Localization of Metallocarboxypeptidase D in AtT-20 Cells

Carboxypeptidase D (CPD) is a recently discovered metallocarboxypeptidase that is predominantly located in thetrans-Golgi network (TGN), and also cycles between the cell surface and the TGN. In the present study, the intracellular distribution of CPD was examined in AtT-20 cells, a mouse anterior pituitary-derived corticotroph. CPD-containing compartments were isolated using antibodies to the CPD cytosolic tail. The immunopurified vesicles contained TGN proteins (TGN38, furin, syntaxin 6) but not lysosomal or plasma membrane proteins. The CPD-containing vesicles also contained neuropeptide-processing enzymes and adrenocorticotropic hormone, a product of proopiomelanocortin proteolysis. Electron microscopic analysis revealed that CPD is present within the TGN and immature secretory granules but is virtually absent from mature granules, suggesting that CPD is actively removed from the regulated pathway during the process of granule maturation. A second major finding of the present study is that a soluble truncated form of CPD is secreted mainly via the constitutive pathway in AtT-20 cells, indicating that the lumenal domain does not contain signals for the sorting of CPD to mature secretory granules. Taken together, these data are consistent with the proposal that CPD participates in the processing of proteins within the TGN and immature secretory vesicles.

Most peptide hormones and neurotransmitters are produced from larger precursors by limited proteolysis. Initially, the prohormone precursors are processed at multiple basic amino acid cleavage sites by a family of endoproteases collectively known as prohormone convertases (PC) 1 (1,2). The subsequent processing step is mediated by carboxypeptidases, which remove the basic amino acids from the C terminus of the peptide to generate either the bioactive product, or a precursor for the formation of the C-terminal amide group (3,4).
Carboxypeptidase E (CPE, also known as carboxypeptidase H and enkephalin convertase) is the major carboxypeptidase involved with the processing of many peptides (3,4). Within several tissues, CPE has been localized to the peptide-contain-ing secretory granules (5)(6)(7). The involvement of CPE in peptide processing is evident from the finding that fat/fat mice have a reduced ability to convert proinsulin into insulin (8). A missense mutation in the CPE gene is the molecular basis for the fat mutation (8). Although a full-length CPE protein is translated, the protein is inactive due to a point mutation in the coding region of the gene (8,9). Despite the absence of functional CPE, the mature forms of numerous neuropeptides are detectable in brain and other tissues of fat/fat mice (8, 10 -13). The presence of correctly processed peptides indicates that an additional enzyme is able to compensate for the deficiency of CPE.
Recently, a second carboxypeptidase designated metallocarboxypeptidase D (CPD) has been identified in the secretory pathway of bovine pituitary glands (14). This protein was independently discovered as gp180, a duck protein that binds duck hepatitis B virus particles (15). In duck and rat, CPD is present in many tissues (15)(16)(17)(18)(19) suggesting a broad function. CPD cDNA has been cloned and sequenced from human, rat, duck, Drosophila, and Aplysia (16, 18, 20 -22). In general, the protein is highly conserved among species, and shows similar enzymatic properties (14,17,22,23). Human, rat, and duck CPD cDNAs encode proteins that contain three carboxypeptidase-like domains, followed by a transmembrane domain and a 58-residue cytosolic tail. The cytosolic tail contains several consensus routing motifs found in other integral membrane peptide processing enzymes such as furin and peptidylglycine-␣-amidating monooxygenase (PAM) (24).
CPD is predominantly located in the TGN, and also cycles between the cell surface and the TGN (25). The co-localization of CPD and the TGN endoprotease furin (25) supports the hypothesis that these two enzymes may function in the same pathway. We were previously unable to detect CPD in mature secretory granules by light microscopy. As mature secretory vesicles contain many of the peptide processing enzymes (prohormone convertases 1 and 2, CPE, PAM), the lack of detectable CPD in this compartment raised doubts as to whether CPD can participate in peptide processing. Although the processing of some neuropeptides begins in the TGN, immature vesicles appear to be the most productive proteolytic compartments. For example, the processing of proinsulin begins in immature secretory granules in ␤-cells (26). Also, the generation of ACTH and ␤-endorphin from the larger precursor proopiomelanocortin (POMC) in AtT-20 cells is predominantly a post-TGN event (27,28), although POMC processing may be initiated in the TGN (29 -31).
Soluble forms of CPD have been identified in various bovine and rat tissues (17). The intracellular distribution of the soluble CPD was not previously examined. Interestingly, both the soluble and membrane-associated forms of PAM are packaged into mature secretory granules in AtT-20 cells (32), indicating that the lumenal domain of PAM contains sorting information.
In the present study, we further characterized the intracellular distribution of CPD using immunoisolation and electron microscopic approaches. We have also examined the intracellular distribution of a soluble truncated form of CPD. Although neither the full-length or the truncated form of CPD are efficiently routed to mature secretory vesicles, both forms are present in immature secretory vesicles. These findings strongly support the proposal that CPD functions in the processing of prohormones and other proteins that transit either the regulated or constitutive secretory pathways.

Generation of Constructs and Expression of Proteins in AtT-20
Cells-gp180 and gp170 constructs previously described for baculovirus expression (23) were subcloned into the pcDNA3 expression vector (Invitrogen) and transfected into AtT-20 cells using the standard calcium phosphate procedure (33). Stable cell lines were selected using 0.6 mg/ml Geneticin (G418). Cells expressing both constructs were identified by Western blot analysis. For each of these constructs, several positive clones were selected and analyzed as described below; the data shown are representative of two or three separate clones.
Antibodies-The antiserum to duck CPD recognizes both the soluble and membrane-associated forms of the protein (i.e. gp170 and gp180) but does not cross-react with mouse CPD (24). The antiserum to the C-terminal cytosolic tail of CPD (17) recognizes both rat and duck proteins. The antiserum raised against a C-terminal peptide of bovine CPE recognizes the mouse protein (9). Antisera to ACTH and ␤-endorphin that recognize both precursor and processed peptide (31)  For immunoisolation, polyclonal antibodies against the C-terminal cytosolic tail of CPD were subjected to affinity purification using glutathione S-transferase-CPD cytosolic tail peptide coupled to Sepharose 4B by the standard cyanogen bromide method (Sigma). Affinity-purified antibodies were concentrated to 1 mg/ml in PBS, and used for the immunoisolation procedure as described below.
Immunoaffinity Isolation of CPD/gp180-containing Vesicles-Typically, six confluent 10-cm plates of AtT-20 cells expressing gp180 were used for the immunoisolation. Cells were washed three times with cold PBS, scraped from the plates, and gently pelleted at 700 ϫ g. . The suspension was passed five times through a 22-gauge needle, and then homogenized in a tight-fitting Dounce homogenizer (25 strokes). The resulting homogenate was centrifuged for 7 min at 1000 ϫ g twice, and the supernatant was layered onto 4 ml of 1.6 M sucrose and centrifuged in a SW41 rotor (Sorvall) for 1 h at 200,000 ϫ g. Typically, 1 ml of a microsomal fraction was collected from the interface of the 0.25 and 1.6 M sucrose, and then subjected to immunoisolation. The affinity resin for the immunoisolation was prepared by incubation of 100 l of Pansorbin (Calbiochem) with 100 g of affinitypurified antibodies to the CPD C-terminal cytosolic tail. An aliquot (300 l) of the microsomal fraction was incubated for 3 h at 4°C with gentle agitation with 100 l of the affinity resin in a final volume of 500 l containing 140 mM NaCl and the protease inhibitors described above. The Pansorbin resin was recovered by low speed centrifugation and was washed three or four times with the same buffer, and then vesicleassociated proteins were subjected to a detergent elution using 1% Triton X-100. IgG-associated gp180 was eluted from the beads by boiling in 1% SDS gel loading buffer for 5 min. Aliquots of each fraction were subjected to Western blot analysis using antisera to various secretory pathway proteins at 1:1000 dilution.
Labeling of AtT-20 Cells with [ 35 S]Met-AtT-20 cells expressing gp170 were labeled with [ 35 S]Met (100 Ci/ml) for 15 min, washed twice with PBS, and then incubated in Dulbecco's modified Eagle's medium for different periods of time. Media were removed, cells were washed with PBS, and then frozen in 10 mM NaAc, pH 5.5, with 1 mM phenylmethylsulfonyl fluoride. The cells and the media were then subjected to immunoprecipitation using the antiserum to either duck CPD or CPE as described (30). To test whether CPD-containing compartments contain the neuropeptide precursor POMC and its proteolytic fragments, AtT-20 cells expressing gp180 were either continuously incubated with [ 35 S]Met (100 Ci/ml) for 8 h, or pulsed for only 20 min and then chased for 0 or 90 min in the presence of unlabeled Met. Following the labeling, the cells were washed three times with cold PBS and then subjected to the immunoisolation procedure described above. The resulting fractions were subjected to immunoprecipitation using antisera to either ACTH or ␤-endorphin, and analyzed on 10% polyacrylamide Tricine gels.
Regulated Secretion-To examine whether gp170 is secreted via the regulated pathway, the cells expressing gp170 were grown on 35-mm cell culture dishes to 90% confluence. The cells were washed twice with PBS, then treated with the secretagogue 5 mM 8-Br-cAMP or control media for 30 min and the secreted proteins analyzed by Western blotting. To examine the regulated secretion of [ 35 S]Met-labeled proteins, cells were labeled for 15 min with [ 35 S]Met, chased for 2 h, washed three times with PBS, and then incubated for 30 min in either control media or with 5 mM 8-Br-cAMP. The media were subjected to immunoprecipitation using the antiserum to either CPE or duck CPD.
Immunofluorescence Analysis-Transfected AtT-20 cells were cultured on growth-supporting glass coverslips (Fisher Scientific). Cells were washed with PBS, fixed in 4% paraformaldehyde for 15 min, and then permeabilized for 15 min in 0.1% Triton X-100 in PBS. After 1 h of blocking in 3% bovine serum albumin, the cells were immunostained for 1 h with the primary antisera (1:1000 dilution). Cells were washed with PBS containing 0.2% Tween 20 and then incubated with fluoresceinlabeled anti-mouse or rhodamine-labeled anti-rabbit secondary antibody (Vector Laboratories Inc., 1:200 dilution) for 1 h, followed by extensive PBS washing. Immunofluorescence staining was examined using a Bio-Rad confocal microscope.
Electron Microscopy Analysis-Stably transfected AtT-20 cells expressing either gp180 or gp170 were fixed for 30 min at room temperature in 4% formaldehyde, 0.2% glutaraldehyde in PBS, then dehydrated, embedded in LR-White resin according to standard procedure, sectioned, and mounted on either gold or nickel grids. Each grid was incubated for 30 min in blocking solution containing 10% goat serum, 2.5% bovine serum albumin, 0.1% Tween 20 in PBS at pH 8.2, and then incubated for 1 h in the same solution containing the antiserum to either duck CPD or CPE (final dilution 1:200). Cells were extensively washed in 0.1% Tween 20 in PBS, and incubated for 2 h with 10-nm colloidal gold (BioCell) coupled to anti-rabbit antibodies in blocking solution (dilution 1:10). After extensive washes, sections were stained for 2 min with uranyl acetate in 30% ethanol and examined with a Jeol 100CX electron microscope. For conventional electron microscopy, the cells were fixed in 2% glutaraldehyde, then postfixed in 1% OsO 4 and embedded in LR-White resin.
To quantitate the intracellular distribution of gp180, gp170, and CPE, electron micrographs that were immunogold-labeled with antibodies to either duck CPD or CPE were scanned in random order. In total, about 1000 gold particles/grid were analyzed, and the percentage of total label that was found in specific compartments was determined. The main morphological criterion to distinguish different subclasses of the secretory granules was the size of the vesicles and the size of the dense cores (34), which were measured from negatives taken at magnification ϫ54,000. An additional morphological criteria was the presence of the lighter core in immature granules surrounded by a broad electron-lucid peripheral zone (this zone was much broader in granules at the low state of condensation). Light aggregates of condensing proteins were observed in the dilated cisterns of the TGN, consistent with previous studies that formation of the dense core aggregates begins at the level of the TGN in AtT-20 cells (35). These electron-dense granules at the low state of condensation, which may represent cross-sections of the TGN or early immature vesicles, had an average diameter of 280 nm and dense cores that were too variable to accurately determine an average. Electron-dense granules at an intermediate state of condensation had an ovoid or spherical shape, and were in the process of detaching or already detached from the TGN. The average diameters of these granules were 170 nm, and their dense cores were 120 nm. These vesicles closely resemble immature secretory granules previously described in AtT-20 cells (35). Another type of vesicle was found in close proximity to the plasma membrane; these vesicles had a regular spherical shape and highly condensed dense cores, which are characteristic features of mature secretory granules in endocrine cells (35). The av-erage diameters of mature secretory granules were typically 150 nm and their dense cores were 110 nm, which is consistent with previous studies on AtT-20 cells (28,36,37).

RESULTS
We have previously demonstrated that endogenous CPD is predominately localized to the perinuclear region in AtT-20 cells (25). To biochemically determine the proteins that colocalize with CPD, we used an immunoisolation technique. For these studies, microsomes from cells stably transfected with full-length duck CPD (gp180) were isolated at the interface between a 1.6 M and a 0.25 M sucrose layer (Fig. 1A, Ms). Typically, about 30% of the total gp180 in the cell homogenate was recovered in the microsomal fraction (Fig. 1B). Similarly, 20 -30% of the calnexin (an ER membrane protein), SNAP-25 (a plasma membrane protein), and syntaxin-6 (a TGN protein) were also recovered in this microsomal fraction (Fig. 1B). None of these proteins were detected in the 0.25 M or the 1.6 M sucrose layers (Fig. 1B), indicating no enrichment or loss of these organelles during the preparation. For immunoisolation, Pansorbin was precoated with the affinity-purified antibodies to the cytosolic tail of CPD, incubated with the microsomal fraction from AtT-20 cells, and then both bound and non-bound materials were analyzed by Western blot using antibodies to various secretory pathway markers (Fig. 1C, right panels). In control experiments, when rabbit IgG was used instead of antibodies to the CPD tail, most of the gp180 immunoreactivity and secretory markers were recovered in non-bound fractions (Fig. 1C, left panels), indicating that nonspecific binding of microsomes to the affinity carrier is very low. When using antibodies to the CPD tail, most of the gp180 (typically Ͼ95%) is recovered in the bound fraction (Fig. 1C, top). Furin, TGN38, and syntaxin 6, which have been previously localized to the TGN and immature secretory granules (38 -40), are exclusively associated with gp180-containing vesicles (Fig. 1C, right panels). In contrast to the TGN proteins, CPE and PC1 which were previously found within mature secretory granules (5,41,42), demonstrate only a partial co-localization with the gp180 compartment (Fig. 1C, right panel). Typically, 20 -30% of CPE, and 40 -50% of PC1 were recovered in non-bound fractions. To further validate the specificity of our immunoisolation technique, we probed bound and non-bound fractions with antibodies to cathepsin B, calnexin, and SNAP-25 (Fig. 1C). Cathepsin B and SNAP-25 are exclusively recovered in the non-bound fraction (Fig. 1C), indicating that lysosomes and plasma membrane are not present in the immunoisolates. Approximately 30% of the calnexin is associated with the CPD-containing FIG. 1. Immunoisolation of the gp180-containing compartment. A, AtT-20 cells were homogenized (H) and then subjected to centrifugation at 1000 ϫ g, as described under "Materials and Methods." The supernatant from this (S1) was then layered onto 4 ml of 1.6 M sucrose and centrifuged for 1 h at 200,000 ϫ g. Microsomes were collected from the interface of the two layers. B, equivalent aliquots of the homogenate, the S1 fraction, the 0.25 M sucrose layer (0.25), the 1.6 M sucrose layer (1.6), and the microsomal fraction (Ms) were analyzed by Western blotting using antisera to the indicated proteins. C, microsomes prepared from AtT-20 cells expressing gp180 were incubated for 3 h at 4°C with Pansorbin precoated with either affinity-purified antibodies to the CPD C-terminal cytosolic tail (␣CPD) or rabbit IgG (IgG). The aliquots of bound (B) and non-bound (NB) fractions were subjected to Western blot analysis using antisera either to the N-terminal portion of duck CPD or to different secretory pathway markers. The positions of prestained protein standards are indicated. Except for the analysis of cathepsin B, calnexin, and SNAP-25, which was performed twice, the entire experiment was repeated five times with similar results. compartment (Fig. 1C), consistent with the expectation that the ER contains newly synthesized CPD. The finding that the plasma membrane and ER membrane proteins are not efficiently isolated in the CPD-containing organelles further suggests that co-sedimentation of the TGN proteins and gp180 is not due to nonspecific binding of membrane proteins to the affinity resin. Identical immunoisolation procedure using wild type AtT-20 cells demonstrated a similar distribution of the various secretory pathway markers (data not shown), indicating that this pattern of distribution is not due to overexpression of gp180 in AtT-20 cells.
To test whether the CPD-containing compartments contain the neuropeptide precursors or their proteolytic fragments, AtT-20 cells were labeled with [ 35 S]Met for 8 h and then subjected to the immunoisolation procedure with antibodies to the CPD cytosolic tail. To reduce the possibility of nonspecific proteolysis, a mixture of protease inhibitors was included throughout the isolation procedure. Typically, 30 -40% of the total POMC and POMC-derived material was recovered in the microsomal fraction (data not shown). Pansorbin-bound and nonbound materials were subjected to immunoprecipitation using antibodies either against ACTH or ␤-endorphin. The antisera against ACTH used in the present study recognizes both POMC and its processed product (31). Typically 70 -80% of the ACTH (both glycosylated and non-glycosylated forms) was recovered in the bound fraction ( Fig. 2A, right panel). In experiments with control IgG, most of the ACTH is found in the non-bound fraction ( Fig. 2A, left panel), although the 30-kDa POMC shows some nonspecific binding ( Fig. 2A, left panel). When immunoisolated material was immunoprecipitated with antibodies to ␤-endorphin, the bulk of the ␤-endorphin-reactive material was recovered in the bound fraction (data not shown). As an additional control for nonspecific proteolysis, we also immunoprecipitated the immunoreactive ACTH peptides in the homogenate, intermediate fractions and microsomes, prior to the vesicle immunoisolation. All fractions showed similar relative amounts of ACTH versus the larger peptides, indicating that no significant proteolysis of POMC occurred during the immunoisolation (data not shown).
To examine the effect of shorter labeling times on the pattern of immunoreactive ACTH peptides, cells were labeled with [ 35 S]Met for 20 min, chased for 0 or 90 min in unlabeled Met, and then subjected to vesicle immunoisolation using an iden-tical procedure. When cells were labeled for only 20 min, without a chase, only POMC-sized material was detected in both the initial homogenate and the immunoisolated vesicles (Fig.  2B, H and Ms). This result further suggests that the POMC processing does not occur during the vesicle immunoisolation procedure. When the cells were chased for 90 min prior to vesicle immunoisolation and immunoprecipitation, ACTHsized peptides were detected (Fig. 2B). The reduction in the amount of [ 35 S]Met label recovered upon immunoprecipitation after the 90 min chase is due to both loss of Met upon cleavage of POMC (ACTH contains 1 of the 3 Met in POMC) and secretion of POMC peptides from AtT-20 cells. Approximately 40% of the ACTH immunoreactivity was recovered in the non-bound fraction after the 90-min chase (Fig. 2B), suggesting that this pool has left the CPD-containing compartment. Taken together, these results demonstrate that a substantial portion of POMC processing by endopeptidases is completed in CPDcontaining compartments.
To further investigate the intracellular distribution of CPD, we performed double-labeling of AtT-20 cells with various antisera (Fig. 3). Co-staining of cells with antibodies to syntaxin 6 and TGN38 demonstrates complete co-localization of these proteins (Fig. 3, bottom) indicating that syntaxin 6 is localized to the TGN in AtT-20 cells, as previously found in other cell lines (40). The majority of the ACTH-reactive material is colocalized with syntaxin 6 in the perinuclear region (Fig. 3). In addition, the anti-ACTH antiserum also labels large discrete vesicles distributed throughout the cytoplasm and at the processes of the cells (Fig. 3, arrows), which presumably represent the mature secretory granules. gp180 is found exclusively in a perinuclear compartment which overlaps with the distribution of syntaxin 6 (Fig. 3, top), suggesting that the majority of gp180 does not enter the mature ACTH-containing vesicles.
Although CPD was originally described as a 180-kDa integral membrane protein, a small amount of CPD was previously purified from bovine and rat tissues as a 170-kDa soluble protein that did not react with the antiserum raised against the C-terminal tail (17). To further explore the intracellular distribution of the soluble form of CPD, we created a deletion within the coding region of gp180 that removes the transmembrane and cytosolic domains of the protein, and stably expressed the resulting construct (named gp170) in AtT-20 cells. Cells expressing gp170 were double-labeled with antibodies against duck CPD and syntaxin 6 and examined by immunofluorescence. A large portion of gp170 is localized to the perinuclear region and co-stained with syntaxin 6 antibodies (Fig. 3). In addition, gp170 also shows a diffuse distribution throughout the cytoplasm (Fig. 3, second row). The diffuse pattern of gp170 is distinct from the punctate distribution of ACTH (Fig. 3).
To obtain further structural information about the CPDcontaining compartments, we used immunogold electron microscopy (Fig. 4). A conventional electron micrograph reveals secretory granules and TGN-derived vacuoles containing condensing material (Fig. 4A). Cells expressing gp180 were processed for immunogold electron microscopy and stained with the antiserum directed against duck CPD (Fig. 4, B-F). The majority of the CPD-reactive material is concentrated along the trans side of the Golgi cisternae (Fig. 4, B and C, filled arrows), and within vesicles containing electron-dense core material at an early state of condensation (Fig. 4, B-D, open arrows). These vesicles may represent cross-sections of the TGN or early im-mature secretory granules. CPD staining was also detected within vesicles with dense cores at an intermediate stage of condensation that resemble immature secretory granules (Fig.  4, C-E, open arrowheads), and even in a small number of dense core vesicles located in close proximity to the plasma membrane that resemble mature secretory granules (Fig. 4F, filled  arrowheads). The small number of gold particles detected associated with dense core vesicles is likely to reflect the real presence of CPD in these compartments rather than background labeling, as the gold particles were often present in clusters of 3 or more. Additionally, virtually no labeling was detected in mitochondria or nucleus, which are not expected to contain CPD. In control experiments in which the primary antiserum was omitted, only background staining was observed in the TGN and vesicles (data not shown). Quantitation of approximately 1000 gold particles demonstrates that 84% of gp180 is located in the TGN, 10% is found within moderatelycondensed granules, 4% is located within granules at an intermediate state of condensation, and only 2-3% is recovered within mature granules (Table I). Taken together, the electron microscopy, immunocytochemical, and immunoisolation results further support the idea that gp180 predominately functions at the level of the TGN, and to a lesser extent in the immature secretory vesicles.
Immunogold electron microscopic analysis was also performed on the AtT-20 cell line expressing gp170 (Fig. 4, G-I). In addition to the TGN localization, gp170 is also found within secretory vesicles containing electron-dense material (Fig. 4I, open arrowheads, and Table I) that are morphologically similar to the gp180-containing vesicles. As with gp180, immunoreactive gp170 was not generally detected in mature granules (Fig.  4H, Table I). A large fraction of gp170 was also found in small clusters (40 -50 nm in diameter) distributed throughout the cytoplasm (Fig. 4G, crossed arrows). These structures may represent small constitutive vesicles that have been previously described (43,44), or cross-sections of the ER. In contrast to the pattern of CPD distribution, the majority of the immunoreactive CPE was localized in the dense core granules that are morphologically similar to immature and mature secretory vesicles (Fig. 4J, Table I).
To further explore the routing pathway of gp170, we determined the secretion kinetics of the transfected gp170 and the endogenous CPE in AtT-20 cells. Cells were labeled for 15 min with [ 35 S]Met, chased for different periods of time, and then both cells and media were subjected to immunoprecipitation using antibodies to either duck CPD or mammalian CPE. The secretion rate of gp170 is fairly rapid; after a 2-h chase approximately 90% of the labeled gp170 is present in the media (Fig.  5, A and B). In contrast, a large portion of CPE (50%) is retained within the cells after 2 h of the chase period, and the CPE secretion approaches a plateau (Fig. 5, A and B). These results are consistent with electron microscopy data, demonstrating that the bulk of gp170 does not usually enter dense core secretory granules, whereas CPE is targeted to this compartment (Fig. 4).
To further characterize the gp170 pathway, we tested whether the secretion of gp170 could be induced by secretagogues. Two different approaches were used to study the secretion of gp170. First, cells were incubated for 30 min in the presence or absence of Br-cAMP and then media were analyzed by Western blot (Fig. 6, left). Although the secretion of CPE was stimulated 2.5-3-fold by the secretagogue treatment, the secretion of gp170 was not significantly affected by the secretagogue treatment (Fig. 6, left). Two other secretagogues (forskolin and a phorbol ester) also did not significantly stimulate the secretion of gp170 from the AtT-20 cells (data not shown). These Stably transfected cells expressing either gp170 (second row) or gp180 (all other rows) were subjected to dual immunofluorescent staining as described under "Materials and Methods" using monoclonal antibodies to syntaxin 6 (right column) or polyclonal antibodies to different proteins (left column). All images represent composites of 1-m optical sections through the cell. To obtain images within the linear range of exposure, the tonal range of each image was adjusted appropriately. Arrowheads, perinuclear staining; arrows, punctate vesicles detected with the ACTH antiserum. Bar, 10 m. results indicate that the bulk of gp170 undergoes constitutive secretion. As a small amount of radiolabeled gp170 was detected within the cells after several hours of chase (Fig. 5A), we tested whether this pool of gp170 could undergo regulated release in response to a secretagogue. Cells were labeled with [ 35 S]Met for 15 min, chased for 2 h, washed, and then incubated for an additional 30 min in either control media or media containing Br-cAMP. Both gp170 and CPE are released to the media in the regulated fashion (Fig. 6, right). Quantitation of three independent experiments demonstrated a statistically significant stimulated release of both gp170 (Fig. 6, right) and CPE (data not shown) in response to the secretagogue, indicating that a small fraction of the gp170 undergoes regulated exocytosis. These findings are consistent with the electron microscopy and immunocytochemistry data, and taken together suggest that the bulk of gp170 does not enter mature secretory vesicles in AtT-20 cells. DISCUSSION CPD was previously localized by light microscopy to a perinuclear compartment that overlaps with the distribution of the TGN endopeptidase furin in AtT-20 cells (25). In the present study, biochemical and electron microscopic analysis supports the localization of CPD to the TGN, and in addition has provided evidence that a fraction of the CPD is present in immature secretory granules. The processing of many neuroen-  docrine peptides begins in the TGN and continues in the immature secretory granules. The acidification of the TGN and secretory granules is thought to play an important role in the activation of the processing enzymes such as prohormone convertase 1 and 2 (28,45). Importantly, CPD has a broad pH optimum (14,23) and could function both in the environment of the TGN and within acidic secretory granules. In contrast, CPE is active only at acidic pH values and is essentially inactive at neutral pH (46), consistent with a role for CPE in the processing of neuroendocrine peptides that occurs in the late secretory pathway. The endopeptidases prohormone convertase 1 and 2 have been also localized to mature secretory granules (28,42), indicating that CPE is likely to be a functional partner for these endopeptidases whereas CPD is a functional partner for furin. Co-localization of CPD with the proteolytic fragments of POMC, together with electron microscopy data, strongly suggest that CPD enters the regulated secretory pathway in AtT-20 cells. However, in contrast to CPE, only a small fraction of CPD routes to mature secretory granules. It has been previously demonstrated that some TGN membrane proteins including furin and mannose 6-phosphate receptor enter the immature secretory granules of the regulated pathway in neuroendocrine cells (47,48). As with CPD, both of these proteins are largely removed from immature granules during the maturation and are not generally detected in mature secretory granules. The mannose 6-phosphate receptor is sorted from immature secretory vesicles to endosomes by an AP-1-and clathrin-dependent process (48). Consistent with this, the furin cytoplasmic tail interacts with AP-1, a component of the TGN clathrin sorting machinery (47). This interaction is dependent on phosphorylation of the furin cytoplasmic tail by casein kinase II (47); mutation of the casein kinase II phosphorylation sites results in mistargeting of furin to mature secretory granules (47). Interestingly, potential casein kinase II phosphorylation sites are also found within an acidic residue-rich cluster of the CPD cytosolic tail. Mutation of these sites altered the trafficking of CPD and resulted in detectable staining of the mutant protein in the tips of AtT-20 cell processes (24). This suggests that a similar mechanism of retrieval from immature secretory granules is involved in CPD trafficking. Our finding that CPD is more abundant within vesicles with a low degree of condensation than in highly condensed vesicles (Fig. 4, Table I) supports the idea that CPD is progressively removed from immature granules in the process of their maturation. The retrieval of CPD from immature vesicles may be required for the targeting of CPD to the recycling pathway and its efficient return to the TGN (Fig. 7).
Another important finding of the present study is that the soluble form of CPD (gp170) also enters immature secretory vesicles, but is not abundant in mature vesicles. As a result, the bulk of this protein is secreted constitutively from AtT-20 cells. A soluble form of CPD has been previously detected in various tissues including liver, kidney, and brain (17). It is likely that this form resembles the truncated gp170 form, as the endogenous soluble form is 170 kDa and does not bind the antiserum directed against the C-terminal tail of CPD (17). Our finding that gp170 also enters the regulated secretory pathway supports the idea that the soluble form of CPD may be also involved in the processing of neuropeptides in immature vesicles. As the majority of gp170 is secreted via the constitutive or constitutive-like pathway, it is also possible that the soluble CPD functions within constitutive vesicles or outside the cell. Our previous finding that gp180/CPD cycles between the plasma membrane and the TGN (24,25) raises the possibility that the full-length form of CPD functions within endocytic compartments. Interestingly, it has been demonstrated that furin cleaves precursor proteins in both the exocytic and endocytic pathways (49). Several viral coat proteins including influenza hemagglutinin and human immunodeficiency virus gp160 are processed by furin when co-expressed with the endopeptidase (49). Furin-mediated cleavage of many toxins occurs either at the plasma membrane or in endosomes (49). As furin cleaves to the C-terminal side of basic amino acids, the product of furin cleavage will contain C-terminal basic residues. If these need to be removed for the products to be biologically active, as is the case for most neuroendocrine peptides, CPD is a likely candidate for this activity based on its tissue distribution, intracellular distribution, and pH optimum. The involvement of CPD in important biological functions is supported by the fact that the mutations in the Silver gene, the Drosophila homologue of CPD, are embryonic lethal (21).
The finding that gp170 is primarily secreted in a constitutive manner from AtT-20 cells also suggests that the lumenal domain of CPD is not sufficient for the efficient targeting of the protein to mature secretory granules, although it is sufficient for entry into immature secretory vesicles. In contrast, soluble forms of PAM are efficiently packaged into storage granules in AtT-20 cells, indicating that the lumenal domain of PAM plays a major role in sorting into the regulated pathway (32). The mechanism of the protein sorting to the regulated secretory pathway remains unclear, and two alternative models have been proposed. The active sorting model postulates that regulated proteins and hormones bind to the sorting receptor in the TGN, and then are delivered to immature secretory vesicles. Proteins lacking specific signals for sorting into the regulated pathway follow the TGN-derived constitutive pathway. In the passive sorting model, protein sorting in the TGN is not selective and both the constitutive and regulated proteins enter immature secretory granules. As sorting proceeds, proteins of the constitutive pathway are selectively removed from immature granules via constitutive-like vesicles, whereas proteins of the regulated pathway undergo further condensation and pack-aging into mature secretory granules. The finding that only a small fraction of CPD is present in mature granules indicates that CPD may fail to undergo further condensation during granule biogenesis.
In conclusion, we have previously demonstrated that CPD is predominantly localized to the TGN, and also cycles between the TGN and the plasma membrane in AtT-20 cells (25). In the present study, we demonstrate that CPD enters immature secretory granules containing neuroendocrine peptides (Fig. 7,  ISG). We propose that CPD is progressively removed from immature granules in the process of their maturation and enters recycling endosomes (Fig. 7, RE). The soluble form of CPD (gp170) may follow a similar immature granule-mediated sorting pathway, and then is efficiently removed from the regulated pathway via constitutive-like vesicles (CLV) (Fig. 7). Alternatively, only a small fraction of gp170 may enter the regulated pathway, whereas the bulk of it follows a constitutive secretion (Fig. 7, CV).