Synaptic integration in excitable CA1 pyramidal neuron dendritesPublic Deposited
A pyramidal neuron receives thousands of inputs spread throughout its dendritic tree, which it must integrate into a decision about whether or not to fire action-potential output. Since action potentials are the primary means by which these neurons communicate with their network partners, understanding this input-output relationship is critical for understanding information processing in the cerebral cortex. Using a combined approach of computational modeling and experimentation, we find that voltage attenuation reduces EPSPs generated at many distal synapses to negligible levels at the soma in CA1 pyramidal neurons. Distance-dependent conductance scaling is insufficient to overcome attenuation for these inputs; instead, they are predicted to communicate using dendritic spikes. Experiments corroborate this prediction: AMPA receptor density increases with distance from the soma, but decreases in the most distal region of the cell. Dendrites are more excitable near their terminal ends than near their branch points, so in the absence of compensatory mechanisms, dendritic spikes would be preferentially initiated at distal locations. Using serial-section electron microscopy to reconstruct apical oblique dendrites of CA1 pyramidal neurons, we find that synapse density and strength is greater near branch points than near terminal ends. Incorporating this result into our computational model, we find that synapses are organized to normalize the contribution of inputs to dendritic spike initiation and optimize the contribution of each branch to axonal output. Both the initiation and propagation of dendritic spikes are affected by inhibition. Using experimentally-constrained computational models to investigate the effect of inhibition targeting different somato-dendritic domains of pyramidal cells, we show how the dynamics of CA1 microcircuits depend on the location, magnitude, timing, and biophysical properties of inhibitory relative to excitatory inputs. In order to understand how the biophysical properties of CA1 pyramidal neuron dendrites contribute to the ability of the hippocampus to navigate space, we directly simulate the spatial alternation task using a network consisting of biophysically realistic model neurons. In the model, the integration of place and temporal context information arriving on the distal and more proximal apical dendrites of CA1 pyramidal neurons respectively make them "splitter cells", cells which fire selectively based on a combination of place and temporal context. These cells store a memory of the previous path through the environment, which the animal uses to navigate to a reward site.