Dopamine beta-monooxygenase signal/anchor sequence alters trafficking of peptidylglycine alpha-hydroxylating monooxygenase.

Dopamine beta-monooxygenase (DBM) and peptidylglycine alpha-hydroxylating monooxygenase (PHM) are essential for the biosynthesis of catecholamines and amidated peptides, respectively. The enzymes share a conserved catalytic core. We studied the role of the DBM signal sequence by appending it to soluble PHM (PHMs) and expressing the DBMsignal/PHMs chimera in AtT-20 and Chinese hamster ovary cells. PHMs produced as part of DBMsignal/PHMs was active. In vitro translated and cellular DBMsignal/PHMs had similar masses, indicating that the DBM signal was not removed. DBMsignal/PHMs was membrane-associated and had the properties of an intrinsic membrane protein. After in vitro translation in the presence of microsomal membranes, trypsin treatment removed 2 kDa from DBMsignal/PHMs while PHMs was entirely protected. In addition, a Cys residue in DBMsignal/PHMs was accessible to Cys-directed biotinylation. Thus the chimera adopts the topology of a type II membrane protein. Pulse-chase experiments indicate that DBMsignal/PHMs turns over rapidly after exiting the trans-Golgi network. Although PHMs is efficiently localized to secretory granules, DBMsignal/PHMs is largely localized to the endoplasmic reticulum in AtT-20 cells. On the basis of stimulated secretion, the small amount of PHMs generated is stored in secretory granules. In contrast, the expression of DBMsignal/PHMs in PC12 cells yields protein that is localized to secretory granules.

Catecholamine synthesis from tyrosine involves several cytosolic enzymes and one secretory granule enzyme, dopamine ␤-monooxygenase (DBM) 1 (1). DBM from adrenal chromaffin cells and neurons occurs in soluble and membrane forms (1)(2)(3). Although soluble DBM is secreted along with its product catecholamines, membrane DBM undergoes endocytosis (4,5). DBM monomers form tetramers composed of two noncovalently bound disulfide-linked dimers (6,7). Soluble and membrane forms of DBM both are heavily glycosylated and derived from a single translation product (8). Phospholipids as well as an uncleaved signal sequence seem to play a role in the attachment of DBM to membranes (8 -10).
Sequence analysis of membrane DBM from bovine adrenal medulla revealed an uncleaved signal sequence in 30% of the protein (11). Similarly, sequence analysis of rat DBM produced by in vitro transcription/translation in the presence of microsomal membranes indicated that the signal sequence was not removed (12). These studies suggest that DBM is a type II integral membrane protein. The NH 2 -terminal sequences of rat, mouse, human, and bovine DBM vary in length, but each contains a stretch of 20 hydrophobic amino acids, which is longer than the hydrophobic domains typically found in cleaved signal sequences (13,14) (Fig. 1A). Although rat and mouse DBM also contain a lengthy stretch of amino acids preceding the hydrophobic domain, human and bovine DBM do not. However, analysis of the genomic sequence encoding human DBM (15,16) identified an in-frame upstream Met codon that is likely to represent the actual translational start site (Fig. 1A) (17,18). With this modification, the similarity of human to mouse and rat DBM is increased greatly. Additional sequence data for the 5Ј end of bovine DBM are not available (19 -21).
Because DBM is a large tetrameric glycoprotein, it has proven difficult to evaluate the role of the DBM signal/anchor sequence in the context of the native protein. To analyze the routing information in the DBM signal/anchor, we appended it to the monooxygenase domain of a homologous enzyme. Peptidylglycine ␣-hydroxylating monooxygenase (PHM) catalyzes the first step of the two-step conversion of peptidylglycine substrates into ␣-amidated products. The reactions catalyzed by DBM and PHM both require copper, ascorbate, and molecular oxygen (22)(23)(24)(25), and the catalytic core of PHM shares 32% identity with a 296-amino acid region of DBM (26 -28). We selected PHM as our reporter protein because it is homologous to DBM in sequence, has the same cofactor requirements, can be assayed easily, is substantially smaller than DBM, and is not glycosylated.
In the DBMsignal/PHMs chimera, the signal sequence of PAM was replaced with residues 1-42 of rat DBM, and a translational stop was inserted after the PHMs domain (Fig.  1B). The metabolism of PHMs has been characterized in detail. In Chinese hamster ovary (CHO) cells, PHMs yields a protein of 39 kDa that retains its pro-region, whereas in AtT-20 cells the pro-region is removed yielding a 38-kDa protein that is stored efficiently in secretory granules and released in response to secretagogue (Fig. 1B) (29 -31). Outside of the catalytic core, the sequences of PAM, the bifunctional protein that contains both PHM and peptidyl ␣-hydroxy-glycine ␣-amidating lyase, and DBM diverge, and their topologies differ. PAM is a type I integral membrane protein and does not form disulfidelinked dimers. Tissue-specific alternative splicing produces in-tegral membrane and soluble forms of PAM (32), and endoproteolysis generates soluble PHM and peptidyl ␣-hydroxy-glycine ␣-amidating lyase from membrane PAM (31).
DBMsignal/PHMs was expressed in AtT-20 corticotrope tumor cells and PC12 pheochromocytoma cells, which contain secretory granules, and in CHO cells, which do not; although PC12 cells produce DBM, AtT-20 cells do not. We monitored PHM activity to evaluate folding of the chimeric protein. The presence or absence of the DBM signal sequence was assessed through changes in mass. In vitro transcription/translation was used to demonstrate exposure of the NH 2 terminus of DBM to cytosol. Metabolic labeling of stably transfected AtT-20 and CHO cell lines demonstrated rapid turnover and limited secretion of DBMsignal/PHMs. DBMsignal/PHMs adopts the topology of a type II membrane protein and is localized to the endoplasmic reticulum in AtT-20 and CHO cells. Although folded into a catalytically active conformation, the PHMs sequence does not localize the chimera to AtT-20 secretory granules. In contrast, DBMsignal/PHMs expressed transiently in PC12 cells was localized to secretory granules.

Construction of Expression
Vector-To generate DBMsignal/PHMs, the gene splicing-by-overlap-extension (SOE) technique was employed (33). cDNA fragments were amplified by polymerase chain reaction using as templates the pBluescript plasmids carrying the cDNAs for PHMs (rPAM-1 (nucleotides 1-1444)) and DBM (rDBM (nucleotides 1-2445)), kindly provided by Dr. E. Sabban (Medical College of New York, Valhalla, NY) (12). The cDNA fragment encoding the DBM signal sequence (42 amino acids) was spliced onto the cDNA encoding rat PHM (26 -382) to generate the DBMsignal/PHMs chimera. The sequence at the splice junction was verified to encode AVAIFLVILVAALQGFR-SPLSVFKRFLETTR, where the DBM signal sequence is shown in bold. The DBMsignal/PHMs cDNA was inserted into a pCIS.2CXXNH expression vector using complementary restriction sites. Construction of the pBluescript plasmid encoding PHMs (pBS.PHMs) was described previously (31).
Tissue Culture and Transfection-AtT-20 and CHO cells were cultured in Dulbecco's modified Eagle's/F-12 medium containing 10% fetal clone serum (HyClone, Logan, UT) and 10% Nu-Serum (Collaborative Research, Bedford, MA). Stable AtT-20 cell lines were established by co-transfecting the pCIS expression vector and pMT.Neo (Stratagene, La Jolla, CA) using LipofectAMINE (Life Technologies, Inc.), followed by drug selection with G418 (0.5 mg/ml) (31). Stable CHO cell lines were generated using Lipofectin and selected in ␣-minimum Eagle's medium containing 20% dialyzed fetal calf serum. Cell lines expressing DBMsignal/PHMs were screened by enzyme assay, immunofluorescence, and Western blot analysis. In addition, DBMsignal/PHMs and rat DBM (34) were transiently expressed in rat PC12 pheochromocytoma cells using GenPORTER 2 (GTS, San Diego, CA). PC12 cells grown in Dulbecco's modified Eagle's/F-12 medium containing 10% fetal clone serum and 10% Nu-Serum were plated on 25-mm culture dishes and transfected with 4 g of DNA/well. After a 6-h incubation with the DNA in CSFM, the cells were plated on chamber slides precoated with 0.1 mg/ml poly-L-lysine and examined 48 h after initial exposure to DNA using immunofluorescence microscopy as described below.
Biosynthetic Labeling and Immunoprecipitations-To study the biosynthesis of the DBMsignal/PHMs chimera, pulse-chase experiments were performed as described previously (34,35). Briefly, cells plated on 12-mm culture dishes coated with 0.1 mg/ml poly-L-lysine were labeled for 20 min with 300 Ci of [ 35 S]methionine (1 mCi/ml, 1000 Ci/mmol; Amersham Pharmacia Biotech) in 300 l of methionine-free CSFM air. Cells were either extracted (pulse) in 20 mM sodium TES, 10 mM mannitol, pH 7.0, and 1% TX-100 containing protease inhibitors (31) or further incubated for varying periods of time (chase) in CSFM air. The CSFM air contained 20 mM sodium HEPES, pH 7.4, instead of NaHCO 3 . For temperature block experiments, the cells were labeled at 37°C for 20 min and then chased at 20 or 37°C for 2 h. Immunoprecipitation of cell extracts and media was carried out using Ab1761 directed against the PHM domain (rPAM-1-(37-382)). The immunocomplexes were isolated by incubation with protein A-Sepharose (Sigma) (35). Immunoprecipitated proteins were fractionated by SDS-PAGE and visualized by fluorography.
In Vitro Transcription and Translation-pBluescript plasmid containing cDNA encoding DBMsignal/PHMs was used as template for in vitro coupled transcription/translation (TNT) and translocation using rabbit reticulocyte lysate and canine pancreas microsomal membranes as specified by the supplier (Promega, Madison, WI). Unless otherwise indicated, 1.3 equivalents of canine pancreas microsomal membranes were added to 50 l of translation reaction. The pBluescript plasmid containing cDNA encoding PHMs was used for comparison. In vitro synthesized proteins were analyzed by SDS-PAGE and fluorography (36).
Analysis of the Topology of DBMsignal/PHMs-DBMsignal/PHMs synthesized in vitro in the presence or absence of microsomal membranes was studied using a trypsin protection assay and biotinylation (37,38). PHMs was used as a control.
Trypsin Protection Assay-Briefly, the translation reaction was stopped by adding 1 l of cycloheximide (100 g/ml) to 15 l of the in vitro translation reaction. The membranes were stabilized by incubation with 1 l of 30 mM tetracaine-HCl (39) for 5 min at room temperature followed by cooling on ice. Samples were digested with 5 g of FIG. 1. DBM signal sequences from several species and construction of DBMsignal/PHMs chimera. A, comparison of the signal sequences of rat (27), mouse (54), human (16), and bovine DBM (11,21). The deduced amino acid sequences of the NH 2 -terminal region are shown. Residues in italics represent additional amino acids encoded by sequence 5Ј to the proposed translational start site in human DBM (17,18). The shaded area shows the hydrophobic region. The arrow indicates the site at which the signal sequence is cleaved to generate soluble DBM. B, a schematic diagram showing construction of the DBMsignal/PHMs chimera. Scale representations of full-length rat DBM, PHMs, and the DBMsignal/ PHMs chimera are shown. Experimentally determined processing patterns for PHMs in CHO and AtT-20 cells are indicated (30,31). The specificity of antibody JH1761 is indicated, as are the predicted molecular masses (in kDa). SS, signal sequence; lollipop, N-glycosylation sites. trypsin (Fluka) for 30 min at 37°C, and digestion was stopped by boiling in 10 mM sodium phosphate, pH 6.0, 0.1% SDS, and 10 mM 2-mercaptoethanol. Where indicated, samples were digested in the presence of 0.1% Triton X-100 (Pierce). Samples were subjected to SDS-PAGE and fluorography (36).
Biotinylation-The in vitro translation reaction (15 l) containing DBMsignal/PHMs protein was diluted to 50 l with PBS (pH 7.4) and incubated with 10 mM N-biotinylaminoethyl methanethiosulfonate (Toronto Research Chemicals, Ontario, Canada) for 10 min at room temperature to label cysteine residues (40). The excess reagent was quenched by incubation with 50 mM NH 4 Cl for 10 min at room temperature. The reaction mixture was diluted with 500 l of 50 mM sodium phosphate, pH 7.4/1% TX-100 (Super E), and the biotinylated proteins were isolated by incubation with immobilized NeutrAvidin (Pierce) for 30 min at room temperature. The NeutrAvidin beads were washed with Super E and 50 mM sodium phosphate buffer, pH 7.4, and protein eluted by boiling in SDS-PAGE sample buffer was analyzed by SDS-PAGE and fluorography.
Immunofluorescence Microscopy-Cells grown on chamber slides precoated with 0.1 mg/ml poly L-lysine were rinsed with PBS and fixed in ice-cold methanol for 15 min. After fixation, the slides were blocked and stained with primary antibody against the PHM domain (Ab1761) for 4 h at room temperature or overnight at 4°C. PHM antibody 1761 was diluted 1:1000 in PBS containing 2 mg/ml bovine serum albumin. A fluorescein-conjugated goat anti-rabbit immunoglobulin G was used as a secondary antibody at a dilution of 1:1000. The distribution of fluorescence was analyzed with a Zeiss (Thornwood, MT) Axioskop epifluorescence microscope using a fluorescein isothiocyanate filter set.

Fully Active PHM Is Formed upon Expression of the DBMsignal/PHMs Chimera-DBMsignal/PHMs was expressed in
AtT-20 cells, which contain secretory granules, and in CHO cells, which do not. DBMsignal/PHMs in the stably transfected cells was analyzed by enzyme assay and Western blot. The soluble PHMs protein expressed in AtT-20 cells was analyzed for comparison. From both cell types, aliquots of cell extract containing an equal amount of PHM activity were fractionated by SDS-PAGE and subjected to Western blot analysis. As shown in Fig. 2A, similar amounts of DBMsignal/PHMs protein and PHMs protein yielded 150 pmol/h of PHM activity. This result indicates that the chimera folds properly when expressed in either cell type and is as active as the control PHMs protein.
Interestingly, the DBMsignal/PHMs protein (43 kDa) has a substantially greater apparent molecular mass than PHMs (38 kDa). DBMsignal/PHMs produced by in vitro transcription/ translation in the absence of microsomal membranes (Fig. 2B), conditions that do not allow cleavage of signal sequences, also has a mass of 43 kDa. The predicted molecular mass for the chimera in the absence of signal peptide cleavage is 43 kDa. PHMs produced by in vitro translation has a mass of 41 kDa as predicted in the absence of signal peptide cleavage and greater than the mass of the cellular product. These data suggest that most of the DBMsignal/PHMs chimera in cell extracts retains the DBM signal sequence.
DBMsignal/PHMs Is an Intrinsic Membrane Protein-To further investigate the properties of the DBMsignal/PHMs chimera, subcellular fractionation was carried out. Stably transfected cells expressing DBMsignal/PHMs or PHMs were separated into soluble and membrane fractions and analyzed by Western blot using the PHM antibody. As shown in Fig. 3, DBMsignal/PHMs was found only in the membrane fraction in both cell types; no DBMsignal/PHMs was recovered in the soluble fraction. In contrast, PHMs was recovered in both the soluble and membrane fractions. The association of monofunctional PHM with membranes is characteristic of the protein and was observed in secretory granules from rat hypothalamus and hippocampus (43).
The association of the DBMsignal/PHMs protein with membranes was investigated by incubating membranes with sodium carbonate, pH 11.5, for 30 min to dissociate peripheral proteins (Fig. 3). After carbonate treatment, all of the DBMsignal/PHMs from both cell types remained in the membrane fraction, indicating that the chimera is an intrinsic membrane protein. As expected for a protein lacking a transmembrane domain, all the PHMs recovered in the membrane fraction was removed by carbonate treatment. Taken together, these data demonstrate retention of the DBM signal sequence, yielding an integral membrane chimeric protein.
DBMsignal/PHMs Turns Over Rapidly-Metabolic labeling was carried out to study the biosynthesis, storage, and secretion of DBMsignal/PHMs in AtT-20 cells and CHO cells. AtT-20 PHMs cells were analyzed for comparison. DBMsignal/PHMs and PHMs were immunoprecipitated from cell extracts and culture medium, fractionated by SDS-PAGE, and detected by fluorography. After a pulse incubation, a single 43-kDa form of DBMsignal/PHMs was detected in both cell types (Fig. 4A); the newly synthesized protein was identical in size to the chimeric protein observed on Western blots and to the in vitro translation product. After 30 min of chase, essentially all the newly synthesized DBMsignal/PHMs protein was recovered and its mass was unaltered, indicating that the DBM signal sequence had not been removed. A dramatic loss of newly synthesized protein occurred at longer chase times. As shown in Fig. 4B, 26% of the newly synthesized DBMsignal/PHMs is retained in AtT-20 cells after the 2-h chase, and almost all the chimera is lost during the 6-h chase. Interestingly, the loss of DBMsignal/ PHMs is less dramatic in CHO cells, with 15% of the protein remaining after the 6-h chase. The radiolabeled product observed in CHO cells after the 6-h chase still had a mass of 43 kDa. In contrast and as expected (31,44), PHMs, which had a mass of 40 kDa after the pulse incubation, matured into a 38-kDa protein upon removal of the pro-region; this cleavage was completed between 30 min and 2 h of chase.
Spent media were examined to evaluate secretion of newly synthesized protein. Secretion of a 38-kDa protein from AtT-20 cells expressing PHMs was readily detectable at 2 h of chase, and all the newly synthesized PHMs was recovered in cells or medium after a 6-h chase. In contrast, the loss of newly synthesized protein from cells expressing the DBMsignal/PHMs chimera was not accounted for by secretion. Secretion of a small amount of 38-kDa PHMs by CHO cells was detectable after 2 h. Secretion from AtT-20 cells expressing DBMsignal/ PHMs was barely detectable even after 6 h (Fig. 4, inset). A 43-kDa protein was also detected in the medium of AtT-20 DBMsignal/PHMs cells (Fig. 4, inset).
Newly synthesized misfolded proteins are known to undergo rapid degradation (45)(46)(47)(48). However, because fully active PHM was observed in cells expressing DBMsignal/PHMs, this type of rapid degradation was not anticipated. The DBMsignal/PHMs chimera presents an acceptable signal peptide cleavage site and contains little sequence outside of the PHM catalytic core that could be misfolded and thus target the chimera for degradation. To determine the site of degradation, temperatureblock experiments were performed. AtT-20 DBMsignal/PHMs cells were incubated with [ 35 S]methionine at 37°C for 20 min and chased for 2 h at 20 or 37°C before extraction. Incubation at 20°C blocks the exit of proteins from the TGN (49, 50). As  shown in Fig. 5A, substantially more DBMsignal/PHMs remained in cells chased at 20°C than in the cells chased at 37°C. In a similar experiment with CHO cells, the degradation of DBMsignal/PHMs was substantially diminished when the chase temperature was 20 instead of 37°C (Fig. 5B).
To further analyze the site at which DBMsignal/PHMs was degraded in AtT-20 cells, we added a variety of protease inhibitors (10 g/ml leupeptin, 100 g/ml N-acetyl-leucinyl-leucinylnorleucinal, 5 mM methylamine/NH 4 Cl, 100 M chloroquine, 5 M MG132, and 20 M lactacystin) during the chase incubation. Proteasome inhibitors MG132 and lactacystin produced a slight increase in recovery, whereas the other inhibitors were without effect (data not shown). No inhibition of degradation was observed after treatment with Brefeldin A (data not shown). Taking into account the effects of temperature and these pharmacological agents, degradation of DBMsignal/ PHMs may occur in the endoplasmic reticulum and/or in a post-Golgi/TGN compartment.
DBMsignal/PHMs Is a Type II Membrane Protein-Our data indicate that the DBM signal is not efficiently removed from the DBMsignal/PHMs chimera in AtT-20 or CHO cells. To verify this conclusion and determine whether DBMsignal/ PHMs has a cytosolic domain, we used in vitro transcription/ translation with microsomal membranes. Whether the in vitro reaction was carried out in the presence or absence of microsomal membranes, a single 43-kDa protein was observed (Fig.  6). The identical sizes of products translated in the absence or presence of microsomal membranes is consistent with retention of the DBM signal. In contrast, translation of PHMs in the presence of microsomal membranes led to the synthesis of a smaller protein (38 kDa), indicating that cleavage of the signal sequence occurred.
With the goal of analyzing translocation of DBMsignal/ PHMs across the membrane, trypsin digestion of the translation product was carried out in the presence of intact microsomal membranes or after detergent permeabilization. Trypsin treatment of DBMsignal/PHMs translated in the presence of intact microsomal membranes decreased the molecular mass of the product by ϳ2 kDa (Fig. 6). Thus a segment of DBMsignal/ PHMs is exposed to the cytosol and susceptible to trypsin digestion. If DBMsignal/PHMs adopts the topology of a type II membrane protein, the residue Arg 19 of DBM would provide a site for cleavage by trypsin (Fig. 6, lower panel). In contrast, PHMs was protected from trypsin digestion as expected for a soluble protein sequestered within the lumen of the microsomal vesicles. In the absence of microsomal membranes or in the presence of microsomal membranes plus detergent, the DBMsignal/PHMs and PHMs products were both digested by trypsin.
These data suggest that the NH 2 terminus of the DBMsignal/ PHMs chimera is exposed to the cytosol and that the chimera has the topology of a type II membrane protein. We verified this prediction by biotinylating the in vitro translation product synthesized in the presence of intact membranes. N-Biotinylaminoethyl methanethiosulfonate, which reacts specifically with cysteine residues, was selected because rat DBM has a Cys residue at position 10. Biotinylated proteins were isolated using streptavidin and analyzed by SDS-PAGE and fluorography. As shown in Fig. 7, biotinylation of DBMsignal/PHMs occurred in the absence and presence of microsomal membranes. The more extensive biotinylation observed in the absence of microsomal membranes reflects the fact that all of the Cys residues in DBMsignal/PHMs should be accessible to N-biotinylaminoethyl methanethiosulfonate. In contrast, in the presence of microsomal membranes, only the single Cys residue preceding the putative transmembrane domain in DBMsignal/PHMs should be accessible, accounting for the less intense signal. As predicted for the topology shown, no biotinylated DBMsignal/ PHMs protein was isolated if the intact microsomes were digested with trypsin. The fact that DBMsignal/PHMs, with its five essential disulfide bonds, yields fully active PHM eliminates the possibility that the chimera adopts the topology of a type I membrane protein. Thus we conclude that the DBMsignal/PHMs chimera adopts a type II membrane topology with ϳ2 kDa of its NH 2 terminus exposed to the cytosol.
The Small Amount of PHMs Generated from DBMsignal/ PHMs Is Localized to Secretory Granules-AtT-20 cells contain secretory granules that store adrenocorticotropic hormone and other products of proopiomelanocortin cleavage along with many of the peptide-processing enzymes (41,51). PHMs is localized to these granules and vesicular structures localized to the perinuclear TGN region (Fig. 8C) (31). We examined the subcellular localization of DBMsignal/PHMs in AtT-20 and CHO cells using immunofluorescence microscopy. A diffuse reticular staining pattern was observed in both cell types (Fig.  8, A and B). DBMsignal/PHMs staining resembled staining for the immunoglobulin-binding protein, which is localized to the endoplasmic reticulum (34) (data not shown). The DBMsignal/ PHMs staining pattern was distinctly different from that of soluble PHMs (Fig. 8C). Interestingly, no PHM staining was FIG. 6. In vitro translation and trypsin protection of DBMsignal/PHMs. In vitro transcription/translation was carried out using cDNAs encoding DBMsignal/ PHMs and PHMs in the presence or absence of microsomal membranes (MM). Aliquots of the translation mixture were digested with 5 g of trypsin for 30 min in the presence (ϩ) or absence (Ϫ) of 1% Triton X-100. The samples were fractionated by SDS-PAGE and analyzed by fluorography. The asterisk indicates a nonspecific product. Similar results were obtained in four independent experiments. A schematic representation of the orientation of DBMsignal/PHMs in the microsomal membranes before and after trypsin digestion is shown. Open circles indicate the microsomal vesicles with DBMsignal/ PHMs represented by the PHMs protein (ribbon), and the hydrophobic region of the DBM signal is indicated by a filled rectangle. The NH 2 -terminal region of DBM is represented by a dark line showing the start Met and Arg 19 , the likely tryptic cleavage site. Small circles represent Cu 2ϩ bound to PHMs.
observed at the tips of the processes of DBMsignal/PHMs cells, suggesting that very little PHMs was localized to secretory granules in these cells.
Neither AtT-20 cells nor CHO cells express endogenous DBM. Because our previous studies on DBM trafficking indicated that AtT-20 cells were unable to store wild-type DBM in secretory granules (34), we investigated the possibility that DBMsignal/PHMs trafficking might be cell type-specific. PC12 cells are derived from adrenal medullary cells, which store DBM in their secretory granules. PC12 cells were transiently transfected with DBMsignal/PHMs and visualized with antiserum to PHM. Punctate staining for PHM was observed throughout the cell and excluded from the nucleus (Fig. 8D). The staining pattern observed for DBMsignal/PHMs in PC12 cells was similar to the pattern observed for VAMP2, an endogenous secretory granule and synaptic-like microvesicle protein (Fig. 8E) (52,53). Furthermore, the DBMsignal/PHMs staining pattern resembled the pattern observed when DBM was expressed transiently in PC12 cells (Fig. 8F) (54). As observed previously for DBM, localization of DBMsignal/PHMs to secretory granules is cell type-specific.
When expressed in AtT-20 cells, exogenous PHMs is stored efficiently in secretory granules; the small amount of PAM expressed endogenously in AtT-20 cells is also stored in secretory granules. We were therefore surprised that tethering active PHMs to the DBMsignal/anchor prevented its storage in granules. We used enzyme assays and Western blots to determine whether any PHM protein derived from DBMsignal/ PHMs reached a stimulatable compartment. Secretion under basal conditions was determined over two 1-h periods, and secretagogue was added for a final 1-h collection period; samples were analyzed by enzyme assay (Fig. 9A) and Western blot (Fig. 9B). The basal secretion rate of PHM activity from DBMsignal/PHMs cells (6.3% of cell content/h) was increased over nontransfected cells (3.3% of cell content/h), and the addition of BaCl 2 stimulated PHMs secretion ϳ4-fold (3.8 Ϯ 0.69, n ϭ 3). Western blot analysis revealed secretion of a 38-kDa PHMs protein during the two basal collection periods, and the addition of BaCl 2 caused a 2-fold increase in secretion of 38-kDa PHMs. When a Western blot of DBMsignal/PHMs cell extract was over-exposed, a small amount of 38-kDa PHMs was detected (Fig. 9B). Thus, although secretory granule localization could not be visualized by immunofluorescence microscopy, the small amount of 38-kDa PHMs derived from the chimera was stored in secretory granules.

DISCUSSION
DBMsignal/PHMs Is a Type II Membrane Protein-Several mechanisms for the membrane attachment of DBM have been tested and ruled out: a glycosyl phosphatidylinositol tail (1), noncovalent association with phosphatidylserine (3,55), and incorporation of myristate (54). The only region of rat (27), human (15,16), mouse (56), or bovine (11,19,21) DBM sufficiently hydrophobic to function as a membrane anchor is a 20-amino acid region near the NH 2 terminus (Fig. 1A). By appending the first 42 amino acid residues of rat DBM to a smaller homologous protein, we show clearly that it serves as an anchor, yielding a type II membrane protein.
In both AtT-20 and CHO cells, mature DBMsignal/PHMs (43 kDa) is ϳ5 kDa larger than mature PHMs (38 kDa). The mass predicted for the intact chimera and the mass observed for DBMsignal/PHMs synthesized in vitro in the absence of microsomal membranes is 43 kDa. Consistent with retention of the signal/anchor, DBMsignal/PHMs remains membrane-associated even after carbonate treatment. Interestingly, expression of bovine DBM in Drosophila Schneider 2 cells (10) and expression of truncated human DBM in AtT-20 cells (9) indicated that the DBM signal was not required for membrane association. Because the authors did not perform further studies to determine the association of DBM with the membrane, it is unclear if the exogenous DBM in these studies is actually anchored to the membrane.
Using von Heijne's analysis program (57) or PSORT (National Institute for Basic Biology), mouse and rat DBM are both predicted to adopt a type II membrane protein topology. Both the length of the hydrophobic core and the length of the sequence preceding the hydrophobic core contribute to predicting a low likelihood for signal sequence cleavage (13,14). Interestingly, analysis of the published human DBM sequence suggests that it has a cleavable signal. However, the 5Ј-untranslated region of human DBM contains an additional in-frame translational initiation site (Fig. 1A). Studies by Kozak (17,18) indicate that translation is initiated at the first acceptable start site, making it highly likely that an additional 15 amino acids are present. In this case, human DBM is also expected to assume the topology of a type II integral membrane protein.
Our results are consistent with earlier studies in which rat DBM mRNA was translated in a cell-free system, and NH 2 -terminal sequence analysis detected an uncleaved signal sequence (12). In addition, some 20 -30% of purified bovine membranous DBM retains its signal sequence (11). It has been shown that DBM exists in both soluble and membrane-bound forms in secretory granules of adrenal chromaffin cells and catecholamine-producing neurons in the central and peripheral nervous systems (5,58). Consistent with its occurrence in soluble and membrane forms, DBM secretion is stimulated from chromaffin cells in response to secretagogue. This indicates that significant cleavage of the signal/anchor domain occurs in chromaffin cells. In contrast, very little cleavage of the signal/ anchor domain was observed in AtT-20 or CHO cells expressing DBMsignal/PHMs. The inability of AtT-20 or CHO cells to remove the signal/anchor domain suggests that this cleavage requires enzymes specific to cells that normally produce DBM. The cell type-specific trafficking of DBMsignal/PHMs may simply reflect the occurrence of more extensive cleavage of the chimera in PC12 cells than in AtT-20 cells. Such cell-type specificity is not unexpected, because trafficking of soluble DBM to secretory granules fails to occur in AtT-20 cells (34).
The DBM signal/anchor Domain Contains Trafficking Information-The specific activity of PHMs secreted into the medium from AtT-20 cells was similar to that of PHMs present in CHO and AtT-20 cells expressing DBMsignal/PHMs. The specific activity of PHMs indicates that attaching the signal/anchor domain of DBM to PHMs did not impair its ability to fold and acquire its normal active conformation. Despite this, DBMsignal/PHMs undergoes rapid degradation. Only 26% of the newly synthesized protein remained in AtT-20 cells after a 2-h chase (Fig. 4, A and B). Interestingly, the degradation process occurs more slowly in CHO cells. When exit from the TGN was blocked by incubation of AtT-20 cells at 20°C, degradation of newly synthesized DBMsignal/PHMs was largely blocked (Fig.  5, A and B), suggesting that degradation occurs in a post-TGN compartment. Incubation of cells with lysosomal inhibitors (NH 4 Cl plus methylamine, chloroquine, and leupeptin) failed to diminish degradation (data not shown). Incubation with proteasome inhibitors (MG132 and lactocystin) did slightly diminish DBMsignal/PHMs degradation, suggesting a role for the proteasome (data not shown).
Taken together, our data indicate that the NH 2 terminus of DBMsignal/PHMs is exposed to the cytosol. Because fully ac- AtT-20 DBMsignal/PHMs cells were incubated in CSFM air for two sequential 1-h periods (B1 and B2) to determine the basal secretion rate; 1 mM BaCl 2 was added during the subsequent 1-h period to assess stimulated secretion. A, PHM activity in the basal and stimulated samples was determined as described under "Materials and Methods." B, secreted PHMs was visualized by Western blot analysis. The asterisk indicates a band detected in both nontransfected and transfected AtT-20 cells. Analysis of the cell extract from transfected cells (T) expressing DBMsignal/PHMs as well as nontransfected cells (N) is also shown. The molecular masses are indicated in kDa. Similar results were obtained in three independent experiments.
tive PHMs is produced in cells expressing DBMsignal/PHMs, it is clear that essential disulfide bonds are formed correctly in the lumen of the endoplasmic reticulum. Trypsin digestion of DBMsignal/PHMs translated in the presence of microsomal membranes decreased its mass by 2 kDa, indicating that a segment of the NH 2 terminus of the chimera would be exposed to the cytosol (Fig. 6). Coincident with these findings, cleavage of the rat DBM signal/anchor at Arg 19 would decrease its mass by this amount. Biotinylation of the in vitro translation product synthesized in the presence of intact membranes also indicated that DBMsignal/PHMs adopts a type II membrane topology with ϳ2 kDa exposed to cytosolic factors (Fig. 7).
The small amount of cellular 38-kDa PHMs protein generated in AtT-20 cells expressing DBMsignal/PHMs is stored in secretory granules and released in response to BaCl 2 (Fig. 9). The amount of PHMs stored in secretory granules is too small to allow visualization by immunostaining, which reveals diffuse reticular staining for DBMsignal/PHMs primarily in the endoplasmic reticulum (Fig. 8, A and B). Similar staining was observed for wild-type DBM expressed in AtT-20 cells where DBM is not stored in secretory granules (34). Furthermore, staining of DBMsignal/PHMs expressed in PC12 cells (Fig. 8D) suggests that its sorting is cell type-specific. Therefore, it is interesting that retention of the DBMsignal/anchor domain apparently prevents storage of the chimera in secretory granules in AtT-20 cells.
The trafficking of soluble and membrane proteins to secretory granules is poorly understood but clearly involves very different signals (59,60). PHMs is efficiently targeted to secretory granules in AtT-20 cells. In contrast, membrane PAM, a type I integral membrane protein, is largely localized to the TGN area. Truncation of the entire cytosolic domain of membrane PAM yields an integral membrane protein that resides largely on the plasma membrane. Although exogenous proinsulin is efficiently targeted to secretory granules of AtT-20 cells (61), appending the transmembrane and cytoplasmic domains of CD5, a plasma membrane protein (62) in human T lymphocytes, yields a chimeric protein that is largely localized to the plasma membrane (63). Based on the limited data available, the targeting signals associated with soluble secretory proteins are dominated by the targeting signals associated with integral membrane proteins. Certainly the DBMsignal/anchor domain blocked access of fully active PHMs to secretory granules. Trafficking of endothelin-converting enzyme, a type II membrane protein, to lysosomes versus plasma membrane is governed by signals in its NH 2 -terminal cytosolic domain (64).
Our data indicate that the signal/anchor domain of DBM, with ϳ2 kDa exposed to the cytosol, contains cell type-specific targeting information. When appended to PHMs, it prevents access to secretory granules. As part of DBM, it is responsible for its membrane association and is likely to play a key role in the targeting of DBM to secretory granules in chromaffin cells and adrenergic neurons.