In endocrine cells, prohormones and granins are segregated in the TGN (trans-Golgi network) from constitutively secreted proteins, stored in concentrated form in dense-core secretory granules, and released in a regulated manner on specific stimulation. The mechanism of granule formation is only partially understood. Expression of regulated secretory proteins, both peptide hormone precursors and granins, had been found to be sufficient to generate structures that resemble secretory granules in the background of constitutively secreting, non-endocrine cells. To identify which segment of CgA (chromogranin A) is important to induce the formation of such granule-like structures, a series of deletion constructs fused to either GFP (green fluorescent protein) or a short epitope tag was expressed in COS-1 fibroblast cells and analysed by fluorescence and electron microscopy and pulse-chase labelling. Full-length CgA as well as deletion constructs containing the N-terminal 77 residues generated granule-like structures in the cell periphery that co-localized with co-expressed SgII (secretogranin II). These are essentially the same segments of the protein that were previously shown to be required for granule sorting in wild-type PC12 (pheochromocytoma cells) cells and for rescuing a regulated secretory pathway in A35C cells, a variant PC12 line deficient in granule formation. The results support the notion that self-aggregation is at the core of granule formation and sorting into the regulated pathway.
In addition to the constitutive secretory pathway present in all cells, endocrine and neuroendocrine cells possess a regulated pathway [1,2]. Peptide hormone precursors and granins are sorted at the TGN (trans-Golgi network) into secretory granules, where they are stored in a densely packed form and processed by granule-specific enzymes. On transport to specific sites in the cell periphery, their fusion with the plasma membrane and the release of their contents are triggered by external stimuli. The mechanisms of granule biogenesis and the machinery required are still incompletely understood; yet, it is clear that an entire gene expression programme is responsible for regulated secretion in endocrine cells .
Granule formation depends at least in part on self-aggregation of cargo proteins under TGN conditions of high calcium and low pH. According to the ‘sorting-by-retention’ model, this could be the major mechanism of segregation, since soluble components of an immature granule are removed by ‘constitutive-like’ secretory vesicles . Alternatively, it has been proposed that sorting receptors, recognizing specific ‘signals’ within soluble or aggregated cargo, recruit granule content in a ‘sorting-for-entry’ mechanism not unlike protein sorting by classical cargo receptors (e.g. the mannose-6-phosphate receptors) . Since several proteins of the regulated pathway have been found to interact with lipid rafts (e.g. prohormone convertase 2 and carboxypeptidase E) [5–7], binding to specific lipid microdomains in the TGN may also play a role in sorting into the regulated secretory pathway.
Granins are a family of granule cargo proteins found throughout the endocrine and neuroendocrine system. In addition to an extracellular function via the release of bioactive peptides, they are believed to play a role in granule biogenesis . This is most prominently the case for CgA (chromogranin A). There is evidence that CgA is required for the formation of granules in sympathoadrenal chromaffin cells. Depletion of CgA reduced the number of secretory granules in the sympathoadrenal PC12 (pheochromocytoma cells) cell line [9,10] and induced the degradation of other granule proteins . Knock-out of the CgA gene or knock-down of CgA expression in mice reduced catecholamine-storage vesicles in adrenal chromaffin cells [11,12] and lowered the content of other chromaffin granule cargo . Furthermore, CgA expression was able to rescue granule formation and the regulated secretory phenotype in the CgA-deficient pituitary variant cells 6T3 and the sympathoadrenal variant cells A35C [9,13], but not in the PC12 cell clone PC12-27 cells, where expression of the regulated secretory programme has been lost completely [3,14]. CgA was recently shown to regulate granule formation in wild-type PC12 cells and in mutant 6T3 cells by inducing the expression of the serine protease inhibitor PN-1 (protease nexin-1) that inhibits the degradation of granule proteins in the Golgi .
Interestingly, it has also been observed that granule cargo proteins, when expressed in various non-endocrine cell lines like COS, CV1 or HEK-293 (human embryonic kidney cells), accumulate in granule-like structures [9,10,16]. This was the case not only for CgA, but also for CgB and SgII (secretogranin II), as well as the peptide hormone precursors pro-vasopressin, pro-oxytocin and POMC (pro-opiomelanocortin). Similar structures have previously also been observed in HepG2 cells expressing proinsulin . The accumulations were membrane-bound and resembled secretory granules morphologically on electron microscopy. They represent post-Golgi structures, devoid of markers of endosomes or lysosomes. The formation of SgII or CgB accumulations was accompanied by significant intracellular storage in COS cells, although storage was hardly measurable for pro-vasopressin . So far, there is no indication that regulated cargo synthesis triggers an endocrine gene expression programme for the regulated secretion machinery. The phenomenon of cargo-induced granule-like structures rather suggests that aggregation of cargo in the TGN is sufficient to trigger the formation of minimal granules with limited functionality in the absence of an endocrine-specific machinery.
For several regulated secretory proteins, it has been possible to identify sequences necessary, and in some cases even sufficient, for granule targeting in endocrine cell lines, for example the propeptide of prosomatostatin , the N-terminal sequence of POMC , and the disulfide-bonded loop segment of CgB [20,21]. Corresponding receptors, however, have remained elusive and it remains to be clarified whether these sequences mediate binding to protein or lipid receptors, and/or enable aggregation.
In the case of CgA, a GFP (green fluorescent protein) fusion protein expressed in neuroendocrine PC12 cells was shown to be sorted to dense-core secretory granules and to be exocytosed in a regulated manner . In a deletion analysis, the information necessary for such a trafficking was found to be contained within the N-terminal but not the C-terminal region of CgA. Residues 1–115 were sufficient for granule sorting. However, unlike for CgB, the conserved N-terminal disulfide-linked loop structure of Cys-17–Cys-38 in CgA was not sufficient, nor was the disulfide bond necessary. The same deletion constructs were also used to test their ability to restore the regulated secretory pathway in the CgA-deficient mutant PC12 cell line A35C . Again, formation of granule structures destined for regulated exocytosis was mediated by a determinant located within CgA's N-terminal segment of residues 1–115, but not its C-terminal region (residues 233–439). This indicated that the same or overlapping sequences are responsible for sorting into the regulated secretory pathway and triggering recovery of A35C to a neuroendocrine phenotype.
In the present study, we have addressed the question as to which parts of CgA are important to induce the formation of granule-like structures in non-endocrine cells. We find that these are essentially the same segments of the protein that were shown to be required for granule sorting in PC12 cells and for rescuing regulated secretion in A35C cells. Since the formation of granule-like structures most likely reflects the ability of a polypeptide to aggregate in the TGN, the result supports the notion that aggregation is at the core of all these processes.
Construction of the expression plasmids pCMV-CgA253-EGFP, pCMV-CgA367-EGFP, pCMV-CgA481-EGFP, pCMV-CgA595-EGFP, pCMV-CgA640-EGFP, pCMV-CgA763-EGFP, pCMV-CgA805-EGFP, pCMV-CgA-EGFP, pCMV-CgAΔC/E-EGFP, pCMV-CgAΔN-EGFP, pCMV-SgP-EGFP-CgA365-481 and pCMV-SgP-EGFP encoding the GFP fusion proteins CgA1-39–GFP, CgA1-77–GFP, CgA1-115–GFP, CgA1-153–GFP, CgA1-168–GFP, CgA1-209–GFP, CgA1-224–GFP, full-length CgA–GFP, CgAC17E–GFP, CgA233-439–GFP, sp-GFP–CgA77-115 and sp-GFP respectively, all with the 18-amino-acid sp (signal peptide) of CgA, has been described previously . To exchange the GFP portion of these constructs for the 14-amino-acid C1 epitope sequence ETELDKASQEPPLL  or for the 11-amino-acid c-Myc epitope EQKLISEEDLN, the corresponding cDNA sequence flanked by KpnI and NotI sites was ligated to the KpnI site at the end of the CgA segment. Construction of expression plasmids encoding SgII C-terminally tagged with a C1 or c-Myc epitope has been described previously .
Cell culture and transfection
COS-1 cells were grown in Dulbecco's minimal essential medium, supplemented with 2 mM L-glutamine, 100 units/ml penicillin, 100 μg/ml streptomycin and 10% (v/v) fetal calf serum, at 37 °C with 7.5% CO2. Cells were transfected using Lipofectin® (Life Technologies) with 2 μg of plasmid DNA per 35 mm dish at ∼40% confluency and processed after 48 h for immunofluorescence microscopy.
We used monoclonal antibodies against the c-Myc epitope (9E10), against LAMP1 (lysosome-associated membrane protein 1) from J. Rohrer (University of Zürich), and against rab5 from R. Jahn (MPI, Göttingen, Germany), and a rabbit antipeptide serum recognizing the C1 epitope. As secondary antibodies, non-cross-reacting cy3-labelled goat anti-mouse, cy2-labelled goat anti-rabbit immunoglobulin antibodies (from Jackson Immunoresearch and Amersham Biosciences) were used as appropriate according to the manufacturers' recommendations.
Cells were grown on glass coverslips, fixed with 3% (w/v) paraformaldehyde in PBS for 15 min at room temperature, washed in PBS, quenched with 50 mM NH4Cl in PBS, and permeabilized with 0.1% Triton X-100 (0.1% saponin for LAMP1 staining) for 10 min. For antibody staining, non-specific antibody binding was blocked with PBS containing 1% (w/v) BSA. The fixed cells were incubated at room temperature with primary antibodies for 1 h, washed in PBS with albumin, and stained with fluorescent secondary antibodies in PBS with albumin for 30 min. After several washes in PBS with albumin, PBS and water, the coverslips were mounted on Mowiol 4-88 (Hoechst). Staining patterns were analysed using a Zeiss Axioplan 2 microscope with a KX Series Imaging System (Apogee Instruments). To stain for LAMP1, saponin was used instead of Triton and 20 mM glycine instead of NH4Cl.
To analyse the stability and motility of granule-like structures, cells expressing GFP-tagged proteins were analysed by live microscopy in medium buffered with 25 mM Hepes on the above microscope inside a temperature-controlled box at 37 °C (Life Imaging Services, Basel, Switzerland). To block protein synthesis, 10 μg/ml cycloheximide was added to the medium and the cells were returned to the CO2 incubator during incubation for 3 h.
To analyse the secretion behaviour, COS-1 cells transfected with C1-tagged CgA constructs were labelled with [35S]methionine (0.1 mCi/ml; from NEN) at 37 °C for 40 min and chased in unlabelled medium for 5 h before immunoprecipitation of the medium and the cells, SDS/PAGE and autoradiography. To more sensitively assess storage, COS-1 cells transfected with CgA-C1, CgAC17E-C1, SgII-C1 or a sulfatable version of A1PiTS [tyrosine sulfation-tagged A1Pi (α1-protease inhibitor)] [23a] were starved for sulfate in sulfate-free medium (from Gibco BRL) for 30 min, labelled with [35S]sulfate (0.5 mCi/ml; from Amersham) for 90 min at 19 °C, followed by a chase at 37 °C for 6 h in a medium containing excess unlabelled sulfate. The labelled protein secreted into the medium as well as that retained in the cells was immunoprecipitated and analysed by gel electrophoresis and autoradiography. Signals were quantified by a phosphoimager.
Transfected cells were fixed in 3% formaldehyde and 0.2% glutaraldehyde for 2 h at room temperature; after the addition of 1% picric acid overnight at 4 °C, the cells were scraped, pelleted, resuspended and washed three times in PBS, treated with 50 mM NH4Cl in PBS for 30 min to quench free aldehydes and washed three times in PBS. The final pellet was resuspended in 2% warm agarose and left on ice to solidify. Agarose pieces were dehydrated and infiltrated with LR-gold resin (London Resin, London, U.K.) according to the manufacturer's instructions and allowed to polymerize for 1 day at 4 °C. For immunogold labelling, sections of ∼60 nm were collected on carbon-coated Formvar-Ni grids, incubated with rabbit anti-Myc antibody (Abcam, Cambridge, U.K.) in PBS, 2% BSA and 0.1% Tween 20 for 2 h, washed with PBS and incubated with 15-nm colloidal gold-conjugated goat anti-mouse immunoglobulin antibodies (BioCell, Cardiff, U.K.) in PBS, 2% BSA and 0.1% Tween 20 for 90 min. Grids were washed five times for 5 min in PBS and then five times in water, before staining for 10 min in 2% uranyl acetate followed by 1 min in Reynolds lead citrate solution. Sections were viewed in a Phillips CM100 electron microscope.
PN-1 knock-down cells and immunoblotting
To perform shRNA (small-hairpin RNA)-mediated silencing of PN-1 in mouse mammary tumour 4T1 cells, the sequence GCCGCGTACCTGTCACTAC was targeted. Briefly, the two complementary oligonucleotides GCCGCGTACCTGTCACTACTTCAAGAGAGTAGTGACAGGTACGCGGCTTTTTT and AATTAAAAGCCGCGTACCTGTCACTACTCTCTTGAAGTAGTGACAGGTACGCGGCGGCC were annealed and the resulting shRNA was cloned into the ApaI and EcoRI sites of pSilencer 1.0-U6 (Ambion). 4T1 cells were co-transfected with pSilencer-PN-1shRNA and the puromycin resistance vector pBABE at the ratio of 20:1. As a mock-transfected control, 4T1 cells were co-transfected with pSilencer and pBABE at the same ratio. Cell lines resistant to 20 μg/ml puromycin were isolated and analysed for PN-1 expression by immunoblotting using the monoclonal anti-PN-1 antibody 4B3.
GFP fusions of CgA and N-terminal CgA segments generate granule-like structures in COS-1 cells
CgA fused to GFP (schematically illustrated in Figure 1) was expressed in COS-1 cells, and its distribution within the cell was analysed by fluorescence microscopy. As is shown in Figure 2(B), the fusion protein accumulated in the Golgi area as well as in punctate structures throughout the cell. Since constitutive secretory vesicles generally cannot be visualized by light microscopy because of insufficient amount and concentration of cargo per vesicle, the appearance of a punctate pattern thus suggested the dense accumulation of the protein in different structures. A secretory form of GFP alone, which is targeted to the ER (endoplasmic reticulum) by the sp (signal peptide) of CgA fused to its N-terminus (sp-GFP), only stained the ER and the Golgi in transit to the cell surface, but did not produce peripheral accumulations (Figure 2I). The fluorescence pattern of the CgA–GFP fusion protein closely resembles that of SgII (Figure 2A) and of other granule cargo that have been shown to generate granule-like structures in various non-endocrine cell lines . Indeed, on co-expression, CgA–GFP and SgII accumulated to a significant extent together in the same structures (Figure 3A). CgA–GFP co-stained with the Golgi marker giantin near the nucleus, but not in the periphery (Figure 4A). Furthermore, the peripheral accumulations stained neither for the ER marker BAP31 (results not shown) nor for the endosomal marker rab5 or the lysosomal protein LAMP1 (Figure 4A). These structures thus do not represent intermediates of lysosomal degradation. Instead, CgA–GFP appears to aggregate in the TGN of COS-1 cells, as it does in endocrine cells, and accumulate in granule-like structures.
Schematic representation of CgA fusion proteins
Formation of granule-like structures by GFP fusions of CgA fragments
Co-expression of CgA–GFP fusion constructs with SgII
Co-staining of CgA–GFP fusion constructs with organelle markers
To identify the segment in CgA necessary for the formation of granule-like structures, GFP fusions of C-terminally truncated CgA were expressed in COS-1 cells. The N-terminal domains spanning residues 1–225, 209, 168, 153 or 115 were sufficient to produce the same fluorescence pattern as full-length CgA–GFP (Figures 2C, 2D and 4C and results not shown). These fusion proteins also co-localized with SgII accumulations when co-expressed (Figures 3B and 3C). In contrast, the 1–16 (results not shown) or 1–39 N-terminal residues were unable to cause peripheral GFP accumulations, and fluorescence was detectable in the ER and in the Golgi (Figure 2F). Even in cells co-expressing SgII, no punctate pattern was observed for GFP fluorescence (Figure 3E), indicating that the N-terminal 39 residues of CgA, which contain the disulfide-linked domain, are neither sufficient to generate granule-like structures nor to sort into existing ones. The disulfide-linked domain is also not necessary for the formation of these structures by full-length CgA–GFP, since mutation of Cys-17 to glutamic acid residue in CgAC17E–GFP did not abolish granule-like structure formation and co-localization with SgII (Figures 3H and 4B).
Surprisingly, the intermediate construct CgA1-77–GFP as well as CgA233-439–GFP, the C-terminal portion of CgA, showed a different fluorescence pattern in transfected COS-1 cells. They produced large unshapely accumulations of often more than 1 μm, mostly near the nucleus, which did not co-localize with co-expressed SgII (Figures 2E, 2H, 3D and 3G); neither did they did not co-localize with Golgi or endosomal/lysosomal markers (Figure 4D). Using an antibody against GFP revealed additional ER staining (results not shown), suggesting that a significant portion of GFP did not fold correctly to acquire fluorescence. It appears most likely that these constructs aggregated early in the secretory pathway.
The segment of residues 77–115 of CgA had been found to be necessary but perhaps not sufficient for sorting to secretory granules in PC12 cells . However, expression of sp-GFP–CgA77-115 in COS-1 cells produced a characteristic Golgi and punctate peripheral distribution pattern (Figures 2G and 4E), indicating that residues 77–115 are sufficient for granule-like structure formation. Interestingly, the extent of co-localization of the chimaeric CgA77-115 domain (Figure 3F) with SgII-containing structures was significantly lower than for the N-terminal constructs (10–20%, compared with 60–80% for the longer fusion proteins like for example CgA1-115–GFP; Figures 3A–3C). Partitioning of different granule cargo into separate granule populations has also been observed in endocrine cells  and was explained by intrinsic biophysical properties of the cargo proteins that govern their (self-) aggregation behaviour. The levels of co-localization with SgII may therefore reflect the specific aggregation properties of the segment of residues 77–115 that are modified for example by the additional contribution of residues 1–77.
Analysis of epitope-tagged fragments of CgA expressed in COS-1 cells
To exclude artefactual effects of GFP, particularly for the two constructs that accumulated in the ER, we analysed a series of CgA constructs with a short C1 epitope tag in place of GFP (Figure 1). C1-tagged full-length CgA and CgA1-115 (Figures 5A and 5B), as well as CgAC17E, but not CgA1-39 (results not shown), generated granule-like structures, confirming the results obtained with the GFP fusion proteins. However, unlike the corresponding GFP fusions, CgA1-77-C1 and CgA233-439-C1 did not produce large aggregates. CgA1-77-C1 generated fine punctate accumulations that co-localized with co-expressed SgII (Figures 5D and 5D′). Residues 1–77 are thus sufficient to produce granule-like structures. In contrast, CgA233-439-C1 stained the ER and Golgi, the distribution pattern typical of a constitutively secreted protein (Figures 5C and 2I). The large aggregates of CgA1-77–GFP and CgA233-439–GFP are thus likely to be artefacts of the particular GFP fusion constructs.
Formation of granule-like structures by epitope-tagged CgA fragments
As summarized in Table 1, our results show that there is redundant information in the N-terminal portion of CgA for producing post-Golgi accumulations. Residues 1–77 (with a necessary contribution of the segment of residues 39–77) and residues 77–115 are sufficient to produce granule-like structures.
|Granule sorting in PC12 cells ||Rescue of granule formation in A35C cells ||Formation of granule-like in COS-1 cells (the present study)|
|Granule sorting in PC12 cells ||Rescue of granule formation in A35C cells ||Formation of granule-like in COS-1 cells (the present study)|
*Formation of large aggregates.
To morphologically characterize the accumulations of wild-type CgA and CgA fragments, Myc-tagged versions of CgA, CgA1-77 and CgA1-115 were expressed in COS-1 cells and analysed by immunogold electron microscopy. Distinct structures with high density of gold labelling were observed in the cell periphery for full-size CgA–Myc (Figures 6A and 6B) as well as for the short fragments CgA1-115–Myc (Figures 6E and 6F) and CgA1-77–Myc (Figures 6G–6I). The dense structures were ∼150–250 nm in size, and often in the centre of or lining the inside of large vesicles. Typically, the structures produced by the deletion constructs were less compact than those of full-size CgA–Myc. Distributed and concentrated labelling was also observed at the Golgi (Figure 6G). In comparison, background labelling and the gold density along ER membranes were very low, testifying to the high concentration of antigen in the labelled structures. Occasionally, small vesicles with gold labelling could also be found (Figure 6D), most likely corresponding to constitutive vesicles.
Immunogold electron microscopy of COS-1 cells expressing wild-type or truncated CgA–Myc
Granule-like structures are long-lived and of low mobility
Live microscopy of cells expressing CgA–GFP and CgA1-115–GFP showed most peripheral fluorescent accumulations to be stable over a time period of at least 30 min (Figure 7). With rare exceptions, they moved only little within a radius of a few micrometres within this time and without specific direction (as illustrated by the tracings of some of the punctate pattern in the rightmost panels). On blocking the synthesis of new proteins with cycloheximide for 3 h, peripheral fluorescent structures were still present (Figure 8). This contrasts with the properties of constitutive secretory vesicles, which have a much shorter half-life (e.g. ∼15 min for a sulfatable version of A1Pi from the TGN to the cell exterior in COS cells, as determined by pulse–chase experiments; [16,23a]). Indeed, the signal at the TGN was significantly reduced during the cycloheximide treatment (Figure 8), suggesting that a large portion of the GFP fusion proteins initially at the TGN was secreted via the constitutive pathway during the incubation period.
Live microscopy of granule-like structures
Stability of granule-like structures in the absence of protein synthesis
Efficiency of storage in granule-like structures in COS-1 cells is low
To analyse biochemically the secretory behaviour of CgA, C1-tagged full-length CgA, CgA1-115 and CgA1-77 were expressed in COS-1 cells, pulse-labelled with [35S]methionine, chased with unlabelled methionine for 5 h and immunoprecipitated from the incubation medium and from the cell lysate (Figure 9, lanes 1–6). Most of the protein was secreted into the medium, suggesting that only a minor fraction was retained in granule-like structures. CgA polypeptides recovered from the cells correspond to the unglycosylated ER form and degradation products. CgA is partially modified by chondroitin sulfate/dermatan sulfate-type glycosaminoglycans at sites of O-glycosylation (indicated by asterisks in Figure 1; [25,26]), giving rise to a somewhat broad band at ∼100 kDa that is efficiently sulfated in the TGN (Figure 9, lane 7). After pulse labelling with [35S]sulfate, approx. 15% of CgA or CgAC17E was still retained in the cells after 6 h of chase (lanes 14–17). This is significantly more than that of a sulfatable version of constitutively secreted A1PiTS (∼5%), but clearly less than that of SgII–C1 (∼43%). This result confirms that in COS-1 cells only a small portion of CgA is forming granule-like structures, while the rest is secreted constitutively.
Secretion behaviour of CgA constructs
Formation of granule-like structures by CgA does not require expression of PN-1
CgA was proposed to regulate granule formation in pituitary neurosecretory-deficient 6T3 cells by inducing the expression of PN-1 to stabilize regulated cargo in the Golgi . COS-1 cells endogenously express PN-1, but no increase in its levels could be detected by immunoblot analysis on transfection of CgA-C1 (N. Beuret, unpublished work). More importantly, we tested whether PN-1 expression is necessary for the formation of granule-like structures, taking advantage of stable cell lines derived from the mouse breast tumour cell line 4T1, in which PN-1 synthesis has been silenced by shRNA. CgA-C1 produced punctate accumulations both in the control line and in the knock-down cell line essentially lacking PN-1 (Figure 10), indicating that granule-like structures are formed by CgA independently of PN-1.
Formation of granule-like structures by CgA in the absence of PN-1
CgA is an extended polypeptide without any sizable folded domains that upon partial deletion might render the protein misfolded and recognizable to the ER quality control and degradation system. This makes CgA a suitable protein to perform deletion analysis to identify segments involved in late Golgi processes. However, it is conceivable that certain combinations of CgA and reporter sequences in a fusion protein might interfere with the folding of the reporter protein and/or cause aggregation even in the ER. This seems to happen with constructs CgA1-77–GFP and CgA233-439–GFP in COS-1 cells. However, when using small peptide tags instead of GFP as a reporter, the two CgA sequences exited the ER and their behaviour in the TGN could be assessed.
Here we analysed CgA fragments for their ability to generate granule-like structures in nonendocrine cells based on the following criteria. They accumulated in dense structures in the cell periphery (and at the TGN) that appeared as bright spots in fluorescence microscopy and as granular, densely gold-labelled structures in electron microscopy. These structures were long-lived; since they could be followed for at least 30 min in live cells and were still present after 3 h in the absence of protein synthesis. They co-localized with co-expressed SgII, another granule cargo, but not with Golgi or endosomal/lysosomal markers, indicating that they do not represent dissociated Golgi elements or intermediates of lysosomal degradation respectively. These properties contrast with those of constitutive secretory cargo, which cannot normally be visualized in constitutive secretory vesicles by light microscopy. Constitutive vesicles have a short lifetime, since transport from the TGN to the cell exterior has a half-life of only ∼15 min in COS cells . The observation that granule-like structures do not move in a directed fashion indicates that in COS cells the machinery for granule transport to sites of regulated fusion is missing.
The efficiency of sorting into granule-like structures in COS cells as determined by pulse–chase experiments is different for different cargo proteins, with storage of more than 40% of newly synthesized SgII after 6 h, but less than 20% for CgA or CgB (Figure 9 and ). It also differs between cell lines (e.g. for SgII and CgB; N. Beuret, unpublished work). This may suggest that the conditions in the TGN vary between cell types and are suboptimal for cargo aggregation in non-endocrine cells. It may also reflect the absence of an auxiliary machinery specific to regulated secretory cells to enhance sorting efficiency. Although a large fraction of CgA and of CgA truncation constructs is therefore constitutively secreted in COS cells, the phenomenon of granule-like structure formation in the context of non-endocrine cells is useful to reveal minimal mechanisms involved in granule biogenesis.
The results from our experiments in COS-1 cells are summarized in Table 1 together with those previously obtained using wild-type PC12  and sympathoadrenal variant A35C cells . All constructs containing the N-terminal 115 residues of CgA are sorted into secretory granules of PC12 cells, rescue granule formation in A35C cells, and are able to generate granule-like structures in non-endocrine COS-1 cells. The N-terminal 39 and the C-terminal 207 residues (amino acids 233–439) are not sufficient for any of these activities, and the disulfide bond near the N-terminus is not necessary.
There are only two constructs where the formation of granule-like structures did not clearly parallel sorting into secretory granules: CgA1-77 and CgA77-115 scored negative for sorting into secretory granules in PC12 cells, but were found to be sufficient to form granule-like structures in COS-1 cells. CgA1-77 with a C1 peptide tag formed the typical post-Golgi accumulations in COS-1 cells that on co-expression also co-localized with SgII. As a GFP fusion protein, however, the same sequence was retained in the ER due to misfolding. ER retention would explain why CgA1-77–GFP was not found in granules of PC12 cells. For the divergent behaviour of CgA77-115 in the two cell models, however, there is no obvious explanation. As has previously been pointed out , this sequence includes a segment potentially forming an amphipathic α-helix of three turns (H79SGFEDELSEVL90) that has been implicated as a general motif in granule sorting [27,28].
The striking correspondence of the sequences able to sort into the regulated pathway of PC12 cells, to rescue granule sorting in A35C cells, and to produce granule-like structures in non-endocrine cells suggests a common mechanism. Aggregation is a well-characterized property of regulated secretory proteins that can be reproduced in vitro in a high-calcium, mildly acidic milieu mimicking TGN conditions. This process is well established as part of the mechanisms underlying the biogenesis of dense-core secretory granules. Aggregation of granule cargo in the TGN of non-endocrine cells is a plausible minimal mechanism to explain the formation of membrane-bounded structures in the absence of an endocrine-specific machinery, whereby soluble protein cargo is stored in highly concentrated form and segregated from constitutively secreted proteins. According to this hypothesis, the deletion analysis performed in COS cells would identify segments mediating self-aggregation in the TGN. Self-aggregation in the TGN may well be sufficient for granule sorting in functional endocrine cells.
In A35C mutant cells, one might be tempted to speculate that aggregating CgA sequences provide the missing cargo to rescue the granulogenesis defect. However, it has been shown that another aggregating cargo protein, growth hormone, did not rescue granule formation when expressed in A35C cells, but was secreted constitutively. Only when co-expressed with CgA was growth hormone rerouted into a regulated secretory pathway. This suggests that CgA sequences exert an additional specific effect in endocrine cells such as A35C cells. Indeed, CgA has been shown in 6T3 cells to regulate expression of PN-1, which in turn prevents granule protein degradation . In COS cells, at least six different granule proteins, both granins and prohormones, produce granule-like structures . A CgA-specific induction of an endocrine expression programme thus cannot explain the phenomenon in COS cells, and we could not observe an increase of PN-1 in CgA–GFP-expressing COS cells. Using 4T1 cells lacking PN-1 showed that this protein is not necessary for granule-like structure formation by CgA–GFP.
It is conceivable that the two functions of aggregation in the TGN of any cell type and of regulation of granulogenic factors in endocrine cells are performed by the same CgA sequences. The two functions may even be connected, if only the aggregated sequences were able to trigger the signal. The molecular details of how signalling is accomplished are yet to be discovered.
We thank Dr Denis Monard for his support and Dr Reinhard Jahn and Dr Jack Rohrer for providing reagents.
tyrosine sulfation-tagged A1Pi
green fluorescent protein
lysosome-associated membrane protein 1
This work was supported by the Swiss National Science Foundation [grant number 31-109424] to M. S. and the NIH (National Institutes of Health) [grant numbers DK59628 and HL58120] to L. T. H. S. was partially supported by the Roche Research Foundation [grant number 216-2002].
These authors have contributed equally to this work.