Roles of Meltrin β/ADAM19 in the Processing of Neuregulin*

Meltrin β/ADAM19 is a member of ADAMs (a d isintegrin andmetalloproteases), which are a family of membrane-anchored glycoproteins that play important roles in fertilization, myoblast fusion, neurogenesis, and proteolytic processing of several membrane-anchored proteins. The expression pattern ofmeltrin β during mouse development coincided well with that of neuregulin-1 (NRG), a member of the epidermal growth factor family. Then we examined whether meltrin β participates in the proteolytic processing of membrane-anchored NRGs. When NRG-β1 was expressed in mouse L929 cells, its extracellular domain was constitutively processed and released into the culture medium. This basal processing activity was remarkably potentiated by overexpression of wild-type meltrin β, which lead to the significant decrease in the cell surface exposure of extracellular domains of NRG-β1. Furthermore, expression of protease-deficient mutants of meltrin β exerted dominant negative effects on the basal processing of NRG-β1. These results indicate that meltrin β participates in the processing of NRG-β1. Since meltrin β affected the processing of NRG-β4 but not that of NRG-α2, meltrin β was considered to have a preference for β-type NRGs as substrate. Furthermore, the effects of the secretory pathway inhibitors suggested that meltrin β participates in the intracellular processing of NRGs rather than the cleavage on the cell surface.

Various intracellular signaling and adhesion molecules govern the cell-cell interactions during the development of multicellular organisms. The actions of these molecules are regulated not only by transcriptional and translational controls but also by post-translational modifications such as phosphorylations and proteolytic processings. Numerous membrane-anchored signaling molecules are subjected to proteolytic processing to release their extracellular domains. Such modifications may cause qualitative and irreversible changes in the functions of these molecules. ADAMs 1 (a disintegrin and metalloproteases; also known as MDC proteins, metalloprotease/disintegrin/cysteine-rich proteins) are a family of membrane-anchored glycoproteins (1,2) which play important roles in sperm-egg binding and fusion (3,4), muscle cell fusion (5), neurogenesis (6), and development of various epithelial tissues (7). At present, more than 30 ADAM cDNAs have been cloned from various species. Since more than half of these have a catalytic site consensus sequence for metalloproteases (HEXGHXXGXXHD), they are predicted to be catalytically active proteases. Genetic and biochemical evidence indicate that some ADAMs participate in the processing of the extracellular domain of membrane-anchored proteins. TACE (tumor necrosis factor-␣ converting enzyme)/ADAM17 was initially identified as the protease responsible for the processing of pro-tumor necrosis factor-␣ (8,9). Furthermore, studies on the disruption of the mouse TACE gene demonstrated that TACE is involved in the processing of extracellular domains of several membrane-anchored proteins including tumor necrosis factor p75 receptor, the adhesion molecule L-selectin, amyloid precursor protein, and transforming growth factor-␣ (7,10). Kuzbanian/ADAM10 is involved in the neurogenesis of Drosophila (6), and processes and releases a soluble form of Delta, a Notch ligand (11). Recently, it has been reported that meltrin ␥/ADAM9 is involved in the processing of heparinbinding EGF-like growth factor (12). Mouse Kuzbanian and meltrin ␥ also cleave amyloid precursor protein (13,14). These findings strongly suggest potential roles of ADAM metalloproteases in the proteolytic processing of various membrane-anchored proteins.
NRGs (also known as acetylcholine receptor inducing activity, glial growth factor, heregulin or neu differentiation factor) are a group of growth factors that are members of the EGF family. NRGs mediate an array of biological effects, including the synthesis of acetylcholine receptors in skeletal muscle (19) and the stimulation of Schwann cell growth (17). These biological effects of NRGs are mediated by the ErbB family of tyrosine kinase receptors (20,21). Gene disruption studies indicate that NRGs are essential for early heart and central nervous system development (22). A variety of different protein isoforms are produced from the single NRG gene via alternative splicing mechanisms. All isoforms contain an EGF-like domain sufficient for biological activity. Although alternatively spliced transcripts also generate some secreted isoforms (17), most soluble NRGs are derived from membrane-anchored precursor proteins via proteolytic cleavage of the extracellular region including EGF-like domain. It has been reported that this processing occurs in intracellular organellas (23). However, the nature of the processing enzyme remains elusive.
In this study, we examined whether meltrin ␤ participates in the processing of membrane-anchored NRGs. First, both meltrin ␤ and NRG proteins were expressed in DRG neurons at the same stages of mouse embryogenesis. Next, overexpression of wild-type meltrin ␤ significantly increased the release of soluble NRGs in culture medium and decreased the cell surface expression of the extracellular domains of NRG-␤1. Furthermore, the processing of NRGs was abrogated by expression of protease-deficient mutants of meltrin ␤. Finally, the enhanced processing of NRGs by meltrin ␤ was blocked by the treatment with brefeldin A but not by monensin, which suggested the action of meltrin ␤ in the Golgi apparatus. Taken together, we concluded that meltrin ␤ (or similar ADAM proteases) participates in the cleavage of membrane-anchored NRGs.
Cell Culture-P19 rat embryonic carcinoma cells were cultured in minimum essential medium ␣ medium supplemented with 10% fetal bovine serum. Mouse L929 fibroblasts were cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum. All cultures were maintained at 37°C in the presence of 5% CO 2 . To induce differentiation, P19 cells were cultured on bacterial grade dishes to form aggregates for 4 days in the presence of 1 M retinoic acid (Sigma) and then replated on tissue culture grade dishes in growing medium.
Expression Plasmids and Transfection-The full-length mouse NRG cDNAs were isolated using the primers corresponding to the nucleotide sequences (5Ј-GGCTCTAGACATGTCTGAGCGCAAAGAAGGCAG-3Ј and 5Ј-GGCTCTAGATTATACAGCAATAGGGTCTTGGTTAGC-3Ј) from murine neonatal muscle and E12.5 mouse embryo trunk cDNAs. The mouse NRG cDNAs were fused with a synthetic DNA cassette coding for the hemagglutinin (HA)-epitope tag (MYPYDVPDYA) and subcloned into pEF-BOS, which has the promoter region of the human EF-1␣ chromosomal gene (24), to obtain pEF-BOS-HA-NRGs. A couple of protease-deficient (E347Q and H346A,H350A) meltrin ␤ cDNAs and a metalloprotease domain-deleted (⌬MP) meltrin ␤ cDNA were constructed by mutagenesis based on a PCR technique using mutated primers. In E347Q mutant meltrin ␤, glutamine is substituted for the conserved glutamic acid at position 347. In H346A,H350A mutant meltrin ␤, alanines are substituted for the conserved histidines at positions 346 and 350. In ⌬MP meltrin ␤, amino acid residues 208-430 are deleted. The nucleotide sequences of the mutants were confirmed by direct sequencing. The cDNAs of wild-type and mutant meltrin ␤ were subcloned into pEF-BOS. Wild-type meltrin ␥ was also subcloned into pEF-BOS (12). pBIE plasmid was generated by deletion of the human cytomegalovirus promoter region of pIRES2-EGFP (CLONTECH) and replaced by the promoter region of pEF-BOS. Wild-type and ⌬MP meltrin ␤ cDNAs were subcloned into pBIE to generate pBIE-meltrin ␤ and pBIE-⌬MP meltrin ␤, respectively. These plasmids were transfected by the Lipo-fectAMINE PLUS method according to the manufacturer's instructions (Life Technologies Inc.).
Metabolic Labeling of Cells-Cells were starved in medium lacking methionine and cysteine (ICN) for 1 h and pulse-labeled with [ 35 S]methionine and -cysteine (EASYTAG express protein labeling mixture, PerkinElmer Life Sciences) at 0.1 mCi/ml. After a 1-h pulse, cells were either extracted in extraction buffer containing Complete TM protease inhibitor mixture or chased with Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum. Where indicated, the secretory pathway inhibitors, brefeldin A (10 g/ml, Wako) and monensin (2 M, Wako), were added during the chase period. After a 5-h chase, cells were extracted in extraction buffer. Cell extracts were clarified by centrifugation at 15,000 ϫ g for 20 min. The supernatants were incubated with anti-HA monoclonal antibody (16B12, Babco) for 30 min on ice. After the addition of protein G-Sepharose beads, the extracts were incubated for 1 h on ice. Immunoprecipitates were washed six times with extraction buffer and subjected to SDS-PAGE.

RESULTS
Coincidental Expression of Meltrin ␤ and NRGs-We previously reported the expression pattern of meltrin ␤ mRNA during mouse embryogenesis (16). Meltrin ␤ mRNA is markedly expressed in the regions where peripheral neuronal cell lineages differentiate including craniofacial and DRG and ventral horns of the spinal cord. In addition, heart, lung, skeletal muscle, and intestine express meltrin ␤ mRNA transiently. This expression pattern of meltrin ␤ coincides well with that of an EGF family growth factor, NRGs (17,18). In this study, we further investigated the precise expression sites of NRGs and meltrin ␤ proteins in the developing mouse nervous system. Adjacent transverse sections through mouse E12.5 embryo were coimmunostained with antibodies against neuronal marker, neurofilament 160 (NF160), and C-terminal domain of meltrin ␤ (Fig. 1, A-C), or EGF-like domain of NRGs (Fig. 1, D-F). As reported previously, high levels of NRG protein were expressed in DRG that give rise to sensory neurons (Ref. 17, Fig. 1E) and the ventral horns of the spinal cord that produce motor neurons (data not shown). Similarly, strong and specific immunoreactivity for meltrin ␤ was observed in the DRG (Fig.  1B) and the ventral horns of the spinal cord (data not shown). This result indicates that meltrin ␤ protein is expressed in the regions where peripheral neuronal cell lineages differentiate during embryogenesis. In addition, immunostainings of both meltrin ␤ and NRGs were detected in most NF160-positive cells in the DRG region (Fig. 1, C and F). These results clearly demonstrate that the majority of NF160-positive neuronal cell lineages in DRG express both meltrin ␤ and NRGs simultaneously.
The mouse embryonic carcinoma P19 cell is a multipotential stem cell, which differentiates into a variety of cell types including neuron and glia. Since this cell line is used as an in vitro model for differentiation of the nervous system, we examined the expression of meltrin ␤ and NRGs in this cell line during differentiation. Differentiation was induced by aggregating P19 cells in the presence of 1 M retinoic acid for 4 days and then the cells were dissociated and plated in the absence of retinoic acid. Cells were harvested at the indicated times, and the cell extracts were subjected to Western blotting using antibodies against the markers of neurons (microtubule-associated protein-2), glial cells (glial fibrillary acidic protein), and smooth muscle cells (smooth muscle actin). The expression of microtubule-associated protein-2 was increased at day 6 and then decreased gradually (data not shown). On the other hand, the expression of glial fibrillary acidic protein was increased during the differentiation period (data not shown). To examine the expression level of NRGs and meltrin ␤ mRNAs, total mRNAs were prepared from cells at the indicated times, and subjected to RT-PCR (Fig. 2). NRGs and meltrin ␤ mRNAs were both expressed at very low levels at day 0. The transcripts of these genes began to appear at day 6, then reached a maximum level at day 8. This similarity in the transcriptional profiles indicates a plausible interaction between meltrin ␤ and NRGs. Many NRGs are derived from membrane-anchored precursor proteins via proteolytic cleavage. We therefore investigated whether the metalloprotease activity of meltrin ␤ is involved in the processing of NRGs.
Proteolytic Processing of NRGs by Meltrin ␤-Alternative splicing of a single gene gives rise to multiple isoforms of NRG. Many of these encode transmembrane, glycosylated precursors of soluble NRGs. In this study, we used three transmembrane isoforms of NRG (␣2, ␤1, and ␤4) and the domain structure of the NRGs used here is shown schematically in Fig. 3A. The extracellular portion of these NRGs contains an immunoglobulin motif (Ig), a glycosylated spacer domain (Glyco.), and an EGF-like domain (EGF) (25). The two major classes of NRGs diverge in the C terminus of the EGF-like domain giving rise to the ␣and ␤-isoforms. Additional variation is seen in the juxtamembrane region following the EGF domain by the insertion of one of three different sequences (numbered 1, 2, or 4). To detect the ectodomains of NRGs released into the culture medium, the N terminus of NRGs was tagged with HA epitope.
Since DRG neurons express meltrin ␤ and NRG simultaneously (Fig. 1), we first examined the processing of a neuronal type of NRG, NRG-␤1. In this study, we used mouse L929 fibroblast which expresses a low level of endogenous meltrin ␤ (data not shown). L929 cells were transfected with an expression plasmid encoding NRG-␤1 and then the conditioned medium (CM) was subjected to Western blotting using anti-HA antibody. Released soluble NRG-␤1 (ϳ46 kDa) was detected in the CM of NRG-␤1 expressing cells (Fig. 3B, lane 2). This released polypeptide could induce the tyrosine phosphorylation of ErbB2 and -3 when added to differentiated muscle cells, C2C12 (data not shown), which shows that this 46-kDa polypeptide is a functionally mature NRG-␤1. The broad ap- pearance of processed NRG-␤1 band might represent the variety of multiple N-linked and O-linked glycosylation in its spacer region (26). We further investigated whether coexpression of meltrin ␤ affects the release of mature NRG-␤1. Overexpression of wild-type meltrin ␤ considerably increased the release of mature NRG-␤1 (Fig. 3B, lane 3). Western blotting of cell extracts using the anti-C terminus of NRGs antibody showed that overexpression of wild-type meltrin ␤ increased the ratio of processed cytoplasmic tail of NRG-␤1 (74 kDa, open triangle) and decreased the ratio of full-length NRG-␤1 (120 kDa, filled triangle) (Fig. 3C, upper panel, lane 3). These results strongly suggest that meltrin ␤ could potentiate the basal processing activity of NRG-␤1.
To investigate whether the meltrin ␤ protease activity is necessary for the processing of NRG-␤1, several mutants of meltrin ␤ were constructed. In E347Q and H346A,H350A meltrin ␤, glutamine, and alanine residues were substituted for the glutamic acid and histidine residues, respectively, which are essential for the metalloprotease activity. In ⌬MP meltrin ␤, metalloprotease domain is completely deleted. Western blotting using anti-meltrin ␤ antibody revealed two immunoreactive species with apparent molecular masses of 125 and 100 kDa in the cell expressing E347Q meltrin ␤ as shown in the cell expressing wild-type meltrin ␤ (Fig. 3C, lower panel, lanes 3  and 4). The 100-kDa form is considered to be generated by removal of the prodomain from the 125-kDa form, probably by a furin-like pro-protein convertase, which cleaves ADAMs at the sequence motif RXKR in a late Golgi compartment (27,28). Western blotting of the cells expressing H346A,H350A meltrin ␤ revealed mainly the 125-kDa unprocessed form (Fig. 3C,  lower panel, lane 5).
Expression of E347Q meltrin ␤ made no change in the basal processing of NRG-␤1 (Fig. 3, B, lane 4, and C, upper panel,  lane 4). This observation clearly demonstrates that protease activity of meltrin ␤ is essential for the increase of NRG-␤1 processing. On the other hand, expression of H346A,H350A meltrin ␤ remarkably suppressed the release of mature NRG (Fig. 3B, lane 5). At the same time, expression of H346A,H350A meltrin ␤ increased the ratio of the unprocessed form of NRG-␤1 and decreased the ratio of its processed cytoplasmic tail in the cells (Fig. 3C, upper panel, lane 5). Expression of ⌬MP meltrin ␤ decreased the production of NRG-␤1 by unknown reasons (data not shown). However, Western blotting of an increased amount of the extract revealed that expression of ⌬MP meltrin ␤ also increased the ratio of the unprocessed form of NRG-␤1 and decreased the ratio of the processed form of NRG-␤1 (Fig. 3C, upper panel, lane 6). Thus, expression of these mutants of meltrin ␤ exert dominant negative effects on the basal processing of NRG-␤1. Taken together, these results indicate that meltrin ␤ participates in the processing of NRG-␤1 through its metalloprotease activity.
Small proportion of unprocessed membrane-anchored NRGs expose their extracellular domains on the cell surface (23). To investigate the effect of meltrin ␤ on appearance of the extracellular domains on the cell surface, the cells expressing HA-NRG-␤1 together with or without meltrin ␤ were stained with anti-HA antibody under the nonpermeabilized condition. In this experiment, another type of expression plasmids were constructed in which wild-type or ⌬MP meltrin ␤ was expressed together with green fluorescent protein (GFP) by inserting internal ribosomal entry site sequence between cDNAs encoding these proteins. The result shown in Fig. 3D showed that efficient exposure of the N-terminal HA-tag of HA-NRG-␤1 significantly decreased in the cells expressing wildtype meltrin ␤. Such an effect could not be seen in ⌬MP meltrin ␤ expressing cells. Thus, enhanced processing of memrane-  We further investigated whether meltrin ␤ participates in the processing of other isoforms of NRGs. The expression plasmids encoding wild-type or H346A,H350A meltrin ␤ were cotransfected into L929 cells with an expression plasmid encoding NRG-␣2, ␤1, or ␤4. Then the amount of released mature NRGs was determined by Western blotting of CM (Fig. 4A). The amount of mature NRG-␤1 and -␤4 was increased by overexpression of wild-type meltrin ␤ and decreased by expression of H346A,H350A meltrin ␤. On the other hand, the amount of mature NRG-␣2 was not affected by overexpression of either wild-type or H346A,H350A meltrin ␤. We then examined the effect of meltrin ␤ on the stability of full-length NRGs by a pulse-chase experiment. Full-length NRG-␤1 is proteolytically cleaved in the presence of wild-type meltrin ␤ (Fig. 4B, lower  panel, lane 5). This cleavage is dependent on the protease activity of meltrin ␤ (Fig. 4B, lower panel, lane 6). On the other hand, full-length NRG-␣2 is not cleaved by meltrin ␤ (Fig. 4B,  upper panel, lane 5). Taken together, these results demonstrate that meltrin ␤ participates in the processing of ␤-type, but not ␣-type, NRGs.
Recently, it has been reported that meltrin ␥ is involved in the processing of a membrane-anchored growth factor, heparinbinding EGF (12). We examined whether meltrin ␥ also partic-ipates in the processing of NRG-␤1. Coexpression of meltrin ␥ resulted in the release of 30-and 35-kDa HA-containing region of NRG-␤1 into CM (Fig. 4C, arrowheads). These polypeptides are much smaller than the mature soluble NRGs reported previously (20,29). This result indicates that meltrin ␥ cleaves NRG-␤1 in a manner different from meltrin ␤.
Brefeldin A-sensitive and Monensin-insensitive Cleavage of NRG-␤1 by Meltrin ␤-To examine whether meltrin ␤ is localized in the cell surface, we carried out a cell surface biotinylation analysis using cells transfected with meltrin ␤-expressing plasmid. However, we could not detect any surface-exposed meltrin ␤ (data not shown). In the same experiment, the fusion meltrin ␤, which has an exogenous signal sequence of human granulocyte colony-stimulating factor, was efficiently biotinylated (data not shown), thereby excluding the possibility that the result was due to experimental failure. These findings indicate that meltrin ␤ is mainly localized inside of the cell. Since it has been reported that some portions of NRGs undergo intracellular proteolysis (23), we investigated the subcellular compartment in which meltrin ␤ processes NRG-␤1 using two inhibitors of the secretory pathway, brefeldin A and monensin. In the presence of brefeldin A, the NRG-␤1 processing induced by meltrin ␤ was completely blocked (Fig. 5, lane 11). On the other hand, monensin did not block the processing of NRG-␤1 induced by meltrin ␤ (Fig. 5, lane 8). Thus the processing of NRG-␤1 induced by meltrin ␤ is a brefeldin A-sensitive and monensin-insensitive event. Brefeldin A blocks traffic from the endoplasmic reticulum to the Golgi by interfering with anterograde transport from the endoplasmic reticulum to Golgi (30,31). On the other hand, monensin is expected to interfere with the transfer across Golgi compartments and compromise secretion from the trans-Golgi (30,32). Taken together, our results suggest that meltrin ␤ participates in the intracellular cleavage of NRG-␤1 within the Golgi apparatus. DISCUSSION NRGs mediate a variety of biological functions including glial cell development, synaptogenesis, and cardiac development through the activation of the ErbB family of tyrosine kinase receptors (33). Most NRG isoforms encode membraneanchored proteins that generate soluble ligands for the ErbB family by proteolytic cleavages. It is not yet clear, however, whether the functions of NRGs depend on actions of processed and released soluble NRGs or whether the transmembrane form is biologically active. Genetic disruption of only the intracellular domain of membrane-anchored NRG isoforms results in a similar phenotype of embryonic maldevelopment to that observed with disruption of the entire gene (34). Furthermore, deletion of the cytoplasmic tail of membrane-anchored NRGs completely abrogated the release of mature NRGs (34). These results strongly suggest that the proteolytic processings of membrane-anchored NRGs are critical regulatory mechanisms FIG. 4. Substrate specificity of meltrin ␤. A, plasmids encoding several isoforms of NRG were transfected to L929 cells together with plasmid encoding wild-type or H346A,H350A (mut) meltrin ␤. CM was subjected to Western blotting using anti-HA antibody as described in Fig. 3B. B, L929 cells were transfected with plasmids encoding NRG-␣2 (upper panel) or NRG-␤1 (lower panel) and meltrin ␤. Cells were labeled for 1 h with [ 35 S]methionine/cysteine (PerkinElmer Life Sciences) and immediately frozen (0) or chased with cold media for 5 h (5). Cells were extracted, and NRGs were immunoprecipitated from extracts with anti-HA antibody. All samples were subjected to SDS-PAGE. Arrowheads indicate the migration of full-length NRGs. C, cells were transfected with plasmids encoding NRG-␤1 and meltrin ␤ or meltrin ␥. CM was subjected to Western blotting using anti-HA antibody.
FIG. 5. Brefeldin A-sensitive and monensin-insensitive cleavage of NRG-␤1 by meltrin ␤. Plasmids encoding NRG-␤1 were transfected to L929 cells together with plasmid encoding wild-type or H346A,H350A (mut) meltrin ␤. Cells were labeled for 1 h and then chased with cold media for 5 h in the absence or presence of 10 g/ml brefeldin A or 2 M monensin. Cells were extracted, and NRGs were immunoprecipitated from extracts with anti-HA antibody. All samples were subjected to SDS-PAGE and autoradiography. Arrowheads indicate the migration of full-length NRGs. of NRG functions.
In the present study, we provided evidence that meltrin ␤ participates in the processing of ␤-type NRGs. Initially, both meltrin ␤ and NRG proteins were found to be expressed in dorsal root ganglia at the same stages during embryogenesis (Fig. 1). During neurogenic differentiation of P19 cells, the expression of meltrin ␤ and NRGs mRNA was activated in a similar fashion (Fig. 2). Next, overexpression of wild-type meltrin ␤ potentiated the release of mature soluble NRG-␤1 (Fig.  3B, lane 3) with a concomitant decrease in the cell surface expression of extracellular domains of NRG-␤1 (Fig. 3D). The protease activity of meltrin ␤ is indispensable for the potentiation of NRG-␤1 processing (Fig. 3, B and C, lane 4). Furthermore, expression of H346A,H350A or ⌬MP meltrin ␤ remarkably suppressed the release of soluble NRG-␤1 (Fig. 3B, lanes 5  and 6) with a concomitant increase in the ratio of full-length NRG-␤1 and a decrease in the ratio of processed forms of NRG-␤1 in the cells (Fig. 3C, lanes 5 and 6). We further confirmed the enhanced processing of NRG-␤1 with meltrin ␤ protease by the pulse-chase experiment shown in Fig. 4B. These results clearly demonstrate that meltrin ␤ has functional processing activity of NRG-␤1 and that the protease activity of meltrin ␤ is necessary for constitutive processing of NRG-␤1. This is the first report on the function of meltrin ␤ and, at the same time, the first report that indicates the involvement of ADAM metalloproteases in the proteolytic processing of membrane-anchored NRGs. It is considered that meltrin ␤ plays a pivotal role in the development of several organs through the processing of NRGs.
As reported previously, NRG-␣2 is the predominant isoform in mesenchymal cells, whereas NRG-␤1 is the major neuronal isoform (35). The main cleavage sites in these NRG molecules are in exon-␣ and exon-␤, respectively (26). While L929 cells possess endogenous proteolytic processing activities for both ␣and ␤-type NRGs, both overexpression of wild-type and H346A,H350A meltrin ␤ only affected the cleavage of ␤-type NRGs. It is plausible that ␣-type NRG is cleaved by a protease(s) other than meltrin ␤ in L929 cells. Alternatively, L929 cells may lack some regulatory factors that cooperate with overexpressed meltrin ␤ to cleave ␣-type NRG efficiently.
Meltrin ␤ expressed in L929 cells was mainly localized in the Golgi apparatus (data not shown) although intracellular localization of meltrin ␤ remains to be determined precisely. Examinations of the effects of brefeldin A and monensin on the processing revealed that meltrin ␤ participates in the intracellular processing of NRGs, probably in the Golgi apparatus or in monensin-insensitive secretory pathways. Recently, several reports demonstrated that some ADAMs are processed and activated in the trans-Golgi network (27,28), and localized mainly in the Golgi apparatus (13,27). Furthermore, Skovronsky et al. (36) have found activity of TACE and/or Kuzbanian in the trans-Golgi network. These observations and our results indicate that multiple ADAMs function in the trans-Golgi network as intracellular processing enzymes.
Expression of H346A,H350A or ⌬MP meltrin ␤ markedly suppressed the basal processing activity of NRG-␤1 (Fig. 3). Genetic and biochemical characterization of other ADAM proteases also indicated such dominant negative effects of protease-deficient mutants (11)(12)(13)(14)37). In preliminary experiments, we found that small proportion of meltrin ␤ and NRGs expressed in L929 cells could be coimmunoprecipitated (data not shown). H346A,H350A and ⌬MP meltrin ␤ might show dominant-negative effects through the interaction with NRGs, thereby blocking the interaction of endogenous proteases with NRGs. On the other hand, expression of E347Q meltrin ␤ did not affect the basal processing activity (Fig. 3). As shown in Fig.   3C, the prodomain of E347Q meltrin ␤ is removed precisely while those of H346A,H350A and ⌬MP meltrin ␤ are not removed. These meltrin ␤ mutants might have different conformation from wild-type or E347Q meltrin ␤, and their conformational abnormality might affect endogenous meltrin ␤ or similar proteases to act on NRG-␤1. The identification of the domain of meltrin ␤ required for the dominant negative effect on the processing will provide further insight into the mechanism by which meltrin ␤ recognizes and processes NRG-␤1.
Phorbol ester induces the processing of several membraneanchored proteins through the activation of protein kinase C (PKC). As reported previously in other cell types, we found that phorbol ester induces the processing and release of mature soluble NRG-␤1 in L929 cells (Ref. 23, and data not shown). This induced processing was not suppressed by expression of H346A,H350A mutant of meltrin ␤ (data not shown). Our observation indicates that meltrin ␤ accounts for the constitutive processing but not for the PKC-regulated processing of NRG-␤1. Thus, distinct pathways for the processing of NRG-␤1 are suggested: one pathway is dependent on meltrin ␤ protease while, in the other PKC-regulated pathway(s), processing is carried out by other proteases. Several reports have demonstrated that TACE, Kuzbanian, and meltrin ␥ take part in PKC-regulated processing (7,10,12,13). As shown in Fig. 4C, meltrin ␥ is not able to process NRG-␤1 as a mature form. Further studies are warranted to determine whether or not other ADAMs such as TACE and Kuzbanian participate in the PKC-regulated processing of NRG-␤1.
In summary, we showed that meltrin ␤ and NRGs are simultaneously expressed in the nervous system during development and meltrin ␤ participates in the proteolytic processing of ␤-type NRG isoforms which are involved in neurogenesis and synaptogenesis. During differentiation of P19 cells the activation of the meltrin ␤ and NRG genes preceded that of glial fibrillary acidic protein (Fig. 2, data not shown), suggesting regulatory roles of meltrin ␤ in glial cell differentiation through the release of mature NRGs. Further analysis including genetic disruption of meltrin ␤ will be required to demonstrate the role of meltrin ␤ in the development of the nervous system.