Sparse force-bearing bridges between neighboring synaptic vesicles

Abstract

Most vesicles in the interior of synaptic terminals are clustered in clouds close to active zone regions of the plasma membrane where exocytosis occurs. Electron-dense structures, termed bridges, have been reported between a small minority of pairs of neighboring vesicles within the clouds. Synapsin proteins have been implicated previously, but the existence of the bridges as stable structures in vivo has been questioned. Here we use electron tomography to show that the bridges are present but less frequent in synapsin knockouts compared to wildtype. An analysis of distances between neighbors in wildtype tomograms indicated that the bridges are strong enough to resist centrifugal forces likely induced by fixation with aldehydes. The results confirm that the bridges are stable structures and that synapsin proteins are involved in formation or stabilization.

Keywords: Presynaptic; RRP; Reserve; Supply-rate depression; Tomography.

Fig. 1

Model of cloud where synaptic vesicles are tethered together into short chains. This scenario was proposed in Gabriel et al. (2011) to explain simplifying mathematical constraints that emerged from electrophysiological studies of rate-limiting steps in synaptic vesicle cycling at Schaffer collateral synapses. Vesicles throughout the terminal are linked together in short chains. Non-docked vesicles attached to docked vesicles serve as an autonomous reserve pool that can be expended/depleted during heavy use. Docked vesicles are replaced stochastically at the slow rate of 1/min with a vesicle at the start of a full chain that was previously not docked

Fig. 2

Graphical user interface used for fine tuning estimates of vesicle center and radius. As a first step in the analysis, rough estimates for the position and size of each vesicle within each tomogram were obtained by modeling each vesicle with a sphere using the 3dmod computer program of the IMOD suite of software (Kremer et al. 1996). A separate mini-tomogram containing each vesicle was then extracted from the whole tomogram. The three images in the user interface are all the means of virtual slices from the same mini-tomogram spanning the estimated center point (i.e., spanning a total of 0.5 median vesicle diameters). The center image is the mean of virtual slices parallel to the horizontal plane without rotating the mini-tomogram, the leftmost image is after rotating the mini-tomogram ${45}^{\circ }$  about the y-axis, and the rightmost image is the mean after rotating about the x-axis. The white circle is the perimeter of the sphere calculated from the center and radius estimates and will match all three images when the estimates are correct. The horizontal sliders directly below the leftmost image determine contrast. The bottom-most horizontal slider allows the observer to adjust the radius estimate. The vertical sliders allow the user to adjust the estimated location of the center point

Fig. 3

Method for evaluating the presence or absence of bridges between vesicles. a Screenshot of graphical user interface for evaluating presence or absence of bridges. Mini-tomograms for each vesicle pair were rotated so that the centers of both vesicles were within the horizontal plane. Images of vesicle pairs were then presented in a random order simultaneously from three angles: not rotated; and rotated $\pm {45}^{\circ }$ about the central axis common to both vesicles. When the “3-D display” radio button is selected, the scroll bar to the right allows the user to scroll through the individual virtual sections. Otherwise, the mean images of all virtual sections between 0.25 median vesicle diameters above and below the central plane are displayed. The lower scroll bars below the images control contrast. b–e Initial survey of 13 WT and 15 DKO tomograms for vesicle pairs separated up to 1.5 vesicle diameters. bZ-angle for a pair of vesicles is defined as the angle between the axis passing through the centers of both vesicles and the plane that is horizontal to the tissue slice. c Fraction of WT vesicle pairs for which images were evaluated as qualitatively “Bad” or “Terrible” vs Z-angle. d Probability of detecting a bridge vs Z-angle in WT synapses. Scores of 1 (“Certain”) and 2 (“Likely”) were counted as bridges. e Probability of detecting a bridge vs distance between the vesicles for Z-angles $\le {25}^{\circ }$ (640 pairs for WT, 374 for DKO). Distance units are median vesicle diameters, which were calculated across all vesicles for each tomogram independently

Fig. 4

Lower probability of detecting a bridge between neighboring vesicles in synapsin DKO synapses. a Images of pairs of vesicles with and without a bridge. Magenta arrows indicate bridges. Scale bars are 30 nm. Image thickness was 8.3 nm (i.e., average of 13 virtual sections with a voxel spacing of $6.4\mathrm{\AA }$). Post-fix staining for these examples was with uranyl acetate, but no lead, which is referred to as “Group 1” below. b Probability of detecting a bridge per pair of neighbors vs genotype. Values were estimated individually for each tomogram by dividing the number of bridges detected by the number of pairs that were analyzed. Circles are the median values across tomograms, boxes delineate the middle two quartiles, ***$p<0.001$ (Kolmogorov–Smirnov; $n=13$ tomograms for WT, 15 for DKO). c No difference in scores of “3 Can’t tell”

Fig. 5

Similar differences between WT and synapsin DKO across several post-fixation staining protocols. a Examples of vesicle pairs with and without bridges after two distinct post-fixation protocols. See Fig. 4a for Group 1 examples. Magenta arrows indicate bridges. Scale bar is 30 nm. b Quantification identical to Fig. 4b for each protocol ($n=5$ and 10, respectively, for WT and DKO tomograms for Group 1; 5 and 1 for Group 2; and 3 and 4 for Group 3; *$p<0.05$, Kolmogorov–Smirnov)

Fig. 6

No decrease in bridges near active zone. a, b Left panels are two-dimensional snapshots of three-dimensional models used to determine vesicle locations. Yellow spheres are docked vesicles, green are non-docked, magenta lines are plasma membrane. Right panels are examples of image slices ($6.4\mathrm{\AA }$ thick) showing docked and non docked vesicles. Magenta arrows indicate bridges. Models and images correspond to the same tomogram, but the tilt angles are slightly different to better illustrate the docked vesicles. Scale bars are 100 nm and pertain to both the model and corresponding image slice. a WT. b DKO. c Fractions of pairs with a bridge vs distance between the geometric center of the vesicle pairs and the center of the closest docked vesicle. Same data set as Figs. 4 and 5; every tomogram had at least two docked vesicle because only synapses where the synaptic cleft was visible were selected for imaging. Mean number of docked vesicles/tomogram was $13.7\pm 1.8$ for WT and $10.3\pm 1.5$ for DKO (not significant, Kolmogorov–Smirnov). Probability of detecting a bridge was calculated for each range of distances for each tomogram separately and then averaged across tomograms; $n=13$ tomograms for WT and $\ge 14$ for DKO instead of the 15 in previous figures because no pairs were present farther than 4 median vesicle diameters from the active zone for one of the DKO tomograms

Fig. 7

WT chains of four vesicles. a Examples of chain where one of the vesicles was docked. b Example where only two bridges are visible in any single plane, but the third can be seen from a different angle. Magenta arrows indicate bridges and scale bars are 30 nm. c Mean number of neighbors within 0.5 median vesicle diameters of each vesicle. The lower value for DKO implies lower density of vesicles in space, which is in-line with previous reports. d Bridges per vesicle estimated by multiplying the probability of a bridge in Fig. 4b by the number of neighbors per vesicle in c

Fig. 8

Comparison of bridges between non-docked and docked vesicles. The analysis included all vesicle pairs where at least one of the two vesicles was docked, along with a randomly selected subset of vesicle pairs where neither vesicle was docked (Z-angle was $\le {25}^{\circ }$ for all pairs). a, b Images of WT and DKO pairs where one of the vesicles is docked and is bridged to the other in the cytoplasm (left panels), and where both vesicles are docked (right panels). Scale bars are 30 nm, thickness of virtual section was 8.3 nm. Magenta arrows indicate bridges. c Quantification: horizontal lines within boxes are fraction of pairs judged to be connected by a bridge across the entire analysis, which is a different type of quantification than used above (see “Results”). Boxes delineate the 90% confidence interval for binomial proportions using the Wilson score interval with correction for continuity (Newcombe 1998); ***$p<0.001$, *$p<0.05$; both using ${\chi }^2$ with Yate’s correction. Raw values are listed in the table

Fig. 9

Inhomogeneities in spatial distribution are predicted by short chains connected by force-bearing bridges. a Example of hyperrectangular subset of vesicles before and after simulating a random walk in space. b Bars are fraction of WT vesicle pairs vs separation distance between vesicles in each pair in hyperrectangles before the random walk. Black circles are the same measurement for the entire WT data set, which includes vesicles inside and outside the hyperrectangles. Green squares represent pairs in hyperrectangles after the random walk. Orange squares represent pairs from simulations of cloud formation where vesicles that were not connected into chains were placed at random locations in a cube. c Analogous histogram for tomograms from Siksou et al. (2009) where tissue was frozen rapidly without chemical fixation ($n=316$ pairs from seven tomograms containing a mean of $36.7\pm 6.3$ vesicles/tomogram; one of eight tomograms was excluded owing to too few pairs). d Histogram of measurements of bridge lengths from the WT data set used in simulations. The blue line is a normal distribution with mean of 0.18 and standard deviation of 0.06. e Left panel is a snapshot of a simulated cloud where vesicles connected into chains of four were placed at random locations in a cube. Each vesicle was shrunk by 13% after cloud formation to simulate the shrinkage that likely occurs during aldehyde fixation. Bridges are represented by the sparsely distributed brown rods. The bars in the right panel are identical to in b; blue circles are the analysis of simulations matching the snapshot to the left; red circles are after adding a small amount of order as described in “Results”. f Analogous histogram for simulations where vesicles were connected into chains of four for a variety of volumetric fractions. The blue circles are the same as in e; the simulations of lower volumetric fractions were then generated by eliminating randomly selected chains. Native data are the same as in b. g Left panel is a snapshot of a simulated cloud where vesicles connected into chains of four were placed in an ordered rather than random formation; volumetric fraction was 0.17. Right panel is histogram of separation distances between vesicles in each pair. h Analogous histogram for DKO tomograms; compare to bars in b. Magenta squares are from simulations of cloud formation where vesicles were not connected into chains, followed by shrinkage by 12%; volumetric fraction was 0.07 for both tomograms and simulation