Simulated dendritic spines influence reciprocal synaptic strengths and lateral inhibition in the olfactory bulb.

Whereas many theories have been proposed for the function of dendritic spines in axodendritic processing, the influence of spines on reciprocal dendrodendritic processing has received relatively little attention. Mitral cells in the olfactory bulb, for example, synapse on granule cell spines (gemmules) which are in turn presynaptic to reciprocal inhibitory synapses back onto the same mitral cells. The postulate that these synapses respond with synaptic strengths graded by presynaptic depolarization results in a sensitivity of the reciprocal response to the local depolarization in the spine head. A biophysical computer simulation was performed to study this effect and the effect of changing the spine neck diameter and cytoplasmic resistance on the reciprocal and lateral inhibitory responses given graded dendrodendritic synapses. Since spine head local potentials are larger than similar inputs on dendritic shafts, spines facilitate the graded reciprocal response even for low levels of activity. Spine heads also reduce the synaptic current, lowering the contribution to the rest of the granule dendritic tree and thus reducing lateral inhibition. In addition, an increase in the effective spine neck axial resistance further increases the reciprocal synaptic response and decreases the lateral inhibitory response. Short-term, reversible, and long-term methods of implementing this resistance-based dendrodendritic plasticity are discussed as well as the partial dependence of the reciprocal increase/lateral decrease effect on a broad synaptic gradation. Candidate memory operations by the bulb are also discussed, including a possible recognition memory pass/block function.

Computation of frequency-to-spatial transform by olfactory bulb glomeruli.

A physiological simulation of 2.5% of the input and inhibitory neurons and 25% of the primary mitral/tufted cells in a single mammalian olfactory bulb glomerulus was constructed. This physiological simulation used the integrate-and-fire paradigm with realistic activation curves and synaptic delays. The dendritic integration incorporated non-linear interactive effects of individual cell excitatory and inhibitory post-synaptic potentials (PSPs) from both axodendritic and dendro-dendritic synaptic contacts. Refractory periods for granule-cell inhibition of mitral/tufted cell activity lead to relatively fixed-frequency rhythmic activity in the glomerulus, independent of the input frequency from the olfactory nerve. Though the frequency of mitral/tufted cell firing in bulb was approximately independent of input frequency, the number of cells active in the glomerulus was a roughly-linear function of input frequency to the glomerulus, indicating the mechanism's ability to function as a frequency-to-spatial encoder.