• Figure 2.
    Download figureOpen in new tabFigure 2. Presynaptic docked vesicle numbers are correlated with PSD areas, spine head volumes, and neck diameter, but not with neck length.

    (A) All 31,377 presynaptic vesicles. (B) En face view of the 24 docked vesicles (gray spheres) viewed through an axon (green) onto the PSD (red) of example spine (yellow). (C) Number of docked vesicles is correlated strongly with both PSD area and (D) spine head volume, weakly with (E) neck diameter, but is not correlated with (F) spine neck length. Regression lines, SEM (from multiple tracers), and r2 are as in Figure 1 n = 236 complete axonal boutons, each associated with one of the 287 complete spines. One human tracer marked PSDs and vesicles, hence no SEM for these two metrics.

    DOI: http://dx.doi.org/10.7554/eLife.10778.005

    Figure 5.
    Download figureOpen in new tabFigure 5. CV of axon-coupled spines on the same dendrite does not vary with spine size.

    There is no significant correlation, which implies that paired small synapses are as precisely matched as paired large synapses.

    DOI: http://dx.doi.org/10.7554/eLife.10778.013

    Figure 7.
    Download figureOpen in new tabFigure 7. Proximity of the glial cell to axon-coupled dendritic spines on either the same or different dendrites.

    Proximity of astrocytic glial processes is not significantly correlated with spine head volumes of axon coupled pairs. (A) Histogram of spine head volume for spines that contain a spinule that is engulfed within the glial process (‘ spinule’). (B) Representation of an engulfed spinule. (C) Histogram of spine head volume for spines that are surrounded by and making contact with a glial process (‘ensheathed’). (D) Representation of ‘ensheathed’ spine. (E) Histogram of spine head volume for spines that are proximal but not contacting a glial process (“‘ adjacent”’). (F) Representation of “‘adjacent”’ spine. (G) Histogram of spine head volume for spines that are distant from any glial process. (H) Representation of a spine “‘distant”’ from the glial process. The KS p value is shown on each inset and indicates that none of these distributions differ from the distribution for the whole population of spines.

    DOI: http://dx.doi.org/10.7554/eLife.10778.017

    Figure 8.
    Download figureOpen in new tabFigure 8. Distinguishable spine sizes.

    Over the factor of 60 range in spine head volumes from the data set there are 26 distinguishable intervals of spine sizes with a discrimination probability of 69% for each interval based on signal detection theory (Green and Swets, 1966; Schultz, 2007). The graph illustrates how distinct Gaussian distributions of spine sizes, each with a certain mean size and standard deviation, covers the entire range of spine head sizes on a log scale. The CV of each distribution is a constant value of 0.083 (Figure 5) and the intervals are spaced to achieve a total of 31% overlap with adjacent intervals giving a 69% discrimination threshold (see Materials and Methods). Note that the constant CV observed in the data set (Figure 5) means that the intervals appear uniform in width and spacing on a logarithmic scale. This is a form of non-uniform quantization which efficiently encodes the dynamic range of synaptic strengths at constant precision.

    DOI: http://dx.doi.org/10.7554/eLife.10778.018

  • Table 1.

    Lower bounds on time window for averaging binomially distributed synaptic input to achieve CV = 0.083.

    DOI: http://dx.doi.org/10.7554/eLife.10778.019

    Release probability
    (pr)
    Presynaptic spikes
    (n)
    Averaging time
    (R = 1 Hz)
    Averaging time
    (R = 25 Hz)
    0.1130621.8 min52.2 sec
    0.25819.68 min23.2 sec
    0.51452.42 min5.8 sec