Using a live-cell imaging approach with the styryl dye FM1-43 we determined the amount of spontaneous and evoked release at individual hippocampal synapses. Therefore synapses underwent a spontaneous and an evoked round of staining, each followed by activity induced destaining (Figure 1A). For spontaneous staining incubation times of 5–240 minutes where used, depending on when saturation was reached. Calcium concentrations ranged from the total absence of calcium in the extracellular medium, physiological calcium concentrations (2.5 mM) to a high calcium concentration (5 mM). Using this paradigm we were able to quantify the recycling pool size as well as the spontaneously released vesicles after different loading times for spontaneously fusing vesicles. Additionally different calcium concentrations for spontaneous loading were tested. First we found that spontaneous release was able to turn over the complete recycling pool in a calcium dependent manner (Figure 1). Although the kinetics to reach saturation was slower at reduced calcium concentrations, all recycling vesicles were labeled at different concentrations of external calcium [8] (Figure 1D). Importantly [9], synapses that were capable of spontaneous vesicle release also underwent evoked vesicle recycling (97.87% ± 3.09% of boutons) and vice versa (98.61% ± 0.96% of boutons; Wilcoxon rank sum test: p = 0.83) and their spontaneous and evoked fluorescence changes were correlated (Figure 1E; Pearson’s r = 0.79 ± 0.06; 3 experiments, 921 boutons). To validate the turnover of the whole recycling pool by spontaneous release we used an alternative dual-color imaging approach with synapto-pHluorin (spH) transfected neurons and an αGFP-CypHer5E™ antibody to label the spontaneously fused vesicles. As both fluorescence markers were pH-sensitive with an inverse characteristic [10, 11], which made them excellent tools for the analysis of exo- and endocytosis due to the acidic pH inside synaptic vesicles. In contrast to the previous FM1-43 experiments evoked and spontaneous release was measured simultaneously by quantifying the fluorescence decay of the αGFP-CypHer5E™ fluorescence signal, resembling the spontaneous labeled vesicles, and on the other hand the spH fluorescence increase, corresponding to activity dependent recycling pool of vesicles. In accordance to the FM1-43 experiments we demonstrated that the whole recycling pool, which corresponds to 37.10 ± 11.10% of the total pool, could be labeled (Additional file 1: Figure S1). In a next step we wanted to determine the fusion rate of a single spontaneously released vesicle to estimate the time to turn over the whole recycling pool of vesicles and thus to proof the kinetics described above. Therefore synapses were stained spontaneously with FM1-43 for only 15 minutes at zero calcium to label only a few vesicles and subsequently destained to monitor the resulting fluorescence decay (Additional file 1: Figure S2A). Analysis of these fluorescence changes using histograms exhibited a clear quantization that corresponds to intensity of a single vesicle, which enabled us to calculate the number of spontaneously fused vesicles during 15 minutes and thus the turnover rate of a single spontaneously fusing vesicle (Additional file 1: Figure S2B). Using this single vesicle quantification, we determined the spontaneous turnover rate to be one vesicle each 80.42 ± 10.77 seconds, which is in accordance with literature [5]. Assuming a recycling pool size of about 130 vesicles [11–13] and neglecting the effect of reuse of vesicles and the faster recycling kinetics of the RRP vesicles [14], this fusion rate resulted in a rough estimate of about 174 minutes to turn over the whole recycling pool by spontaneous fusion, which agrees with the time course described above. Furthermore similar to evoked release [14] spontaneous vesicles were instantly re-releasable and thus immediately repopulate the readily releasable pool (Additional file 1: Figure S2).

In order to determine the strength of spontaneous release at the soma and the neuronal processes, we adopted a dual-color imaging approach with spH transfected neurons and an αGFP-CypHer5E™ antibody to label spontaneously fused synaptic vesicles (Figure 2A, B). Additionally we made use of a low transfection rate and chose neurons that fluoresced isolated among untransfected neighboring neurons for the analysis. The soma and the neurites were set manually (Figure 2C). Sholl analysis revealed a normal branching pattern (Σ_spH = 1.88 ± 0.34; Σ_mCherry = 1.26 ± 0.33) [15]. We found that the size of the recycling pool as well as the number of spontaneously fused vesicles is higher at the soma than in distal cellular compartments indicating a higher absolute release. However, the relative amount of spontaneous turnover, normalized on the recycling pool size, was significantly higher in the periphery (Figure 2D). Therefore the smaller synapses in the periphery have a lower absolute spontaneous turnover, but release relative to their recycling pool more vesicles spontaneously. Distinction of the axon, the dendrites and the soma via MAP2 immunostaining also showed that in absolute values synapses near the soma released the highest number of spontaneous vesicles during a 120 minute period (Additional file 1: Figure S3). The fact that synapses located at the dendrite released more spontaneous vesicles compared to the axon can be explained by the shorter length of the dendrite and thus their closer average proximity to the soma. These results fit to the notion that vesicle recycling is more effective at smaller synapses [11] arguing for neuronal outgrowth in distal segments. Spontaneous release showed a strong correlation with the size of recycling and total pool at neurites and soma, respectively (Figure 2E, G). On the other hand, no clear relationship between spontaneous release and the reserve pool was observed (Figure 2F). The fact that spontaneous release correlates with the size of recycling and total pool, but not with the size of the reserve pool seems controversial. This can be due to the high variability of the pool sizes [16] especially the reserve pool [17] and heterogeneous release probabilities among synapses [18]. Similar to evoked release, the frequency distribution of spontaneous release is more right-skewed at the soma when compared to the process (Additional file 1: Figure S4A-C), which is consistent with more recycling vesicles at the soma-near synapses (Figure 2D). No differences between synapses at the dendrite and the axon could be observed (Additional file 1: Figure S4D). Synapses were then pooled and sorted into the 20% largest, smallest and the 20% around the medium size. Regarding these different synapse sizes and locations of individual boutons, these diverse distributions caused distinct fractions of the recycling and reserve pool size with respect to the total pool size (Additional file 1: Figure S4E). In absolute numbers spontaneous release declined with distance to the soma (Additional file 1: Figure S5A), but remained largely constant if spontaneous turn-over was normalized on synapse size (Additional file 1: Figure S5B).