Physical determinants of vesicle mobility and supply at a central synapse

STRD: Standard ACSF. -drift: data corrected for tissue drift (Figure 2—figure supplement 1); all other measurements are not drift corrected. CD + LB: 10 µM cytochalasin-D plus 10 µM latrunculin-B. Jaspla: jasplakinolide. OA: okadaic acid. Rosco: roscovitine. Values for D_long and t_1/2 (± STDV) were computed by fitting experimental FRAP curves (Figure 2A,B) to Equation (2).

The effect of OA on vesicle mobility in the MFT is in close agreement with that reported by Shtrahman et al. (2005) who report D_long = 0.10 µm²/s for hippocampal boutons in OA. While our results do show a reduction in the immobile vesicle fraction, this reduction is not enough to account for the large increase in D_long. Instead, the increase in D_long is more likely due to a reduction in protein interactions between the vesicles and cytoskeleton, as suggested by Shtrahman et al., in which case the effects of OA will be reflected in a change in D_cyto. Using data from Figure 5, we estimate D_cyto = 0.515 µm²/s in OA, a four-fold increase from control conditions (0.127 µm²/s).

For the MFT, 3D vesicle (ves) densities were computed from 2D densities by diving by the slice thickness (0.03 µm). The total vesicle volume % was computed assuming a 44 nm vesicle diameter in fixed tissue. It was assumed the immobile vesicle volume fraction near the AZ was the same as in the MFT centre (4%). For the MFT centre computation, the non-diffusible volume (vol) is the mitochondria volume fraction; for the cloud and AZ face computations, the non-diffusible volume is the non-diffusible space within the vesicle clouds computed from the 14 AZ reconstructions. D₀ was computed via the Stokes-Einstein equation assuming a 49 nm vesicle diameter for in vitro conditions at 35°C. Diffusion constants and ratios are from Results (see Figure 5). Hydrodynamic (hydro) effects from the membrane wall near the AZ were computed via Equations (9) and (10), and are average β between 50 and 100 nm from the wall, where β = (2β_||+β⊥)/3. Measured D_long near the AZ face is not available (NA): vesicles close to AZs are too small to be detected by our FRAP measurements.

For the NMJ, the 2D vesicle density is the average of those reported in Mantilla et al. (2004) and Coleman et al. (2008). The total vesicle volume % was computed assuming a proportional relationship with the MFT vesicle density and volume fraction. The immobile vesicle % and measured D_long is from Gaffield and Betz (2007). D₀ was computed assuming a 49 nm vesicle diameter and 37°C.

Data for the ribbon synapse is from Rea et al. (2004). The 3D vesicle density was computed assuming 250,000 vesicles with 50 nm diameter inside a hemisphere with 6 µm diameter. The non-diffusible volume was set to zero since Figure 3A of Rea et al. shows few mitochondria. D₀ was computed assuming a 50 nm vesicle diameter and 22°C. Note, our estimate of D_long is 10-fold smaller than the measured D_long of Rea et al., but is comparable to the measured D_long of another study of ribbon-type synapses in bipolar cells (value shown in brackets; 0.015 µm²/s; Holt et al., 2004).

For the hippocampal boutons, the 2D vesicle density is from Li et al. (1995) and Schikorski and Stevens (2001). The total vesicle volume % was computed assuming a proportional relationship with the MFT vesicle density and volume fraction. The immobile vesicle % is from Shtrahman et al. (2005). D₀ was computed assuming a 49 nm vesicle diameter and 37°C. Values for measured D_long derived from fluorescence correlation spectroscopy (FCS) vary widely, depending on the model used to fit to the data (5 × 10⁻⁵ to 0.054 µm²/s; Figure 2D, pink symbols), but our predicted D_long most closely matches that of a fit to pure diffusion (0.0043 µm²/s; Shtrahman et al., 2005) and the measured D_long of Lee et al. (0.003 µm²/s; 2012) who tracked single vesicles using quantum dots.

For all terminals, D_cyto/D₀ was assumed to equal that in the MFT centre (0.01). D_short/D_cyto was computed as the blue line in Figure 5B. D_long/D_short was computed using data in Figure 5A.