Transport of meprin subunits through the secretory pathway: role of the transmembrane and cytoplasmic domains and oligomerization.

The meprin alpha subunit, a multidomain metalloproteinase, is synthesized as a type I membrane protein and proteolytically cleaved during biosynthesis in the endoplasmic reticulum (ER), consequently losing its membrane attachment and COOH-terminal domains. The meprin alpha subunit is secreted as a disulfide-linked dimer that forms higher oligomers. By contrast, the evolutionarily related meprin beta subunit retains the COOH-terminal domains during biosynthesis and travels to the plasma membrane as a disulfide-linked integral membrane dimer. Deletion of a unique 56-amino acid inserted domain (the I domain) of meprin alpha prevents COOH-terminal proteolytic processing and results in the retention of this subunit within the ER. To determine elements responsible for this retention versus transport to the cell surface, mutagenesis experiments were performed. Replacement of the meprin alpha transmembrane (alphaT) and cytoplasmic (alphaC) domains with their beta counterparts allowed rapid movement of the alpha subunit to the cell surface. The meprin alphaT and alphaC domains substituted into meprin beta delayed movement of this chimera through the secretory pathway. Replacement of glycines in the meprin alphaT domain GXXXG motif with leucine residues, alanine insertions in the meprin alphaT domain, and mutagenesis of basic residues within the meprin alphaC domain did not enhance the movement of the alpha subunit through the secretory pathway. By contrast, a mutant of meprin alpha (C320AalphaDeltaI) that did not form disulfide-linked dimers or higher order oligomers was transported through the secretory pathway, although more slowly than meprin beta. Taken together, the data indicate that the meprin alphaT and alphaC domains together contain a weak signal for retention within the ER/cis-Golgi compartments that is strengthened by oligomerization.

hetero-oligomeric complexes containing meprin ␣ subunits are referred to as meprin A (EC 3.4.24.18); homo-oligomeric complexes of meprin ␤ subunits are referred to as meprin B (EC 3.4.24.63) (4). Mouse, rat, and human meprin A homo-oligomers consist of disulfide-linked meprin ␣ dimers that selfassociate to form high molecular mass multimers of ϳ1-6 MDa (3,5,6). Whereas the high molecular mass forms of meprin A are secreted proteins, meprin B is a disulfide-linked, dimeric type I plasma membrane-bound protein (3,4). The heterooligomeric meprin A complex remains membrane-bound by virtue of the meprin ␤ transmembrane anchor and is a tetramer (3,4,7). Meprin B and hetero-oligomeric meprin A are found on the brush-border membranes of the kidney proximal tubules and intestines of mice, rats, and humans, whereas homo-oligomeric meprin A is secreted and found in the urine and intestinal lumina (8,9). Meprins are also expressed in certain cancer cells and in the leukocytes of the intestinal lamina propria, implying roles in the growth and metastasis of cancer cells and in inflammatory processes (9 -11).
The deduced amino acid sequences of meprin ␣ and ␤ subunits (Structure 1) contain a signal sequence (S) at the NH 2 terminus, followed by a prosequence (Pro), a catalytic domain (Protease), and the interaction domains MAM (meprin, A-5 protein, receptor protein-tyrosine phosphatase ) and TRAF (tumor necrosis factor receptor-associated factor; previously referred to as the MATH (meprin and TRAF homology) and AM (after MATH) domains), which are important for the folding, secretion, activity, and oligomerization of the meprin proteins (12,13). Both meprin ␣ and ␤ subunits contain epidermal growth factor-like (E), 1 transmembrane (T), and COOH-terminal cytoplasmic (C) domains. Meprin ␣ contains a unique 56amino acid sequence (the I domain) inserted between the TRAF and E domains. This subunit is proteolytically cleaved within or near the I domain during subunit maturation in the endoplasmic reticulum (ER) (7,14). This allows the meprin ␣ subunit to be released from its membrane anchor and to be secreted from cells. Previous studies have demonstrated that the I domain is sufficient and necessary to direct the proteolytic release and secretion of the mature meprin ␣ ectodomains (14). Insertion of the I domain into the meprin ␤ subunit (␤/␣I/␤) leads to proteolysis and secretion of this subunit. Deletion of the I domain from meprin ␣ (␣⌬I) causes the retention of the mutant protein in the secretory pathway prior to the acquisition of complex glycosylation that occurs in the medial-Golgi network (14,15).
Studies of chimeric meprin constructs of mouse ␣/rat ␤ and human ␣/␤ subunits indicated that the COOH-terminal region (encompassing the E, T, and C domains) of meprin ␣ is important for retention within the ER (14,16). However, no systematic study demonstrating which of the COOH-terminal domains or elements therein are essential for the localization and trafficking of the subunits has been conducted. Thus, the studies herein were designed to determine whether any one domain or element in the meprin ␣ or ␤ COOH terminus is responsible for retention within the ER/cis-Golgi compartments or movement of the subunits to the cell surface.
The T domain of the meprin ␣ (but not ␤) subunit contains a repeating glycine motif (GXXXG) that is present in the transmembrane region of proteins such as glycophorin A, ␤ 2 -adrenergic receptor, N-syndecan, and major histocompatibility complex class II ␣ and ␤ subunits and is essential for dimerization of these proteins and multimerization of the VacA toxin membrane channel of Helicobacter pylori (17)(18)(19)(20)(21)(22). The glycines are positioned along one face of the transmembrane ␣ helix, allowing for tight packing with a second transmembrane helix, thereby stabilizing the helix-helix interactions (see, for example, Fig. 4B) (17). Previous studies with glycophorin A and the VacA toxin have shown that mutation of the glycine residues disrupts dimerization and multimerization (21,22). In addition, alanine scanning insertion mutagenesis disrupts the alignment of the glycine residues within the transmembrane ␣ helix and thereby disrupts glycophorin A dimerization (23). It seemed possible that the transmembrane structure of meprin ␣ is important for protein-protein interactions (multimerization or hetero-oligomerization) that could lead to retention of the subunit in the ER/cis-Golgi compartments and that mutation of the meprin ␣T domain by site-directed or alanine insertion mutagenesis could disrupt this retention. Therefore, this type of study was conducted with meprin ␣.
There are two well characterized ER retention motifs found in the cytoplasmic region of ER-retained type I transmembrane proteins, a dilysine motif (KKXX COOH ) and an RXR motif (24,25). Basic residues in meprin ␣C domain (KXRX COOH ) share similarities with both motifs. For this reason, the basic residues of the meprin ␣ COOH-terminal tail were modified, and the effects on retention or movement of meprin ␣ out of the ER were determined.

EXPERIMENTAL PROCEDURES
Reagents and Materials-[ 35 S]Methionine/cysteine was purchased from PerkinElmer Life Sciences. Pansorbin was from Calbiochem. The mammalian expression plasmid pcDNA3.1 ϩ , Dulbecco's modified Eagle's medium, Opti-MEM, methionine-free Dulbecco's modified Eagle's medium, and Dulbecco's modified Eagle's medium/Ham's F-12 medium were from Invitrogen. The QuikChange site-directed mutagenesis kit was from Stratagene. Protein A-Sepharose was from Sigma. Endoglycosidase H (Endo H) and peptide N-glycosidase F (PNGase F) were from New England Biolabs Inc. Complete mini EDTA-free protease inhibitor tablets were from Roche Applied Science. n-Octyl glucoside was from Fisher.
Plasmid Construction and Mutagenesis-All cDNAs were cloned into the expression vector pcDNA3.1 ϩ . The full-length wild-type mouse meprin ␣ subunit, the ␣⌬I mutant, and the ␣⌬I/␤ETC mutant were described previously (14). Wild-type meprin ␣ and ␣⌬I were subcloned into pcDNA3.1 ϩ by restriction digestion with HindIII and XbaI. The COOH-terminally His 6 -tagged wild-type meprin ␣⌬I mutant was constructed by PCR amplification with an internal 5Ј-primer and a 3Јprimer containing a Gly-Gly-Gly-Ser spacer, a His 6 tag sequence, and an XhoI restriction site. This PCR fragment was restriction-digested with BspEI and XhoI and used to replace the corresponding region of the ␣⌬I mutant. The COOH-terminally His 6 -tagged full-length wild-FIG. 1. Endo H sensitivity of the meprin ␣⌬I mutant and meprin ␣⌬I/␤ chimeras. A, diagram of the meprin ␣⌬I/␤ chimeras. Meprin ␣ domains are indicated by white boxes, and meprin ␤ domains by black boxes. S, signal sequence; P, prosequence. B, deglycosylation analysis of the meprin ␣⌬I protein containing meprin ␤ COOH-terminal domains. Secreted wild-type meprin ␣ (WT␣) and solubilized total membrane fractions of the meprin ␣⌬I mutant and various chimeric ␣⌬I/␤ mutants were analyzed for susceptibility to endoglycosidase treatment. Media or membrane fractions were subjected to enzymatic deglycosylation for 2 h at 37°C in the absence (Ϫ) or presence of Endo H (H) or PNGase F (F) as indicated. After deglycosylation treatment, the samples were subjected to SDS-PAGE and Western blot analysis using anti-meprin ␣ antibodies. The relative positions of the molecular mass markers (in kilodaltons) are shown. type meprin ␤ subunit was constructed by Dr. Greg P. Bertenshaw (Pennsylvania State University). Briefly, the entire wild-type meprin ␤ cDNA was PCR-amplified with a 5Ј-primer containing a KpnI restriction site and a 3Ј-primer containing a Gly-Gly-Gly-Ser spacer, a His 6 tag sequence, and an XhoI restriction site. This PCR fragment was digested with KpnI and XhoI and cloned into pcDNA3.1 ϩ digested with the same enzymes. The ␤/␣TC and ␤/␣C mutants were constructed by Dr. Petra Marchand by fusion PCR using primers designed to replace the T and C domains of meprin ␤ with those of meprin ␣ and to replace the C domain of meprin ␤ with that of meprin ␣, respectively. These mutants were subcloned into pcDNA3.1 ϩ by restriction digestion with HindIII and XhoI. The C320A␣⌬I mutant was constructed by replacing the NH 2 -terminal fragment containing Cys 320 of ␣⌬I with the corresponding region of the C320A␣ mutant, described previously (26), by restriction digestion with HindIII and BspEI.
The ␣⌬I/␤TC mutant was constructed by fusion PCR in which the wild-type ␤T and ␤C domains replaced the corresponding fragments of meprin ␣. Fusion PCR was also used to construct the ␤/␣T/␤C mutant, in which the T domain of meprin ␤ was replaced with the corresponding region of meprin ␣, and the ␤/␣ETC mutant, in which the E, T, and C domains of meprin ␤ were replaced with the corresponding domains of meprin ␣. The ␣⌬I/␤C mutant was constructed by restriction digesting the ␤/␣T/␤C mutant with PmlI and ApaI and cloning the ␣T/␤C fragment into ␣⌬I digested with the same enzymes. The ␣⌬IAXAX COOH mutant was created by site-directed mutagenesis using the QuikChange site-directed mutagenesis kit and mutagenic primers designed to mutate KLRQ COOH in the meprin ␣C domain to ALAQ COOH .
Cell Culture and Transient Transfection-Human embryonic kidney 293 cells (CRL-1573, American Type Culture Collection) were cultured as described previously (12). Cells were transiently transfected with 10 g of DNA/10-cm 2 dish using the HEPES-buffered saline calcium phosphate-mediated transfection method (27). Briefly, 500 l of 2ϫ HEPESbuffered saline (50 mM HEPES, 1.5 mM Na 2 HPO 4 , and 280 mM NaCl, pH 7.05) was added dropwise to 10 g of DNA in 500 l of 125 mM CaCl 2 . After DNA-containing precipitates were allowed to form for 30 min at 25°C, the reaction mixtures were then added to the 293 cells and incubated overnight at 37°C. If meprin expression in the culture medium was to be assayed, the culture medium was changed 16 h posttransfection to serum-free Dulbecco's modified Eagle's medium/Ham's F-12 medium or serum-free Opti-MEM. Transfection efficiencies were ϳ10% for mutant and wild-type proteins. Transfected proteins had similar levels of expression as determined by Western blot analysis of media using similar volumes, antibody dilutions, and film exposure times. The results shown in Fig. 1, 2, and 4 -6 are representative of at least three separate experiments.
Preparation of Media and Membrane Fractions and Azocaseinase Activity Assay-Tissue culture media and membrane fractions were prepared 48 h post-transfection and analyzed for the expression of meprins. Media samples were collected; the serine protease inhibitor phenylmethylsulfonyl fluoride (1 mM) was added; and samples were subjected to centrifugation at 16,000 ϫ g for 10 min. Supernatant fractions were concentrated to 500 l with Centriplus-50 concentrators (Millipore Corp.). Cells were washed twice with phosphate-buffered saline (PBS), removed from the plate by scrapping using a plate scrapper in 2 ml of PBS, and centrifuged at 200 ϫ g for 5 min. Cell sediment was suspended in 500 l of PBS containing 1ϫ Complete mini EDTAfree protease inhibitor tablet mixture and sonicated for 30 s to disrupt the cells. Total membrane fractions were isolated by centrifugation at 100,000 ϫ g for 30 min. Membranes were washed twice with PBS and solubilized in 200 l of 1.0% n-octyl glucoside in 1ϫ PBS and 1ϫ Complete protease inhibitor tablet mixture. Azocaseinase activity was measured as described previously (28).
SDS-PAGE and Immunoblotting-Concentrated media and solubi- lized total membrane fractions were subjected to electrophoresis in the presence of ␤-mercaptoethanol on 7.5% Ready gels (Bio-Rad) in SDS gel electrophoresis buffer and transferred to nitrocellulose membranes. The polyclonal antibody used to detect meprin ␣ subunits was HMC14 (14). Polyclonal antibodies against meprin ␤ were produced in rabbits using purified recombinant rat meprin ␤ protein (PSU56). Horseradish peroxidase-conjugated anti-rabbit secondary antibodies (Amersham Biosciences) were detected by chemiluminescence using the SuperSignal Dura substrate (Pierce).
Endoglycosidase Treatment-Proteins were prepared for deglycosylation using Endo H and PNGase F following the manufacturer's instructions. Briefly, protein samples were denatured by boiling in 1ϫ denaturation buffer for 10 min. Deglycosylation buffers containing 1ϫ Complete protease inhibitor tablet mixture were added, and proteins were incubated for 2 h at 37°C. Deglycosylated proteins were analyzed by SDS-PAGE and immunoblotting.
Pulse-Chase Experiments-Pulse radiolabeling was performed as described previously (12). Briefly, cells were radiolabeled for 30 min with [ 35 S]methionine/cysteine (100 Ci/ml/dish). The cells were washed once with serum-free medium and incubated for 0, 2, 4, and 8 h in the presence of 2 ml of serum-free Opti-MEM. Cells were washed thoroughly with PBS and lysed. Cell lysates were incubated with Pansorbin for 1 h at 4°C and centrifuged at 6500 ϫ g for 30 min. Supernatant fractions were incubated with anti-mouse meprin ␤ antibodies (PSU56) for 10 min at 37°C and then overnight at 4°C. Immune complexes were immunoprecipitated with 40 l of a 50% suspension of protein A-Sepharose for 3 h at 4°C. Beads were washed, boiled in 1ϫ denaturation buffer for deglycosylation for 5 min, and incubated for 2 h at 37°C in the presence or absence of Endo H. Following deglycosylation, samples were subjected to SDS-PAGE. Gels were dried, and the immunoprecipitated proteins were visualized by fluorography.
Size Exclusion Chromatography of Membrane-bound Meprins-Total membrane fractions were prepared by a modification of the method of Booth and Kenny (29). Briefly, stably transfected 293 cell lines grown to confluency on 10-cm 2 dishes were scraped and washed once with 10 ml of 1ϫ PBS, pH 7.5. Cells were sedimented at 200 ϫ g for 5 min at 4°C. PBS was removed by aspiration, and the sediment was resuspended in 1 ml of 2 mM Tris and 10 mM mannitol, pH 7.0, and sonicated in an ice bath for 1 min. CaCl 2 was added to a final concentration of 10 mM to precipitate the membranes. Membranes were sedimented at 16,000 ϫ g for 30 min at 4°C; the supernatant fraction was removed; and precipitates were solubilized overnight in 25 mM HEPES, 150 mM NaCl, and 1% n-octyl glucoside, pH 7.5, on a rotating platform at 4°C. Unsolubilized membrane components were sedimented at 16,000 ϫ g for 30 min at 4°C, and the supernatant fraction was subjected to chromatography on a Superose 6 size exclusion column (Amersham Biosciences) equilibrated in 25 mM HEPES, 150 mM NaCl, and 1% n-octyl glucoside, pH 7.5. Fractions (1 ml) were collected, and the presence of meprin protein was detected by Western blot analysis using HMC14 polyclonal antibodies. Western blots were analyzed by densitometry.

RESULTS
The Transmembrane and Cytoplasmic Domains of the Meprin ␣ and ␤ Subunits Are Important for Transport of the Proteins through the Secretory Pathway-Chimeras of meprins FIG. 4. Mutation of glycine residues in the mouse meprin ␣T domain. A, sequence alignment and schematic representation of the T domains of rat meprin ␤ and glycophorin A; the T domains of rat, human, and mouse wild-type meprin ␣ (WT␣); and the equivalent T domain sequences of the glycine mutants of mouse ␣⌬I. The T domain sequences of mouse, rat, and human meprin ␣ were aligned using ClustalW and were compared with the T domain sequences of rat meprin ␤ and glycophorin A. The conserved glycine repeat residues in the meprin ␣T domain sequences are indicated by black boxes. The Gly-to-Leu mutations are underlined. B, helical wheel projection of the mouse meprin ␣T domain. The diagram illustrates that the glycine repeat motif residues (Gly 730 , Gly 734 , Gly 738 , and Gly 742 ) align along one face of the transmembrane helix. C, sensitivity of secreted wild-type meprin ␣ and cell-associated meprin ⌬I mutants to Endo H and PNGase F. Wild-type meprin ␣, the ␣⌬I mutant, and the Gly-to-Leu ⌬I mutants were transfected into 293 cells and allowed to grow for 48 h. For wild-type meprin ␣, the medium was collected; for the meprin ⌬I mutants, membrane fractions were prepared. Media and membranes were subjected to enzymatic deglycosylation for 2 h at 37°C in the absence (Ϫ) or presence of Endo H (H) or PNGase F (F) as indicated. After deglycosylation treatment, the mutants were subjected to SDS-PAGE and Western blot analysis using anti-meprin ␣ antibodies. The relative positions of the molecular mass markers (in kilodaltons) are shown.
␣⌬I and ␤ were constructed to map the domains responsible for intracellular retention versus cell-surface expression (Fig. 1A). Endoglycosidase treatments allowed the determination of the steady-state subcellular localization of the meprin proteins. Endo H removes the high-mannose oligosaccharides that are found in the ER, whereas PNGase F removes both highmannose and complex oligosaccharides that arise in the Golgi apparatus (15). Mutants susceptible to Endo H treatment (such as meprin ␣⌬I) are not complex glycosylated, indicating retention within a compartment prior to the medial-Golgi compartment (i.e. ER or cis-Golgi). Secreted wild-type meprin ␣ was resistant to Endo H, but sensitive to PNGase F (Fig.  1B), indicating that complex glycosylation occurred. The replacement of all three COOH-terminal ␣⌬I domains with those of meprin ␤ (␣⌬I/␤ETC) or the last two domains (␣⌬I/ ␤TC) resulted in resistance of the protein to Endo H, indicating that the proteins effectively moved through the medial-Golgi compartment. The meprin ␤C domain alone (␣⌬I/␤C) was partially effective in allowing movement of the protein out of the ER/cis-Golgi compartments, but a significant amount of Endo H sensitivity was observed, indicating highmannose oligosaccharides were present in this chimeric protein. Thus, both the meprin ␤T and ␤C domains play roles in movement of the ␤ subunit out of the ER/cis-Golgi compartments.
Chimeras of meprin ␤ containing meprin ␣ COOH-terminal domains were then constructed and transfected into 293 cells (Fig. 2A). The Endo H/PNGase F profiles of these mutants indicated that the meprin ␤ constructs containing meprin ␣ COOH-terminal domains were primarily resistant to Endo H, indicating movement out of the ER, but showed varying degrees of high-mannose glycosylation (Fig. 2B). Untreated membrane fractions of all the meprin ␤ mutants existed as two species (unlike wild-type meprin ␤) that most likely represent complex glycosylated (higher molecular mass) and high-mannose glycosylated (lower molecular mass) species. After deglycosylation with PNGase F, one band was usually observed. Occasionally, there appeared to be two bands that migrated very closely to each other. The two bands could result from incomplete deglycosylation of some subunits, proteolytic clipping, or post-translational modifications of some subunits such as amidation and phosphorylation. Overall, the results indicate that the meprin ␣ COOH-terminal domains impede the movement of the meprin ␤ chimeras through the secretory pathway, but do not effectively retain these chimeras within the ER/cis-Golgi compartments.
Pulse-chase experiments were performed to determine the rate of complex glycosylation of the meprin ␤ subunit and chimeric meprin ␤/␣ mutants. Wild-type meprin ␤ acquired complex glycosylation, as evidenced by resistance to Endo H, within 2 h of pulse radiolabeling (Fig. 3). This finding is consistent with determinations of the steady-state distribution of wild-type meprin ␤ indicating rapid transport out of the ER/ cis-Golgi compartments (Fig. 2B). Pulse-chase studies of chimeras of meprin ␤ with the meprin ␣TC or ␣C domain detected significant amounts of high-mannose oligosaccharides even at 8 h, indicative of impeded transport, and correlated with their steady-state Endo H profiles (Fig. 2B).
Mutation of the Glycine Residues in the Meprin ␣T Domain-To determine whether the glycine repeat motif within the meprin ␣T domain is responsible for ER/cis-Golgi retention of the ␣⌬I mutant, one or more glycine residues were systematically mutated to leucine (Fig. 4A). Fig. 4A shows that the meprin ␣ glycine repeat motif is not present in meprin ␤ and the similarity to the glycophorin A glycine repeat. Wild-type and mutant mouse meprin ␣ cDNAs were transiently transfected into 293 cells, and the meprin proteins associated with the cells or media were detected by immunoblotting (Fig. 4C). The SDS-PAGE mobility of the subunits after treatment with Endo H and PNGase F was again used to determine whether the subunits contained high-mannose oligosaccharides or were complex glycosylated as an indication of retention prior to or movement beyond the cis-Golgi apparatus. The deglycosylation profiles of the single and multiple Gly-to-Leu mutants of the meprin ␣⌬I transcripts were very similar to those of the meprin ␣⌬I mutant, indicating that the repeating glycine motif is not responsible for retention of the subunit in the ER/cis-Golgi compartments.
Alanine Scanning Insertion Mutagenesis of the Transmembrane Domain Has No Effect on Retention of the ␣ Subunit in the ER/cis-Golgi Compartments-Insertion of an alanine residue within the transmembrane domain will cause a 90°shift of the residues NH 2 -terminal to the insertion and consequently disrupt helical alignments (Fig. 5A) (21). Alanine residues were inserted at various positions in the meprin ␣T domain within the glycine repeat region and COOH-terminal to it (Fig. 5A). The meprin ␣⌬I alanine insertion mutants all showed similar deglycosylation patterns compared with the ␣⌬I mutant, indicating that they did not pass through the medial-Golgi compartment. These results are consistent with those obtained with the glycine mutants, indicating that the structure of the T domain alone is not a major determinant of ER/cis-Golgi retention.
Motifs in the Meprin ␣C Domain Resembling Known Cytoplasmic ER Retention Motifs Are Not Involved in the Inability of the Meprin ␣ Subunit to Move through the medial-Golgi Compartment-The meprin ␣C domain contains only six amino acids (Fig. 6A), in contrast to meprin ␤, which has 26 amino acids. There are two basic residues in the meprin ␣C domain (XXKXRQ COOH ), and although they do not exactly conform to the known ER retention motifs (e.g. KDEL or a dilysine motif at positions Ϫ3 and Ϫ4 or positions Ϫ3 and Ϫ5 prior to the COOH terminus) (30), the basic residues were replaced with Ala to test whether the basic residues in the meprin ␣C domain affect ER retention. A mutant that extended the COOH terminus to 16 amino acids (␣⌬I-His 6 ) was also constructed (Fig. 6A). Both mutants had a similar Endo H profile compared with the ␣⌬I mutant (Fig. 6B), indicating that the basic residues present in the meprin ␣C domain and the length of the tail do not affect subunit retention.
The Oligomeric State Affects the Intracellular Retention of the Meprin ␣⌬I Mutant-To determine the effects of the oligomeric state of meprin ␣ on ER/cis-Golgi retention of the ␣⌬I mutant, a mutant lacking the ability to form disulfide-linked dimers was constructed. Previous studies have demonstrated that Cys 320 is responsible for the formation of intersubunit disulfide-bonded dimers in wild-type meprin ␣; mutation of this residue (C320A␣) abrogates the ability to form higher order oligomeric species (24). Therefore, Cys 320 was mutated to Ala in FIG. 6. Endo H sensitivity of the meprin ␣C domain mutants. A, diagrammatic representation of the meprin ␣C domain and its mutants. Addition of a Gly-Gly-Gly-Ser spacer and a His 6 tag to the ␣⌬I mutant (␣⌬I-His 6 ) is indicated. Mutation of the basic residues in the meprin ␣C domain (␣⌬IAXAX COOH ) is also indicated. TM, transmembrane domain. B, deglycosylation of the meprin ␣⌬I C domain mutants. The meprin ␣⌬I C domain mutants were analyzed for susceptibility to endoglycosidase treatment. Solubilized membranes were subjected to enzymatic deglycosylation for 2 h at 37°C in the absence (Ϫ) or presence of Endo H (H) or PNGase F (F) as indicated. After deglycosylation treatment, the mutants were subjected to SDS-PAGE and Western blot analysis using anti-meprin ␣ antibodies. The relative positions of the molecular mass markers (in kilodaltons) are shown.
the ␣⌬I mutant (C320A␣⌬I), and the oligomeric size and glycosylation state of this mutant were determined.
The ␣⌬I and C320A␣⌬I mutants were solubilized from transiently transfected 293 cell membrane fractions and analyzed by size exclusion chromatography on a Superose 6 column. The elution profiles were determined by densitometry of Westernblotted Superose 6 fractions and compared with the elution profiles of standard proteins (Fig. 7A). The ␣⌬I mutant eluted with a peak in fraction 12 corresponding to a molecular mass of ϳ620 kDa, whereas the C320A␣⌬I mutant eluted with a peak in fraction 15 corresponding to a molecular mass of ϳ270 kDa (Fig. 7A). It is likely that the 270-kDa protein is a dimeric form of the C320A␣⌬I protein because no evidence could be found for another protein interacting with the mutant after immunoprecipitation, two-dimensional gel electrophoresis, NH 2 -terminal sequencing, and mass spectrophotometry analyses (data not shown).
To determine the effect of the different oligomeric states of the ␣⌬I and C320A␣⌬I mutants on movement through the secretory pathway, the deglycosylation patterns of these mu-tants were compared. Approximately half of the C320A␣⌬I mutant contained Endo H-resistant oligosaccharides, indicating movement out of the ER/cis-Golgi compartments, whereas the ␣⌬I mutant was primarily Endo H-sensitive, indicating residence within a pre-medial-Golgi compartment (Fig. 7B). Wild-type meprin ␣ protein secreted into the culture medium was completely Endo H-resistant, as expected of a protein that has traversed the secretory pathway. From this result, we concluded that the oligomeric state of the meprin ␣ subunit is a determinant in the ability of this protein to move into the medial-Golgi compartment. In addition, the ␣⌬I mutant, like the mature wild-type protein, was resistant to extensive degradation by trypsin, indicating that it is properly folded (data not shown). The activated ␣⌬I mutant was able to hydrolyze the protein substrate azocasein and had a similar specific activity for this substrate (1083 Ϯ 164 units/mg) compared with wild-type meprin ␣ (977 Ϯ 268 units/mg). Thus, although the ␣⌬I mutant is retained in the ER/cis-Golgi compartments, it has similar enzymatic and oligomerization properties compared with the secreted wild-type protein. DISCUSSION This work establishes that the meprin ␣T domain in combination with the meprin ␣C domain is responsible for the retention of the subunit in the ER/cis-Golgi compartments and that the E domain does not contribute to the retention process. In addition, the meprin ␤T and ␤C domains together promote the movement of meprin proteins through the secretory pathway (Fig. 8). No one motif or domain in the meprin ␣ COOH terminus was found to be responsible for the retention; however, the formation of disulfide-linked dimers and higher oligomerization of meprin ␣ subunits enhanced the retention of the complex. One interpretation of these results is that the meprin ␣T and ␣C domains together contain a weak retention signal and that oligomerization serves to cluster multiple signals and to increase the strength for retention. The retention signal is not likely to involve the GXXXG motif of the T domain or the basic residues of the six-member cytoplasmic tail, however, because mutations of these residues did not enhance movement out of the ER/cis-Golgi compartments. But higher oligomerization of subunits containing these motifs may increase interactions with resident ER proteins or disrupt interactions with proteins that are instrumental in moving the proteins out of the ER. This would explain why the dimeric C320A␣⌬I mutant and the ␤/␣TC chimera are only partially retained within the early secretory pathway compartments.
This work also demonstrates that higher order oligomeric complexes of meprin ␣ form in the ER. There are several other known protein complexes that assemble in the ER; for example, measles virus envelope glycoproteins, the T cell receptor, and IgE all form oligomers in the ER (20,31,32). In these examples, however, proper oligomerization in the ER leads to exit from the ER rather than retention; multimerization of these proteins apparently masks ER retention motifs (32). By contrast, multimerization of meprin ␣ leads to unmasking or strengthening retention signals and retaining the protein unless the luminal protein domains are proteolytically cleaved from the membrane. Thus, the purpose of meprin ␣ retention appears to be to allow the complex to form properly for the COOH-terminal hydrolysis rather than for membrane-associated transport through the secretory pathway. Previous studies have established that multimerization of meprin ␣ is essential for stability and enzymatic activity against substrates; thus, this process has functional significance (26). It is also known from previous studies that meprins must contain the MAM domain for proper folding (12). Mutants lacking the MAM domain, either by truncation after the protease domain or by deletion of FIG. 7. Effects of oligomeric state on the Endo H sensitivity of the meprin ␣⌬I and C320A␣⌬I mutants. A, solubilized membrane fractions of ␣⌬I and C320A␣⌬I (C320A⌬I) were subjected to size exclusion chromatography. The fractions were collected and analyzed for the presence of meprin proteins by Western blot analysis. The elution profile was determined by densitometric analysis of the lanes testing positive for meprin expression and by plotting the percentage of total meprin protein present per lane. B, media samples from wild-type meprin ␣ (WT␣)-transfected cells and solubilized total membrane fractions of the meprin ␣⌬I and C320A␣⌬I mutants were collected and subjected to enzymatic deglycosylation for 2 h at 37°C in the absence (Ϫ) or presence of Endo H (H) or PNGase F (F) as indicated. After deglycosylation treatment, the mutants were subjected to SDS-PAGE and Western blot analysis using anti-meprin ␣ antibodies. The relative positions of the molecular mass markers (in kilodaltons) are shown.
FIG. 8. Diagrammatic representation of the intracellular trafficking of wild-type meprin ␣ and ␤ subunits and mutants. From top to bottom: wild-type meprin ␣ is secreted into the culture medium; wild-type meprin ␤ is cell surface-expressed; ␣⌬I is retained in the ER/cis-Golgi compartments; ␣⌬I/␤TC is rapidly transported in the secretory pathway; and C320A␣⌬I and ␤/␣TC are slowly transported out of the ER/cis-Golgi compartments. this domain from the rest of the protein, are degraded by the ubiquitin/proteasome pathway (12,33). Thus, the quality control system of the ER monitors for proper folding of the meprin ␣ protein and for proteolytic release of the protein from the membrane.
The meprin ␣TC and ␤TC domains have little in common, in contrast to the other domains of the subunits, for which amino acid identity ranges from 35 to 55% (34). This indicates that these COOH-terminal domains have evolved for different functions. The meprin ␣T and ␣C domains appear to have evolved for retention of the subunit in the ER/cis-Golgi compartments for proper proteolytic processing and for formation of a large secreted protease. By contrast, the meprin ␤TC domains are endowed with features that maintain transmembrane association and that enable interactions with cytosolic proteins for transport and possibly for signal transduction. For example, the 26-amino acid meprin ␤C domain (YCTRRKYRKKARAN-TAAMTLENQHAF) has two potential phosphorylation sites. One is a potential protein kinase C phosphorylation site in the consensus sequence TRR; the other is a potential site for calmodulin kinase II phosphorylation in the consensus sequence RXXT (35). In addition, it has been shown that meprin ␤ transiently interacts with a cytoplasmic protein (OS9) implicated in ER-to-Golgi transport (36,37). Truncation of 10 -15 residues from the meprin ␤C domain and a Tyr-to-Pro substitution within this domain result in retention of the mutant meprin ␤ proteins in the ER, indicating that the length and secondary structure of the meprin ␤C domain influence movement of meprin ␤ out of the ER (38). Our results are consistent with those reported previously, indicating a role for the meprin ␤C domain in transport out of the ER. The studies with meprin ␣ herein indicate that increasing the length of its C domain by 10 amino acids did not enhance movement of this subunit out of the ER; however, both the number of amino acids and composition of the peptide are likely to be important for transport.
Meprin ␤ serves to localize meprin ␣ to the plasma membrane when the subunits are expressed cotranscriptionally (39). This has been demonstrated in recombinant systems as well as in vivo (40,41). Meprin ␤ knockout mice express normal levels of meprin ␣ mRNA and protein, but show no meprin ␣ associated with apical membranes of kidney proximal tubules or intestinal epithelial cells in contrast to controls. Analyses of recombinant and wild-type mouse kidney hetero-oligomers indicate that the ␣ and ␤ subunits may be disulfide-linked, or that homodimers of meprin ␣ may interact noncovalently with meprin ␣/␤ heterodimers that are membrane-bound (3,39). Therefore, one of the functions of the meprin ␤TC domains is to localize meprin ␣ to the plasma membrane. This is important because meprin ␣ has been shown to be destructive to extracellular matrix proteins and cytotoxic if it is mislocated to basolateral membranes (42). This is one of many examples indicating that the localization and concentration of proteases are key elements in the regulation of proteolytic activity. Thus, the potentially destructive activity of meprins is regulated by zymogen formation, dimerization, and multimerization and delivery of the proteases to specific locales.