Neurotrophins Redirect p75NTR from a Clathrin-Independent to a Clathrin-Dependent Endocytic Pathway Coupled to Axonal Transport

Authors

  • Katrin Deinhardt,

    1. Molecular Neuropathobiology Laboratory, Cancer Research UK, London Research Institute, 44 Lincoln’s Inn Fields, WC2A 3PX, London, UK
    2. Current address: Molecular Neurobiology Program, Skirball Institute of Biomolecular Medicine, New York University School of Medicine, New York, NY 10016, USA
    Search for more papers by this author
  • Alessandra Reversi,

    1. Molecular Neuropathobiology Laboratory, Cancer Research UK, London Research Institute, 44 Lincoln’s Inn Fields, WC2A 3PX, London, UK
    Search for more papers by this author
  • Otto Berninghausen,

    1. Department of Biological Sciences, Imperial College London, SW7 2AZ, London, UK
    2. Current address: LMU Gene Center, University of Munich, Feodor-Lynen-Str 25, 81377 Muenchen, Germany
    Search for more papers by this author
  • Colin R. Hopkins,

    1. Department of Biological Sciences, Imperial College London, SW7 2AZ, London, UK
    Search for more papers by this author
  • Giampietro Schiavo

    Corresponding author
    1. Molecular Neuropathobiology Laboratory, Cancer Research UK, London Research Institute, 44 Lincoln’s Inn Fields, WC2A 3PX, London, UK
    Search for more papers by this author

Abstract

The p75 neurotrophin receptor (p75NTR) plays multiple roles in neuronal physiology through interactions with many ligands and coreceptors. However, its intracellular neuronal trafficking prior to and after neurotrophin activation is still poorly characterized. We have previously shown that in response to nerve growth factor (NGF), p75NTR is retrogradely transported along the axons of motor neurons (MNs) in carriers shared with NGF, brain-derived neurotrophic factor and the tyrosine kinase receptor TrkB. Here, we report that NGF does not enhance the internalization or degradation of p75NTR, which undergoes a rapid dynamin-dependent and clathrin-independent recycling process in MNs. Instead, incubation of cells with NGF leads to the redirection of a pool of plasma membrane p75NTR into clathrin-coated pits. The subsequent internalization of p75NTR via clathrin-mediated endocytosis, as well as the activity of Rab5, are essential for the sorting of the p75NTR-containing endosomes to the axonal retrograde transport pathway and for the delivery of p75NTR to the soma. Our findings suggest that the spatial regulation of p75NTR signalling is controlled by these ligand-driven routes of endocytosis.

The p75 neurotrophin receptor (p75NTR) plays important roles in regulating neuronal death and survival. It binds all four neurotrophins with approximately equal affinity in the nanomolar range (1) as well as the proneurotrophins with higher affinity (2). The binding of mature neurotrophins to p75NTR causes a variety of physiological responses, including neuronal survival and differentiation (3), whereas the interaction with proneurotrophins induces apoptosis in sympathetic neurons (4). The ability of p75NTR to bind a large number of ligands, including proteins not belonging to the neurotrophin family (5,6), and thus to mediate distinct physiological responses is partially because of its interaction with several coreceptors coupled to a variety of signalling cascades. For example, p75NTR forms heterodimers with Trk receptors, thereby increasing Trk’s affinity for individual neurotrophins and influencing neuronal survival. Sortilin has been identified as the coreceptor for p75NTR in pro-nerve growth factor (NGF) and pro-brain-derived neurotrophic factor (BDNF) binding and is required for conveying their proapoptotic responses (4,7). Alternatively, p75NTR can form a complex with Nogo receptor, the polysialoganglioside GT1b and leucine repeat rich and Ig domain containing Nogo receptor interacting protein 1 (LINGO), which binds myelin-based growth inhibitors. These interactions may be important in both neuronal outgrowth and myelination (8,9).

Neurotrophin receptor complexes activated at distal sites on the axonal plasma membrane are, at least in part, internalized and retrogradely transported to carry out their biological functions in the soma (10–12). Despite their physiological importance, the endocytic mechanism controlling the uptake and intracellular trafficking of p75NTR and other neurotrophin receptors is still controversial. The p75 neurotrophin receptor has been detected in detergent-resistant membrane fractions in NGF-stimulated pheochromocytoma cells (PC12) and cerebellar and hippocampal neurons, indicating its ability to partition into cholesterol-enriched membrane microdomains (13,14). However, it is internalized via a clathrin-dependent mechanism in PC12 cells upon NGF or BDNF stimulation (15,16), suggesting that the interplay of membrane raft- and clathrin-based pathways could be required for internalization and sorting of the neurotrophin-receptor complexes.

Recently, we showed that in spinal cord motor neurons (MNs), NGF, BDNF and p75NTR share an axonal retrograde transport compartment with the carboxy-terminal-binding fragment of tetanus neurotoxin (17,18). Tetanus neurotoxin binds to the neuromuscular junction and, upon internalization, is retrogradely transported to the soma of MNs located in the spinal cord. This trafficking pathway is highly specific and essential for tetanus pathogenesis (19). Endocytosis of the carboxy-terminal-binding fragment of tetanus neurotoxin into MNs is dependent on detergent-resistant membranes as well as on a specific clathrin-dependent pathway (20,21). Here, we report that NGF binding is not required for p75NTR internalization but induces its relocalization into clathrin-coated pits (CCPs). Although an intact clathrin machinery is dispensable for the bulk uptake of p75NTR into MNs, clathrin-dependent endocytosis is essential for the sorting of p75NTR to the axonal retrograde transport route, which also requires the activity of Rab5 downstream of clathrin-dependent receptor internalization.

Results

p75NTR undergoes NGF-independent internalization in MNs

Neurotrophins have been shown to promote p75NTR internalization into PC12 cells via a clathrin-dependent mechanism (15,22). However, it is presently unclear if this finding has general validity and applies to primary neurons, including spinal cord MNs. To investigate if this is the case, we chose NGF for p75NTR stimulation as MNs in culture do not express TrkA (23 and data not shown) and therefore p75NTR is the only NGF receptor in this system. Uptake of p75NTR was assessed by surface biotinylation of MNs on ice, followed by incubation at 37°C for 30 min with or without NGF. This relatively long internalization time was necessary given the very slow endocytosis of p75NTR with no visible uptake at time-points earlier than 15 min (Figure S1). This is not a unique feature of MNs because similar internalization kinetics have been reported in PC12 cells (15). Cells were then shifted back to ice and treated with sodium 2-mercaptoethanesulfonate (MESNA) to remove the remaining surface-bound biotin (20,21). Internalized biotinylated proteins were then isolated on avidin beads. Samples were analyzed by Western blot for the presence of p75NTR as well as the constitutively recycling transferrin receptor (TfR). As shown in Figure 1A,B, p75NTR was taken up into MNs with or without addition of NGF to comparable amounts. To follow p75NTR internalization in the soma as well as in neurites, MNs were then incubated with an antibody directed against the extracellular domain of p75NTR [p75NTR(EC), (17)] on ice, with or without NGF, for 45 min at 37°C. Cells were acid washed to remove all surface-bound antibody (see Figure S2A for controls) and analyzed by immunofluorescence. There was no visible difference in internalization between the somatodendritic and axonal compartments in the presence or absence of NGF (Figure 1C). To demonstrate that the antibody binding did not interfere with the ability of NGF to recognize p75NTR, we performed internalization experiments in three different ways: by adding p75NTR(EC) and NGF simultaneously on ice for 30 min, by preincubating MNs with the antibody followed by the addition of NGF or the other way around. Strikingly, we did not observe any difference in p75NTR uptake in the presence or absence of NGF on using full medium (Figure 1C) or under starvation conditions (Figure S2B). These findings are not dependent on the antibody used to monitor p75NTR internalization because the same receptor behaviour has been observed using the antibody IgG192, which also binds to the extracellular domain of p75NTR(15) (Figure S2C).

Figure 1.

The role of NGF in p75NTR endocytosis. A–C) p75 neurotrophin receptor is internalized in MNs in presence or absence of NGF. A) Mixed ventral horn cultures were surface biotinylated and incubated with or without 100 ng/mL NGF for 30 min. Cells were then MESNA treated to remove remaining surface-bound biotin. Biotinylated proteins were isolated on avidin-sepharose beads and analyzed by Western blot for p75NTR and TfR. B) The ratio of the p75NTR signal versus TfR signal was used to normalize the uptake of p75NTR. Shown is the mean of two independent experiments. C) Cells were incubated with the p75NTR(EC) antibody only (a), with p75NTR(EC) and NGF simultaneously on ice for 30 min (b), with p75NTR(EC) followed by the addition of NGF (c), or with NGF followed by addition of p75NTR(EC) (d). Cells were then washed and shifted to 37°C for 45 min, acid washed, fixed and stained with an Alexa488-coupled anti-rabbit IgG secondary antibody to visualize internalized p75NTR(EC). Scale bar, 10 μm. D, E) p75 neurotrophin receptor is not degraded in response to NGF. D) Mixed ventral horn cultures were surface biotinylated and incubated with NGF for indicated amounts of time. Biotinylated proteins were isolated on avidin-sepharose beads and analyzed by Western blot for p75NTR and TfR. E) Levels of p75NTR relative to TfR were determined at different times points. Bars indicate the mean of two independent experiments.

In contrast to TrkA, p75NTR is not degraded in response to NGF in PC12 cells (22,24). To ascertain whether or not this holds true in MNs, we first looked at the level of colocalization between internalized p75NTR and the lysosomal marker lysosome-associated membrane protein (LAMP-1). While dendrites and soma contain numerous LAMP-1 positive structures, we observed very few LAMP-1-containing organelles in axons (Figure S3A). After 90 min of uptake, no significant overlap between LAMP-1 and p75NTR was observed either in axons (Figure S3B) or in the somatodendritic compartment (Figure S3C) in the presence or absence of NGF. This was also the case at later time-points (i.e. 180 min of internalization; see Figure S3D,E), suggesting that p75NTR is not targeted to lysosomes in response to NGF. The degradation of p75NTR was directly assessed by surface biotinylation on ice, followed by the treatment of MNs with NGF at 37°C for the indicated times. Upon isolation of biotinylated proteins on avidin beads, samples were analyzed by Western blot for the presence of p75NTR and TfR. In agreement with previous studies (22,24), p75NTR is not degraded for at least 6 h in response to NGF (Figure 1D,E).

p75NTR recycles at the plasma membrane

Given the lack of targeting to lysosomes and degradation of the internalized p75NTR in MNs, we considered the possibility of its recycling back to the plasma membrane. For this, we adapted a microscopy assay that was used previously to demonstrate Trk recycling (25). To this end, we incubated MNs with Cy3-p75NTR(EC) on ice before washing and warming up for 30 min at 37°C in absence or presence of NGF, followed by an acid wash to remove the surface-bound antibody. Motor neurons were then incubated in presence of Alexa647-coupled anti-rabbit IgG for 60 min at 37°C, which allowed the detection of the pool of antibody and as a consequence of p75NTR, that recycled back to the neuronal surface. Motor neurons were then fixed and imaged by confocal microscopy. In contrast to previous observations in PC12 cells (22), a significant proportion of p75NTR recycles both in absence and presence of NGF (Figure 2). No signal was detected in control MNs where Cy3-p75NTR(EC) was omitted, indicating that the staining with the secondary anti-rabbit IgG was specific for the primary antibody bound to recycling p75NTR and not the result of fluid-phase uptake (Figure 2A, bottom panel). Therefore, the structures that were only labelled by the secondary antibody and not for Cy3-p75NTR(EC) (in green in Figure 2A) most likely represent unconjugated p75NTR(EC) and were quantified as recycled p75NTR structures. The recycling of p75NTR occurs both in the somatodendritic compartment and axons to a similar extent (Figure 2C). Interestingly, we repeatedly observed a slight reduction in the levels of p75NTR recycling within axons upon NGF treatment (Figure 2C). The p75 neurotrophin receptor recycling was also observed using a different p75NTR antibody (IgG192; Figure S4), strongly suggesting that this type of NGF-independent receptor trafficking is an intrinsic property of p75NTR and was not induced by the binding of the p75NTR(EC) antibody to the receptor.

Figure 2.

p75 neurotrophin receptor undergoes NGF-independent recycling in spinal cord MNs. A, B) Motor neurons were incubated with 0.4 μg/mL Cy3-p75NTR(EC) (‘internalized’, in red) with or without 100 ng/mL NGF for 30 min, acid washed and incubated for further 60 min in presence of Alexa647-coupled anti-rabbit IgG (‘recycled’, green). Cells were fixed and imaged by confocal microscopy. A portion of internalized p75NTR recycles both in the somatodendritic compartment (A) and axons (B, arrowheads). In the bottom row of A (control), Cy3-p75NTR(EC) was omitted. Scale bar, 10 μm. C) Quantification of the p75NTR-containing puncta that are positive for secondary antibody versus the total p75NTR-positive structures. p75 neurotrophin receptor recycling is not significantly altered by NGF. Columns show the mean of three (axon) and two (soma) independent experiments. Error bars indicate standard deviation, while numbers in brackets, the total numbers of puncta observed.

NGF directs p75NTR to a fast axonal transport pathway

A recently reported effect of NGF on the fate of p75NTR(EC) is the induction of a robust retrograde axonal transport of this antibody and its bound receptor in spinal cord MNs (17). However, the molecular mechanism at the basis of this observation and its dependence on different neurotrophins remain unclear. To gain insights into this process, MNs were incubated with p75NTR(EC) alone or in presence of 100 ng/mL NGF or 50 ng/mL BDNF, washed and imaged by time-lapse confocal microscopy. In absence of neurotrophins, p75NTR carriers were rarely seen (Figure 3A). However, upon the addition of either NGF (Figure 3B) or BDNF (Figure 3C), high-frequency fast retrograde transport of p75NTR(EC) was observed. Figure 3 shows the displacement charts of representative movies, where the progression of individual carriers, each one represented by a single trace, is plotted over time. Therefore, even though the endocytosis and recycling of p75NTR are largely independent of the presence of neurotrophins, its efficient sorting to the axonal retrograde transport pathway is strictly neurotrophin dependent.

Figure 3.

Neurotrophins induce retrograde axonal transport of p75NTR. Motor neurons were incubated with 0.4 μg/mL Cy3-p75NTR(EC) in control conditions (A) or upon addition of 100 ng/mL NGF (B) or 50 ng/mL BDNF (C) for 30 min, washed and imaged by time-lapse confocal microscopy. Cy3-p75NTR(EC)-positive carriers were tracked, and individual carriers were displayed as traces showing displacement over time. A) The p75 neurotrophin receptor carriers were rare in absence of neurotrophins (four carriers; 10 axons; n = 3 independent experiments). In contrast, Cy3-p75NTR(EC) showed robust retrograde transport after exposure either to NGF (B; 198 carriers; 11 axons; n = 3 independent experiments) or to BDNF (C; 127 carriers; six axons; n = 2 independent experiments).

NGF stimulation targets p75NTR to clathrin-coated invaginations

Next, we were interested in understanding the level at which this sorting is initiated. In MNs, p75NTR and carboxy-terminal-binding fragment of tetanus neurotoxin (TeNT HC) share retrograde transport carriers, which TeNT HC reaches upon internalization via a specialized clathrin-dependent process (17,20). Whether the same endocytic route was responsible for p75NTR internalization in response to NGF is, however, unclear. Therefore, we first determined the level of codistribution of p75NTR and clathrin by fine structural analysis. Because endocytosis and subsequent uncoating of clathrin-coated vesicles may occur very quickly at 37°C (26), we incubated MNs with p75NTR(EC) at 12°C, a temperature that allows the lateral movement of ligand–receptor complexes within the plasma membrane and invagination but blocks the pinching off of vesicles (20). In absence of NGF, the vast majority of p75NTR was found in uncoated invaginations (94%, n = 50 structures) (Figure 4Aa). In contrast, p75NTR could be found both in clathrin-coated and uncoated structures upon NGF stimulation (Figure 4Ab,c; arrowheads and arrows, respectively), suggesting that NGF triggers the recruitment of p75NTR to CCP. Quantification showed that in the presence of NGF, 39% of p75NTR-associated gold was found in clathrin-coated structures (n = 66 structures) (Figure 4B).

Figure 4.

p75 neurotrophin receptor enters coated and uncoated structures in MNs. A) Motor neurons were incubated on ice with 2 μg/mL p75NTR(EC) without (a) or with (b, c) 100 ng/mL NGF for 45 min, washed, incubated with 10 nm anti-rabbit immunogold for 30 min on ice and then chased in medium for 45 min at 12°C. Gold label was found in uncoated (arrows and b′) and coated structures (arrowheads and c′). Scale bar, 0.2 μm. B) Quantification of coated and uncoated invaginations containing gold particles. Nerve growth factor induces a shift of p75NTR towards CCP. Motor neurons show mixed populations of endocytic structures containing p75NTR upon NGF treatment (n = 15 MNs). Numbers in brackets indicate numbers of structures observed.

Dynamin, but not AP-2 or AP180, is essential for p75NTR uptake into MNs

The process of clathrin-mediated endocytosis involves the coordinated action of a large number of proteins, including the adaptors AP-2 and AP180, and dynamin, a GTPase that plays an essential role in several internalization pathways. Incubation of MNs with the cell-permeable peptide P4, which inhibits dynamin function (27), significantly reduced uptake of Cy3-labelled p75NTR(EC), while treatment with the scrambled peptide P4S showed no effect (data not shown). The requirement for dynamin in p75NTR endocytosis was confirmed by expression of the dominant-negative dynamin2K44A mutant (28). After 25 h of expression, cells were incubated with Cy3-labelled p75NTR(EC), 100 ng/mL NGF and 0.2 mg/mL dextran for 30 min at 37°C. Upon acid wash and fixation, MNs were immunostained to visualize the expression of the tagged dynamin mutant (Figure 5Ae). Control MNs readily internalized p75NTR(EC) (Figure 5Aa), transferrin (Figure 5Ac) and dextran (Figure 5Af, cross). In contrast, the dynamin2K44A mutant completely abolished the uptake of Cy3-p75NTR(EC) (Figure 5Ad, asterisk), while a neighbouring control cell readily internalized Cy3-p75NTR(EC) (Figure 5Ad, cross). In addition, fluid-phase endocytosis was not significantly affected in dynamin2K44A-expressing MNs (Figure 5Af, asterisk), therefore confirming the specificity of the inhibition.

Figure 5.

Internalization of p75NTR is dynamin dependent but does not require a functional clathrin machinery. A) After 25 h following microinjection of a plasmid encoding dynamin2K44A (d–f), AP-2 μ2T156A and tTA (g–i) or 26 h after microinjection of a construct expressing AP180-C (k–m), MNs were incubated with 0.4 μg/mL Cy3-p75NTR(EC), 100 ng/mL NGF and 20 μg/mL Alexa647-transferrin or 0.2 mg/mL Alexa647-dextran (molecular weight 10 000) for 30 min, acid washed, fixed, stained for the epitope tags of the expressed proteins and imaged. Uninjected MNs were imaged as a control (a–c). Scale bar, 10 μm. The asterisk marks a dynamin2K44A-expressing MNs, whereas the cross points to an uninjected cell (d–f). B) Quantification of the effects of the overexpression of dominant-negative mutants on p75NTR endocytosis. Numbers in brackets indicate the MNs observed per condition.

Next, we assessed the requirement for AP-2 and AP180. The overexpression of dominant-negative mutants of these two adaptors impairs clathrin-mediated endocytosis in cell lines (29,30) and primary MNs (20). After 25–26 h from the microinjection of AP-2 μ2T156A (Figure 5Ag–i) or AP180-C (Figure 5Ak–m) expressing plasmids, MNs were incubated with Cy3-labelled p75NTR(EC), 100 ng/mL NGF and 20 μg/mL transferrin for 30 min at 37°C and processed as described above. Expression of either AP-2 μ2T156A or AP180-C in MNs completely blocked transferrin uptake (Figure 5Ai,m), indicating that clathrin-mediated endocytosis was completely inhibited under these conditions. However, the uptake of p75NTR was not visibly affected in these cells (Figure 5Ag,k). These results, which were quantified in Figure 5B, showed that even though the addition of NGF causes a redistribution of p75NTR into CCP and other clathrin-coated endocytic structures, the majority of p75NTR uptake into MNs follows a dynamin-dependent, clathrin-independent pathway.

AP180 is required for directing p75NTR to the retrograde transport route

The binding of p75NTR to different ligands and coreceptors may result in the multiple internalization pathways observed. Sorting of the distinct ligand–receptor complexes along these endocytic routes could therefore be important for the final physiological readout of p75NTR-dependent signalling. In this light, the ligand-dependent sorting of p75NTR may already start at the level of the plasma membrane, as suggested by its increased association with CCP upon NGF stimulation. To test this hypothesis, we microinjected MNs with an AP180-C-expressing plasmid and imaged the axonal transport of endogenous p75NTR after internalization of p75NTR(EC) in the presence of NGF (Figure 6). The kinetic analysis of p75NTR-containing vesicles along an MNs axon is here represented as a kymograph, which was generated by drawing a line along the axon, and stacking pixels derived from subsequent frames of the video as a function of time. In the resulting diagram, stationary compartments appear as vertical lines, while moving carriers are shown as diagonal traces, the slope of which is inversely proportional to the speed of the moving organelle (Figure 6).

Figure 6.

Axonal retrograde transport of p75NTR requires internalization via CCP. After 24 h following microinjection of a plasmid encoding AP180-C, MNs were incubated with 0.4 μg/mL Cy3-p75NTR(EC) and 100 ng/mL NGF, washed and imaged. A) The p75 neurotrophin receptor was readily transported in an uninjected control MNs, whereas p75NTR-positive structures are stationary in a MNs expressing AP180-C (B). Top panels in A and B show the GFP expression of injected cells. Four individual frames from a confocal time series of an axon, corresponding to the areas boxed in the differential interference contrast (DIC) image, display the kinetics of transport of Cy3-p75NTR(EC). Arrows point out a p75NTR-positive moving vesicle and arrowheads a moving tubule (A). Asterisks highlight a stationary organelle (B). Scale bars, 10 μm. Kymographs of the time-lapse sequences are also shown (lower part of A and B). The kymographs were generated by drawing a line along the axon and aligning vertically the horizontal single line scans of subsequent frames. The time scale is indicated to the left. See also Videos S1 and S2.

Control MNs readily transported Cy3-p75NTR(EC) (Figure 6A and Video S1). In contrast, even though expression of AP180-C did not abolish p75NTR uptake (Figure 5Ak), its axonal retrograde transport was completely blocked in MNs expressing this dominant-negative mutant (Figure 6B and Video S2). Cell functionality was assessed by incubating MNs immediately after transport analysis with tetramethylrhodamine ethyl ester (TMRE), a dye that accumulates in mitochondria with intact membrane potential. Mitochondrial distribution and morphology were identical in axons of MNs expressing AP180-C (Figure S5) and control cells (data not shown), indicating that the blockade of the retrograde transport of p75NTR was specific and not because of cell suffering.

Rab5 is required for the sorting of p75NTR to axonal retrograde transport carriers

Nerve growth factor and p75NTR, as well as TeNT HC, use a retrograde transport route in MNs axons that is regulated by the small GTPase Rab7 (17). In addition, the sorting of TeNT HC to the axonal transport pathway is Rab5 dependent (17). To assess whether p75NTR also requires Rab5 for its targeting to this axonal route, we incubated MNs with p75NTR(EC) in the presence or absence of NGF for 30 min before fixation and stained for endogenous Rab5. Under these conditions, the two probes displayed a partial overlap, indicating that at least a portion of p75NTR is sorted into Rab5-positive compartments following internalization (Figure 7A). At this relatively late time-point of internalization, we found a larger overlap of p75NTR and Rab5 in NGF-treated samples (87.2%, n = 193 structures) compared with untreated samples (15.3%, n = 210 structures). This slow sorting of p75NTR in NGF-treated samples correlates with the late onset of axonal transport, which typically reaches steady state not before 45 min after start of the incubation. In contrast, the recycling pool of p75NTR may transiently associate with Rab5-positive endosomes (see below). Next, we microinjected green fluorescent protein (GFP)-Rab5wt or the constitutively active mutant GFP-Rab5Q79L-expressing vectors into MNs and after 16–20 h incubated the cells with Cy3-p75NTR(EC) and NGF before live imaging. GFP-Rab5wt displayed localized short-range movements within axons, with no evidence of vectorial movement over long distances (Figure 7Ba,d) (17). The p75 neurotrophin receptor was found in some of these stationary or oscillating Rab5-positive structures (Figure 7Ba,c,f; asterisks), whereas p75NTR moving carriers were negative for Rab5 (Figure 7Bc,f; arrowheads; see also Video S3). Expression of either GFP-Rab5wt or GFP-Rab5Q79L did not alter the frequency or speed profile of transported p75NTR compartments (Figure S6 and data not shown). In contrast, expression of a dominant-negative version of Rab5 (GFP-Rab5N133I) blocks the retrograde transport of Cy3-p75NTR(EC) (Figure 8A), suggesting that the association of p75NTR with Rab5-positive compartments was functionally relevant for its recruitment to the axonal retrograde transport pathway. Given that internalization of the probe into MNs was not affected by GFP-Rab5N133I expression (Figure 8B), this block is likely to occur after completion of the pinching off of the p75NTR-containing endosome. These findings indicate that the targeting of p75NTR to the retrograde transport route, but not its uptake into MNs, is dependent on Rab5 activity.

Figure 7.

Rab5 and p75NTR colocalize at early stages of internalization. A) Motor neurons were incubated with the antibody p75NTR(EC) for 30 min on ice in presence or absence of 100 ng/mL of NGF and then shifted to 37°C for 45 min. Cells were acid washed to remove the surface-bound p75NTR(EC) and fixed and stained with an antibody against endogenous Rab5. An increase of double-positive organelles is observed after treatment with NGF (see inset). B) Motor neurons were microinjected with a plasmid encoding GFP-Rab5wt, and after overnight expression, cells were treated with 0.4 μg/mL Cy3-p75NTR(EC) (in red) and 100 ng/mL NGF for 30 min at 37°C, washed and imaged by time-lapse confocal microscopy. Green fluorescent protein-Rab5wt does not interfere with p75NTR transport and localizes to oscillating p75NTR-positive compartments (B, see asterisks), but not to p75NTR-positive carriers (B, arrowheads). Kymographs in B have been prepared as described in Figure 6. See also Video S3. Scale bar, 10 μm (inset, 2 μm).

Figure 8.

Rab5 is essential for sorting of p75NTR into the axonal transport carriers. A) Motor neurons were microinjected with a plasmid encoding GFP-Rab5N133I, and after overnight expression, cells were treated with 0.4 μg/mL Cy3-p75NTR(EC) (in red) and 100 ng/mL NGF for 30 min at 37°C, washed and imaged by time-lapse confocal microscopy. Motor neurons overexpressing GFP-Rab5N133I show nearly complete inhibition of the axonal retrograde transport of p75NTR. The kymograph in A was prepared as described in Figure 6. See also Video S4. B) Overexpression of GFP-Rab5N133I in MNs does not impair p75NTR entry. After 16 h expression of GFP-Rab5N133I (in green), MNs were incubated with 0.4 μg/mL Cy3-p75NTR(EC) (in red) and 100 ng/mL NGF for 30 min at 37°C, acid washed, fixed, stained for GFP and imaged.

Discussion

The neurotrophin receptor p75NTR interacts with many different ligands and coreceptors, which modulate its signalling properties and direct its intracellular fate. Accordingly, p75NTR has been shown to be internalized and retrogradely transported in an intact form (12) or to undergo regulated proteolysis, resulting in the shedding of the extracellular domain prior to internalization (31). In this study, we have addressed the trafficking of the unprocessed receptor in response to neurotrophins using antibodies raised against the extracellular domain of p75NTR(17). At variance with the neurotrophin-dependent uptake and recycling of p75NTR observed in PC12 cells (15,22), we found that NGF does not alter bulk p75NTR internalization, recycling or degradation in MNs. Ligand-independent endocytosis and recycling have been described not only for house-keeping receptors such as TfR, but also for the signalling receptor Notch and may function in preventing its ligand-independent activation (32,33). Even though NGF did not trigger p75NTR endocytosis in MNs, we found that this neurotrophin targeted a pool of p75NTR into CCP. In addition, the stimulation with NGF or BDNF was strictly required for the sorting of p75NTR into the axonal retrograde transport pathway. This transported population of p75NTR most likely transmit signals from distal parts of the axon to the cell soma. While the amount of p75NTR recycling in the soma is not significantly different with or without NGF, we consistently observed a slight reduction of the axonal recycling p75NTR upon NGF addition. This may represent the portion of p75NTR that has been redirected from a local recycling pathway towards long-range retrograde transport and suggests that p75NTR undergoes a dynamic equilibrium on the plasma membrane of MNs that can be modulated by neurotrophins.

Previous work on epithelial cells demonstrated that cargo sorting in clathrin-mediated endocytosis is initiated at the level of plasma membrane (34). More specifically, cargoes targeted for degradation are preferentially internalized into dynamic endosomes, whereas recycling ligands are enriched in a slowly maturing endosomal population (34). In agreement with this model, p75NTR has previously been shown to enter PC12 cells via a slow, clathrin-mediated endocytic route (15,16). However, we found that the clathrin machinery is not required for bulk p75NTR endocytosis into MNs because interfering with either AP-2 or AP180 function did not inhibit its uptake. In spite of this result, clathrin-mediated endocytosis was strictly required for the internalization of the fraction of receptor undergoing retrograde transport because inhibition of the clathrin machinery completely abolished the long-range axonal movement of p75NTR triggered by NGF. Altogether, these findings indicate that sorting of p75NTR towards the retrograde transport pathway is triggered by neurotrophin binding and is initiated at the plasma membrane.

We have previously shown by biochemical and imaging approaches that neurotrophins and their receptors, as well as TeNT HC, share an axonal transport compartment (17), which is characterized by a neutral pH (18,35), an environment that would preserve the ligand–receptor complexes during the long-distance journey from the nerve terminal to the MNs soma. Interestingly, both p75NTR and TeNT HC display similar kinetics for the onset of retrograde transport and require the activity of Rab7 (17). Here, we have demonstrated that just like TeNT HC(17,20), p75NTR also requires Rab5 activity as well as a functional clathrin machinery for its targeting to fast retrograde transport carriers, strongly suggesting that both proteins follow an overlapping trafficking route.

The shared axonal transport compartment contains full-length p75NTR together with BDNF and TrkB (17). Interestingly, NGF or BDNF are equally effective in triggering the axonal transport of p75NTR in MNs, irrespective of the profile of expression of Trks in these cells. MNs do not express TrkA, which acts as the coreceptor for NGF binding, but contain TrkB, the coreceptor for BDNF (23). The overlapping effect of NGF and BDNF on the axonal transport of p75NTR indicates that this process requires neurotrophin binding to p75NTR but is not dependent on, or modulated by, Trks. Conversely, p75NTR influences Trk trafficking by delaying their internalization and degradation (36).

Traditionally, the main role of endocytosis in signal transduction was seen in terminating the signal through degradation of the activated receptor complexes. Over the last few years, this view has significantly changed as the spatial regulation of signalling is increasingly recognized (37,38). For example, in the case of transforming growth factor β (TGF-β), the mechanism of receptor internalization determines target activation versus receptor turnover. Uptake via a clathrin-mediated pathway directs TGF-β receptors into an early endosome containing downstream targets, therefore promoting the activation of the TGF-β signalling cascade, whereas internalization via caveolae leads to ubiquitin-dependent receptor downregulation (39). A similar mechanism has been shown for epidermal growth factor receptor (EGFR) signalling, where low doses of EGF, which are fully competent in the activation of downstream targets, direct EGFR to a clathrin-mediated pathway. High concentrations of EGF do not enhance this downstream signalling but instead promote its uptake via a clathrin-independent mechanism, resulting in increased receptor degradation (40). Receptor uptake via a non-clathrin-mediated pathway is not always associated with receptor downregulation. For example, localization to membrane microdomains is instrumental for efficient downstream signalling through the glial-derived neurotrophin factor receptor RET (41). Moreover, not only the endocytic route but also the precise location at which the signalling complex is activated contributes to the modulation of signal intensity. Accordingly, the mitogen-activated protein kinase cascade is sensitive to low-input signals if triggered at the level of the plasma membrane, whereas it is much less responsive when activated in the cytoplasm (42).

Our finding that neurotrophins do not induce an increase in p75NTR internalization, but instead trigger the recruitment of a pool of p75NTR to a long-range transport pathway and its delivery to the soma strengthens the view of a spatial regulation of p75NTR signalling in response to neurotrophins. This model exemplifies how a single receptor is able to mediate different signals, leading to distinct physiological outputs. In neurons, spatial regulation of signalling may influence whether a certain ligand induces local changes, for example, in growth cone guidance, or acts distally at the nuclear level, influencing cell survival and/or differentiation.

Materials and Methods

Reagents

Chemicals were from Sigma, BDH or Molecular Probes unless stated otherwise. Antibodies IgG192, 9E10, 4E12 and 12CA5 were from the Cancer Research UK antibody facility, anti-Rab5 was from Synaptic Systems, and anti-LAMP-1 was from Abcam. Plasmids encoding GFP-Rab5wt, GFP-Rab5Q79L and GFP-Rab5N133I were a gift from C. Bucci (Università di Lecce, I), the plasmid encoding AP-2 μ2T156A was a gift from E. Smythe (University of Sheffield, UK), and plasmids encoding dynamin2K44A and AP180-C were a gift from H.T. McMahon (LMB Cambridge, UK).

Microinjection, internalization and degradation assays

Motor neuron cultures were prepared from E14.5 rat embryos and maintained in culture as described previously (43). In some preparations, the 6.5% (w/v) metrizamide in L-15 (Gibco) was replaced by 10.4% (v/v) Optiprep (Axis-Shield) in phenol-red-free L-15. For biochemical experiments, mixed ventral horn cultures were used instead of purified MNs.

For internalization studies, cells were incubated with the polyclonal antibody p75NTR(EC) (17) and NGF (Alomone) simultaneously on ice for 30 min, with NGF followed by the addition of the antibody, or with the antibody followed by addition of NGF, washed and shifted to 37°C. To release surface-bound p75NTR(EC), MNs were acid washed by incubation in 100 mm citrate NaOH, pH 2.0, 140 mm NaCl for 5 min at room temperature before further processing by immunofluorescence or Western blot. Alternatively, cells were treated with 100 μm sulfo-NHS-SS-biotin (Pierce) in phosphate-buffered saline (PBS) for 30 min on ice, washed with PBS and incubated in medium with or without NGF at 37°C for the 30 min. The surface-bound biotin was removed by treatment with 25 mm MESNA in neurobasal medium, 20 mm Tris–Cl, pH 8.3, three times for 15 min. Cells were washed and lysed, and biotinylated proteins were recovered from postnuclear supernatants with tetrameric avidin-sepharose (Affiland) and analyzed by Western blot with a polyclonal antibody against the p75NTR intracellular domain [1:500, (17)] and TfR (1:1000; Zymed).

Motor neurons were microinjected between 4 and 7 days in vitro with 0.05 mg/mL of plasmid. In case of microinjection of multiple plasmids (e.g. the pTRE-μ2T156A plasmid that requires a transactivator ptTA for expression; Clontech), 0.04 mg/mL of each construct were premixed before injection. After 16–26 h of expression, MNs were incubated with 0.4 μg/mL Cy3-p75NTR(EC), 100 ng/mL NGF and 0.2 mg/mL AlexaFluor647-dextran (molecular weight 10 000) for 30 min at 37°C prior to the acid wash and fixation.

To assess p75NTR degradation, mixed ventral horn cultures were surface biotinylated as described above and incubated in medium with or without NGF at 37°C for the indicated time. Cells were lysed, and biotinylated proteins were recovered from postnuclear supernatants with tetrameric avidin-sepharose and analyzed by Western blot with antibodies against the p75NTR intracellular domain and TfR as described above. Recycling experiments were adapted from Chen et al. (25). Briefly, MNs were incubated with 0.4 μg/mL Cy3-p75NTR(EC) with or without 100 ng/mL NGF for 30 min, acid washed and incubated with Alexa647-coupled anti-rabbit IgG (1:1000) for additional 60 min before fixation.

Immunofluorescence and confocal microscopy

Motor neurons were fixed in 4% paraformaldehyde, 20% sucrose in PBS for 15 min at room temperature, permeabilized with 0.1% Triton X-100 in PBS for 5 min, blocked in 2% bovine serum albumin, 10% normal goat serum, 0.25% fish skin gelatin in PBS for 30 min and incubated with the relevant antibodies (anti-p75NTR 0.1 μg/mL; anti-HA (12CA5) 1:1000; anti-myc (9E10) 1:250; anti-GFP (4E12) 1:500; anti-LAMP-1 (LY1C6) 1:500; anti-Rab5 1:500; secondary antibodies 1:500) for 30 min in the blocking solution. Cells were mounted in Mowiol-488 and imaged using a Zeiss LSM510 confocal microscope equipped with a 63×, 1.4 c Plan Apochromat oil-immersion objective (Zeiss). For time-lapse confocal microscopy, MNs were incubated with 0.4 μg/mL Cy3-p75NTR(EC) and 100 ng/mL NGF for 30 min at 37°C in Neurobasal medium, washed three times with E4 supplemented with 30 mm HEPES–NaOH, pH 7.4 and imaged after 15 min at 37°C. To assess cell viability of AP180-C-expressing MNs, cells were incubated with 5 nm TMRE for 5 min after imaging and washed and imaged again with reduced laser power (HeNe 543: 0.5% output, low gain; no Cy3 signal was observed with these settings). Axons were identified by morphology as (1) the longest processes extending from a soma that (2) have very regular diameter with little to no branching and very few filopodia, which (3) show vectorial long-range transport. In selected cases, this identification was validated by the absence of a microtubule-associated protein (MAP-2) counterstain.

Images were processed using the Zeiss lsm 510 software, and kymographs were generated using MetaMorph (Molecular Devices) after rotation of the image stack to align the neuronal process vertically. Horizontal single line scans through the thickness of each process were plotted sequentially for every frame in the time series. Colocalization was quantified in MetaMorph using the ‘manually count objects’ option. For this, all Cy3-p75NTR-positive structures were manually marked and automatically counted and then the Alexa647 channel was overlaid and double-positive structures were highlighted and counted.

Electron microscopy

Motor neurons grown on glass cover slips were incubated with 2 μg/mL anti-p75NTR(EC) and 100 ng/mL NGF in neurobasal medium for 45 min at 4°C. Cells were then washed and incubated with a 10-nm gold-conjugated anti-rabbit antibody (BB International) on ice for additional 30 min and subsequently chased at 12°C for 45 min. Cells were fixed with 2% paraformaldehyde, 1.5% glutaraldehyde in 100 mm sodium cacodylate, pH 7.5 for 15 min and postfixed and embedded in Epon as described previously (44). Cells were then sectioned en face, and 60-nm sections stained with lead citrate were viewed in a Philips CM12 electron microscope.

Acknowledgments

We thank C. Bucci, H. T. McMahon and E. Smythe for mammalian expression plasmids. We are grateful to L. Bilsland, C. Bucci, M. Fainzilber, V. Neubrand, S. Salinas and S. Tooze for critical reading of the manuscript. This work was supported by Cancer Research UK (K. D., A. R. and G. S.) and the Medical Research Council (O. B. and C. R. H.).

Ancillary