Frontiers in Systems Neuroscience (Apr 2014)
Computational models to explore morphological diversity of pyramidal neurons from monkey visual and prefrontal cortices
Abstract
Layer 3 (L3) pyramidal neuron dendrites and dendritic spines vary significantly across neocortical areas. Our group has previously described substantial differences in the structure and function of these neurons in in vitro slices of the primary visual cortex (V1) and dorsolateral prefrontal cortex (PFC) of the rhesus monkey (Amatrudo et al., 2012). Our analyses combine 3D morphological reconstructions of entire dendritic arbors from high-resolution confocal laser scanning microscopy, electrophysiological data from whole-cell patch-clamp recordings of the same neurons, and computational modeling of these cells. The result is a unique database of monkey cortical neurons, some of which are already available on NeuroMorpho.org and ModelDB. Here we present our computational models constrained by empirical data, and their predictions for the contribution of morphology to the function of L3 pyramidal neurons in V1 vs. PFC. L3 pyramidal neurons in V1 are much smaller than PFC neurons, with simpler dendritic arbors and lower spine densities. Physiological differences include higher input resistance and action potential (AP) firing rates and spontaneous excitatory postsynaptic currents (sEPSCs) with faster kinetics and lower amplitudes in V1 than in PFC neurons. We have created multicompartment models of V1 and PFC neurons based on their 3D reconstructions, including Hodgkin-Huxley conductances with equal distributions in the soma and dendrites, and AMPA- and GABAA ionotropic channels inserted uniformly through the dendrites. In these simple models, differences in passive physiological properties of V1 vs. PFC neurons are largely explained by morphology. However, when model parameters are held constant in the two model types, AP firing rates in the models differ more than observed empirically, and simulated EPSCs are contrary to empirical results. Thus the models predict that active membrane properties differ between V1 and PFC neurons. Dendritic channel distributions of monkey pyramidal cells have yet to be explored empirically; our computational models will test various plausible scenarios in the future. In recent experiments we have further investigated potential mechanisms underlying the difference between sEPSCs in V1 and PFC neurons. Specifically, action potential independent glutamate release onto L3 neurons in PFC vs. V1 neurons was assessed by recording TTx insensitive miniature EPSCs. The data suggest that quantal synaptic events are larger in PFC than in V1 neurons, due to both postsynaptic and presynaptic factors. The distribution and morphology of excitatory synapses on these neurons was studied with confocal and serial electron microscopy and 3D reconstruction. Compared to V1, PFC neurons had larger spines across the entire dendritic arbor, and the neuropil of L2-3 of PFC exhibited significantly larger excitatory asymmetric postsynaptic densities and presynaptic boutons, as well as a higher proportion of more powerful perforated synapses. Using a simple model with spines as distinct compartments attached to a cylindrical dendrite, we show that these empirical findings can largely account for differences in sEPSC properties between V1 and PFC neurons. The unique features of V1 and PFC neurons are likely fundamental determinants of area-specific network behavior. Future experiments and models will continue to explore these important structure/function relationships in individual neurons and in small networks.
Keywords
