|
Coherent patterns of population activity, such as oscillations in various frequency ranges, are abundant in most brain areas, and are believed to make important contributions to information processing in the cortex. In most models to date, these population oscillations arise largely as a result of synaptic interactions in the local network, possibly with some contribution from the subthreshold resonance properties of single neurons. Many of these studies employed relatively simple neuronal models, often consisting of a single compartment, and focused on the complexities of network interactions. However, it is now clear that the dendrites of many neurons, notably including cortical pyramidal neurons, harbor a multitude of voltage-gated conductances which enable complex nonlinear processing within local dendritic regions, and have a major impact on the output of the neuron. In addition, subcortical modulatory inputs to cortex, which reflect the behavioral state of the animal, change the properties of these active conductances, and may switch single cortical neurons between different information-processing modes. Since dendritic processing and neuromodulation alter neuronal firing patterns and input-output relationships, they may be expected to have a significant effect on network dynamics. Our goal was to quantify the extent of such influences in the context of theta-frequency oscillations in the hippocampus.
We developed a detailed network model of the hippocampal CA1 region. First, we constructed a simplified compartmental model of CA1 pyramidal neurons based on our existing, morphologically and biophysically realistic models using a systematic reduction procedure. We combined a large population of such pyramidal cell models with active single-compartment models of basket cells and O-LM interneurons in simulations of the CA1 network. We examined possible mechanisms of the generation of theta rhythmicity both in the presence and in the absence of oscillatory external inputs from the entorhinal cortex, area CA3, and the medial septum, and in the presence/absence of cholinergic modulatory input from the medial septum. In the absence of cholinergic input, the network did not generate oscillations autonomously. Oscillatory activity could be generated in response to rhythmic external input, and the firing phases of various cell classes could be matched to experimental data; however, the spatio-temporal distribution of current source density was markedly different from the measured distribution. Cholinergic input modulated various active conductances, and enabled CA1 pyramidal cells to fire repetitive calcium spikes in response to distal dendritic depolarization. In the network, such repetitive calcium spikes in the apical dendrites were synchronized across the pyramidal cell population by inhibitory feedback mediated by O-LM cells, resulting in the autonomous generation of theta-frequency oscillations even in the absence of phasic input. When rhythmic external inputs were included, these entrained the local rhythm such that neural activity satisfied the dual experimental constraints of neuronal spike timing and current source density pattern. Our modeling results suggest that subcortical inputs can profoundly affect cortical population activity through the modulation of dendritic active conductances, and explain some previously controversial characteristics of the hippocampal theta rhythm. |
|