Endocytosis of Activated TrkA: Evidence that Nerve Growth Factor Induces Formation of Signaling Endosomes

MATERIALS AND METHODS

Materials. Bis(sulfosuccinimidyl)suberate (BS³), disuccinimidyl suberate (DSS), and NHS-SS-biotin were obtained from Pierce (Rockford, IL). Potassium hydroxide (99%) was from Aldrich (Milwaukee, WI). Type IV collagen was from Collaborative Biomedical Products (Bedford, MA). Normal goat serum and Vectastain mounting medium were from Vector Laboratories (Burlingame, CA). Except as noted, Sigma (St. Louis, MO) was the source of all other reagents and chemicals.

NGF was prepared as described previously (Mobley et al., 1986). NGF was labeled with ¹²⁵I (Amersham, Arlington Heights, IL) using lactoperoxidase, as modified from Vale and Shooter (1985). Iodinated protein was separated from free ¹²⁵I on a PD-10 column (Pharmacia, Uppsala, Sweden) preequilibrated with binding buffer [PBS, pH 7.4, containing 1 mg/ml glucose and 1 mg/ml bovine serum albumin (BSA)]. Final specific activity was 25–100 cpm/pg. Radioactivity was quantified using a Beckman 2000 gamma counter.

1088 is a rabbit antibody against the C terminus of human TrkA. It was purified using protein A–Sepharose (Pierce) and has been characterized previously (Zhou et al., 1995). Sc11 is another rabbit antibody to the C terminus of human TrkA (Santa Cruz Biotechnology, Santa Cruz, CA). Both antibodies recognize full-length receptors whose kinase domains can be activated. RTA is a rabbit antibody against the extracellular domain of rat TrkA (Clary et al., 1994; Lucidi-Phillipi et al., 1996). X22 is a mouse monoclonal antibody to the clathrin heavy chain (Brodsky, 1985). AP.6 is a mouse monoclonal antibody to the α-adaptin 100 kDa subunits (Chin et al., 1989). GM10 is a mouse monoclonal antibody that stains lysosomes (Grimaldi et al., 1987; Grady et al., 1995). The antibody to PLC-γ1 was a mixed monoclonal antibody; 4G10 is a mouse monoclonal antibody to phosphorylated tyrosine (both from UBI, Lake Placid, NY). ¹²⁵I-labeled goat anti-mouse was prepared using Na-¹²⁵I (Amersham), iodobeads (Pierce), and goat anti-mouse (Pierce) according to the manufacturer’s instructions, and desalted on a PD-10 column (Pharmacia Biotech, Piscataway, NJ).

NGF binding and cross-linking to surface receptors.PC12 cells (a gift of Lloyd A. Greene, Columbia University) were grown on collagen-coated plates in RPMI 1640 with 10% horse serum and 5% fetal calf serum (both from HyClone Laboratories, Logan, UT). Cells were washed and harvested in warm PBS (without Ca²⁺ and Mg²⁺) and resuspended in cold (4°C) binding buffer. In all experiments, equal amounts of cells were aliquoted for each condition. Cells were incubated with [¹²⁵I]NGF (1 nm = 26.5 ng/ml) at 4°C for 1 or 2 hr, warmed to 37°C in the presence of [¹²⁵I]NGF for 5, 10, or 30 min, and then rapidly chilled (4°C). For cross-linking [¹²⁵I]NGF to surface receptors, the membrane-impermeant cross-linking reagent BS³ was added at a final concentration of 0.8 mm, while rotating at 4°C for 30 min. To correct for nonspecific binding and cross-linking, unlabeled NGF (1 μm = 26.5 μg/ml) was included during binding and cross-linking. The reactions were quenched with 1 mm lysine for 10 min. Cells were pelleted and then lysed for 20 min on ice in 1 ml of lysis buffer #1 (20 mm Tris, pH 8.0, 150 mm NaCl, 1% NP-40, 10% glycerol, and proteinase inhibitors: 1 mm PMSF, 10 μg/ml benzamidine, 1 μg/mlO-phenantholine, and 0.1 μg/ml each pepstatin, chymostatin, leupeptin, and aprotinin). After centrifuging at 16,000 × g for 30 min, the supernatant was removed and assayed for protein (BCA Assay, Pierce). Samples were normalized for protein and immunoprecipitated with 1088 (12 μg/ml), rotating overnight at 4°C. Protein A–Sepharose beads (Pierce), 120 μl of a 20% solution per ml of lysate, were added and incubated at 4°C for 2 hr. After washing two times in lysis buffer #1 and once in H₂O, 50 μl of 7 m urea SDS-PAGE sample buffer (125 mm Tris, pH 6.95, 7 m urea, 2% SDS, 1 mm EDTA, 0.1% Bromphenol Blue) with 100 mm DTT was added, and the sample was heated to 65°C for 15 min. Samples were loaded onto 8–12% SDS-PAGE. Fixed gels (10% acetic acid, 10% isopropanol, 20% methanol) were dried and exposed to the PhosphorImager and then to XAR-5 film (Eastman Kodak, Rochester, NY).

Cell surface biotinylation. Cells were incubated with or without NGF (1 nm) for 30 min at 4°C in PBS with 1 mg/ml glucose, and then NHS-SS-biotin (0.5 mg/ml) was added. The mixture was incubated with gentle rocking for 90 min at 4°C. Cells were pelleted (1000 rpm for 5 min) and then washed three times in cold PBS containing 1 mm lysine. Samples were resuspended in binding buffer, and equal amounts were aliquoted for three different treatments. One sample (designated 100%) was held at 4°C. Another (bkg = background) was treated at 4°C with 50 mm reduced glutathione in 50 mm Tris, pH 8.6, 100 mm NaCl, 1 mg/ml glucose, and 1 mg/ml BSA for 30 min. This treatment was repeated twice. The third sample (int = internalized) was warmed to 37°C for either 10 or 20 min to allow endocytosis, and then treated with glutathione as above. All samples were then incubated 1 hr on ice in 0.2 ml of lysis buffer #2 (20 mm Tris, pH 8.0, 150 mm NaCl, 1% NP-40, 1 mm EDTA) containing 1% BSA and 1 mg/ml iodoacetamide. The lysates were centrifuged 10 min at 10,000 × g. SDS (final concentration 0.5%) was added to the supernatant, and the lysates were boiled 5 min. Lysis buffer #2 (0.8 ml) was then added, and TrkA was immunoprecipitated with 1088 (12 μg). Each lysate was divided into two aliquots. The first was used to detect the amount of biotinylated TrkA. Proteins were separated on nonreducing 7.5% SDS-PAGE, transferred to nitrocellulose (Hoefer Pharmacia Biotech, San Francisco, CA), and blotted with [¹²⁵I]streptavidin (Amersham). Biotinylated TrkA was quantified by PhosphorImager (Molecular Dynamics, Sunnyvale, CA). The second aliquot was submitted to 7.5% SDS-PAGE in a reducing environment, transferred to nitrocellulose, and blotted with 1088 (1:500). The signal was developed using [¹²⁵I]protein A (Amersham) and quantified using the PhosphorImager. The signal for biotinylated TrkA was normalized to the amount of TrkA protein. The TrkA available for internalization was (100% − bkg). TrkA internalized during warming was not susceptible to reduction with glutathione. The amount of TrkA internalized was (int − bkg). The percent of TrkA internalized was (int − bkg)/(100% − bkg) × 100.

Immunostaining and confocal microscopy. Cells were plated on 8-well chamber slides (Nunc, Naperville, IL) that had been coated with Matrigel (Collaborative Biomedical Products) using a 1:200 dilution in PBS (Ca²⁺- and Mg²⁺-free) overnight at 4°C. Wells were washed three times with cold PBS. PC12 cells were plated in DME H-21 medium with 10% horse serum and 5% fetal calf serum 1–2 d before experiments. After aspirating the medium, NGF (2 nm = 53 ng/ml) was added to cells for 30 sec, 2 min, or 60 min in 300 μl of DME H-21 containing 0.5 mg/ml BSA and 10 mm HEPES at 37°C. This medium, without NGF, was added to controls.

Cells were fixed on ice with 1% paraformaldehyde in PBS for 15 min. Cells were permeabilized in PSS (PBS with 10% normal goat serum and saponin at 1 mg/ml) at room temperature for 30 min, changing the solution every 10 min. Primary antibodies were diluted in PSS (sc11 at 1 μg/ml; X22 at 3.1 μg/ml; AP.6 at 8 μg/ml; GM10 at 1:6000) and were incubated with cells overnight at 4°C. After three 10 min washes with PSS, secondary antibodies diluted in PSS (FITC-conjugated goat anti-rabbit IgG at 1:100; Texas Red-conjugated goat anti-mouse immunoglobulins, 1:200; both from Cappel Research Products, Durham, NC) were applied for 45 min at room temperature. After three washes in PBS, coverslips were mounted using Vectashield mounting solution. No staining was evident when primary antibodies were excluded. For Sc11, preliminary incubation (overnight, 4°C) with the peptide immunogen (10 μg/ml) eliminated staining.

Cells were observed with an MRC 1000 Laser Scanning Confocal Microscope (Bio-Rad, Hercules, CA) equipped with a krypton/argon laser and attached to a Zeiss Axiovert microscope. A Zeiss Neofluor ×100 oil-immersion objective with a numerical aperture of 1.3 (0.17°) was used, and images were collected using an aperture of 3–4 mm and a zoom of 2–3. Typically, 10–20 optical sections were taken at 0.5 μm intervals through the cells. The resolution of the confocal microscope in the x–y-axis was 170–200 nm, and in thez-axis was 230–400 nm. Images of 768 × 520 pixels were obtained. Images were processed using Adobe Photoshop 3.0 (Adobe Systems, Mountain View, CA) and printed using a Techtronix Printer. In experiments in which markers were colocalized, colocalization was confirmed by examining individual organelles at higher magnification. In addition, we ensured that colocalization was eliminated by merging overlaying but noncoincident sections. The images shown correspond to optical sections through cells at mid-height.

To count TrkA-positive vesicles near the surface of cells, images of individual optical sections were examined. The edge of the cell was defined as the outermost limit of staining for the clathrin heavy chain, and this was marked with a line. A second line was drawn 0.5 μm interior to the first, and all bright, punctate TrkA-positive vesicles between the lines were counted. The number of these vesicles was then divided by the perimeter of the surface of the cell to yield a value for the number per micrometer cell perimeter.

Cell fractionation. Figure 4 depicts the cell fractionation strategy. Cells grown and harvested as above were first incubated with or without NGF (1 nm) in binding buffer at 4°C for 1 hr. They were then either washed briefly in binding buffer, or not washed and warmed in binding buffer for 10 min at 37°C. Cells were then chilled (4°C) and washed in PBS with 1 mm EDTA and 1 mm EGTA, and then in a cytoplasm-like buffer (buffer B; 38 mm each of the potassium salts of aspartic, gluconic, and glutamic acids, 20 mm MOPS, 5 mm reduced glutathione, 10 mm potassium bicarbonate, 0.5 mm magnesium carbonate, 1 mm EGTA, 1 mm EDTA, pH-adjusted to 7.1 at 37°C with potassium hydroxide). Cells were resuspended in 0.5 ml of buffer B containing proteinase inhibitors (as for lysis buffer #1). Sodium orthovanadate (1 mm) was included to inhibit phosphatase activity for the experiments shown in Figures 8, 9, 10. To permeabilize cells, we used a ball homogenizer obtained from the European Molecular Biology Laboratory (Heidelberg, Germany) and tungsten carbide grade-25 balls obtained from Industrial Tectonics (Ann Arbor, MI). The cell suspension was passed through the homogenizer (8.020 mm cylinder with an 8.0186 mm ball). More than 98% of the cells stained with trypan blue after this procedure. By centrifuging at 1000 × g for 10 min, cell ghosts (P1) were pelleted, thus separating them from the cytosol and the released vesicles. Membranes in the supernatant were isolated using two different protocols. In the first, they were layered over a 0.4 ml pad of 10% sucrose in buffer B (with inclusions, as above) and then centrifuged at 100,000 × g for 1 hr in a Beckman Ti SW50.1 rotor. The membrane pellet was referred to as P2′, and the supernatant (S2′) was cytosol. In the second protocol, the 1000 ×g supernatant was centrifuged at 8000 × gfor 35 min in a TiSW50.1 rotor. The pellet (P2) contained large organelles. The supernatant (S2) was layered over a 0.4 ml pad of 10% sucrose in buffer B (with inclusions, as above) and centrifuged at 100,000 × g for 1 hr to produce the pellet, P3. S3 was the cytosol.

To determine whether fragments of plasma membrane contaminated released intracellular organelles, we biotinylated cell surface proteins and determined whether amyloid precursor protein (APP), a relatively abundant protein (Haass et al., 1992), was detected in P2 and P3. PC12 cells were harvested, and cell surface proteins were biotinylated, as indicated above, for 20 min at 4°C. Cells were washed three times with 1 mm lysine in PBS and then permeabilized and fractionated. P1, P2, and P3 were lysed in lysis buffer #2 and immunoprecipitated with α-C7, an antibody to the C terminus of APP (Podlisny et al., 1991). Immunoprecipitates were submitted to 7.5% SDS-PAGE and blotted. [¹²⁵I]streptavidin was used to probe the blots, and the PhosphorImager was used for detection.

Electron microscopy of cell fractionations. PC12 cells treated with NGF, as just described, were permeabilized and fractionated. The P1, P2, and P3 pellets were fixed in 2% glutaraldehyde in 100 mm sodium cacodylate buffer, pH 7.4, for 90 min at room temperature. Each pellet was washed sequentially in the sodium cacodylate buffer, pH 7.4, and in 50 mm veronyl acetate buffer, pH 7.4, before fixation in 1% osmium tetroxide in the veronyl acetate buffer at 0°C for 45 min. After a final wash in the same veronyl acetate buffer, the pellets were dehydrated and embedded in Epon–Araldite. Thin sections were cut and examined in a Zeiss EM 10CA microscope. Micrographs were taken at ∼20,000×. Several areas, each 20.5 μm², were chosen for vesicle measurements.

Fractionation of internalized NGF. To quantify [¹²⁵I]NGF in intracellular organelles, cells incubated with [¹²⁵I]NGF (1 nm) for 1 hr were washed, warmed at 37°C for 10 min, chilled, permeabilized, and fractionated. Fractions were assayed as above for [¹²⁵I]NGF. In some cases, we determined the amount of [¹²⁵I]NGF associated with the cytoskeleton, which was defined as the NP-40-insoluble pellet of P1 (Vale and Shooter, 1985). For these studies, P1 was resuspended in 0.45 ml of PBS, 1 mm EGTA, 1 mm EDTA containing the protease inhibitors listed above. NP-40 (1%) was then added, and the suspension was incubated on ice for 1 hr before centrifuging for 10 min at 10,000 × g in a microcentrifuge. In some experiments, acid washing was used to measure surface-bound [¹²⁵I]NGF (Bernd and Greene, 1984). To examine the distribution of [¹²⁵I]NGF in intracellular organelles, P2′ was resuspended in 10% sucrose in buffer B (with inclusions) using a 25 G needle and then applied to a 10–40% (w/w) sucrose gradient with a 0.4 ml 50% sucrose pad. The gradient was centrifuged at 100,000 × g for 1 hr in a Beckman Ti SW50.1 rotor. Gradient fractions (4 drops each) were collected from the bottom of the tube. Each fraction was surveyed for radiolabeled NGF using the gamma counter.

NGF cross-linking to intracellular TrkA. Cells incubated with [¹²⁵I]NGF (1 nm) in binding buffer for 1 hr at 4°C were washed and then warmed for 10 min at 37°C before permeabilization, as above. Unlabeled NGF (1 μm) was added to control for nonspecific binding and cross-linking. DSS (2 mm) was added to the permeabilized cells and released membranes, and the mixture was incubated while rotating for 30 min (4°C). The reaction was quenched with lysine (10 mm) for 10 min. P1, P2, and P3 were then prepared, as above. P1 was solubilized by extracting in 1% NP-40 in buffer B for 1 hr (4°C). After centrifuging at 1000 × g (10 min), SDS was added to bring the final concentration to 0.5%. P2 and P3 were resuspended in H₂O with 0.5% SDS. S3 was brought to the same SDS concentration. All samples were boiled for 5 min, chilled (4°C), and then brought to a volume of 1 ml and a final concentration of 0.1% SDS by diluting with immunoprecipitation (IP) buffer (20 mmHEPES, pH 7.4, 0.15 m NaCl, 1% NP-40, 0.5% DOC, 1 mm EDTA). To this was added 1088 (17 μg/ml). After incubating at 4°C overnight, one-tenth of the volume of 20% protein A–Sepharose beads (Pierce) was added for 1 hr, with rotation. The beads were washed twice with IP buffer and once with water before resuspending in 7 m urea SDS-PAGE sample buffer with 20 mm DTT. Samples were heated to 65°C for 10 min and run on a 5–12% SDS-PAGE. Dried gels were exposed to XAR film for 1–3 weeks.

Immunoprecipitation and blotting. Immunoprecipitation of cell fractions was performed by dissolving P1, P2, P3, or S3 in 1 ml of lysis buffer #1 with 1 mm Na-orthovanadate. To this was added 12 μg of 1088, 12 μg of RTA, or 5 μg of anti-PLCγ1. The mixture was incubated overnight at 4°C, and one-tenth of the volume of protein A– or protein A/G–Sepharose beads was added for 2 hr at 4°C. Sepharose beads were washed three times in lysis buffer #1 and once with water, then treated with 50 μl of 7 m urea sample buffer and heated (55°C, 15–30 min). Samples were loaded on 7.5% SDS-PAGE. After transferring to nitrocellulose, blots were probed with anti-phosphotyrosine (4G10), RTA, or anti-PLCγ1. Immune complexes were detected with horseradish peroxidase-conjugated anti-mouse IgG or anti-rabbit IgG and chemiluminescence (Amersham) or with ¹²⁵I-labeled goat anti-mouse IgG. Data were quantified from multiple chemiluminescent exposures using National Institutes of Health Image or using a Molecular Dynamics PhosphorImager.

RESULTS

DISCUSSION

NGF signaling must be communicated from the target of responsive neurons to their cell bodies. Our studies were aimed at exploring the hypothesis that endocytosed activated TrkA receptors serve as the retrograde message (Grimes et al., 1993). We discovered that (1) NGF induced rapid internalization of TrkA in PC12 cells, (2) NGF and TrkA were both found in intracellular vesicles, (3) NGF was bound to TrkA in these vesicles, and (4) TrkA receptors in vesicles were activated, as assessed by tyrosine phosphorylation and association with PLC-γ1. Our findings raise the possibility that it is through the creation of signaling endosomes containing activated TrkA that NGF signals retrogradely to regulate neuronal survival and differentiation.

Target-derived NGF is critical for the normal survival and differentiation of several populations of PNS and CNS neurons (Levi-Montalcini, 1987; Longo et al., 1993; Crowley et al., 1994; Li et al., 1995). Given that NGF gene expression is localized to target tissues (Longo et al., 1993), a mechanism must exist to carry the NGF signal retrogradely from the processes of neurons to their cell bodies. Studies characterizing retrograde NGF signaling showed that although NGF and the signal were similar with respect to both the time course for retrograde transport and the requirement for microtubules (Hendry et al., 1974b; Paravicini et al., 1975; Hendry and Bonyhady, 1980), NGF itself was not the signal (Heumann et al., 1984). One possibility for the NGF signal is an intracellular signaling intermediate that is distinct from the receptor. Another is an activated NGF receptor or NGF–NGF receptor complex that continues to signal after endocytosis. Whether p75^NTR signals, and if so, whether it could serve as a retrograde signal for NGF, is an active area of investigation (Bothwell, 1996; Carter et al., 1996). Significantly, p75^NTR is retrogradely transported in NGF-responsive CNS and PNS neurons (Taniuchi and Johnson, 1985; Johnson et al., 1987;Raivich et al., 1991; Kiss et al., 1993). However, there is no evidence that NGF induces p75 internalization and endocytosis. Indeed, in earlier studies using PC12 cells, there was little change in the amount of p75^NTR at the surface of cells incubated with NGF for up to 5 hr at 37°C (Hosang and Shooter, 1987). Furthermore, Curtis et al. (1995) have shown recently that disrupting p75^NTRfunction had little effect on the retrograde transport of NGF in sensory and sympathetic neurons. Thus, current data provide little support that the NGF retrograde signal is carried by p75^NTR.

Internalized, activated TrkA is an attractive candidate for the NGF retrograde signal. Ehlers et al. (1995) have shown recently that NGF induced an increased accumulation of tyrosine-phosphorylated TrkA distal to a ligature on the sciatic nerve. The studies reported herein suggest that NGF-mediated induction of rapid, extensive endocytosis of TrkA in the distal processes of DRG neurons was responsible for increased retrograde transport of activated TrkA. Using two different methods to assess the disposition of surface receptors on PC12 cells, the internalization of TrkA was increased significantly after NGF addition; indeed, surface-biotinylated TrkA was decreased by >60% after 20 min. These data are consistent with earlier studies showing that NGF downregulated surface TrkA receptors (Hosang and Shooter, 1987; Zhou et al., 1995) and extends them by demonstrating that endocytosed receptors are intact, at least at the treatment times assayed. These biochemical observations were complemented by confocal microscopy studies that showed an increase in TrkA in bright, punctate structures near the cell surface after NGF addition. These organelles were evident soon after NGF treatment and persisted through 60 min. Using Sc11, an antibody to the C terminus of TrkA, there was comparatively little TrkA staining at the surface of cells. However, we know that TrkA is present at the cell surface because of our biotinylation and cross-linking studies (Zhou et al., 1995). Also, we were able to stain the surface of live cells with RTA, a TrkA extracellular domain antibody (Clary et al., 1994) (D. Hall and W. Mobley, unpublished observations), which suggests that certain epitopes are more easily detected than others at the cell surface. In some TrkA-positive organelles, staining with antibodies to TrkA colocalized with staining for the clathrin heavy chain and for α-adaptin. Colocalization of TrkA with these markers indicates that TrkA internalization is mediated, at least in part, through clathrin-coated pit-mediated endocytosis. Many of the TrkA-positive organelles of the same size and distribution that failed to stain with antibodies to clathrin and α-adaptin may also have been derived from this pathway. Taken together, our findings suggest that activation of TrkA enhances recruitment of the receptor into clathrin-coated pits. In this respect, TrkA may behave as do other receptor tyrosine kinases (Lamaze and Schmid, 1995).

Surface downregulation targets other receptor tyrosine kinases to lysosomes and is believed to serve an important role in regulating signaling (van der Geer et al., 1994). There was evidence in our studies that endocytosed TrkA was also targeted to lysosomes. In confocal microscopy, we noted marked redistribution of TrkA to the juxtanuclear region in NGF-treated cells. By 60 min, staining for TrkA at this site was quite intense. Using confocal microscopy, some juxtanuclear TrkA staining was colocalized with GM10, a lysosomal marker. This finding suggests that these TrkA molecules were destined for degradation, a view consistent with earlier studies in which NGF treatment for 60 min markedly decreased total cellular TrkA levels (Zhou et al., 1995). The significance of the perinuclear TrkA staining not present in lysosomes is uncertain. It is likely these receptors are also destined for degradation. However, earlier studies suggest an additional possibility. Using EM, Schwab (1977) detected an NGF–HRP conjugate in nonlysosomal smooth vesicles in the perinuclear cytoplasm of adrenergic neurons. Bernd and Greene (1983), using EM autoradiography to detect labeled NGF, found grains well above background at the nuclear membrane. It is possible, although not proven, that the TrkA-immunopositive organelles present in the juxtanuclear region also contained NGF. If so, they would be ideally positioned to initiate NGF signaling leading to changes in gene expression.

It is possible that activated TrkA receptors internalized at the tips of axons would be targeted to lysosomes. However, if activated TrkA is the retrograde NGF signal, it must avoid degradation in the axon. Current evidence suggests that it would. Studies on the endosomal–lysosomal pathway in neurons (Parton et al., 1992;Hollenbeck, 1993; Parton and Dotti, 1993; Nixon and Cataldo, 1995) suggest that late endosomes and lysosomes are located predominantly in the cell body and proximal dendrites. Indeed, there was no evidence for these organelles in the axons or presynaptic terminals of cultured hippocampal neurons (Parton et al., 1992). Further evidence to suggest that degradative activity in axons is limited is that a minority of endocytic organelles in axons are acidic, and those that are present have an average pH of 5.4, a value consistent with that for late endosomes (Overly et al., 1995). These observations suggest that endocytosed NGF and TrkA would remain intact during retrograde transport. Retrogradely transported NGF was shown to be intact in one study (Claude et al., 1982), and in another, using EM, an NGF–HRP conjugate was found in multivesicular bodies in cell bodies and dendrites but not in axons (Schwab, 1977). Moreover, TrkA retrogradely transported in the sciatic nerve was apparently intact (Ehlers et al., 1995). These data suggest that NGF, and the TrkA receptors endocytosed in response to NGF, resist degradation during retrograde transport.

Cell fractionation studies were used to characterize the organelles that contained NGF and TrkA. A very gentle method of homogenization was chosen in order to avoid contaminating intracellular organelles with bits of plasma membrane (Martin and Walent, 1989; Grimes and Kelly, 1992b). Based on data for the kinetics of NGF and TrkA internalization and degradation in PC12 cells (Bernd and Greene, 1983, 1984; Layer and Shooter, 1983; Hosang and Shooter, 1987; Eveleth and Bradshaw, 1988;Zhou et al., 1995), we examined organelles produced after a brief period of internalization (10 min) so as to restrict our attention to primary endocytic vesicles, endosomes, and the vesicles derived from them. NGF and TrkA were both present in the large and small vesicle fractions. Indeed, using a membrane-permeable cross-linking reagent, we found that NGF was bound to TrkA in these fractions. Given the need for receptor dimerization to induce TrkA activation (Jing et al., 1992), it is possible that persistent NGF binding to TrkA may be required to maintain TrkA kinase activation. In recent studies using 3T3 cells expressing TrkA, we have found that most NGF bound to surface receptors at pH 7.4 remains bound at pH 5.5 (J. Zhou and W. Mobley, unpublished observations), i.e., near the average pH for acidified endocytic organelles in axons (Overly et al., 1995). Thus, some NGF could remain bound to TrkA during retrograde transport. It will be important to define further the organelles that contain NGF and TrkA, those in which NGF is bound to TrkA, and those that carry activated TrkA retrogradely. In earlier studies of NGF retrograde transport in axons, NGF was found in smooth-walled tubules (45–60 nm) and in clear vesicles ranging in diameter from 50 to 150 nm (Schwab, 1977; Claude et al., 1982). Recent data (M. Grimes, E. Beattie, and W. Mobley, unpublished observations) in which in vitro reactions were used to characterize organelles that emerged from permeabilized cells suggest that P2 contains clathrin-coated vesicles and that P3 contains uncoated primary endocytic vesicles as well as transport vesicles (Grimes and Kelly, 1992b). We do not know which organelle(s) carries the NGF signal, but small transport vesicles are an attractive possibility (Grimes et al., 1993).

NGF binding to TrkA in intracellular vesicles suggested that activation of TrkA receptors persisted after endocytosis. Evidence for this was the presence of tyrosine-phosphorylated TrkA in both P2 and P3 after NGF treatment. NGF markedly increased the specific activity of tyrosine-phosphorylated TrkA in all cell fractions. TrkA signaling is communicated through the activation of signaling intermediates, including PLC-γ1. Immunoprecipitation of TrkA in intracellular organelles of NGF-treated cells showed that activated TrkA formed a complex with tyrosine-phosphorylated PLC-γ1. In preliminary studies, we have shown that tyrosine-phosphorylated SHC is also associated with activated TrkA in intracellular organelles (J. Zhou and W. Mobley, unpublished observations). These data link activated intracellular TrkA to important signaling cascades (Stephens et al., 1994) and thereby suggest strongly that internalized activated TrkA receptors are capable of signaling. Together with the data of Ehlers et al. (1995), our observations support the hypothesis that through endocytosis of activated TrkA, NGF creates signaling endosomes that convey its retrograde signal. To test this idea, studies must be done to determine whether activated TrkA in endocytic vesicles can initiate NGF signal transduction in the neuron cell body.