neurotrophin receptor contains only one member, termed p75NTR. This receptor is a member of TNF receptor superfamily that also includes CD27, CD30, CD40, 4-1BB, OX40, the Fas antigen, and the tumor necrosis factor receptors TNFR1 and TNFR2 (reviewed in Ref. 8). The physiological role of the p75NTR has proven elusive, but recent data support at least two functions. First, p75NTR functionally interacts with TrkA to enhance neurotrophin responsiveness, particularly at low ligand concentrations (9-11). p75NTR appears to mediate this effect in part by increasing the amount of NGF that becomes bound to the TrkA receptor (12, 13). Second, recent studies suggest that the p75NTR may play an autonomous signaling role. Neurotrophin binding to p75NTR results in sphingomyelinase activation and ceramide production (14, 15), and recent reports suggest that NGF binding to p75 produces activation of the NF-kB transcriptional complex (16) as well as a stimulation of Jun kinase (17). These results indicate that p75NTR may be functionally grouped with other members of the TNF receptor superfamily, such as TNFR1 and Fas.

Previous analyses of p75NTR-TrkA interactions have focused on ligand binding events that might account for modulation of receptor function. In this report, we tested the hypothesis that a p75NTR-activated signaling mechanism alters TrkA function. Using receptor-selective ligands, we found that BDNF binding to p75NTR results in a strong attenuation of TrkA signaling. TrkA signaling was also reduced following a short exposure to ceramide, suggesting that ceramide production which results from p75NTR activation may be responsible for this effect. To test if the reduction in TrkA signaling correlated with serine or threonine phosphorylation of the TrkA intracellular domain, TrkA was subjected to phosphoamino acid analysis; both BDNF and ceramide treatment were found to increase phosphoserine content. These findings indicate that ligand binding to p75NTR activates a ceramide-dependent signaling cascade that regulates TrkA activity through serine phosphorylation of the TrkA intracellular kinase domain.

EXPERIMENTAL PROCEDURES

NGF was purchased from Collaborative Research, TNF was purchased from R & D Systems, and C₂-ceramide and dihydrosphingosine were obtained from BioMol. BDNF was provided by Regeneron Pharmaceuticals (Tarrytown, NY), the pan-Trk antibody 203 was a gift from David Kaplan (Montreal Neurological Institute), and NGF3T was prepared as described (12). All other reagents were purchased from either Sigma or ICN.

PC12 cells (18) and PC12-6/24 cells (19) were maintained in Dulbecco's modified Eagle's medium containing 6% horse serum and 6% bovine calf serum in 7% CO₂ at 37 °C. MG139-2 cells (20), a subline of 3T3 fibroblasts stably expressing rat TrkA (but not p75NTR), were maintained in Dulbecco's modified Eagle's medium containing 10% bovine calf serum and 200 µg/ml G418 in 7% CO₂ at 37 °C.

PC12 cells or MG139-2 cells were plated on 100-mm dishes and 24 h later were washed twice in Dulbecco's modified Eagle's medium containing 0.1% bovine serum albumin (DMEB) and then incubated in 5 ml of DMEB supplemented as described in the figure legends. Control incubations received appropriate vehicle control solutions. After induction, the medium was removed, and plates were placed on ice, rinsed with ice-cold Tris-buffered saline (20 mM Tris (pH 8.0), 137 mM NaCl), and then lysed with 1 ml of lysis buffer (20 mM Tris (pH 8.0), 137 mM NaCl, 1% Nonidet P-40, 10% glycerol, 1 µg/ml leupeptin, 100 µM phenylmethylsulfonyl fluoride, and 1 mM orthovanadate). Lysates were normalized for protein content, incubated with 2 µl of rabbit pan-Trk antibody 203 overnight at 4 °C, supplemented with 30 µl of protein A-Sepharose, and then incubated at 4 °C for 90 min. Beads were pelleted, washed three times in lysis buffer, resuspended in Laemmli sample buffer, and boiled for 5 min. Immunoprecipitates were separated on 10% Laemmli acrylamide gels and transferred to nitrocellulose. Phosphotyrosine blots were blocked in 10 mM Tris (pH 7.4), 150 mM NaCl, 2% bovine serum albumin, 0.2% Tween 20, and primary and secondary antibody incubations were performed in 10 mM Tris (pH 7.4), 150 mM NaCl, and 0.2% Tween 20. Blocking, primary (using antibody 203), and secondary (using protein A-conjugated horseradish peroxidase) incubations for TrkA immunoblots were all performed in 10 mM Tris (pH 7.4), 150 mM NaCl, 0.2% Tween 20, and 5% (w/v) dry skim milk powder. Reactive bands on all immunoblots were detected using enhanced chemiluminescence (ECL) according to the manufacturer's instructions (DuPont).

PC12 cells were grown to confluence in 150-mm plates, washed once in serum-free medium, and then preincubated 2 h in phosphate-free DMEB containing 750 µCi of [³²P]orthophosphate/ml. Cells were then supplemented with 11 × stock solutions made in phosphate-free DMEB to produce final NGF, BDNF, or C₂-ceramide concentrations of 100 ng/ml, 200 ng/ml, and 100 µM, respectively. After a 30-min incubation, the medium was removed, and cells were washed once with Tris-buffered saline (pH 7.4). Wash buffer was replaced with lysis buffer (10 mM Tris (pH 7.4), 10% glycerol, 150 mM NaCl, 1 mM EGTA, 1% Nonidet P-40, 1 mM phenylmethylsulfonyl fluoride, 10 µg/ml aprotinin, 10 µg/ml leupeptin, and 1 mM orthovanadate). Lysates were centrifuged at 16,000 × g, and the supernatant was precleared with preimmune serum-coated protein A-Sepharose beads for 1 h at 4° C. The beads were removed by centrifugation, and TrkA was immunoprecipitated as described above. Immunoprecipitates were split into two aliquots that were both separated on SDS-polyacrylamide gel electrophoresis and transferred to polyvinylidene difluoride membrane. One set of labeled lysates was analyzed for TrkA content by immunoblot as described above, and the other was exposed to XAR film to detect labeled TrkA protein. Polyvinylidene difluoride membrane fragments containing the labeled TrkA samples were cut from the transfer membrane and hydrolyzed under vacuum in 6 N HCl vapor for 2.5 h at 110° C. Phosphoamino acids were eluted from the membrane with H₂O for 16 h then were concentrated in a Speed-Vac. Phosphoamino acids were separated using two-dimensional thin layer electrophoresis on a Hunter apparatus. Phosphoamino acid standards were visualized by spraying the plates with a solution of 0.2% ninhydrin in acetone (w/v), and ³²P-labeled phosphoamino acids were identified by autoradiography. Quantitation of phosphoamino analysis was performed on scanned autroradiographs using NIH IMAGE.

RESULTS

Signaling pathways originating from p75NTR and TrkA can be activated in PC12 cells (15, 19). We have previously shown that BDNF treatment of PC12 cells, which binds p75NTR but not TrkA, results in a dramatic reduction in TrkA receptor activation that is at least partially due to its ability to disrupt ligand transfer between p75NTR and TrkA (12). To test whether other mechanisms might also contribute to the effect of BDNF on TrkA activation, we have used ligands that selectively bind either TrkA or p75NTR. A mutant form of NGF (NGF3T) that does not bind p75NTR (21) was used to selectively activate TrkA and BDNF was used to selectively activate p75NTR. To determine if a p75NTR-derived signal may affect TrkA activation, PC12 cells were preincubated with various concentrations of BDNF for 3 h, exposed to NGF3T for 5 min, and then analyzed for TrkA tyrosine phosphorylation levels by immunoblotting (Fig. 1A). A clear reduction in NGF3T-stimulated TrkA tyrosine phosphorylation levels was observed in the presence of BDNF; densitometry reveals reductions of 53, 70, and 81% at BDNF concentrations of 50, 100, and 500 ng/ml, respectively. Direct TrkA immunoblotting showed that this decrease in phosphotyrosine signal did not reflect a reduction in TrkA protein levels because these were maintained throughout all treatment conditions (Fig. 1B). This effect of BDNF was rapid; a BDNF preincubation lasting 5 min (Fig. 1C) produced effects on TrkA tyrosine phosphorylation similar to that observed using a 3-h preincubation (Fig. 1A).

To test if the BDNF-mediated reduction in TrkA tyrosine phosphorylation is reflected in downstream signaling events, NGF3T-stimulated c-fos mRNA accumulation was examined in the presence and the absence of BDNF. Fig. 2A shows that the robust increase in c-fos mRNA expression, which normally occurs in response to NGF3T, was reduced (by 71%) following BDNF treatment. We next considered the possibility that the effect of BDNF on TrkA activation and signaling could be due some direct interference with NGF3T binding. We have previously shown that BDNF does not directly interact with TrkA and does not interfere with the interaction of wild type NGF with TrkA (9, 12), but to ensure that BDNF has no effect on the interaction of TrkA with NGF3T, a mutant form of NGF, we examined the effect of BDNF on the activation of TrkA in fibroblasts that do not express p75NTR. Fig. 2B shows that the NGF3T-mediated increase in TrkA tyrosine phosphorylation in these cells was unaffected by BDNF preincubation.

Binding of BDNF to p75NTR on PC12 cells activates sphingomyelinase activity and results in the accumulation of intracellular ceramide (15). To test if BDNF-mediated ceramide accumulation might mediate the reduction in TrkA activity, PC12 cells were pretreated with increasing concentrations of C₂-ceramide, a cell-permeable ceramide analogue, and then supplemented with either NGF or NGF3T for 5 min and analyzed for Trk phosphotyrosine levels. Fig. 3A shows that C₂-ceramide levels greater than 15 µM cause a profound reduction in the level of Trk tyrosine phosphorylation induced by NGF, similar to the effects observed with BDNF. Densitometry shows an average decrease in NGF-stimulated tyrosine phosphorylation of 61% at 20 µM C₂-ceramide, 71% at 50 µM C₂-ceramide, and 80% at 100 µM C₂-ceramide. Very similar results were obtained when NGF3T was used in place of NGF (data not shown). Levels of TrkA protein remain unchanged under these conditions (Fig. 3B), indicating that C₂-ceramide reduced the ability of TrkA to respond to NGF. To determine whether this represents a specific effect of ceramide or a nonspecific lipid effect, these experiments were repeated using dihydrosphingosine, the inactive saturated analog of C₂-ceramide. Dihydrosphingosine had no effect on NGF-mediated TrkA activation (Fig. 3C).

Tyrosine kinase receptors are ligand-regulated transmembrane enzymes, but kinase activity of the intracellular domain is subject to intracellular regulation, with phosphorylation of key intracellular serine or threonine residues playing an important regulatory role (22-25). To determine if BDNF or ceramide treatment results in alterations in the level of serine or threonine phosphorylation within the TrkA intracellular domain, we performed phosphoamino acid analysis of the TrkA receptor. PC12-6/24 cells, which overexpress the TrkA receptor, were grown in [³²P]orthophosphate for 2 h, treated with NGF, BDNF, or C₂-ceramide for 30 min, and then analyzed for phosphoamino acid content. Fig. 4 shows that TrkA obtained from untreated cells contained low levels of phosphoserine, and as expected, TrkA from cells treated with NGF treatment contained abundant phosphotyrosine and also showed an increase in phosphoserine levels of about 50-fold. TrkA receptor from cells treated with either BDNF or C₂-ceramide in the absence of NGF also showed a clear increase in phosphoserine content, with maximal increases in phosphoserine content approaching 12-fold with either reagent. These results indicate that the phosphorylation state of the TrkA intracellular domain is regulated by BDNF-activated signal and by ceramide and indicate that ligand-dependent activation of the TrkA receptor may be regulated by increased serine phosphorylation of its intracellular domain.

DISCUSSION

p75NTR and TrkA have a complex functional interaction. TrkA activation is increased in cells coexpressing the two receptors compared with those expressing TrkA alone; at least part of this effect is due to a p75NTR-mediated increase in the amount of NGF that ultimately becomes bound to the TrkA receptor (9, 12, 13). Intriguingly, whereas p75NTR increases the ability of TrkA to become activated by preferred ligands such as NGF, correlative evidence suggests that p75NTR reduces the ability of TrkA to become activated by nonpreferred ligands such as neurotrophin-3 and neurotrophin-4 (26-29). In our studies, we have found that treatment of PC12 cells with BDNF results in a pronounced attenuation in subsequent signaling through TrkA. This effect of BDNF is not due to disruption of NGF binding to p75NTR, since the NGF3T mutant used in these experiments does not bind this receptor. Furthermore, BDNF does not interfere with NGF3T-mediated activation of TrkA on cells that do not express p75NTR, indicating that the effects observed on PC12 cells are not due to disruption of NGF3T binding to TrkA. Instead, these effects are most easily explained by a mechanism that involves BDNF binding to p75NTR and subsequent activation of a signaling cascade that regulates TrkA activity.

Regulation of receptor activity by intracellular signaling paths can occur through a process termed receptor transmodulation. Selective phosphorylation of specific serine and threonine residues within the epidermal growth factor and insulin receptors results in a reduction in subsequent ligand-mediated receptor activity (22, 24, 30, 31). The mechanisms controlling these regulatory events are not well understood, but recent analyses of the effect of TNF receptor activation on insulin receptor signaling has been revealing. TNF treatment results in a reduction in subsequent ligand-stimulated insulin receptor signaling, and recent reports have shown that this attenuation is due to TNF-induced serine and threonine phosphorylation of the insulin receptor and its downstream tyrosine kinase signaling partner, IRS1. TNFR1 and TNFR2-mediated activation of sphingomyelinase activity and the subsequent accumulation of ceramide appear to be key components of this signaling cascade (32-38). Because BDNF, acting through p75NTR, also stimulates sphingomyelinase activity within PC12 cells (15), we reasoned that the BDNF-mediated attenuation in TrkA activation may likewise function through ceramide-dependent phosphorylation events. Consistent with this, we found that C₂-ceramide, a cell-permeable ceramide analogue, mimics the effect of BDNF on TrkA activation and that BDNF and C₂-ceramide lead to an increase in TrkA phosphoserine content. NGF treatment also results in a profound increase in TrkA phosphoserine, but p75-mediated activation of sphingomyelinase is apparently suppressed when TrkA is activated (15). Therefore, NGF-dependent serine phosphorylation of TrkA is unlikely to be a ceramide-dependent event but may instead be an indirect consequence of TrkA activation. Comparing the specific TrkA phosphoserine residues phosphorylated in response to NGF with those phosphorylated by BDNF and C₂-ceramide treatment will allow us to define the specific activities mediating these effects.

We have not yet determined the mechanism by which BDNF and ceramide lead to an increase in TrkA phosphoserine content. BDNF and ceramide may activate cascades that result in activation of a regulated serine-directed kinase or could result in the stable association of some constitutively active kinase with TrkA. Alternatively, the increase in phosphoserine content observed in response to BDNF or ceramide treatment may reflect an induced dephosphorylation of TrkA with subsequent replacement with labeled phosphate through the process termed "back-phosphorylation." Identification of the serine residues within the TrkA intracellular domain that are substrates of this modulatory activity should allow us to dissect the mechanism by which ceramide and BDNF alter levels of TrkA phosphoserine.

Our studies indicate that a p75-activated signaling pathway that involves ceramide acts to negatively regulate ligand-stimulated TrkA activity. Combined with work by others that suggests that a TrkA-derived signal may reduce the capacity of p75NTR to signal (15), this suggests a complex trans-receptor regulatory loop between the two receptors. When exposed to NGF, a preferred ligand, the primary function of p75NTR appears to be to increase TrkA signaling, and the transmodulatory activity of p75NTR is likely inhibited under these conditions. Conversely, in the presence of nonpreferred ligands, our results suggest that p75NTR may inhibit nonpreferred ligand-stimulated TrkA activity through ceramide-dependent receptor transmodulation.