Imaging of evoked dense-core-vesicle exocytosis in hippocampal neurons reveals long latencies and kiss-and-run fusion events

A detailed understanding of how a nervous system functions requires analysis of its chemical signaling, which is largely mediated by regulated vesicle exocytosis. Regulated exocytosis in neurons involves at least two types of secretory vesicles: synaptic vesicles (SVs) and dense-core vesicles (DCVs) (De Camilli and Jahn, 1990). SVs contain classical neurotransmitters and mediate fast synaptic transmission. DCVs store and release a diverse array of modulators, including neuropeptides, monoamines and neurotrophins (Bean et al., 1994; Lessmann et al., 2003; Li et al., 2005), that regulate many crucial processes, such as neuronal survival, synaptic plasticity and learning (Chen and Strickland, 1997; Poo, 2001; Strand et al., 1991). Whereas exocytic and endocytic events have been extensively studied for SVs in neurons and DCVs in endocrine cells (Neher, 1998; Penner and Neher, 1989; Ryan, 2001), the vesicle cycle for DCVs in neurons is still poorly understood. Little is known about the cellular location of DCV exocytosis, the latency between stimulation and DCV fusion, release probabilities for DCVs, the kinetics of individual DCV exocytic events or the termination of exocytosis and DCV-membrane retrieval.

Traditional electrophysiological studies are restricted in their ability to monitor single-vesicle release events. Capacitance and amperometric measurements achieve single-vesicle resolution in only specialized conditions (Henkel and Almers, 1996; Penner and Neher, 1989; Staal et al., 2004). By contrast, recently developed imaging techniques have allowed the tracking of individual vesicles and the visualization of single SV exocytic events in neurons (Aravanis et al., 2003; Gandhi and Stevens, 2003; Murthy et al., 1997; Ryan et al., 1997; Zenisek et al., 2000) and DCV exocytic events in endocrine cells (Steyer et al., 1997; Takahashi et al., 2002; Taraska et al., 2003). DCV transport in central neurons has also been visually tracked by fluorescence microscopy (Shakiryanova et al., 2006; Silverman et al., 2005), but the exocytosis of peptidergic DCVs has not been studied by single-vesicle imaging.

In the current work, we selectively labeled peptidergic vesicles in primary cultured hippocampal neurons by targeting GFP-tagged propeptides to DCVs and studied the vesicle cycle by total internal reflectance fluorescence (TIRF) microscopy. This approach allowed the imaging of single exocytic events from plasma-membrane-proximal DCVs in both the soma and neurites. Our results indicate that peptidergic DCV exocytosis exhibits long latencies from stimulation to fusion, occurs with high probability in somatic regions and principally uses a kiss-and-run mode. Fusion-pore openings and re-closures were regulated by Ca²⁺ levels selectively meditated by L-type Ca²⁺-channel entry.

To directly image stimulus-secretion coupling of DCVs in neurons, primary cultured hippocampal neurons were transfected with plasmids encoding atrial-natriuretic-factor–EGFP (ANF-EGFP), which is specifically targeted to DCVs (Burke et al., 1997; Shakiryanova et al., 2006). A confocal z-series revealed a punctuate distribution of ANF-EGFP throughout the soma and neurites of the neurons (supplementary material Fig. S1). The region within ∼185 nm of the plasma membrane was selectively illuminated by TIRF microscopy (evanescent-field penetration depth ∼80 nm) to visualize the vesicles. ANF-EGFP fluorescence near the coverslip appeared as uniformly sized fluorescent puncta in both the soma and neurites of the hippocampal neurons (Fig. 1A, left panel). The ANF-EGFP-containing puncta were the same apparent size as 100-nm beads but were significantly smaller than 400-nm beads (Fig. 1A, right panel). DCVs have diameters of ∼100-200 nm (Grabner et al., 2006; Klyachko and Jackson, 2002), so the fluorescent puncta probably corresponded to single DCVs. All of the ANF-EGFP-containing puncta in the evanescent field were positive for secretogranin II (SgII) (Fig. 1B, first row), a DCV-content marker (Huttner et al., 1991), which confirmed their identity as DCVs. All of the ANF-EGFP-containing puncta in the evanescent field were also positive for the Ca²⁺ sensor synaptotagmin I (SytI) (Fig. 1B, second row) (Brose et al., 1992), implying that the vesicles could participate in Ca²⁺-dependent exocytosis. A different set of vesicles was identified as being weakly positive for vesicular monoamine transporter 2 (VMAT2) (Fig. 1B, third row), which was consistent with the finding that hippocampal neurons transiently express monoamine transporters until at least postnatal day 10 (Lebrand et al., 1998). The ANF-EGFP-containing DCVs probably store and release neurotrophins, because most of them were stained well for brain-derived neurotrophic factor (BDNF), an abundant neurotrophin (Fig. 1B, fourth row). In confirmation of this, expression of BDNF-RFP in the cells resulted in >85% colocalization with ANF-EGFP, as assessed by TIRF microscopy (not shown). Overall, these results implied that ANF-EGFP is targeted to peptidergic DCVs in hippocampal neurons, which enabled single-vesicle visualization by TIRF microscopy.

Little lateral movement was detected for most ANF-EGFP-containing DCVs within the evanescent field in both resting and stimulated cells (supplementary material Movie 1). When each vesicle was circled with a 400-nm diameter circle for intensity measurement, most vesicles stayed inside the circle, indicating that their lateral movement was restrained by cellular components. We refer to these relatively immobile DCVs within the evanescent field as plasma-membrane-proximal DCVs, although whether these are `docked' at the plasma membrane or `tethered' to subplasmalemmal cytoskeletal elements was not determined in this study. To stimulate DCV exocytosis, the external [K⁺] was elevated from 2.5 mM to 90 mM to promote depolarization and Ca²⁺ influx via voltage-gated Ca²⁺ channels. This led to a rapid elevation of intracellular Ca²⁺ level within 2 seconds (Fig. 2; supplementary material Fig. S2), and DCV exocytic events were recorded and analyzed by TIRF microscopy in the subsequent 100 seconds. Ca²⁺ influx triggered numerous DCV exocytic events, which were detected as transient fluorescence increases accompanied by the release of the contents (fluorescent ANF-EGFP) (Fig. 3C). The initial increased fluorescence of an exocytic event arises from a combination of the vesicle moving closer in the evanescent field, the neutralization of the vesicular pH upon fusion, with the brightening of EGFP, and the secretion of ANF-EGFP into the exponential evanescent field. In contrast to the large number of exocytic events seen in the cell soma region, very few events were observed in the neurites (supplementary material Movie 1). DCV-fusion probabilities for soma and neurites following stimulation were calculated as the number of exocytic events in 100 seconds following depolarization divided by the number of total immobile plasma-membrane-proximal vesicles within the evanescent field. Release probabilities were more than 50 times higher for DCVs in the soma compared with those in the neurites (Fig. 3A). All events that were observed in the soma and neurites were dependent upon stimulation, and no spontaneous release events were observed. The markedly increased fusion probability for DCVs in the soma compared with those in neurites was unexpected because the traditional view is that most DCVs formed in the cell body are transported to nerve terminals for exocytosis (Kelly and Grote, 1993; Shakiryanova et al., 2006), unlike SVs, which are formed and released locally at nerve terminals. Our results indicate that the neuronal cell body, which is the DCV-assembly site, is also a dominant release site for DCVs. Although somatic vesicle release has previously been documented in ultrastructural, biochemical and electrophysiological studies (Huang et al., 2007; Huang and Neher, 1996; Nirenberg et al., 1996; Pudovkina et al., 2001), the predominant distribution of DCVs with high release probabilities in soma has not previously been appreciated.

Latency histograms of exocytic events for DCVs in the soma (Fig. 3B) revealed a time constant (τ) of ∼16 seconds between stimulation and fusion, whereas the latency for DCVs in neurites was even much longer (a τ value could not be calculated but a significant percentage of events appeared 1 minute after stimulation) (Fig. 3B). This was in spite of the fact that Ca²⁺ elevations with depolarization were maximal within 2 seconds and exhibited no latency (Fig. 2). Long latencies to fusion indicate that the coupling between Ca²⁺ influx and exocytosis by plasma-membrane-proximal DCVs in hippocampal neurons is surprisingly `loose'. By contrast, time constants for SV exocytosis in boutons are <0.001 seconds (Sabatini and Regehr, 1999), whereas those for DCV exocytosis in endocrine cells are 0.01-1 seconds (Barg et al., 2002; Chow et al., 1996; Elhamdani et al., 1998; Martin, 2003). Long latencies to fusion might reflect a lack of spatial localization between vesicles and Ca²⁺ channels (Chow et al., 1996; Elhamdani et al., 1998). This suggests that plasma-membrane-proximal DCVs in the soma of cultured hippocampal neurons are not in close proximity to L-type Ca²⁺ channels (see below).

Stream acquisition allowed us to image single DCV exocytic events at a time resolution of 0.1 seconds. Individual DCV exocytic events triggered by depolarization were typically characterized by a transient increase in fluorescence followed by a decrease in fluorescence to pre-exocytosis levels or even lower (Fig. 3C,D). The initial increase in fluorescence results from fusion-pore formation, with proton efflux from the acidic DCVs, which increases EGFP fluorescence (Taraska et al., 2003). Additional increased fluorescence was caused by the release of ANF-EGFP into the evanescent field, which was evident as the appearance of fluorescent clouds around the DCV (supplementary material Movies 2, 3). The subsequent decrease in fluorescence in the event time course (Fig. 3C,D) results from the rapid diffusion of ANF-EGFP from the sites of exocytosis. For events in which content release is partial, the ANF-EGFP retained in DCVs after fusion-pore closure exhibits slower decreases in fluorescence, owing to re-acidification of the DCVs. The overall post-fusion decrease in fluorescence of ANF-EGFP (ΔF) (Fig. 3C,D) is due to content release rather than movement of DCVs from the plasma membrane, because exocytosing DCVs containing BDNF-EGFP, which is not released, exhibited no net change in fluorescence (see below). Overall, the single DCV exocytic events detected with ANF-EGFP cargo in hippocampal neurons were quite similar to those described for endocrine cells (Taraska et al., 2003).

For most exocytic events in the soma for ANF-EGFP-containing DCVs (Fig. 3Ci; supplementary material Movie 2), the fluorescence intensity of the vesicles after exocytosis decreased by only a small extent (∼30%), indicating that there was incomplete cargo release. This type of event, previously observed for DCV exocytosis in endocrine cells, corresponds to `kiss-and-run' (Neher, 1993) or `cavicapture' (Henkel and Almers, 1996; Palfrey and Artalejo, 1998) events, in which fusion-pore re-closure prevents full cargo release. In several cases, the same vesicle was observed to release cargo multiple times within 1 second (Fig. 3Cii). In the neurites, DCV exocytic events were associated with significantly more loss of fluorescence than in the soma (Fig. 3D; supplementary material Movie 3). We determined the extent of cargo release (ΔF) by subtraction of the fluorescence recorded 15 seconds after an event, when vesicle re-acidification was complete, from that 2 seconds before an event. Plotting the distribution of loss of fluorescence for neurite and soma DCV exocytic events (Fig. 3E) revealed a unimodal distribution for neurite DCVs at ∼75% loss, whereas that for soma DCVs was bimodal with most DCVs exhibiting ∼30% loss and a smaller number with ∼65% loss of fluorescence. Thus, all DCVs in the soma exhibited incomplete release, whereas 50% of the DCVs in the neurites exhibited >75% fluorescence loss. A possible explanation for the greater content loss for DCVs in neurites was that exocytic events in neurites involved fuller or more prolonged fusion-pore dilation so that more content was secreted. An analysis of the time course of fusion events supported this explanation. Individual exocytic events in the neurites exhibited slower and more prolonged time courses (Fig. 3D). The rising phase of the fluorescence increase (designated t_r in Fig. 3Ci,D), which reflects proton efflux through the fusion pore as well as cargo release into the evanescent field, was much more prolonged for DCV exocytic events in the neurite compared with events in the soma (Fig. 3F). This might be due to more-prolonged dilation of the fusion pore in neurite events, which would account for increased cargo release.

Although some DCVs in neurites released the majority of their ANF-EGFP content during exocytosis, additional studies indicated that full fusion of DCVs did not occur. We expressed BDNF-EGFP (Brigadski et al., 2005) in the hippocampal neurons to image the DCVs. Colocalization studies indicated that BDNF-EGFP was targeted to the same DCVs as ANF-RFP (not shown). Again, evoked exocytic events with BDNF-EGFP cargo were detected mostly in the soma upon stimulation. Mature BDNF is a larger protein than ANF (118 versus 28 amino acids) and is a dimer that is highly condensed in the peptide core of secretory granules (Brigadski et al., 2005). Correspondingly, BDNF-EGFP may be dimerized as well as condensed in the DCV core and would require fuller fusion-pore dilation than ANF-EGFP for release. Nonetheless, if full fusion of DCVs were to occur, full release of the cargo BDNF-EGFP would be expected. Depolarization of the neurons resulted in rapid fluorescence increases of DCVs in both soma (Fig. 4A) and neurites (Fig. 4B), indicating fusion-pore formation and de-protonation of the vesicle; however, no fluorescent clouds were detected around the DCVs (supplementary material Movie 4), which indicated that little release of the BDNF-EGFP cargo occurred. After the initial brightening due to de-protonation, fluorescence declined very slowly to pre-exocytosis values as the result of vesicle re-acidification (Fig. 4). There was little net loss of fluorescence from the DCVs at 20 seconds after the event, which is consistent with little cargo release in single exocytic events. It should be noted that net release of BDNF-EGFP can be observed from the soma and dendrites of hippocampal neurons upon more prolonged depolarization (Hartmann et al., 2001; Kolarow et al., 2007), which might involve multiple DCV exocytic events. Moreover, the lack of BDNF-EGFP release in our studies is not reflective of the behavior of BDNF. Nonetheless, our results indicated that, under the present conditions, BDNF-EGFP largely served as an unreleased DCV-content marker for single exocytic events that detected vesicle alkalinization and re-acidification accompanying fusion-pore opening and re-closure, respectively. Thus, exocytosis of DCVs in both soma and neurites is followed by recapture (e.g. kiss-and-run) but, in neurites, the fusion pore dilates more or pore opening is prolonged, which allows more ANF-EGFP but not BDNF-EGFP release.

Ca²⁺ is one of the major factors that affects the opening and dilation of the fusion pore. A Ca²⁺-induced shift of exocytic modes to kiss-and-run has been reported for endocrine cells (Ales et al., 1999; Elhamdani et al., 2006). To determine whether DCV exocytosis in the soma of hippocampal neurons is regulated by Ca²⁺ levels, we decreased the extent of depolarization by reducing K⁺ levels from 90 mM to 30 mM and monitored individual exocytic events. Reducing the extent of depolarization and the accompanying Ca²⁺ influx (see Fig. 2) dramatically reduced fusion probabilities in the soma as anticipated (Fig. 5A), but the reduced Ca²⁺ level also altered the kinetics of individual DCV exocytic events, measured as changes in ANF-EGFP fluorescence (Fig. 5B). Many release events in the soma were significantly prolonged (representative example in Fig. 5B), with an average t_r value of 0.58 seconds, which was very similar to that in neurites (0.65 seconds) and significantly longer than that in soma with strong depolarization (0.17 seconds) (Fig. 5C). Analysis of the loss of fluorescence accompanying exocytosis showed that a majority of DCVs in the soma exhibited fuller content release (∼80%) at lower depolarization (Fig. 5D). Thus, whereas DCV fusion probabilities are increased at higher stimulus strength as anticipated (Fig. 5A), the shift to faster fusion-pore re-closure at higher stimulus strength reduced the amount of peptide secreted per event (Fig. 5D). This implies that complete DCV cargo release at higher stimulus strength would require several rounds of exocytosis (as in Fig. 3Cii), which would slow the time course of evoked peptide secretion.

Several pharmacologically distinct Ca²⁺ channels, including L-, N-, Q- and P-type channels (Trimmer and Rhodes, 2004; Wheeler et al., 1994), have been identified in hippocampal neurons. To determine which Ca²⁺ channels mediate Ca²⁺ influx for DCV exocytosis in the soma, we used selective blockers. Verapamil, an antagonist for L-type Ca²⁺ channels (Godfraind et al., 1986), completely blocked all DCV exocytic events (Fig. 6), which is consistent with previous reports that nifedipine blocks evoked BDNF-EGFP release (Kolarow et al., 2007). By contrast, 10 μM ω-conotoxin MVIIC, which blocks both N- and Q-type Ca²⁺ channels (Sather et al., 1993), and 200 nM ω-agatoxin TK, an antagonist that is specific for P-type Ca²⁺ channels (Kuwada et al., 1994), had no effect on release probabilities (Fig. 6). These results suggest that L-type Ca²⁺ channels exclusively mediate the Ca²⁺ influx that triggers DCV exocytosis in the soma of hippocampal neurons. Previous work indicated that L-type Ca²⁺ channels are largely localized to somatodendritic regions (Pravettoni et al., 2000), where they mediate depolarization-induced entry of Ca²⁺, which regulates gene transcription, as well as dynorphin and BDNF secretion in cultured hippocampal neurons (Brosenitsch et al., 1998; Kolarow et al., 2007; Simmons et al., 1995; Yamamoto et al., 2005). Our results indicate that L-type channels mediate the endocrine-like secretion of neuropeptides and neurotrophins that occurs in the soma of central neurons. The distribution of L-type Ca²⁺ channels might also underlie the differences in release probability and fusion-pore kinetics observed above for DCVs in the cell body and neurites.

Our results of imaging single peptidergic DCV exocytic events in hippocampal neurons led to four major observations. First, DCVs in the soma of neurons exhibited significantly higher release probabilities than those in neurites. Second, the kinetics of DCV fusion events in the soma and neurites differed and the kinetics of DCV fusion events in the soma were tightly regulated by levels of Ca²⁺ influx through L-type Ca²⁺ channels. Third, virtually all DCV fusion events in the soma and neurites used a kiss-and-run or cavicapture mode. Fourth, depolarization-evoked DCV exocytic events exhibited surprisingly long latencies to fusion. Overall, these findings contribute to understanding a major mode of chemical signaling in the nervous system that employs neuropeptides and neurotrophins.

We used peptide-EGFP constructs that are properly targeted to DCVs in neurons (Fig. 1). Although the ANF-EGFP and BDNF-EGFP fusion proteins might exhibit release properties that differ from those of the native peptides, we took advantage of the DCV-targeted EGFP to determine the characteristics of single DCV exocytic events. Fusion-pore formation with vesicle de-protonation at the early stage of exocytosis was detected as a brightening of the pH-sensitive EGFP. For the larger, less readily released BDNF-EGFP, fusion-pore re-closure and slow vesicle re-acidification were detected by the subsequent decrease in fluorescence. Fluorescence losses for the smaller releasable ANF-EGFP indicated the relative extent or duration of fusion-pore dilation under different experimental conditions. Thus, fluorescence changes of the peptide-EGFP enabled the detection of individual exocytic events, their latencies following stimulation, and the kinetics of fusion-pore opening and re-closure.

Evoked DCV exocytosis occurred mainly from the soma of hippocampal neurons. This is consistent with electron-microscopic studies that revealed high densities of DCVs in the dendrites and soma of central neurons, as well as exocytic profiles distant from synaptic regions (Ludwig and Leng, 2006; Morris and Pow, 1991). The surprisingly high release probability of DCVs in the soma in response to stimulation (Fig. 3A) has not previously been appreciated and indicates a major physiological role for somatic peptide release. An important role of somatic neuropeptide and amine release is thought to be to provide negative feedback via autoreceptors on the cell body (Huang et al., 2007). In addition, peptides secreted from the soma in an endocrine-like manner would diffuse to adjacent neurons and exert paracrine regulation allowing the coordination of neuron populations innervating similar or different targets (Zaidi and Matthews, 1999). Although we did not detect substantial release of EGFP-modified BDNF in single DCV exocytic events, native BDNF release does occur from the soma. BDNF release from the soma is physiologically important because presynaptic GABAergic terminals, which are highly regulated by BDNF (Kohara et al., 2007), are formed preferentially onto the soma.

DCV exocytic events in neurites, where SV exocytic events predominate, were rare and exhibited even longer latencies to fusion than somatic events. However, a limitation of the TIRF microscopy is that only DCV exocytic events within 100-200 nm of the coverslip would be detected, which might preclude imaging of release events that occurred on dorsal regions of the neuron. However, a previous study using epifluorescence microscopy found that only a small fraction (∼10%) of neurotrophins are released synaptically during prolonged stimulation (>300 seconds) with a time constant >100 seconds (Brigadski et al., 2005) or in response to high-frequency tetanic stimulation (Hartmann et al., 2001), which is consistent with the current results. It is possible that DCVs in neurites, which exhibit low release probabilities, are older, having undergone longer transport, than those in the soma, and it has been shown in chromaffin cells that younger DCVs undergo preferential release (Duncan et al., 2003). The rare DCV exocytic events that we detected for ANF-EGFP-containing DCVs in neurites resulted in fuller peptide release per event, apparently owing to wider or more-prolonged fusion-pore opening. The mechanism underlying differences in release probabilities and fusion-pore kinetics in soma and neurites was further investigated and found to probably reside with the different effective Ca²⁺ levels specifically mediated by L-type Ca²⁺ channels. In endocrine cells, Ca²⁺ levels during stimulation regulate DCV fusion probabilities in addition to fusion-pore dilation and closure (Ales et al., 1999; Elhamdani et al., 2006; Wang et al., 2006). Our results indicated a strong dependence of DCV fusion probabilities on stimulus strength and Ca²⁺ levels, as would be anticipated. However, higher stimulus strength also promoted faster fusion-pore kinetics, which might result from either Ca²⁺ stimulation of fusion-pore dilation accompanied by spontaneous closure (Wang et al., 2006), or from Ca²⁺ stimulation of `rapid endocytosis' or pore re-closure (Ales et al., 1999; Elhamdani et al., 2006). In either case, the prevalence of L-type Ca²⁺ channels in soma is well known (Trimmer and Rhodes, 2004), which would provide somatic DCVs with higher release probabilities as well as increased fusion-pore closure rates characteristic of high Ca²⁺ stimulation.

Previous studies found that high-K⁺ depolarization and high-frequency stimulation elicited similar rates of BDNF-EGFP secretion from hippocampal neurons (Hartmann et al., 2001). Action potentials have been shown to be a sufficient stimulus for release of BDNF-GFP from dendrites (Kuczewski et al., 2008). However, because of the uncertainty of how far action potentials propagate and promote Ca²⁺ elevations in various neuronal regions, we employed high-K⁺ solutions as a more-defined stimulus. Nonetheless, because of the strong Ca²⁺ dependence of the kinetics, it will be of interest to re-examine the release probabilities and kinetics of individual DCV exocytic events in response to action potentials.

Despite the difference in fusion-pore kinetics, DCV exocytic events in both soma and neurites occurred by `kiss-and-run' or `cavicapture' events. This was indicated by the incomplete release of peptide cargo, and by the sequential brightening and dimming of unreleased BDNF-EGFP that marked the initial de-protonation and subsequent re-acidification of the DCV. The prevalence of `kiss-and-run' versus full fusion for SVs has been controversial and estimated at <20% to >80% depending on the neuron type and stimulation conditions (Ales et al., 1999; Aravanis et al., 2003; Elhamdani et al., 2006; Richards et al., 2005; Stevens and Williams, 2000). Alternatively, the `kiss-and-run' mode for SVs in nerve terminals has been attributed to the selective sampling of a stochastic endocytosis process (Balaji and Ryan, 2007). By contrast, most DCVs in cultured endocrine cells use `kiss-and-run' or a `cavicapture' mode of exocytosis (Elhamdani et al., 2006; Taraska et al., 2003). Our results indicate that `kiss-and-run' is the major mode for the DCV exocytotic events observed by TIRF in cultured hippocampal neurons in soma as well as neurites. By using this mode of exocytosis, multiple bouts of peptide secretion can occur over an extended time period without the need to reassemble vesicular components following endocytosis or the need to package new cargo in the Golgi complex (Henkel and Almers, 1996; Palfrey and Artalejo, 1998). Repeated fusion-pore openings that were observed in the hippocampal neurons (Fig. 3Cii) provide a way to control the amount of peptide released and regulate the composition of the released peptide mixture depending upon the extent of fusion-pore opening (Staal et al., 2004). It remains to be determined how relevant the kiss-and-run mode would be for any specific neuropeptide because release kinetics during fusion-pore opening are determined by the size, aggregation state and diffusibility of the peptide (Barg et al., 2002; Taraska et al., 2003). It can be calculated that BDNF-EGFP would diffuse ∼1.8 times more slowly than BDNF based on size differences alone (86 vs 28 kD). A need for repeated openings to enable substantial release of large neuropeptide and neurotrophin cargo might be the basis for the observed strong dependence of net release on prolonged stimulation (Hartmann et al., 2001). The functional relevance of different exocytic kinetics in soma and neurites might reside with the preferential release of peptides of different sizes from the similar peptide cargo mixtures in the DCVs of the soma and neurites.

DCV exocytosis in soma and neurites exhibited surprisingly long latencies for stimulation-secretion coupling. Previous studies found that the stimulated secretion of neuropeptides and neurotrophins is slow compared with rapid, synchronous synaptic transmitter release (Brigadski et al., 2005). Slow evoked neuropeptide secretion has previously been attributed to the cytoplasmic location of DCVs, with long transit times to the plasma membrane (Shakiryanova et al., 2006), or to the slow dissolution of DCV core contents (Brigadski et al., 2005) and the size-dependent diffusion of peptides through a limiting fusion pore (Barg et al., 2002). Our results indicate that a major contributing factor to the slow kinetics of peptide release is the long latency to DCV fusion even for plasma-membrane-proximal DCVs and especially for DCVs in neurites. A recent study showed similar long latencies to DCV fusion-pore openings (Kolarow et al., 2007), but did not provide a quantitative analysis of these events. A second major contributing factor to the slow kinetics of peptide release would be rapid fusion-pore re-closure at high stimulus strengths that would necessitate multiple rounds of `kiss-and-run' to mediate net peptide release. Overall, our studies imply a predominant endocrine-like release process for neuropeptides that occurs mainly from somatic regions. In contrast to the extensive studies of depolarization evoked SV exocytosis in similar hippocampal cultures (Aravanis et al., 2003; Gandhi and Stevens, 2003; Sabatini and Regehr, 1999; Wheeler et al., 1994), our results highlight striking differences in the Ca²⁺-triggered exocytosis of neuronal DCVs and SVs, motivating additional mechanistic studies.

Cultures of rat hippocampal CA3-CA1 neurons were prepared from embryonic day 18 (E18) Sprague-Dawley rats by described methods (Banker and Goslin, 1998). Cultures contained <10% astrocytes as indicated by morphology and the finding that the majority of cells expressed GFP driven from the neuron-specific synapsin promoter in a lentiviral construct (Scott and Lois, 2005). In addition, the depolarization-evoked ANF-EGFP secretion reported herein was characteristic of neurons but not astrocytes when similar studies were conducted with purified astrocyte cultures. Neuronal cultures were transfected with constructs encoding ANF-EGFP [kindly provided by Edwin Levitan (Burke et al., 1997; Shakiryanova et al., 2006)] or BDNF-EGFP (Brigadski et al., 2005) using calcium phosphate at 6-9 days in vitro (div 6-9) and imaged 2 days later. Before stimulation, cells were washed with 25 mM HEPES buffer containing 119 mM NaCl, 2.5 mM KCl, 2 mM CaCl₂, 2 mM MgCl₂, 30 mM glucose, 10 μM 6-cyano-7-nitroquinoxaline-2,3-dione (CNQX) and 50 μM D,L-2-amino-5-phosphonovaleric acid (D,L-AP5) (pH 7.4). The stimulation buffer had the same composition except that the KCl concentration was elevated to 90 mM (or 30 mM) and the NaCl concentration adjusted to 31.5 mM (or 91.5 mM) to keep ionic strength constant. All buffers were balanced with sucrose to 315 mmol kg^–1 osM^–1. Experiments were conducted on the microscope stage in a room at 23°C but stimulation buffers were pre-warmed to 37°C. Control studies showed that the same numbers of evoked DCV exocytic events were observed when the experiments were carried out on a heated stage at 37°C.

Cells were grown in glass-bottom dishes (MatTek) for immunocytochemistry. Cells were washed with phosphate buffered saline (PBS), fixed with 4% formaldehyde (wt/vol), permeabilized with PBS containing 0.3% Triton X-100, and blocked with 10% donkey serum before antibody incubations. Primary and secondary antibodies were diluted in serum blocking solution. Cells were washed extensively with PBS after secondary-antibody incubation and left in PBS for visualization by TIRFM. Antibodies used were anti-SgII monoclonal (Abcam); anti-SytI monoclonal (clone 604.1; Synaptic Systems); anti-VMAT2 polyclonal (Pel-Freez); and anti-BDNF polyclonal (Promega).

Cells were grown on No. 1.5 cover glasses (thickness 0.16-0.19 mm) and viewed with an inverted microscope (Nikon Eclipse TE2000-U) equipped for TIRF with 488 nm and 543 nm lasers with a 1.45-numerical-aperture TIRF objective (Nikon plan Apo TIRF, 100×). The field in the TIRF microscope was calibrated by measuring the intensity of 0.1 μm and 0.5 μm fluorescent Tetraspeck beads (Invitrogen, Carlsbad, CA) and 1.0 μm Fluoresbrite beads (Polysciences, Warrington, PA) by epifluorescence and in the evanescent field at 480 nm excitation in BSA-containing buffers with a refractive index of 1.37 equivalent to cytosol (Steyer and Almers, 1999). The exponential evanescent field exhibited a calculated penetration depth (1/e) of ∼80 nm at the incident angle used for these studies. Fluorescent images were recorded by a charge-coupled device (CCD) camera (Photometrics CoolSNAP ES, Roper Scientific, Tucson, AZ). Images were acquired at up to 10 Hz using the stream acquisition function of the Metamorph software (Universal Imaging, Downingtown, PA). Regions of interest (ROI) were automatically determined by Metamorph and refined manually. Each vesicle was encompassed by a 400-nm diameter circle for intensity measurement and the circle was manually moved to an adjacent area without granules to determine local background. Fluorescence intensity was quantified by the region measurement function of Metamorph and exported to Microsoft Excel for further analysis. Exocytotic events were defined as abrupt fluorescence increases (>50% increase) followed by a decrease in fluorescence intensity. Unpaired t-test was used to evaluate statistical significance. Non-linear curve fitting was done using Microcal Origin 6.0 software (OriginLab, Northampton, MA).