G-protein-coupled receptors (GPCR) play important roles in controlling neurotransmitter and hormone release. Inhibition of voltage-gated Ca2+ channels (Ca2+ channels) by G protein βγ subunits (Gβγ) is one prominent mechanism, but there is evidence for additional effects distinct from those on calcium entry. However, relatively few studies have investigated the Ca2+-channel-independent effects of Gβγ on transmitter release, so the impact of this mechanism remains unclear. We used carbon fiber amperometry to analyze catecholamine release from individual vesicles in bovine adrenal chromaffin cells, a widely used neurosecretory model. To bypass the effects of Gβγ on Ca2+ entry, we stimulated secretion using ionomycin (a Ca2+ ionophore) or direct intracellular application of Ca2+ through a patch pipette. Activation of endogenous GPCR or transient transfection with exogenous Gβγ significantly reduced the number of amperometric spikes (the number of vesicular fusion events). The charge (“quantal size”) and amplitude of the amperometric spikes were also significantly reduced by GPCR/Gβγ. We conclude that independent from effects on calcium entry, Gβγ can regulate both the number of vesicles that undergo exocytosis and the amount of catecholamine released per fusion event. We discuss possible mechanisms by which Gβγ might exert these novel effects including interaction with the soluble N-ethylmaleimide-sensitive factor attachment protein receptor (SNARE) complex.
G-protein-coupled receptors (GPCR) play key roles in orchestrating the dynamic regulation of neurotransmitter and hormone release. On activation, GPCR catalyze guanine nucleotide exchange on the α subunit of heterotrimeric G proteins leading to dissociation from Gβγ subunits (Gβγ). Subsequently Gβγ interacts with many effectors and is known to modulate transmitter release through direct binding to and inhibition of voltage-gated Ca2+ channels (Ca2+ channels) (De Waard et al. 2005; Evans and Zamponi 2006; Ikeda and Dunlap 1999). There is increasing evidence that GPCR/Gβγ can also modulate exocytosis independently from Ca2+ entry (Blackmer et al. 2001, 2005; Capogna et al. 1996; Chen et al. 2005; Gerachshenko et al. 2005; Man-Son-Hing et al. 1989; Photowala et al. 2006; Scholz and Miller 1992). One possible target for these effects is the core exocytotic machinery: Gβγ binds to syntaxin 1A, SNAP25, and synaptobrevin vesicle-associated membrane protein (VAMP) and can compete with synaptotagmin-1 for binding to the ternary soluble N-ethylmaleimide-sensitive factor attachment protein receptor (SNARE) complex in vitro (Blackmer et al. 2005; Gerachshenko et al. 2005; Jarvis et al. 2002; Yoon et al. 2007). In lamprey reticulospinal neurons, synaptic transmission is inhibited by 5-HT/Gβγ signaling and this involves a decrease in the quantal size of glutamate release (Photowala et al. 2006; Schwartz et al. 2007). However, relatively few studies have investigated the Ca2+-channel-independent effects of Gβγ, so the details of this mechanism and its impact on transmitter release in other mammalian neurosecretory models remain unclear.
We have used bovine adrenal chromaffin cells, a well-characterized neurosecretory model that provides a number of experimental advantages including the ability to quantify the amount and kinetics of catecholamine release from individual large dense core vesicles (LDCV) using carbon fiber amperometry. Several other transmitters are co-released with catecholamines from LDCV including ATP, enkephalin, and various peptides (Winkler et al. 1988). The cells express autoreceptors for several of these transmitters including P2Y and μ-opioid receptors that inhibit exocytosis (Ennion et al. 2004; Harkins and Fox 2000; Powell et al. 2000; Ulate et al. 2000). These previous studies of feedback inhibition used changes in membrane capacitance to assay exocytosis and concluded that Ca2+-channel inhibition is the dominant (if not sole) mechanism (but see Lim et al. 1997). One recent study using amperometry provided evidence that “quantal size” but not the number of release events could be reduced independently from Ca2+-channel modulation (Chen et al. 2005).
In this study, we stimulated secretion using ionomycin (a Ca2+ ionophore) or direct intracellular application of Ca2+ through a patch pipette to bypass the effects of Gβγ on Ca2+ entry. Acute activation of endogenous GPCR or transient transfection with exogenous Gβγ subunits significantly reduced the number of amperometric spikes (vesicular fusion events). Gβγ also reduced the amount of transmitter released from individual fusion events. Thus in addition to reducing Ca2+ entry, Gβγ can also control transmitter release by another pathway(s). We speculate that this might involve direct interaction with SNARE proteins to modulate the number and nature of vesicular fusion events.
The plasmids pEYFP-C1, pEGFP-C1, and pECFP-C1 were obtained from Clontech. The Gβ1 gene was cloned into pEYFP-C1 and pEGFP-C1. A point mutant Gγ2 (F66C) that generates an additional palmitoylation site to increase Gβγ subunit targeting to the plasma membrane (Takida and Wedegaertner 2003) was used for the experiments. Gγ2 (F66C) was cloned into pECFP-C1 to visualize the expression.
Chromaffin cell culture and transfection
Adult bovine adrenal glands were obtained from a local slaughterhouse, and chromaffin cells were prepared by digestion with collagenase followed by density gradient centrifugation based on previously published protocols (Fenwick et al. 1978; Greenberg and Zinder 1982). The cells were plated onto coverslips coated with collagen (at a density of ∼2–5 × 105 cell/ml) and maintained in an incubator at 37°C with 95% air-5% CO2 and a relative humidity of ∼90%. Fibroblasts were effectively suppressed with cytosine-arabinoside (10 μM) (Sigma-Aldrich; St Louis, MO), leaving relatively pure chromaffin cell cultures. The culture medium consisted of DMEM\ F12 (1:1) supplemented with fetal bovine serum (10%), glutamine (2 mM), penicillin (100 unit/ml)/streptomycin (100 μg/ml), cytosine arabinoside (10 μM) and 5-fluorodeoxyuridine (10 μM). Chromaffin cells were recorded from 2 to 5 days following cell isolation. Plasmid transfection was performed within 24 h of chromaffin cell isolation using a calcium phosphate transfection kit (Invitrogen, Carlsbad, CA) following manufacturer instructions. DNA was incubated for 4 h before a 3-min glycerol (12.5%) shock was performed. The cells were then washed twice with normal medium and used for recording 48–72 h after transfection.
Chromaffin cells were patch-clamped in the conventional whole cell recording configuration using an Axon Instruments Axopatch 200B amplifier and custom software written in Axobasic (kindly provided by Dr Aaron Fox, University of Chicago). Data were filtered at 2 kHz and sampled every 100 μs. Analyses were performed using custom-written programs and OriginPro software. Electrodes were pulled from microhematocrit capillary tubes (Fisher Scientific) and coated with silicone elastomer (Sylgard; Dow Corning, Midland, MI). After fire polishing, electrodes had resistances of <2 MΩ. Series resistance was partially compensated using the Axopatch circuitry. The standard patch pipette solution contained (in mM) 125 CsCl, 4 MgCl2, 20 HEPES, 0.3 or 10 EGTA, 0.35 GTP (sodium salt), 4 ATP (sodium salt), and 14 creatine phosphate, pH 7.3, 310 mosM. The recording bath (total volume ∼250–300 μl) was continually washed with fresh extracellular solution at a rate of ∼4 ml/min from gravity-fed reservoirs. The extracellular bathing solution had the following composition (in mM): 150 NaCl, 2 KCl, 2 MgCl2, 10 glucose, 10 HEPES, 2 or 10 CaCl2, and 0.1 TTX, pH 7.3, 305 mosM. Whole cell voltage-gated calcium channel currents (ICa) were stimulated by a 20-ms voltage step from − 80 to +10 mV. To determine the extent of Gβγ-mediated inhibition, prepulse facilitation was measured by including a 50-ms prepulse to +100 mV before the test pulse to +10 mV (see Fig. 4B).
Carbon fiber amperometry
Cells were placed into the recording bath and continuously washed with extracellular buffer containing (in mM) 150 NaCl, 2 KCl, 2 MgCl2, 10 glucose, 10 HEPES, and 2 or 5 CaCl2, pH 7.3, 305 mosM. The carbon fiber amperometry electrodes (CFE) were purchased from Dagan Instruments. The electrode was backfilled with 3 M KCl and positioned so that it just touched the surface of the cell. A potential of +700 mV was applied to the carbon fiber using HEKA-EVA8 amplifier. Amperometric electrodes were changed frequently (every 1–2 cells) and discarded if they did not display low noise. Data were acquired using custom-software written in Visualbasic and kindly provided by Dr Aaron Fox (University of Chicago). Amperometric currents were filtered at 0.7 kHz using the Bessel filter of the EVA8 amplifier and continuously sampled at 5 kHz for the duration of the stimulus episode. This filtering level places a limit on the temporal resolution, including the minimum duration (2 ms) for identification of a stable foot (see following text). To evoke exocytosis, 10 μM ionomycin was applied to the cells as detailed in results. Alternatively, cells were patch-clamped as described in the preceding text with elevated Ca2+ in the pipette. The intent was to elevate free calcium such that it evoked robust secretion rather than to accurately buffer free calcium to a known concentration. Thus we simply added calcium (450 μM) to our standard patch-pipette solution that consisted of (in mM) 125 CsCl, 4 MgCl2, 20 HEPES, 0.3 EGTA, 0.35 GTP (sodium salt), 4 ATP (sodium salt), and 14 creatine phosphate, pH 7.3, 310 mosM. The estimated free [calcium] in this solution was ∼50 μM (WebmaxC standard software— http://maxchelator.stanford.edu), but it should be noted that this is a “ballpark” figure and the actual concentration is not accurately known.
Amperometric data were analyzed using the Synaptosoft (Decatur, GA) mini analysis program, OriginPro software (OriginLab, Northampton MA) and GraphPad Prism (version 5, GraphPad Software, San Diego, CA). Events were detected using minianalysis based on a threshold of five times the rms noise of the trace (mean rms noise was 0.62 ± 0.02 pA, n = 120 cells) and were also confirmed by visual inspection. To assess overall secretory activity, all detected events were counted. To analyze the individual spike parameters (charge, amplitude etc), spikes had to meet additional criteria. All overlapping spikes were excluded from analysis. To minimize the inclusion of events that occurred distant from the carbon fiber and thus would not be reliably collected, we also excluded all spikes with rise times >5 ms. Minianalysis was used to calculate the spike amplitude, charge, 35–90% rise time and half-width (duration at half-maximal amplitude). Again each spike was visually inspected to ensure accurate detection of these parameters. Spike slope was calculated from the rise time and amplitude data in Origin. As discussed in results, there is considerable cell-to-cell variability in the number of events recorded by amperometry. If the parameters from all spikes in all cells are simply pooled for statistical comparison, then those cells with a high number of events will have greater weight than those cells with a low number of events. To avoid this, we calculated average (mean or median) values for the individual spike parameters within each cell or each recording window within a particular cell. This approach means that each cell has the same weight during statistical comparisons (for full discussion, see Colliver et al. 2001; Mosharov and Sulzer 2005). These average values were then pooled and compared between groups using paired or independent tests as appropriate. We also present the population data (all spikes from all cells) as frequency distribution histograms and cumulative frequency distributions. It is clear from these distributions that spike amplitude and charge did not follow a normal Gaussian distribution, and this was confirmed using D'Agostino and Pearson omnibus normality test (Prism version 5 software). Hence we used nonparametric statistical tests (Mann-Whitney test or Wilcoxon matched pairs test). Alternatively, the data were subjected to log-transformation to normalize the distribution and enable the use of parametric statistical tests (paired or independent Student's t-test). In all cases, data were considered to be significantly different if P < 0.05. A prespike foot was identified in some events. In those cells analyzed for foot signals, the rms noise was <0.5 pA. Based on this value and our filtering of the data at 0.7 kHz, we defined a readily identifiable foot as having an amplitude of >1 pA (at least twice rms noise) and duration of >2 ms. Events that did not meet these criteria were not included in the analysis of foot parameters (see results).
P2Y receptors reduce the number of amperometric spikes in cells stimulated by direct application of Ca2+ via a patch-pipette
Chromaffin cells express P2Y receptors that inhibit ICa and transmitter release (Albillos et al. 1996; Carabelli et al. 1998; Currie and Fox 1996; Harkins and Fox 2000; Powell et al. 2000; Ulate et al. 2000). To determine if P2Y receptor activation can inhibit transmitter release independently from the well documented modulation of Ca2+ entry, we directly increased intracellular Ca2+ via a patch-pipette and monitored catecholamine release using carbon fiber amperometry (Fig. 1A). Cells were voltage-clamped at −80 mV throughout the experiment to ensure that voltage-gated Ca2+ channels could not open. The standard patch-pipette solution was nominally Ca2+-free, and when cells were recorded under these conditions, there was no secretory activity. In contrast, when Ca2+ was added to the patch-pipette solution (free concentration ∼50 μM), robust secretion was stimulated (Fig. 1A). The recording bath was continuously washed with fresh medium for the entire recording period (300 s). For the first 100 s, the bath was perfused with NaCl-based control medium, then 1 mM ATP was applied for 100 s and finally the cells were washed with NaCl-based control medium for 100 s. Control cells were treated in an identical manner except they were perfused with control medium for the entire recording period. To gauge the overall level of secretion, we counted the number of amperometric spikes for the entire 300-s recording period. Each spike represents release of transmitter from one vesicle fusing with the plasma membrane. The total number of spikes over the 300-s recording period showed considerable cell-to-cell variability and was not significantly different in control cells compared with cells transiently exposed to ATP (Fig. 1B). However, closer inspection of the time course revealed there was a decrease in secretory activity during application of ATP. Figure 1C plots a time course of the cumulative number of amperometric spikes from 11 cells exposed to ATP (left) and 9 control cells (right). The cumulative rate of secretion (slope of the curve) decreased reversibly during application of ATP. The delay in onset of ATP action was largely due to the “dead space” in the bath perfusion system which was estimated to be ∼20 s. In control cells, the rate of secretion did not change during the drug application period (100–200 s).
To analyze the inhibition of secretion in more detail, we calculated the rate of release events (spikes per minute) in each cell before, during, and after application of ATP. The three time periods are indicated by the shaded boxes in Fig. 1C and the mean rate plotted in D. The rate of release events was significantly reduced from 30 ± 4.5 spike/min before ATP to 19 ± 4.6 spike/min during ATP (P < 0.02; n = 11). After washout of ATP, the spike rate fully recovered to 32 ± 6.2 spike/min. In control cells, the rate of release events did not change over the same time period (37 ± 12 spike/min before vs. 36 ± 12 spike/min during; n = 9; Fig. 1D, right).
As already noted, there was considerable cell-to-cell variability in the total number of amperometric spikes that were recorded (Fig. 1B). To account for this variability, we normalized the number of amperometric spikes before, during, and after ATP to the total number of spikes that occurred within the same cell (Fig. 1E). After this normalization, the rate of release events was still significantly and reversibly reduced by application of ATP (Fig. 1E, left), whereas in control cells, there was no change over the same time periods (Fig. 1E, right). We also compared the data across experimental groups (i.e., cells exposed to ATP compared with control cells). The percentage of total spikes released by cells before application of ATP was identical to that seen in control cells during the same time period (24 ± 2.8%; n = 11 vs. 24 ± 3.6%; n = 9). However, the percentage of spikes released during application of ATP was significantly less than in control cells during the same time period (13 ± 1.9%; n = 11 vs. 22 ± 2.9%; n = 9; P < 0.03). The rate of release recovered after washout of ATP and was not significantly different from controls (22 ± 2.1%, n = 11 vs. 19 ± 1.8%, n = 9).
P2Y or μ-opioid receptors reduced the number of amperometric spikes in cells stimulated by ionomycin
We used ionomycin as an alternative method to bypass Ca2+ channels and directly elevate intracellular [Ca2+] to stimulate secretion (Fig. 2). Ionomycin is a Ca2+ ionophore and when applied to cells leads to influx of extracellular Ca2+ and triggering of secretion. In the absence of extracellular Ca2+, ionomycin produced no secretory activity (data not shown), but in the presence of extracellular Ca2+, secretion was observed (Fig. 2A). Cells were exposed to 10 μM ionomycin in NaCl-based medium containing 5 mM extracellular Ca2+, and secretion was recorded for 240 s. The experiment was divided into a “drug application window” (from 60 to 160 s) and a “washout window” (from 160 to 240 s; see Fig. 2B). Control cells were continuously perfused with NaCl-based medium for the entire period. A second group of cells was perfused with 1 mM ATP during the drug application window and with NaCl-based medium during the washout window. A third group of cells was perfused with 10 μM [D-Ala2, N-Me-Phe4, Gly5-ol]-enkephalin (DAMGO) (a μ-opioid receptor agonist) during the drug application window and with NaCl-based medium during the washout window). μ-opioid receptors have previously been shown to inhibit ICa and secretion from chromaffin cells (Albillos et al. 1996; Chen et al. 2005; Currie and Fox 1996; Kitamura et al. 2002).
The mean rate of release events (spike/minute) was calculated during the drug application window and the washout window for each cell. The data were then pooled and are presented in Fig. 2B. The rate of release events during the drug application window was 27 ± 6.3 spike/min in control cells (n = 20). This was significantly reduced in cells exposed to ATP (14 ± 5.6 spike/min; n = 20; P < 0.01) and cells exposed to DAMGO (13 ± 3.3 spike/min; n = 21; P < 0.04; Fig. 2B, left). In contrast, during the washout window, the rate of secretion was not significantly different in cells previously exposed to ATP or DAMGO compared with control cells (57 ± 11 spike/min in control cells; 62 ± 9 spike/min in ATP-treated cells; 60 ± 12 spike/min in DAMGO-treated cells; Fig. 2B, right). To account for the cell-to-cell variability in secretory activity, we normalized the number of spikes that occurred during the drug application window to the total number of spikes that occurred during the entire recording period within the same cell (Fig. 2C). In control cells 33 ± 4% of spikes occurred in this time period, and this was significantly reduced in cells exposed to ATP (16 ± 3%; P < 0.002) or DAMGO (20 ± 3%; P < 0.02).
Taken together, the data in Figs. 1 and 2 demonstrate that P2Y and μ-opioid receptors can reduce the number of amperometric spikes independently from effects on Ca2+ influx through voltage-gated Ca2+ channels.
P2Y and μ-opioid receptors reduced the charge (quantal size) of amperometric spikes
Each amperometric spike represents an individual vesicular fusion event and the charge (integral) of a spike is directly proportional to the number of catecholamine molecules released during the vesicle fusion event (i.e., “quantal size”). We analyzed the individual amperometric spikes triggered by ionomycin in control cells and cells exposed to ATP or DAMGO to determine if spike charge was altered. For these analyses, we were careful to exclude overlapping spikes and spikes with excessively slow rise times (>5 ms). These slowly rising spikes likely represent release events that occur distant from the carbon fiber, and so their charge and kinetics cannot be quantified reliably.
We calculated an average value of spike charge from all eligible spikes in the drug application window and washout window for each cell. These data were then pooled, and the spike charge during the drug application and washout windows was compared (Fig. 3, B–D). This ensured that each cell had the same statistical weight in the final analyses (see methods) (see also Colliver et al. 2001; Mosharov and Sulzer 2005 for discussion). To enable the use of parametric statistical analyses, the data were log-transformed to normalize the distribution. In control cells (n = 10), the mean amperometric charge did not change over time and was 0.62 ± 0.09 pC during the drug application window versus 0.65 ± 0.09 pC during the washout window (Fig. 3B). In contrast, spike charge in the presence of ATP was significantly reduced compared with after washout of ATP (0.64 ± 0.21 vs. 0.94 ± 0.13 pC; n = 8; P < 0.04; Fig. 3C). Similarly, when DAMGO was present during the drug application window, the spike charge was significantly smaller than during the washout window (0.43 ± 0.04 vs. 0.60 ± 0.04 pC; n = 9; P < 0.04; Fig. 3D). These differences are also apparent when looking at the population distribution histograms (Fig. 3, E–G). All the spikes from all cells in each experimental group were pooled and relative frequency distributions plotted. The inset box in each panel plots cumulative frequency distributions. Comparison of the pooled population data also showed that charge was significantly reduced by both P2Y and μ-opioid receptors (P < 0.001; Mann-Whitney U test).
There was no clear effect of ATP on spike charge when we used direct application of Ca2+ via the patch-pipette to stimulate secretion (see Fig. 1). In these cells, the charge increased by ∼40% from 0.54 ± 0.08 pC before ATP to 0.76 ± 0.15 pC (n = 8; P < 0.04) during application of ATP. However, this increase was irreversible and the charge remained unchanged after washout of ATP (0.75 ± 0.21 pC). Furthermore, when spike charge was compared over the same time periods in control cells, there was a similar (50 ± 24%; n = 6) increase. This suggests that ATP had little effect on spike charge and the increase was a time-dependent phenomenon due to other factors associated with this recording configuration.
Transient expression of exogenous Gβγ in chromaffin cells
To determine if Gβγ could mimic the effects of ATP and DAMGO on the number and charge of the amperometric spikes, we used CaPO4 to transfect chromaffin cells with EGFP-Gβ1 and Gγ2(F66C). We used the F66C point mutant of Gγ2 as this generates an additional palmitoylation site that increases targeting of the Gβγ dimer to the plasma membrane (Takida and Wedegaertner 2003). Cells expressing EGFP were visually identified for use in patch-clamp and amperometry experiments. To confirm that these “green cells” expressed functional Gβγ dimers, we used patch-clamp recording of ICa (Fig. 4, A and B). A defining feature of Gβγ-mediated inhibition of ICa is reversal by a strongly depolarizing voltage-step (Bean 1989; Elmslie et al. 1990; Penington et al. 1991). This reversal, termed prepulse facilitation, is thought to reflect transient dissociation of Gβγ from the channel at the depolarized membrane potential. Functionally, prepulse facilitation can be used to estimate the extent of Gβγ-mediated inhibition of ICa. Therefore this provided a convenient test to determine the extent of functional Gβγ expression in the transfected cells. A prepulse was applied under basal conditions (the cells were perfused with standard NaCl based buffer) and then again during application of 100 μM ATP to activate endogenous P2Y receptors. We have previously shown that 100 μM ATP produces maximal inhibition of ICa in these cells (Currie and Fox 1996, 2000). In control cells, there was no prepulse facilitation of ICa under basal conditions, but with ATP present, there was strong facilitation reflecting the inhibition produced by endogenous Gβγ-activated by P2Y receptors (Fig. 4, A and B). Three of 11 cells transfected with Gβγ behaved in the same manner as control cells: no prepulse facilitation under basal conditions, but strong facilitation with ATP present (not shown). The remaining eight cells transfected with Gβγ all showed prepulse facilitation under basal conditions, indicating that Gβγ was functionally expressed and produced tonic inhibition of ICa (Fig. 4, A and B). Application of ATP to these Gβγ-transfected cells increased facilitation to the same level seen in control cells with ATP present (Fig. 4B).
We also used confocal imaging to examine expression of Gβγ. For these experiments, we transfected cells with EYFP-Gβ1 and ECFP-Gγ2(F66C). In some cells, both Gβ1 and Gγ2 were clearly expressed and colocalized at the plasma membrane (Fig. 4C). However, in some cells, we noted that Gβ appeared to be expressed alone and was not targeted to the plasma membrane. This may explain our observation in the patch-clamp experiments that 3/11 cells expressing GFP-tagged Gβ did not exhibit prepulse facilitation of ICa.
Taken together, these data suggest that most (∼75%) but not all transfected cells visually identified by expression of GFP-tagged Gβ express functional Gβγ dimers. Furthermore, the data suggest that the expression level of Gβγ is not excessively high, and at least in the vicinity of the Ca2+ channels (and presumably release sites) is nonsaturating and lower than that achieved by activation of endogenous GPCRs.
Transient transfection with Gβγ inhibited the number and charge (quantal size) of amperometric spikes
To assess the ability of Gβγ to inhibit secretion independently from its effects on Ca2+ channels, we used ionomycin to directly increase intracellular Ca2+ and evoke exocytosis. Cells were transfected with EGFP-Gβ1 and Gγ2(F66C). Control cells were transfected with EGFP alone. In both groups, green fluorescent cells were visually identified for amperometry recordings. No release events were observed before application of ionomycin in either EGFP control cells or cells transfected with Gβγ (data not shown). The number of amperometric spikes elicited from each cell was monitored for four minutes following application of ionomycin. The rate of release events (spike/minute) was significantly reduced in Gβγ-transfected cells (13 ± 3.3 spike/min, n = 19) compared with EGFP controls (27 ± 4.3 spike/min, n = 18; P < 0.02; Fig. 4D).
We also analyzed the parameters of the individual spikes recorded from EGFP control cells and compared them to spikes recorded from Gβγ-transfected cells (Fig. 5). Any overlapping spikes and spikes with slow rise times (>5 ms) were excluded from this analysis. The median spike parameter for each cell was calculated, and these cell-averaged data were pooled and compared (Fig. 5, A–D). For statistical comparisons, the spike charge and amplitude were log-transformed to normalize the distribution and enable the use of parametric statistical tests (unpaired t-test). Cells transfected with Gβγ had a significantly smaller amperometric charge than control (EGFP-transfected) cells (0.28 ± 0.08 pC, n = 13 vs. 0.41 ± 0.08, n = 16; P < 0.05; Fig. 5A). The amplitude of the spikes was also significantly smaller in Gβγ-transfected cells (17 ± 3.3 vs. 23 ± 2.3 pA; P < 0.05; Fig. 5B). The rising slope of the spike, although somewhat slower in Gβγ-transfected cells, was not statistically different from controls (6.9 ± 1.6 pA/ms, n = 13 vs. 9.4 ± 1.5 pA/ms, n = 16; P = 0.19; Fig. 5C), and there was no difference in the duration at half-peak amplitude (Fig. 5D). The same effects of Gβγ are also apparent when all eligible spikes from all cells in each experimental group were plotted as population distribution histograms (Fig. 5, E–H). Statistical comparison of the pooled spike populations showed that spike charge, amplitude, and slope were all significantly reduced by Gβγ (P < 0.0001; Mann-Whitney U test), but half-width was not altered (P = 0.74; Mann-Whitney U test).
When looking at the population data the percentage of small spikes was increased by Gβγ (Fig. 5E) and the raw data traces also appeared to have fewer large spikes (not shown). We wanted to test if this apparent shift in the spike population was evident in individual cells or if it was simply a consequence of pooling spikes from multiple cells with different numbers of events. For these analyses, we included only the faster-rising spikes used to analyze charge, amplitude etc as explained in the preceding text. We wanted to determine if Gβγ transfection produced a shift in the ratio of smaller and bigger spikes in each cell or if the number of events was uniformly decreased regardless of spike size. We also wanted to determine if the charge of the “smaller” and “bigger” spikes was differentially altered by Gβγ. Although skewed from normal the distribution of spike charges is not clearly bimodal (Fig. 5E) (skewness coefficient calculated in Prism 5 software was 2.18 for EGFP-transfected cells and 3.29 for Gβγ-transfected cells). Therefore to be as objective as possible, the threshold for designating spikes as “smaller” or “bigger” was based on the population distribution of spike charge in EGFP control cells (Fig. 5E). The 25th percentile value of this distribution was 0.142 pC (i.e., 25% of the spikes in control cells were smaller than this value). Based on this value, events in both EGFP- and Gβγ-transfected cells were classified as being smaller if spike charge was <0.15 pC or bigger if spike charge was >0.15 pC. Figure 6 plots the percentage of spikes in each cell that fell into the two categories (smaller vs. bigger) as well as the mean number of smaller and bigger spikes per cell. This confirmed that Gβγ produced a shift in the spike distribution even within individual cells. In control cells, ∼25% of spikes were smaller and ∼75% were bigger, but in Gβγ-transfected cells, ∼51% of spikes were in the smaller group (<0.15 pC; Fig. 6A). This reflected a decrease in the mean number of bigger spikes in Gβγ-transfected cells compared with EGFP controls (12.7 ± 4.2, n = 13 vs. 28.5 ± 6.9, n = 16; P < 0.04, Mann-Whitney test), whereas the mean number of smaller spikes did not change (13 ± 3.7; n = 13 in Gβγ-transfected cells compared with 10.4 ± 2.9; n = 16 in EGFP-transfected control cells; Fig. 6B).
We also compared the median spike charge per cell for the smaller (<0.15 pC) and bigger (>0.15 pC) spikes (Fig. 6, C and D). The charge of the smaller spikes was significantly reduced from 85 ± 7 fC (n = 12) in EGFP-transfected control cells to 55 ± 5 fC (n = 11) in Gβγ-transfected cells (P < 0.005). In contrast, the charge of the bigger spikes was not significantly different between EGFP-transfected control cells (0.49 ± 0.06 pC) and Gβγ-transfected cells (0.6 ± 0.09 pC; P = 0.12).
Effects of Gβγ on the “prespike foot”
Some amperometric events display an initial slower rising phase or plateau-like feature prior to the fast rising spike. This prespike “foot” (see Fig. 7A) is thought to represent slower release of catecholamine through the fusion pore before it expands more fully to generate the faster spike (Chow et al. 1992; Zhou et al. 1996). It has been reported that activation of G proteins reduced the charge and duration of the foot signal (Chen et al. 2005), suggesting an effect of Gβγ on the fusion pore. Therefore we compared the properties of the prespike foot in control (EGFP transfected) and Gβγ-transfected cells. For these purposes, a stable and readily identifiable “foot” was required to be >1 pA in amplitude (>2 times rms noise of the trace) and the foot duration was >2 ms (see methods and Fig. 7A). All spikes in the EGFP-expressing cells were pooled together, and those events with a prespike foot identified. Similarly, all events from Gβγ-expressing cells were pooled and those with a foot identified. First, we calculated the mean values for foot charge, duration, and amplitude for all the spikes from EGFP- and Gβγ-transfected cells. These data are presented as bar graph insets to Fig. 7, B–D. From these data, Gβγ appeared to significantly reduce the foot parameters and also the percentage of spikes that display a foot (Fig. 7E). However, there was also a positive correlation of foot charge, amplitude, and duration with overall spike charge. As spike charge increased, so did all the foot parameters (P < 0.001; Spearman's rank sum test; data not shown). Given this positive correlation, we separated the spikes into groups with a similar overall spike charge (<0.2, 0.2–0.4, 0.4–0.6, 0.6–1.0, and 1.0–1.6 pC). Figure 7 plots foot charge, amplitude, and duration versus overall spike charge for each of these groups (as already noted the mean data for all spikes are included as a bar graph inset to each panel). All the foot parameters clearly increase with an increase in overall spike charge. Furthermore, when comparing spikes of similar charge, there was little difference in the foot parameters between EGFP- and Gβγ-expressing cells. Similarly, when events were grouped by similar size (<0.1, 0.1–0.2, 0.2–0.4, 0.4–0.6, and 0.6–1.0 pC), the percentage of spikes that displayed a foot increased with overall spike charge and showed little difference between control and Gβγ-transfected cells (Fig. 7E). Thus although it is possible that Gβγ directly modulates the foot properties/fusion pore, it is also possible that these effects are secondary to the decrease in overall spike charge produced by Gβγ.
In addition to playing important physiological roles, adrenal chromaffin cells are a well-characterized neurosecretory model that provide important insight into the mechanisms that regulate exocytosis. Carbon fiber amperometry can be used to quantify the amount and kinetics of transmitter release from single vesicles. One recent study provided evidence that G proteins can reduce the quantal size of amperometric spikes triggered by release of intracellular calcium stores (Chen et al. 2005). In contrast, most previous studies used changes in membrane capacitance to monitor release and concluded that inhibition of voltage-gated Ca2+ channels is the dominant, if not sole, mechanism by which GPCRs reduce secretion (Harkins and Fox 2000; Powell et al. 2000; Ulate et al. 2000; but see Lim et al. 1997). In this paper, we stimulated secretion by methods that bypass Ca2+ channels, so the effects we observed were not due to modulation of Ca2+ influx through Ca2+ channels. When we used ionomycin to stimulate secretion, the charge (“quantal size”) of amperometric spikes was reduced by ATP or DAMGO, consistent with the findings of Chen et al. (2005). In contrast to this previous report, we also show that the number of amperometric spikes was significantly reduced (∼40–50%) by ATP or DAMGO (see Figs. 1–3). We also investigated the effects of transiently transfected Gβγ on the amperometric spikes. Control experiments examining prepulse facilitation of ICa suggest that most but not all (8 of 11) transfected cells express functional Gβγ dimers. The concentration of exogenous Gβγ in the vicinity of the Ca2+ channels, and presumably the transmitter release sites, was within the range of endogenous Gβγ produced by activation of GPCRs (Fig. 4). Gβγ expression recapitulated the effects of GPCRs—both the number and charge of the amperometric spikes were significantly reduced. Hence our data show that Gβγ can control exocytosis by a mechanism(s) independent from calcium channel modulation.
It is interesting to speculate on how Gβγ might produce these effects. The reduction of spike charge could simply reflect the presence of less catecholamine loaded into the vesicles, but this seems unlikely given the rapidly reversible nature of the ATP/DAMGO effects (Fig. 3). Another possibility is that Gβγ preferentially inhibits the release of a distinct subset of larger vesicles. There is evidence for two populations of vesicles in mouse chromaffin cells that differ by a factor of ∼5 based on estimates of vesicle size (volume) and charge of amperometric spikes (Grabner et al. 2005). In bovine chromaffin cells, there is no clear evidence for two vesicle populations, but it remains possible that this could contribute to the effects that we observe. Indeed, Gβγ reduced the number of larger spikes (charge >0.15 pC) elicited in each cell with little effect on the number of smaller spikes (Fig. 6).
Another possibility is that Gβγ might shift the mode of exocytosis and favor the occurrence of transient rather than full fusion events. By definition, full fusion will result in complete emptying of the vesicle, but it is possible that transient fusion may result in partial release of vesicular contents. Work from lamprey reticulospinal synapses supports the idea that GPCRs can modulate the mode of exocytosis. Inhibition of glutamate release from these synapses occurs very rapidly following photolysis of caged 5-HT (<20 ms) and persists following cleavage of synaptobrevin with botulinum neurotoxin B, suggesting an effect on the readily releasable pool of vesicles (Gerachshenko et al. 2005). Combined imaging of FM 1–43 quenching and electrophysiological recording suggests that 5-HT promotes transient fusion (Photowala et al. 2006), thereby altering the peak concentration of glutamate in the synaptic cleft independent from release probability (Schwartz et al. 2007). In chromaffin cells, dynamic shifts in the mode of exocytosis from transient fusion to full collapse events have been reported to occur with increased stimulation frequency/Ca2+ entry (Elhamdani et al. 2001, 2006; Fulop and Smith 2006; Fulop et al. 2005). Moreover, it has been proposed that GPCRs might reduce the charge of amperometric spikes by modulating the fusion pore (Chen et al. 2005).
Chen et al. (2005) showed that G proteins reduced the charge and duration of the prespike foot, suggesting an effect on the fusion pore stability. Therefore we analyzed the effects of Gβγ on the prespike foot. When all spikes were pooled, the mean foot duration, foot amplitude and foot charge were reduced in Gβγ-transfected cells compared with controls (Fig. 7). However, as overall spike charge decreased so did the amplitude, duration, and charge of the foot in addition to the percentage of spikes that displayed a foot (Fig. 7). When spikes of similar overall charge were compared, there was little difference in the foot parameters in Gβγ- and EGFP-transfected control cells (Fig. 7). It remains possible that Gβγ modulates the fusion pore to reduce the foot parameters and spike charge in parallel. However, we cannot exclude the possibility that Gβγ reduces spike charge by another mechanism and that the changes in foot parameters are secondary to the decrease in overall spike charge.
Our data do clearly show that Gβγ reduced spike charge, consistent with an effect on the mode of exocytosis. We also show that the charge of smaller spikes (<0.15 pC) was significantly reduced in Gβγ-transfected cells compared with EGFP controls, whereas the charge of the bigger events (charge >0.15 pC) was not significantly altered. One possible explanation for this difference is that the bigger spikes are more likely to reflect full fusion events and the smaller spikes more likely to reflect transient fusion events (Fulop et al. 2005). Assuming no effect on vesicle loading, then Gβγ should have little effect on spike charge when a full fusion event occurs (i.e., the total amount of catecholamine in the vesicle is not altered). However, Gβγ could alter the charge of transient release events by restricting release of catecholamine through the fusion pore.
The molecular targets that underlie the effects of Gβγ on exocytosis are not completely elucidated, but evidence in the literature provides support for the idea that one possible mechanism involves interaction with the SNARE complex. Gβγ can bind to syntaxin-1A, synaptobrevin, SNAP25, and the ternary SNARE complex in vitro (Blackmer et al. 2005; Gerachshenko et al. 2005; Jarvis et al. 2002; Yoon et al. 2007). As already outlined, 5-HT reduces glutamate release from lamprey reticulospinal synapses. Notably, presynaptic injection of Botulinum Toxin A, which removes nine amino acids from the C-terminus of SNAP25, decreased the amplitude of postsynaptic currents by ∼50% and completely eliminated the inhibitory effect of 5-HT (Gerachshenko et al. 2005). In the same study, injection of a 14 amino acid peptide from the C-terminus of SNAP25 also eliminated the inhibitory effect of 5-HT. Furthermore, Gβγ- and Ca2+-bound synaptotagmin-1 compete for binding to the SNARE complex in vitro, and in lamprey reticulospinal synapses, the 5-HT-mediated inhibition is diminished by elevating intracellular Ca2+, consistent with increased affinity of synaptotagmin-1 for the SNARE complex (Yoon et al. 2007). Synaptotagmins have also been implicated in fusion pore modulation (Bai et al. 2004; Moore et al. 2006; Wang et al. 2001, 2003). Based on these collective findings, it is tempting to speculate that one action of Gβγ is to interfere with the triggering and/or modulation of exocytosis by Ca2+-bound synaptotagmin-1. The SNARE proteins in conjunction with a variety of other interacting proteins also play important roles in vesicle docking/priming (Banerjee et al. 1996; Borisovska et al. 2005; Gulyas-Kovacs et al. 2007; Sorensen 2004), so it is possible that binding of Gβγ could modulate multiple stages in the exocytotic process.
It is well established that Gβγ signaling can inhibit calcium entry through Ca2+ channels to control catecholamine release from chromaffin and other cells. We postulate that Gβγ can also interact with additional targets, perhaps the SNARE complex, to exert precise control over the number and nature of exocytotic fusion events. One possible consequence of Gβγ modulation is a shift in the mode of exocytosis to transient fusion. This has the potential to impact the identity as well as the amount of transmitter release from chromaffin cells. In addition to small molecule transmitters (catecholamines), the large dense core vesicles also contain peptidergic transmitters (Winkler 1988). Full fusion/collapse of the vesicle will by its nature release all of these contents, but transient fusion that restricts catecholamine release can also prevent release of some these larger peptide transmitters by a simple size exclusion mechanism (Fulop et al. 2005). Ongoing investigations will dissect the precise molecular interactions and the relative contribution of these different pathways to the complex effects of Gβγ on exocytosis.
This work was supported by National Institutes of Health Grants R01-NS-052446 to K.P.M. Currie and R01-EY-010291 to H. E. Hamm, American Heart Association Southeast Affiliate Grant AHA 0665219 to K.P.M. Currie, and a pre-doctoral fellowship from the American Heart Association to E.-J. Yoon.
We thank S. McDavid for expert preparation of adrenal chromaffin cells and Dr. Aaron Fox (University of Chicago) for providing data-acquisition and analysis programs.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2008 by the American Physiological Society