Evidence for a Critical Role of the Tumor Necrosis Factor (cid:1) Convertase (TACE) in Ectodomain Shedding of the p75 Neurotrophin Receptor (p75 NTR )*

Protein ectodomain shedding, the proteolytic release of the extracellullar domain of membrane-tethered proteins, can dramatically affect the function of cell surface receptors, growth factors, cytokines, and other proteins. In this study, we evaluated the activities involved in ectodomain shedding of p75 NTR , a neurotrophin receptor with critical roles in neuronal differentiation and survival. p75 NTR is shed in a variety of cell types, including dorsal root ganglia cells and PC12 cells. In Chinese hamster ovary cells, inhibitors

Protein ectodomain shedding is emerging as an important post-translational mechanism for regulating the function of membrane-anchored proteins. Shedding involves the proteolytic processing of membrane-tethered proteins leading to the release of their extracellular-or ectodomain (reviewed in Refs. [1][2][3][4]. About 2-4% of the proteins on the cell surface are released by metalloproteases in response to phorbol ester stimulation (5). These molecules comprise a variety of structurally and functionally distinct proteins, including epidermal growth factor receptor ligands such as TGF␣ 1 and HB-EGF, TNF family members, and other cytokines, receptors such as the p55 and p75 TNF receptors and the interleukin-6 receptor, and several other proteins, including the amyloid precursor protein, Notch, and Delta (see Ref. 3 and references therein). Ectodomain shedding has been shown to be essential for proper signaling via epidermal growth factor receptor ligands (6 -15), Notch-mediated lateral inhibition (16 -23), limiting TNFR-mediated inflammatory reactions (24), and regulating axonal guidance (25)(26)(27). Even if the functional consequences of ectodomain release remain to be evaluated for most shed proteins, it seems likely that ectodomain shedding will affect the function of the majority of its substrate proteins.
Several intracellular signaling pathways have been implicated in mediating constitutive and stimulated shedding. Among these, the MEK/ERK MAP kinase pathway has been shown to mediate the receptor tyrosine kinase-and phorbol ester-stimulated shedding of the TACE substrates TGF␣ and TNF␣ (42), and of HB-EGF (43), syndecan-1 and -4 (41), and the L1 adhesion molecule (44). In contrast, the constitutive shedding of TGF␣ depends on the p38 MAP kinase pathway (42).
In this study, we evaluated the proteolytic and signaling activities involved in ectodomain shedding of the p75 neurotrophin receptor (p75 NTR ), which serves as a common receptor for all known neurotrophins. The p75 NTR receptor has diverse functions, including the promotion of cell survival in some cell types, and triggering apoptosis in others (45)(46)(47)(48). The p75 NTR also refines ligand specificity of a second class of neurotrophin receptors, the TRK receptors (49). Furthermore, p75 NTR reportedly regulates myelination by Schwann cells in the developing peripheral nervous system (50). Finally, recent work has demonstrated that p75 NTR binds to the Nogo receptor, thereby inhibiting axon growth mediated by three myelin-associated proteins, MAG, OMgp, and Nogo-A (51,52). One way to regulate p75 NTR levels on the cell surface of Schwann cells was shown to occur via truncation by cellular metalloproteases (53). Whereas the relevance of p75 NTR ectodomain shedding in vivo is unclear, it has been proposed that shedding regulates the amount of cell surface p75 NTR and may therefore affect axonal outgrowth (54). Furthermore, recent studies demonstrate that p75 NTR undergoes regulated intramembrane proteolysis (RIP), which may in turn affect the association with TRK receptors and regulate the function of p75 NTR (55,56). Like in other well characterized cases of RIP, such as in the cleavage of Notch, amyloid precursor protein, or sterol regulatory element-binding protein (57), a membrane proximal cleavage is considered a prerequisite for the subsequent intramembrane cleavage of p75 NTR (55). These studies highlight the significance of identifying the enzyme responsible for membrane proximal cleavage and ectodomain shedding of p75 NTR . Our results uncover two distinct activities involved in the release of p75 NTR ectodomain, and define characteristic features of these two shedding activities. Furthermore, we evaluate the role of different candidate sheddases of the ADAM (a disintegrin and metalloprotease) family of metalloproteases in p75 NTR ectodomain shedding.

EXPERIMENTAL PROCEDURES
Materials-Phorbol 12-myristate 13-acetate (PMA), hydrogen peroxide, and sodium vanadate were purchased from Sigma. Pervanadate was generated immediately prior to each experiment by mixing sodium vanadate and hydrogen peroxide to a final concentration of 100 mM each. Protein A-and Protein G-Sepharose were purchased from Pharmacia. Restriction enzymes, Hi-Fidelity Taq, and Klenow fragment were purchased from Roche Diagnostics. Protein kinase C inhibitor bisindolylmaleimide I, MEK inhibitor U0126, p38 MAP kinase inhibitor SB202190, compound E, lactacysteine, and EndoH and PNGaseF were purchased from Calbiochem. BB-94 metalloprotease inhibitor was kindly provided by Dr. J. D. Becherer (GlaxoSmithKline, Research Triangle Park, NC). Active recombinant human TIMP-1 and -2 were purified as described (58).
Constructs and Antibodies-The cDNA encoding for the rat p75 NTR cDNA (59) was placed into pcDNA3 using EcoRI and XbaI restriction sites. The expression construct encoding full-length rat p75 NTR with the NH 2 -terminal signal peptide followed by the Strep-tag II and hemag-glutinin epitope tag (St75) was generated as previously described (56). Site-directed mutagenesis was performed with untagged wild-type p75 NTR or St75 cDNA as a template and deletions were introduced with PCR and appropriate oligonucleotides using the QuikChange system (Stratagene). All of the resulting mutations were confirmed by cDNA sequencing. The anti-rat p75 NTR polyclonal antibody (anti-REX) and monoclonal antibodies (mc192), and anti-TACE cytoplasmic domain polyclonal antibodies have been previously described (60 -63). The p75 NTR antibodies recognize the extracellular portion of the rat p75 NTR receptor protein (60,61). Affinity capture of secreted St75 ectodomain and Western blot analysis were performed as described (56).
All cells were transfected in 6-well tissue culture plates with Lipo-fectAMINE (Invitrogen) following the manufacturer's recommendations. Cells were allowed to recover for 6 to 18 h prior to metabolic labeling. For metabolic pulse labeling, cells were washed twice in Dulbecco's modified Eagle's medium lacking cysteine and methionine and then incubated in the same media supplemented with 10% dialyzed fetal calf serum for half an hour. 200 Ci/ml of [ 35 S]Pro-mix (Amersham Biosciences) was then added for 30 min. For overnight labeling experiments, the 30-min starvation period was omitted and cells were labeled with 200 Ci/ml of [ 35 S]Pro-mix for 10 -14 h. In all cases, after labeling, the cells were washed twice with phosphate-buffered saline, pH 7.4, and immediately lysed in cell lysis buffer (TBS, pH 7.4, 1% (v/v) Nonidet P-40, 10 mM NaF, 1 mM Na 3 VO 4 , and protease inhibitors 1 mM 1,10phenanthroline, 2 g/ml leupeptin, 0.4 mM benzamidine, 10 g/ml soybean trypsin inhibitor, 0.5 mM iodoacetamide), or incubated in Opti-MEM (Invitrogen) for the indicated number of hours. Conditioned media was collected after 1 h and cell debris was removed by centrifugation at 13,000 ϫ g for 20 min followed by a further 100,000 ϫ g spin for 45 min. p75 NTR protein was then immunoprecipitated from the media and lysates with anti-REX IgG. p75 NTR protein from dorsal root ganglia cultures was immunoprecipitated with mc192 IgG. In all cases, immunoprecipitations were performed at 4°C for at least 6 hours followed by washing, and samples were then analyzed by SDS-PAGE and autoradiography as described below.
In experiments in which PMA, pervanadate, or BB-94 were added, cells were first chased for 2 h followed by an additional chase of 15 min or 1 h in fresh Opti-MEM containing the indicated concentration of additive(s). In experiments examining the effect of bisindolylmaleimide I or of the MAP kinase inhibitors U0126 and SB202190, the indicated concentration of inhibitor was present both during the initial 2-h chase and during the following 1-h chase. Supernatants were then collected and cells lysed as before.
For immunoprecipitations, the appropriate antibody and Protein A-Sepharose were added to the lysate or supernatant and incubated at 4°C from 2 h to overnight. The beads were then washed three times with lysis buffer (lysates) or PBS, pH 7.4, 0.05% (v/v) Nonidet P-40 (supernatant). 2ϫ sample loading buffer supplemented with 10 mM dithiothreitol was added and the samples were incubated at 95°C for 5 min prior to SDS-PAGE analysis. Gels were fixed in 50% (v/v) methanol, 10% (v/v) acetic acid for 15 min, rehydrated in water for 15 min, dried, and exposed to Kodak Bio-MAX MR film. Quantification was performed using a Fuji BAS2500 BioImaging analyzer.
Cell surface biotinylation of p75 was carried out as described (66, 67) and subsequent shedding was performed at 37°C for 15 min. Deglycosylation experiments were carried out using endoglycosidase H (En-doH) and peptide N-glycosidase F as described (66,67).
Determination of the Cleavage Site for TACE in Peptides Corresponding to the Juxtamembrane Domain of p75 NTR -Two overlapping peptides corresponding to the juxtamembrane region of p75 NTR were synthesized by Synpep (Dublin, CA). Both peptides contained a dinitrophenyl group at the their NH 2 terminus to facilitate detection at 350 nM (peptide 1, dinitrophenyl-VTTVMGSSQPVVTRGT-CONH 2 ; peptide 2, dinitrophenyl-VVTRGTTDNLIP-CONH 2 ). Peptide cleavage assays were run as previously described (68). Briefly, peptides were incubated with the catalytically active purified recombinant extracellular domain of TACE (catalytic domain, disintegrin domain, and cysteine-rich region) and the cleavage products were analyzed by liquid chromatography matrix-assisted laser desorption ionization time-offlight mass spectrometry. Only peptide 1 was cleaved by TACE under these conditions in vitro, between residues QP and VV (data not shown).

RESULTS
To evaluate the shedding of the p75 neurotrophin receptor (p75 NTR ), we assayed for the release of its 35 S-labeled ectodomain into the tissue culture supernatant. Initially, we used wild-type CHO cells and the previously described M2 mutant CHO cells, which are defective in the metalloprotease-mediated shedding of several transmembrane proteins (5). The mutation in the M2 cells inactivates the TNF␣ convertase (TACE, ADAM17) 2 and abolishes the stimulated shedding of all TACE substrates examined to date (69,70). Cells transiently transfected with either a control vector or a plasmid encoding for rat p75 NTR were pulse-labeled. Labeled proteins were recovered from the media and cell lysates immediately after the labeling period or after a 6-h chase (Fig. 1A). p75 NTR is initially synthesized as a protein with a molecular mass of ϳ70 kDa (lanes 2 and 5) that is completely converted to the 75-kDa mature form after the 6-h chase period (lanes 3 and 6). This shift is likely because of differences in glycosylation (see below), and is not affected in the M2 mutant cells. After the 6-h chase period, a significant fraction of the labeled p75 NTR was recovered as a 50-kDa ectodomain in the culture supernatant of wild-type CHO cells (lane 7), whereas no p75 NTR ectodomain was detected in the media of the M2 mutant cells (lane 8).
The shedding of numerous transmembrane proteins, including p75 NTR (54,55), is mediated by metalloproteases. To determine whether p75 NTR shedding involves metalloproteases in CHO cells, constitutive shedding was examined in the presence or absence of the hydroxamate-based metalloprotease inhibitor, BB-94 (batimastat, Fig. 1B). The constitutive release of the p75 NTR ectodomain was severely diminished in the presence of inhibitor (Fig. 1B, top panel), with a concomitant increase in the amount of labeled full-length protein recovered in the lysate (Fig. 1B, lower panel). Thus, similar to the behavior of p75 NTR in Schwann cells (54,55), the constitutive shedding of p75 NTR is apparently dependent upon the action of one or more metalloproteases in CHO cells. Because BB-94 does not completely block p75 NTR release ( Fig. 2A), it is also possible that a small amount of shedding is because of activities that are not metalloprotease-dependent. Alternatively, it is also possible that the relevant metalloprotease activities cannot be completely inhibited by BB-94, for example, if some shedding originates from a location in the secretory pathway that is not completely accessible to exogenously added BB-94.
Next, we examined the effect of either the phorbol ester PMA or pervanadate, a tyrosine phosphatase inhibitor, on the shedding of p75 NTR ( Fig. 2A). Both stimuli have been shown to  1 and 2) and a rat pheochromocytoma cell line (PC12, lanes 3 and 4). Both cell types were metabolically labeled and stimulated as described above. Endogenous p75 NTR was immunoprecipitated from supernatants incubated for 1 h in the presence (lanes 2 and 4) or absence (lanes 1 and 3) of 25 ng/ml PMA. up-regulate the metalloprotease-mediated shedding of several proteins (3,40,71). In wild-type CHO cells, both PMA and pervanadate are able to increase the release of the p75 NTR ectodomain ( Fig. 2A, top panel). Furthermore, the activated shedding events are also inhibited by BB-94. The effect of pervanadate on shedding is specific, as neither hydrogen peroxide nor vanadate enhances shedding (Fig. 2B). The stimulated shedding of numerous proteins, including several TACE substrates, in response to agents such as phorbol esters is absent in the M2 CHO cell line (5,10,70,72). We therefore assessed the ability of this cell line to shed p75 NTR when stimulated. M2 cells failed to release significant amounts of p75 NTR ectodomain either constitutively or in response to either PMA or pervanadate ( Fig. 2A, bottom panel). p75 NTR is also shed in a constitutive and stimulated manner from several other cells, including transiently transfected COS-7 cells (see below), as well as two non-transfected cell types that endogenously express this receptor, cultured primary dissociated mouse dorsal root ganglia cells and the PC12 rat pheochromocytoma cell line (Fig. 2C).
Numerous signaling pathways have been implicated in metalloprotease-mediated ectodomain shedding. We used pharmacological inhibitors to assess the role of protein kinase C and the MEK/ERK and p38 MAP kinase pathways in the release of p75 NTR ectodomains. The protein kinase C inhibitor bisindolylmaleimide I blocks the PMA-stimulated shedding of p75 NTR , but does not affect either the constitutive or pervanadatestimulated release of these proteins (Fig. 3A). The constitutive shedding of p75 NTR is inhibited by SB202190, an inhibitor of the p38 MAP kinase pathway, but is not affected by inhibition of MEK 1 and 2 by U0126 (Fig. 3B). Both U0126 and SB202190 partially reduced the release of the p75 NTR ectodomain in cells treated with PMA, the latter likely by inhibiting the underlying constitutive shedding pathway. However, even in the presence of both inhibitors, the shedding of the p75 NTR ectodomain is still increased by PMA compared with unstimulated shedding levels, indicating that an additional shedding activity not dependent on the p38 or MEK/ERK MAP kinase pathways is activated. In contrast, the increase in p75 NTR shedding seen in response to pervanadate treatment, compared with constitutive shedding, is completely abrogated by the MEK 1 and 2 inhibitor, U0126. Furthermore, pervanadate does not enhance the SB202190-sensitive shedding activity above what is seen in unstimulated cells.
To further characterize the proteolytic activities involved in p75 NTR shedding, we assessed their sensitivity to tissue inhibitors of metalloproteases TIMP-1 and TIMP-2 (Fig. 3C). Neither inhibitor was able to significantly affect either the constitutive or stimulated shedding of p75 NTR . In contrast, in parallel experiments, TIMP-2 was found to effectively inhibit pervanadate-stimulated TRANCE shedding (40). These results indicate that matrix metalloproteases are unlikely to be involved in this process, as most matrix metalloproteases are inhibited by one or both of these TIMPs. However, we cannot rule out that the sheddases involved are not accessible to the TIMPs.
To address the issue of where in the secretory pathway shedding of p75 NTR occurs, we examined the sensitivity of membrane-anchored and soluble p75 NTR to treatment with En-doH and PNGaseF. N-Linked carbohydrate residues acquire resistance to treatment with EndoH by conversion of high mannose glycans into complex carbohydrates during passage through the medial Golgi apparatus. When membrane-anchored p75 NTR was immunoprecipitated from the cell lysate of transiently transfected COS-7 cells, we found that the 75-kDa form of p75 NTR (marked by an asterisk in Fig. 4A, lanes 1 and  2) was mainly resistant to EndoH treatment, whereas the 70-kDa immature proform of the receptor was mostly or completely EndoH-sensitive (marked by an arrow in Fig. 4A, lanes  1 and 2). Soluble p75 NTR immunoprecipitated from the supernatant was resistant to treatment with EndoH (Fig. 4A, lane 5), as would be expected following passage through the secretory apparatus and release from cells. Membrane anchored as well as soluble p75 NTR could be deglycosylated with PNGaseF, which removes all N-linked carbohydrate residues (Fig. 4A,   FIG. 3. Effect of protein kinase C and MAP kinase inhibitors and of TIMP-1 and TIMP-2 on shedding of p75 NTR . A, effect of the protein kinase C inhibitor bisindolylmaleimide I (bisind I) on constitutive, PMA-, and pervanadate-stimulated p75 NTR ectodomain release. CHO cells transfected with p75 NTR expression vectors were pulse-labeled and chased for 6 h in fresh media alone (lanes 1 and 2), with 25 ng/ml PMA (lanes 3 and 4), or with 100 M pervanadate (lanes 5 and 6). For each condition, 5 M bisindolylmaleimide I was either included (lanes 2, 4, and 6) or omitted (lanes 1, 3, and 5). B, effect of MAP kinase inhibitors on constitutive and stimulated shedding. CHO cells were transfected with p75 NTR expression vector. Cells were metabolically labeled for 30 min and then chased for 2 h in media containing either carrier, 5 M U0126 (a MEK 1 and 2 inhibitor), 10 M SB202190 (a p38 MAP kinase inhibitor), or both inhibitors. Cells were then placed in fresh media containing the same inhibitors with no further additions (-), 25 ng/ml PMA (PMA), or 100 M pervanadate (PV) and chased for an additional hour. Full-length and shed p75 NTR were recovered from the lysate and conditioned supernatant, respectively, by immunoprecipitation and analyzed by SDS-PAGE. The amount of shed protein recovered from the media was quantified using a phosphorimager. Values were normalized to the amount observed in the untreated cells. Values represent the average of 3 to 5 experiments; error bars indicate one standard deviation. C, p75 NTR expressing CHO cells were labeled overnight and chased for 1 h in Opti-MEM alone (lanes 1-3), with 25 ng/ml PMA (lanes 4 -6), or with 100 M pervanadate (lanes 7-9). TIMP-1 (15 nM) or TIMP-2 (18 nM) were present in the chase media as indicated. Soluble ectodomains were isolated from the media by immunoprecipitation and analyzed by SDS-PAGE followed by autoradiography. lanes 3 and 6). Identical results were obtained when cell-associated and shed p75 NTR was immunoprecipitated from PC12 cells and subjected to treatment with EndoH and PNGaseF (data not shown). The EndoH resistance of most or all of the 75-kDa form of p75 NTR suggests that shedding of p75 NTR occurs after passage through the medial Golgi apparatus. Finally, biotinylated p75 NTR can be immunoprecipitated from lysates of cell surface-biotinylated COS-7 cells, which confirms that p75 NTR is present on the cell surface (Fig. 4B, lanes 1-3).
When cell surface-biotinylated p75 NTR expressing COS-7 cells were incubated in tissue culture medium for 15 min at 37°C after cell surface labeling, soluble biotinylated p75 NTR was released (Fig. 4B, lane 4, arrowhead marks position of soluble p75 NTR ). Shedding of cell surface-biotinylated p75 NTR was slightly enhanced by treatment with PMA (Fig. 4B, lane 5), and strongly enhanced by treatment with PV (Fig. 4B, lane 6). The increased shedding of cell surface-labeled p75 NTR following PV stimulation was also reflected in a concomitant decrease in cell-associated p75 NTR in these cells (Fig. 4B, lane 3). Taken together, these results suggest that shedding of p75 NTR occurs at or near the cell surface in transiently transfected COS-7 cells.
The stimulated shedding of several proteins is mediated by the TNF␣-converting enzyme (TACE/ADAM17). To assess the role of ADAM17 as well as other candidate sheddases of the ADAM family in the shedding of p75 NTR , we investigated release of the p75 NTR ectodomain in murine fibroblasts isolated from adam9/12/15Ϫ/Ϫ, adam10Ϫ/Ϫ, or adam17Ϫ/Ϫ mice. When transfected into these cells, p75 NTR is shed constitutively at low levels (Fig. 5). However, whereas stimulation with PMA strongly enhances p75 NTR shedding in wild-type murine fibroblasts and in adam9/12/15Ϫ/Ϫ or adam10Ϫ/Ϫ cells, no PMAdependent stimulation of p75 NTR shedding was observed in adam17Ϫ/Ϫ cells. In separate experiments, we confirmed that pervanadate stimulates p75 NTR shedding in mouse fibroblasts, and that this stimulation is also abolished in adam17Ϫ/Ϫ cells (data not shown). Thus, TACE is essential for both phorbol ester-and pervanadate-stimulated shedding of p75 NTR in murine fibroblasts.
What is the cleavage site that is processed to generate soluble p75 NTR ? This question was initially addressed by in vitro cleavage of peptides corresponding to the juxtamembrane domain of p75 NTR by TACE, and then by assessing how mutagenesis of the membrane-proximal stalk domain of p75 NTR affects its shedding from cells. When two peptides corresponding to partially overlapping fragments of the juxtamembrane domain of p75 NTR were incubated with purified TACE in vitro, one of these peptides was cleaved between residues Pro 234 and Val 235 (data not shown, see "Experimental Procedures" for details). However, when two mutant forms of p75 NTR carrying point mutations at this site (QPVV Ͼ QAAV, QPVV Ͼ QPAV) were expressed in COS-7 cells, no defect in constitutive and stimulated shedding compared with the wild-type receptor was observed (Fig. 6, lanes 19-22). Furthermore, deletion of 8 aa residues including the QPVV sequence (⌬8aa) also had little or no effect on constitutive or stimulated p75 NTR shedding (Fig.  6B, lanes 9 and 10).
One interpretation of these results is that the cleavage site  2 and 4) or peptide N-glycosidase F (PNGaseF, lanes 3 and 5). p75 NTR was immunoprecipitated from cell lysates of transiently transfected COS-7 cells (lanes 1-3), and soluble p75 NTR was immunoprecipitated from the corresponding culture supernatant (lanes 4 -6). Aliquots of the immunoprecipitated material were treated with EndoH (lanes 2 and 4) or PNGaseF (lanes 3 and 5). The migration of the 75-kDa form of p75 NTR (lane 2, asterisk) is largely unaffected by EndoH treatment, whereas the immature 70-kDa form of the receptor (arrow in lane 1) is sensitive to deglycosylation by EndoH ( compare lanes 1 and 2). Because passage through the medial Golgi compartment usually renders N-linked carbohydrate residues resistant to treatment with EndoH, this demonstrates that the 75-kDa form of p75 NTR accumulates in an uncleaved form after passage through the medial Golgi apparatus (i.e. in the trans-Golgi network or on the cell surface). Treatment with PNGaseF removes all N-linked carbohydrate residues from p75 NTR (lanes 3). Shed p75 NTR in the supernatant is resistant to treatment with EndoH (lanes 4 and 5), but sensitive to treatment with PNGaseF (lane 6). Open arrowheads mark the position of untreated, fully glycosylated transmembrane or soluble p75 NTR , and solid arrowheads mark the position of PNGaseF-treated forms of p75 NTR . B, shedding of cell surface-biotinylated p75 NTR . COS-7 cells expressing p75 NTR were biotinylated with a non-membrane-permeable biotinylation reagent (see ''Experimental Procedures'' for details). After quenching the biotinylation reaction, the cells were incubated for 15 min in growth medium with no additions (lanes 1 and 4), with 25 ng/ml PMA (lanes 2 and 5), or with 100 M pervanadate (lanes 3 and 6). Biotinylated p75 NTR was immunoprecipitated from cell lysates (lanes 1-3), and shed p75 NTR was immunoprecipitated from the corresponding supernatants (lanes 4 -6). Arrowheads, membrane-bound p75 NTR in lanes 1-3 and shed p75 NTR ectodomain in lanes 4 -6; asterisk, nonspecific background band in lanes 4 -6.
FIG. 5. Ectodomain shedding of p75 NTR in wild-type or adam9/ 12/15؊/؊, adam17؊/؊, or adam10؊/؊ mouse embryonic fibroblasts. Primary mouse embryonic fibroblasts from wild-type mice, adam9/12/15Ϫ/Ϫ triple knockout mice, adam17Ϫ/Ϫ mice, or immortalized adam10Ϫ/Ϫ embryonic cells (65) were transfected with p75 NTR , metabolically labeled overnight and chased for 1 h in media with carrier (-), 25 ng/ml PMA, or 25 ng/ml PMA in the presence of 2 M BB94 as indicated. Soluble p75 NTR ectodomain was immunoprecipitated from the media and analyzed by SDS-PAGE. Full-length p75 NTR was immunoprecipitated from the corresponding lysates to confirm that the expression levels were comparable in wild-type and adamϪ/Ϫ cells (data not shown).
for p75 NTR peptides in vitro does not match the physiological cleavage site in cells. Alternatively, the cleavage site used to release p75 NTR from cells may not have been present in the two peptide sequences, or may not be efficiently processed by TACE in vitro. To determine whether the cleavage site in cells is amino-terminal or carboxyl-terminal to the QPVV 236 sequence, we compared the migration of the soluble forms of wild-type p75 NTR and the ⌬8aa mutant in culture supernatant, and of the remaining COOH-terminal membrane stubs in cell lysates of transiently transfected COS-7 cells. The samples were separated on 4 -20% SDS-acrylamide gels to facilitate detection of a shift in migration caused by deletion of 8 aa residues. As shown in Fig. 6C (lanes 1-4), the shed form of ⌬8aa p75 NTR (solid arrowhead) migrates faster than the soluble wild-type receptor FIG. 6. Evaluation of the effect of mutations in the juxtamembrane region of p75 NTR on its constitutive and stimulated shedding. A, schematic illustration of specific point mutations and deletions in the membrane-proximal region of p75 NTR . The designation of each mutant is listed on the left and corresponds to the designations in panels B and C. Deleted aa residues are replaced by a hyphen, and the two point mutations are indicated by lowercase bold letters. Amino acid residues that are predicted to be part of the hydrophobic transmembrane domain of p75 NTR are lowercase. The cytoplasmic domain is predicted to begin with the two charged aa residues Lys-Arg (KR). B, effect of mutations on constitutive and stimulated shedding of p75 NTR . COS-7 cells expressing wild-type or mutant forms of p75 NTR were labeled for 1 h, then chased for 1 h in fresh medium with carrier (Ϫ). This was followed by a 1-h chase in fresh medium containing 25 ng/ml PMA. p75 NTR was immunoprecipitated from supernatants of unstimulated and PMA-stimulated cells as indicated, and separated on 10% SDSpolyacrylamide gels. The expression of wild-type and mutant forms of p75 NTR was similar in all cases (data not shown). C, initial mapping of the position of the cleavage site of p75 NTR . Western blot analysis of the culture supernatant (lanes 1-4) and cell lysate (lanes 5-8) of COS-7 cells expressing p75 NTR . Samples were separated on a 4 -20% SDS gel to visualize the effect of deleting 8 aa residues on soluble p75 NTR and the corresponding COOH-terminal membrane stub. The presenilin inhibitor compound E (100 nM) and the proteasome inhibitor lactacysteine (1 M) were included in the tissue culture medium to prevent degradation of the COOH-terminal stubs after shedding of the ectodomain. Shed wild-type p75 NTR (lanes 1 and 3, open arrowhead) migrates slower than the shed ⌬8aa mutant (lanes 2 and 4, solid arrowhead) in the presence of carrier (lanes 1 and 2) or 25 ng/ml PMA (lanes 3 and 4). The COOH-terminal stubs generated by ectodomain shedding of p75 NTR in the presence of carrier (lanes 5 and 6) or 25 ng/ml PMA comigrate. This suggests that p75 NTR is processed within the remaining membrane-proximal sequence of the ⌬8aa mutant (TDNLIPV), even though mutations within this sequence do not abolish constitutive or stimulated p75 NTR shedding.
(open arrowhead), both with and without PMA stimulation. On the other hand, the membrane-anchored COOH-terminal stubs that remain after ectodomain release of wild-type and ⌬8aa p75 NTR co-migrate (Fig. 6C, lanes 5-8, position of COOH-terminal stubs marked by an arrowhead). At least three closely co-migrating COOH-terminal stubs are generated by shedding of wild-type and ⌬8aa p75 NTR , which indicates that more than one cleavage site may be used to release p75 NTR . Taken together, these results suggest that the p75 NTR cleavage site(s) in COS-7 cells are located between the deleted sequence in the ⌬8aa mutant and the transmembrane domain (see Fig. 6A).
This possibility was addressed by generating additional mutant forms of p75 NTR with various overlapping deletions in the very membrane-proximal region of p75 NTR . Immunoprecipitations of p75 NTR from cell lysates of transiently transfected COS-7 cells confirmed that all mutants were expressed at similar levels compared with wild-type p75 NTR in the experiment presented in Fig. 6B (data not shown). When soluble mutant forms of p75 NTR with short deletions (3-4 aa residues) or with longer deletions (up to 11 aa residues) were immunoprecipitated from the culture supernatant of these cells, no defect in constitutive or stimulated shedding was evident (Fig.  6B, lanes 1-4, 7, 8, and 13-16). Only a large deletion encompassing 15 aa residues of the juxtamembrane region of p75 NTR substantially reduced both constitutive and stimulated shedding (Fig. 6B, lanes 5 and 6). Finally, removal of the cytoplasmic domain of p75 NTR did not affect shedding, demonstrating that this domain is not required for regulated or constitutive release of p75 NTR (Fig. 6, lanes 17 and 18). DISCUSSION Protein ectodomain shedding is a ubiquitous process that results in the release of a variety of functionally and structurally distinct proteins from cells. In this study, we evaluated the properties of the sheddases for the p75 NTR , a neurotrophin receptor with diverse roles in the developing nervous system, ranging from axonal growth and retrograde transport to cell survival and apoptosis of different neuronal populations (45)(46)(47)(48). We found that p75 NTR is shed constitutively by a metalloprotease-dependent activity that requires an intact p38 MAP kinase pathway in CHO cells. Furthermore, metalloprotease-dependent p75 NTR shedding can be stimulated by the phorbol ester PMA and the tyrosine phosphatase inhibitor pervanadate. In CHO cells, the phorbol ester-stimulated shedding of p75 NTR requires protein kinase C, whereas the pervanadate-stimulated shedding does not. However, the pervanadate-stimulated shedding is blocked by MEK/ERK pathway inhibitors, which also partially reduce shedding in response to PMA.
The effect of MAP kinase pathway inhibitors on constitutive and activated shedding of p75 NTR suggests that distinct signaling pathways affect these processes. The constitutive shedding of p75 NTR is dependent on signaling via the p38 MAP kinase pathway in CHO cells. In contrast, stimulated shedding apparently acts independently of the p38 MAP kinase, as the absolute decrease in p75 NTR release mediated by the p38 MAP kinase inhibitor is similar in the presence or absence of PMA or pervanadate. Conversely, most of the increase in p75 NTR shedding seen in response to these stimuli can be blocked by inhibiting the MEK/ERK MAP kinase pathway.
These pharmacological inhibitors of ectodomain shedding raised the possibility that the constitutive and stimulated p75 NTR sheddases may be distinct. We decided to further explore this possibility using cells that lack different candidate sheddases of the ADAM family of metalloproteases, and a mutant CHO cell line with a specific defect in the processing and activation of TACE (ADAM17) (70). Wild-type mouse fibro-blasts released p75 NTR constitutively, and this shedding could be strongly enhanced by stimulation with PMA and pervanadate. Mouse fibroblasts lacking ADAMs9, -12, and -15 (40) or ADAM10 (65) did not show any defect in constitutive or stimulated shedding, arguing against an essential role for these ADAMs in p75 NTR shedding in these cells. However, in adam17Ϫ/Ϫ mouse embryonic fibroblasts (6), both PMA-and PV-stimulated p75 NTR shedding was abrogated, whereas a small amount of BB94-insensitive constitutive shedding remained. Thus, whereas TACE is responsible for the PMA-and pervanadate-stimulated release of p75 NTR in primary mouse embryonic fibroblasts and CHO cells, the role of TACE in constitutive shedding is unclear.
There are several plausible models to explain constitutive shedding of p75 NTR in CHO cells. One model entails distinct sheddases responsible for constitutive and stimulated shedding. TACE would represent the MEK/ERK-dependent sheddase, whereas a second BB94-sensitive, TIMP-1-and TIMP-2insensitive protease is responsible for constitutive shedding. The M2 mutant CHO cell line would then necessarily carry a defect that affects both of these sheddases. Alternatively, TACE could be responsible for constitutive and stimulated p75 NTR shedding in CHO cells. This is plausible, because the basal release of two other TACE substrates, TGF␣ and TNF␣, has been shown to depend on p38 MAP kinase activity in CHO cells (42).
To explore the cleavage site requirements for shedding of p75 NTR , several point mutations and deletions were introduced into the juxtamembrane region of p75 NTR , which is predicted to contain the cleavage site(s) for TACE and the constitutive sheddase. Stimulated shedding of p75 NTR was not detectably affected by the point mutations or by overlapping deletions of up to 11 aa residues within the juxtamembrane domain. Only a large deletion of 15 aa residues led to a strong decrease in constitutive and PMA-dependent shedding of p75 NTR . These results are reminiscent of studies of how mutations affect shedding of the TACE substrate TNF␣ (73), or of angiotensinconverting enzyme (74), which is not a TACE substrate (39). Shedding of TNF␣ and ACE was also only affected by relatively large deletions in the juxtamembrane domain, but not by shorter deletions. As has been suggested for TNF␣ and ACE, large deletions within the juxtamembrane domain of p75 NTR may place its ectodomain so close to the plasma membrane that the cleavage site is no longer accessible. In this context it is noteworthy that mutagenesis of the juxtamembrane domains of other shed molecules, including L-selectin (75), p55 TNF receptor (76), and the amyloid precursor protein (77) has uncovered a relaxed sequence specificity of their respective sheddases (3). However, in these studies, certain mutations in or around the predicted cleavage site of each protein did abolish ectodomain shedding. This raises the question of why none of the shorter deletions in the juxtamembrane domain of p75 NTR abolished its shedding? As shown in Fig. 6C, lanes 5-8, TACE and the constitutive sheddase(s) appear to cleave the receptor at several closely adjacent sites. Perhaps shorter deletions are not sufficient to abolish all of these potential cleavage sites. Another possibility is that TACE and the constitutive sheddase(s) have a highly relaxed sequence specificity toward p75 NTR . Furthermore, the mutations generated in this study may have created new sites for TACE or other sheddases (an example of such a scenario is described in Ref. 78). Finally, a cleavage site may be COOH-terminal to the deleted regions, and thus within the predicted transmembrane domain of p75 NTR . Further studies will be necessary to address these and other possibilities.
Ectodomain shedding is likely to occur in most or all cell types. In the case of the p75 NTR , shedding may have several different functional consequences. It could conceivably lead to inactivation of the receptor and generation of soluble decoy receptors. This might serve to decrease the concentration of neurotrophins at the cell surface, or lower the concentration of soluble neurotrophins, or both. Furthermore, recent studies demonstrate that p75 NTR undergoes RIP (55), which may in turn regulate the function of p75 NTR (56). Like in other cases of RIP, luminal membrane-proximal cleavage may be a prerequisite for the intramembrane cleavage of p75 NTR (55,56). However, although p75 NTR has been implicated in myelination, it was not possible to evaluate potential defects in this process in adam17Ϫ/Ϫ mice, because the perinatal lethality of these animals precludes proper analysis of myelination. Therefore future studies, including generation of a conditional TACE knockout mouse will be necessary to address the function of TACE-dependent p75 NTR shedding in vivo.
In summary, we present the evaluation of the activities that are involved in the ectodomain shedding of p75 NTR in different cell types, including CHO cells, COS-7 cells, and mouse embryonic fibroblasts. Pharmacological inhibitors were used to define different sheddases in these cells and a mutant CHO cell line. In addition, cells lacking ADAMs were used to identify TACE as the major stimulated p75 NTR sheddase in these cells. Further studies will be necessary to determine the identity of the constitutive p75 NTR sheddase, and to evaluate how shedding affects the function of p75 NTR . In light of recent evidence suggesting that p75 NTR is subjected to regulated intramembrane proteolysis, it will be interesting to learn whether this process is initiated and regulated by the membrane proximal cleavage by TACE or other metalloproteases in the same manner that RIP is controlled by membrane proximal proteolysis in other systems. The identification of TACE as a major p75 NTR sheddase is thus likely to contribute to a better understanding of the regulation of this neurotrophin receptor.