High-speed imaging reveals the bimodal nature of dense core vesicle exocytosis

Abstract

During exocytosis, the fusion of secretory vesicle with plasma membrane forms a pore that regulates release of neurotransmitter and peptide. Heterogeneity of fusion pore behavior has been attributed to stochastic variation in a common exocytic mechanism, implying a lack of biological control. Using a fluorescent false neurotransmitter (FFN), we imaged dense core vesicle (DCV) exocytosis in primary mouse adrenal chromaffin cells by total internal reflection fluorescence microscopy at millisecond resolution and observed strikingly divergent modes of release, with fast events lasting <30 ms and slow events persisting for seconds. Dual imaging of slow events shows a delay in the entry of external dye relative to FFN release, suggesting exclusion by an extremely narrow pore <1 nm in diameter. Unbiased comprehensive analysis shows that the observed variation cannot be explained by stochasticity alone, but rather involves distinct mechanisms, revealing the bimodal nature of DCV exocytosis. Further, loss of calcium sensor synaptotagmin 7 increases the proportion of slow events without changing the intrinsic properties of either class, indicating the potential for independent regulation. The identification of two distinct mechanisms for release capable of independent regulation suggests a biological basis for the diversity of fusion pore behavior.

Keywords: dense core vesicle; exocytosis; fluorescent false neurotransmitter; fusion pore; synaptotagmin 7.

Fig. 1.

High-frequency imaging reveals distinct classes of exocytic event. (A) Micrographs of an isolated primary mouse adrenal chromaffin cell, illuminated by bright field (Left) or total internal reflection, revealing FFN206 puncta (Right). (Scale bar: 5 µm.) (B–D) Fluorescence time traces showing fast (B) and slow (C) release of FFNs with reduced post-event baseline (indicating docked vesicles), as well as rapid release with unchanged baseline (no detectable docked vesicle) (D). Insets in each panel show select frames of the corresponding event, marked by the time relative to initial FFN signal increase (time 0).

Fig. 2.

FFN release is exocytic in nature. (A) Removal of Ca⁺⁺ from the external solution abolishes FFN release from chromaffin cells. (B) Lentiviral transduction of tetanus toxin light chain (TeTX-LC) abolishes FFN release. FFN release was imaged at an acquisition rate of 800 Hz. Error bars indicate SD. (C) Quantitative Venn diagram shows the vast majority of FFN release events coincide with unquenching of BDNF-pHluorin. (D–G) Fluorescence time traces show that unquenching of BDNF-pHluorin (green) follows FFN release (blue) for both fast (D and F) and slow events (E) with reduced post-event baseline, as well as for release with an unchanged post-event baseline (G). The prolonged exposure time required for dual imaging of BDNF-pHluorin and FFN limited capture of the FFN event to a single (D and G) or no (F) frame. (H and I) Kymographs showing the spread (or lack thereof) of FFN following rapid (H) or slow (I) release. Note the difference in timescale.

Fig. 3.

FFN and Alexa Fluor 488 dual imaging. (A–F) Fluorescence time traces showing Alexa (green) dye entry occur simultaneously with a fast FFN event (blue) (A) but during (B) or after a slow FFN event (C–F). (G) Quantitative Venn diagram showing that the vast majority of FFN events are accompanied by Alexa dye entry. (H) Onset of external dye entry relative to peak FFN fluorescence. Negative delay means external dye entry occurred before FFN reached peak fluorescence.

Fig. 4.

Quantitative analysis of FFN events. (A) Events with a spike were fit by least-squares regression. (B) Events without spikes were fit using free-knot B-splines. Event kinetics are characterized using the FWHM, defined as the time difference between the half-rise and half-decay points (green crosses). (C) The observed cumulative frequency of FWHM for 252 events from 12 cells and the cumulative probability function determined by fitting the data using MLE. Inset: P-P plot showing the goodness of fit. (D) The probability density functions of the two lognormal distributions used to fit the data in (C). (E) Relative contribution of the two components shown in (D). Error bars: 95% CI.

Fig. 5.

Synaptotagmin 7 knockout increases the proportion of slow events without affecting their kinetics. (A) Cumulative frequencies and MLE best fit of event full width half-maximum (FWHM) in wild-type (Left, 281 events from 14 cells) and syt 7 knockout (Right, 144 events from ten cells). Insets show the respective P-P plots. (B) Loss of syt 7 does not alter the properties of individual kinetic components. Error bars represent the 2.5 to 97.5% quantile region of each distribution, calculated using best-fit parameters. Note that the probability density function of a lognormal distribution only appears symmetrical when plotted using a logarithmic-scaled x-axis (e.g., Fig. 4D); the mean (statistical expectation) of the distribution is in fact much greater than the peak of the probability density function. (C) Loss of syt 7 significantly reduces the relative contribution of the fast component to the total. Error bars indicate 95% CI. (D) Traditional view that syt 7 mediates both membrane fusion and pore expansion. (E) The results suggest two distinct pathways to exocytosis, leading to fast (solid black arrows) or slow release (dashed arrows). Syt 7 may either act to promote the targeting of vesicles for fast release (green arrow), or to increase specifically the fusion probability of vesicles forming large pores (red arrow).