Abstract
Purpose:
The spatial arrangement of vesicles and release-related proteins along the rod synaptic ribbon suggests that its geometry subserves a functional specialization. Our purpose was to incorporate ribbon release sites as neural network elements in a molecular simulation of this synapse, and to investigate the dynamic effects of different interactions, reflecting putative mechanisms for glutamate release inhibition and transport.
Methods:
We created a 3D model of the synaptic cleft and postsynaptic transduction molecules using the molecular simulation software mcell. Molecular properties (diffusion constants, EC50s, concentrations, etc.) and structural properties (cleft depth, dendritic tip size, etc.) were estimated from the literature. We simulated vesicle release as either a purely random process (Poisson), a regular process (clockwork), or a random process with feedback derived from localized release inhibition (quasi-regular). For model light responses, the time-course of release probability was proportional to the membrane voltage measured for dim flash responses from mouse rods. For each process, we compared time-averaged release rate, glutamate concentration near postsynaptic mGluR6 receptors, and mGluR6 receptor activation.<br />
Results:
The Poisson process required high release rates (>~200 vesicles/s) to ensure the uninterrupted presence of glutamate in darkness. For release simulated at 100 vesicles/s for 8 s, glutamate concentration averaged 26 μM and ~50% of mGluR6 receptors were activated (i.e., doubly glutamate-bound). Interruptions in glutamate concentration and mGluR6 activation were prominent for the Poisson process, which produced four "events" where fewer than 25% of mGluR6 receptors were activated for more than 10 ms, while the other processes produced none. Our quasi-regular process produced about tenfold lower variance in the number of released vesicles than the Poisson process. Similar observations held for model light responses. Kinetics of glutamate concentration and mGluR6 activation were consistent with measurements from mouse rod bipolar cells, whether or not EAAT5 activity was modeled.
Conclusions:
Our model suggests that local release inhibition induces quasi-regular release from rods, similar in mathematical form to network oscillations, to suppress fluctuations in release rate that would frequently cause false-positive photon detections near visual threshold.