Sorting of the Neuroendocrine Secretory Protein Secretogranin II into the Regulated Secretory Pathway

Secretogranin II (SgII) belongs to the granin family of prohormones widely distributed in dense-core secretory granules (DCGs) of endocrine, neuroendocrine, and neuronal cells, including sympathoadrenal chromaffin cells. The mechanisms by which secretory proteins, and granins in particular, are sorted into the regulated secretory pathway are unsettled. We designed a strategy based on novel chimeric forms of human SgII fused to fluorescent (green fluorescent protein) or chemiluminescent (embryonic alkaline phosphatase) reporters to identify trafficking determinants mediating DCG targeting of SgII in sympathoadrenal cells. Three-dimensional deconvolution fluorescence microscopy and secretagogue-stimulated release studies demonstrate that SgII chimeras are correctly targeted to DCGs and released by exocytosis in PC12 and primary chromaffin cells. Results from a Golgi-retained mutant form of SgII suggest that sorting of SgII into DCGs depends on a saturable sorting machinery at the trans-Golgi/trans-Golgi network. Truncation analyses reveal the presence of DCG-targeting signals within both the N- and C-terminal regions of SgII, with the putative α-helix-containing SgII-(25-41) and SgII-(334-348) acting as sufficient, independent sorting domains. This study defines sequence features of SgII mediating vesicular targeting in sympathoadrenal cells and suggests a mechanism by which discrete domains of the molecule function in sorting, perhaps by virtue of a particular arrangement in tertiary structure and/or interaction with a specific component of the DCG membrane.

In neuroendocrine cells, newly synthesized secretory proteins are targeted to the constitutive or regulated branches of the secretory pathway (1,2). Secretory proteins entering the regulated pathway are concentrated into dense-core secretory granules (DCGs) 2 prior to release by exocytosis in response to secretagogue. Two basic models have been proposed, namely "sorting-for-entry" and "sorting-by-retention," and suggest that multimerization/aggregation of a cargo of secretory proteins in a mildly acidic, cation-rich environment is a nexus to both processes (3,4). Sorting-for-entry proposes that a structural/conformational motif within an aggregate of secretory proteins mediates subsequent association to a protein receptor or a lipid component of the membrane of the trans-Golgi network (TGN). The quest for identifying a consensus sorting signal has proven elusive, and it now appears that such process may rather depend on a variety of motifs (5-7), including hydrophobic disulfide-bonded loop structures (8 -12) or amphipathic ␣-helices (13)(14)(15)(16). In the sorting-by-retention model, selective aggregation/condensation occurs in post-TGN immature secretory granules (ISGs), resulting in retention of regulated secretory proteins, whereas nonaggregated proteins are removed into constitutive-like secretory vesicles (3). These two models might not be mutually exclusive. Indeed, sorting-by-retention may perhaps be viewed as an auxiliary condensation mechanism refining the segregation process in a more distal compartment of the secretory pathway, which is the maturing ISG.
The chromogranins/secretogranins (granins) family of regulated secretory proteins are widely distributed in the core of DCGs throughout the neuroendocrine and neuronal systems (17). Index members include chromogranin A (CgA), chromogranin B (CgB), and secretogranin II (SgII). Granins function as prohormones (17), and there is evidence that CgA, and perhaps other granins, mediates the biogenesis of DCGs (18 -21). Earlier reports have investigated the presence of specific DCG sorting motifs within CgA and CgB (8, 10 -13), such as a conserved N-terminal hydrophobic disulfide-bonded loop (8, 10 -13). Although such structure appears necessary and sufficient for sorting of CgB (8,10), DCG targeting of CgA might require an additional amphipathic ␣-helix (13). Whether one or many discrete regions in primary structure mediate the vesicular targeting of SgII in the primary sequence of the protein remains undetermined and might not merely be inferred from studies of CgA or CgB trafficking. Indeed, SgII shows little homology with CgA and CgB, with respect to primary structure and genomic organization (17). Most notably, SgII lacks the N-terminal hydrophobic disulfide-bonded loop structure found in CgA and CgB.

Cell Culture
Pheochromocytoma PC12 cells and primary chromaffin cells isolated from fresh bovine medulla were cultured as described previously (22,23). In some experiments, PC12 cells were treated with nerve growth factor (NGF; 2.5 S, 100 ng/ml; Invitrogen) 48 h prior to transfection and further differentiated by NGF for an additional 48 h.

DNA Transfection
Supercoiled plasmid DNA for transfection was grown in Escherichia coli strain DH5␣ (Invitrogen) and purified on columns (Qiagen). In nucleofection experiments (Amaxa), plasmid DNA was further purified using the MiraCLEAN endotoxin removal kit (MirusBio). Two days before transfection, PC12 cells were split onto either poly-L-lysine (Sigma) plus collagen (Upstate)-coated 15-mm round glass coverslips (Fisher number 1) in 12-well Costar plates or onto poly-L-lysine plus collagen-coated 6-or 12-well Costar plates. Cells were transfected with 1 g (12-well plate) or 2 g (6-well plate) of DNA per well using the cationic lipid transfection reagent GenePorter 2 (Genlantis). Five hours after the beginning of the transfection, the culture medium was replaced, and cells were further cultured for 48 h. Nucleofection was performed using 2 ϫ 10 6 bovine chromaffin cells resuspended in 100 l of Basic Nucleofector solution for primary neurons, 3 g of DNA, and program A-33. Immediately after transfection, 500 l of RPMI 1640 medium was added to the cells, which were allowed to recover 20 min (37°C, 5% CO 2 ), before plating (6 ϫ 10 5 cells per well) onto coated glass coverslips in 12-well Costar plates.
A truncated domain (EAP) of the full-length human secreted EAP (SEAP) gene devoid of 17-amino acid signal peptide of SEAP was obtained as described previously (25). To design SgII domain-EAP fusion chimeras, EGFP was substituted by EAP at the KpnI and NotI cloning sites of each of the SgII domain/ EGFP plasmids described above. pCMV-SgII-(254 -304)-EAP and pCMV-SgII-(1181-1225)-EAP were obtained by inserting EAP into the BamHI and NotI cloning sites of EGFP. pECFP-Golgi (Clontech) encoding the membrane-anchoring region of glycoprotein ␤1,4-galactosyltransferase (GalT-CFP) was used to design pECFP-Golgi-SgII. The ochre codon of enhanced CFP (TAA at position 1581-1583) was mutated to a serine codon (TCA) by site-directed mutagenesis (QuikChange, Stratagene) to produce mutated pECFP-Golgi. Next, a human SgII domain lacking SIG was amplified from pCMV-SgII-EGFP using specific primers incorporating a NotI restriction site at the 5Ј end and a TAA ochre codon followed by a NotI restriction site at the 3Ј end. The amplified fragment was purified, digested with NotI, and subcloned in-frame into the same site of mutated pECFP-Golgi to produce pECFP-Golgi-SgII encoding GalT-CFP-SgII. All the constructs were verified by restriction and nucleotide sequence analysis.

Chimeric Photoprotein Fluorescence and Immunocytochemistry
Transfected cells cultured on poly-L-lysineand collagencoated glass coverslips were fixed for 1 h at room temperature with 2% paraformaldehyde in PBS, pH 7.4, permeabilized for 10 min with 0.1% Triton X-100 in PBS, and exposed to 1 g/ml nucleic acid stain Hoechst 33342 (Molecular Probes) for nuclei visualization. Prior to fixation, NGF-differentiated PC12 cells were incubated for 30 min at 37°C with 12.5 M CellTracker Orange CMRA (Molecular Probes) in serum-free medium to visualize cell bodies and neurites and washed for 30 min at 37°C in complete medium. For immunocytochemistry, fixed cells were permeabilized for 10 min with 0.5% Triton X-100 in PBS, incubated for 5 min with 150 mM glycine in PBS, and exposed for 30 min to 5% fetal bovine serum in PBS. Cells were then incubated for 90 min at room temperature with goat polyclonal anti-CgB (C19, 1:100; Santa Cruz Biotechnology), rabbit polyclonal anti-human SgII (1:2000) (26), or rabbit polyclonal antihuman placental alkaline phosphatase antibody (1:50; Biomeda) in PBS containing 1% bovine serum albumin. Cells were subsequently washed and incubated for 30 min with a Alexa Fluor 594-conjugated donkey anti-goat IgG (1:350; Molecular Probes) or goat anti-rabbit IgG (F(abЈ) 2 at 1:250; Molecular Probes) together with 1 g/ml Hoechst 33342 in PBS containing 1% bovine serum albumin. Coverslips were washed with PBS, mounted in buffered Celvol (Celanese), and processed for three-dimensional imaging by deconvolution fluorescence microscopy.

Three-dimensional Imaging by Deconvolution Fluorescence Microscopy
Images were captured on a DeltaVision deconvolution microscopy system operated by SoftWoRx software (Applied Precision), using oil immersion objectives (60 or 100ϫ, N.A. 1.4). The system included a Photometrics CoolSNAP HQ CCD camera mounted on a Nikon TE2000 fitted with a mercury arc lamp light source. Pixel intensities were kept in the linear response range of the digital camera (Յ4095). Optical xy sections along the z axis were acquired with increments of 0.2 m. The fluorescent data sets were deconvolved to generate optical sections and three-dimensional images of the data sets. Adobe Photoshop 7.0 or ImageJ were used for additional processing of the images.

Analyses of the Images Data Sets
Quantification of Fluorescence Colocalization-To assess quantitatively the extent of colocalization between SgII fragment/GFP fusion proteins, DCG marker CgB, or CFP chimeras, 9 -25 optical xy sections along the z axis of 3-6 cells were analyzed using the pixel-per-pixel colocalization algorithm of ImageMaster 5.0 imaging software (Photon Technology International). A scatter plot was generated for each xy section and represented the intensity of the green channel (x coordinate, GFP signal) versus the intensity of the red or blue channel (y coordinate, Alexa Fluor 594 or CFP signal) for each pixel of the section. Based on the two-dimensional histogram, the software calculates the Pearson's correlation coefficient (R p ), which is mathematically related to Pearson's 2 test, and the overlap coefficient (R o ). R p and R o are quantitative of the degree of overlap between two signals S1 and S2. R p is calculated according to Equation 1, where Ϫ1 Յ R p Յ1 and 0 Յ R o Յ1. A value of 1 stands for complete positive correlation between the S1 and S2. R o is not sensitive to the differences in pixel intensity between S1 and S2. S1 i is signal intensity in the first channel; S1 avg is average intensity in the first channel; S2 i is signal intensity in the second channel; S2 avg is average intensity in the second channel. The averaged R p and R o values of each cell were used for statistical analysis of variance by ANOVA with Dunnett's post test. Differences between correlation within each data group of Table 1 were also analyzed by 2 for a row by column contingency table using pixel counts from a representative xy optical section taken in the mid-cell region. Differences were considered significant when p Ͻ 0.05.

Quantification of Fluorescence Distribution
To assess quantitatively the distribution of SIG-SgII-GFP puncta in PC12 cells coexpressing GalT-CFP-SgII chimera, we analyzed the abundance and size of green fluorescent particles using ImageJ. For each xy section, green puncta that contained above-threshold green pixel values within a specified area range of 0 -1.15 m 2 (i.e. a diameter range of 0 -600 nm) were outlined and counted, and their diameter was measured. The average diameter was calculated over the total number of identified particles for each condition as follows: SIG-SgII-GFP and GalT-CFP, 2814 counts; SIG-SgII-GFP and GalT-CFP-SgII, 997 counts. A statistical analysis of variance was done by t test.

Chemiluminescence Detection of EAP Secretion Assay
PC12 cells grown onto poly-L-lysine-coated plus collagencoated 6-or 12-well plates were transiently transfected with an expression plasmid for SgII domain/EAP chimeras or SEAP. Cells were washed twice with calcium saline buffer (CaSB: 150 mM NaCl, 5 mM KCl, 2 mM CaCl 2 , and 10 mM HEPES, pH 7.4) and subsequently exposed for 15 min to CaSB or barium secretion buffer (BaSB: 150 mM NaCl, 5 mM KCl, 2 mM BaCl 2 , and 10 mM HEPES, pH 7.4). Supernatants were collected, and cells lysates were prepared by quick freeze/thaw in CaSB containing 0.1% Triton X-100. Detection of EAP enzymatic activity in the culture supernatant and cell lysate was achieved with a high sensitivity chemiluminescence assay (Phospha-Light, Applied Biosystems). The secretion rate of EAP chimeras was calculated as a percentage of the total EAP activity present in the cells before stimulation. Total EAP activity is the sum of the amount released plus the amount remaining in the cells. In the figures, the release of EAP chimeras is expressed as percent EAP activity secretion, or relative to EAP secretion rate in nonstimulating condition (CaSB, mock control).

Secondary Structure Prediction
The secondary structure of mature human SgII was predicted with NNPREDICT (27) and further analyzed with the NPS@ consensus secondary structure prediction program (28) that includes 11 secondary structure prediction algorithms as follows: DPM, DSC, GOR1, GOR3, GOR4, HNNC, PHD, Predator, SIMPA96, SOPM, and SOPMA. Helical wheel projection was accomplished using the PEPWHEEL software of the EMBOSS application package (29).

Presentation of Data and Statistics
Values are given as the mean Ϯ S.E. for at least triplicate determinations. In the figures, data are representative of a typical experiment repeated independently two or more times. The number of independent experiments or analyses (n) per-   Fig. 9, nՆ 3; Fig. 10, n Ն 2. Statistical analysis was performed by t test, or ANOVA with Dunnett's or Bonferroni post test, or by 2 for a row by column contingency table. Differences were considered significant when p Ͻ 0.05.

A Human SgII-GFP Chimeric Photoprotein Localizes to Dense-core Secretory Granules in Sympathoadrenal Cells
Intracellular Distribution in Sympathoadrenal PC12 Cells and Primary Chromaffin Cells-We designed expression plasmids encoding the predicted 30-amino acid signal peptide of human SgII ( Ϫ30 MAEAKTHWLGAALSLIPLIFLISGAE-AASF Ϫ1 , or SIG; pCMV-SgII181-EGFP, see Fig. 1), as defined in Ref. 24, or full-length human SgII (including SIG; pCMV-SgII-EGFP, see Fig. 1) fused in-frame at their C termini to the N-terminal of enhanced GFP (EGFP). The subcellular distribution of the resulting chimeric photoproteins was first examined by three-dimensional deconvolution fluorescence microscopy after transient expression in PC12 cells and primary bovine chromaffin cells ( Fig. 2A). SIG-GFP accumulated in the perinuclear region of the cells, consistent with the localization of the chimera in the Golgi apparatus ( Fig. 2A). Indeed, quantification in three dimensions of the colocalization between SIG-GFP and the trans-Golgi/TGN resident glycoprotein ␤1,4-galactosyltransferase fused to cyan fluorescent protein (GalT-CFP), in sets of 200-nm xy image stacks, revealed high overlap in their distribution (Fig. 5A), with a Pearson's correlation coefficient R p of 0.73 Ϯ 0.07 and an overlap coefficient R o of 0.74 Ϯ 0.07 between the two molecules (n ϭ 27; Table 1). In addition, the subcellular distribution of SIG-GFP in PC12 and primary chromaffin cells ( Fig. 2A) was remarkably similar to that of the constitutive secretory pathway markers SgP-GFP (where SgP is the predicted 18-amino acid signal peptide of human CgA) (13) and SEAP (21,25) in the sympathoadrenal cell lineage, which further indicates that SIG-GFP is likely routed to the constitutive branch of the secretory pathway.
In sharp contrast, the distribution of full-length SgII fused to GFP (SIG-SgII-GFP) in PC12 cells and primary chromaffin cells was highly punctate ( Fig. 2A), suggesting storage of the chimera in secretory granules. Examination of individual 200-nm xy optical sections of SIG-SgII-GFP-expressing PC12 cells revealed a ring of subplasmalemmal vesicular fluorescence ( Fig. 2A) characteristic of the DCG distribution previously reported for other regulated secretory GFP fusion proteins after transient expression in PC12 cells (i.e. CgA-GFP, atrial natriuretic factor-GFP, or neuropeptide Y-GFP (13,30,31)). On the other hand, intracellular distribution of SIG-SgII-GFP in primary chromaffin cells appeared more uniform, with punctate fluorescence present throughout the cell body, likely produced by secretory granules ( Fig. 2A).
Colocalization with a Marker of DCGs-To further define the subcellular localization of SIG-SgII-GFP in sympathoadrenal cells, we assessed its colocalization with CgB, a classical marker Cells were transfected for 48 h with pCMV-SgII181-GFP or pCMV-SgII-GFP and examined by deconvolution microscopy. A series of xy optical sections along the z axis was acquired with increments of 0.2 m using 60 or 100ϫ oil immersion objectives (1.4 N.A.). The data set was processed to generate three-dimensional (3D)/ volume or representative xy section (0.2 m) views of the distribution of the GFP chimeras (green) (A). PC12 cells expressing SIG-GFP or SIG-SgII-GFP were processed for immunocytochemistry using a polyclonal anti-CgB primary and an Alexa Fluor 594-conjugated secondary antibody (B). The distribution of SIG-GFP or SIG-SgII-GFP (green) and the endogenous secretory granule marker CgB (red) was compared in merged images. Yellow regions are indicative of colocalization. Quantification of fluorescence overlap is reported in Table 1. Nuclei were visualized with Hoechst 33342 (blue). Scale bar, 5 m.
of neuroendocrine and neuronal secretory granules (17). Detection of endogenous CgB was achieved by immunocytochemistry in PC12 cells expressing SIG-SgII-GFP or SIG-GFP. As anticipated, immunostaining of CgB and fluorescence of SIG-SgII-GFP displayed highly punctate patterns (Fig. 2B), and the extent of colocalization between the two molecules was substantial (R p ϭ 0.39 Ϯ 0.05 and R o ϭ 0.41 Ϯ 0.05; n ϭ 121; Table 1), although incomplete, as compared with that of the maximum possible colocalization of endogenous SgII and CgB assessed by immunofluorescence (R p ϭ 0.76 Ϯ 0.02 and R o ϭ 0.75 Ϯ 0.02; n ϭ 254; Table 1; visual output not shown). On the contrary, colocalization of SIG-GFP with endogenous CgB was poor ( Fig. 2B) with R p ϭ 0.13 Ϯ 0.05 and R o ϭ 0.17 Ϯ 0.04 (n ϭ 72; Table 1), as predicted for a fusion protein primarily destined to the constitutive secretory pathway (21,25). Partial rather than absolute overlap in the distribution of SIG-SgII-GFP with endogenous CgB can best be explained by the extended half-life of DCGs, which gives rise to pools of old granules (pre-transfection; lacking SIG-SgII-GFP) and young post-transfection granules (i.e. containing SIG-SgII-GFP). Lack of CgB immunostaining of granules positive for SIG-SgII-GFP may also be the result of incomplete penetration of the antibodies into the secretory granule core. Nonetheless, such partial colocalization is consistent with values reported in studies localizing other GFP fusion proteins, such as CgA-GFP (13) or pro-atrial natriuretic factor-GFP (13,32), with endogenous markers of DCGs.
NGF Differentiation to Neurites-Redistribution of DCGs away from the peripheral/subplasmalemmal region of the cell body into the tips of neurite-type processes has been documented in PC12 cells after neuronal differentiation by NGF (13,31). PC12 cells transiently expressing SIG-GFP or SIG-SgII-GFP were exposed to NGF (100 ng/ml, 96 h), and neurite extensions were visualized with the cytosolic fluorescent dye Cell-Tracker Orange CMRA. Consistent with routing of SIG-SgII-GFP to DCGs, substantial accumulation of chimeric granin fluorescence was seen at the termini of growth cones (Fig. 3), whereas NGF treatment did not affect the juxtanuclear/Golgi apparatus distribution of SIG-GFP (Fig. 3). Taken together, these findings provide evidence that human SgII flanked by a signal peptide sequence and fused in-frame to GFP is correctly targeted into DCGs of sympathoadrenal cells.

A Human SgII-EAP Fusion Protein Is Sorted to DCGs and Released by Exocytosis in Response to Secretagogue
We previously reported that CgA fused to an engineered form of secreted EAP is trafficked to DCGs in PC12 cells and release by exocytosis (25). We reasoned that an SgII/EAP fusion protein might also become a soluble component of DCGs when expressed in PC12 cells. We designed plasmids encoding EAP fused at the C termini of full-length SgII (SIG-SgII-EAP; Fig. 1) or SgII's signal sequence SIG (SIG-EAP; Fig. 1). We first examined the subcellular distribution of SIG-EAP and SIG-SgII-EAP by immunofluorescence microscopy after transient expression in PC12 cells (Fig. 4A). Both SIG-EAP and SIG-SgII-EAP chimeras showed an intracellular distribution similar if not identical to their GFP-labeled counterparts. Thus, SIG-EAP exhibited a perinuclear distribution highly indicative of a Golgi apparatus localization, although the punctate pattern of SIG-SgII-EAP was likely because of storage of the recombinant protein into DCGs. Consistent with visual inspection, the extent of colocalization between endogenous CgB and SIG-EAP was poor (R p ϭ 0.19 Ϯ 0.02 and R o ϭ 0.25 Ϯ 0.02; n ϭ 85; Table 1),

Colocalization between endogenous DCG markers SgII and CgB are shown (A). Colocalization between SgII domain/GFP (B) or SgII domain/EAP (C) fusion proteins and endogenous DCG marker CgB is shown. Colocalization between SgII domain-GFP chimeras and trans-Golgi/TGN-retained CFP chimeras GFP is shown (D)
. xy optical sections were acquired with increments of 0.2 m along the z axis for the indicated number of cells. The degree of colocalization between fluorescent molecules was measured on a pixel-per-pixel basis for the two wavelength sources by comparing the position of each pixel in each of the 0.2-m xy image of a given z stack. R p and R o were determined using the colocalization algorithm of ImageMaster image processing software, as described under "Experimental Procedures." Statistical significance of correlation within B, C, and D data sets was also tested by 2 for a 2 ϫ 4 contingency table, using pixel counts from a representative xy optical section taken in the mid-cell region. Group B, c 2 ϭ 5925, p Ͻ 0.001; group C, 2 ϭ 6548, p Ͻ 0.001; group D, c 2 ϭ 1642, p Ͻ 0.001. whereas SIG-SgII-EAP showed substantial overlap with CgB (R p ϭ 0.49 Ϯ 0.05 and R o ϭ 0.52 Ϯ 0.05; n ϭ 95; Table 1; visual output not shown).
To further ascertain the localization of SIG-SgII-EAP into the lumen of DCGs, we measured secretion of SIG-SgII-EAP and SIG-EAP from transfected PC12 cells stimulated with Ba 2ϩ . Secretion of SIG-SgII-EAP was low (2.8 Ϯ 0.2%) under basal, nonstimulating condition, consistent with efficient storage of the fusion protein within DCGs ( Fig. 4B and Table 1). Ba 2ϩ (2 mM, 15 min) triggered 39.5 Ϯ 1.8% secretion of SIG-SgII-EAP into the extracellular medium, which represented a highly significant ϳ14.3-fold increase over basal (p Ͻ 0.0001; Fig. 4B), clearly demonstrating regulated trafficking of the chimera. In contrast, the release of the constitutive secretory protein SEAP (11,25) and that of SIG-EAP was elevated (8.6 -8.9%), even in the absence of secretagogue (Fig. 4B), and was slightly augmented to 15.2-16.8% in response to secretagogue.
Although such a secretory profile might suggest some degree of nonselective entry of SEAP and SIG-EAP from the constitutive into a regulated pathway of secretion, as suggested previously for SEAP in PC12 cells (26), the marginal increase in relative stimulated release (ϳ1.7-1.8-fold) of the chimeras was not significant as compared with control (mock) treatment (p Ͼ 0.05; Fig. 4B), suggesting that SEAP and SIG-EAP may primarily traffic through the constitutive pathway of secretion.
Thus, these results indicate that a SgII/EAP chimera may be expressed in PC12 cells where it is targeted to DCGs and released by exocytosis in response to secretagogue. We show

. Trafficking of SIG-EAP and SIG-SgII-EAP chimeras in PC12 cells.
A, subcellular distribution of SIG-EAP and SIG-SgII-EAP. PC12 cells transfected with pCMV-SgII181-EAP or pCMV-SgII-EAP were processed for immunocytochemistry using an anti-human placental alkaline phosphatase polyclonal primary and Alexa Fluor 594-conjugated secondary antibodies and analyzed by deconvolution microscopy. B, chemiluminescence detection of secretagogue-evoked secretion of EAP chimeras. Cells transiently expressing SEAP, SIG-EAP, or SIG-SgII-EAP were exposed for 15 min to secretion medium alone (mock) or to 2 mM Ba 2ϩ . EAP secretion was calculated relative to total enzymatic activity present in the cells before stimulation. Total enzymatic activity is the sum of the amount released plus the amount remaining in the cells. Release of EAP is expressed either as % EAP activity secretion or relative to enzymatic activity release in the absence of secretagogue. Values are given as the means Ϯ S.E. of triplicate determinations. †, p Ͼ 0.05; ***, p Ͻ 0.0001 as compared with control (mock), ANOVA with Dunnett's post test. Scale bar, 5 m. that fluorescent and EAP-mediated chemiluminescence SgII chimeras may be used as tracers of the regulated secretory pathway in living cells, providing both qualitative (GFP) and quantitative (EAP) means to study the mechanisms by which SgII is sorted into DCGs and to determine whether such a process requires discrete regions within the primary structure of the molecule.

Dominant Inhibitory Effect of a trans-Golgi/TGN Resident Form of SgII on the Targeting of Soluble SgII to DCGs
On the premise that sorting of SgII into the regulated pathway may depend on aggregative signals and/or recognition signals, we questioned whether a mutant form of SgII designed to be selectively retained within the distal regions of the Golgi apparatus might compromise targeting of a soluble cargo of SgII to DCGs. We fused the C-terminal end of the trans-Golgi/ TGN retention signal of ␤1,4-galactosyltransferase (GalT) to the N terminus of full-length SgII (excluding its signal sequence SIG) to create GalT-CFP-SgII (Fig. 1). Analysis of the subcellular distribution of GalT-CFP-SgII revealed a cluster of CFP fluorescence in a perinuclear region of PC12 cells characteristic of the Golgi apparatus (Fig. 5). The distribution of GalT-CFP-SgII was virtually identical to that of SIG-GFP, as shown by fluorescence microscopy image overlays of volume views and 0.2-m xy optical sections (Fig. 5A). The extent of the overlap between GalT-CFP-SgII and SIG-GFP was high, with colocalization parameters  Table 1). These results provide evidence that SgII fused to the membrane-anchoring region of GalT is efficiently retained/retrieved within the lumen of the trans-Golgi/TGN.
We assessed the influence of GalT-CFP-SgII on the trafficking of SIG-SgII-GFP in PC12 cells. First, we examined the distribution of SIG-SgII-GFP in cells coexpressing GalT-CFP. As expected, SIG-SgII-GFP showed a typical DCG localization, whereas GalT-CFP accumulated in the trans-Golgi/TGN region of the cells (Fig. 5B). Consequently, the extent of colocalization between SIG-SgII-GFP and GalT-CFP was poor, as judged by visual inspection of the image overlays (Fig. 5B) and determination of Pearson's and overlap coefficients (R p ϭ 0.15 Ϯ 0.06 and R o ϭ 0.17 Ϯ 0.06; see Table 1). In sharp contrast, expression of SIG-SgII-GFP together with GalT-CFP-SgII caused substantial redistribution of SIG-SgII-GFP to the perinuclear region of the cells (Fig. 5B). In such conditions, SIG-SgII-GFP showed high colocalization with GalT-CFP-SgII (R p and R o ϭ 0.77 Ϯ 0.03; see Fig. 5B and Table 1), indeed suggesting retention of SIG-SgII-GFP into the trans-Golgi/ TGN in the presence of GalT-CFP-SgII. To further probe the dominant inhibitory effect of the trans-Golgi/TGN-resident form of SgII, we also quantified the size and abundance of SIG-SgII-GFP puncta in three dimensions. As shown in Fig. 5C, the size of fluorescent SIG-SgII-GFP puncta was similar in either GalT-CFP-SgII-or GalT-CFP-expressing PC12 cells, with an average diameter of ϳ308 nm. Considering that the resolution of fluorescence puncta containing the photoprotein is diffraction-limited, this value is consistent with our earlier analyses by electron microscopy, reporting a DCG diameter of ϳ100 -130 nm in PC12 cells (21,25). The number of SIG-SgII-GFP puncta per xy optical section was significantly reduced in cells coexpressing GalT-CFP-SgII (22 Ϯ 4 puncta/xy plan, n ϭ 43; p Ͻ 0.05; Fig. 5C), as compared with PC12 cells expressing SIG-SgII-GFP and GalT-CFP (66 Ϯ 14 puncta/xy plan; n ϭ 43; Fig.  5C). Thus, these results clearly suggest that transient expression of GalT-CFP-SgII may reduce targeting of coexpressed SIG-SgII-GFP to DCGs, while promoting its retention within the trans region of the Golgi apparatus.
We also analyzed the secretory profile of SIG-EAP and SIG-SgII-EAP in the presence of GalT-CFP or GalT-CFP-SgII (Fig.  5D). The release of SIG-EAP was unchanged in the presence of either photoproteins and was not significantly stimulated by 2 mM Ba 2ϩ (p Ͼ 0.05; Fig. 5D), consistent with trafficking of SIG-EAP into the constitutive branch of the secretory pathway (Fig.  4B). In contrast, expression of SIG-SgII-EAP together with GalT-CFP-SgII lowered Ba 2ϩ -evoked secretion of SIG-SgII-EAP by ϳ48% (p Ͻ 0.001; Fig. 5D), as compared with cells coexpressing GalT-CFP. It is noteworthy that expression of GalT-CFP (lacking SgII) did not reduce the vesicular localization of coexpressed SIG-SgII-GFP (Fig. 5B) nor Ba 2ϩ -stimulated secretion of SIG-SgII-EAP (Figs. 4B and 5D). This clearly establishes that impaired regulated trafficking of soluble SgII chimeras specifically results from the expression of a trans-Golgi cisternae membrane-anchored form of SgII and is not merely a consequence of protein overexpression that could, for instance, exacerbate GFP and CFP susceptibility to self-associate/cluster (33).
Altogether, these findings provide evidence in cella that a chimeric form of SgII designed to be selectively retained/recycled to the Golgi apparatus may act as dominant negative, impairing the trafficking of coexpressed, secreted forms of SgII (i.e. SIG-SgII-GFP or SIG-SgII-EAP) from the lumen of the trans-Golgi/TGN into the regulated secretory pathway.

DCG Targeting of N-terminal and C-terminal Half-domains of SgII
Earlier studies in PC12 cells have identified sorting determinants for the regulated secretory pathway within the N-terminal region of CgA and CgB (8,11,13,25,34). We questioned whether a signal for the vesicular targeting of SgII might also locate within a discrete region of the primary structure of granin.
SgII-EAP (ϳ14.5-fold over basal; Fig. 6B). Taken together, these results demonstrate that both the N-terminal and C-terminal half-regions of SgII are able to steer GFP and EAP reporters into DCGs for exocytosis, suggesting that SgII contains independent sorting signals for the regulated secretory pathway.

The Domain SgII-(334 -348) Is a Sufficient Sorting Signal for Trafficking into the Regulated Secretory Pathway
To test whether the SgII-(333-348) domain might be also be a sufficient DCG-sorting determinant, we generated GFP-and EAP-tagged chimeric proteins consisting of SgII's signal peptide sequence SIG fused to residues 334 -348 (SIG-SgII-(334 -348)-GFP and SIG-SgII-(334 -348)-EAP; Fig. 1). Expression of SIG-SgII-(334 -348)-GFP or SIG-SgII-(334 -348)-EAP in PC12 FIGURE 7. Trafficking of progressively truncated N-terminal domains of SgII fused to GFP or EAP in PC12 cells. Cells transfected with expression plasmids encoding the indicated photoproteins were processed for deconvolution microscopy (A). A transition from a punctate/vesicular distribution (e.g. SIG-SgII-GFP) to a juxtanuclear localization (e.g. SIG-GFP) is seen as the N-terminal domain of SgII is reduced from ϩ41 to ϩ24 residues. B, secretagogue-evoked release of SgII domain-EAP chimeras. PC12 cells expressing the indicated EAP fusion proteins were exposed for 15 min to secretion medium alone (mock) or to 2 mM Ba 2ϩ . The enzymatic activity of EAP chimeras was assayed in the culture supernatant and cell lysate, and relative secretion was determined as described in the legend of Fig. 4. Values are given as the means Ϯ S.E. of triplicate determinations. †, p Ͼ 0.05; **, p Ͻ 0.01; ***, p Ͻ 0.001 as compared with control (mock), ANOVA with Dunnett's post test. Scale bar, 5 m. cells resulted in a highly punctate distribution of the chimeras throughout the cell body, which showed substantial overlap with endogenous CgB (Fig. 10, A and C). Quantitative analysis of such colocalization yielded an R p of 0.39 Ϯ 0.04 and an R o of 0.44 Ϯ 0.01 for SIG-SgII-(334 -348)-GFP versus CgB, and an R p of 0.67 Ϯ 0.08 and an R o of 0.72 Ϯ 0.06 for SIG-SgII-(334 -348)-EAP versus CgB (Fig. 10, A and C; Table 1). Note that these values were comparable and even exceeded the colocalization coefficients of full-length SgII-GFP (SIG-SgII-GFP) versus endogenous CgB (R p ϭ 0.39 Ϯ 0.05 and R o ϭ 0.41 Ϯ 0.05; Fig.  2B and Table 1). Accumulation of SIG-SgII-(334 -348)-GFP fluorescence at the termini of neurite processes was obvious in cells differentiated by NGF (Fig. 10B), consistent with DCG storage of the chimera. Finally, we found that SIG-SgII-(334 -348)-EAP was secreted in response to secretagogue. As shown in Fig. 10C, Ba 2ϩ triggered the release of SIG-SgII-(334 -348)-EAP from transfected PC12 cells by ϳ4-fold over basal (p Ͻ 0.001). The stimulated secretion of SIG-SgII-(334 -348)-EAP was lower than that of full-length SgII-EAP chimera (SIG-SgII-EAP; ϳ14.3-fold over basal, p Ͻ 0.001) but significantly higher than the release of SIG-EAP, which showed no significant response to Ba 2ϩ (p Ͼ 0.05; Fig. 10C). Taken together, the results provide evidence that SgII-(334 -348) ( 334 DLIEMLKTGEKP-NGS 348 ) functions as a necessary determinant, in the context of sorting of the larger SgII-(302-371) domain into the regulated pathway, and contains a sufficient (transferable) granule-targeting signal, regardless of the reporter tag (GFP or EAP).

DISCUSSION
Understanding the molecular mechanisms by which secretory proteins are sorted into DCGs has been the subject of intensive research for many years, and it underscores the remarkable heterogeneity of this process (6,7). This study  (Table 1). SIG-SgII-(25-41)-EAP was stained using a polyclonal anti-human placental alkaline phosphatase primary and an Alexa Fluor 488conjugated secondary antibodies (B). Yellow regions are indicative of colocalization between SIG-SgII-(25-41)-EAP (green) and CgB (red). Quantification of fluorescence overlap is reported in Table 1. C, regulated secretion of SIG-SgII-(25-41)-EAP. PC12 cells transiently expressing SIG-SgII-EAP or SIG-SgII-(25-41)-EAP were incubated for 15 min in secretion medium alone (mock) or 2 mM Ba 2ϩ . The enzymatic activity of EAP chimeras was assayed in the culture supernatant and cell lysate, and relative secretion was determined as described in the legend of defines sequence features of the secretory prohormone SgII that are involved in the targeting of the molecule to DCGs in sympathoadrenal cells. We find that information necessary for such a process is dependent upon a saturable sorting machinery within the lumen of the trans-Golgi/TGN, and recruitment of SgII-(25-41) and SgII-(334 -348) regions, which are predicted to contain ␣-helical structural motifs that may function as sufficient, independent sorting domains for the regulated pathway.
Trafficking of SgII Fusion Proteins into the Regulated Secretory Pathway-Taking our cue from the experimental paradigm developed in our earlier studies on the vesicular trafficking of CgA in sympathoadrenal cells (13,21,25), we fused full-length human SgII to GFP or EAP, for fluorescent or enzymatic tracing of the regulated secretory pathway. This study establishes that SgII flanked by a signal sequence is able to convert GFP or EAP to soluble (releasable) components of DCGs, and it thus reveals that SgII contains a dominant sorting signal for the regulated secretory pathway, whereas its signal peptide alone does not (Figs. 2-7, 9, and 10). We find that the intracellular distribution and trafficking of SIG-SgII-GFP and SIG-SgII-EAP may reflect the behavior of endogenous regulated secretory proteins, providing unique tools to study qualitatively FIGURE 9. Trafficking of progressively truncated C-terminal domains of SgII fused to GFP or EAP in PC12 cells. PC12 cells expressing the indicated photoproteins were processed for deconvolution microscopy (A). A transition from a punctate/vesicular distribution (e.g. SIG-SgII-GFP) to a juxtanuclear localization (e.g. SIG-GFP) is seen as the N-terminal domain of SgII is reduced from ϩ348 to ϩ333 residues. B, secretagogue-evoked release of SgII domain-EAP chimeras. PC12 cells expressing the indicated EAP fusion proteins were exposed for 15 min to secretion medium alone (mock) or to 2 mM Ba 2ϩ . The enzymatic activity of EAP chimeras was assayed in the culture supernatant and cell lysate, and relative secretion was determined as described in the legend of Fig. 4. Values are given as the means Ϯ S.E. of triplicate determinations. †, p Ͼ 0.05; *, p Ͻ 0.05; ***, p Ͻ 0.001 as compared with control (mock), ANOVA with Dunnett's post test. Scale bar, 5 m. . Cells transfected with pCMV-SgII-(1181-1225)-GFP were processed for immunocytochemistry using a polyclonal anti-CgB primary and an Alexa Fluor 594-conjugated secondary antibodies, followed by imaging by deconvolution microscopy. Yellow regions indicate an overlap in the distribution of SIG-SgII-(334 -348)-GFP (green) and endogenous secretory granules marker CgB (red), and the extent was determined using ImageMaster 5.0 (Table 1). B, accumulation of SIG-SgII-(334 -348)-GFP puncta along and at the tip of neurite processes. NGF-differentiated PC12 cells were transiently transfected with pCMV-SgII-(1181-1225)-GFP and labeled with CellTracker Orange CMRA dye (red) prior to fixation to visualize the cell body and neurite processes. Deconvolution microscopy imaging reveals a substantial accumulation of SIG-SgII-(334 -348)-GFP chimera at the tip of neurites, as shown in the enlarged inset. C, colocalization between SIG-SgII-(334 -348)-EAP and secretory granule marker CgB. Cells expressing SIG-SgII-(334 -348)-EAP were processed for immunocytochemistry using polyclonal anti-CgB and anti-human placental alkaline phosphatase primary and Alexa Fluor 594-conjugated and 488-conjugated secondary antibodies. Quantification of fluorescence overlap is reported in Table 1. D, regulated secretion of SIG-SgII-(334 -348)-EAP. PC12 cells expressing SIG-EAP, SIG-SgII-EAP, or SIG-SgII-(334 -348)-EAP were incubated for 15 min in secretion medium alone (mock) or 2 mM Ba 2ϩ . The enzymatic activity of EAP chimeras was assayed in the culture supernatant and cell lysate, and relative secretion was determined as described in the legend of Fig. 4. Values are given as the means Ϯ S.E. of triplicate determinations. †, p Ͼ 0.05; ***, p Ͻ 0.001 as compared with control (mock), ANOVA with Dunnett's post test. Nuclei are stained with Hoechst 33342 (blue). Scale bar, 5 m. and quantitatively the intracellular fate of SgII (or a domain therein) and characterize the targeting mechanism of the protein in sympathoadrenal cells.
Two Independent, Minimum Sorting Domains within the Nand C-terminal Region of SgII-Peptide sequences acting as independent sorting motif for the regulated secretory pathway have been documented for several regulated proteins. A prime example is proinsulin, in which DCG-targeting determinants may include a binding domain for carboxypeptidase E (CPE), PCs cleavage sites, and aggregation domains (35,36). Similarly, two distinct peptide sequences in the C-terminal region of PC1 may function as sufficient sorting signals to reroute a constitutively secreted IgG Fc fragment to DCGs in pituitary cells (15). Sorting of CgA into DCGs may rely on cell-specific mechanisms mediated by distinct sorting domains. Studies in PC12 cells have shown that information for sorting of CgA into the regulated pathway may be confined within the N-terminal region of the molecule that contains a hydrophobic disulfide-bonded loop and an ␣-helical motif (11,13,25), whereas only a C-terminal domain appears necessary for DCG targeting of CgA in pituitary cells (11). Our results point to the presence of multiple sorting signals, but with two discrete regions SgII- (25)(26)(27)(28)(29)(30)(31)(32)(33)(34)(35)(36)(37)(38)(39)(40)(41) and SgII-(334 -348), functioning independently for sorting of SgII within the same cell type.
What might be the physiological significance of two domains acting independently to traffic SgII within the regulated pathway of PC12 cells? An intriguing parallel may be drawn from the vesicular targeting of the ELH prohormone in Aplysia californica neurons, in which peptides derived from proteolytic processing of ELH are sorted into separate DCGs, suggesting the existence of independent sorting domains within the prohormone (37). Thus, distinct structural motifs within a prohormone precursor may function independently in the segregation of processing intermediates, thereby determining their routing through distal compartments of the secretory pathway. SgII is also a precursor protein that can be proteolytically processed by SgII-(334 -348) (B) domains and the surrounding region are shown. Identical amino acids are highlighted by gray shading. Secondary structure of human SgII was predicted with NNPREDICT and further analyzed with the NPS@ consensus secondary structure prediction program that includes the algorithms DPM, DSC, GOR1, GOR3, GOR4, HNNC, PHD, Predator, SIMPA96, SOPM, and SOPMA. Consensus secondary structure prediction is indicated below the multiple sequence alignment of each region. Also shown is the helical wheel projection (PEPWHEEL) of predicted ␣-helices within SgII- (25)(26)(27)(28)(29)(30)(31)(32)(33)(34)(35)(36)(37)(38)(39)(40)(41) and SgII-(334 -348). The hydrophobic face of the ␣-helices is highlighted with light gray shading, and hydrophobic amino acids residues are indicated in black boxes.
PCs at dibasic amino acid cleavage sites into smaller peptides with important biological function. For instance, secretoneurin (SgII-(152-184)) acts as a chemotactic/angiogenic cytokine in vivo, with activity comparable in potency to vascular endothelial growth factor (38,39). Whether full-length SgII and/or processing intermediates are uniformly distributed, or localize to distinct populations of secretory granules, remains an open question (40), and it is also unclear if proteolytic processing of SgII begins in the TGN prior to sorting or in later compartments of the regulated secretory pathway, for instance ISGs (41).
Integration of the Results and Putative Mechanisms-Considering a mechanism based on a sorting-for-entry model, we propose that particular regions of SgII, for instance SgII- (25)(26)(27)(28)(29)(30)(31)(32)(33)(34)(35)(36)(37)(38)(39)(40)(41) and SgII-(334 -348), act as discrete sorting signals binding to particular lipid components of the membrane of the TGN or the nascent secretory granule. Consistent with this hypothesis, SgII may interact specifically with isolated vesicle membranes at a mildly acidic pH (44), a property shared by several secretory granule cargo proteins, including CgA and CgB (45,46). A mechanism recruiting a specific sorting receptor at the membrane of the TGN may also be considered. For instance, CPE is thought to act as a common sorting receptor for proopiomelanocortin, proinsulin, and proenkephalin (35,47). Earlier studies have excluded a role of CPE in the sorting of CgA (35,48), but new evidence suggests that a sorting complex consisting of CPE and SgIII mediates targeting of several prohormones in endocrine cells (49). However, interaction of SgII with SgIII has not been found (50), and whether CPE may contribute to the sort-ing process of SgII has yet to be determined. Another candidate sorting receptor is the inositol 1,4,5-trisphosphate receptor (IP3R). CgA and CgB have been documented to bind IP3R in vitro at the intravesicular pH of 5.5 and modulate the channel activity in living cells along several compartments of the secretory pathway, including DCGs (51,52). Hence, the pH-dependent and functional interaction of IP3R with CgA and CgB raises the possibility of such a receptor serving as a membrane target for the vesicular sorting of other calcium storage proteins such as SgII.
Although our findings provide evidence that SgII-(25-41) and SgII-(334 -348) domains contain sorting information sufficient to redirect GFP and EAP into the regulated pathway, whether these two regions uniquely contribute to the sorting of SgII within the regulated pathway remains to be seen. Indeed, our results from secretagogue-stimulated release of SgII domain-EAP chimeras show a positive correlation between the sorting efficiency and the length of the SgII domain under test, i.e. the longer the domain, the better the sorting efficiency, with the highest efficiency achieved for full-length SgII (Figs. 6, 7, and 9). This secretory behavior does not prove but certainly points to a substantial contribution of other regions of the SgII to achieve optimal sorting efficiency, perhaps by promoting the ability of SgII to aggregate/multimerize within the TGN or ISGs and enhance segregation by a sorting-by-retention mechanism. One example of a synergistic action between sorting signals might be that of CgB. As outlined in the Introduction, an N-terminal disulfide-bonded loop domain acts as a transferable sorting determinant (8 -10) but does not contribute to the aggregation/multimerization of the granin in PC12 cells (34). However, this sorting domain may not be functional in the absence of endogenous CgB, which underlies the contribution of aggregation in the sorting process of CgB (10).
Our results from the trans-Golgi resident GalT-CFP-SgII chimera are consistent with a contribution of aggregative signals to the sorting of SgII (Fig. 5). One possibility is that membrane-anchored SgII might compete with a multimerization/ aggregation step common to both models for sorting, behaving as a bait for the nucleation and growth of an aggregate of preysoluble SgII, and recycle such aggregate within the trans-Golgi/ TGN, thereby impairing further trafficking into the regulated secretory pathway. In the simplest scenario, membrane-anchored SgII would homomerize with endogenous and/or exogenous soluble SgII (GFP-or EAP-labeled), which is consistent with studies in vitro reporting pH-and Ca 2ϩ -regulated selfaggregation of SgII (24,44). However, the situation could be more complex, with a role of heteromeric association with other cargo proteins to the overall multimerization/aggregation process. For instance, CgA and CgB form homo-as well as hetero-aggregates in vitro in a mildly acidic pH and millimolar Ca 2ϩ environment (53). However, heteromeric interaction of SgII with other granins has yet to be reported in sympathoadrenal cells, although studies in pituitary cells showed no interaction of SgII with CgA or secretogranin III (SgIII) (50). An equally plausible explanation is that GalT-CFP-SgII may compete out binding of soluble SgII to a putative membrane receptor at the TGN, leading to the accumulation of soluble SgII within this compartment. Indeed, sorting of SgII via a sorting-for-entry mechanism is expected to be limited by the number of receptors and therefore saturable.
In conclusion, this work elucidates some of the molecular details that underlie sorting of SgII within the regulated secretory pathway of sympathoadrenal cells. We find that discrete domains within the primary structure of SgII act as sufficient sorting signals, perhaps by virtue of aggregation/multimerization properties and/or by interaction with a specific component of the membrane of the budding secretory granule. The ability of the TGN-resident SgII chimera to readily saturate the sorting machinery at the trans-Golgi/TGN may provide essential information for future attempts to identify protein-protein interaction and/or putative receptor that mediate efficient and optimal sorting of SgII into DCGs.