Synaptotagmin isoforms confer distinct activation kinetics and dynamics to chromaffin cell granules

Regulated exocytosis in chromaffin cells is triggered by membrane depolarization and subsequent Ca²⁺ influx through voltage-gated channels. The level of Ca²⁺ accumulation is commensurate with the strength of stimulation (Douglas and Rubin, 1961; Neher and Augustine, 1992; Fulop and Smith, 2007; de Diego et al., 2008). Ca²⁺ drives exocytosis through the Ca²⁺-binding synaptotagmin (Syt) protein family (Brose et al., 1992; Voets et al., 2001a; Schonn et al., 2008). The Syt protein family includes 17 isoforms, but only two of these isoforms (Syt-1 and Syt-7) are known to be expressed on chromaffin cell dense core granules (Schonn et al., 2008). Both Syt isoforms harbor an N-terminal transmembrane domain that extends into the lumen of the chromaffin granule, followed by two cytosolic C2 domains (C2A and C2B) connected by a short linker region (Perin et al., 1990, 1991; Chapman, 2002). The Ca²⁺- and membrane-binding properties of these isoforms are determined primarily by the amino acid sequence within the tandem C2 domains (Sutton et al., 1995; Ubach et al., 1998; Fernandez et al., 2001). Biochemical studies have established several differences in how these isoforms respond to Ca²⁺. For example, Syt-7 is capable of binding a total of six Ca²⁺ ions, while Syt-1 can bind to only five (Südhof and Rizo, 1996; Ubach et al., 1998). Although both proteins bind membranes in a Ca²⁺-dependent manner, Syt-7 does so with a 10-fold higher sensitivity for Ca²⁺ ions compared with Syt-1 (Sugita et al., 2002; Bhalla et al., 2005).

The notion that granule or vesicle proteins may confer spatiotemporal heterogeneity to fusion events has recently become more widely appreciated. At synapses, there is evidence that vesicle-associated membrane protein/synaptobrevin isoforms may act to sort vesicles into synchronous, asynchronous, and spontaneously fusing populations (Raingo et al., 2012; Bal et al., 2013; Crawford and Kavalali, 2015). Syt isoforms may serve similar functions in neurons and neuroendocrine cells (Walter et al., 2011; Raingo et al., 2012; Bacaj et al., 2013; Bal et al., 2013; Weber et al., 2014; Crawford and Kavalali, 2015; Lee and Littleton, 2015; Luo et al., 2015). In chromaffin cells, Syt-1 and Syt-7 are thought to drive the bulk of Ca²⁺-triggered exocytosis. When both isoforms are eliminated, the overall secretory capacity of the cell is reduced by greater than 70% (Schonn et al., 2008). Elimination of one isoform at a time revealed that Syt-7 likely accounts for the slow phase of exocytotic release (as revealed by membrane capacitance measurements), while Syt-1 likely accounts for the fast phase (Schonn et al., 2008).

The biochemical differences between Syt-1 and Syt-7 may endow granules with different functional properties during exocytosis. Previous work from our group demonstrated that Syt-1 and Syt-7 are sorted to different populations of chromaffin granules (Rao et al., 2014). Syt-7 granule fusion is triggered by milder membrane depolarization than Syt-1 granule fusion, and fusion pores of granules harboring Syt-1 expand more rapidly than pores of granules expressing Syt-7. Although the underlying mechanisms of these phenomena are unclear, they may result from differential affinities of Syt isoforms for Ca²⁺, phospholipids, and/or effector proteins. The striking differences observed between Syt-1 and Syt-7 granules motivated the work described here, which further clarifies the nature of Syt-mediated differences in granule properties.

In this study, we assessed whether the distinct dynamics and activation kinetics of chromaffin granules could be attributed to the selective sorting of Syt isoforms. Our data reveal that although the cytosolic distribution of granules bearing Syt-1 or Syt-7 is largely overlapping, their mean distances from the membrane were not identical. Although “newcomer” (i.e., vesicles not evident in the evanescent field at the start of stimulation) fusion events accounted for a relatively small percentage of total fusion events, their number was greater for granules with Syt-1 than Syt-7. Syt-1 granules also exhibited a greater tendency to fuse when chromaffin cells were repeatedly challenged with a strongly depolarizing stimulus, suggesting differences in their fusion “readiness” or, potentially, differences in their distribution with respect to sites of exocytosis on the membrane. Indeed, image analysis reveals that Syt-7 granules tend to fuse in clusters, whereas fusion locations of Syt-1 granules are more dispersed. The two populations of granules were also differentially dependent on Ca²⁺ channel subtypes for fusion, consistent with the theme of exocytotic site heterogeneity. Strikingly, the two granule populations responded differentially to increases in cytoplasmic Ca²⁺. Direct cytosolic introduction of Ca²⁺ at defined concentrations demonstrated that the Syt-7 granule population is likely to drive the bulk of exocytosis when cells are weakly stimulated. Syt-1 granules required a comparatively much larger increase in intracellular Ca²⁺ for fusion. At the highest concentrations of Ca²⁺ tested, comparable to a strongly depolarizing electrophysiological stimulus, Syt-1 granules fused with faster kinetics than Syt-7 granules. Overall, these findings demonstrate that chromaffin granules are not homogeneous but are composed of at least two populations with distinct functional properties specified by different Syt isoforms and their responses to Ca²⁺.

The chromaffin cells used for these experiments were isolated from the adrenal medullae of adult cows (Bos taurus) obtained from JBS Packing Company. Transient transfections were performed as described earlier (Wick et al., 1993). In brief, cells were plated on 35-mm glass-bottom tissue culture dishes (refractive index 1.51; World Precision Instruments) precoated with poly-d-lysine and bovine collagen. Cells were transiently transfected by electroporation using the Neon transfection system (Invitrogen). After standardization of the electroporation protocol, the chromaffin cells were transfected with the desired plasmid(s) with a single pulse of 1,100 mV for a period of 40 ms. The media on the cells were changed after overnight incubation, and all the experiments were performed 48 h after transfection to ensure expression of the fusion protein. All procedures were performed after approval was obtained from the Institutional Animal Care and Use Committee.

All fluorescent proteins fused to Syt were located upstream of the N-terminal lumenal domain. For cargo proteins (e.g., neuropeptide Y [NPY]), the fluorescent protein followed the C-terminal region. The parent NPY plasmid was a gift of Wolfhard Almers (Vollum Institute, Portland, OR). All Syt constructs used were made in pCI vectors with coding sequences from Rattus norvegicus. The linker sequence between the upstream fluorescent protein tag for both Syts is 5′-GAATTCTAGACAGGCGGAAGCGGAGGCTCCGGCGGGAGCGGCGGA-3′. A bovine preprolactin signal sequence is located further upstream of the pHluorin/GFP/pHuji tag. The pHluorin and pHuji tags can be removed from the plasmid by cutting with upstream (Acc65I, KpnI, or MluI) and downstream (XbaI) enzymes. The GFP used here is the “enhanced” GFP variant.

Immunofluorescence microscopy was performed to detect the distribution of Syt isoforms within chromaffin cells. Cells were first fixed with 4% PFA in PBS for 30 min at 37°C. The cells were quickly rinsed after the incubation and quenched in a solution of 50 mM NH₄Cl in PBS for 30 min at 37°C. After washing off the quenching solution, the cells were permeabilized using acetone solution for 7 min. After the incubation, the cells were washed with TBS twice and blocked with 1% gelatin in TBS for 20 min, followed by 4% donkey serum for 30 min, both at 37°C. A solution of 2–4 mg/ml BSA made in TBS was used to dilute the primary and secondary antibodies. Cells were incubated for 2 h at 37°C with a combination of the following primary antibodies: monoclonal mouse anti–Syt-1 antibody (antibody 41.1, catalog no. 105 011) and polyclonal rabbit anti–Syt-7 antibody (catalog no. 105 173) from Synaptic Systems. After incubation, the cells were washed five times with TBS and incubated for 70 min with Alexa-conjugated anti-rabbit and anti-mouse secondary antibodies (Molecular Probes/Invitrogen). Finally the cells were washed five times after the secondary incubation and imaged by confocal microscopy.

Images of immunostained endogenous Syt puncta were acquired on a Leica Biosystems TCS SP5 confocal microscope with a 63×/1.40 NA oil objective. Images of cells transfected with either Syt-1 or Syt-7 GFP were acquired on a ZEISS LSM 780 confocal microscope with a W Plan-Apochromat 63×/1.0 NA M27 water immersion objective. For imaging, a 405-nm diode laser, a 488-nm argon laser, and a HeNe red 633-nm laser were used on both the confocal microscopes. The images were analyzed with Imaris (Bitplane) software (see image analysis below). Statistical calculations where needed were performed using Prism 6 software (GraphPad Software).

Cells were quickly washed two or three times in basal physiological salt solution (PSS) to get rid of any residual media and then used for the secretion experiments. For the total internal reflection fluorescence (TIRF) experiments, cells were perfused with PSS containing 145 mM NaCl, 5.6 mM KCl, 2.2 mM CaCl₂, 0.5 mM MgCl₂, 5.6 mM glucose, and 15 mM HEPES, pH 7.4, for 10 s and then stimulated with high-K⁺ solution containing either 56 mM KCl (95 mM NaCl, 56 mM KCl, 5 mM CaCl₂, 0.5 mM MgCl₂, 5.6 mM glucose, and 15 mM HEPES, pH 7.4) or 25 mM KCl (126 mM NaCl, 25 mM KCl, 5 mM CaCl₂, 0.5 mM MgCl₂, 5.6 mM glucose, and 15 mM HEPES, pH 7.4) to trigger exocytosis for ∼60 s. Individual cells were perfused with a needle (100-µm inner diameter) under positive pressure applied using the ALA-VM4 perfusion system (ALA Scientific Instruments).

For repeated stimulation experiments, chromaffin cells transfected with GCaMP5G, Syt-1, or Syt-7 GFP were depolarized with elevated K⁺ (56 mM KCl) during subsequent rounds of stimulation. The cells were initially perfused with 56 mM KCl for 30 s and then exposed to basal PSS for either 10 or 60 s to allow clearance of Ca²⁺ from within the cytoplasm. After this recovery period, the cells were again exposed to elevated K⁺ one or more times for a period of 15 s, with intermittent application of basal PSS (10 s; see Fig. 6).

TIRF imaging was performed on an inverted microscope (IX81; Olympus) equipped with an oil 60× objective (NA 1.49), two additional lenses (1.6× and 2×) in the emission path between the microscope, and a cooled electron-multiplying charge-coupled device camera (iXon3; Andor Technology). Two-color TIRF microscopy was performed by exciting the samples using a 488-nm argon ion laser and a 561-nm solid-state Sapphire laser (Coherent). A detailed description of this TIRF setup (along with specifics of the optical filter assembly) is provided elsewhere (Passmore et al., 2014). Lambda SC smart-shutter controllers were used for rapid switching between the excitation wavelengths (Sutter Instrument). Images were acquired with Metamorph software (Molecular Devices) successively in the green and red channels with an exposure time of 30 ms and an electron-multiplying gain of 300. The frame acquisition rate was ∼4–5 Hz for a pair of images. The final pixel size was ∼80 nm.

An Olympus cellTIRF-4Line unit was used to obtain the dual-color imaging results reported in Fig. S2. GFP and mCherry fluorescent fusion proteins were sequentially excited with 488- and 561-nm lasers. Emission was separated on an OptoSplit II system from Cairn Research. Optical filters in the OptoSplit were obtained from Semrock (FF01-514/30-25, FF02-617/73, and dichroic FF580-FDi01). A sequential pair of images were acquired at 4–5 Hz on an electron-multiplying charge-coupled device (iXon 897; Andor Technology). The final pixel size of the images was ∼80 nm. The microscope and all peripherals were controlled using Olympus CellSens software. All TIRF experiments were performed at ∼30°C.

We used the genetically encoded Ca²⁺ indicator GCaMP5G (Akerboom et al., 2012) to determine changes in cytosolic free Ca²⁺ evoked by different depolarizing solution challenges. Chromaffin cells transfected with GCaMP5G were plated on a 35-mm glass cover dish and imaged by TIRF microscopy (see “TIRF microscopy” above). The standard bath solution was a basal PSS (see “Cell stimulation” above). To depolarize the cells to evoke changes in cytosolic Ca²⁺, the standard stimulating solution was adjusted to contain 5.6, 10, 25, 40, 56, 80, or 100 mM KCl while maintaining overall osmolarity of the solution. Calculations for determining the intracellular Ca²⁺ concentrations were performed as described previously (Neher and Augustine, 1992; Neher, 1995; Maravall et al., 2000). A brief explanation for the same has been provided here. The mean fluorescence change of GCaMP5G in response to the different depolarizing stimuli was estimated as the ratio ΔF/F = (F − F_base)/(F_base), where F is the peak fluorescence intensity of GCaMP5G after stimulation, and F_base is the mean fluorescence intensity in the same cell before stimulation. The averaged ΔF/F values from all KCl concentrations were used to estimate the fold change in fluorescence at GCaMP5G-Ca²⁺ binding saturation (see below, [ΔF/F]_plateau). Calcium concentrations were then calculated using the following equations (Maravall et al., 2000):

where Δ[Ca²⁺] is the rise in calcium concentration, K_d is the dissociation constant (we used a K_d value of 447 ± 10 nM for GCaMP5G; Chen et al., 2013), F_max is the peak fluorescence value for each cell, F₀ is the mean baseline fluorescence before stimulation for each cell, R_f is the dynamic range of the indicator (we used an R_f value of 45.4 ± 0.9 for GCaMP5G; Chen et al., 2013), ΔF/F is the change in fluorescence divided by the baseline fluorescence, (ΔF/F)_max is the corrected ΔF/F at saturation (see below), and (ΔF/F)_max was calculated on the basis of the equation below. Saturation of GCaMP5G occurred at 80 and 100 mM KCl (see Fig. 7 A), and these values were used to determine the correction factor, x:

where x is a correction factor:

where υ1 and υ2 are the lower and higher KCl concentrations driving stimulation (80 and 100 mM) at GCaMP5G saturation, and Q is the ratio of plateau fluorescence at higher and lower KCl stimulus (Maravall et al., 2000).

The mean resting calcium concentration ([Ca²⁺]₀) within control unstimulated cells was determined using the following equation (Maravall et al., 2000):

We used the ratiometric Ca²⁺ indicator fura-2 (TEFLabs) to determine changes in cytosolic free Ca²⁺ evoked by different depolarizing solution challenges. Cells cultured on glass coverslips were used to form the bottom of a recording chamber and loaded at room temperature for 45 min with 2 µM fura-2 AM in physiological saline. Local application of bath or stimulus solutions through small-bore glass capillary tubing was achieved using a pressure-driven reservoir. The standard bath solution was a basal PSS (see “Cell stimulation” above). To depolarize the cells to evoke changes in cytosolic Ca²⁺, the standard bath solution was adjusted to contain 10, 25, or 56 mM KCl while maintaining overall osmolarity of the solution. Dye fluorescence was monitored using a monochromator-based digital imaging system (T.I.L.L.-Photonics) coupled to a Nikon TE-2000 microscope. Cells were alternately excited at 340/380 ± 15 nm reflected onto the plane of focus using a DM400 dichroic mirror and Super Fluor 40×, 1.3 NA oil immersion objective. The emitted fluorescence was collected through a 510 ± 25 nm band-pass filter (Chroma Technology Corp.).

For dye calibration, we used the in vivo procedure in which the calibration constant values for R_min, R_max, and K_eff were determined from three to five cells dialyzed through patch pipette with standard internal recording solution adjusted to contain either minimal Ca²⁺ (10 mM EGTA, no added Ca²⁺) or saturating Ca²⁺ (3 mM Ca²⁺) concentrations and containing 150 µM fura-2 (Giovannucci and Stuenkel, 1997). Values for R_min, R_max, and K_eff were 0.44, 5.52 and 972 nM, respectively.

Permeabilized chromaffin cell experiments were performed as described elsewhere (Bittner and Holz, 1992). First the nutrient media were washed off the dish and replaced with sodium glutamate solution (NaGEP) containing 139 mM sodium glutamate, 20 mM PIPES, 2 mM MgATP, and 0.2 mM EGTA, pH 6.6, with no added Ca²⁺. The EGTA in the bath solution helps control the levels of free calcium in a precise and reproducible manner. To extend the range of concentrations tested above 30 µM, nitrilotriacetic acid was used in place of EGTA to chelate Ca²⁺. Individual cells were then perfused with NaGEP solution (0.2 mM EGTA) for 10 s and then permeabilized by switching to NaGEP solution (5 mM EGTA) containing 40 µM digitonin for 50 s with varying concentrations of free Ca²⁺ (1, 5, 10, 30, and 100 µM). The amount of free Ca²⁺ in the solution was calculated using the Martell and Smith constants for EGTA-Ca²⁺ binding. A stock NaGEP solution with 30 µM free Ca²⁺ was made using 4.693 mM CaCl₂ (total Ca²⁺). Other concentrations of Ca²⁺ were made by mixing the Ca²⁺-free and the 30-µM Ca²⁺ stock solutions together in the appropriate ratios. NaGEP solution with 100 µM free Ca²⁺ was also made using 4.97 mM CaCl₂ (total Ca²⁺).

Stock solution of nifedipine (L-type; Sigma-Aldrich) was prepared in DMSO at a concentration of 10 mM and diluted with the bath solution (PSS; see “Cell stimulation”) to the required final concentration of 3 µM. ω-Conotoxin GVIA (N-type) and ω-conotoxin MVIIC (N/P/Q-type; Sigma-Aldrich) were prepared in distilled water to a stock concentration of 1 mg/ml and kept in aliquots at −20°C until use. ω-Conotoxin GVIA and ω-conotoxin MVIIC were used at final concentrations of 1 and 3 µM in PSS, respectively. Chromaffin cells transfected separately with GCaMP5G, Syt-1, or Syt-7 GFP were incubated with the above-listed Ca²⁺ channel blockers alone or in combination. The effect of channel blockers was then determined using TIRF microscopy after stimulation of exocytosis with 56 mM KCl. The control cells were imaged before application of the blockers. The percentage reduction in exocytosis is based on the difference in the number of fusion events between control and treated cells.

The images obtained from confocal microscopy were analyzed using a software-based approach with Imaris using the “spot function” module for object-based colocalization analysis. To determining the distance of the Syt granules from the nearest point on the cell surface, confocal z-sections were used to render a 3D surface in Imaris. The Syt GFP granules within the evanescent field were tracked using a custom script (Degtyar et al., 2007) developed in IDL (ITT).

The images obtained from TIRF microscopy were analyzed in multiple ways. Determination of Syt GFP fusion events were identified primarily on the basis of the increase in fluorescence after fusion with the plasma membrane (see explanation below). We selected events on the basis of a minimum 20% increase in GFP fluorescence after fusion. In some cases, the initial rise in fluorescence was followed by a decrease as Syt GFP diffuses along the plasma membrane. Fusion of GFP containing granules with the plasma membrane results in brightening of fluorescence for two reasons: (1) GFP (pKa 5.9) present within the granule lumen encounters the neutral pH (7.4) of the extracellular medium, causing slight dequenching of the fluorophore, and (2) the increased strength of the evanescent field at the plasma membrane–coverslip interface causes greater excitation of the GFP molecules (Kneen et al., 1998; Tsien, 1998; Campbell and Choy, 2001).

The Syt-1 pHluorin/Syt-7 pHuji emission images were captured sequentially using Metamorph software. Individual granules undergoing exocytosis were evident by an increase in the pHluorin/pHuji intensity after fusion with the plasma membrane because of the change in pH of the granule. The changes in the fluorescence intensity over time for the different fusion events were calculated after normalization for every fusion event using custom software written in IDL.

The possible tendency of exocytotic sites to cluster in localized regions was examined by quantitatively analyzing the distribution of exocytotic sites in comparison with random positions as generated by a computer. This was done with a custom program written in IDL, with a multistep protocol that built on algorithms written for a previous study (Allersma et al., 2004). On a stack of sequential experimental images, the (x,y) positions of exocytotic events are located by eye as sudden increases in local fluorescence. These locations are recorded by mouse click without regard to which frame the event occurred during. The total number of selected events is recorded. Then, the distance r = (x² + y²)^1/2 between every pair of events is calculated and grouped in bins of range nΔr to (n + 1)Δr, where Δr is a user-settable width of each bin, and n is an integer ranging from zero to some maximum value such that the maximum (n + 1)Δr is larger than the maximum dimension of a cell. The number of events that fall into each bin, plotted against the radius of each bin, is the pairwise distance histogram of the experimental exocytotic events.

That experimental histogram must be compared with the histogram expected from a random distribution of simulated event locations. The simulated histogram can be generated by assigning locations independently to x and y with a random number generator. It is essential, however, that randomly selected (x,y) locations be accepted only if they fall within the perimeter of the cell footprint (which must be outlined in the operation of the software program). Any other (x,y) position must be rejected. The generation of successful locations continues until there are exactly the same number of simulated events as was seen in the experimental data. Then, a histogram for simulated random positions is generated exactly the same way as for experimental data. Because each simulation produces a different scatter of event locations, the simulation procedure can be repeated very quickly as often as necessary to gain a good estimate of the mean and SD of the simulated histograms. A set of “difference histograms” (H_D vs. r) can be produced, showing the difference between the actual experimental histogram and each simulated histogram. The set of H_D versus r histograms has the same SD as the simulated histograms themselves. If H_D falls within an SD of zero, then the experimental event distribution is not distinguishable from a random distribution. If H_D versus r is at least several SDs different from zero for at least some radii, then the experimental distribution cannot be random. Depending on which range of radii shows H_D significantly greater than zero allows us to conclude that whether the actual events are clustered (H_D > 0 for small radii) or anticlustered (H_D > 0 for large radii).

Fig. S1 shows quantitation of intracellular Ca²⁺ levels in depolarized chromaffin cells using TIRF microscopy and epifluorescence imaging techniques. Fig. S2 shows dual-color imaging experiments performed to determine mobility and fusion efficacy of granules harboring NPY mCherry and either Syt-1 or Syt-7 GFP. Fig. S3 shows the stimulation dependence of exocytosis from docked and newcomer Syt-1 or Syt-7 granule populations. Fig. S4 shows analysis of the intracellular distribution of granules harboring Syt-1 or Syt-7. Fig. S5 shows the effects of repeated stimulation on the near-membrane levels of Ca²⁺using gCaMP5G as a molecular indicator.

Chromaffin cells granules have traditionally been segregated into “pools” defined by their fusion kinetics (measured by increases in membrane capacitance) in response to electrophysiological stimulation (Voets et al., 1999; Voets, 2000; Sørensen, 2004). We sought to approximate this approach using a combination of chemical stimulation and TIRF imaging of individual Syt fusion events. Our initial purpose was to determine if there were differences in activation kinetics of individual granules that could be attributed specifically to the presence of Syt-1 and Syt-7 on the granule membrane. We previously reported that Syt isoforms show little to no colocalization on granules, even when overexpressed as fluorescent fusion proteins (Rao et al., 2014). To assay for differences in fusion kinetics, we cotransfected plasmids encoding Syt-1 pHluorin and Syt-7 pHuji (Shen et al., 2014) so that both isoforms could be observed and their respective time courses of fusion precisely known. Cells expressing fluorescent Syts were identified after a brief incubation with NH₄Cl (Fernández-Alfonso et al., 2006; Rao et al., 2014) and subsequently depolarized with 25, 56, or 80 mM KCl for 40 s. There were only a handful of fusion events for Syt-1 with 25 mM KCl, but at both 56 and 80 mM depolarization, the number of fusion events was sufficiently large for us to evaluate fusion kinetics.

We observed that Syt-7 granules fused earlier than Syt-1 after depolarization with elevated KCl (Fig. 1, A–C). This can be seen in the representative images and traces of Syt-1 and Syt-7 events from a single cotransfected cell, where Syt-7 pHuji events begin almost immediately after application of 56 or 80 mM KCl (Fig. 1 B). Syt-1 events follow after a brief but measurable delay. We also compared the frequency distribution of all fusion events on the basis of the time at which they occurred after depolarization (Fig. 1 C). The events were binned into 1-s intervals and fit with a Gaussian for 56 mM KCl and 80 mM KCl. Notably, the mean for Syt-7 events was shifted to shorter times at both 56- and 80-mM stimulation conditions (7.01 and 6.10 s, respectively), in relation to Syt-1 events in the same cells (8.86 and 8.11 s, respectively).

Depolarization with high extracellular KCl drives exocytosis by elevating intracellular Ca²⁺ near fusion sites. But how comparable is the Ca²⁺ increase with KCl depolarization to what might be measured after electrophysiological depolarization? To answer this question, we quantitated Ca²⁺ levels in the evanescent field using the fluorescent Ca²⁺-binding protein GCaMP5G (Akerboom et al., 2012). GCaMP5G fluorescence increases upon Ca²⁺ binding, and this increase can be represented as a fold change (ΔF/F) in fluorescence intensity (Fig. S1 A). Binding of GCaMP5G to Ca²⁺ was saturated upon application of stronger KCl solutions (≥56 mM). Using this information, we were able to approximate the total internal Ca²⁺ concentration ([Ca²⁺]_i) resulting from application of each KCl solution on the basis of the equations described in Materials and methods (Maravall et al., 2000). We observed that the rise in total internal Ca²⁺ was directly related to the concentration of KCl used to depolarize the cell (Fig. S1 B). Even with the strongest depolarization (100 mM KCl), the [Ca²⁺] saturates at ∼5 µM. This is likely in the lower range of what has been reported for Ca²⁺ near fusion sites after electrophysiological depolarization of chromaffin cells (Augustine and Neher, 1992). We also measured intracellular Ca²⁺ changes using the ratiometric Ca²⁺ indicator fura-2 and observed a similar concentration-dependent change in [Ca²⁺]_i levels on the basis of the strength of the KCl solution (Fig. S1 C; Neher, 1995; Giovannucci and Stuenkel, 1997).

Ca²⁺ ostensibly enters cells through voltage-gated channels upon membrane depolarization. To verify that the source of Ca²⁺ for exocytosis was extracellular, we incubated chromaffin cells expressing GCaMP5G with Ca²⁺ channel blockers nifedipine (L-type), ω-conotoxin GVIA (N-type), and ω-conotoxin MVIIC (N/P/Q-type; Hernández-Guijo et al., 1998). Each blocker applied alone significantly reduced the influx of Ca²⁺ upon membrane depolarization. When the blockers were applied together, increases in Ca²⁺ fluorescence were almost completely absent (Fig. 2 A).

As one might expect, inhibition of Ca²⁺ channel activity also had downstream effects on Ca²⁺-triggered exocytosis. Interestingly, subtype-selective channel blockade differentially inhibited fusion in Syt-1 and Syt-7 granules with respect to both degree (Fig. 2 B) and kinetics (Fig. 2, C–E) of fusion. For example, all the blockers significantly reduced the fusion probability of Syt-1 GFP. But the application of nifedipine had almost no effect on Syt-7 granule fusion and, unsurprisingly, also had no effect on shifting Syt-7 fusion kinetics. In both groups, fusion kinetics were most significantly slowed after application of MVIIC (see dotted lines shifted to the right of the solid lines).

The above results suggest that Syt-1 and Syt-7 granules are likely to be differentially dependent on Ca²⁺ channel subtypes for exocytosis. To look more directly at spatial heterogeneity in granule fusion sites, we mapped fusion sites of Syt-1 and Syt-7 granules on the chromaffin cell footprint. We used an automated procedure (custom written in IDL) to determine distances between every pair of fusion locations (Fig. 3). The program compares the experimental distances with those obtained from the same number of spots placed at random within a hypothetical cell of the same size and shape footprint. The number of times a particular distance r between fusion locations occurs is plotted on the ordinate axis, while the actual r-distance (in pixels; 1 pixel = 80 nm) is plotted on the abscissa. In this way, several histograms are formed for each cell: one histogram for the actual experimental fusion locations and a whole set of histograms for the repeated sets of random fusion placements. Cartoons of an actual chromaffin cell fusion location set and footprint are shown in Fig. 3 (A and B). Fig. 3 A shows the distribution of Syt-1 GFP fusion locations, whereas Fig. 3 B shows the distribution of Syt-7 pHuji fusion locations.

From the numerous histograms generated from simulated random event locations (typically 50), a reliable SD can be established for simulated histograms within each r-distance bin. After analysis, it is evident that the experimental histogram for Syt-7 fusion events (Fig. 3 D) does not overlap much with the set of histograms generated from hypothetical random locations. In fact, the occurrence of experimental Syt-7 granule fusion events separated by 57 pixels is roughly 70% higher than what would be expected if the events occurred at random locations. A “difference histogram” H_D versus r between the experimental histogram and each of the simulated histograms can be established and plotted (Fig. 3 E for Syt-1 and Fig. 3 F for Syt-7). The H_D versus r curves have a spread given by the SD of the simulated histograms and should be significantly different from zero at any r only if the experimental events are clustered or anticlustered. The difference histograms of Fig. 3 (E and F) can be replotted to show the number of SDs by which they depart from zero, thereby showing the statistical significance of the experimental data’s departure from the mean of the simulations (Fig. 3, G and H). For many of the individual Syt-7 cells, H_D versus r is significantly greater than zero at smaller r-distances, by six deviations, showing the fusion locations tend to be closer together than would be expected from random placement. Averaging these difference histograms over all the Syt-7 cells (Fig. 3 J) still clearly shows a significant tendency toward clustering. On the other hand, the Syt-1 fusion event histograms and the random histograms are largely indistinguishable (Fig. 3, G and I). This analysis shows that Syt-1 fusion locations are randomly dispersed, and Syt-7 fusion locations are clustered.

We next asked whether granules bearing either Syt-1 or Syt-7 demonstrated differences in mobility. If some subset of granules (e.g., Syt-1) must travel a greater distance before fusion, fusion kinetics are likely to be delayed as a consequence (Fig. 1). Although z-direction movements of granules could not be reliably measured, movements of granules in x and y within the TIRF evanescent field can be readily observed and quantitated. To perform these studies, Syt-1 or Syt-7 was expressed in cells as GFP fusion proteins and tracked for tens of seconds during membrane depolarization with 56 mM KCl. First, the mobility of all granules within the evanescent field was measured, irrespective of whether they would eventually fuse. An example of a Syt-1 cell’s granule tracks is shown in Fig. 4 A. We found that docked Syt-1 granules exhibited more movement on average than docked Syt-7 granules (100 ± 12 nm from a start position to end position for Syt-1, compared with a mean of 63 ± 7.8 nm for Syt-7; Fig. 4 B). The mean x and y velocities of granules harboring Syt-1 or Syt-7 isoforms were also measured. The velocities were small, consistent with the relative overall lack of granule movement (Fig. 4, C and D).

Next, we considered only the mobility of those granules destined to fuse. Before stimulation, both Syt-1 and Syt-7 granules undergo small x and y (R) motions in the plane of the cell membrane. It is clear from monitoring these motions that granules tend to fuse at sites that are tens to hundreds of nanometers displaced from their initial positions at the start of stimulation (Fig. 5 A, solid circle). Granules were tracked within the evanescent field using a custom script developed in IDL (Degtyar et al., 2007). Representative tracks for Syt-1 and Syt-7 granules in the seconds before fusion are shown in Fig. 5 B. The red X indicates the site of fusion. We quantified this motion by taking the frame-by-frame, poststimulation ΔR expressed relative to its ΔR before stimulation. This normalization was necessary because it was apparent that certain granules are inherently more mobile than others. Fig. 5 C shows the mean ΔR (poststimulation)/ΔR (prestimulation) for fusing Syt-1 and Syt-7 granules. In the moments leading to fusion, Syt-1 and Syt-7 mobility is similar. However, in the last 200 ms before fusion, Syt-1 granules tend to travel greater distances than Syt-7 granules. The sudden increased motion is specific to fusing granules. For example, when the motion of nonfusing granules (typically within 2–5 granule diameters of fusing granules) was tracked alongside their fusing neighbors, there was no measurable increase in movement (not depicted). The displacement of individual Syt-1 and Syt-7 granules from the start position to the fusion site was also measured (Fig. 5 D). Again, we found that Syt-7 granules tend to travel a shorter distance to fusion sites, on average, than Syt-1 granules.

We validated these results with two-color imaging experiments in which both NPY mCherry and Syt GFP were expressed in chromaffin cells (Fig. S2 A). The NPY mCherry allowed us to know unambiguously that a Syt-bearing granule had actually fused and released its contents (Fig. S2, C–E). Syt and NPY exhibited a colocalization frequency of ∼20%, irrespective of the isoform expressed (Fig. S2 B). The mobility of specifically those Syt granules destined for fusion and colabeled with NPY Cherry were tracked (Fig. S2 G). Qualitatively, these results are similar to those shown in Fig. 5 D. Although both Syt-1 and Syt-7 granules exhibited more motion just before fusion, movement was greater for Syt-1 granules than for Syt-7 granules.

The previous studies showed that chromaffin granules exhibit relatively small x and y motions in the plane of the membrane before fusion. However, a minority of fusion events result from granules not initially present in the evanescent field before stimulation. Instead, these granules move to the plasma membrane from deeper within the cell during depolarization. We initially observed “newcomer” fusion events in granules harboring NPY GFP (Fig. S3 A). Interestingly, there is a greater tendency for newcomer fusion events to occur the more strongly a cell is depolarized (Fig. S3 A).

We thus asked whether Syt-1 or Syt-7 granule fusion is more likely to occur with newcomers. An example of a Syt-1 fusion event from a granule not initially in the evanescent field at the start of imaging is shown in Fig. 6 A. The granule moves into the field by 5 s after stimulation and fuses at 14.4 s after stimulation. When many cells are analyzed, it is evident that newcomers constitute a greater proportion of the total number of Syt-1 fusion events than do Syt-7 newcomers for Syt-7 events (Fig. 6 C). We also found that the likelihood of a newcomer Syt-1 granule fusing (73.9% with 56 mM KCl) is greater than the likelihood of a newcomer Syt-7 granule fusing (64.5% with 56 mM KCl). In both cases, the fusion probability is already quite high compared with the fusion probability for docked vesicles (Fig. 7). The time course of newcomer fusion shows a similar trend for both isoforms, with a greater number of newcomers fusing later in the stimulus (Fig. 6 D). Over time, as the docked pool is depleted and the membrane potential remains depolarized, additional granules from within the cell are progressively recruited for fusion.

Because a greater proportion of Syt-1 fusion events result from newcomers, we sought to identify whether there were differences in Syt granule distribution. Such differences may suggest preferential sorting to a docked or, conversely, more mobile pool found deeper within the cell. We thus used confocal microscopy to directly measure distances of fluorescent granules from a cell surface. We first analyzed the distribution of endogenous Syt within chromaffin cells using commercially available isoform-specific antibodies. After identifying fluorescent Syt puncta, confocal z-sections (0.71 µm) were used to render a 3D surface representing the approximate contours of the plasma membrane (Fig. S4 A). The images showed Syt-1 and Syt-7 puncta to share a roughly overlapping distribution, with small but significant differences in their respective distance from cell surface (Fig. S4 C). We repeated the analysis in cells overexpressing Syt-1 or Syt-7 GFP and stained with the carbocyanine dye diD. Again, the distances measured between the respective Syts and the fluorescent surface membrane were significantly different (Fig. S4, D and E).

Granule population “releasability” has typically been defined by curve fitting to exponential components of a membrane capacitance increase (Voets, 2000; Sørensen, 2004). The “mature” or “readily releasable” pool of granules, which fuses with fast kinetics, may be the first to be depleted after depolarization. Conversely, granules that fuse more slowly with respect to the start of stimulation (“slowly releasable pool”) may undergo changes that allow them to be mobilized and transition to a “fusion-ready” state. We reasoned that using TIRF microscopy, it may be possible to identify differences in releasability of granule pools by measuring differences in fusion kinetics and probability in response to elevated KCl depolarization.

Chromaffin cells expressing either Syt-1 or Syt-7 GFP were depolarized with repeated pulses of 56 mM KCl and 80 mM KCl, as shown in Fig. 7. We then calculated the percentage of docked Syt-1 and Syt-7 granules that underwent fusion. We sought to determine whether one population of granules was depleted more readily than the other, particularly at later time points in the stimulation cycle. When cells were depolarized with 56 mM KCl, the percentage of docked Syt-1 granules that fused is higher than it is for Syt-7 granules at every time point (Fig. 7 A). With 80 mM KCl, the percentage of docked Syt-1 granules fusing is higher for the second and third time points (Fig. 7 B). Interestingly, when we extended the interstimulus interval from 10 s to 60 s (the stimulation protocol was otherwise identical) fusion occurred with equal probability in both populations (Fig. 7, C and D).

We also visualized intracellular Ca²⁺ during the stimulus protocol to determine whether it continued to accumulate with time or whether it effectively recovered to baseline levels in the subplasmalemma between stimulus applications. Indeed, high amounts of residual Ca²⁺ between stimulus applications might exaggerate the level of Syt-7 granule fusion given this isoform’s higher Ca²⁺ affinity. For that reason, we monitored intracellular Ca²⁺ with the genetically encoded Ca²⁺ indicator GCaMP5G in cells subjected to multiple rounds of depolarization with 56 mM KCl as before (Fig. S5). The data show that depolarization-induced elevations in intracellular Ca²⁺ do largely recover to baseline, but not completely. Even this relatively small level of Ca²⁺ may be enough to elevate Syt-7 granule fusion probability to levels that are near Syt-1 at stimulus cycles 2 and 3.

Both Syt-1 and Syt-7 granules fuse with high efficacy when cells are depolarized with 56 or 80 mM KCl (Figs. S2 F and S3 B). At depolarization intensities that are more modest, as with 25 mM KCl, Syt-7 granules fuse with greater efficacy than Syt-1 granules (Figs. S2 F and S3 B). These experiments demonstrate that Syt isoforms do have graded responses to KCl. In biochemical assays, the Syt cytoplasmic domains are predicted to vary by greater than 10-fold in their Ca²⁺ affinities (Hui et al., 2005; Wang et al., 2005), but it is not clear that the observed KCl responses are due to differential Ca²⁺ sensitivity in Syt-1 and Syt-7.

We sought to confirm the Ca²⁺ dependence of graded, Syt-mediated fusion by adjusting, incrementally, the level of Ca²⁺ in the cytosol. To precisely control cytosolic Ca²⁺, the cells were permeabilized by a brief incubation with digitonin and perfused with bath solutions containing 1–100 µM Ca²⁺. We first calculated the fusion probabilities of Syt granule populations as a function of [Ca²⁺] (Fig. 8 A). At concentrations of 1 µM Ca²⁺ or less, fusion of Syt-1 granules is almost completely absent, while fusion of Syt-7 granules is reliably detected. Ca²⁺ must be elevated to 5 µM and higher to observe any Syt-1 granule fusion events. At much higher Ca²⁺ levels (30 and 100 µM), docked granules fuse with roughly equivalent probability. We next plotted individual fusion times of granules bearing Syt-1 or Syt-7 GFP. Histograms for Syt-1 and Syt-7 granule fusion kinetics are significantly shifted to the left as Ca²⁺ is increased (Fig. 8, B and C). Rate constants from single-exponential curves fit to the histograms are shown in Fig. 8 D. The data indicate that Syt-1 activation kinetics are more sensitive to increases in Ca²⁺ than are activation kinetics for Syt-7. As Ca²⁺ increases from 10 to 100 µM, first-order rate constants increase fourfold for Syt-1 and only twofold for Syt-7. At the higher Ca²⁺ concentrations, Syt-1 granules fused more rapidly than Syt-7 granules (Fig. 8 D).

The idea that the dynamics of the chromaffin cell secretory response may be determined by the properties of individual granules was previously posited by Chow and colleagues (Duncan et al., 2003). Using time-resolved optical imaging, their work suggested that chromaffin granules are segregated spatially and functionally in ways that may reflect sorting to slowly releasable and readily releasable pools. Before this work, the notion of functionally segregated granule pools had largely been inferred from the kinetics of membrane capacitance increases in response to flash uncaging of Ca²⁺ or strong electrophysiological stimulation (Heinemann et al., 1993, 1994).

In this study, we identified several differences in the functional properties of chromaffin granules that could be attributed specifically to the selective sorting of Syt. Among these is the finding that Syt-1 and Syt-7 differ significantly in their activation kinetics in response to membrane depolarization. With a strong stimulus (56 or 80 mM KCl), Syt-7 granules fused with faster kinetics than Syt-1 granules in the same cell (Fig. 1). This observation was made in cells coexpressing isoforms fused to pH-sensitive fluorescent tags (pHuji and pHluorin). The kinetics of activation and fusion were highly sensitive to Ca²⁺. We showed this in two ways: (1) by selective inhibition of voltage-gated channels and (2) by application of Ca²⁺ after membrane permeabilization. The channel blockade experiments demonstrated that the Syts were not equally dependent on Ca²⁺ channel subtypes for fusion. Given the highly localized nature of ion channel–mediated [Ca²⁺] changes, this result suggested that Syt-1 and Syt-7 do not share the same distribution with respect to channels. This is indicated by the fact that channel inhibition did not have identical effects on the fusion probability of Syt-1 and Syt-7 granules. It should be noted that the degree to which exocytosis is inhibited (Fig. 2 B) is not well correlated with the degree to which Ca²⁺ fluorescence is reduced by the channel blockers (Fig. 2 A). Although each of the blockers reduced Ca²⁺ fluorescence by roughly threefold, the percentage reduction in fusion was more variable. This may have been because of differences in the overexpression of GCaMP in cells, differences in the channels’ relative contribution to exocytosis, or differences in the localization of isoforms with respect to channels. These possibilities are also not mutually exclusive. Interestingly, although nifedipine and GVIA both inhibit exocytosis, the fusion kinetics of Syt-1 and Syt-7 were essentially unaffected. The most dramatic effects on exocytosis were observed with MVIIC, in terms of both fusion inhibition and slowing of fusion kinetics. For Syt-1, the rate constant in the presence of MVIIC was slowed to 39.2 s⁻¹ from 20.1 s⁻¹; for Syt-7, the rate constant with MVIIC was slowed to 8.6 s⁻¹ from 6.1 s⁻¹. It is not obvious why the N/P/Q-type blocker MVIIC had such dramatic effects on fusion kinetics but not the N-type blocker GVIA. One explanation may be tighter coupling of P/Q channels with granules (Lara et al., 1998). As a consequence of blocking these channels, exocytosis must be triggered by Ca²⁺ sources farther away, thereby slowing fusion kinetics. In bovine chromaffin cells, P/Q-type channels carry the largest proportion of total Ca²⁺ current (50%), while N- and L-type carry 30% and 20% current, respectively (García et al., 2006). Numerous in vitro studies have also shown that Syt-1 can directly bind to P/Q-, N-, and L-type Ca²⁺ channels (Charvin et al., 1997; Sheng et al., 1997; Wiser et al., 1999; Rosa et al., 2011; Mahapatra et al., 2012). One can speculate that Syt-1 granules are localized closer to Ca²⁺ channels to compensate for their relatively low Ca²⁺-binding affinity (Hui et al., 2005; Bhalla et al., 2008). In contrast, multiple in vitro studies demonstrate that Syt-7’s Ca²⁺ sensitivity for binding membranes exceeds that of Syt-1 by more than an order of magnitude (Hui et al., 2005; Wang et al., 2005; Bhalla et al., 2008). Syt-7 granules may thus not be as dependent on close coupling to Ca²⁺ channels because of Syt-7’s higher Ca²⁺ sensitivity versus that of Syt-1.

Interestingly, we do find evidence for geographic differences in preferred fusion locations of Syt-1 and Syt-7 granules. Syt-7 granules tend to fuse in locations that deviate significantly from random, consistent with clustering. On the other hand, Syt-1 granules tend to show no such fusion location bias (Fig. 3). What this means from a molecular perspective is not yet clear. It may well reflect a level of organization that places particular granule populations close to specific channels or components of the fusion machinery. The analysis approach developed in this study will be useful to interrogating this issue in subsequent studies. Indeed, this analysis approach may be broadly useful to study the clustering (or anticlustering) of proteins or protein complexes important for dynamic cellular processes of the kind described here.

Syt-7 granules are less mobile than Syt-1 granules and fuse with faster kinetics when depolarized with KCl (Fig. 1). One might assume that this satisfies the conditions of Syt-7 granules’ belonging to a “docked” and “primed” population. However, even with a strongly depolarizing KCl stimulus (100 mM KCl), we estimate that the [Ca²⁺]_i saturates at only ∼5 µM in the subplasmalemma (Fig. S1). This does not approach the typically much higher concentrations of Ca²⁺ used in flash uncaging experiments (Schonn et al., 2008). In permeabilized cells, when cytosolic Ca²⁺ was elevated to concentrations that exceeded 30 µM, Syt-1 granule fusion kinetics were faster than those for Syt-7 (Fig. 8). As such, the faster kinetics observed for Syt-7 granules under KCl stimulation may directly result from Syt-7’s higher calcium affinity rather than a tendency toward “docked” or “primed” states. It should be noted that the proportion of docked granules that fuse with high KCl depolarization (Fig. S3 B) and with “matched” concentrations of Ca²⁺ in permeabilized cells (Fig. 8 A) are not identical. However, regardless of the method used to elicit exocytosis, there is a clear distinction in the secretory responses of Syt-1 and Syt-7 granules. Small elevations from baseline in cytosolic Ca²⁺ are generally ineffective at driving fusion of granules bearing Syt-1 and much more effective at driving fusion of granules bearing Syt-7. This is again consistent with prior in vitro studies, which demonstrate that Syt-7 binds membranes and triggers fusion at far lower Ca²⁺ concentrations as compared Syt-1(Hui et al., 2005; Wang et al., 2005; Bhalla et al., 2008). As Ca²⁺ rises (in cells, ostensibly as a result of membrane depolarization), both populations of granules fuse with higher efficacy (Fig. 8 A). These results strongly suggest that the differences in Syt-1 and Syt-7 granule fusion are mediated by differential Ca²⁺-phospholipid affinities between the two isoforms.

Our data also show that not all granules that undergo fusion are present in the evanescent field before stimulation. Moreover, these newly arriving vesicles or “newcomers” have distinct properties depending on their corresponding Syt isoform. Our study indicated that a larger proportion of fusing Syt-1 granules are newcomers in comparison to Syt-7 (Fig. 5). In order for a newcomer granule to undergo fusion, it must first undergo interactions with proteins at the plasma membrane, including t-SNAREs. These interactions are well characterized in chromaffin cells and are believed to be mediated in part by the activity of Munc18-1. For example, using Munc18-1 knockout mice, Neher and colleagues showed that lack of Munc18-1 dramatically reduced the proportion of granules docked at the plasma membrane (Voets et al., 2001b). A more recent study found that overexpression of syntaxin-3 (Syn-3) in INS-1 cells resulted in increased recruitment of newcomer granules during stimulation (Zhu et al., 2013). A possible explanation for our findings is thus that Syt-1 granules have a higher affinity for proteins such as Munc18-1 and/or Syn-3 that mediate rapid recruitment of granules from within the cell. However, not much is known regarding isoform-specific interactions of Syts and SNAREs/SMs that impinge on the late steps of exocytosis.

Two points with regard to spatial and temporal resolution of these studies should be noted. First, although we do find that Syt granules are located at different distances from the membrane, our ability to reliably resolve those distances may be limited by the thickness of the confocal sections. Alternative imaging approaches, such as electron microscopy or stochastic optical reconstruction microscopy, may allow higher resolution measurements of distance from the membrane. Second, there appears to be no upper limit (at least within our imaging time resolution) to how quickly a granule can fuse after initially entering the evanescent field. This raises questions about what constitutes docking from a molecular and even morphological perspective. In this study we apply a phenomenological definition of docking that only requires a granule to be present in the evanescent field at the start of stimulation.

The distinct, Ca²⁺-dependent properties of chromaffin granules shown here indicate that Syt mediates differences in the behavior of individual secretory granules in an isoform-specific manner. These differences provide a molecular framework for the mechanisms that underlie stimulus specificity of secretion. In particular, the ubiquity of Ca²⁺ as a secretory switch demands a highly tuned set of Ca²⁺ sensors for exocytosis. Our work suggests that Syt-1 and Syt-7, via distinct trafficking pathways and graded Ca²⁺ affinities, allow tunable responses to a wide range of stimuli that differentially elevate intracellular Ca²⁺.

We would like to thank Mary Bittner for assistance on the digitonin permeabilization experiments and Ronald Holz for helpful suggestions on many of the experiments that were conducted. We also thank Dr. Lawrence G. Palmer for providing helpful feedback on the manuscript.

This research was supported by the National Institutes of Health (grant GM111997 to A. Anantharam, grant GM110289 to D. Axelrod, and grant MH061876 to E.R. Chapman), the American Heart Association (grant SDG14420049 to A. Anantharam), and the Richard Barber Summer Research Program (to A. Anantharam, T.C. Rao, and P.J. Dahl). E.R. Chapman is an investigator of the Howard Hughes Medical Institute.

The authors declare no competing financial interests.

Angus C. Nairn served as editor.