Nerve Growth Factor-dependent Sorting of Synaptotagmin IV Protein to Mature Dense-core Vesicles That Undergo Calcium-dependent Exocytosis in PC12 Cells*

Synaptotagmin IV (Syt IV) is a fourth member of the Syt family and has been shown to regulate some forms of memory and learning by analysis of Syt IV null mutant mice (Ferguson, G. D., Anagnostaras, S. G., Silva, A. J., and Herschman, H. R. (2000) Proc. Natl. Acad. Sci. U. S. A. 97, 5598–5603). However, the involvement of Syt IV protein in vesicular trafficking and even its localization in secretory vesicles are still matters of controversy. Here we present several lines of evidence showing that the Syt IV protein in PC12 cells is normally localized in the Golgi or immature vesicles at the cell periphery and is sorted to fusion-competent mature dense-core vesicles in response to short nerve growth factor (NGF) stimulation. (i) In undifferentiated PC12 cells, Syt IV protein is mainly localized in the Golgi and small amounts are also present at the cell periphery, but according to the results of an immunocytochemical analysis, they do not colocalize with conventional secretory vesicle markers (Syt I, Syt IX, Rab3A, Rab27A, vesicle-associated membrane protein 2, and synaptophysin) at all. By contrast, limited colocalization of Syt IV protein with dense-core vesicle markers is found in the distal parts of the neurites of NGF-differentiated PC12 cells. (ii) Immunoelectron microscopy with highly specific anti-Syt IV antibody revealed that the Syt IV protein in undifferentiated PC12 cells is mainly present on the Golgi membranes and immature secretory vesicles, whereas after NGF stimulation Syt IV protein is also present on the mature dense-core vesicles. (iii) An N-terminal antibody-uptake experiment indicated that Syt IV-containing vesicles in the neurites of NGF-differentiated PC12 cells undergo Ca2+-dependent exocytosis, whereas no uptake of the anti-Syt IV-N antibody was observed in undifferentiated PC12 cells. Our results suggest that Syt IV is a stimulus (e.g. NGF)-dependent regulator for exocytosis of dense-core vesicles.

Synaptotagmin (Syt) 1 is a family of C-terminal-type (C-type) tandem C2 proteins with an N-terminal single transmembrane domain and is thought to regulate membrane traffic (reviewed in Refs. [1][2][3][4][5]. To date, 13 distinct syt genes have been identified in mice, rats, and humans, and several syt genes have been identified in invertebrates (6 -9). Syt I is evolutionarily conserved and is the best characterized isoform of Syt. Abundant Syt I is found on synaptic vesicles, and it has been shown to be essential for synaptic vesicle exocytosis and endocytosis in neurons by genetic analysis of syt mutants (reviewed in Refs. [2][3][4][5] and by antibody or peptide inhibition experiments (10 -13). Syt I is also present on dense-core vesicles in some neuroendocrine cells and has been shown to regulate Ca 2ϩ -dependent densecore vesicle exocytosis (14 -22). However, the precise subcellular localizations and functions of other Syt isoforms (Syts III-XIII) are still a matter of controversy (see Discussions in Refs. 23 and 24).
Syt IV was first described as a fourth member of the Syt family (25), and the syt IV gene was subsequently identified as an immediate early gene induced by membrane depolarization in brain and in PC12 cells (26 -28). Syt IV mRNA expression is developmentally regulated (29,30) and rapidly changes in response to a variety of extracellular stimuli (31)(32)(33)(34)(35). Since Syt IV null mutant mice exhibit abnormalities in motor performance and some forms of memory related to the hipoccampus (36), it has been suggested that Syt IV protein is crucial to learning and memory (or synaptic plasticity) (28,37). However, the role of Syt IV on the molecular level during learning and memory and its involvement in vesicular trafficking remain to be elucidated. Although Syt IV protein was first proposed to be a synaptic vesicle protein and function in concert with Syt I (38 -40), recent subcellular fractionation studies and immunoelectron microscopy have clearly demonstrated that rather than being a synaptic vesicle protein, Syt IV is present on uncharacterized vesicular/organelle structures in both axons and dendrites and in the Golgi of developing mouse brain (27,37,41). Perinuclear localization of Syt IV protein and distinct subcellular localizations of Syts I and IV have also been observed in some endocrine cells (PC12, pituitary AtT-20, and pancreatic ␣-cells) (27,42,43). In addition, Syt IV protein has been shown to be localized on immature vesicles rather than mature secretory vesicles in AtT-20 cells, suggesting a role of Syt IV protein as a keeper of the switch for the change from unregulated to regulated secretory vesicles (42). Whether endogenous Syt IV-containing vesicles/organelles fuse plasma membrane in response to Ca 2ϩ , however, had never been determined despite this information being important to learning whether Syt IV acts as a positive regulator or negative regulator of exocytosis.
In this study, we observed a nerve growth factor (NGF)-de-pendent redistribution of Syt IV protein in PC12 cells and discovered that NGF stimulates the sorting of Syt IV protein to mature dense-core vesicles that undergo Ca 2ϩ -dependent exocytosis in NGF-differentiated PC12 cells. Based on our findings, we discuss the distinct roles of Syt I and Syt IV in secretory vesicle trafficking in the brain.

MATERIALS AND METHODS
Antibody Purification-The anti-Syt I mouse monoclonal antibody (SYA148) was from StressGen (Victoria, British Columbia, Canada). The anti-Syt IX-C2A, anti-Syt IX-N, and anti-Syt IV-C2A rabbit polyclonal antibodies were prepared as described previously (21,27). The antibody specific for the N-terminal domain of the mouse Syt IV (anti-Syt IV-N) was raised against the following synthetic peptide with a C-terminal artificial Cys residue: MAPITTSRVEFDEC (Syt IV-N amino acids 1-14) (44). The antibody was affinity-purified by exposure to the antigenic peptide bound to FMP-activated Cellulofine (Seikagaku Co.) as described previously (44). The specificity of the antibody was checked by immunoblotting with recombinant T7-tagged Syts I-XIII expressed in COS-7 cells (21,22,27,45). Under our experimental conditions, immunoblotting did not reveal any evidence of cross-reactivity between the anti-Syt IV-N antibody and other Syt isoforms including with closely related isoform Syt XI (46) (data not shown). The protein concentration was determined with a Bio-Rad protein assay kit with bovine serum albumin as a reference. Immunoblotting was performed as described previously (7).
Immunoelectron Microscopy-The pre-embedding silver enhancement immunogold method was performed as described by Yoshimori et al. (48) with a slight modification. PC12 cells cultured on collagen type IV-coated plastic coverslips were fixed in 4% paraformaldehyde in sodium phosphate buffer (PB) (pH 7.4) for 2 h. The cells were washed in the buffer three times and were incubated for 30 min in PB containing 0.25% saponin and 5% bovine serum albumin and then for 30 min for blocking in PB containing 0.005% saponin, 10% bovine serum albumin, 10% normal goat serum, and 0.1% cold water fish skin gelatin. The cells were then exposed to the anti-Syt IV-C2A rabbit IgG (1/500 dilution) in the blocking solution overnight. After washing in PB containing 0.005% saponin six times for 10 min, cells were incubated with the FabЈ fragment of goat anti-rabbit IgG that had been conjugated to colloidal gold (1.4-nm diameter) in the blocking solution for 2 h. The cells were then washed with PB six times for 10 min and fixed with 1% glutaraldehyde in PB for 10 min. After washing, the gold labeling was intensified with a silver enhancement kit for 6 min at 20°C in the dark. After washing in distilled water, the cells were postfixed in 0.5% OsO 4 for 90 min at 4°C, washed in distilled water, incubated with 50% ethanol for 10 min, and stained with 2% uranyl acetate in 70% ethanol for 1 h. The cells were further dehydrated with a graded series of ethanol and embedded in epoxy resin. Ultrathin sections were doubly stained with uranyl acetate and lead citrate.

Less Colocalization of Synaptotagmin IV Protein with Conventional Secretory Vesicle Markers in NGF-differentiated
PC12 Cells-In a previous study using Syt IV-specific antibody, we showed that endogenous Syt IV protein is mainly present in brefeldin A-sensitive perinuclear regions (probably the Golgi) of undifferentiated PC12 cells (27,51). However, when relatively high concentrations of the anti-Syt IV-C2A antibodies (Ͼ1.5 g/ml) were used for immunocytochemical analysis, we also detected weak dot-like Syt IV signals at the cell periphery ( Fig. 1 in green), and in some cells, the Syt IV signals were accumulated at the tips of the cellular processes ( Fig. 1A, arrowhead). To investigate whether Syt IV signals at the cell periphery correspond to conventional secretory vesicles (densecore vesicles and/or synaptic-like microvesicles), we compared the Syt IV signals with six different secretory vesicle markers (21, 24, 27, 52) ( Fig. 1 in red; Rab27A (B and C), Rab3A (E and F), Syt I (H and I), Syt IX (K and L), synaptophysin (N and O), and VAMP-2 (data not shown)). To our surprise, none of the secretory vesicle markers tested colocalized with Syt IV protein even in the cellular processes (Fig. 1C, inset), indicating that Syt IV protein is unlikely to be present on conventional secretory vesicles in undifferentiated PC12 cells (Fig. 1C, F, I, L, and O) (53).
Consistent with this finding, we found that the Syt IV expression levels in the CGA AS -5 cell lines that almost completely lack dense-core vesicles (Ͻ10% of the number in wild-type PC12 cells) as a result of antisense knock down of chromogranin A (49) were almost unchanged (Ͼ80% control cell level). By contrast, the expression of Syts I and IX, two major Syt isoforms abundantly expressed on dense-core vesicles (21,22), was dramatically reduced to ϳ20 -30% of their expression levels in the control cells (data not shown).
After exposing the PC12 cells to NGF, we observed that Syt IV protein was also localized in the distal parts of neurites where dense-core vesicles are known to be accumulated (Fig. 2 in green), although the majority of the Syt IV signals remained in the perinuclear region. We again compared Syt IV signals with the six secretory vesicle markers, especially focusing on the distal portions of neurites (Fig. 2, insets). It should be noted that only a small population of Syt IV signals colocalized with dense-core vesicle markers (yellow dots in Fig. 2, panels C, F, I, and L) and none with synaptic-like microvesicle markers (Fig.  2O, synaptophysin, inset), although the majority of the Syt IV signals (green) in the neurites still did not coincide well with the dense-core vesicle markers (red) (Fig. 2, panels C, F, I, and L, insets). These observations were in great contrast to Syt I and Syt IX, two major Syt isoforms in PC12 cells that colocalize well in the neurites (21). Because NGF did not increase the protein expression levels of Syt IV (27), the Syt IV protein in the distal parts of neurites may have been transported from the cell body (i.e. newly forming vesicles from the TGN (trans-Golgi network)) or may have been redistributed locally (i.e. sorting of immature vesicles to dense-core vesicles at the cell periphery) but were unlikely to have been synthesized locally de novo. Therefore, we hypothesized that some populations of Syt IV protein are sorted to dense-core vesicles in response to NGF stimulation, and we tested this hypothesis by immunoelectron microscopy (Fig. 3) because it is impossible to judge the localization of Syt IV protein on dense-core vesicles by immunocytochemistry alone.
Synaptotagmin IV Protein Is Sorted to Dense-core Vesicles in PC12 Cells after NGF Stimulation-Immunoelectron microscopic analysis was then performed with highly specific anti-Syt IV antibody to reveal the exact localization of Syt IV protein in PC12 cells (27,37). As shown in Fig. 3A, Syt IV protein was abundantly localized on the Golgi membrane in undifferentiated PC12 cells, the same as in neocortical neurons of the developing mouse brain (37). The Syt IV signals were promi-nent in the cisterns of the trans-Golgi, TGN, and the vacuoles presumably formed from TGN, but some signals were also observed in the cis-Golgi. Consistent with the immunocytochemical findings described above, there were virtually no Syt IV signals on the mature dense-core vesicles around the plasma membrane (Fig. 3A, upper inset, arrowheads), but in some cases, the Syt IV signals were observed on the lighter vesicles presumably corresponding to immature vesicles (Fig. 3A, lower inset, arrowheads) (42). The majority of the Syt IV signals in NGF-differentiated PC12 cells was still observed around the Golgi membranes (Fig. 3B), but some Syt IV signals were clearly localized on the dense-core vesicles in addition to the immature vesicles, especially in the distal parts of neurites and around the plasma membrane (Fig. 3B, arrowheads, inset). Interestingly, in many specimens, the Syt IV-containing densecore vesicles were much lighter than the non-Syt IV-containing dense-core vesicles, suggesting that Syt IV-containing dense-  core vesicles may form immediately after NGF stimulation. Therefore, we concluded that at least some populations of Syt IV protein are indeed sorted to dense-core vesicles in PC12 cells after NGF stimulation.
NGF-dependent Sorting of Synaptotagmin IV Protein to Mature Dense-core Vesicles That Undergo Ca 2ϩ -dependent Exocytosis-Finally, we attempted to determine whether Syt IVcontaining dense-core vesicles in the presence of NGF are fully mature and thus capable of exocytosis in response to Ca 2ϩ stimulation because PC12 cells contain various types of secretory vesicles (54). To visualize the dynamics of endogenous Syt IV molecules during Ca 2ϩ -dependent exocytosis, an N-terminal antibody-uptake experiment was performed as described previously (21,23). Antibodies against the luminal domain of Syt IV (or IX) were added to the extracellular medium, and PC12 cells were stimulated either with a low or high concentration of KCl. If the Syt IV-containing vesicle undergoes exocytosis in response to Ca 2ϩ stimulation, the N terminus of Syt IV would be accessible on the outer surface of the cell membrane and therefore should be recognized by anti-Syt IV-N antibody in the culture medium. The Syt IV-antibody complex would then be incorporated into the cell by endocytosis.
The uptake of the anti-Syt IX-N antibody into the cell body and neurites occurred only at depolarizing KCl concentrations regardless of NGF exposure as described previously (Fig. 4, I-L). The high KCl-dependent uptake of the antibody should be Ca 2ϩ -dependent but not depolarization-dependent, because no uptake was observed in the presence of extracellular EGTA even in response to high KCl stimulation (data not shown) (23). By contrast, the uptake of the anti-Syt IV-N antibody was both NGF-and high KCl-dependent (Fig. 4, A-H). The anti-Syt IV-N antibody was not incorporated very much into the cell body of undifferentiated PC12 cells even in response to the high KCl stimulation (Fig. 4, A-D), consistent with our immunoelectron microscopic observations that Syt IV protein is mainly localized in the Golgi or immature secretory vesicles in undifferentiated PC12 cells (Fig. 3A). It should be noted that after NGF stimulation, high KCl-dependent uptake of the anti-Syt IV-N antibody into neurites was prominent (Fig. 4, E-H).
We also used fluorescence-labeled antibodies (i.e. rhodaminelabeled anti-Syt IV-N and fluorescein-labeled anti-Syt IX-N antibodies) to investigate whether Syt IV and Syt IX proteins are incorporated into the same sites or different sites via endocytosis. There was obvious colocalization of the fluoresceinanti-Syt IX-N and rhodamine-anti-Syt IV-N antibodies in the neurites, but some signals were Syt IX-N antibody-specific (Fig. 5, A-C, arrows) or Syt IV-N antibody-specific (Fig. 5, A-C,  arrowheads). It should be noted that colocalization of Syt IX FIG. 3. Localization of synaptotagmin IV protein by immunoelectron microscopy in PC12 cells cultured with and without NGF. Panels show representative views of the perinuclear region containing the Golgi stained with affinity-purified anti-Syt IV-C2A antibody followed by silver enhancement. Note that Syt IV signals (i.e. gold particles) were abundant at the Golgi membranes of both the undifferentiated PC12 cells (A) and NGF-differentiated PC12 cells (B). In the absence of NGF, hardly any Syt IV signals were detected on the mature dense-core vesicles near the plasma membrane (arrowheads, upper inset in A), but some Syt IV signals were associated with immature vesicles (arrowheads, lower inset in A). By contrast, in the presence of NGF, Syt IV signals are often found on the mature dense-core vesicles (arrowheads, inset in B) in the neurites. Scale bars in B and inset in A are 1 m and 500 nm, respectively. N, nucleus; G, Golgi. and Syt IV clearly increased after high KCl stimulation (compare Fig. 2 with 5). This change may be explained by the notion that even if Syt IV-and Syt IX-containing vesicles undergo exocytosis at different sites, Syt IV and Syt IX proteins can be retrieved at the same sites and sorted to the same dense-core vesicles.
We further investigated whether the Syt IV-N antibody uptake occurs after only a short exposure to NGF, and the results showed that only a 1-h exposure to NGF is adequate to detect the uptake of the rhodamine-Syt IV-N antibody (Fig. 5F), suggesting that Syt IV proteins at the cell periphery may be rapidly sorted into dense-core vesicles rather than being formed from the TGN and transported to the cell periphery. Unlike the N-terminal antibody uptake seen in the neurites of NGF-differentiated PC12 cells, the fluorescein-Syt IX-N and rhodamine-Syt IV-N antibody signals were often somewhat different (Fig. 5G, insets). DISCUSSION Although both the Syt IV mRNA and protein expression levels rapidly increase after exposure to depolarizing stimuli (26,27) and Syt IV null mutant mice exhibit abnormalities in some forms of memory and motor performance (36), whether Syt IV is actually involved in membrane traffic related to learning and memory had never been elucidated. All of the previous studies on the function of Syt IV protein had been conducted by overexpression (39,53) or exogenous addition of recombinant proteins (40). The former approach is sometimes unreliable for studies on Syt function because exogenously expressed Syt proteins or fragments (especially produced by forced overexpression) often result in mislocalization when compared with endogenous protein (23,24,55). Actually, one study (53) reports that overexpressed Syt IV protein is localized on dense-core vesicles in undifferentiated PC12 cells, whereas others show (42,43,51) that it is localized in Golgi and/or immature vesicles. The latter approach (so-called "dominant negative approach") also has some drawbacks, because recombinant proteins from Syt isoforms that are not endogenously expressed inhibit Ca 2ϩ -dependent secretion in PC12 cells more strongly than recombinant proteins from endogenous Syt isoforms (22,56) and recombinant Syt proteins bind various mol-ecules important for Ca 2ϩ -dependent secretion (e.g. SNARE protein) (18,19). In addition, recombinant C2 fragments from bacteria are often contaminated by non-proteinaceous components (57), and contradictory results have been reported even when the same cDNA constructs have been used (18,56). Therefore, it is crucial to determine the exact localization of endogenous Syt IV protein and its dynamics during Ca 2ϩ -dependent exocytosis. In this study, we used PC12 cells to study Syt IV function and localization, because Syt IV is more abundantly expressed in PC12 cells than in brain (27) and PC12 cells are often used as a good system for studying Ca 2ϩ -dependent exocytosis. Immunoelectron microscopic analysis (Fig. 3) and the Nterminal antibody-uptake experiment (Fig. 4) clearly demonstrated that the endogenous Syt IV protein in PC12 cells is mainly localized in the Golgi and/or immature secretory vesicles (42) and that after NGF stimulation some populations of Syt IV protein are sorted to mature dense-core vesicles. Because the Syt IV N-terminal antibody uptake occurs only after a 1-h exposure to NGF, Syt IV protein present at the cell periphery rather than TGN-derived Syt IV protein is most likely to be sorted to mature dense-core vesicles. Since exposure to NGF did not alter the expression levels of Syts I and IV (27), newly formed Syt IV-containing dense-core vesicles are expected to carry a single Syt IV isoform, not Syts I and IX. Consistent with this finding, colocalization of Syts I (or IX) and IV in the distal parts of neurites was very limited even in NGF-differentiated PC12 cells (Fig. 2, I and L). However, after exocytosis, Syt IV and Syt IX proteins are likely to be retrieved at the same or similar sites and then be sorted to the same dense-core vesicles because of the obvious colocalization of Syt IV and IX after high KCl stimulation (i.e. colocalization of the fluorescein-anti-Syt IX-N and rhodamine-anti-Syt IV-N antibodies in the neurites) (Fig. 5, A-C). Consistent with our findings, Ng et al. (58) and Amino et al. (59) recently reported that short NGF stimulation of PC12 cells enhances releasable pools of peptide hormone. Thus, it is highly possible that Syt IV is involved in the enhancement of some releasable pools in response to NGF.
At present, the mechanism of the NGF-dependent Syt IV sorting to mature dense-core vesicles at the cell periphery remains unknown, but we speculate that certain properties (e.g. phosphorylation and/or protein interaction) in the unique spacer domain of Syt IV (51) qualitatively change after NGF exposure. We recently suggested (60) that the interaction between Syt I and VAMP-2 may be involved in sorting of Syt I protein to secretory vesicles. Thus, it may be possible that interaction between Syt IV and other VAMP isoforms (e.g. VAMP-4 that is also localized in the Golgi and immature secretory vesicles) (42) regulates Syt IV protein sorting since Syt IV interacts with certain VAMP isoforms in vitro. 2 The physiological meaning of the Syt-VAMP interaction is now under investigation in our laboratory.
How does Syt IV protein function during Ca 2ϩ -dependent exocytosis? Based on an analysis of Syt IV overexpression in the fruit fly, Drosophila Syt IV was first proposed to be a synaptic vesicle protein and to negatively regulate neurotransmitter release by binding to Syt I via the C2B domain (39). However, a subcellular fraction study and electron microscopic analysis have shown that the mouse Syt IV protein is not localized on synaptic vesicles and is present in the much denser vesicles/organelles (27,29,37). In addition, the Ca 2ϩ -dependent and -independent oligomerization activity of the mouse Syt IV is very weak compared with that of mouse Syt I (50,(61)(62)(63), indicating that Ca 2ϩ -dependent hetero-oligomerization of the mouse Syts I and IV is unlikely in vivo. Because the isolated recombinant C2A domain of Syt IV lacks Ca 2ϩ -dependent phospholipid binding activity due to one amino acid substitution (Ser-244) at the putative Ca 2ϩ -binding loop 3 (46,64), Syt IV was often thought to be a Ca 2ϩ -independent type Syt that may negatively regulate exocytosis (39,53). However, recent findings have strongly contradicted this notion and suggested that Syt IV is a positive Ca 2ϩ -dependent regulator of exocytosis, the same as Syt I. First, two C2 domains of Syts I and VII have redundant Ca 2ϩ binding sites, and the mutation of a single Asp residue responsible for Ca 2ϩ binding in one C2 domain is neu-tral for Ca 2ϩ -dependent phospholipid binding and Ca 2ϩ -dependent oligomerization (19,65,66). Actually, the Syt IV C2B domain has five Asp residues that may be crucial for Ca 2ϩ binding, and the full cytoplasmic region of Syt IV interacts with negativelychargedphospholipids(phosphatidylserine)inaCa 2ϩdependent manner (64). Second, the Syt I mutant carrying Asp-to-Ser substitution in the C2A domain (mimics Syt IV) can fully rescue Syt I null mutant animals (for review see Refs. 5, 67, and 68). More recently, Drosophila Syt IV was found to be capable of rescuing neurotransmitter release at the neuromuscular junction in Syt I null flies (69), strongly indicating that Syt IV is a positive regulator for Ca 2ϩ -regulated exocytosis, the same as Syt I. Third, our N-terminal antibody-uptake experiments showed that Syt IV-containing dense-core vesicles can fuse plasma membrane in response to Ca 2ϩ stimulation, indicating that the effect of Syt IV on dense-core vesicle exocytosis is not entirely inhibitory.
In summary, we showed for the first time that Syt IV protein is NGF-dependently sorted to fusion-competent mature densecore vesicles in PC12 cells. Our results suggest that Syt IV protein regulates stimulus (e.g. NGF)-dependent membrane trafficking that may be involved in plastic changes at the synapses in brain in contrast to the role of Syt I protein in synaptic vesicle trafficking.