Figure 7. High-speed imaging of vGluT1-pHluorin corroborates the presence of ultrafast retrieval for single synaptic vesicle fusion events.

(A) Example traces for single vesicle fusion events after de-noising, measured with vGluT1-pHluorin imaging at a speed of 40 Hz. Arrows and insets: expansion of the peak showing dwell time lengths in detail. (B) Distribution of dwell time durations at 24°C and 2 mM extracellular Ca²⁺. Black line: double exponential decay fit (R-square = 0.8377; RRS = 0.00115). Arrows: decay constants for the ultrafast and fast component of the exponential. (C) Distribution of dwell time durations at 34°C and 2 mM extracellular Ca²⁺. Black line: double exponential decay fit (R-square = 0.7768; RRS = 0.00121). Arrows: decay constants for the ultrafast and fast component of the exponential. Inset: cumulative histogram comparing the effect of temperature on dwell times at 2 mM Ca²⁺, there is no significant effect. (D) Distribution of dwell time durations at 24°C and 8 mM extracellular Ca²⁺. Black line: double exponential decay fit (R-square = 0.8667; RRS = 0.00119). Arrows: decay constants for the ultrafast and fast component of the exponential. (E) Distribution of dwell time durations at 34°C and 8 mM extracellular Ca²⁺. Black line: double exponential decay fit (R-square = 0.770; RRS = 0.0023). Arrows: decay constants for the ultrafast and fast component of the exponential. Inset: cumulative histogram comparing the effect of temperature on dwell times at 8 mM Ca²⁺, **=p < 0.01. (F) Pie charts depicting the relative contribution of the three modes of retrieval to total endocytosis for all experimental groups. Green: ultrafast retrieval (dwell time duration between 0 and 1 s). Blue: fast endocytosis (dwell time of 1 to 20 s). Yellow: ultra-slow retrieval (>20 s). Note that the percentage of each type of endocytosis is not greatly affected by changes in Ca²⁺ or temperature. For B to E. Kolmogorov-Smirnov test: 24°C 2 mM Ca²⁺ vs. 24°C 8 mM Ca²⁺: p<0.0001; 34°C 2 mM Ca²⁺ vs. 34°C 8 mM Ca²⁺: p=0.0043; 24°C 2 mM Ca²⁺ vs 34°C 2 mM Ca²⁺: p=0.08271; 24°C 8 mM Ca²⁺ vs 34°C 8 mM Ca²⁺: p<0.0001. Kruskal-Wallis test: p<0.0001; Dunn’s post-test: 24°C 2 mM Ca²⁺ vs. 24°C 8 mM Ca²⁺: p<0.0001; 34°C 2 mM Ca²⁺ vs. 34°C 8 mM Ca²⁺: non-significant; 24°C 2 mM Ca²⁺ vs 34°C 2 mM Ca²⁺: non-significant; 24°C 8 mM Ca²⁺ vs 34°C 8 mM Ca²⁺: p<0.0001. For all the data presented in this figure: 24°C – 2 mM Ca²⁺: 558 boutons from eight coverslips; 24°C – 8 mM Ca²⁺: 474 boutons from seven coverslips; 34°C – 2 mM Ca²⁺: 327 boutons from seven coverslips; 34°C – 8 mM Ca²⁺: 209 boutons from five coverslips. At least three independent experiments (cultures). Also see Figure 7—figure supplement 1 for analysis of negative controls and simulated traces. Legends to the figure supplements.

Figure 7—figure supplement 1. False positive artifacts are negligible at fast (40 Hz) imaging speeds.

(A) Comparison of original not processed vGluT1-pHluorin traces imaged at 40 Hz with the resulting signal after bleaching, background and noise corrections, at different Ca²⁺ concentrations and temperatures. The failure analysis used to quantify probability of release is exemplified. (B) Example of a fluorescence trace over time, before and after de-noising, of a negative control bouton from neurons pretreated with EGTA-AM 100 μM for 15 min in a 0 mM Ca²⁺ medium, and then imaged at 34°C with 0 mM extracellular Ca²⁺ and similar stimulation paradigm applied to the positive control group. (C) Cumulative histogram of probability of release measured at 34°C, for positive control boutons (2 mM extracellular Ca²⁺) and negative control group measured as described in B. The release probability of the negative control is <0.02, which corresponds to <15% of the mean release probability in the positive control group. Positive control: 418 boutons; Negative control – EGTA-AM: 120 boutons. All data are from 2 to 3 independent experiments. (D) and (E) Distribution of experimental noise amplitudes (empty circles) measured at 2 mM extracellular Ca²⁺ and 24°C or 34°C, respectively. Red lines: Gaussian fit. The noise distribution after de-noising is superimposed (grey histogram). Note the dramatic effect of the filtering in noise reduction. (F) Top: simulated fluorescence trace with a release probability of 0.5. Middle: simulated trace with added random Gaussian noise with an amplitude of ~75% the amplitude of the signal (to properly simulate real experimental traces). Bottom: recovered trace after de-noising, note that most of the fusion events are recovered with minimal morphological alterations. (G) Comparison of cumulative histograms of dwell times for the original simulated trace (before adding noise, shown in black) and for the recovered simulated trace after de-noising (shown in red). Insets: corresponding Tukey plots of dwell times and probability of detection. (H) Relative proportion of different types of simulated events (with no dwell, with a dwell followed by retrieval – including fast and ultrafast events –, with no retrieval – similar to ultra-slow endocytosis –) before (original simulated trace) and after de-noising (recovered), showing similar distribution. Note the presence of <15% contribution of false detection or false positives (yellow). For G and H. Data from two independent simulations with 100 boutons (traces) and 47 events per simulation.