A continuous-time model that required docking matched published fusion-state counts at an assumed age of 2 ms

We tested whether published synaptic fusion-state counts exclude a required tight-docking step. We used the exact count inputs of a released model from mouse hippocampal cryo-electron tomography. At the same assumed age of 2 ms, an irreversible continuous-time model matched every state frequency when every path passed through docking. Its mean docking time was 84.05 microseconds before stalk formation, or 41.68 microseconds after stalk formation. These are model times, not measured dwell times. For the irreversible comparison, a discrete rule that permits only one adjacent jump per interval cannot fit the first two occupancies at a 0.2 ms step. The interval must be at most 88.115 microseconds. Exact sampling of the continuous model at 0.2 ms retains the fit because several jumps can occur within an interval. A model with a docking branch matched the full profile for every route fraction above 1/77 through 1. Its mean time to pore opening has no finite upper bound as the route fraction approaches the lower limit. A constructive proof, independent ODE calculations, and sampled event paths check the results. The pooled counts are not treated as independent biological samples. This audit limits one kinetic inference from one structural profile. It does not identify the biological fusion route or test other evidence about membrane mechanics. The package preserves code, source bytes and hashes, all model states, results, figures, proofs, and repeat instructions.

synaptic vesicle fusion · tight docking · cryo-electron tomography · kinetic inference · continuous-time Markov chain · structural identifiability · transmitter release