Calcium-Evoked Dendritic Exocytosis in Cultured Hippocampal Neurons. Part I: Trans-Golgi Network-Derived Organelles Undergo Regulated Exocytosis

FM1-43 labels dendritic compartments

Brief exposure of cultured neurons to FM1-43, combined with depolarizing stimuli, can produce a selective loading of presynaptic vesicles (Fig. 1A,C) (Ryan et al., 1993). In contrast, we found that cultured hippocampal neurons incubated with FM1-43 for 16–36 hr in the absence of stimulation loaded intracellular compartments in all processes and the soma (Figs. 1D, 2, 4). We focused our analysis on the characterization of FM1-43-labeled compartments in processes, anticipating that these may be more likely to be associated with synaptic plasticity.

To determine the distribution of FM1-43 fluorescence in pre- and postsynaptic compartments after the two labeling protocols, we compared its labeling pattern with that of a presynaptic marker protein, synapsin I (Fig. 1B,E; n = 3). After standard acute exposure and depolarizing stimuli, FM1-43 and synapsin I showed significant colocalization, confirming the presynaptic localization of FM1-43 with this loading protocol (Fig.1A,B). However, with long exposure and no stimulation, FM1-43 and synapsin I showed little colocalization (Fig.1D,E). This indicates that the longer exposure selectively labels intracellular compartments that are not presynaptic terminals. Quantitated immunocytochemical analysis (see Materials and Methods) confirmed that, with overnight exposure to the dye, FM1-43 fluorescence in synapsin I-labeled compartments accounts for only 10% of the dye (Fig. 3A; R = 0.13 ± 0.06;n = 3).

Dendritic compartments labeled with FM1-43 often were seen at the base of small filopodial structures, proposed to be synaptic spine precursors (Fig. 1F) (Ziv and Smith, 1996). Time-lapse imaging over 1 hr revealed that FM1-43-loaded compartments maintained this association although the filopodial structures moved over time (∼2 μm/min; Fig. 1F). In contrast, presynaptic vesicles labeled with acute loading of FM1-43 did not move within 1 hr (n = 3; Fig. 1C), again supporting the conclusion that overnight FM1-43 exposure loads compartments that are not presynaptic terminals.

The dendritic localization of FM1-43-labeled compartments was confirmed further by correlative electron and fluorescent microscopy (Fig.2). Neurons loaded with FM1-43 overnight were imaged under high magnification (100× lens; Fig. 2-1,6) and then processed for electron microscopy (EM). The same regions were identified on the EM preparations (Fig. 2-3,8) and aligned with the light microscope images by using fiduciary points (Fig. 2-4,9). Under EM, FM1-43-fluorescent sites were dendritic and often close to synapses. The sites included different organelles, such as coated pits, endosomes, multivesicular bodies, endoplasmic reticulum, Golgi apparatus, and mitochondria (Fig. 2-5,10). No dense core granules were observed.

To identify which dendritic organelles are labeled by overnight exposure to FM1-43, we performed high-resolution immunocytochemical studies on FM1-43-loaded neurons (Fig.3A). After FM1-43 imaging, the cultures were fixed and labeled with different organelle-specific markers: transferrin receptor (Moos, 1996), cathepsin D (Nakanishi et al., 1994), and mannose-6-phosphate receptor (Couce et al., 1992), specific for endosome/lysosome organelles; IP₃ receptor (Moschella et al., 1995) and BiP (Huovila et al., 1992), specific for endoplasmic reticulum; and NBD C₆-ceramide (Pagano et al., 1989), a trans-Golgi fluorescent marker. Quantitative immunocytochemical analysis (see Materials and Methods) indicated little colocalization between FM1-43-labeled compartments and transferrin receptor (R = 0.2 ± 0.14;n = 3), cathepsin D (R = 0.4 ± 0.16; n = 3), mannose-6-phosphate receptor (R = 0.21 ± 0.1; n = 3), IP₃ receptor (R = 0.14 ± 0.07;n = 3), and BiP (R = 0.3 ± 0.09;n = 3). However, we observed a consistently high correlation of FM1-43-labeled dendritic sites only with NBD C₆-ceramide (R = 0.77 ± 0.11;n = 5; Fig. 3). This indicates that, after overnight exposure to FM1-43, dendritic organelles loaded with the FM1-43 are primarily dendritic Golgi-like or Golgi-derived structures.

Dendritic organelles undergo regulated exocytosis

To determine whether FM1-43-labeled compartments exhibit calcium-evoked exocytosis, we challenged neurons loaded with FM1-43 for 16–36 hr with the calcium ionophore A23187 (calcimycin) for 1 min. This produced a marked loss of FM1-43 fluorescence (Fig.4A) that was attributable to the exocytosis of internalized dye (see below). We measured this CEDE as the percentage of loss of background-subtracted FM1-43 fluorescence as compared with pre-ionophore background-subtracted fluorescence (Fig. 4B). Exposure to the carrier solution alone (0.1% DMSO) produced no loss of fluorescence (Fig. 5; n = 12). Similarly, calcimycin did not have any effect when it was applied in the presence of 5 mm EGTA (Fig. 5; n = 6). A23187 induced CEDE only in the presence of 2.5 mmcalcium. It could be recorded within 5 min after ionophore application and persisted for 15 min.

Ionophore-induced exocytosis from neuronal cultures loaded overnight with FM1-43 was confirmed by recovering dye in the extracellular medium (Henkel et al., 1996). Extracellular medium aliquots from 10 pooled cultures were obtained before (A_pre) and 15 min after (A_post) A23187 application. Dye from pooled samples was extracted with butanol and dissolved in 2% CHAPS. FM1-43 fluorescence intensity was measured in microcapillaries. The relative FM1-43 fluorescence intensity of solution after A23187 treatment was 3.32 ± 0.7 (n = 7;p < 0.05). This shows directly that a significant amount of FM1-43 was released by calcium ionophore treatment and could be recovered in the extracellular medium. Calibration curves generated from 1.5 nm to 1.5 μm FM1-43 indicate that the observed exocytosis corresponds to ∼35 nm FM1-43. This is ∼100 times more than the estimated FM1-43 release measured after similar experiments conducted on presynaptic exocytosis (0.3 nm; Henkel et al., 1996).

If the organelles undergoing CEDE are derived from the trans-Golgi network (TGN), then one may expect that endocytosed FM1-43 may require several hours before reaching such compartments. To examine this possibility, we measured the amount of CEDE as a function of the FM1-43 loading time (Fig. 6A). Mature cultures were exposed to FM1-43 for different periods of time (1–16 hr) and challenged with A23187. Only when neurons were exposed to FM1-43 for >8 hr was CEDE observed (Fig. 6A). Notably, FM1-43 showed little colocalization with NBD C₆-ceramide fluorescence after 4 hr of loading (n = 2) and significant colocalization after 16 hr of FM1-43 loading (Fig. 6B). These results indicate that the passive loading of FM1-43 requires 8 hr before the dye reaches CEDE-competent organelles, at which time the dye colocalizes with NBD C₆-ceramide. This result supports the view that organelles undergoing CEDE are TGN-like or TGN-derived structures.

We tested if CEDE uses exocytotic machinery with tetanus toxin, a specific inhibitor of exocytosis (Montecucco and Schiavo, 1994). CEDE was blocked completely by pretreatment with 10 nm tetanus toxin (see Fig. 5; n = 7), indicating regulated exocytosis rather than some nonspecific loss of the dye from intracellular compartments.

We found that CEDE is regulated developmentally. Although neurons of all ages that were examined (5–21 DIV) showed comparable loading after overnight exposure to FM1-43, only neurons that were 9 DIV or older were capable of producing CEDE (see Fig. 5; n = 11;p < 0.05). At this stage the cultured neurons have already established synaptic contacts and are fully differentiated. At earlier ages A23187 occasionally induced regrouping of labeled sites or intensified movement of fluorescently labeled organelles but caused no detectable CEDE.

Dendritic exocytosis of FM1-43 also could be elicited with repetitive field electrical stimulation (Fig. 7). We tested different numbers and frequencies of stimuli. Low-frequency stimuli (1 Hz) did not produce any detectable loss of fluorescence (n = 3). Medium-frequency stimuli (10 Hz) could produce loss of fluorescence, but only if sufficient stimuli were delivered. High-frequency stimuli (50 Hz) generally produced exocytosis (n = 10). Figure 7 shows a pattern of CEDE observed in a neuron stimulated with different parameters. This pattern generally was observed in the tested neurons. In addition, the application of 1 mm glutamate for 1 min could evoke the exocytosis of some dendritic structures (n = 2; data not shown). We characterized CEDE by using calcium ionophore as the stimulating agent, because it gave the most robust effect.

FM1-43 labels dendritic compartments

In this study we report on a novel biological process of regulated dendritic exocytosis. Cultured hippocampal neurons were exposed to FM1-43 for >16 hr with no stimulation. Because FM1-43 immerses into the outer leaflet of the plasma membrane, it is internalized inside the cells whenever there is endocytosis. On washout of the uninternalized dye from the outer surface, the remaining FM1-43 fluorescence marks the inner membrane of intracellular membranous compartments. Via the fusion of endosomes and endosome-derived compartments, over time, the dye has access to numerous intracellular compartments. Because the neurons are exposed continuously to FM1-43 for >16 hr, some FM1-43 can be found in intracellular compartments that are derived from endocytic organelles—like early endosomes (expressing transferrin receptor), late endosomes and lysosomes (expressing cathepsin D and mannose-6-phosphate receptor), and endoplasmic reticulum in soma (expressing IP₃ receptor and BiP). However, quantitative analysis of the colocalization between FM1-43-labeled structures and intracellular markers of various organelles shows the greatest degree of colocalization with NBD C₆-ceramide, indicating the labeling of TGN-like or TGN-derived structures (see Fig. 3).

The immunocytochemical analysis of labeled compartments is supported by correlative electron and light microscopy. Electron microscopy of FM1-43-labeled regions indicates that the majority of labeled sites clearly is in dendrites and not presynaptic terminals, has no dense core vesicles, has smooth endoplasmic reticulum (SER)/TGN organelles, and often is close to synapses. SER/TGN-like structures have been identified in spine apparatus and parent dendrites (Spacek and Harris, 1997). In addition, numerous organelles potentially involved in endocytosis and exocytosis have been visualized in spines under electron microscopic analysis (Harris and Kater, 1994; Spacek and Harris, 1997), although their function has only been hypothesized. Smooth vesicles fusing with the plasma membrane were seen in some spines, and exocytosis was suggested on the basis of these observations (Spacek and Harris, 1997).

In contrast to acute loading protocols, overnight exposure of neurons to FM1-43 appears to load little dye in presynaptic sites. This is indicated by several observations. First, organelles labeled with the overnight exposure to FM1-43 do not colocalize with synapsin I, a marker for presynaptic vesicles (see Fig. 1D,E). In addition, in associated electron micrographs the fluorescent regions are in dendrites and only occasionally in presynaptic terminals (see Fig. 2). Organelles loaded with overnight exposure to FM1-43 show considerable movement (∼2 μm/min) when observed during the 1 hr period, whereas organelles loaded with brief exposure to high potassium (which loads presynaptic terminals) do not move (see Fig.1F). After the neurons are exposed to FM1-43 in the absence of depolarization, 8 hr are required before the dye reaches exocytosis-competent organelles (see Fig. 6A). Presynaptic vesicles require only seconds after endocytosis to become exocytosis-competent (Ryan, 1996). Finally, compartments labeled with overnight exposure to FM1-43 colocalize with NBD C₆-ceramide, which is found only in soma and dendrites and not in presynaptic terminals (see Figs. 3,6B). The significantly lower labeling of presynaptic sites during prolonged dye exposure may be attributable to their lower volume, mechanical perturbation (moving the coverslip to the recording chamber could induce presynaptic release), or extensive washing (generally >1 hr may preferentially induce presynaptic loss).

Calcium-evoked dendritic exocytosis

We find that a brief challenge of hippocampal neurons with calcium ionophore A23187 produces a robust dendritic exocytosis. CEDE of FM1-43 was demonstrated directly by measuring dye fluorescence in the extracellular medium. Calcium is required for CEDE, because ionophore did not produce any effect in the presence of the calcium chelator EGTA (see Fig. 5). CEDE was blocked by tetanus toxin, an agent known to block exocytosis (Montecucco and Schiavo, 1994). This indicates that the target of tetanus toxin, synaptobrevin, cellubrevin, or a homolog (Yamasaki et al., 1994; Chilcote et al., 1995), is necessary for CEDE.

CEDE can be detected within 5 min after ionophore application and reaches a plateau ∼15 min after stimulation. Individual sites do not disappear suddenly but destain continuously over this time. This is comparable to what has been described with presynaptic destaining, in which a single spot observed by light microscopy destains slowly rather than abruptly (Betz et al., 1992). The slow time course is explained by the presence of many small (50 nm) structures (labeled presynaptic vesicles) that cannot be viewed individually by light microscopy (Betz et al., 1992). The slow time course of CEDE may owe to a similar underlying process. For instance, CEDE destaining is consistent with the budding of small vesicles from dendritic SER/TGN-like structures and their fusion with the local dendritic plasma membrane. Such a process could occur over a time frame of minutes and would cause a relative decrease in the fluorescence of the SER/TGN-like compartments, rather than their complete destaining. Although small vesicles may be formed, they appear not to be transported for long distances along the dendrite, because all regions along a dendrite lose fluorescence with a similar time course. If vesicles were to move along a dendrite, then one would expect bright regions to lose fluorescence and neighboring dark regions (in the dendrite) to gain fluorescence. However, we see homogeneous loss of FM1-43 fluorescence over time, which indicates a delivery to the immediate surface. Therefore, organelles appear to be released locally.

Intact microtubules are required for CEDE, because nocodazole, an agent producing microtubule depolymerization (Goslin et al., 1989), prevents CEDE (see Fig. 5). An association between CEDE-competent compartments and microtubules is supported by the observation that these compartments undergo rapid movements that appear synchronized to local filopodial movements (see Fig. 1F). Thus, microtubules may provide a delivery system for TGN-like organelles to reach appropriate sites as well as a docking station for additional components that may be necessary to execute the numerous processes required for exocytosis (budding, docking, fusion, etc.).

Exocytosis of FM1-43-labeled sites also was observed in the soma, which is consistent with the recent publication demonstrating calcium-dependent exocytosis of compounds from the cell body of dorsal root ganglion cells (Huang and Neher, 1996).

Our experiments indicate that passive loading of FM1-43 requires 8 hr before the dye reaches structures that are competent to undergo CEDE. Studies of intracellular trafficking indicate that 8 hr is sufficient for endocytosed material to reach biosynthetic pathways like SER/TGN in cultured cells (Green and Kelly, 1992). Thus, these experiments further support the view that the organelles undergoing CEDE are TGN-derived structures and part of the biosynthetic pathway.

Under experimental conditions described in this study, in which calcium is raised everywhere in the cell, CEDE is widespread. However, the homogeneous loss of the dye in a nonhomogeneously labeled structure indicates that exocytosis is local. These observations suggest the existence of a general mechanism that is present throughout the cell. It is well known that physiological stimuli can produce a localized rise in postsynaptic Ca²⁺ concentration. Under such conditions one may expect localized CEDE, thereby ensuring spatial specificity to this process.

Possible Roles of CEDE

CEDE may underlie several previously described phenomena. Some of them include the release of dopamine from dendrites in substantia nigra neurons (Heeringa and Abercrombie, 1995) and somatic release of substance P observed after strong stimulation (Huang and Neher, 1996). Furthermore, there is evidence supporting a role for activity or calcium in the release of growth factors and β-amyloid (Querfurth and Selkoe, 1994; Blochl and Thoenen, 1995). It will be interesting to determine whether CEDE releases such compounds.

In addition, CEDE may be involved in the activity-dependent expression of transmembrane proteins. There are several reports on the activity-dependent expression of membrane-bound proteins, like neural cell adhesion molecules (NCAM) (Kiss et al., 1994). CEDE also might be involved in the redistribution of substance P receptors in spinal neurons after somatosensory stimulation (Mantyh et al., 1995) or the homeostatic control of transmembrane glucose transporter proteins (Cushman and Wardzala, 1980; Suzuki and Kono, 1980). Similarly, CEDE might play a role in the distribution of ion channels at the surface membrane, observed under different stimulus conditions (Harris et al., 1991; Turrigiano et al., 1994).

Could CEDE be involved in long-term potentiation (LTP)? Indeed, a recent report indicates that postsynaptic exocytosis is required to generate LTP (Lledo et al., 1998). Localized calcium entry during LTP induction might trigger the insertion of glutamate receptors (stored in intracellular compartments or perisynaptic regions) (Hampson et al., 1992; Baude et al., 1995) into synapses by a CEDE-like process. This would produce larger responses to a synaptically released transmitter. A similar process triggered by insulin appears to deliver GABA receptors to the cell surface (Wan et al., 1997). In addition to delivering receptors to synapses, substances concentrated inside the postsynaptic membranous structures could be released into the synaptic cleft, potentially affecting presynaptic function.

Finally, CEDE could provide new membrane at postsynaptic sites, allowing for the growth and formation of new synapses (Lisman and Harris, 1993). Indeed, SER and smooth vesicles have been localized at the base and inside dendritic spines (Spacek and Harris, 1997), but their function has not yet been clarified.

Conclusion

In summary, in this manuscript we have described a novel process of regulated exocytosis in cultured hippocampal neurons—CEDE. We have shown that CEDE has a number of important, physiologically relevant, properties: it is calcium-evoked and it requires mature neurons, exocytotic machinery, and intact microtubules. Compartments that undergo CEDE derive from SER/TGN-like organelles in dendrites and postsynaptic sites. This phenomenon thus could play a role in synaptic plasticity, where it might involve changes in glutamate sensitivity, release of active compounds having effect on pre- and postsynaptic function, and/or growth of the plasma membrane.