Introduction. In vivo, cerebellar mossy fibre boutons (cMFBs) can sustain high-frequency transmission which is well-suited for rate coded signalling, while hippocampal mossy fibre boutons (hMFBs) exhibit strong facilitation which is well-suited for burst coding. Despite sharing a common name and gross morphological features, it is unclear whether differences in presynaptic ultrastructure contributes to their distinct functional properties. 

Aims. We compared the presynaptic ultrastructure and vesicle mobility in cMFBs and hMFBs to determine whether they contribute to the different vesicle supply rates required for sustained rate and burst signalling in these axons. 

Methods. Boutons from wild type mouse cerebellar and hippocampal tissue were analysed by transmission electron microscopy and electron tomography. Mitochondrial size and area fraction, active zone (AZ) surface area, and vesicle size and density were quantified stereologically [1]. Vesicle mobility was measured in acute slices from VGLUT1-Venus mice using fluorescence recovery after photobleaching (FRAP; 35°C) [2]. Biologically constrained 3D Monte Carlo simulations were used to model FRAP data and to predict sustained vesicle supply to multiple AZs. 

Results. Mitochondria in cMFBs were larger (short-axis diameter: 243.8 ± 5.7 nm, n = 36 images) than in hMFBs (191.2 ± 3.0 nm, n = 26; p < 0.001, t-test) and more densely packed (area fraction: 0.224 ± 0.012 vs 0.076 ± 0.005; p < 0.001). AZs were significantly smaller in cMFBs (0.040 ± 0.006 µm², n = 8 AZs) than hMFBs (0.110 ± 0.026 µm², n = 8, p = 0.03, t-test), with lower docked vesicle density (102.4 ± 8.0, n = 20 vs 160.3 ± 10.6 vesicles/µm², n = 20, p < 0.001). Vesicle diameters were larger (45.4 ± 0.5 nm, n = 41 VOIs vs 44.0 ± 0.2 nm, n = 31, p = 0.007, t-test) and more variable in cMFBs, while vesicle cluster volume fraction was lower (0.340 ± 0.015 vs 0.418 ± 0.007; p < 0.001, t-test). FRAP revealed a 9-fold slower long-time diffusion coefficient in hMFBs (D_long = 2.7 ± 1.1 × 10^-3 µm²/s, n = 79 boutons) compared to cMFBs (24.5 ± 3.6 × 10^-3 µm²/s, n = 63; p < 0.001, model F-test), with 2-fold smaller mobile fraction (40 ± 6% vs 76 ± 1%; p < 0.001). Monte Carlo simulations predicted that the higher vesicle density and larger immobile fraction can fully account for the reduced mobility in hMFBs. Simulations of vesicle supply to multiple AZs predicted that the smaller, spatially distributed AZs of cMFBs can support sustained vesicle delivery longer than the larger AZs of hMFBs. 

Conclusions. Our results show that cMFBs and hMFBs differ substantially in mitochondrial content, AZ size, vesicle packing and mobility. These ultrastructural differences, and 9-fold difference in vesicle mobility, contribute to their distinct functional properties, supporting higher energy and vesicle supply requirements for sustained rate coding at cerebellar synapses and approximately 4-fold larger pool of docked vesicles required for burst coding at hippocampal synapses. 

Ethics. Procedures were carried out in accordance with the UK Animals (Scientific Procedures) Act 1986, approved by relevant institutional ethics committees.