Stimulation-induced Down-regulation of Tumor Necrosis Factor-α Converting Enzyme

The extracellular domains of many proteins, including growth factors, cytokines, receptors, and adhesion molecules, are proteolytically released from cells, a process termed “shedding.” Tumor necrosis factor-α converting enzyme (TACE/ADAM-17) is a metalloprotease-disintegrin that sheds tumor necrosis factor-α and other proteins. To study the regulation of TACE-mediated shedding, we examined the effects of stimulation of cells on TACE localization and expression. Immunofluorescence microscopy revealed a punctate distribution of TACE on the surface of untreated cells, and stimulation of monocytic cells with lipopolysaccharide did not affect TACE staining. Phorbol 12-myristate 13-acetate (PMA), a potent inducer of shedding, decreased cell-surface staining for TACE. Surface biotinylation experiments confirmed and extended this observation; PMA decreased the half-life of surface-biotinylated TACE without increasing the turnover of total cell-surface proteins. Soluble fragments of TACE were not detected in the medium of cells that had down-regulated TACE, and TACE was not down-regulated when endocytosis was inhibited. Antibody uptake experiments suggested that cell-surface TACE was internalized in response to PMA. Surprisingly, a metalloprotease inhibitor prevented the PMA-induced turnover of TACE. Thus, PMA activates shedding and causes the down-regulation of a major “sheddase,” suggesting that induced shedding may be regulated by a mechanism that decreases the amount of active TACE on the cell surface.

A wide variety of proteins, including cytokines, growth factors, and their receptors, as well as cell adhesion molecules, are synthesized as transmembrane proteins that can be released from cells by proteolysis, a process termed "ectodomain shedding" (reviewed in Ref. 1). Release of soluble growth factors and cytokines, in addition to relaxing spatial constraints on their action, may qualitatively change their biological activities (2)(3)(4)(5). Shedding of transmembrane receptors may render cells unresponsive to particular ligands, and the resulting soluble receptors may act to modulate the activity of their ligands (6,7). Thus, the process of ectodomain shedding may regulate the interaction of cells with their environment in multiple ways (8). The importance of ectodomain shedding in mammalian development was underscored by the observation of perinatal lethality in mice lacking activity of the first identified "sheddase," tumor necrosis factor-␣ (TNF-␣) 1 converting enzyme (TACE) (9). TACE (ADAM-17) is a member of the ADAM (a disintegrin and metalloprotease) family of metalloprotease-disintegrins. It was originally identified by its ability to cleave transmembrane proTNF-␣ resulting in the release of mature, soluble TNF-␣ from cells (10,11). Subsequent studies using mice and cell lines lacking TACE activity implicated TACE in the shedding of transforming growth factor-␣, L-selectin, and the p75 TNFreceptor (9). Both ADAM-10 and TACE may also participate in the ␣-secretase processing of ␤-amyloid precursor protein (12,13). Another ADAM, ADAM-9 (meltrin-␥, MDC9), has recently been suggested to mediate the shedding of heparin-binding epidermal growth factor-like growth factor (HB-EGF) (14). Kuzbanian, the Drosophila homolog of mammalian ADAM-10, is thought to release the Notch ligand Delta from cells (15) and may also be required for proteolytic cleavage of Notch itself (16). Thus, ADAM proteins containing functional metalloprotease domains may constitute a family of sheddases that regulate plasma membrane composition and release soluble signaling molecules and receptors from cells.
The mechanisms by which shedding is regulated are poorly understood. Many shedding events are induced by phorbol esters (1), suggesting a role for protein kinase C (PKC). However, a number of metalloprotease-mediated shedding events induced by other stimuli are insensitive to PKC inhibitors, indicating that there are multiple signaling pathways capable of inducing the shedding of a particular molecule (17)(18)(19). In addition, it is not generally known whether induced shedding events occur via an activation or mobilization of metalloproteases or via some alteration in the conformation or localization of substrates. PKC␦ associates with the cytoplasmic domain of MDC9 and is required for phorbol ester-induced, MDC9-mediated HB-EGF shedding in Vero cells, suggesting that PKC may directly activate ADAMs in response to phorbol esters (14). In contrast, L-selectin shedding can be induced by cross-linking of L-selectin molecules (20) or by treatment of cells with calmodulin inhibitors, which are thought to antagonize the function of calmodulin molecules associated with the cytoplasmic tail of L-selectin (21). In most cases, however, induced shedding occurs even if the cytoplasmic domain of the substrate has been deleted entirely (22)(23)(24)(25)(26)(27)(28)(29).
Shedding enzymes are also thought to be regulated by naturally occurring inhibitors. ADAM proteins are synthesized as pro-enzymes. The N-terminal pro-domain is thought to main-tain the metalloprotease in a catalytically inactive state via a cysteine-switch mechanism until it is removed by proteolysis (30,31). Thus, posttranslational processing of ADAMs may regulate activity. In addition, purified TACE is inhibited by tissue inhibitor of metalloproteases-3 (TIMP-3) (32), and TIMP-3 can inhibit a number of shedding events (33)(34)(35). Similarly, TIMP-1 can partially inhibit the shedding of an alkaline phosphatase-HB-EGF fusion protein (17). Presumably, other sheddases can be similarly regulated, suggesting that an interplay between stimulatory and inhibitory regulators determines the shedding activity of a given cell.
In an attempt to understand the mechanism by which TACEmediated shedding events are activated, we examined the effects of lipopolysaccharide (LPS) and phorbol 12-myristate-13acetate (PMA) on the localization and abundance of cell-surface TACE. LPS treatment of monocytic cells appeared to have no effect on TACE expression or localization. Surprisingly, TACE was down-regulated in cells treated with PMA. This downregulation occurred via the internalization and degradation of TACE molecules; TACE appeared quite stable in untreated cells. The PMA-induced internalization and degradation of TACE was inhibited by a metalloprotease inhibitor, and the processed form of TACE was preferentially down-regulated in response to PMA, suggesting that inducible shedding events may be under feedback regulation.

EXPERIMENTAL PROCEDURES
Cells, Antibodies, and Reagents-THP-1 and Jurkat cells were grown in RPMI 1640 medium containing 10% fetal bovine serum, 1 mM Lglutamine, 50 units/ml penicillin, and 50 g/ml streptomycin. HeLa cells were grown in Dulbecco's modified Eagle's medium supplemented as above. All cell lines were grown at 37°C in a 5% CO 2 incubator.
Monoclonal antibody M220 was raised from a mouse immunized with recombinantly expressed human TACE extracellular domain (containing the catalytic and disintegrin/cysteine-rich regions) fused to the human IgG1 Fc region (10). The P1 polyclonal antiserum was from a rabbit immunized with the same recombinant human TACE extracellular domain-Fc fusion protein. The crude P1 antiserum was affinitypurified as described (36) on a column of immobilized recombinant human TACE extracellular domain. The P7 antiserum was from a rabbit immunized with a synthetic peptide corresponding to amino acids 19 -31 of human TACE (RPPDDPGFGPHQR, within the pro domain of TACE). The mouse monoclonal antibody directed against human L-selectin was purchased from R&D Systems. Fluorescently conjugated goat anti-mouse secondary antibodies were purchased from Molecular Probes.
Recombinant TACE extracellular domain (containing both catalytic and disintegrin/cysteine-rich domains) was purified from supernatants of Chinese hamster ovary cells that had been transfected with a plasmid encoding amino acids 1-670 of human TACE. The purified protein consisted primarily of processed material (lacking the pro-domain). The expression and purification of recombinant TACE catalytic domain has been described (37). A protein containing the disintegrin/cysteine-rich domain of TACE (residues 478 -671 of human TACE) fused to human IgG1 Fc was purified from the supernatants of transfected COS cells.
Characterization of Anti-TACE Antibodies-Recombinant TACE extracellular domain, catalytic domain, and disintegrin/cysteine-rich domain-Fc fusion protein were separated on non-reducing SDS-PAGE gels and Western-blotted by standard methods onto Immobilon-P membranes (Millipore). The blots were blocked with a solution of 5% nonfat dry milk and 0.1% Tween 20 in phosphate-buffered saline (PBS) for 1 h at room temperature and then incubated with primary antibodies diluted in PBS containing 2.5% bovine serum albumin and 0.1% Tween 20 for 1 h at room temperature. The M220 monoclonal antibody and affinity-purified P1 polyclonal antibodies were each used at a final concentration of 0.1 g/ml. The blots were then washed three times for 5 min with PBS, 0.1% Tween 20 and incubated in horseradish peroxidase (HRP)-conjugated donkey anti-mouse or donkey anti-rabbit secondary antibodies (Jackson Immunoresearch) as appropriate. After four more washes, the blots were visualized using enhanced chemilumines-cence reagents (ECL; Amersham Pharmacia Biotech).
Immunofluorescence Microscopy-THP-1 or Jurkat cells treated as indicated were pelleted at 200 ϫ g for 5 min. The cells were resuspended in 500 l of ice-cold PBS containing 5% serum and incubated for 15 min on ice. For THP-1 cells, heat-inactivated human serum was used as a blocking agent to saturate Fc receptors; goat serum was used to block Jurkat cells, which were not as susceptible to nonspecific staining. The cells were then pelleted, resuspended in primary antibody diluted to a final concentration of 5 g/ml in blocking solution, and incubated for 30 min on ice. In some experiments, soluble TACE extracellular domain was used as a competitor at a final concentration of 25 g/ml in the primary antibody solution. After incubation with primary antibodies, the cells were washed twice in ice-cold PBS, fixed in a solution of 3.2% paraformaldehyde in PBS for 10 min at room temperature, and then quenched in 50 mM NH 4 Cl in PBS for 10 min at room temperature. They were then incubated with fluorescently conjugated secondary antibodies diluted to 5 g/ml in PBS containing 5% goat serum for 1 h at room temperature in the dark with gentle mixing. Following three washes with PBS, the cells were either mounted on chambered coverslips (Nalge Nunc International) in a solution of 50% glycerol in PBS or on standard microscope slides using ProLong antifade reagent (Molecular Probes). Samples were viewed and images acquired on a confocal laser scanning microscope (Nikon/Molecular Dynamics) using a 60ϫ objective and appropriate filters.
Cell-surface Biotinylation and Immunoprecipitation-Cells from cultures treated as indicated were pelleted for 5 min at 200 ϫ g, washed three times with ice-cold PBS C/M (PBS containing 1 mM MgCl 2 and 0.1 mM CaCl 2 ), and incubated for 30 min at 4°C with 100 g/ml D-biotinoyl-⑀-aminocaproic acid N-hydroxysuccinimide ester (Roche Molecular Biochemicals) in PBS C/M, pH 8.4. The cells were then washed 3 times with an ice-cold solution of 50 mM Tris, pH 7.5, 150 mM NaCl, 25 mM KCl. Depending on the experiment, the biotinylated cells were then either lysed immediately or returned to culture medium and incubated as indicated.
Lysates were prepared from pelleted cells by resuspending the cells in ice-cold 50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1% Triton X-100 containing 1 mM phenylmethylsulfonyl fluoride, 20 M pepstatin A, and 10 M leupeptin (Sigma) and incubating on ice for 30 min. Insoluble material was pelleted in a microcentrifuge at 20,800 ϫ g for 10 min at 4°C, and the resulting supernatant was transferred to a fresh microcentrifuge tube. Protein concentrations were determined by BCA assay (Pierce).
Normalized amounts of total protein (typically 200 g) were used as input for each immunoprecipitation. Lysates were incubated with 5 g of antibody/sample for 2 h at 4°C. Immune complexes were then collected on protein G-Sepharose beads (Amersham Pharmacia Biotech) for 2 h at 4°C with gentle mixing, after which the beads were washed four times with immune precipitation buffer (50 mM Tris, pH 7.5, 150 mM NaCl, 1% Nonidet P-40, 0.5% sodium deoxycholate, 0.1% sodium dodecyl sulfate) containing protease inhibitors as above. The washed beads were resuspended in SDS-PAGE sample buffer containing ␤-mercaptoethanol, heated to 95°C for 5 min, and the resulting supernatants were separated by SDS-PAGE on pre-cast gels (Novex). Following electrophoresis, Western transfers onto Immobilon-P membranes (Millipore) were performed by standard methods, and biotinylated proteins were detected by probing blots with streptavidin conjugated to either HRP (Amersham Pharmacia Biotech) or alkaline phosphatase (Molecular Probes). Blots probed with HRP conjugates were visualized using ECL reagents and exposure to x-ray film. Blots probed with alkaline phosphatase conjugates were developed using enhanced chemifluorescence (Amersham Pharmacia Biotech) and were scanned on a Storm scanner (Molecular Dynamics) in blue fluorescence mode.
Detection of Soluble TACE in Culture Supernatants-Jurkat cells (5 ϫ 10 6 cells/sample) were pelleted and resuspended in culture medium lacking serum at a final concentration of 2.5 ϫ 10 6 cells/ml in the presence or absence of 100 ng/ml PMA. To estimate the loss of soluble TACE during the culture period and subsequent preparation of the culture supernatants, purified recombinant TACE extracellular domain protein was added to the culture medium of some samples. After a 2-h incubation at 37°C, the cells were pelleted and lysates were prepared in ice-cold immune precipitation buffer containing 1 mM phenylmethylsulfonyl fluoride, 20 M pepstatin A, 10 M leupeptin, 2 mM 1,10-phenanthroline (Sigma), and 25 M IC-3. The culture supernatants were concentrated approximately 25-fold using Centricon-10 centrifugal filter devices (Amicon). Equivalent fractions of the cell lysates and culture supernatants were separated by non-reducing SDS-PAGE and Western-blotted as above. To estimate the sensitivity of the blots, 1:3 serial dilutions of lysates of untreated cells were prepared in non-reducing sample buffer and analyzed in parallel to treated samples and culture supernatants. The blots were probed with the affinity-purified P1 antibodies diluted to 0.1 g/ml or with a 1:5000 dilution of the P7 polyclonal antiserum as described above for characterization of anti-TACE antibodies.
Detection of Cell-surface L-selectin-Jurkat cells were incubated in culture medium or hypertonic culture medium (culture medium with sucrose added to a final concentration of 450 mM) in the presence or absence of 100 ng/ml PMA and 25 M IC-3 as indicated. After a 30-min incubation at 37°C, the cells were pelleted, washed once with ice-cold PBS, and incubated with a monoclonal antibody directed against Lselectin or an isotype-matched control antibody diluted to a final concentration of 5 g/ml in PBS containing 5% goat serum. After a 30-min incubation on ice, the cells were pelleted and washed three times with ice-cold PBS. They were then fixed and reacted with secondary antibodies as described above for immunofluorescence detection of TACE. The final stained cell preparations were resuspended in PBS and analyzed on a Becton Dickinson FACScan flow cytometer.
Antibody Uptake Experiments-HeLa cells were plated onto glass coverslips 2 days prior to each experiment. THP-1 cells were grown and treated in suspension. On the day of the experiment, the culture medium was removed and replaced with medium containing 100 ng/ml PMA and 5 g/ml anti-TACE M220 or an isotype-matched control antibody as indicated. The cells were then incubated at 37°C for 1 h, washed once with PBS, and fixed with 3.2% paraformaldehyde in PBS for 10 min at room temperature, followed by a 10-min incubation in Ϫ20°C methanol. Residual paraformaldehyde was quenched by incubation in 50 mM NH 4 Cl in PBS for 10 min at room temperature. The fixed cells were incubated with fluorescently labeled secondary antibodies and processed for microscopy as described above.
Endocytosis Assays-For measurement of transferrin uptake, cells were incubated with 25 g/ml BODIPY-FL transferrin (Molecular Probes), diluted in RPMI 1640 medium lacking serum for 30 min at 37°C. The cells were washed twice with ice-cold PBS, once with ice-cold mild acid wash buffer (50 mM MES, pH 5.5, 280 mM sucrose), and once more with ice-cold PBS. They were next fixed with paraformaldehyde and analyzed by flow cytometry as above.
For measurement of fluid-phase endocytosis, Jurkat cells were incubated with 40-kDa lysine-fixable fluorescein-conjugated dextran (Molecular Probes) diluted to 500 g/ml in culture medium containing 100 ng/ml PMA and 25 M IC-3 as indicated. After 30 min at 37°C, the cells were pelleted and washed three times with ice-cold PBS. The cells were then fixed with 3.2% paraformaldehyde for 10 min, pelleted and resuspended in PBS, and analyzed by flow cytometry.

RESULTS
Characterization of Anti-TACE Antibodies-Monoclonal and polyclonal antibodies directed against the extracellular domain of TACE were raised in mice and rabbits immunized with a recombinant TACE extracellular domain-human IgG1 Fc fusion protein. To further characterize the resulting antibodies, we examined their specificity on Western blots of recombinantly expressed fragments of TACE. Fig. 1A shows the domain structure of TACE. Western blots of recombinant proteins containing the catalytic domain (cat only), the disintegrin/cysteine-rich domain (dis only), or both domains (catϩdis) were probed with the anti-TACE antibodies (Fig. 1B). The M220 monoclonal antibody detected the recombinant catalytic domain plus disintegrin/cysteine-rich domain protein and the disintegrin/cysteine-rich domain protein alone, but not the catalytic domain protein alone. Thus, M220 recognizes an epitope within the disintegrin/cysteine-rich domain of TACE. The affinity-purified rabbit polyclonal antiserum designated P1 recognized all three recombinant proteins. Therefore, this polyclonal antiserum contains antibodies that recognize epitopes within both the catalytic and disintegrin/cysteine-rich domains of TACE.
We next used THP-1 cells to evaluate the reactivity of the anti-TACE antibodies toward cellular forms of TACE. M220 and P1 each recognized processed and unprocessed forms of TACE on Western blots of THP-1 cell lysates (Fig. 1C, cell lysates). Unprocessed forms of TACE (containing the pro-domain) were identified by probing Western blots with P7, a polyclonal antiserum raised against a synthetic peptide derived from the pro-domain of TACE. M220 and P1 also immunoprecipitated both processed and unprocessed forms of TACE (Fig.  1D). However, only the processed form of TACE was detected when immunoprecipitates of surface-biotinylated cells were probed with streptavidin-HRP conjugates, regardless of the precipitating antibody used (Fig. 1D). Therefore, M220 and P1 recognize unprocessed and processed forms of TACE, but only processed TACE is detectable on the surface of cells, as has been suggested previously (10).
We next used THP-1 cells to evaluate the utility of the M220 antibody for immunofluorescence detection of cell-surface TACE. The antibody produced a punctate pattern of staining on the surface of THP-1 cells (Fig. 1E, left panel). This pattern of staining was not observed with an isotype-matched control antibody (Fig. 1E, middle panel), and including excess soluble TACE as a competitor in the primary antibody reaction effectively blocked the cell-surface staining (Fig. 1E, right panel). Therefore, M220 appears to specifically detect TACE on the surface of cells.
Effect of Stimulation on Cell-surface TACE Expression-To begin to address the mechanism by which shedding is activated, we characterized the levels and distribution of TACE on the surface of cells by fluorescence microscopy. Treatment of monocytic cells with LPS induces synthesis of transmembrane proTNF-␣, which is cleaved by TACE, resulting in the release of mature TNF-␣ into the culture supernatant. TNF-␣ biosynthesis is subject to both transcriptional and posttranscriptional regulation (38). However, it is not known whether activation or increased expression of TACE is a component of the posttranscriptional regulation of TNF-␣ production. Previously, biotinylation experiments indicated that LPS treatment did not increase the amount of TACE on the plasma membrane of primary human monocytes (10).
To further address the question of whether LPS-stimulated TNF-␣ production is regulated at the level of shedding, we examined the effect of LPS on expression and localization of cell-surface TACE in THP-1 cells. Treatment of the cells for 3 h with 1 g/ml LPS, at which time the cells are releasing large amounts of soluble TNF-␣ into the medium, 2 did not affect the amount or distribution of cell-surface TACE, as detected by immunofluorescence (Fig. 2, A and B). Thus, the unstimulated level and distribution of TACE may be sufficient to shed TNF-␣ from LPS-stimulated monocytic cells, and the effect of LPS on this system may be solely to induce synthesis of TNF-␣, which is then released by constitutive TACE activity.
Phorbol esters such as PMA are potent inducers of ectodomain shedding. Because PMA-induced shedding events occur rapidly and because PMA is known to stimulate protein secretion (39,40), we speculated that the induction of shedding by PMA might be due to stimulated movement of active TACE molecules from intracellular stores to the plasma membrane.
To test this possibility, we examined the effect of PMA treatment on cell-surface levels and localization of TACE by microscopy. Jurkat cells, which shed L-selectin in response to PMA, were treated with PMA, stained with an anti-TACE monoclonal antibody, and examined by confocal microscopy as before. The distribution of TACE on the surface of unstimulated Jurkat cells (Fig. 2C) was similar to that described previously for THP-1 cells, although the amount of TACE was lower on Jurkat cells. Surprisingly, PMA treatment caused a decrease in staining for cell-surface TACE (Fig. 2D). THP-1 cells showed a similar loss of TACE in response to PMA (Fig. 2, E and F).
To confirm that TACE was lost from the surface of PMA-treated cells, we examined the effects of PMA on TACE by cell-surface biotinylation. As was the case for THP-1 cells, only the processed form of TACE was detected on the surface of Jurkat cells. Supporting the immunofluorescence results, treatment of Jurkat cells with PMA led to a down-regulation of cell-surface TACE as measured by biotinylation and immunoprecipitation (Fig. 3A). A similar down-regulation was seen in THP-1 cells (Fig. 3B). Additional experiments showed that a variety of cell lines down-regulated cell-surface TACE in response to PMA, including T84, HeLa, and CV-1 (data not shown), suggesting that the loss of TACE is a general feature of the response of cells to PMA, rather than a peculiarity of cells of hematopoietic origin. PMA Treatment Results in the Turnover of Pre-existing TACE Molecules-To examine the fate of pre-existing TACE molecules in PMA-treated cells, we reversed the steps in the previous experiment such that Jurkat cells were surface-biotinylated before treatment with PMA. This approach is akin to a pulse-chase experiment; TACE molecules present on the surface of cells are labeled with biotin at the beginning of the experiment, and their fate is followed over time.
In the absence of PMA, surface-biotinylated TACE was stable in Jurkat cells, with a half-life of at least 8 h (Fig. 4A). PMA treatment decreased the half-life of surface-biotinylated TACE to approximately 90 min (Fig. 4B) without appreciably affecting the stability of total biotinylated cell-surface proteins (Fig.  4, C and D). These observations argue that cell-surface TACE is degraded (or shed) rather than redistributed to an intracellular compartment in response to PMA, and that the loss of TACE does not merely reflect a bulk turnover of cell-surface proteins, since total surface-biotinylated proteins do not show a similar destabilization in PMA-treated cells.
TACE Does Not Appear to Be Shed in Response to PMA-The disappearance of TACE from the cell surface in response to PMA raised the possibility that TACE itself is shed. Although shedding of membrane-bound TACE has not been previously reported, the catalytic domains of several ADAMs are known to be removed by proteolysis (41)(42)(43). If TACE were being shed in our experiments, we would expect to detect (with appropriate antibodies) a cell-associated transmembrane fragment remaining after proteolysis and a soluble secreted fragment of TACE. We did not detect any lower molecular weight fragments of Lysates of THP-1 cells were immunoprecipitated with either M220 or P1 antibodies as indicated. Precipitated proteins were separated by non-reducing SDS-PAGE, Western-blotted, and probed with the indicated antibodies or streptavidin-HRP (SA). Samples probed with streptavidin-HRP were derived from cells that had been surface-biotinylated prior to lysis. In C and D, open and closed triangles indicate the migration of unprocessed and processed forms of TACE, respectively. E, immunofluorescence detection of cell-surface TACE. THP-1 cells were stained with monoclonal antibody M220 (IgG1), an IgG1 isotype control antibody, or with M220 in the presence of excess soluble human TACE, followed by a goat anti-mouse Alexa-488 conjugate. Staining was visualized by confocal microscopy. TACE in immunoprecipitations from cell lysates using antibody M220, which recognizes an epitope within the disintegrin/ cysteine-rich domain of TACE (Fig. 4B).
We also attempted to detect soluble forms of TACE in the supernatants of PMA-treated cells. We first determined how much cell-associated TACE was lost in response to PMA, thereby establishing the amount of TACE expected in the medium if it were shed. The P1 antibodies detected three forms of TACE in lysates of untreated Jurkat cells (Fig. 5A); the upper doublet was identified as unprocessed TACE, based on its reactivity with the P7 antiserum (Fig. 5A, lane 1), which was raised against a peptide derived from the pro-domain of TACE. The amount of processed TACE (the lower band) in cell lysates decreased in response to PMA treatment, consistent with data presented above, indicating that the loss of TACE detected previously with M220 was not due simply to the loss of the particular epitope recognized by the monoclonal antibody. Interestingly, PMA treatment selectively decreased the abundance of processed TACE without affecting the amount of unprocessed TACE (Fig. 5A).
No soluble forms of TACE were detected in the culture supernatants of untreated or PMA-treated cells (Fig. 5A), even when Western blots were overexposed (Fig. 5B). However, purified TACE protein added to cell cultures during the incuba-tion with PMA was readily detected. Greater than 50% of the material spiked into the culture medium was recovered (Fig.  5A, compare culture supernatants to input), suggesting that a soluble form of TACE shed by the cells would have been detected by this method. With prolonged exposures, we could detect less than 4% (a 1:27 dilution) of the TACE present in unstimulated cells (Fig. 5B, cell lysate). Despite this sensitivity, we did not detect any soluble TACE on prolonged exposures of the culture supernatant blots (Fig. 5B, culture supernatants). Thus, if PMA-treated cells shed TACE, the hypothetical shed form of TACE is either a small fraction of the amount present in untreated cells, or it is somehow rendered undetectable in culture supernatants by a mechanism that does not affect TACE spiked into the medium.
Hypertonic Medium Prevents PMA-induced Turnover of TACE-Because we did not detect soluble forms of TACE in supernatants of PMA-treated cells, we speculated that TACE might be internalized in response to PMA. To test for a role of endocytosis in the down-regulation of TACE, we measured the effect of hypertonic medium, which inhibits clathrin-mediated endocytosis (44), on TACE turnover. As in Fig. 4, PMA induced the loss of biotinylated TACE from cells incubated in isotonic medium (Fig. 6A). In contrast, cells incubated in hypertonic medium did not appreciably down-regulate TACE in response to PMA (Fig. 6A, second panel).
To confirm that hypertonic treatment was having the intended effect, we also assayed receptor-mediated endocytosis and L-selectin shedding. Hypertonic medium strongly decreased transferrin endocytosis, and the extent of inhibition was similar in the absence or presence of PMA (Fig. 6B). However, L-selectin was still shed by cells incubated in hypertonic medium (Fig. 6C). In fact, hypertonic medium alone was sufficient to induce shedding of cell-surface L-selectin, as has been described (45). This loss was enhanced by PMA treatment, indicating that the cells had not lost the ability to respond to PMA. The metalloprotease inhibitor IC-3, a peptide hydroxamate that inhibits TACE (and other metalloproteases) by binding to its active site and chelating a zinc ion required for catalytic activity (37), prevented both the hypertonic medium-induced and the PMA-induced loss of L-selectin from cells. Therefore, the activation of shedding is not sufficient for the down-regulation TACE, and the loss of TACE is impaired under conditions that inhibit clathrin-mediated endocytosis.
PMA Treatment Induces the Internalization of TACE-To address directly the possibility that TACE was being internalized, we measured the uptake of anti-TACE antibodies by cells treated with PMA. HeLa and THP-1 cells were incubated in the presence or absence of PMA in medium containing the M220 monoclonal antibody. HeLa cells were used because, as adherent cells, their morphology was more amenable to analysis of intracellular protein distribution than that of non-adherent THP-1 cells. After various periods of incubation, the cells were fixed and permeabilized, incubated with fluorescently tagged anti-mouse secondary antibodies, and viewed by confocal microscopy.
TACE staining was readily detected on the surface of untreated HeLa cells (Fig. 7A). After 1 h of incubation with PMA, very little surface staining was detected (Fig. 7B). Instead, the anti-TACE antibody had accumulated intracellularly in structures likely to represent endosomes or lysosomes. Cells incubated with M220 in the absence of PMA retained surface staining for TACE and displayed only a low level of intracellular staining (Fig. 7C). The intracellular fluorescence seen in unstimulated cells may reflect either basal internalization of TACE or a low level of internalization induced by the binding of antibodies to TACE molecules on the cell surface. Cells incu- bated for 1 h with an isotype control antibody in the presence of PMA did not show significant surface or intracellular staining (Fig. 7D), indicating that the internalization of M220 could not be ascribed to nonspecific fluid-phase uptake of antibodies induced by PMA. THP-1 cells also showed uptake of M220 into punctate intracellular structures in response to PMA (Fig. 7, E  and F). The difference in staining patterns between the two cell types may arise in part from differences in the intracellular morphology of non-adherent monocytic cells (THP-1) and adherent epithelial cells (HeLa). Additionally, because they are non-adherent, the THP-1 cells were oriented randomly in the samples, making it impossible to obtain a uniform plane of section through all of the cells in a given field. However, the observation that PMA treatment increased internalization of anti-TACE antibodies was consistent between the two cell types. Thus, PMA appears to induce the internalization of cell-surface TACE molecules.
A Metalloprotease Inhibitor Prevents TACE Down-regulation-Because we initially suspected that TACE might be shed in response to PMA, we tested the effects of the metalloprotease inhibitor IC-3 (10) on the PMA-induced down-regulation of TACE. Surface biotinylation experiments showed that IC-3 inhibited the PMA-induced disappearance of pre-existing TACE molecules (Fig. 8A).
Typically, inhibition of the loss of membrane proteins by hydroxamates is an indication of metalloprotease-mediated shedding. However, the results shown in Figs. 5-7 suggest that TACE is primarily internalized, not shed, in response to PMA. Although we could not find evidence in the literature for a role of metalloproteases in endocytosis, it remained possible that IC-3 prevented the down-regulation of TACE by inhibiting endocytosis via an unknown mechanism. Therefore, we measured the effect of IC-3 on fluid-phase and receptor-mediated endocytosis. Jurkat cells were incubated with PMA in the presence or absence of IC-3 for 30 min. Uptake of fluorescent dextran and transferrin conjugates was used to measure fluidphase and receptor-mediated endocytosis, respectively (Fig. 8, B and C). IC-3 had no effect on dextran or transferrin uptake, indicating that its protective effect on cell-surface TACE was not due to a nonspecific inhibition of endocytosis. Therefore, it is likely that IC-3 protects TACE from degradation by inhibiting either TACE itself or another metalloprotease. DISCUSSION The experiments described here were undertaken in an attempt to elucidate the mechanisms by which ectodomain shedding is activated. LPS stimulation of THP-1 cells appeared to have no effect on TACE localization or expression (Fig. 2),

FIG. 3. Loss of cell-surface TACE measured by surface biotinylation.
Jurkat cells (A) or THP-1 cells (B) were treated with 100 ng/ml PMA. At the indicated times, aliquots of the cultures were removed and the cells were surface-biotinylated. The cells were then lysed, and protein was immunoprecipitated with anti-TACE M220 or a control antibody (c). Immunoprecipitated biotinylated TACE was detected on Western blots probed with streptavidin-HRP. Migration of molecular size markers, in kDa, is indicated.

FIG. 4. PMA-induced turnover of biotinylated TACE protein.
Jurkat cells were surface-biotinylated and then returned to culture medium and incubated in the presence or absence of 100 ng/ml PMA. At the indicated times, cell lysates were prepared from aliquots of the cultures and immunoprecipitated with M220. Immunoprecipitated TACE or total biotinylated proteins were detected on Western blots probed with streptavidinalkaline phosphatase. A and B show anti-TACE immunoprecipitates from untreated and PMA-treated cultures, respectively. C and D show total biotinylated proteins from untreated and PMAtreated cultures, respectively. Migration of molecular size markers, in kDa, is indicated.
suggesting that the basal level of TACE activity may be sufficient for the release of TNF-␣ from LPS-stimulated cells. In support of this, COS-7 cells transfected to express proTNF-␣ shed significant amounts of mature TNF-␣ into their culture medium in the absence of any stimulation (46). Thus, shedding per se may not be activated in the case of LPS-induced TNF-␣ release.
The shedding of other apparent TACE substrates is activated by PMA. However, we have shown here that treatment of cells with PMA causes the down-regulation of plasma membrane TACE rather than an up-regulation. Treatment of cells with PMA caused 1) a decrease in the amount of cell-surface TACE, as measured by both immunofluorescence microscopy and cell-surface biotinylation (Figs. 2-4), and 2) a decrease in the total cellular amount of processed TACE, as measured by direct Western blotting (Fig. 5). The loss of cell-surface TACE appeared to be less rapid than the shedding response itself; PMA-induced shedding of L-selectin occurs within minutes (47). In addition, a low level of processed TACE is detected in cells even after several hours of incubation with PMA (see Figs.  3 and 4). Thus, the PMA-induced loss of TACE is not incompatible with the action of TACE as a broad specificity sheddase.
Antibody uptake experiments (Fig. 7) and the inhibition of induced TACE turnover by hypertonic medium (Fig. 6) suggested that PMA induced the internalization of plasma membrane TACE molecules. We cannot exclude the possibility that some cellular TACE is shed in response to PMA, but to date we have been unable to detect soluble or cell-associated fragments of TACE that shedding would be expected to generate (Fig. 5). The internalization of TACE may occur via clathrin-coated pits since 1) the cytoplasmic domain of TACE contains a potential YXXØ internalization motif (YESL, residues 702-705 of hu-FIG. 5. TACE is not detected in supernatants of PMA-stimulated cells. Jurkat cells were incubated in serum-free medium in the presence or absence of 100 ng/ml PMA as indicated for 2 h at 37°C. To evaluate the stability and recovery of TACE in culture medium, purified recombinant soluble TACE extracellular domain protein (sol. TACE) was added to some cultures prior to incubation at a final concentration of 100 ng/ml. At the end of the culture period, cell lysates were prepared, and culture supernatants were concentrated using Centricon-10 filter devices. A, equal fractions (10%) of the cell lysates and concentrated culture supernatants were separated by non-reducing SDS-PAGE and blotted. In parallel, dilutions of recombinant TACE protein in serum-free medium were prepared and immediately electrophoresed and blotted (input). Western blots were probed with anti-TACE polyclonal antibody P7 (directed against the pro-domain, anti-pro) or P1 (directed against the catalytic and disintegrin/cysteine-rich domains, cat ϩ dis) as indicated. Open and closed triangles indicate the migration of unprocessed (pro-containing) and processed forms of TACE, respectively. For the cell lysate and culture supernatant samples, the theoretical amount (in nanograms, assuming 100% recovery) of soluble TACE present in the fraction of culture supernatant loaded on the gels is indicated. For the input samples, the actual amount (in nanograms) of recombinant soluble TACE loaded in each lane is indicated. B, the lysate of untreated cells was serially diluted in non-reducing sample buffer and electrophoresed, Western-blotted, and probed with the P1 antibody in parallel with concentrated culture supernatants from untreated and PMA-treated cells. The amount of undiluted cell lysate (left panel, first lane) and concentrated culture supernatants (right panel) loaded were equivalent fractions (10%) of the total material. Exposure times were 1 min (A) and 45 min (B). man TACE) (48) and 2) the PMA-induced loss of TACE was inhibited when cells were incubated in hypertonic medium, which disrupts clathrin organization and impairs endocytosis (44).
Remarkably, the metalloprotease inhibitor IC-3 prevented the PMA-induced disappearance of TACE molecules (Fig. 8). The protective effect of IC-3 was not due to an inhibition of endocytosis, since IC-3 did not affect uptake of fluorescently tagged dextran or transferrin molecules. Because IC-3 is a relatively nonspecific metalloprotease inhibitor, we do not know whether inhibition of TACE itself is sufficient to prevent PMA-induced down-regulation; the relevant target of IC-3 may be either TACE or another metalloprotease. Inhibiting metalloproteases other than TACE could act to stabilize TACE if a hydroxamate-sensitive metalloprotease participated in the degradation of internalized TACE molecules. By blocking such an activity, IC-3 might allow the recycling to the cell surface of TACE molecules that would otherwise be degraded. However, lysosomal metalloproteases have not been identified in mammalian cells, arguing against this possibility.
The interaction of IC-3 with TACE itself might protect TACE molecules from internalization and degradation. Binding of IC-3 to TACE might alter the conformation of the TACE extracellular domain in a manner that prevents the internalization of TACE. Alternatively, TACE activity might be required for its internalization. In one scenario, TACE could remain associated with the transmembrane remnants of some shed proteins, and these complexes might be preferentially internalized and degraded. In this case, IC-3 might stabilize cell-surface TACE by preventing the formation of this hypothetical complex. Another possibility is that a protein cleaved by TACE (or another metalloprotease) in response to PMA generates a signal that induces TACE down-regulation. Both of these hypothetical mechanisms suggest that the amount of TACE activity on the surface of cells is under feedback control; the down-regulation of shedding enzymes may be a direct consequence of their FIG. 6. Effect of hypertonic medium on TACE turnover. A, Jurkat cells were surface-biotinylated and then incubated in culture medium (isotonic) or culture medium containing 450 mM sucrose (hypertonic), each containing 100 ng/ml PMA. At the indicated times, aliquots of the cultures were removed, and cell lysates were immunoprecipitated with M220. The immunoprecipitates were separated by SDS-PAGE, Western-blotted, and probed with streptavidin-HRP. B, Jurkat cells were incubated in isotonic (upper panels) or hypertonic (lower panels) serum-free medium containing 25 g/ml BODIPY-FL transferrin in the presence or absence of 100 ng/ml PMA. After 1 h at 37°C, the cells were washed and transferrin uptake was analyzed by flow cytometry. Open and shaded histograms are cells incubated with and without BODIPY-FL transferrin, respectively. C, Jurkat cells were incubated in isotonic (upper panels) or hypertonic (lower panels) culture medium containing 100 ng/ml PMA and 25 M IC-3 as indicated. After 30 min at 37°C, the cells were stained for surface L-selectin and analyzed by flow cytometry. Open and shaded histograms are from cells stained with anti-L-selectin and isotype matched control antibodies, respectively.
FIG. 7. PMA-induced uptake of anti-TACE antibodies. HeLa or THP-1 cells were treated with 100 ng/ml PMA in the presence of 5 g/ml M220 for 1 h at 37°C. At the end of the incubation, the cells were fixed and permeabilized, incubated with fluorescently tagged anti-mouse secondary antibodies, and viewed by confocal microscopy. Shown are images of untreated HeLa cells stained for cell-surface TACE (A), cells incubated with PMA and M220 for 60 min (B), HeLa cells incubated with M220 for 1 h in the absence of PMA (C), HeLa cells incubated with PMA and an isotype-matched control antibody for 1 h (D), and THP-1 cells were incubated with M220 for 1 h in the absence (E) and presence (F) of 100 ng/ml PMA.

activity.
The observation that only the mature form of TACE is downregulated in response to PMA (Fig. 5) suggests either that catalytically active forms of TACE are preferentially subject to down-regulation or that cell-surface TACE is selectively downregulated. Our data do not distinguish between these two possibilities. However, the selective down-regulation of mature TACE and the protective effect of IC-3 are both consistent with a requirement for TACE activity.
The consequence of TACE down-regulation may be a decrease in the rate of shedding, allowing the replenishment of the cell's repertoire of surface proteins. In support of this idea, in studies of cells stably expressing the p55 TNF receptor, a cell-surface receptor thought to be shed by TACE, 3 cell-surface p55 expression rapidly decreased via shedding in response to PMA but then gradually increased from its lowest level over longer periods of incubation (49). Additionally, the release of soluble TNF-␣ from cells transfected with plasmids encoding proTNF-␣ initially increases dramatically but then plateaus over time in response to PMA treatment, even though the cells should be continuously synthesizing large amounts of proTNF-␣. 4 These observations are consistent with the hypothesis that the induced down-regulation of TACE results in decreased shedding activity.
It is not known whether other ADAMs are down-regulated by PMA-treated cells. Similar down-regulation of other ADAMs would bolster the idea that the shedding response is under feedback control at the level of metalloprotease expression on the cell surface. Alternatively, if only a subset of sheddases are down-regulated in response to PMA, then the specificity and kinetics of the shedding response may change over longer periods of treatment. It will also be of interest to determine whether other stimuli that induce shedding also cause the down-regulation of shedding enzymes, as a feedback model would predict. nism following cell lysis in the absence of metalloprotease inhibitors (Schlöndorff, J., Becherer, J. D., and Blobel, C. P. (2000) Biochem. J. 347, 131-138). Detection of TACE with our antibodies directed against the extracellular domain is unaffected. However, the apparent molecular weight of processed TACE shown in Figs. 3 and 4 (but not Figs. 1 and 5, in which metalloprotease inhibitors were present) is probably decreased slightly as a result of this post-lysis phenomenon.