Synaptic Integration and its Control in Neocortical Pyramidal Cells

The main goal of this thesis is to investigate the input/output relationship of single regular-firing neocortical pyramidal neurons and how this relationship can be controlled by external inputs. The thesis can be divided into three main parts. First, a detailed single cell model was developed, based on the morphology of reconstructed cells and experimental values for membrane conductances and synaptic input distributions. This model was used for the following investigations. Second, the spatio-temporal integration of single and multiple inputs was studied. Several measures for the efficacy and time delay of single synapses were defined and shown to vary dramatically. For example, a somatic synapse was only 2.2 times stronger than a very distal synapse using the charge attenuation measure, but more than 450 times stronger in the voltage attenuation measure. The effect that temporal synchronicity of multiple inputs had on firing rate was shown to vary with the number of inputs: for just-threshold input rates, synchronicity increased firing rate; for large inputs, high synchronicity strongly reduced firing rate, due to inputs being "wasted" during the refractory period. Third, a subset of the inputs were considered to constitute a control signal, and their effect on other inputs was studied for three cases. The first case considers the level of synaptic background activity to be a control signal; since each synapse is a small conductance change, and not a voltage-independent current source, the sum total of all "background" synapses will constitute the lion's share of the membrane conductance. The background firing rate, f_b, will therefore determine the electrotonic structure of the cell. For f_b in the range of 0-10 H z, a more than 10-fold decrease was seen in both input resistance (50.4-5.1 MΩ) and membrane time constant (33.7-1.6 msec). Electrotonic length and resting potential were similarly affected. The second case treats input to the apical trunk as the control signal; when this input was weak and excitatory, the more distal input to the apical tuft could be facilitated, but when this input was strong or combined with inhibition, more distal input were reduced. The third case involves distributing two types of active conductances throughout the apical dendrites. The activation curves of these conductances were "designed" to ensure that the current delivered to the soma was linear in the input rate and amplified, since a passive tree strongly attenuates large apical inputs. The linearization was implemented with a persistent potassium conductance in the superficial layer I- III and the amplification with a persistent calcium conductance in the apical trunk (layer IV). The amplification gain could be set arbitrarily by modulating the channel density of either the potassium or calcium conductance.