Heregulin-dependent Trafficking and Cleavage of ErbB-4*

Heregulin was shown to promote the proteolytic cleavage of its receptor, ErbB-4, in several cell lines. The growth factor also rapidly promoted the transient translocation of ErbB-4 to a detergent-insoluble fraction, in which the receptor was hyper-tyrosine-phosphorylated compared with the receptor present in the detergent-soluble pool. However, an 80-kDa proteolytic fragment of ErbB-4 was found in the detergent-soluble fraction, but not in the detergent-insoluble fraction. Although the heregulin-induced cleavage of ErbB-4 produced a fragment of ErbB-4 very similar to that induced by 12-O-tetradecanoylphorbol-13-acetate or pervanadate (each of which is blocked by metalloprotease inhibitors), the growth factor-induced cleavage was not sensitive to these inhibitors under the same conditions. The heregulin-induced cleavage of ErbB-4 could be blocked by conditions that prevent clathrin-coated pit formation, suggesting that heregulin-mediated ErbB-4 cleavage occurs subsequent to internalization. When reagents that prevent acidification of endosomes were employed, heregulin-induced ErbB-4 cleavage was sensitive to metalloprotease inhibitors. The results imply that during ligand-dependent receptor trafficking, activated ErbB-4 receptors are subject to proteolytic cleavage involving an intracellular metalloprotease.

ErbB-4 is a member of the epidermal growth factor family of receptor tyrosine kinases (1) and is expressed in many normal adult and fetal tissues, particularly nerve and heart (2). However, its expression in carcinoma cells is limited compared with other ErbB receptors (2)(3)(4). Mice lacking ErbB-4 die during mid-embryogenesis from aborted development of heart ventricle myocardial trabeculae and are also deficient in axon guidance during development of the central nervous system (5,6). Although all the EGF 1 receptor family members are able to stimulate cell proliferation, ErbB-4 is also implicated as a positive regulator of the differentiation of certain epithelial and neuronal tissues (7)(8)(9). Ligands for the ErbB-4 receptor include various heregulin isoforms, which also bind to ErbB-3 (10,11), and betacellulin (10,12), which also binds to the EGF receptor (12). Following ligand stimulation, the ErbB-4 receptor under-goes homodimerization and/or heterodimerization with other receptors of the ErbB family and initiates cellular responses (13,14).
Desensitization and down-regulation of activated growth factor receptors have an important role in regulating cellular events triggered by ligand binding and determine signaling duration and/or potency. A common pathway for down-regulation of many activated growth factor receptors is clathrincoated pit-mediated endocytosis (15). In the case of the EGF receptor, the activated receptor rapidly enters clathrin-coated pits and is internalized and subsequently sorted from endosomes to lysosomes, where the receptor and its ligand are degraded (15,16). However, studies with NIH 3T3 cells expressing wild-type ErbB receptors or chimeric receptors and some ErbB-expressing human carcinoma cell lines showed that all other activated ErbB receptors (ErbB-2, ErbB-3, and ErbB-4) are endocytosis-impaired and are inefficiently internalized (17).
Earlier studies from this laboratory showed that the ErbB-4 receptor is slowly and constitutively cleaved, producing a membrane-anchored cytoplasmic domain fragment of 80 kDa (18). ErbB-4 receptor cleavage is greatly enhanced by the protein kinase C activator TPA or by pervanadate, a potent phosphotyrosine phosphatase inhibitor (19,20). These results suggest the presence of intracellular signaling pathways that can regulate this cleavage event. Both the TPA-and pervanadateinduced cleavages of ErbB-4 are completely inhibited by the hydroxamate-based metalloprotease inhibitor BB-94. Recently, it has been shown that the tumor necrosis factor-converting enzyme (TACE) is required for TPA-or pervanadate-induced ErbB-4 cleavage (21). Also, a 23-residue domain in the extracellular juxtamembrane domain of ErbB-4 is necessary for this cleavage (21,22). However, the exact site of cleavage of ErbB-4 by TACE is not known. TACE and other metalloproteases have been shown to be involved in ectodomain shedding of a variety of membrane proteins, including the precursor for transforming growth factor ␣, an EGF receptor ligand (23). In many of these TACE-dependent cases, cleavage can be stimulated by TPA (24), as reported for ErbB-4 (19). However, studies of the regulated cleavage of cell-surface proteins by hormones, growth factors, and pervanadate indicate that these proteolytic events are protein kinase C-independent and involve signaling pathways that are not well defined. Fan and Derynck (25) have shown that shedding of the transforming growth factor can be stimulated by several different growth factors acting through receptor tyrosine kinase activation and the MAPK signaling pathway.
In contrast to stimulation by TPA or pervanadate, the proteolytic sensitivity of the ErbB-4 receptor to ligand stimulation has not been reported. It is of particular interest to know if protease cleavage could be a mechanism for down-regulation of activated ErbB-4 receptor in the absence of rapid endocytosis and lysosomal degradation. In this study, we have found that in several tumor cell lines, ligand stimulation of ErbB-4 produces a reduction of the ErbB-4 receptor level and accumulation of an 80-kDa fragment, suggesting that a proteolytic cleavage of the ErbB-4 receptor occurs. Evidence that this cleavage may occur in an intracellular compartment is presented.
Cell Culture-The human ovarian carcinoma cell lines OVCA429 and OVCA432 were obtained from Dr. Robert Bast (M. D. Anderson Hospital), and OVCAR3 cells were a gift from Dr. Roy Jensen (Vanderbilt University). The human lung carcinoma cell lines H661 and H1155 and the human breast carcinoma line T47D were purchased from American Tissue Culture Collection. These cells were routinely grown in 5% CO 2 at 37°C in Dulbecco's modified Eagle's medium (DMEM) containing 10% fetal bovine serum and penicillin/streptomycin (Life Technologies, Inc.). The rat glioma cell line C6 was kindly provided by Dr. Mary Ann Thompson (Vanderbilt University), and the cells were grown in medium F-10 containing 15% equine serum and penicillin/streptomycin. SK-Br-3 cells were obtained from Dr. Matthias Kraus (European Institute of Oncology) and were grown in McCoy's medium containing 10% fetal bovine serum. T47-14 cells, transfected NIH 3T3 cells that overexpress human ErbB-4, have been described elsewhere (17). The growth medium for T47-14 cells was DMEM containing 10% calf serum and penicillin/streptomycin. Atrial tumor myocytes (AT-1 cells) derived from T antigen transgenic mice (27,28) were generously provided by D. M. Roden (Vanderbilt University) and were prepared and grown in culture as described previously (29). Experimental cultures were generally grown in 60-or 100-mm diameter culture dishes until nearly confluent.
Cell Lysis and Fractionation-Cell lysates were obtained essentially as described previously (18). Briefly, newly confluent cell monolayers in 60-mm culture dishes were incubated overnight in DMEM containing 0.5% serum. Cells were then rinsed once with DMEM and incubated with the indicated additions for the indicated times at 37°C in DMEM supplemented with 0.1% BSA and 20 mM Hepes (pH 7.2). Next, cells were washed with Ca 2ϩ /Mg 2ϩ -free phosphate-buffered saline and incubated on ice for 30 min in 400 l of TGH lysis buffer (1% Triton X-100, 10% glycerol, 20 mM Hepes (pH 7.2), 100 mM NaCl, 1 mM phenylmethylsulfonyl fluoride, 10 g/ml aprotinin, 10 g/ml leupeptin, 1 mM Na 3 VO 4 , and 50 mM NaF). Lysates were then centrifuged at 14,000 ϫ g for 10 min at 4°C. The resulting supernatant was termed the Triton X-100-soluble fraction. The pellet was dissolved in TGH buffer supplemented with 0.5% SDS in a volume equal to its supernatant or as indicated. The solution was then passed through a syringe with a 23-gauge needle five to eight times to shear DNA and was termed the Triton X-100-insoluble fraction. Prior to immunoprecipitation, the SDS concentration was adjusted to 0.1%.
Immunoprecipitation and Immunoblotting-For ErbB-4 immunoprecipitation, 1 mg of cell lysate from the Triton X-100-soluble fraction or an equal volume of the Triton X-100-insoluble fraction was incubated with 3 g of anti-ErbB-4 polyclonal antibody for 3 h at 4°C before incubation with protein A-Sepharose for 1 h at 4°C. Subsequently, the immunocomplexes were washed three times with TGH buffer and resuspended in Laemmli buffer. After boiling, proteins in the samples were electrophoretically separated on a reducing SDS-7.5% polyacrylamide gel and transferred to nitrocellulose membranes for Western blotting. Membranes were blocked with 5% milk in TBST buffer (50 mM Tris (pH 7.4), 150 mM NaCl, and 0.05% Tween) for 1 h. For antiphosphotyrosine blotting, membranes were blocked with 3% BSA in TBST buffer for 1 h. The membranes were then incubated with either ErbB-4 antiserum or anti-phosphotyrosine polyclonal antibody in TBST buffer containing 1% BSA for 2 h at room temperature. After the antibody incubation, the membranes were washed three times with TBST buffer for 30 min and incubated with horseradish peroxidaseconjugated protein A in TBST buffer containing 1% BSA at room temperature. Bound antibodies were detected by ECL. For direct Western blots, a lysate aliquot of ϳ100 g was loaded onto SDS-7.5% polyacrylamide gels, electrophoresed, transferred to nitrocellulose membranes, and blotted as described above.
Potassium Depletion-The use of potassium depletion conditions to prevent internalization by clathrin-coated pits has been described elsewhere in detail (30,31). In brief, cell monolayers incubated overnight in low serum were washed once with depletion buffer (50 mM Hepes (pH 7.4), 100 mM NaCl, 1 mM CaCl 2 , and 1 mM MgCl 2 ) and incubated in DMEM/H 2 O (1:1) for 5 min. Cells were then washed with depletion buffer once again and incubated in depletion buffer for 2 h at 37°C. After incubation in depletion buffer, cells were incubated with heregulin ␤1 or TPA in either depletion buffer containing 0.1% BSA or in regular medium (DMEM containing 0.1% BSA, and 20 mM Hepes) as the control.

Ligand-induced Accumulation of the 80-kDa ErbB-4 Fragment-
To determine whether growth factor binding to the ErbB-4 receptor tyrosine kinase initiates proteolytic cleavage of this receptor, as previously reported for TPA (19) and pervanadate (20), several cell lines were surveyed. These included two breast carcinoma cell lines (T47D and SK-Br-3), three ovarian carcinoma cell lines (OVCAR3, OVCA432, and OVCA429), two lung carcinoma cell lines (H661 and H1155), one glioma cell line (C6), a mouse cardiomyocyte cell line (AT-1), and the NIH 3T3-derived cell line T47-14. These cells were chosen on the basis of literature reports indicating that they express moderate to high levels of ErbB-4, which is not commonly expressed in cell lines. The presence of ErbB-4 in these lines was confirmed by Western blotting, and each was tested for heregulin-induced cleavage of the ErbB-4 receptor using an antibody to a cytoplasmic domain epitope. As shown in Fig. 1A, heregulin-induced the accumulation of an 80-kDa ErbB-4 fragment in three cell lines: T47D, OVCAR3, and OVCA432. A similar-sized fragment of ErbB-4 was produced by treatment of these cells with TPA, as reported previously for other cell lines (19). In the other cell lines tested, no heregulin-induced fragment was detectable, although TPA-induced ErbB-4 cleavage was always detectable, and heregulin did induce ErbB-4 tyrosine phosphorylation (data not shown). Subsequent experiments were conducted with T47D cells.
The time course of heregulin-induced accumulation of the 80-kDa fragment in T47D cells is shown in Fig. 1B. Increased levels of the 80-kDa fragment were detectable at 15 min after the addition of the growth factor and were maximal at 60 min. Thereafter, the amount of the 80-kDa ErbB-4 fragment decreased. Beginning at 60 min after heregulin addition, there was also a decrease in the level of the native form of ErbB-4. This time course is similar to that reported previously for TPAor pervanadate-induced cleavage (19,20). From the data in Fig.  1A, it is clear that TPA produces a more substantial level of ErbB-4 cleavage than does heregulin. Scanning densitometry indicates that at 60 min, the amount of the 80-kDa fragment in cells treated with heregulin is equal to 20% of the total cellular content of ErbB-4.
Several ligands have been reported to bind to ErbB-4 and to act as activators of this receptor. These include the ␣ and ␤ forms of heregulin (10, 11), betacellulin (12), and epiregulin (32). The heregulins also recognize ErbB-3, whereas betacellulin and epiregulin also bind to the EGF receptor. Therefore, we tested these ligands for their capacity to stimulate ErbB-4 cleavage. As shown in Fig. 2A, accumulation of the 80-kDa ErbB-4 fragment was enhanced by treatment of the cells with heregulin ␤, betacellulin, or heregulin ␣, in the approximate order of efficacy. As shown in Fig. 2B, each of these ligands also stimulated ErbB-4 tyrosine phosphorylation in the same cells. Epiregulin or EGF, a negative control, did not alter the basal level of this fragment, and as shown in Fig. 2B, neither of these two growth factors stimulated ErbB-4 autophosphorylation.
Translocation of ErbB-4 to the Detergent-insoluble Domain-ErbB-1 (EGF) receptors have been reported to be present in both the detergent (Triton X-100)-soluble and -insoluble fractions of fibroblasts (33), A-431 cells (34), and PC12 cells (35), whereas ErbB-4 presence in both of these fractions has been reported for cardiomyocytes (ErbB-4) (36). The addition of ligand has been reported to promote the migration of ErbB-1 (34) and ErbB-4 (36) out of these detergent-insoluble domains. The exact nature of the detergent-insoluble domains is unclear, and in some cases, they are referred to as caveolae (37).
We have examined ErbB-4 movement between these different fractions to determine whether heregulin-induced ErbB-4 cleavage is associated with these trafficking events. The results, shown in Fig. 3A, demonstrate that in the absence of heregulin, all detectable ErbB-4 molecules were present in the detergent-soluble fraction (compare lanes 1 and 7). When heregulin was added to T47D cells, there was a rapid and transient increase in the amount of ErbB-4 in the detergentinsoluble fraction; however, no 80-kDa ErbB-4 fragment was detectable in this fraction (lanes 2-6). In contrast, the 80-kDa fragment transiently accumulated in the detergent-soluble fraction following ligand addition (lanes 8 -12).
The data shown in Fig. 3A were quantitated to show the influence of time on the amount of native ErbB-4 that appeared in the detergent-insoluble fraction and the amount of the 80-kDa fragment that was found in the detergent-soluble fraction (Fig. 3C). Clearly, the native 180-kDa molecule appeared in the detergent-insoluble fraction more rapidly than the 80-kDa fragment was found in the detergent-soluble fraction following heregulin addition. These data do not suggest that translocation of ErbB-4 to the detergent-insoluble fraction is the mechanism by which proteolytic fragmentation occurs. It is possible, however, that the 80-kDa fragment is produced in the detergent-insoluble fraction and very rapidly migrates out of this fraction. To test this, we also determined whether TPA or pervanadate treatment of T47D cells, each of which provokes more extensive fragmentation of ErbB-4 than heregulin, induced translocation of ErbB-4 to the detergent-insoluble fraction. As shown in Fig. 4, neither TPA nor pervanadate, in contrast to heregulin, brought about detectable levels of ErbB-4 in the detergent-insoluble fraction.
This ligand-dependent distribution of ErbB-4 between the detergent-soluble and -insoluble fractions was also examined for the tyrosine phosphorylation of ErbB-4, as shown in Fig. 3B. Following heregulin addition, tyrosine-phosphorylated ErbB-4 rapidly appeared in both the detergent-insoluble and -soluble fractions. Tyrosine phosphorylation of the 80-kDa fragment was not detectable in this system. We have used scanning densitometry to compare the level of tyrosine phosphorylation at each time point in both fractions with the amount of ErbB-4 present (Fig. 3D). These results indicate that the ErbB-4 present in the detergent-insoluble fraction is more highly phosphorylated (ϳ5-fold) than the ErbB-4 present in the detergent- Characteristics of Heregulin-dependent ErbB-4 Cleavage-The basal level of cleavage of ErbB-4 in untreated cells (18) as well as the TPA-induced (19) or pervanadate-induced (20) cleavage of this receptor are all sensitive to BB-94, a broadspectrum metalloprotease inhibitor that blocks the ectodomain cleavage of a number of cell-surface proteins. However, the heregulin-induced cleavage of ErbB-4 was not diminished by preincubation of the cells with BB-94, as shown in Fig. 5A. In this experiment, BB-94 did prevent ErbB-4 cleavage induced by TPA. Similar results were obtained with the metalloprotease inhibitor N-d-l-[2-(hydroxyaminocarbonyl)-methyl]-4methylpentanoyl-1-3-(tert-butyl)-alanyl-1-alanine, 2-aminoethylamide (data not shown) for both heregulin-and pervanadate-induced ErbB-4 cleavages. Also, other metalloprotease inhibitors such as EDTA, BB-2116, BB-3105, and 1,10-phenanthroline did not influence heregulin-induced ErbB-4 cleavage under these conditions.
Since the protein kinase C activator TPA is a potent stimulator of ErbB-4 cleavage as well as other cell-surface proteolytic events and is likely activated by heregulin binding to ErbB-4, we used the protein kinase C inhibitor GF109203X to test whether protein kinase C is required for heregulin-induced ErbB-4 cleavage. As shown in Fig. 5B, GF109203X blocked TPA-induced ErbB-4 cleavage, but did not alter heregulininduced fragmentation of the receptor.
The results with metalloprotease inhibitors were surprising in view of the similar-sized ErbB-4 fragment produced by heregulin, TPA, or pervanadate. This suggested either that the heregulin-induced cleavage is metalloprotease-independent or that perhaps receptor trafficking, induced uniquely by the growth factor, might modify the sensitivity of the cleavage to the metalloprotease inhibitors. Following ligand binding, ErbB-4 receptors are slowly endocytosed compared with occupied EGF receptors (17). However, this slowed rate of internal- ization might be sufficient to accommodate the rate and extent of heregulin-induced cleavage of ErbB-4. A widely employed technique to prevent internalization of receptors is the K ϩ depletion method, which prevents the formation of clathrincoated pits (30,31). The data presented in Fig. 6 show that K ϩ depletion effectively prevented heregulin-induced ErbB-4 cleavage (panel A), but did not prevent cleavage induced by  TPA (panel B).
The data obtained with K ϩ depletion implicate the endocytic pathway as a means by which ErbB-4 trafficking may be required for heregulin-induced cleavage of the receptor. Therefore, we assessed several inhibitors that prevent the acidification of endocytic vesicles. These include NH 4 Cl and chloroquine, which enter endosomes, are protonated, and thereby prevent acidification of these vesicles (38,39), plus folimycin, which inhibits the ATPase pump that is required for endosome acidification (40). The results with these agents are shown in Fig. 7. The data in Fig. 7 (left panel) show that each of these inhibitors raised the basal level of ErbB-4 cleavage, such that the amount of heregulin-dependent cleavage was less and more difficult to detect. These results suggest that the basal level of cleavage of ErbB-4 is enhanced within the microenvironment of an intracellular vesicle whose pH is not acidified. Next, the sensitivity of heregulin-induced cleavage to BB-94 was tested in the absence and presence of these inhibitors of endosome acidification. As shown in Fig. 7 (right panel), heregulin-induced ErbB-4 cleavage was not detectable in the presence of BB-94 and NH 4 Cl, chloroquine, or folimycin. Therefore, under these conditions, the ligand-dependent cleavage of ErbB-4 is sensitive to metalloprotease inhibition. DISCUSSION A substantial number of transmembrane proteins are subject to ectodomain cleavage by metalloproteases, and among these are several growth factor receptor tyrosine kinases (41,42). In addition to the heregulin receptor ErbB-4, this includes TrkA, the nerve growth factor receptor (43), the orphan receptor Tie-1 (44), fibroblast growth factor receptor-1 (45), and the co-receptor ErbB-2 (46), which heterodimerizes with other ErbB receptors including ErbB-4. In each case, receptor cleavage is promoted by TPA and/or pervanadate and is inhibited by a metalloprotease inhibitor. In addition to these, there are additional receptor tyrosine kinases whose ectodomains are shed when TPA is added to cells, but in these cases, metalloproteases have not yet been tested for involvement.
Although the biological significance of ectodomain cleavage is clear for some membrane proteins such as growth factor/ cytokine precursors and certain adhesion molecules (24,41,47), the physiological importance of receptor cleavage is less understood. In some cases, activation of protein kinase C by heterologous ligands can lead to receptor cleavage and may represent a means of receptor cross-talk. For example, the platelet-derived growth factor acts in a protein kinase C-dependent manner to facilitate the cleavage of ErbB-4 (19). In only two cases has it been reported that the homologous ligand can promote cleavage of its receptor. The nerve growth factordependent cleavage of TrkA in PC12 cells has been reported (43), but not characterized. In this study, we report and characterize the cleavage of ErbB-4 by its ligands.
Based on the similarity of the 80-kDa fragments produced by TPA-induced (19), pervanadate-induced (20), and heregulininduced ErbB-4 cleavage, it would be expected that all three agonists would act through the metalloprotease TACE, which has very recently been shown to be required for ErbB-4 cleavage by TPA (21). However, heregulin-induced cleavage is insensitive to the metalloprotease inhibitor BB-94 under the same conditions that TPA-or pervanadate-stimulated cleavage is blocked by BB-94. This difference is not particular to BB-94, as other metalloprotease inhibitors show the same pattern.
Growth factor binding to receptors not only elicits the activation of signal transduction pathways, but also results in the modulation of receptor distribution on the cell surface as well as receptor trafficking into cells through endocytosis. Data on ErbB-4 in cardiomyocytes (36) and the EGF receptor (ErbB-1) in fibroblasts (33) and PC12 cells (35) show that in the absence of their ligand, a fraction of these receptors is localized to caveolae (33,36) or a caveola-like detergent-insoluble fraction (35); and then following ligand addition, the receptors rapidly migrate out of this fraction. Our data indicate that in several tumor cell lines, ErbB-4 is entirely present in the detergentsoluble fraction in the absence of its ligand and is rapidly and transiently translocated to a detergent-insoluble fraction following heregulin addition. The nature of this fraction is not clear, but seems akin to detergent-insoluble domains or rafts recorded in several receptor systems (37,48). The T47D cells employed do not express detectable levels of caveolin; hence, this fraction does not represent caveolae.
Interestingly, the data show that ErbB-4 is hyper-tyrosinephosphorylated in this detergent-insoluble fraction compared with the receptor in the detergent-soluble fraction. Analogously, the T cell receptor is observed to enter detergent-insoluble lipid rafts following activation and is also hyper-phosphorylated (49). In that system, preliminary studies suggest that phosphotyrosine phosphatases are excluded from these receptor-containing membrane microdomains, and this may explain, in part, the observed receptor hyperphosphorylation (50). This and other aspects of ErbB-4 translocation to detergent-insoluble domains are currently being investigated.
Our present data do not suggest that the translocation to the detergent-insoluble fraction is part of the mechanism of heregulin-dependent ErbB-4 cleavage. However, we also cannot completely discount its involvement due to the following correlation. We have assayed eight ErbB-4-expressing cell lines fortranslocationtothedetergent-insolublefractionandheregulindependent cleavage. In three cell lines (T47D, OVCAR3, and OVCA432), there is both heregulin-dependent cleavage and translocation. In the other five lines, neither cleavage nor translocation is observed.
Our data do indicate that endocytic processing of the heregulin⅐ErbB-4 complex is necessary for ligand-induced cleavage of ErbB-4. This is based, in part, on the fact that K ϩ depletion, a standard technique employed to disorganize clathrin-coated pits (31,32), prevents heregulin-induced ErbB-4 cleavage. Of course, K ϩ depletion may have unknown pleiotropic effects. Therefore, we investigated other agents that modify the endocytic process. When cells are treated with any of three chemicals that modify the acidic pH of endosomes and that prevent endosomal trafficking of the receptor, but not receptor internalization, the basal level of ErbB-4 cleavage is increased, and surprisingly, the cleavage is sensitive to BB-94 under these conditions. We interpret these results to indicate that heregulindependent trafficking of ErbB-4 leads to its slow internalization and that in an early endocytic compartment, prior to acidification, cleavage of ErbB-4 by a BB-94-sensitive metalloprotease occurs. It has been suggested elsewhere that activation of proteolysis by metalloproteases could be due to topological factors that influence the physical relationship of protease and substrate (24,51). It is unclear why the observed cleavage is not sensitive to BB-94 in the absence of inhibitors of endosome acidification.
The subcellular distribution of metalloproteases and TACE, in particular, has not been well defined. There is evidence of the presence of TACE and other metalloproteases within intracellular compartments, such as the Golgi, related to their biosynthesis as transmembrane proteins (52). It is clear that these proteases may be active within such compartments. For example, TACE mediates tumor necrosis factor cleavage, which has been reported to occur in a post-endoplasmic reticulum compartment (53). Localization of TACE in endocytic compartments derived from clathrin-coated pits has not been described. However, the ectodomain cleavage of the amyloid precursor protein by an ␣-secretase recently discovered to be ADAM-10, a protease related to TACE (ADAM-17), does occur in endosome compartments subsequent to internalization of this protein (54 -57).
In some systems, growth factors are able to activate the metalloprotease-dependent cleavage of target plasma membrane proteins. For example, EGF induces the cleavage of the transmembrane form of the transforming growth factor by TACE (23, 25) as well as the metalloprotease-dependent ectodomain shedding of syndecan-1 and -4 (58). These data indicate the existence of tyrosine kinase-initiated signaling pathways that promote ectodomain shedding. In the case of EGF-induced processing of the transforming growth factor, this regulation is attributed to signaling from the Ras/MAPK cascade and not to the activation of protein kinase C (25). That ErbB-4 cleavage may be regulated by tyrosine phosphorylation events was indicated by our previous results with pervanadate (20). The experiments now reported show that heregulin-induced cleavage of ErbB-4 is independent of protein kinase C activation, as was shown previously for the pervanadate-stimulated cleavage (20). At present, however, we have not ruled out whether activation of other ErbB-4 signaling pathways, such as the Ras/MAPK cascade, is involved in this proteolytic event either directly or indirectly through modulation of ErbB-4 trafficking.