Characterization of sequential exocytosis in a human neuroendocrine cell line using evanescent wave microscopy and “virtual trajectory” analysis

Abstract

Secretion of hormones and other bioactive substances is a fundamental process for virtually all multicellular organisms. Using total internal reflection fluorescence microscopy (TIRFM), we have studied the calcium-triggered exocytosis of single, fluorescently labeled large, dense core vesicles in the human neuroendocrine BON cell line. Three types of exocytotic events were observed: (1) simple fusions (disappearance of a fluorescent spot by rapid diffusion of the dye released to the extracellular space), (2) “orphan” fusions for which only rapid dye diffusion, but not the parent vesicle, could be detected, and (3) events with incomplete or multi-step disappearance of a fluorescent spot. Although all three types were reported previously, only the first case is clearly understood. Here, thanks to a combination of two-color imaging, variable angle TIRFM, and novel statistical analyses, we show that the latter two types of events are generated by the same basic mechanism, namely shape retention of fused vesicle ghosts which become targets for sequential fusions with deeper lying vesicles. Overall, ∼25% of all exocytotic events occur via sequential fusion. Secondary vesicles, located 200–300 nm away from the cell membrane are as fusion ready as primary vesicles located very near the cell membrane. These findings call for a fundamental shift in current models of regulated secretion in endocrine cells. Previously, sequential fusion had been studied mainly using two-photon imaging. To the best of our knowledge, this work constitutes the first quantitative report on sequential fusion using TIRFM, despite its long running and widespread use in studies of secretory mechanisms.

Appendix

Distribution of Δz for transient fusions

The probability density function of the fraction of vesicle contents released was calculated by Jones and Friel (2006), assuming (1) fusion pore lifetimes are exponentially distributed (τ _P ), as expected for simple channel openings, and (2) vesicle contents are lost through the fusion pore with an exponential time course (τ _D ):

$$ {\rm d}P/{\rm d}F = A (1-F)^{A-1}, $$

(1)

where A ≡ τ _D /τ _P , and F is the fraction of vesicle contents released in a transient pore opening. For our experiments, F = 1−r, where r = I _a/I _b is the ratio of the fluorescence intensity of a spot just after (I _a) to just before (I _b) a fusion. Thus, (1) can be written as

$$ y = A r^{A-1}, $$

(2)

where y = dP/dF. We wish to calculate the distribution of “virtual jump” sizes, Δz = −δln r, in the z-space (see “Materials and methods”). This is obtained trivially by substituting r = e^−Δz/δ into (2):

$$ y = A {\rm e}^{-(A-1) \Delta z/\delta}. $$

(3)

In our experiments, the great majority of fluorescent spots disappeared in a single fusion (F = 1), implying A ≡ τ _D /τ _P <  1. That is, the factor −(A−1) in (3) is >0, and y increases exponentially as a function of Δz without bound. The experimentally measured distribution of Δz, shown in Fig. 6c, is clearly not exponential, implying that transient fusions do not contribute to our observations in any significant way.

Frequency of orphan events generated by “ballistic” vesicles

Orphan events generated by ballistic vesicles should be observed more often when thinner evanescence depths, δ, are used. For such vesicles, there are two extreme cases, depending on the time to cross the evanescent field (τ_cross) and the time spent at the membrane before fusion (τ_mb). A vesicle would be visible for a duration that is the total of these two timescales, i.e. τ_vis = τ_cross + τ_mb. The crossing time depends on the evanescence depth, δ, becoming longer with larger δ (τ_cross ∝ δ/v, where v is the average speed with which a vesicle moves toward the membrane), whereas τ_mb is independent of it. In the limit τ_cross ≫ τ_mb, we expect that the fraction of detected orphan events, f _orphan, scales inversely with τ_cross, i.e. f _orphan ∼ δ⁻¹. Thus, comparing data acquired at δ₁ = 100 nm and δ₂ = 150 nm, we expect f _orphan(δ₁)/f _orphan(δ₂) = δ₂/δ₁ = 150 nm/100 nm =  1.5, that is ∼50% more orphan events should be detected at δ₁ = 100 nm compared to δ₂ = 150 nm. In the other extreme, i.e. τ_cross ≪ τ_mb, orphan events should be detected with the same efficiency regardless of the value of δ since ballistic vesicles should spend most of their short stays in the evanescent field at the cell membrane. Overall, at most 50% more ballistic orphans should be detected at δ = 100 nm compared to δ = 150 nm.