* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Download The unique proline-rich domain of parotid proline
Survey
Document related concepts
Hedgehog signaling pathway wikipedia , lookup
Cell growth wikipedia , lookup
Magnesium transporter wikipedia , lookup
Protein moonlighting wikipedia , lookup
Cytokinesis wikipedia , lookup
Organ-on-a-chip wikipedia , lookup
Cell encapsulation wikipedia , lookup
Cell culture wikipedia , lookup
Extracellular matrix wikipedia , lookup
Endomembrane system wikipedia , lookup
Cellular differentiation wikipedia , lookup
Signal transduction wikipedia , lookup
Type three secretion system wikipedia , lookup
Proteolysis wikipedia , lookup
Transcript
Journal of Cell Science 109, 1637-1645 (1996) Printed in Great Britain © The Company of Biologists Limited 1996 JCS1183 1637 The unique proline-rich domain of parotid proline-rich proteins functions in secretory sorting Lisa E. Stahl, Rhonda L. Wright, J. David Castle and Anna M. Castle* Department of Cell Biology and the Molecular Biology Institute, University of Virginia Health Sciences Center, Charlottesville, VA 22908, USA *Author for correspondence (e-mail: [email protected]) SUMMARY When expressed in pituitary AtT-20 cells, parotid prolinerich proteins enter the regulated pathway. Because the short N-terminal domain of a basic proline-rich protein is necessary for efficient export from the ER, it has not been possible to evaluate the role of this polypeptide segment as a sorting signal for regulated secretion. We now show that addition of the six-amino acid propeptide of proparathyroid hormone to the proline-rich protein, and especially to a deletion mutant lacking the N-terminal domain, dramatically accelerates intracellular transport of these polypeptides. Under these conditions the chimeric deletion mutant is stored as effectively as the full-length protein in dense core granules. The propeptide does not function as a sorting signal in AtT-20 cells as it does not reroute a constitutively secreted reporter protein to the regulated pathway. During transit, the propeptide is cleaved from the chimeric polypeptides such that the original structures of the full-length and the deletion mutant proline-rich proteins are reestablished. We have also found that the percentage stimulated secretion of the proline-rich proteins increases incrementally (almost twofold) as their level of expression is elevated. The increase reflects an enrichment of these polypeptides in the granule pool and its incremental nature suggests that sorting of proline-rich proteins involves an aggregation-based process. Because we can now rule out contributions to sorting by both N- and C-terminal segments of the proline-rich protein, we deduce that the unique proline-rich domain is responsible for storage. Thus at least some of the determinants of sorting for regulated secretion are protein-specific rather than universal. INTRODUCTION cells and types of secretory products (Colomer et al., 1994). Thus the sorting machinery may be more diverse than originally anticipated. Several efforts have been made to identify structural determinants within secretory proteins that may function in sorting. Among endocrine and exocrine proteins, structural regions have been identified that are either necessary or sufficient for targeting to granules (Stoller and Shields, 1989; Chu et al., 1990; Cool et al., 1995; Arrandale and Dannies, 1994; Castle et al., 1992). Many, but not all, of these regions are located at or near the N termini of the polypeptides. No bona fide consensus structural signal has been identified, although common structural motifs involved in sorting have been postulated (Kizer and Tropsha, 1991; Gorr and Darling, 1995). However, there is limited insight regarding how these domains function in sorting. We have been examining the mechanisms of sorting of salivary proline-rich proteins (PRPs). PRPs are elongate proteins lacking disulfide bonds and they have an unusually simple structure consisting of three domains: a short Nterminal segment, a central proline-rich domain of repeating cassettes, and a short C-terminal segment. They are efficiently targeted to secretion granules in parotid acinar cells (Arvan and In endocrine and exocrine secretory cells, storage granules are formed from a subset of the proteins that are transported through the secretory pathway. Molecular sorting is required to determine which proteins are retained in these granules for stimulus-dependent exocytosis. Several studies have implicated the function of specific cellular machinery and interactions among secretory proteins, especially selective aggregation (Chanat and Huttner, 1991; Kuliawat and Arvan, 1994; Colomer et al., 1994), in these sorting events (reviewed by Arvan and Castle, 1992; Bauerfeind and Huttner, 1993). Transfection of cultured cells with cDNAs encoding secretory proteins from different cell types has shown that several exogenous proteins are sorted as efficiently as the endogenous secretory products (e.g. Moore and Kelly, 1985; Burgess et al., 1985; Moore et al., 1983; Dickerson et al., 1987), suggesting that targeting for regulated secretion might involve a common mechanism. However, other proteins are not sorted as efficiently as endogenous products (e.g. Fennewald et al., 1988; Castle et al., 1992; Schmidt and Moore; 1994; Reaves et al., 1990; Stoller and Shields, 1989), and recent work has suggested that sorting processes may differ among types of Key words: Proline-rich protein, Protein sorting, Regulated secretion, AtT-20 cell 1638 L. E. Stahl and others Castle, 1986; Blair et al., 1991), and when expressed from cloned cDNAs in a pituitary cell line, AtT-20, they are stored in modest amounts in the dense core granules (Castle et al., 1992; Castle and Castle, 1993). Deletion of the N-terminal thirteen amino acid segment (transition region) from a basic PRP eliminated storage of the protein but also resulted in its inefficient transport from the ER. Because the ability to detect regulated secretion of biosynthetically labeled proteins is compromised by slow ER export, we were not able to determine whether the transition region was truly needed for storage in granules or whether the deletion mutant (PRP∆T) would be stored if its ER export was efficient. Consequently, we have sought ways to restore more rapid ER export of PRP∆T so as to allow examination of its storage in dense core granules. We now report that rapid export of PRP∆T can be achieved by attaching the hexapeptide prosequence of parathyroid hormone (PROPTH) to its N terminus. We selected PROPTH for this approach for several reasons. First, PROPTH is normally removed rapidly and efficiently from PTH during intracellular transport (Habener et al., 1979; Hellerman et al., 1984) and we hoped that similar processing might occur from the chimera, thereby generating PRP∆T. Second, PROPTH is thought to function in translocation into the ER but does not affect postER transport through the secretory pathway in GH4 cells (Wiren et al., 1988). Third, PROPTH is short and its strongly basic and hydrophilic character differs from the amphipathic nature of hydrophobic helical motifs that have been postulated to function in post-Golgi sorting for regulated secretion (Kizer and Tropsha, 1991; Gorr and Darling, 1995). Finally, the PTH signal and propeptide have been used in chimeras previously and have enabled efficient transport of recombinant fibronectin polypeptides (Schwarzbauer et al., 1987). We now show that upon restoring efficient ER export to PRP∆T, PROPTH is indeed cleaved and PRP∆T is stored in secretory granules. Moreover, the fraction of PRP∆T present in a stimulusreleasable (granule) pool approximately doubles with increased expression. Thus we have clarified that the native Nterminal domain of PRP is not necessary for granule storage provided that ER export is efficient. Rather the storage of PRP is probably mediated by self-interactions of its unique prolinerich domain. MATERIALS AND METHODS Antibodies Anti-ACTH (adrenocorticotropic hormone) antiserum and the polyclonal antibody against PRPs were characterized previously (Blair et al., 1991; Castle and Castle, 1993). The polyclonal antibody against the vesicular stomatitis protein G (Indiana strain) was a gift from Dr Robert Wagner (University of Virginia). Construction of vectors PROPTH-PRP∆T and PROPTH-PRP The signal and pro sequences of preproPTH were synthesized by the polymerase chain reaction (PCR) with a HindIII restriction site included at the 5′ end and an EcoRI site at the 3′ end. The prepro fragment was cloned into HindIII and EcoRI sites of pBluescript (Stratagene, Inc., CA). Fragments encoding PRP and PRP∆T lacking the signal sequence were synthesized by PCR with an EcoRI site at the 5′ end and cloned into pBluescript containing the prepro sequence of PTH. The extra bases resulting from inclusion of the EcoRI site were deleted by oligonucleotide-directed mutagenesis according to the instructions from Bio-Rad (Muta-Gene kit, Bio-Rad Laboratories, Richmond, CA). Mutated clones were screened by limited DNA sequencing, and the selected constructs were sequenced in entirety to confirm their structure. For expression in eukaryotic cells, the cDNA fusion constructs were excised from pBluescript and cloned into the HindIII/BamHI site of the pRhR1100 expression vector (provided by Dr Tim Reudelhuber, Montreal Cancer Research Institute, Quebec) which is driven by the Rous sarcoma virus long terminal repeat (Chu et al., 1990). HISPTH-PRP∆T HISPTH-PRP was obtained by site-directed mutagenesis of a restriction fragment containing the signal and propeptide sequences of preproPTH and 93 base pairs of PRP∆T. The sequence of the entire mutagenized fragment was confirmed by sequencing. The mutagenized fragment was then attached to the remainder of PRP∆T using standard cloning techniques. tG and PROPTH-tG The cDNA for tG was a gift from Dr J. Rose and was cloned into the pLEN expression vector (Neufeld et al., 1987). The signal and propeptide sequences of preproPTH with added 5′ SalI and 3′ HindIII sites were synthesized by PCR, inserted into pBluescript, and the primary structure was confirmed by DNA sequencing. tG cDNA lacking nucleotides encoding a signal sequence was also synthesized by PCR, and it was inserted into the plasmid containing the signal and propeptide sequence of preproPTH. The amino acid sequence across the junction of PTH propeptide and tG was KRSKLKTS where the underlined amino acids result from HindIII. The final construct was cloned into the pLEN expression vector. Cell culture and transfection Mouse pituitary AtT-20 D16v cells were cultured as described previously (Castle et al., 1992). DNA transfections were carried out by calcium phosphate precipitation. Selection of stable transfectants and screening for expression of polypeptides by western blotting were also performed as described (Castle et al., 1992). Metabolic labeling and immunoprecipitation Cells expressing PRP, PROPTH-PRP, PROPTH-PRP∆T or HISPTHPRP∆T were plated in 24-well plates with 1×105 cells/well. After 4872 hours in culture, cells were labeled for 15 hours with 0.4 mCi/ml [3H]proline (Amersham Corp.) and chased as specified in individual experiments. The cells and medium containing secreted proteins were harvested and radiolabeled PRP-related polypeptides or ACTH were immunoprecipitated and quantitated by SDS-PAGE and scintillation counting as previously described (Castle et al., 1992). Level of expression was measured following a 15 hour labeling period with [3H]proline. Radiolabeled PRP-related polypeptides in the medium and cell lysates were quantitated and the level of expression was calculated as the total amount of radiolabeled PRP-related polypeptide synthesized during the labeling (sum of radiolabeled polypeptide from the medium and cell lysate). The data were corrected for the number of prolines in each molecule and total trichloroacetic acid-precipitable counts in each sample. The level of expression of ACTH-related peptides was estimated from the same experiment as the sum of radiolabeled proopiomelanocortin, ACTH biosynthetic intermediate and ACTH. The data were corrected for the number of prolines in each molecule and total trichloroacetic acidprecipitable counts in each sample. To measure the percentage stimulated secretion, duplicate wells of cells were labeled for 15 hours, chased first for 6 hours in the absence of secretagogue and then chased for an additional 3 hours in the absence (basal secretion) or the presence (stimulated secretion) of 5 mM 8-Br-cAMP. Radiolabeled PRP- or ACTH-related polypeptides from both chase media and cell lysates were quantitated and the per- Sorting of proline-rich proteins 1639 centage stimulated secretion was calculated as the difference between stimulated and basal secretion of radiolabeled polypeptides, expressed as the percentage of total (chases + cell extract). Cells expressing tG and PROPTH-tG were plated in 35 mm dishes at a density of 3×105 cells per dish. After 48-72 hours, the cells were labeled for 1 hour with 0.25 mCi/ml Expre35S35S-label (ICN Radiochemicals, Irvine, CA) in minimal essential medium (Gibco BRL, Gaithersburg, MD) supplemented with 15% dialyzed Nuserum, 20 mM Hepes, 4 mM glutamine or for 15 hours with 0.15 mCi/ml Expre35S35S-label in the same medium supplemented with 2.5 µg/ml of methionine and 4 µg/ml of cystine. Chases were carried out in the same medium containing excess cold methionine and cystine plus Trasylol as proteinase inhibitor (50 kallikrein units/ml). Determination of ER exit rates To estimate ER exit rates, cells were labeled with [3H]proline for 15 hours and chased for different lengths of time. At each timepoint, medium and cells were harvested, immunoprecipitated with anti-PRP antibody and digested with Endoglycosidase H (Endo H) and analyzed as described previously (Castle et al., 1992). Endo Hsensitive and Endo H-resistant bands were quantitated by densitometric scanning. Half-times of exit from the ER were obtained from log-linear plots of the fraction of each polypeptide that was Endo Hsensitive versus time. Quantitation of stimulated secretion using western blotting AtT-20 cells expressing PROPTH-PRP and PROPTH-PRP∆T were plated at equal cell densities in duplicate dishes. After 2-3 days, cell monolayers were washed thoroughly and incubated in serum-free medium with or without 5 mM 8-Br-cAMP for 3 hours. Cells were lysed and assayed for protein (BCA protein assay; Pierce, Rockford, IL) and DNA content (West et al., 1985). Aliquots of the medium were analyzed by SDS-PAGE followed by western blotting with the anti-PRP antibody. Bound antibody was detected with 125I-conjugated secondary antibody and the amount quantitated by phosphoimager analysis using ImageQuant software (Molecular Dynamics). The values were normalized to DNA content, and the net stimulated secretion was calculated as the amount of PRP secreted in the presence minus the amount secreted in the absence of 8-Br-cAMP. Automated microsequencing of PROPTH-PRP∆T AtT-20 cells expressing PROPTH-PRP∆T were biosynthetically labeled for 20 hours with 1 mCi/ml [3H]proline and labeled PROPTHPRP∆T polypeptides were immunoprecipitated. The immunoprecipitates were heated for 15 minutes at 55°C in SDS sample buffer (Laemmli, 1970) containing 5 mM dithiothreitol and then treated with 10 mM iodoacetamide for 2 hours at 20°C in the dark. Samples were resolved on 12.5% tube gels which were sliced into 1 mm slices and eluted with 5 mM ammonium acetate. Bovine serum albumin (0.5 nM) was added as a carrier and the eluted material was exchanged into 0.01% SDS and less than 0.38 mM Tris using a Centricon microconcentrator (Amicon, Beverly MA). This preparation was loaded onto a sequencer (Applied Biosystems 470-A,Foster City, CA) and sequenced by Edman degradation at the University of Virginia Biomolecular Facility. RESULTS As discussed in the Introduction, we have added PROPTH to PRP∆T in an attempt to overcome its very slow rate of ER export. Initially, two chimeric cDNAs were constructed, PROPTH-PRP∆T and PROPTH-PRP (Fig. 1). PROPTH-PRP∆T, encodes the PTH signal and propeptide (Hendy et al., 1981) attached to amino acid residue 17 of PRP∆T. Amino acids 21- 33 of PRP are deleted in the PRP∆T portion of the construct (Castle et al., 1992). PROPTH-PRP, encodes the same PTH segment attached to the N terminus (residue 17) of full-length PRP. Full-length PRP containing its native signal sequence and transition domain was included in some of the studies (Castle et al., 1992). PROPTH-PRP∆T and PROPTH-PRP cDNAs were transfected into AtT-20 cells and stable cell lines were created. We selected cell lines with different levels of expression of PROPTH-PRP∆T, PROPTH-PRP and PRP. The level of expression of all PRP-related polypeptides varied over a 100fold range (Tables 1, 2 and 3) but in all cases was less than 20% of the level of endogenous ACTH-related peptides. Notably, the level of expression of ACTH-related peptides was nearly the same in all the clones examined (Table 1, legend). Acceleration of ER export by PROPTH Rates of export from the ER were estimated by following the kinetics of the loss of sensitivity to Endo H digestion (Castle et al., 1992). As shown by the values of t1/2ER in Table 1, the presence of PROPTH causes a dramatic acceleration in transport of PRP∆T. PROPTH-PRP∆T exited the ER with a t1/2ER of 1.31.4 hours in contrast to >7 hours for PRP∆T (Table 1). Similarly, PROPTH-PRP exhibited a t1/2ER of 1.1 hours. The independence of the export rate on level of expression for both PROPTH-PRP∆T and PROPTH-PRP also contrasts with the results obtained for PRP, for which t1/2ER was ~6 hours for a Fig. 1. Schematic structure of PRP and tG constructs. Full-length basic PRP consists of the signal sequence (PREPRP; wavy line), the transition domain (TPRP; filled in rectangle), the proline-rich domain (PPRP; elliptical shapes) and the C-terminal domain, (CPRP; open rectangle) (Castle et al., 1992). PROPTH-PRP is a chimera of the signal (PREPTH; wavy line) and propeptide of PTH (PROPTH; rectangle) with full-length PRP. PROPTH-PRP∆T consists of the PTH signal and propeptide fused to the N terminus of PRP∆T, which lacks amino acids 21-33 (Castle et al., 1992). The sequence of PROPTH is KSVKKR. Note that amino acids 17-20 of the transition domain (small filled-in rectangle) are present within this construct. HISPTH-PRP∆T has the identical structure as PROPTH-PRP∆T except that the sequence of HISPTH propeptide is HSVLHR. PROPTH-tG is a chimera of the signal and propeptide of PTH with the truncated G protein of vesicular stomatitis virus, tG (Moore and Kelly, 1985). 1640 L. E. Stahl and others Table 1. Kinetics of exit PRP-related polypeptides from the ER Construct PROPTH-PRP∆T PROPTH-PRP∆T PRP∆T PRP∆T PROPTH-PRP PROPTH-PRP PRP PRP PRP PRP Clone LOEPRP 20 43 26 1 39 4 A2 1A1 B9 B35 3.3 66.0 2.9 12.0 1.0 60.0 2.9 21.5 59.4 97.5 Table 2. 8-Br-cAMP-dependent secretion of PRP-related polypeptides t1/2ER (hours) 1.3 1.4 7.0 9.7 1.1 1.1 6.0* 2.5* 1.8 1.8 LOEPRP represents level of expression of PRP-related polypeptides corrected to total trichloroacetic acid-precipitable counts. It was determined as described in Materials and Methods. The values shown were normalized to the lowest expressing clone, PROPTH-PRP 39. The level of expression of ACTH-related polypeptides determined from the same experiments had an average value of 506 ± 20 (s.e.m.). t1/2ER is the halftime of exit of PRP-related polypeptides from the ER. It was determined as described in Materials and Methods. *Values of t1/2ER were taken from Table 3 of Castle et al. (1992). low expressing clone and increased to a plateau of 1.8 hours for progressively higher expressing clones (Table 1 above; Castle et al., 1992). Thus the presence of the propeptide bypasses the concentration-dependent step during transport from the ER, which is normally mediated by the transition domain of PRP. Secretion and storage of PROPTH-PRP∆T and PROPTH-PRP As the PROPTH chimeras were transported from the ER with rapid and essentially identical kinetics, it was now possible to evaluate whether the presence of the transition region of PRP was truly necessary for targeting to storage granules. We used stimulation of secretion of radiolabeled cells by a secretagogue to assess whether the chimeric polypeptides were routed to the regulated pathway. Secretion of both chimeric polypeptides was stimulated with 8-Br-cAMP, indicative of their presence in secretory granules. The percentage of total radiolabeled PROPTH-PRP∆T and PROPTH-PRP undergoing stimulated secretion is shown in Table 2 and will be referred to as the percentage stimulated secretion throughout this work. The results show that the percentage stimulated secretion of PROPTHPRP∆T and PROPTH-PRP is the same (10-20%) for comparable levels of expression. Therefore, in this context, the transition region appears unnecessary. For both chimeric polypeptides, we also observed that the percentage stimulated secretion exhibited a single-step increase with increased level of expression. For several different clones expressing low levels of PROPTH-PRP∆T or PROPTH-PRP (representative examples are clones 20 and 39, respectively, in Table 2) the percentage stimulated secretion was ~10%. At higher levels of expression (e.g. clones 10 and 8 in Table 2), the percentage stimulated secretion increased to ~20% but did not rise significantly beyond this level with yet higher levels of expression (e.g. clones 43 and 4 in Table 2). To confirm that the differences in stimulus-dependent secretion of the chimeric polypeptides do not reflect variations in intrinsic storage capacity or secretory response, we measured 8-Br-cAMP-stimulated secretion of radiolabeled Construct PROPTH-PRP∆T PROPTH-PRP∆T PROPTH-PRP∆T PROPTH-PRP PROPTH-PRP PROPTH-PRP PRP PRP PRP PRP Clone LOEPRP 20 10 43 39 8 4 A2 1A1 B9 B35 3.3 13.2 66.0 1.0 10.0 60.0 2.9 21.5 59.4 97.5 % Stim. sec. 10.3±1.2 18.3±1.3 21.0±2.1 9.4±1.1 24.6±1.6 18.2±1.1 6.1±0.5* 5.2±0.8* 11.0±2 † 9.0 ‡ LOEPRP represents level of expression of PRP-related polypeptides normalized to total trichloroacetic acid precipitable cpm. It was determined as described in Materials and Methods. % Stim. sec. represents the difference between stimulated (+ 8Br-cAMP) and basal (−8-Br-cAMP) secretion of radiolabeled PRP-related polypeptides expressed as percentage of total (chases + cell extract). Detailed description of the experiment is presented in Materials and Methods. Data are presented as the mean of four experiments ± s.e.m. *Data taken from Castle et al. (1992). †Mean of two experiments ± deviation from the mean. ‡Result of a single experiment. ACTH-related peptides in untransfected AtT-20 cells and selected clones expressing PROPTH-PRP∆T. As shown in Table 3, the percentage stimulated secretion of ACTH-related peptides was essentially the same in all cell populations. The studies of stimulated secretion suggested that low and high expressing clones contain different fractions of PROPTHPRP∆T and PROPTH-PRP in the stimulus-releasable (granule) pool. These differences should be reflected in the fraction of PROPTH-PRP∆T and PROPTH-PRP in the post-Golgi (Endo Hresistant) pool. We compared the percentage of cell-associated Endo H-resistant PROPTH-PRP∆T and PROPTH-PRP at different times of chase following a long-term labeling for high and low expressing clones (Fig. 2). For each clone this value decreases rapidly during the first hour probably reflecting release of PROPTH-PRP∆T and PROPTH-PRP via the constitutive pathway and reaches a plateau with longer times of chase. At all timepoints, the percentage of cell-associated Endo Hresistant PROPTH-PRP∆T and PROPTH-PRP is higher (by ~10% on average) in a high expressing clone than in a low expressing clone. Our results using radiolabeled polypeptides imply that the level of expression has a significant influence on the distribution of PROPTH-PRP and PROPTH-PRP∆T among the major Table 3. 8-Br-cAMP-dependent secretion of ACTH Construct Clone LOE PROPTH-PRP∆T % Stim. sec. ACTH PROPTH-PRP∆T PROPTH-PRP∆T 20 43 AtT-20 3.3 66.0 NA 49 47 49 LOE PROPTH-PRP∆T refers to the level of expression of PROPTH-PRP∆T in each clone. % Stim. sec. of ACTH was calculated as described for PRP-related polypeptides (Materials and Methods). NA, not applicable. Sorting of proline-rich proteins 1641 Fig. 2. Cell-associated Endo H-resistant PROPTHPRP∆T and PROPTHPRP in low- and high-expressing clones. Low expressing clones, PROPTH-PRP∆T 20 (d) and PROPTH-PRP 39 (.), and high expressing clones, PROPTH-PRP∆T 43 (j) and PROPTH-PRP 4 (m) were labeled with [3H]proline for 15 hours to approach steady state and chased for varying lengths of time up to four hours. At each timepoint, radiolabeled secreted and cell-associated polypeptides were immunoprecipitated and subjected to Endo H digestion followed by SDSPAGE and fluorography. Radiolabeled PROPTH-PRP∆T and PROPTH-PRP were quantitated by densitometry, and Endo H-resistant, cell-associated PROPTH-PRP∆T and PROPTH-PRP were expressed as the percentage of total (secreted + cell-associated) radiolabeled protein. intracellular pools, the granules and the ER. Because steadystate labeling enriches for the slowest turnover pools of protein, to a first approximation, the granule and ER pools may be represented by Endo H-resistant and Endo H-sensitive PRPs, respectively. Using the clones expressing PROPTHPRP∆T as examples, at steady state the fraction of Endo Hresistant PROPTH-PRP∆T is 0.62 in clone 43 (high level of expression) and only 0.40 in clone 20 (low level of expression) (Fig. 2). Based on these fractions and the difference in level of expression, it is possible to predict that the granule pool in clone 43 should be 31-fold higher (0.62/0.4×20) than in clone 20 while the ER pool (calculated on basis of Endo H-sensitive PROPTH-PRP∆T should be only 13-fold higher (0.38/0.6×20) in clone 43 than in clone 20. We have confirmed the predicted increase in the granule pool experimentally by comparing the stimulated discharge of PROPTH-PRP∆T by clones 43 and 20 using quantitative western blotting (see Table 4). The results indicate a 33-fold higher stimulated secretion of PROPTHPRP∆T by clone 43. Applying the same analysis to PROPTHPRP clones 39 and 4, the predicted fold difference in stimulated secretion of 88 (0.53/0.36×60) is in good agreement with the experimental value of 83 (see Table 4). Secretion and storage of full-length PRP Our previous study showed that the percentage stimulated secretion of full-length PRP did not increase when levels of expression of PRP varied over the range corresponding to 0.64% of ACTH (Table 2, this work; Castle et al., 1992). In view of the increased percentage stimulated secretion of the PROPTH chimeras observed at higher levels of expression, we have now measured 8-Br-cAMP-stimulated secretion of full-length PRP in clones expressing it at yet higher levels. Interestingly, an approximately twofold increase in the percentage stimulated secretion was observed when the level of expression of PRP was increased (Table 2). Yet, as in the case of PROPTH-PRP∆T and PROPTH-PRP, no further significant rise in the percentage stimulated secretion occurred at still higher levels of expression (Table 2). Thus the secretion of full-length PRP shows the same trend as that of the chimeric polypeptides. But overall, the percentage stimulated secretion is lower at all levels of expression. The role of the PTH propeptide The presence of the chimeric polypeptides in regulated secretion has lead us to examine in more detail the potential role(s) of PROPTH in sorting of PRPs. We considered two possibilities. First, we examined whether storage might relate to cleavage of the propeptide, reflecting an interaction with the processing enzyme (Journet et al., 1993). Second, we examined whether the propeptide might serve as a sorting signal. Proteolytic processing of the PTH propeptide The kinetics of processing of native proPTH to PTH are very rapid and are consistent with proteolytic cleavage occurring within or just after exiting the Golgi complex (Habener et al., 1979; Hellerman et al., 1984). If processing of PROPTH is occurring in the chimeric polypeptides, then the propeptide should be absent in secreted polypeptides. For cell-associated polypeptides, the propeptide should be present in Endo Hsensitive forms (ER forms) and absent in Endo H-resistant forms (Golgi/post-Golgi forms). Using cells expressing PROPTH-PRP∆T and PRP∆T, we compared the electrophoretic mobilities of secreted and cell-associated [3H]proline-labeled polypeptides. Fig. 3A shows that secreted PROPTH-PRP∆T (PRO∆T in Fig. 3A) and PRP∆T (∆T in Fig. 3A) have identical apparent Mr values suggesting that the propeptide has been removed from PROPTH-PRP∆T. The cell-associated Endo H sensitive PROPTH-PRP∆T has higher apparent Mr than the Endo H sensitive PRP∆T, indicating that the propeptide is still attached. The cell-associated Endo H resistant PROPTHPRP∆T has an identical Mr as the secreted PRP∆T indicating the absence of the propeptide in the post-Golgi pool of PROPTH-PRP∆T. No band of higher Mr was ever observed in samples of secreted or cell-associated Endo H-resistant PROPTH-PRP∆T, suggesting that the propeptide is cleaved off from the majority of transported PROPTH-PRP∆T, whether or not it is stored in the granules. This is illustrated by the identical mobility of PROPTH-PRP∆T released in the absence and presence of 8-Br-cAMP (Fig. 3B). Together, these results imply that the polypeptide stored in the granules of cells expressing PROPTH-PRP∆T corresponds to PRP∆T, the deletion mutant that was judged to be defective in storage and transport previously (Castle et al., 1992). Analysis of secreted and cell-associated PROPTH-PRP yielded similar results (not shown). To confirm that the propeptide was cleaved at the predicted processing site (the carboxyl side of arginine-1 of the hexapeptide prosequence; Fig. 1), we performed limited N-terminal sequencing of secreted radiolabeled PROPTH-PRP∆T. Using [3H]proline for biosynthetic labeling, radioactivity was antici- 1642 L. E. Stahl and others Table 4. Western blotting of stimulated secretion of PROPTH-PRP∆T and PROPTH-PRP PROPTH-PRP∆T 43 Net stimulated secretion (AU) Stimulated secretion ratio Mean simulated secretion ratio 20 130,000 3,600 36 33±6 PROPTH-PRP 4 39 100,000 1,370 73 83±9 Net stimulated secretion represents the difference in secretion of PROPTHPRP∆T or PROPTH-PRP in the presence and absence of 5 mM 8-Br-cAMP. The data are from a sample experiment performed as described in Materials and Methods. The values presented in arbitrary units (AU) are taken from phosphoimager analysis and normalized for DNA content. Stimulated secretion ratio is: (net stimulated secretion PROPTH-PRP∆T 43)/(net stimulated secretion PROPTH-PRP∆T 20) and (net stimulated secretion PROPTH-PRP 4)/net stimulated secretion PROPTH-PRP 39). This is the fold difference in stimulated secretion between high and low expressing clones. Mean stimulated secretion ratio is an average for three separate experiments ± s.e.m. Fig. 3. Processing of PTH propeptide from PROPTHPRP∆T chimera. (A) Comparison of secreted (SEC) and cell-associated polypeptides (CELL) from cells expressing PROPTH-PRP∆T (PRO∆T) or PRP∆T (∆T) following a 15 hour labeling with [3H]proline. Immunoprecipitated polypeptides were digested with Endo H and immunoprecipitates were analyzed by 10% SDS-PAGE and fluorography. Positions of the Endo H resistant (r) and Endo H sensitive (s) forms are indicated. (B) Comparison of PROPTH-PRP∆T polypeptides secreted in the absence or presence of a secretagogue. Two wells of cells expressing PROPTH-PRP∆T were labeled for 15 hours with [3H]proline and chased for 6 hours in the absence of stimulation (not shown). Subsequently the medium was replaced with medium with (+) or without (−) 5 mM 8-Br-cAMP. Radiolabeled PROPTH-PRP∆T polypeptides were immunoprecipitated and analyzed by 10% SDS-PAGE and fluorography. (C) N-terminal sequencing of radiolabeled secreted PROPTH-PRP∆T. Cells expressing PROPTH-PRP∆T were labeled for 20 hours with [3H]proline. Secreted PROPTH-PRP∆T was prepared for N-terminal sequencing as described in Materials and Methods. Six cycles of Edman degradation were performed and the radioactivity recovered in each cycle is shown. pated in the 5th and 6th residues from the N terminus (Fig. 1; Castle et al., 1992). As seen in Fig. 3C, these expectations were confirmed, indicating correct cleavage of the PTH propeptide. Mutagenesis of the processing site in PROPTHPRP∆T Having established correct cleavage of PROPTH during intracellular transport, we sought to test the effect of preventing the cleavage of the chimeras by mutagenesis of the processing site. The PTH propeptide of chimeric PROPTH-PRP∆T and PROPTH-PRP may be cleaved by PC3 or the more ubiquitous furin (Smeekens et al., 1991; reviewed by Steiner et al., 1992; Seidah et al., 1993). In order to alter amino acid residues commonly recognized by these enzymes and at the same time retain the same positive charge of the PTH propeptide, the lysines at positions -2 and -6 were changed to histidines (Fig. 1) by in vitro mutagenesis of PROPTH-PRP∆T. The resulting construct (HISPTH-PRP∆T) was transfected into AtT-20 cells and stable lines with different levels of expression were selected. To demonstrate that the propeptide is not cleaved, we examined the fate of HISPTH-PRP∆T after labeling with [3H]histidine. As shown in Fig. 4, both secreted and cell-associated HISPTH-PRP∆T were labeled with [3H]histidine. Since PRP∆T contains no histidine residues, this indicates that the propeptide was not removed. Further, the ratio of [3H]histidine:[3H]proline labeling for secreted HISPTH-PRP∆T is the same as that for cell-associated HISPTH-PRP∆T (Fig. 4), consistent with no cleavage. Using cells expressing HISPTH-PROPTH, we tested whether secretion of HISPTH-PROPTH was stimulated with 8-Br-cAMP. Results presented in Table 5 show that the secretion of HISPTHPRP∆T was stimulated with 8-Br-cAMP; however, the percentage stimulated secretion was slightly lower than observed for PROPTHPRP∆T for all levels of expression examined (Table 2). As in the case of PROPTH-PRP∆T, PROPTH-PRP and PRP, there was an approximately twofold jump in the percentage of stimulated secretion of HISPTH-PROPTH as the level of expression was increased. Thus removal of the propeptide is not essential for storage in granules. Effect of PROPTH on the storage of truncated VSVG To address the second possibility that PROPTH carries independent sorting information, we fused it to the truncated G protein of vesicular stomatitis virus (tG), which lacks its Nterminal cytoplasmic and transmembrane segment (Fig. 1). By itself, tG is secreted constitutively in AtT-20 cells (Moore and Kelly, 1985), but it is rerouted to the regulated pathway when fused to growth hormone (Moore and Kelly, 1986). Both tG and the PROPTH-tG chimera (Fig. 1) were transfected into AtTTable 5. 8-Br-cAMP-dependent secretion of HISPTHPRP∆T Construct HISPTH-PRP∆T HISPTH-PRP∆T HISPTH-PRP∆T Clone LOE HISPTH-PRP∆T % Stim. sec. 16 15 10 5.4 21.5 86.0 7±1 14±2 11±2 LOE HISPTH-PRP∆T and % Stim. sec. were determined as described in Table 2. The mean ± s.e.m. of three experiments are presented. Sorting of proline-rich proteins 1643 Fig. 4. Labeling of HISPTH-PRP∆T with [3H]histidine. Cells expressing HISPTH-PRP∆T were labeled for 15 hours with 0.2 mCi/ml [3H]proline (pro) or 0.4 mCi/ml of [3H]histidine (his). Secreted (SEC) and cell-associated (CELLS) HISPTH-PRP∆T was recovered by immunoprecipitation and analyzed by 12.5% SDSPAGE and fluorography. 20 cells and stable cell lines secreting each construct were created. The glycosylation pattern of PROPTH-tG suggests that it correctly traverses the intracellular transport pathway. As is the case for tG (see Fig. 5A; Fig. 2, Moore and Kelly, 1985), the predominant intracellular form of PROPTH-tG is Endo Hsensitive (Fig. 5A) indicating its presence in the ER. A minor fraction of both cell-associated tG and PROPTH-tG is Endo Hresistant (not visible at the exposure shown) and represents the post-Golgi pool. The small amount of intracellular Endo Hresistant tG and PROPTH-tG is consistent with the absence of both polypeptides in secretory granules. In contrast to the intracellular polypeptides, secreted tG and PROPTH-tG are entirely resistant to Endo H digestion (Fig. 5B). We examined the secretion in the presence and absence of 8-Br-cAMP to determine whether PROPTH-tG is rerouted to the regulated secretory pathway. The effect of secretagogue was examined following a 1 hour pulse with Expre35S35S label and a 5 hour chase. As shown in Fig. 6, no 8-Br-cAMPdependent stimulation of PROPTH-tG was observed. The same result was obtained with tG (Fig. 6) as was expected based on previous findings (Moore and Kelly, 1985). These results obtained with a 1 hour labeling were confirmed with cells that were labeled for 15 hours to approach steady state (data not shown). Our findings suggest that PROPTH does not contain independent sorting information for the regulated pathway. Fig. 5. Endoglycosidase H digest of tG and PROPTHtG. Clones expressing tG and PROPTH-tG were labeled for 2 hours with Expre35S35S label and the secreted and cell-associated tG-related polypeptides were immunoprecipitated. The immunoprecipitates were subjected to Endo H digestion as indicated. (A) Cell-associated tG and PROPTH-tG polypeptides. (B) Secreted tG and PROPTH-tG polypeptides. Fig. 6. Secretion of tG and PROPTH-tG. Two wells of cells expressing tG or PROPTH-tG were labeled for 1 hour with Expre35S35S label and chased for 5 hours without a secretagogue (1st chase). Subsequently, cells were chased for an additional 2.5 hours in the absence (−) or presence (+) of 5 mM 8-Br-cAMP (2nd chase). tG-related polypeptides were immunoprecipitated and analyzed by 10% SDS-PAGE and fluorography. The slightly greater amount of PROPTH-tG in the presence of 8-Br-cAMP during the second chase reflects a greater number of cells in this well; the amount of PROPTHtG recovered from this well in the first chase was also greater by the same factor. Quantitation indicates that the average stimulation of secretion of tG and PROPTH-tG was 0% of total (average of two experiments). DISCUSSION In this study we have achieved rapid intracellular transport of a deletion mutant, PRP∆T, and of full-length PRP by attaching to them the propeptide of PTH. Using this approach we have gained new insight about the intracellular transport and sorting of these unusual secretory proteins. Our findings argue against the presence of a universal sorting signal for regulated secretory proteins and are consistent with the notion that protein-specific interactions are important. Intracellular transport The relatively slow ER export of full-length PRP suggested the presence of an ER retention mechanism for PRPs (Castle et al., 1992). Since polypeptides lacking either the transition domain (PRP∆T) or the C-terminal domain (PRP∆C) also exit the ER slowly (Castle et al., 1992), this ER retention most likely involves the proline-rich repeats. Proline-rich domains are known to function in protein-protein interactions (Williamson, 1994), and PRPs may bind to resident ER protein(s) via their proline-rich repeats. For full-length PRP, ER retention decreases with increasing concentration (level of expression) of PRP (Castle et al., 1992). Based on this observation, we have inferred that self-association of PRPs competes with retention in the ER. This self-association appears to be mediated by the transition region as its deletion causes extensive retention of PRP∆T regardless of concentration (Castle et al., 1992). Replacement of the transition region with the short and basic PTH propeptide causes rapid ER export that is independent of level of expression, suggesting that the ER retention mechanism has been bypassed. The same results are obtained when the PTH propeptide and the transition regions are both present (PROPTH-PRP), suggesting that the propeptide has a dominant effect. As pointed out by Mains et al. (1995) the PTH propeptide exhibits a striking sequence similarity to the propeptides of PAM and of albumin, both of which are thought to accelerate the rate of secretion of the respective polypep- 1644 L. E. Stahl and others tides, most likely as a consequence of enhanced ER export (Mains et al., 1995; McCracken and Kruse, 1989). Together with the finding that the effect of the PAM propeptide is transferrable to another protein, PC2 (Mains et al., 1995), these results are suggestive of a specialized mechanism involving recognition of the propeptide sequence. However, deletion of the propeptide from PTH does not alter its own rate of transport (Wiren et al., 1988) suggesting that not all proteins are affected by the presence of the propeptide. Clearly, further studies will be required to elucidate the mode of action of these propeptides. Using both electrophoretic analysis and N-terminal microsequencing, we have demonstrated that the propeptide is cleaved from the chimeric PROPTH-PRP∆T before secretion. Further, processing was observed regardless of the pathway of secretion, suggesting that cleavage occurred before or concomitantly with sorting. Therefore, the post-Golgi and secreted forms of PROPTH-PRP∆T and PROPTH-PRP have identical primary structures to the original PRP∆T and PRP, validating the strategy of propeptide attachment as a bypass for slow ER export. Sorting and storage of PRP-related polypeptides It is clear that by accelerating ER export, PROPTH enhances the ability to detect granule storage of PRP and PRP∆T. We regard the possiblity that PROPTH contains sorting information for the regulated secretory pathway as unlikely for two reasons. First, deletion of the propeptide from the precursor of PTH does not affect the pathway of secretion of PTH in GH4 cells (Wiren et al., 1988). Second, when the PTH propeptide was fused to a reporter protein, tG, the chimeric polypeptide was not targeted to the regulated pathway of AtT-20 cells (Fig. 6, this work). We also considered the possibility that the propeptide might affect sorting through its interaction with the proteolytic processing machinery. An intact cleavage site has been shown to be necessary for the storage of von Willebrand factor in Weibel-Palade bodies, and it has been proposed that the processing enzyme may function in sorting either by facilitating local concentration and aggregation of propeptide-containing polypeptides or by serving as a ‘receptor’ at sites of sorting (Journet et al., 1993). However, we have shown that disruption of PTH propeptide processing by mutagenesis of the cleavage site did not abrogate regulated secretion. While this would appear to rule out a major role of the processing enzymes in targeting of the propeptide-containing chimeras, we can’t discount the possibility that processing enzymes might affect the degree of storage because the percentage of 8-Br-cAMP dependent secretion of HISPTH-PRP∆T was slightly lower than that of PROPTHPRP∆T. Because the major role of PROPTH is to accelerate ER export of PRPs rather than to serve as a sorting signal, we believe that sorting of PRPs for stimulated secretion must rely on structural characteristics of the PRP itself. While we previously regarded the transition domain of PRP as necessary for storage in secretory granules (Castle et al., 1992), the present findings argue convincingly that this is not the case. Rather the transition domain mainly facilitates efficient transport of the native protein from the ER. Thus sorting of PRP for regulated secretion must depend on one of the two remaining domains, either the central proline-rich repeats or the C-terminal segment. Our earlier study ruled out any effect of the C- terminal segment on sorting, so by elimination, we are left with the proline-rich repeats as being responsible. Since this domain is unique in primary and probably higher order structure among regulated secretory proteins, any interactions that it mediates are likely to be specific to the PRP family. Strikingly, regulated secretion of PRP-related polypeptides shows a dependence on level of expression. An incremental increase in the percentage stimulated secretion is observed when expression is increased beyond a certain level. While both ER and storage granule pools of PRP-related polypeptides are elevated when the level of expression is increased, our data clearly indicate that the storage pool is preferentially enriched (Table 4). In contrast, the fractional storage of the endogenous hormone ACTH was unaffected by the expression of PRPrelated polypeptides. Consequently, PRP-related polypeptides either are stored in increased amounts in the ACTH-containing granules or are segregated into a second storage compartment with the same regulation. Because the incremental increase in storage was observed for the full-length PRP as well as for the chimeric polypeptides, the presence of PROPTH is not necessary for this effect. We propose that the jump in storage is driven by the increased steady state concentration at sites of secretory granule formation and reflects homo- or heterotypic aggregation involving PRP-related polypeptides. The nonlinear dependence on concentration (level of expression) is consistent with a cooperative process which is catalyzed once a threshold concentration is reached. The jump in storage for PROPTH-PRP∆T and PROPTH-PRP occurs at a 5- to 10-fold lower expression level than for full-length PRP, suggesting that PROPTH may promote somewhat different interactions than occur for fulllength PRP. In any case, the concentration-dependent storage of PRPs is best explained as an aggregation process, and it may reflect the same type of associations that cause PRPs to concentrate visibly within the content of parotid acinar granules (Kousvelari et al., 1982). In conclusion, our studies put in question the existence of a universal sorting signal for regulated secretion as has been postulated previously (Kizer and Tropsha, 1991; Gorr and Darling, 1995). On the other hand, we wish to emphasize that the fractional storage that we have observed for PRPs and that others have observed for several nonendocrine proteins (e.g. Fennewald et al., 1988; Castle et al., 1992; Schmidt and Moore, 1994; Reaves et al., 1990; Stoller and Shields, 1989; Castle et al., 1995) and free glycosaminoglycan chains (Matsuuchi and Kelly, 1991) is typically well below the extent of storage achieved by the endogenous hormone, ACTH. Thus we believe that it will be important in future studies to evaluate sorting for regulated secretion on the basis of the amount of polypeptide stored. Cell- or protein-specific sorting signals (Cool et al., 1995; Arrandale and Dannies, 1994; Stoller and Shields, 1989) or highly efficient aggregation and retention mechanisms (Kuliawat and Arvan, 1994) may be superimposed on a bulk flow pathway leading to forming granules, and proteins that are either absent or poorly represented in regulated secretion may be progressively excluded from the maturing storage compartment. We are grateful to Yun Shim for his skillful assistance in the mutagenesis and cloning of HISPTH-PRP∆T. We appreciate the helpful advice of Drs R. Mains and B. Eipper (Johns Hopkins Medical Sorting of proline-rich proteins 1645 School), regarding the mutagenesis of the PTH propeptide, and of the reviewers of the submitted manuscript. We thank Dr Robert Wagner (University of Virginia) for his gift of an antibody against the VSV protein G, and Dr Jack Rose for his gift of the cDNA of the truncated G protein. This work was supported by NIH Grant DE08491. REFERENCES Arrandale, J. M. and Dannies, P. S. (1994). Inhibition of rat prolactin (PRL) storage by coexpression with human PRL. Mol. Endocrinol. 8, 1083-1090. Arvan, P. and Castle, J. D. (1986). Isolated secretion granules from parotid glands of chronically-stimulated rats possess an alkaline internal pH and inward-directed H+ pump activity. J. Cell Biol. 103, 1257-1267. Arvan, P. and Castle, D. (1992). Protein sorting and secretion granule formation in regulated secretory cells. Trends Cell Biol. 2, 327-331. Bauerfeind, R. and Huttner, W. B. (1993). Biogenesis of constitutive secretory vesicles, secretory granules and synaptic vesicles. Curr. Opin. Cell Biol. 5, 628-635. Blair, E. A., Castle, A. M. and Castle, J. D. (1991). Proteoglycan sulfation and storage parallels storage of basic secretory proteins in exocrine cells Am. J. Physiol. 261, C897-C905. Burgess, T. L., Craik, C. S. and Kelly, R. B. (1985). The exocrine protein trypsinogen is targeted into the secretory granules of an endocrine cell line: studies by gene transfer. J. Cell Biol. 101, 639-645. Castle, A. M., Stahl, L. E. and Castle, J. D. (1992). A 13-amino acid Nterminal domain of a basic proline-rich protein is necessary for storage in secretory granules and facilitates exit from the endoplasmic reticulum. J. Biol. Chem. 267, 13093-13100. Castle, A. M. and Castle, J. D. (1993). Novel secretory proline-rich proteoglycans from rat parotid: cloning and characterization by expression in AtT-20 cells. J. Biol. Chem. 268, 20490-20496. Castle, A. M., Schwarzbauer, J. E., Wright, R. L. and Castle, J. D. (1995). Differential targeting of recombinant fibronectins based on their efficiency of aggregation. J. Cell Sci. 108, 3827-3837. Chanat, E. and Huttner, W. B. (1991). Milieu-induced, selective aggregation of regulated secretory proteins in the trans-Golgi network. J. Cell Biol. 115, 1505-1519. Chu, W. N., Baxter, J. D. and Reudelhuber, T. L. (1990). A targeting sequence for dense secretory granules resides in the active renin protein moiety of human preprorenin. Mol. Endocrinol. 4, 1905-1913. Colomer, V., Lal, K., Hoops, T. C. and Rindler, M. J. (1994). Exocrine granule specific packaging signals are present in the polypeptide moiety of the pancreatic granule membrane protein GP2 and in amylase: implications for protein targeting to secretory granules. EMBO J. 13, 3711-3719. Cool, D. R., Fenger, M., Snell, C. R. and Loh, Y. P. (1995). Identification of the sorting motif within pro-opiomelanocortin for the regulated secretory pathway. J. Biol. Chem. 270, 8723-8729. Dickerson, I. M., Dixon, J. E. and Mains, R. E. (1987). Transfected human neuropeptide Y cDNA expression in mouse pituitary cells. J. Biol. Chem. 262, 13646-13653. Fennewald, S. M., Hamilton, R. L and Gordon, J. I. (1988). Expression of human preproapoAI and pre(∆pro)apoAI in murine pituitary cell line (AtT20). J. Biol. Chem. 263, 15568-15577. Gorr, S. U. and Darling, D. S. (1995). An N-terminal hydrophobic peak is the sorting signal of regulated secretory proteins. FEBS Lett. 361, 8-12. Habener, J. F., Amherdt, M., Ravazzola, M. and Orci, L. (1979). Parathyroid hormone biosynthesis: correlation of conversion of biosynthetic precoursers with intracellular protein migration as determined by electron microscope autoradiography. J. Cell Biol. 80, 715-731. Hellerman, J. G., Cone, R. C., Potts, J. T., Rich, A., Mulligan, R. C. and Kronenberg, H. M. (1984). Secretion of human parathyroid hormone from rat pituitary cells infected with a recombinant retrovirus encoding preproparathyroid hormone. Proc. Nat. Acad. Sci. USA 81, 5340-5344. Hendy, G. N., Kronenberg, H. M., Potts, J. T. and Rich, A. (1981). Nucleotide sequence of cloned cDNAs encoding human preproparathyroid hormone. Proc. Nat. Acad. Sci. USA 78, 365-7369. Journet, A. M., Saffaripour, S., Cramer, E. M., Tenza, D. and Wagner, D. D. (1993). Von Willebrand factor storage requires intact prosequence cleavage site. Eur. J. Cell Biol. 60, 31-41. Kizer, J. S. and Tropsha, A. (1991). A motif found in propeptides and prohormones that may target them to secretory vesicles. Biochem. Biophys. Res. Commun. 174, 586-592. Kousvelari, E. E., Oppenheim, F. G. and Cutler, L. S. (1982). Ultrastructural localization of salivary acidic proline-rich proteins from Macaca fascicularis. J. Histochem. Cytochem. 30, 274-278. Kuliawat, R. and Arvan, P. (1994). Distinct molecular mechanisms for protein sorting within immature secretory granules of pancreatic B-cells J. Cell Biol. 126, 77-86. Laemmli, U. K. (1970). Cleavage of structural proteins during the assembly of the head of bacteriophage T4. Nature 227, 680-685. Mains, R. E., Milgram, S. L., Keutmann, H. T. and Eipper, B. A. (1995). The NH2-terminal proregion of peptidyl-amidating monooxygenase facilitates the secretion of soluble proteins. Mol. Endocrinol. 9, 3-13. Matsuuchi, L. and Kelly, R. B. (1991). Constitutive and basal secretion from the endocrine cell line, AtT-20 J. Cell Biol. 112, 843-852. McCracken, A. A. and Kruse, K. B. (1989). Intracellular transport of rat serum albumin is altered by a genetically engineered deletion of the propeptide. J. Biol. Chem. 264, 20843-20846. Moore, H.-P. H. and Kelly, R. B. (1985). Secretory protein targeting in a pituitary cell line: differential transport of foreign secretory proteins to distinct secretory pathways. J. Cell Biol. 101, 1773-1781. Moore, H.-P. H. and Kelly, R. B. (1986). Re-routing of a secretory protein by fusion with human growth hormone sequences. Nature 321, 443-446. Moore, H.-P. H., Walker, M. D., Lee, F. and Kelly, R. B. (1983). Expressing a human proinsulin cDNA in a mouse ACTH secreting cell. Intracellular storage, proteolytic processing and secretion on stimulation. Cell 35, 531538. Neufeld, G., Mitchell, R., Ponte, P. and Gospodarowicz, D. (1987). Expression of human basic fibroblast growth factor cDNA in baby hamster kidney-derived cells results in autonomous cell growth J. Cell Biol. 106, 1385-1394. Reaves, B. J., Van Itallie, C. M., Moore, H.-P. M. and Dannies, P. S. (1990). Prolactin and insulin are targeted to the regulated pathway in GH4C1 cells, but their storage is differentially regulated. Mol. Endocrinol. 4, 1017-1026. Schmidt, W. K. and Moore, H.-P. H. (1994). Synthesis and targeting of insulin-like growth factor-I to the hormone storage granules in an endocrine cell line. J. Biol. Chem. 269, 27115-27124. Schwarzbauer, J. E., Mulligan, R. C. and Hynes, R. O. (1987). Efficient and stable expression of recombinant fibronectin polypeptides Proc. Nat. Acad. Sci. USA 84, 754-758. Seidah, N. G., Day, R. and Chretien, M. (1993). The family of pro-hormone and pro-protein convertases. Biochem. Soc. Trans. 21, 685-690. Smeekens, S. P., Avruch, A. S., LaMendola, J., Chan, S. J. and Steiner, D. F. (1991). Identification of a cDNA encoding a second putative prohormone convertase related to PC2 in AtT-20 cells and islets of Langerhans. Proc. Nat. Acad. Sci. USA 88, 340-344. Steiner, D. F., Smeekens, S. P., Ohagi, S. and Chan, S. J. (1992). The new enzymology of precursor processing Endoproteases. J. Biol. Chem. 267, 23435-23438. Stoller, T. J. and Shields, D. (1989). The propeptide of preprosomatostatin mediates intracellular transport and secretion of alpha-globin from mammalian cells. J. Cell Biol. 108, 1647-1656. West, D. C., Sattar, A. and Kumar, S. (1985). A simplified in situ solubilization procedure for the determination of DNA and cell number in tissue cultured mammalian cells. Analyt. Biochem. 147, 289-295. Williamson, M. P. (1994). The structure and function of proline-rich regions in proteins. Biochem. J. 297, 249-260. Wiren, K. M., Potts, J. T. and Kronenberg, H. M. (1988). Importance of the propeptide sequence of human preproparathyroid hormone for signal sequence function. J. Biol. Chem. 263, 19771-19777. (Received 25 October 1995 - Accepted 10 March 1996)