A conserved alpha-helix at the amino terminus of prosomatostatin serves as a sorting signal for the regulated secretory pathway.

Mammalian prosomatostatin (PSST) contains the bioactive peptides SST-14 and SST-28 at the COOH-terminal end of the molecule and a putative sorting signal in the propeptide segment for targeting the precursor to the regulated secretory pathway. The NH(2)-terminal segment of PSST consists of an amphipathic alpha-helix, which has been totally conserved throughout vertebrate evolution. We have analyzed the PSST-(3--15) region for sorting function by alanine scanning and deletional mutagenesis. Mutants created were stably expressed in AtT-20 cells. Regulated secretion was studied by analyzing basal and stimulated release of SST-14 LI and by immunocytochemistry for staining of SST-14 LI in punctate granules. Deletion of the PSST-(3--15) segment blocked regulated secretion and rerouted PSST for constitutive secretion as unprocessed precursor. Alanine scanning mutagenesis identified the region Pro(5)--Gln(12) as being important in precursor targeting, with Leu(7) and Leu(11) being critical. Molecular modeling demonstrated that these two residues are located in close proximity on a hydrophobic surface of the alpha-helix. Disruption of the alpha-helix did not impair the ability of PSST to be processed at the COOH terminus to SST-14 and SST-28. Processing, however, was shifted to the early compartments of the secretory pathway rather than storage granules and was relatively inefficient.

Secretory cells such as neuroendocrine, exocrine, and mast cells contain two distinct pathways for protein secretion, a constitutive secretory pathway (CSP) 1 that transports proteins to the cell surface by bulk flow and a regulated secretory pathway (RSP) that releases secretory proteins from a granular storage pool in response to specific stimuli (1,2). Proteins destined for secretion are initially synthesized as precursors on ribosomes, translocated into the lumen of the endoplasmic re-ticulum, and transported through the Golgi stacks to the trans-Golgi network (TGN). Here the protein is sorted via clathrincoated vesicles into the RSP consisting of dense core secretory granules or the CSP through small nonclathrin-coated vesicles, which exit from the TGN and rapidly migrate to the plasma membrane (2)(3)(4)(5)(6). A major unanswered question is the mechanism for sorting prohormone and proneuropeptide precursors in the TGN into either the CSP or RSP (4 -6). Sorting is an active process that requires some form of recognition of the secretory protein (2, 4 -6). It is one step in a multistep cascade during which the prohormone is concentrated over 100-fold, packaged with other granular proteins, extruded into budding secretory vesicles, and proteolytically processed into smaller mature products. These events may be interdependent, and their temporal and spatial relationship remains poorly understood (5,6). Three models have been proposed to explain how proteins are sorted to the RSP. The first proposes that regulated secretory proteins possess an intrinsic ability to form aggregates leading to packaging of condensed products into secretory granules, thereby sorting them away from soluble proteins that are carried off by bulk flow in small vesicles. Support for this model comes from the tendency of a number of secretory granule proteins such as prolactin, growth hormone, the chromogranins, carboxypeptidase E (CPE), and prohormone convertase 2 (PC2) to aggregate at the mildly acidic pH in the TGN (7)(8)(9)(10)(11)(12). However, other proteins, such as fibronectin, that aggregate easily are not targeted into the RSP, and modifications on proteins such as chromogranin B and insulin-like growth factor-1 result in missorting without affecting aggregation (13)(14)(15). Furthermore, GH does not aggregate in the acidic environment of the TGN in COS-7 cells but does so in AtT-20 cells, and blockade of acidification with chloroquin and bafilomycin A1 is without effect on the ability of these hormones to aggregate in secretion granules in GH 4 C 1 cells, suggesting that aggregation alone is not sufficient for sorting into secretory granules (7). The second model assumes that regulated secretory proteins contain sorting signals in the form of specificsequence motifs or conformational epitopes that allow them to be sorted from constitutive secretory proteins by a receptormediated mechanism at the level of the TGN (1,2). The third model combines features of the first two mechanisms and assumes that there is initial interaction of the regulated secretory protein with a receptor, which then triggers the formation of an aggregate that is packaged into secretion granules. Several lines of evidence suggest that the propeptide is recognized by the sorting apparatus and that the structural domains that serve as recognition signals are dominant, since fusion of a constitutively secreted protein to a hormone (e.g. GH) targets the hybrid protein to the RSP and deletion of sorting signal domains results in mistargeting to the CSP (16 -22). A sorting sequence domain has been described in the prosegment of POMC, enkephalin, SST, chromogranins, and PC1 (18 -23). The most compelling arguments for a specific sequence sorting motif have come from studies of POMC and prosomatostatin (PSST) (18, 19, 21, 24 -28). In the case of POMC, structurefunction and molecular modeling studies have identified a sorting signal motif in the NH 2 -terminal segment made up of a disulfide bond constrained amphipathic hairpin loop that binds to a sorting receptor identified as membrane-associated CPE ( Fig. 1) (21,25). Molecular modeling has revealed a similar putative sorting motif in two other precursors, proenkephalin and proinsulin (25). Mutation of the binding site on CPE or in vitro antisense depletion of CPE or genetic obliteration of CPE in the CPE fat mouse leads to missorting of POMC, proenkephalin, and proinsulin (25,26,29). Not all secretory proteins, however, are recognized for sorting by CPE. For instance, chromogranin A, which possesses a RSP sorting domain similar to that in POMC, does not use CPE as a sorting receptor, suggesting the existence of other sorting receptors (26). PSST is another well characterized precursor that has been suggested to harbor a sorting signal (18,19,28). Mammalian PSST is processed post-translationally at COOH-terminal dibasic and monobasic sites to yield SST-14 and SST-28, respectively (30,31). In addition, cleavage at an unknown site at the NH 2terminal region has been implicated in generating the decapeptide PSST-(1-10) without any known biological activity (32). The PSST-(1-10) sequence is conserved throughout vertebrate evolution (33), and deletion of this region results in missorting of the mutant precursor (28) (Fig. 1). A comparison of the amino acid sequence and secondary structure of the PSST NH 2 -terminal segment with that of 14 other prohormones that have been shown experimentally to be sorted to secretory vesicles in AtT-20 cells has identified a common motif consisting of a degenerate amphipathic ␣-helix (34). This consensus sorting sequence in the case of PSST lies within residues 3-15 and differs from the disulfide bond containing hairpin loop struc-ture in POMC, proenkephalin, and proinsulin ( Fig. 1). In the present study, we have analyzed the PSST-(3-15) segment as a sorting signal by alanine scanning and deletional mutagenesis. We show that Leu 7 and Leu 11 , which form part of a contiguous hydrophobic patch on the surface of the ␣-helix, are critical for sorting function and that COOH-terminal processing of PSST to SST-14 and SST-28 can occur constitutively but is relatively inefficient in the absence of correct precursor targeting to the RSP.

EXPERIMENTAL PROCEDURES
Materials-Synthetic peptides were obtained as follows. SST-14 and SST-28 were from Bachem (Marina del Rey, CA); Tyr 0 SST-14 was from Peninsula Laboratories (Belmonte, CA); acetonitrile and trifluoroacetic acid were purchased from Fisher; heptafluorobutyric acid was obtained from Pierce; pepstatin-A, 12-O-tetradecanoylphorbol-13-acetate (TPA) and phenylmethylsulfonyl fluoride were from Sigma. Forskolin (FSK) was purchased from Calbiochem; Dulbecco's modified Eagle's medium and fetal bovine serum were purchased from Life Technologies, Inc. Ser-X-tend was obtained from Irvine Scientific (Santa Anna, CA). All other reagents were of analytical grade and were obtained from various suppliers.
Construction of Wild Type and Mutant PSST cDNAs-cDNA for wild type rat prepro-SST was constructed in the expression vector pTEJ8. Using rat prepro-SST as template, a series of mutants were created by the PCR overlap extension technique (35) (Fig. 2): (i) alanine scanning mutagenesis substituting Ala for each of the 13 residues from Ser 3 to Leu 15 in PSST; (ii) NH 2 -terminal deletion mutant deleting residues 3-15 of PSST (⌬NPSST); and (iii) insertional mutants substituting Lys 13 with KR or RTKR. To construct the mutants, two fragments were created separately, which included a 5Ј fragment containing the desired mutation using primer A and a reverse primer and a 3Ј fragment using primer B and a forward primer where the forward primer and reverse primers are mirror images of each other. The fragments were then ligated in a third PCR to generate the full-length mutant PPSST cDNA. For example, to create Ser 3 3 Ala, primers A and 2 were used to synthesize the 5Ј fragment of PPSST, and primers B and 1 were used to generate the 3Ј fragment. Primer A (5Ј-ATT CATA AGC TTG CCG CCA CCA TGC TGT CCT GCC GT-3Ј (forward)) was designed to contain HindIII endonuclease restriction site, Kozak consensus sequence, and initiation codon. Primer B (5Ј-TAG TAG ATG AAT TCC TAA CAG GAT GTG GAA TGT-3Ј (reverse)) contained 3Ј-flanking sequence and stop codon followed by an EcoRI restriction site.
PCR was carried out with 50 ng of PPSST cDNA in 100 l containing 20 mM Tris-HCl, 200 M dNTPs, 1.5 mM MgCl 2 , 6% Me 2 SO, and 2 units of Pfu (Stratagene) using the following conditions: denaturation at 94°C for 80 s, anealing at 59°C for 50 s, and extension at 72°C for 60 s for 25 cycles followed by extension at 72°C for 10 min. PCR products were separated by agarose gel electrophoresis, and the amplified bands were electroeluted and purified. Fragments B-1 and A-2 were then fused in a ligation reaction using flanking primer pair A and B. After PCR ligation, the products were digested to completion with EcoRI and HindIII, and the purified fragments were subcloned into HindIII-EcoRI multiple cloning sites of pTEJ8. All recombinant plasmid constructions were verified by sequencing of double-stranded DNA (University Core DNA Service, University of Calgary, Alberta, Canada), and at least two independent clones of each mutant were independently transfected.
Cell Culture and Transfection-AtT-20 mouse anterior pituitary cells were cultured in Dulbecco's modified Eagle's medium with 5% fetal bovine serum supplemented with Ser-X-tend in an atmosphere of 5% CO 2 and 95% air in a humidified incubator at 37°C. Cells were plated in 100 ϫ 20-mm Petri dishes and transfected at 50% confluency with 3-5 g of the appropriate plasmid construct by LipofectAMINE (Life Technologies) and stable G418 (0.861 mg/ml)-resistant nonclonally selected cells were propagated for study.
Secretion Studies-Stably transfected AtT-20 cells were cultured in 35-mm diameter six-well plastic Petri dishes and grown to 80 -90% confluency, after which they were prepared for studies of basal and stimulated secretion of immunoreactive SST-14 (SST-14 LI). Following removal of the feeding medium, groups of five wells were incubated with Dulbecco's modified Eagle's medium plus 1% bovine serum albumin containing phenylmethylsulfonyl fluoride and pepstatin-A (20 g/ml each). To study regulated secretion, cells were incubated with 20 M FSK or 10 Ϫ7 M TPA for 4 h. Media were then harvested and centrifuged at 1000 ϫ g for 5 min to remove detached cells, and the supernatant was acidified to pH 4.8 with 1 M acetic acid and stored at Ϫ20°C pending radioimmunoassay (RIA) analysis of SST-14 LI. Attached cells were extracted by scraping into 1 M acetic acid containing phenylmethylsulfonyl fluoride and pepstatin-A (20 g/ml each) on ice. The cell suspension was further extracted by sonication followed by centrifugation at 5000 ϫ g for 30 min. The supernatant was stored at Ϫ20°C for RIA and HPLC analysis.
HPLC-Pooled acidified secretion media and cell extracts were diluted 1:7 with 0.1% trifluoroacetic acid and concentrated using Waters Sep-Pak C18 cartridges. The adsorbed peptides were analyzed by HPLC on a C18 Bondapak reverse phase column using a Waters HPLC system as previously described (30,31). The column was eluted at room temperature (21°C) at 1 ml/min with 12-55% acetonitrile and a 0.2% heptafluorobutyric acid gradient over 150 min. The column effluent was monitored for UV absorbance at 214 and 280 nM. Fractions were spiked with 10 l of 10% bovine serum albumin, and stored at Ϫ20°C until further use. 30 -100-l aliquots from each fraction were rotary-evaporated with a Speedvac and assayed for SST-14 LI by RIA.
RIA of SST-14 LI-RIA for SST-14 LI was performed using a rabbit anti-SST antibody (R149), [ 125 I]Tyr 0 SST-14 radioligand, synthetic SST-14 standards, and a bovine serum albumin-coated charcoal separation method (30,31). Antibody R149 is directed against the central segment of SST-14 and detects SST-14 as well as the molecular forms extended at the amino terminus of the peptide such as SST-28 and PSST.
Immunofluorescence Microscopy-The cellular localization of SST-14 LI in AtT-20 cells expressing wild type and mutant PSST forms was characterized by fluorescence immunocytochemistry (36). Stably transfected AtT-20 cells were plated at 1.25 ϫ 10 5 cells/well in 24-well plates coated with 50 mg/ml polyornithine. On day 3 at ϳ60 -70% confluency, cells were washed twice in PBS and fixed in 2% paraformaldehyde (in 0.1% PBS) for 20 min on ice. Cells were then permeabilized with 0.2% Triton X-100 in (0.1% PBS) for 5 min at room temperature, washed three times in PBS and incubated with R149 anti-SST-14 antibody (diluted 1:1000) for 8 -12 h at 4°C. The cells were washed with PBS and incubated for 90 min at 20°C with Cy3-conjugated goat anti-rabbit secondary antibody (1:200). For staining the Golgi apparatus, cells were washed twice in PBS and incubated for 5 h with wheat germ agglutinin conjugated to fluoroscein (1:1000). Finally, cells were washed twice with PBS, mounted with immunofluor and viewed under a Zeiss LSM 410 confocal microscope. Images were obtained as single optical sections taken through the middle of cells and averaged over 32 scans/frame. Secondary Structure Prediction and Model Building-The secondary structure of rPSST (residues Ala 1 -Cys 92 ) was predicted with the NPS@ consensus secondary structure prediction algorithm (37) using 11 secondary structure prediction methods: SOPM, SOPMA, HNN, DPM, DSC, GOR-I, GOR-III, GOR-IV, PHD, PREDATOR, and SIMPA96. A structural model of the predicted ␣-helical region Pro 5 -Thr 19 was constructed from standard geometries using the BIOPOLYMER module in SYBYL 6.6 molecular modeling software (Tripos Inc., St. Louis, MO). NH 2 and COOH termini were blocked with acetyl and methylamino groups, respectively. Structural refinement was carried out by energy minimization using an AMBER 4.1 all-atom force field (38) and a distance-dependent (4R) dielectric constant.
Statistical Analysis-Results are expressed as mean Ϯ S.E. Statistical analysis was carried out by one-way analysis of variance followed by Dunnet's significance test. Significance was indicated by a p value of Ͻ 0.05.

RESULTS
Basal and Stimulated Release of WT PPSST-AtT-20 cells expressing WT PSST released total SST-14 LI at a low basal rate of 0.54 Ϯ 0.08 ng/ml/4 h representing 8.3% of total cell content (Fig. 3A). FSK stimulated SST-14 LI secretion 1.9-fold, whereas TPA demonstrated a 3.3-fold stimulation of SST-14 LI release. By immunocytochemistry, SST-14 LI displayed punctate localization in vesicular structures in both the main cell body throughout the cytoplasm as well as in cell processes (Fig.  4, A-C). These results provide both morphological and functional evidence that PSST is properly sorted to the RSP in AtT-20 cells displaying low basal secretion and positive response to secretagogue stimulation, thereby making these cells an appropriate model for studying PSST sorting to the RSP.
⌬NPSST, KR, and RTKR Substitution Mutants-To assess the sorting function of the NH 2 -terminal domain of PSST, we created a deletion mutant in which the Ser 3 to Leu 15 residues were removed. In addition, two other mutants were created, replacing the putative monobasic Lys 13 processing site with RTKR (a classic furin motif) or the dibasic motif KR to enhance NH 2 -terminal PSST cleavage endogenously by the prohormone convertase furin or PC1/PC2, respectively. AtT-20 cells stably expressing the ⌬NPSST mutant released SST-14 LI at a high basal rate (7.1 Ϯ 0.33 ng/ml/4 h) representing ϳ50% of total cell content (Fig. 3B). Release was unresponsive to FSK or TPA stimulation during a 4-h incubation (7.12 Ϯ 0.38 and 7.83 Ϯ 0.35 ng/ml SST-14 LI, respectively) (Fig. 3B). Similar results were obtained with the RTKR and KR substitution mutants, which showed even higher basal release of SST-14 LI of 72 and 81% of cell content, respectively, with no response to FSK and TPA stimulation (Fig. 3, C and D). These results were correlated with immunocytochemistry. Contrary to WT PSST expression in AtT-20 cells, SST-14 LI in cells expressing the KR substitution mutant was localized to a perinuclear area that was immunopositive for wheat germ agglutinin (WGA) and corresponded to the TGN (Fig. 4, D-F). Similar results were obtained with ⌬NPSST and RTKR mutants. Constitutive secretion, absence of secretagogue responsiveness, and lack of SST-14 LI staining in punctate granules suggest that the PSST-(1-15) domain harbors important information that is essential for sorting PSST correctly to the RSP.
Alanine Substitution Mutants-Having found that the amino-terminal 3-15 domain of PSST contains a potent sorting signal, we proceeded to map specific amino acid residues involved by alanine scanning mutagenesis. Mutants were stably expressed in AtT-20 cells and characterized for basal and regulated secretion, and the results were correlated with immunocytochemistry. Basal release of SST-14 LI from the Ser 3 , Asp 4 , Lys 13 , Ser 14 , and Leu 15 mutants was Ͻ10% of total cellular content, comparable with that of WT PSST (Fig. 5). Pro 5 , Arg 6 , Arg 8 , Gln 9 , Phe 10 , and Gln 12 mutants, however, exhibited somewhat higher levels of basal SST-14 LI release compared with wild type (ϳ15% of cell content/4 h). Substitution of the Leu 7 and Leu 11 residues with Ala resulted in a dramatic increase in basal secretion to 72 and 70% of total cell content, respectively, comparable with the amounts found with the  Table I). The Ser 3 , Asp 4 , Lys 13 , Ser 14 , and Leu 15 mutants all displayed increased release of SST-14 LI in response to both FSK and TPA. Like WT PSST, the Lys 13 , Ser 14 , and Leu 15 mutants showed a 2-fold increase in secretion in response to FSK, whereas the Ser 3 and Asp 4 mutants exhibited somewhat reduced 1.6 -1.7-fold stimulation. Both WT and the five responsive mutants displayed differentially greater sensitivity to TPA compared with FSK stimulation. Thus, TPA induced a 3-fold increase in SST-14 LI release from WT PSST and a 2-4-fold increase in the case of the Ser 3 , Asp 4 , Lys 13 , Ser 14 , and Leu 15 mutants. In contrast, FSK produced an approximate doubling of SST-14 LI release from WT and the five responsive mutants. Ala substitution of the 8 amino acid residues from Pro 5 to Gln 12 rendered all of these mutants totally unresponsive to both FSK and TPA stimulation ( Table I). The ability of the mutants to respond to secretagogues was correlated with the granular morphology of the cells. Fig. 4 depicts the subcellular distribution of immunofluorescent SST-14 LI in representative point mutants. As an example of a mutant displaying high basal secretion and loss of regulated secretion, the Leu 7 point mutant showed immunofluorescent SST-14 localized in a perinuclear area, which overlapped the distribution of wheat germ agglutinin staining (Fig. 4, G-I). Unlike WT PSST cells, SST-14 LI was not identified in the cell body of the two mutants. As an example of a point mutant that continued to display regulated secretion, the Lys 13 mutant displayed a punctate pattern of staining throughout the cytoplasm similar to WT PSST, implying proper PSST targeting to secretory granules (Fig. 4, J-L).
Effect of NH 2 -terminal PSST Mutations on COOH-terminal Processing to SST-14 and SST-28 -To characterize the prod-ucts of PSST processing, cell extracts and media from AtT-20 cell transfectants were fractionated by HPLC followed by RIA of the eluting fractions (Fig. 6). The elution positions of the peaks obtained were compared with those of synthetic SST-14 and SST-28 or of purified PSST chromatographed under identical conditions. Table II compares the percentages of SST-14, SST-28, and unprocessed PSST derived from HPLC chromatograms. Extracts of cells of WT transfectants displayed three peaks coeluting with synthetic SST-14 (retention time 67 min), SST-28 (retention time 73 min), and PSST (retention time 111 min), representing 65, 28, and 7% of total immunoreactivity, respectively (Fig. 6B). SST-14 LI released basally consisted entirely of two peaks corresponding to SST-14 (70%) and SST-28 (30%) (Fig. 6B). A similar ratio of SST-14 to SST-28 was obtained in FSK and TPA stimulated release medium (data not shown). The Leu 15 to Ala mutant displayed comparable HPLC profiles to WT PSST in both cell extracts and media. Thus, PSST was efficiently processed intracellularly to SST-14 and SST-28 (67 and 24% of SST-14 LI, respectively). SST-14 and SST-28 were also the principal immunoreactive species released into the medium; the peak corresponding to PSST released from these cells comprised 9% of the total released immunoreactivity. In contrast, mutants characterized by diversion of PSST from the RSP to the CSP (⌬NPSST, Lys 13 to KR, Lys 13 to RTKR, Leu 7 to Ala, Leu 11 to Ala) displayed a different HPLC profile of SST-14, SST-28, and unprocessed PSST (Fig.  6, C-F, Table II). In the case of the Leu 7 to Ala mutant, despite the missorting of PSST to the RSP, the precursor was efficiently cleaved intracellularly to SST-14 and SST-28 (59 and 31%, respectively); a third peak corresponding to full-length PSST accounted for 10% of the total intracellular immunoreactivity (Fig. 6E, left panel). In contrast to cell extracts, however, the HPLC profile of SST-14 LI released basally in the medium was very different, with only small amounts of processed SST-14 and SST-28 (14 and 18%, respectively); the major product released into the medium of these cells was full-length PSST, accounting for 68% of total SST-14 LI (Fig.  6E, right panel). As expected, the pattern of release after FSK or TPA stimulation was identical to that of basal release, since neither secretagogue provoked regulated release from these cells (data not shown). Similar results were observed in the case of the Leu 11 to Ala mutant and the ⌬NPSST, Lys 13 to KR, and Lys 13 to RTKR mutants, all of which displayed efficient intracellular PSST processing to SST-14 and SST-28 (ϳ62 and 25-31%, respectively), with a small 7-13% peak corresponding to unprocessed PSST. However, the major form released into the medium both basally and in response to secretagogue stimulation was unprocessed PSST, accounting for 51-73% of total released immunoreactivity. These results indicate that PSST that fails to be targeted to the RSP can still be processed to SST-14 and SST-28 in TGN compartments. However, PSST targeting is critical for efficient processing of the releasable pool of SST-14 and SST-28.
Molecular Modeling of rPSST-We constructed a structural model of the Pro 5 to Thr 19 sequence of rPSST based on the  secondary structure prediction data (Fig. 7A). This model reveals an amphipathic ␣-helix with a hydrophobic face formed by the side chains of Leu 7 , Phe 10 , Leu 11 , and Leu 15 residues and a polar face comprising the side chains of Arg 6 , Arg 8 , Gln 9 , Gln 12 , and Lys 13 residues (Fig. 7B). The side chains of Leu 7 and Leu 11 residues that are essential for high activity are located in close proximity to each other on the hydrophobic surface of the ␣-helical structure. It is noteworthy that for all of the point mutants created for this study that contain a single amino acid residue mutated to alanine, the ␣-helical structure is highly probable due to the strong propensity of alanine to adopt the ␣-helical conformation (39). The inactivity of the Leu 7 to Ala and Leu 11 to Ala mutants can be attributed to the removal of critical side chains from the hydrophobic surface, which prob-ably forms a binding interface for the propeptide, rather than to global conformational changes introduced by the mutations.

DISCUSSION
In this study, we have shown that the PSST-(3-15) segment, which comprises an amphipathic ␣-helix, acts as a sorting signal for directing PSST to the RSP and that residues Leu 7 and Leu 11 separated by one turn on the ␣-helix are critical determinants of precursor sorting. Disruption of the NH 2 -terminal ␣-helix does not impair the ability of PSST to be processed at the COOH terminus to SST-14 and SST-28. Processing, however, is shifted to early compartments of the secretory pathway instead of storage granules and is relatively inefficient.

Comparison of the percentage of SST-14, SST-28 and unprocessed PSST derived from HPLC chromatograms of cell extracts and media from AtT-20 cells expressing WT or mutant PSST
Several previous studies have shown that the prosegment of PSST harbors a sorting signal (18,19,28). For instance, an SST fusion protein consisting of the signal peptide and proregion of anglerfish PSSTI fused to ␣-globin is sorted to the RSP in transfected GH 3 cells, whereas the ␣-globin gene joined to the ␤-lactamase signal peptide is degraded in the secretory compartment (18). Anglerfish PSSTI transfected in Rin5F cells is directed to the RSP, whereas anglerfish PSSTII is mainly targeted to the CST (19). A fusion protein comprising the first 54 residues of rPSST and the last 48 amino acids of anglerfish PSSTII is correctly targeted to the RSP (19). Deletion of the rPSST-(1-10) segment results in selective blockade of the mutant precursor from sorting into a TPA-responsive (but not cAMP-responsive) secretory compartment (28). These findings suggest that NH 2 -terminal sequences of rPSST (and probably anglerfish PSSTI but not anglerfish PSSTII) contain intracellular targeting information (19). Molecular modeling of rPSST reveals an ␣-helix at residues 5-19 with the side chains of residues Leu 7 , Phe 10 , Leu 11 , and Leu 15 forming a contiguous hydrophobic patch on the helix surface (Fig. 7). This domain is highly conserved in all known vertebrate PSST molecules as well as in the SST-related precursor procortistatin (PCST) (where it is located not at the NH 2 terminus but further downstream at residues 19 -35) but is not present in anglerfish PSSTII consistent with the targeting data obtained for this precursor experimentally (Fig. 1). The secondary structure predictions of a dozen other prohormones that are known to be targeted to the RSP also reveal a common amphipathic ␣-helix similar to the one in PSST that qualifies as a putative sorting signal (34). We have analyzed the PSST NH 2 -terminal ␣-helix as a sorting signal by detailed mutagenesis. Deletion of the PSST-(3-15) segment blocked regulated secretion of SST-14 LI in response to both TPA and FSK and rerouted PSST for constitutive secretion as unprocessed precursor. Similar results were obtained with two other mutants in which the Lys 13 residue was substituted with RTKR (a classic furin motif) or the dibasic motif KR to enhance NH 2 -terminal PSST cleavage endogenously by furin or PC1/PC2, respectively. Analysis of NH 2 -terminal processing by NH 2 -terminus-specific RIA confirmed a 2-and 3-fold increase in cleavage of a PSST-(1-10)like product from these mutant precursors (data not shown). An additional possibility for the missorting is that insertion of extra basic residues at Lys 13 disrupted the ␣-helix. Thus, removal of the PSST NH 2 -terminal ␣-helix by mutagenesis or endogenously by endoproteolysis resulted in precursor missorting. The complete abrogation of TPA-and FSK-stimulatory responses by the three deletion mutants differs from the results of Sevarino et al., who found that deletion of the rPSST-(1-10) segment induced only partial loss of regulated secretory responses to TPA but not FSK (28). Since the NH 2 -terminal domain of PSST is crucial for precursor targeting to secretory granules, it is surprising to find that the precursor is normally processed at the NH 2 terminus to generate PSST-(1-10). The site of processing has been postulated to be Lys 13 , although this region does not qualify as a substrate for monobasic cleavage by a prohormone convertase (PC)-like enzyme (32,40). Recent studies suggest that NH 2 -terminal processing of PSST is effected by the novel protease subtilisin-kexin isoenzyme SKI-1, which cleaves at Leu 11 (40,41). The biological significance of PSST NH 2 -terminal processing is unclear, although it is known that PSST-(1-10) cleavage is relatively inefficient compared with that of SST-14 and SST-28 and is highly tissue-specific with moderate production of the peptide in stomach and brain and virtually none in islet cells or intestinal mucosa. Even in antral D-cells, the site of maximum PSST-(1-10) synthesis, only a small subpopulation of secretory granules (30% in rats, 10% in humans) contain the peptide (41). Our finding that deletion of the PSST- (3)(4)(5)(6)(7)(8)(9)(10)(11)(12)(13)(14)(15) domain results in missorting of the precursor suggests that endogenous NH 2 -terminal PSST processing must be a late event, distal to the TGN sorting process, and its function may be to target the precursor to a subpopulation of secretory granules. Alanine scanning mutagenesis identified the region Pro 5 -Gln 12 as being important in precursor targeting, with Leu 7 and Leu 11 being critical. These results complement the modeling data and suggest that these two residues located in close proximity on the hydrophobic surface of the ␣-helix may provide a binding interface for interaction with a putative sorting receptor. Recently, an amphipathic ␣-helix in the COOH-terminal segment of PC1 with critical leucine residues at Leu 745 and Leu 749 has also been reported to mediate targeting of the convertase to the RSP (22). Unlike PSST, the COOH-terminal region of PC1 contains two segments of ϳ40 residues, each of which can independently target the convertase to the RSP and which both harbor ␣-helices. These results provide direct evidence that an ␣-helix in PSST and PC1 mediates the targeting of the two proproteins to the RSP and suggest that an ␣-helical structure common to a number of prohormones may serve as a general sorting signal. The ␣-helix sorting signal differs from the disulfide bond constrained amphipathic hairpin loop structure shown to be a sorting signal for POMC (21,24,26). The critical elements of this motif comprise residues DLEL at the apex of the loop and Cys 8 /Cys 20 residues that form a disulfide bridge (21). Molecular modeling has revealed a similar putative disulfide bond constrained sorting motif in proenkephalin and proinsulin (26).
If there is a sorting signal, does it bind to a specific sorting receptor? Thus far, two proteins have been proposed to function as sorting signal receptors. One is the inositol 1,4,5triphosphate receptor, which binds chromogranin A (43). This receptor, however, is only weakly expressed in secretory granules of neuroendocrine cells and therefore is unlikely to function as a general sorting receptor. The second is membrane-associated CPE expressed in high concentrations in TGN and secretory granule membranes of neuroendocrine cells (24 -27). CPE binds to the POMC sorting signal motif and acts as a low affinity, high capacity sorting signal receptor (24). CPE interacts at Arg-Lys basic residues with the acidic residues in the POMC sorting signal (25). Mutation of the binding site on CPE, in vitro antisense depletion of CPE, or genetic ablation of CPE in the CPE fat mouse leads to missorting of POMC, proenkephalin, and proinsulin (25,26). Other studies, however, have found that proinsulin is sorted to the RSP in pancreatic islets from CPE-deficient fat mice as well as in cell lines derived from pancreatic ␤ cells of these mice (44,45). Additionally, chromogranin A, which possesses a POMC-like sorting signal, does not use CPE as a sorting receptor (26). Thus, not all sorting signals are recognized by CPE, suggesting the existence of other sorting receptors. Whether there is a putative receptor that interacts with the ␣-helical sorting signal that we have identified in PSST and that is common to a number of other neuroendocrine precursors remains to be determined.
Processing of prohormones at basic residues is effected by a family of subtilisin-related mammalian Ca 2ϩ -dependent serine proteinases known as PCs with seven current members: furin, PACE4, PC1, PC2, PC4, PC5A/B, and PC7 (40). Furin, PC5B, and PC7 are membrane-bound and along with PACE4 process proteins in the CSP, whereas PC1, PC2, and PC5A process neuroendocrine precursors that are targeted to secretory granules. The cellular compartment in which cleavage occurs is controversial. Proteolytic processing of several hormone precursors (e.g. proinsulin, propresophysin, and POMC) occurs largely or exclusively in secretory granules. Immunogold labeling studies have shown that proinsulin cleavage is a post-Golgi event initiated in acidic clathrincoated immature secretory vesicles and completed in mature uncoated granules (46). Several recent studies, however, have demonstrated that limited to extensive proteolytic cleavage of some hormone precursors can also occur proximally in the TGN (47)(48)(49)(50). This comes as no surprise, since the converting enzymes already exist in an active form in this compartment (furin, PC1, PC5, PC7, and PACE4), and the weakly acidic (pH ϳ6.5) milieu would favor proprotein processing (31,40). Conversion of SST-14 from PSST is mediated by either PC1 or PC2 (31). Although both convertases are present in secretory granules, PC1 also exists in an active form in the TGN and is therefore capable of SST-14 conversion in this compartment, whereas PC2 is optimally active in secretory granules (22,40). Monobasic cleavage of SST-28 is effected by furin and/or PACE4 (51,52). Blockade of PSST targeting to secretory granules by the NH 2 -terminal deletion and Leu 7 and Leu 11 PSST mutations led to an escape of large quantities of unprocessed PSST through the CSP. The remainder of the precursor, however, was retained in the TGN, where it underwent relatively efficient processing to both SST-14 and SST-28, presumably through the action of PC1 (for SST-14) and furin/PACE4 (for SST-28). These results are consistent with previous studies that have shown significant processing of PSST in the absence of secretory granules in TGN compartments (48,50). Overall, this means that the NH 2 -terminal PSST conformation does not influence enzyme recognition and PSST cleavage at the COOH terminus. The main consequence of the blockade of PSST entry into secretory granules is incomplete precursor processing and retention of the cleaved mature products in Golgi vesicles. Targeting of PSST to secretory granules, therefore, subserves two purposes: to optimize processing and to package and store the mature products for regulated release.