Phorbol 12-Myristate 13-Acetate-induced Ectodomain Shedding and Phosphorylation of the Human Meprinβ Metalloprotease*

Shedding of proteins localized at the cell surface is an important regulatory step in the function of many of these proteins. Human meprin (N-benzoyl-l-tyrosyl-p-aminobenzoic acid hydrolase, PPH, EC 3.4.24.18) a zinc-metalloendopeptidase of the astacin family is an oligomeric protein complex of α- and β-subunits and is expressed abundantly in the intestine and kidney as well as in leukocytes of the lamina propria and in cancer cells. In transfected cells intracellular proteolytic removal of the membrane anchor results in the secretion of the meprin α-subunit. In rats and mice, the β-subunit exists in a membrane-anchored form. In contrast, human meprinβ is constitutively converted into a secretable form. We now show that phorbol 12-myristate 13-acetate (PMA) stimulates an increased release of hmeprinβ from transfected COS-1 cells, whereas hmeprinα secretion is not influenced. This stimulatory effect is inhibited by the protein kinase C (PKC) inhibitor staurosporine, suggesting that activation of PKC mediates PMA-induced hmeprinβ shedding. The use of different protease inhibitors shows that two different metalloprotease activities are responsible for the constitutive and the PMA-stimulated hmeprinβ shedding. We identified tumor necrosis factor α-converting enzyme (TACE or ADAM17) as the protease that mediates the PMA-induced release. We also demonstrate that hmeprinβ is phosphorylated by PMA treatment on Ser687 within a PKC consensus sequence in the cytosolic domain of the protein. This phosphorylation of hmeprinβ is not, however, implicated in the enhanced secretion by PMA treatment.

prin is capable of hydrolyzing many substrates in vitro, such as extracellular matrix proteins, hormones, and small peptides like gastrin 17, cerulein, and sCCK 8 of the gastrointestinal tract (6,9,10). Meprin may therefore be involved in the modulation of the activity of important cellular and extracellular proteins and peptides. We have observed elevated levels of meprin in patients with Crohn's disease 1 and colon cancer (11). In addition we have found expression of meprin in leukocytes in the lamina propria of inflamed intestine, where it may be involved in cytokine processing (12). Moreover a novel mRNA isoform (meprin␤Ј) was identified in a variety of human cancer cell lines like MCF7, SK-BR-3, U2OS, and BxPC3 (13).
Studies of recombinant forms of meprin expressed in mammalian cells have yielded important information about the biosynthesis of both subunits (14 -19). In cells transfected with meprin␣, cDNA proteolytic processing of the subunit occurs within the ␣-specific insertion domain. C-cytosolic and transmembrane domains are involved in retention and proteolytic processing of meprin␣ in the endoplasmic reticulum and are essential for subsequent intracellular transport (19). The cleaved ␣-subunit is constitutively secreted into the medium when expressed alone and is largely retained at the cell surface in the presence of the ␤-subunit. Whereas mouse meprin␤ is localized only at the cell surface (5,8), human meprin␤ is found at the cell surface and in a processed form in the culture medium of transfected cells and of cultured intestinal explants (12,18).
We have recently shown that the soluble form of hmeprin␤ is released from transfected cells by cleavage in front of the epidermal growth factor-like domain. A peptide sequence 13 amino acids long is necessary for this cleavage (20). The enzyme(s) involved and the regulation of this release are still unknown. Soluble forms of transmembrane proteins diverse in structure and function are generated by a process referred to as "ectodomain shedding." The enzymes catalyzing this shedding process are named "membrane protein secretases," "membrane protein convertases," or "sheddases" and belong to the superfamily of zinc-dependent proteases that include MMPs 2 (matrix metalloproteases) and ADAMs (a disintegrin and metalloprotease). The release of membrane-bound proteins like ␤-amyloid precursor (21,22), angiotensin-converting enzyme (23,24), transforming growth factor-␣ (25,26), tumor necrosis factor-␣, Fas ligand (27), tumor necrosis factor receptor (28), CD30 (29), and interleukin 6 receptor (30 -32) is usually enhanced by phorbol 12-myristate 13-acetate (PMA), indicating common mechanisms. PMA is thought to activate protein kinase C (33,34), a family of isoenzymes (35,36) by substituting 1,2-diacylglycerol in its binding domain or by binding simultaneously to the C1 region of the kinase (37)(38)(39). These serine/threonine kinases are activated in response to numerous hormones, mitogens, and neurotransmitters (40 -42). In contrast to the ␣-subunit and the homologous mouse meprin␤ subunit, human meprin␤ has potential regulatory elements in the cytosolic domain including two phosphorylation sites (Ser 687 and Ser 688 ) in PKC consensus sequences (18,43). We have now analyzed the regulation of hmeprin␤ secretion. Our results provide evidence that hmeprin␤ cleavage at the cell surface can be markedly enhanced by PMA via the PKC pathway and that the constitutive and inducible release is mediated by two different metalloprotease activities. Furthermore, we provide evidence that tumor necrosis factor-␣ converting enzyme (TACE or ADAM) is the sheddase responsible for the PMA-induced hme-prin␤ release. We also find that PMA induces phosphorylation of hmeprin␤ and we have identified Ser 687 as the main phosphorylation site. This phosphorylation step is not directly involved in the PMA-induced shedding of hmeprin␤. It does, however, provide the potential for a coupling between signaling pathways and meprin activity. The potential of regulated shedding of hmeprin␤ in contrast to the constitutive secretion of hmeprin␣ from the cell surface suggests different biological roles for the two meprin subunits in man.

EXPERIMENTAL PROCEDURES
Materials-[ 35 S]Methionine (1,000 Ci/mmol) was from PerkinElmer Life Sciences, [ 32 P]orthophosphate (carrier-free) from Amersham Biosciences. Cell culture media, penicillin, streptomycin, fetal calf serum, and LipofectAMINE were obtained from Invitrogen. Phenylmethanesulfonyl fluoride, pepstatin, aprotinin, leupeptin, benzamidine, dimethyl sulfoxide (Me 2 SO), and molecular weight standards were from Sigma. PMA and staurosporine, both from ALEXIS Corp. (Lä ufelfingen, Switzerland), were resolved in Me 2 SO and kept as stock solutions of 1 mg/ml and 20 M, respectively. TIMP1, TIMP2, and TIMP3 were obtained from R&D Systems (Minneapolis, MN). TAPI was a gift from Immunex, RO111-3456 a gift from Roche Diagnostics, and BB94 a gift from British Biotech. Endo-␤-N-acetylglucosaminidase, N-glycosidase F, neuraminidase, and O-glycosidase were purchased from Roche Diagnostics. The protein A-Sepharose beads were obtained from Amersham Biosciences. All other chemicals were analytical grade from Merck (Dietikon, Switzerland). Primers used for mutagenesis were synthesized by MWG Biotech (Ebersberg, Germany).
Construction of Hmeprin␤ Mutants-The potential phosphorylation sites in hmeprin␤ for protein kinase C: Ser 687 and Ser 688 were mutated into Arg 687 and Ala 688 (the respective amino acids in mouse meprin␤) by the method of recombinant PCR (44). The mammalian expression vector pPPH␤ (18) was used as a template. Two PCR rounds were performed with the following primers (Ser 687 mutant): upper primer, 5Ј-A-GCGTATGTCCAATCAGC-3Ј; PPH␤ArgL, 5Ј-TGGTCGATTTGATCTC-ATCCTTTCACG-3Ј; and PPH␤ArgU, 5Ј-TCGTGAAAGGATGAGATCA-AATCGACC-3Ј; lower primer: 5Ј-TAATACGACTCACTATAGGGCG-3Ј. Representative fluorographs shown in A and C were densitometrically scanned (B and D). COS-1 cells incubated in the presence (f) and absence (ࡗ) of phorbol 12-myristate 13-acetate (10 ng/ ml). The amount of immunoprecipitable medium forms was determined as described under "Experimental Procedures." Signals from cell extracts and the corresponding media were added and set as 100% of total protein before the distribution of the medium forms were calculated.
In a second round these two PCR products were combined and amplified using the upper and lower primer. Finally the mutated fragment was cut with StuI and NotI and ligated back into the corresponding sites of pPPH␤. The mutations were confirmed by sequencing. The Ser 688 mutation and the Ser 687 -Ser 688 mutation were indentically done with the following primers: PH␤AlaL, 5Ј-TGGTCGATTTGCGCTCATCCTTTCA-CG-3Ј and PPH␤AlaU, 5Ј-GAAAGGATGAGCGCAAATCGACC-3Ј and PPH␤Arg/AlaL, 5Ј-TGGTCGATTTGCTCTCATCCTTTCACG-3Ј and PPH␤Arg/AlaU, 5Ј-TCGTGAAAGGATGAGAGCAAATCGACC-3Ј.
Western Blot Analysis-COS-1 cells transfected with hmeprin␤ were stimulated with PMA (10 ng/ml) for 30 min before analyzing the effects of potential protease inhibitors on hmeprin␤ shedding. After incubation of cells for 6 h in the presence of the inhibitors, the media and cells were collected. The cells were lysed in 1% deoxycholate, 1% Nonidet P-40 in homogenization buffer (50 mM NaCl, 25 mM Tris-HCl, pH 8) containing complete Mini EDTA-free protease inhibitor mixture (Roche) and centrifuged (13,000 rpm, 4°C, 5 min). Aliquots of the media and cells were removed and subjected to 7.5% SDS-PAGE followed by transfer to Hybond-P membrane (Amersham Biosciences). The membrane was blocked in TBST containing 5% blocking agent NIF833 (Amersham Biosciences), followed by incubation with hmeprin␤ antibody raised against its intervening region in blocking solution for 2 h at room temperature. After four washes in TBST, the membrane was incubated with anti-rabbit horseradish peroxidase (1:30000) in blocking solution. The ECL plus detection system (Amersham Biosciences) was used according to the guidelines to detect horseradish peroxidase activity.
Cell Culture and Transfection-COS-1 cells were grown in Eagle's minimal essential medium supplemented with 10% (v/v) fetal calf serum, 100 units/ml penicillin, and 100 g/ml streptomycin at 37°C in 95% air and 5% CO 2 . 5 ϫ 10 5 cells were seeded onto 6-cm dishes 24 h before transfection with DEAE-dextran (45). The transfection mixture (1 g of recombinant DNA dissolved in 50 l of DEAE-dextran stock solution (10 mg/ml H 2 O) and 950 l of NaCl/P i was added to previously washed COS-1 cells. The cells were incubated for 30 min at 37°C before adding 4 ml of the complete culture medium containing 100 M chloroquine and subsequent incubation for 2.5 h at 37°C. The medium was replaced with medium containing 10% (v/v) Me 2 SO followed by a 2.5min incubation at 37°C. Cells were grown in complete medium for 2 days before labeling. TACE(Ϫ/Ϫ) cells kindly provided by R. Black (Immunex) were cultured in Dulbecco's modified Eagle's medium/F-12 without sucrose, supplemented with 1% fetal calf serum, 100 units/ml penicillin, and 100 g/ml streptomycin and were transfected with Li-pofectAMINE exactly as described by the supplier.
Metabolic Labeling of COS-1 and TACE(Ϫ/Ϫ) Cells-Cells were washed twice with NaCl/P i , incubated for 1 h at 37°C in methioninefree medium, and labeled with 50 Ci of [ 35 S]methionine for 1 h at 37°C. After washing twice with methionine-free medium, incubation was done for the indicated times with chase medium (complete medium with 10 mM L-methionine). Cells were harvested and the medium was collected and filtered (0.2 m) before immunoprecipitation. In experiments where PMA and/or the kinase inhibitor staurosporine were applied, labeled cells were treated for 30 min in chase medium containing 10 ng/ml PMA (100 ng/ml PMA in the case of TACE(Ϫ/Ϫ) cells, respectively) alone or together with staurosporine (20 nM), followed by the indicated chase times. In untreated cells addition of Me 2 SO served as control. Labeling with [ 32 P]phosphate was done after starving the cells for 2 h at 37°C in phosphate-free medium before addition of 0.4 mCi/ml [ 32 P]orthophosphate for another 3 h in the presence of 10 ng/ml PMA.
Immunoprecipitation and SDS-PAGE-Immunoprecipitation was conducted as described previously (18,19). In [ 32 P]phosphate-labeled FIG. 1-continued cells, lysis was performed in the presence of 10 mM NaF, 20 nM calyculin, and 1 mM Na 3 VO 4 to inhibit phosphatases, followed by centrifugation in a Microspin column P-30 (Bio-Rad) to remove unincorporated nucleotides before immunoprecipitation. Immunoprecipitated proteins were analyzed on 7.5% polyacrylamide gels under reducing conditions. Electrophoresis, fixation, and fluorography were carried out as described before (19).
Glycosidase Treatments-After immunoprecipitation the samples were treated with endoglycosidase H essentially as described before (18,19) and prepared for electrophoresis.  1-7). Hmeprin␤ was detected with a specific antibody using ECL plus (Amersham Biosciences). Media were ultracentrifuged (100,000 ϫ g) to remove insoluble cell components. B, TACE(Ϫ/Ϫ) cells transfected with hme-prin␤ cDNA were metabolically labeled for 1 h. The PMA-stimulated cells were incubated with/without 50 M TAPI (lanes 2 and 3). After a chase of 6 h hme-prin␤ was immunoprecipitated from cells and media and analyzed by SDS-PAGE. C, Western blot of COS-1 cells transfected with hmeprin␤. The cells were stimulated with PMA for 30 min followed by treatment with TIMP1, TIMP2, and TIMP3 for 6 h. Hmeprin␤ was detected with a specific antibody using ECL plus (Amersham Biosciences).  Fig. 1, followed by a chase of 0 h (A) and 3 h (B). After immunoprecipitation, protein samples were divided and the aliquots were treated with the indicated enzyme(s) before separation on 7.5% SDS-PAGE and analysis by fluorography. C, fluorographs shown in A and B were densitometrically scanned. Protein bands of uncleaved (Ⅺ) and cleaved (f) hmeprin␤ in cell extracts were added and set as 100% of total protein, before the ratio of cleaved and uncleaved protein was calculated.
Densitometric measurements were carried out with Lumi Analyst version 3.0 from Roche. Fluorographic signals were scanned densitometrically with the Lumi Imager F1 from Roche using Roche's Lumi Analyst software version 3.0. Total protein amount (100%) was calculated by adding hmeprin signal from cell extracts and corresponding media.
In COS-1 cells expressing hmeprin␤ the cell fraction consisted of two bands with apparent molecular masses of 105 and 95 kDa (Fig. 1C, lane 1-6). Treatment with PMA resulted in an enhanced secretion of the major species of hmeprin␤ of 95 kDa (Fig. 1C, lane 10 and 12), which was matched by a corresponding decrease of cell-associated protein. PMA-induced stimulation of hmeprin␤ secretion was verified by densitometric analysis (Fig. 1D). In untreated COS-1 cells most of the synthesized hmeprin␤ remained associated with the cellular membrane and only a minor fraction was secreted. After 3 h of chase the amount of hmeprin␤ secreted into the medium was ϳ6.5-fold higher in PMA-treated cells compared with untreated cells Role of Metalloproteases in Secretion of Hmeprin␤-To characterize the enzymes responsible for constitutive and PMAinduced shedding of hmeprin␤, we have analyzed the effects of several protease inhibitors on the shedding of hmeprin␤ in transfected COS-1 cells (Fig. 2). The possibility of autocatalytic cleavage can be ruled out; in transfected cells hmeprin␤ is expressed as a zymogen and therefore enzymatically inactive. Besides the general metalloprotease inhibitors EDTA, 1,10phenanthroline, or dithiothreitol (data not shown) shedding was also inhibited by the MMP-and ADAM-targeting hydroxamate inhibitors RO111-3456 (a derivative of RO31-970), BB94, and TAPI ( Fig. 2A, lanes 4-6). The natural tissue inhibitors of MMPs, TIMP1, TIMP2, and TIMP3 had no effect on hmeprin␤ release ( Fig. 2A, lanes 1-3). The activity of the TIMPs was ascertained in a MMP-2 control assay (data not shown). The inhibition of hmeprin␤ shedding by RO111-3456, BB-94, and TAPI was accompanied by an accumulation of the mature cell-associated form of hmeprin␤ (105 kDa) ( Fig. 2A,  lanes 4-6). These results practically exclude an involvement of MMPs in the constitutive shedding of hmeprin␤.
The hydroxamate-type metalloprotease inhibitor TAPI (47) also inhibited hmeprin␤ release. To test a possible involvement of TACE in hmeprin␤ shedding, TACE(Ϫ/Ϫ) cells were transfected with hmeprin␤ cDNA. These transfected TACE(Ϫ/Ϫ) cells still secreted hmeprin␤ (Fig. 2B, lane 1) but this secretion was unaffected by PMA treatment (Fig. 2B, lane 2). This and the inhibition of hmeprin␤ secretion in TACE(Ϫ/Ϫ) cells by TAPI (Fig. 2B, lane 3) indicates that TACE function is required for PMA inducible shedding but not for the constitutive hme-prin␤ shedding. The constitutive shedding is mediated by a yet unknown metalloprotease. In contrast to unstimulated conditions where TIMP3 had no effect on hmeprin␤ release, the stimulatory effect of PMA in secretion of hmeprin␤ was inhibited by TIMP3 (Fig. 2C, lane 4) (48). As TIMP3 has also been shown to inhibit TACE, these data support the involvement of TACE in the PMA-induced shedding of hmeprin␤.
PMA Treatment Does Not Affect the Ratio of Intracellular Cleavage of Human Meprin␤-To study the effect(s) of PMA treatment on proteolytic processing and secretion of the meprin ␤-subunit, pulse-chase experiments in the presence or absence of PMA followed by subsequent digestion with different endoglycosidases were carried out. The two 105-and 95-kDa forms were detected after a pulse of 1 h with [ 35 S]methionine (Fig. 3A, lanes 1 and 2). The 95-kDa band was sensitive to Endo H treatment and shifted to 78 kDa (Fig. 3A, lanes 3 and 4), indicating that this was the high mannose form of hmeprin␤. The 105-kDa band was partially resistant to Endo H digestion and represents a complex glycosylated form of hmeprin␤ that still contains some high-mannose glycans. After digestion with N-glycosidase F we observed three bands with molecular masses of 83, 78, and 70 kDa, respectively (Fig. 3A, lanes 5 and  6). These represent uncleaved and cleaved forms of hmeprin␤ with differing glycosylation status. Complete deglycosylation with N-glycosidase F/neuraminidase/O-glycosidase yielded two forms at 78 and 68 kDa, respectively (Fig. 3A, lanes 7 and 8). The same molecular weight pattern was observed after PMA treatment of these cells. After 3 h of chase secreted hmeprin␤ was detectable in the culture medium (Fig. 3B, lanes 9 and 10). This secreted form did not contain O-glycans as seen by the insensitivity to O-glycosidase digestion (Fig. 3B, lanes 15 and 16).
To analyze the ratio of cleaved and uncleaved hmeprin␤ in cell extracts in the presence and absence of PMA only the signals after complete deglycosylation (Fig. 3, A and B, lanes 7  and 8) were compared densitometrically (Fig. 3C). PMAtreated cells thus showed no significant differences in the ratio of proteolytic processing of hmeprin␤. These results indicate that the intracellular ratio of uncleaved and cleaved protein is unchanged by PMA treatment, whereas the ectodomain shed- ding at the cell surface is enhanced after phorbol ester incubation (Fig. 3B, lanes 15 and 16).
Involvement of PKC in the PMA-induced Release of Human Meprin␤-The involvement of protein kinase in the PMA-induced generation of soluble hmeprin␤ was validated using the general protein kinase inhibitor staurosporine, one of the most potent protein kinase C inhibitors (49, 50) (Fig. 4). Unstimulated secretion of hmeprin␤ occurred at a low level (Fig. 4, lanes  10 and 13). This secretion was strongly enhanced by PMA (Fig. 4,  lanes 11 and 14), and the induced release was reduced to basal levels of secreted protein by staurosporine (Fig. 4, lane 12 and  15). These findings suggest that activation of a protein kinase, presumably PKC, mediates PMA-induced hmeprin␤ shedding.
PMA Treatment Induces Phosphorylation of Human Me-prin␤-To clarify the action of PMA, we investigated the phosphorylation of hmeprin␤ in transfected COS-1 cells. The cytoplasmic domain of hmeprin␤ contains PKC consensus sequences with two potential PKC phosphorylation sites (Ser 687 and Ser 688 ). COS-1 cells were transiently transfected with hmeprin and metabolically labeled with [ 32 P]orthophosphate in the presence of PMA followed by immunoprecipitation with a polyclonal antibody that recognizes both meprin subunits (Fig. 5). In COS-1 cells expressing hmeprin␣ (the cytoplasmic tail of hmeprin␣ lacks a PKC phosphorylation sequence) no incorporation of [ 32 P] was observed (Fig. 5, lane 6). In contrast, COS-1 cells transiently transfected with the ␤-sub-unit of hmeprin led to a 105-kDa precipitable phosphorylated protein after PMA treatment (Fig. 5, lane 7). By labeling COS-1 cells transfected with mutant constructs lacking the potential phosphorylation sites, we identified Ser 687 as the main 32 Pacceptor. These data provide evidence that the cytoplasmic domain of hmeprin␤ can function as a PKC substrate.
Up-regulation of Hmeprin␤ Release Is Independent of the Cytoplasmic Tail or Its Phosphorylation-To study if the phosphorylation of hmeprin␤ is associated with the regulation of its shedding, we analyzed the potential of PMA to induce secretion of hmeprin␤ containing the mutated phosphorylation sites. Elimination of Ser 687 and Ser 688 had no considerable influence on the enhanced shedding of hmeprin␤ following PMA stimulation (Fig. 6, A and B). This result supports the view that the release of human meprin␤ is independent of its phosphorylation status and that shedding occurs by an ADAM (in all probability TACE) whose activity is stimulated by PMA via the PKC pathway. PMA treatment of cells transfected with a tail switch mutant where the transmembrane and cytosolic domains were replaced by the homologous counterparts from the ␣-subunit (19), also resulted in induced shedding of the hme-prin␤-ectodomain into the medium (data not shown). This clearly shows that neither the cytoplasmic domain nor the phosphorylation site contained herein is involved in the PMA induced up-regulation of hmeprin␤ secretion.

DISCUSSION
We have previously reported that meprin␣ and meprin␤ are secreted in a polarized fashion from transfected Madin-Darby canine kidney cells (14,18). Secretion of the meprin␤ subunit was specific to the human enzyme, as mouse meprin␤ was not secreted. Proteolytic processing of hmeprin␤ occurs intracellularly and at the cell surface and requires a stretch of 13 amino acids (QIQLTPAPSVQDL) in the intervening sequence (20). This is in contrast to hmeprin␣, which is cleaved in the endoplasmic reticulum by a process that is dependent on the ␣-spe- cific insertion domain and the transmembrane region (15-17, 19, 51). In the present study we show that secretion of hme-prin␤, in contrast to the ␣-subunit, is affected by phorbol ester treatment. PMA treatment dramatically increases the amount of secreted hmeprin␤, whereas secretion of hmeprin␣ is not influenced. These results clearly support the notion that processing and secretion of the two subunits are different. A large number of secreted proteins are derived from integral plasma membrane protein precursors (52)(53)(54)(55)(56) and are released upon treatment of cells with phorbol esters such as PMA, presumably by activating protein kinase C, a cytosolic protein that becomes membrane associated after activation and rapidly phosphorylates intracellular target proteins (42). To clarify the involvement of a kinase in the regulation of hmeprin␤ shedding, staurosporine inhibition experiments were carried out. The effect of PMA-induced shedding of hmeprin␤ was antagonized by staurosporine, confirming that protein kinase C is involved in the regulation of hmeprin␤ shedding.
Shedding appears to occur at or near the cell surface. The proteases involved are often ADAMs. They have been shown to play important roles in the phorbol ester-stimulated shedding or release of membrane-anchored precursor proteins (57). The observation that TIMP1, TIMP2, nor TIMP3 inhibit constitutive hmeprin␤ shedding practically rules out an involvement of MMPs in this process. This notion is further supported by the data with TAPI that showed an inhibition of hmeprin␤ shedding. One metalloprotease that is inhibited by TAPI is ADAM 17 (TACE), identified by its ability to release the inflammatory cytokine tumor necrosis factor ␣ (47). Other structurally unrelated proteins like transforming growth factor ␣ and L-selectin are also shed by TACE (58). This prompted us to analyze the ability of hmeprin␤-transfected TACE(Ϫ/Ϫ) cells to secrete hmeprin␤. An undiminished secretion of hmeprin␤ was observed in these cells not stimulated by PMA, indicating that constitutive and stimulated hmeprin␤ shedding are mediated by different metalloproteases. The protease responsible for the induced shedding is in all probability TACE. This is supported by the fact, that TIMP3, an inhibitor of TACE, has no effect on the constitutive shedding but inhibits the PMA-stimulated shedding of hmeprin␤ in transfected COS-1 cells. Such observations have also been made by others. Garton et al. (59) demonstrated that shedding of fractalkine (CX3CL1) may be stimulated by PMA and that TACE is the shedding enzyme that mediates the PMA-induced but not the constitutive release. The constitutive sheddase of hmeprin␤ remains to be determined. It may be an ADAM or another TIMP3-sensitive metalloprotease.
As we have shown recently, shedding of hmeprin␤ is also influenced by its glycosylation status (60). In contrast to hme-prin␣ and mouse meprin␣ and meprin␤, hmeprin␤ bears Oglycan side chains. If the formation of O-glycan side chains during post-translational processing of hmeprin␤ in transfected Madin-Darby canine kidney cells was prevented by benzyl-N-acetyl-␣-D-galactosamide an increase in secretion of the enzyme was observed, suggesting that these carbohydrate chains have a "protective" role against proteolytic cleavage within the intervening region of hmeprin␤. Two O-glycosylation sites at Thr 599 and Ser 603 were identified within the 13amino acid peptide found essential for hmeprin␤ shedding. Transfection of COS-1 cells with a mutant enyzme in which the 13 amino acids had been deleted resulted in a protein that was (a) not O-glycosylated and (b) not secreted. As the secreted ectodomain of hmeprin␤ (wt) was not O-glycosylated (Fig. 3, lane 16), these two O-glycosylation sites must be in the Cterminal stub remaining bound to the plasma membrane. As the meprin␤ mutant lacking this 13-amino acid peptide sequence is not shed (60), the cleavage site leading to the release of the enzyme may be narrowed down further and must be localized between Thr 594 and Thr 599 (Fig. 7).
Human meprin␤, in contrast to hmeprin␣, has two potential phosphorylation sites (Ser 687 and Ser 688 ), within the protein kinase C phosphorylation site motives in the cytosolic tail: XR/KXXS 687 /TXR/KX and R/KXS 688 /T (18,43,61). This suggests a role of hmeprin␤ in signal transduction and/or regulation of proteolytic activity. This phosphorylation motif is present exclusively in human meprin␤, but not in mouse meprin␤, indicating a possible involvement of phosphorylation/dephosphorylation reactions in the species-specific function of this subunit. We have now demonstrated phosphorylation of hme-prin␤ in transfected COS-1 cells after stimulation with PMA. This is the first report showing phosphorylation of a member of the astacin family. Serine 687 was identified as the main 32 P acceptor site. Phosphorylation of proteins on serine and threonine residues is seen to trigger changes in allosteric conformations (62,63). A conformational change in hmeprin␤ because of phosphorylation events may mediate easier access of an appro- priate shedding enzyme to the ectodomain-cleavage site. To address this issue, the amounts of secreted hmeprin␤ in COS-1 cells, transfected with the mutants, were compared. However, neither the single nor the double mutants showed any significant difference in secretion, ruling out a direct influence of the phosphorylated cytoplasmic tail in the secretion of hmeprin␤. The role of phosphorylation of hmeprin␤ thus remains open to speculation. Indirect evidence has been presented for the regulation of catalytic activity, by phosphorylation, of ADAM24 (testase1), a plasma membrane-anchored sperm protease (64). It is conceivable that phosphorylation may, under certain physiologic conditions, affect the activity of hmeprin␤. The only direct activation step, known so far for meprins is the removal of the pro-domain by trypsin or plasmin (Fig. 7) (14, 65). Another intriguing possibility is that modification of hmeprin␤ by phosphorylation and dephosphorylation may reflect a role of hmeprin␤ as a cell surface receptor in signal transduction mechanisms, as proposed by Bauvois (66). If this is the case, identification of ligands that bind to hmeprin␤ will be an important task in the future. In recent years signaling molecules and domains have been identified that bind to phosphoserine/ threonine motifs, as is the case in the 14-3-3 proteins (63). These proteins have been shown to recognize a sequence that is similar to the sequence around the phosphorylated serine residue in hmeprin␤. Whether this observation has a biological meaning has to be determined. The 14-3-3 proteins are small acidic proteins that bind and regulate key proteins that are otherwise quite different from the meprins. These key proteins are involved in intracellular signaling, cell cycling, apoptosis, and transcriptional regulation (67). Clearly, more experimental data are needed to understand the function of the cytoplasmic domain phosphorylation in meprin␤ and other membranebound metalloproteases such as ADAM12.
The data presented here demonstrate that enhanced secretion as well as phosphorylation of hmeprin␤ occurs upon PMA stimulation. The physiologic role of hmeprin␤ shedding in health and/or disease is not known. In the small intestine, where both meprin␣ and -␤ are expressed in enterocytes, both subunits are detected on the brush-border membrane of these cells by immunohistochemistry (12). The meprin␤ subunit thus largely remains membrane-bound and the meprin␣ subunit is retained at the brush-border membrane, linked to the ␤-subunit by covalent disulfide bonds. It is likely that localization of the proteolytic potential on the luminal surface of enterocytes is an advantage for the degradation of luminal peptides and thus shedding probably plays a subordinate role in intestinal cells. In small intestinal tissue, both meprin subunits are also expressed in leukocytes of the lamina propria (12). Differential expression of soluble and membrane-bound meprin␣/␤ heterodimers in these cells may provide a unique mechanism of directing the proteolytic activity either to the cell surface or to the extracellular matrix. Depending on the localization of meprin, proteolytic activity will be directed toward different substrates. Shedding of hmeprin␤ may also be a means of downregulating the proteolytic activity in a given cell. While hmeprin␣ expression is limited to the gastrointestinal organs, hmeprin␤ exhibits a more ubiquitous expression pattern 3 and in vitro studies have shown that a range of different peptides and proteins, including cytokines, growth factors as well as extracellular matrix proteins are hydrolyzed by hmeprin (6,9,10,68). It is likely that substrates such as these are the targets for hmeprin␤ in tissues other than the gut. In this context it is of considerable interest to analyze if an imbalance of distribution between cell-associated and secreted hmeprin␤, in relation to potential substrates in a given cell or tissue, may have consequences that lead to a pathological situation.