Intra- and interlaminar excitatory synaptic connections of layer 4 spiny neurons and layer 6A pyramidal cells in rat barrel cortex

In the primary somatosensory (barrel) cortex of rodents, layer 4 (L4) and 6A are the main recipient layers of thalamocortical projections. In addition, a subset of L6A pyramidal neurons provide a direct corticothalamic feedback to the thalamus. Thus, neurons in layer 4 and 6A are an integral part of a thalamo-cortical-cortico-thalamic feedback circuit. To better understand the role of the intracortical unit in this circuit, we studied the anatomical and functional properties of excitatory synaptic connections from layer 4 to layer 6A in the rat barrel cortex by making dual whole-cell recordings with dye injection from L4 spiny neurons and L6A pyramidal cells in acute brain slices. Interlaminar monosynaptic L4-to-L6A excitatory connections (n = 17) were relatively rare. They were of low efficacy with an average excitatory postsynaptic potentials (EPSPs) of 0.32 ± 0.19 mV (n = 17) but of moderately high reliability with failure rate of 24.2 ± 17.7% (n = 16) and coefficient of variation (CV) of 0.56 ± 0.16 (n = 16). The EPSP amplitude was either depressing or weakly facilitating with paired-pulse ratio (PPR) of 0.45 - 1.38 (n = 17) at an interstimulus interval of 100 ms. Notably, we found a spatial separation of synaptic inputs on the dendritic domain of the postsynaptic L6A pyramidal cells depending on the presynaptic L4 neuron type: L4 spiny stellate neurons innervated predominantly the distal apical tuft dendrites of L6A pyramidal cells with synapse-to-soma distance of 591 ± 137 μm (n = 6) and elicited slow EPSPs (20-80% rise time = 6.7 ± 2.1 ms and latency = 3.8 ± 1.6 ms, n = 6) in L6A somata, while most of L4 star pyramidal neurons preferentially innervated the proximal basal and apical oblique dendrites with synapse-to-soma distance of 86 ± 54 µm (n = 7) and elicited fast EPSPs (20-80% rise time = 1.5 ± 0.9 ms and latency = 1.7 ± 0.2 ms, n = 7) in L6A somata with some star pyramids also forming synapses on the L6A apical tuft or oblique dendrites (synapse-to-soma distance = 524 ± 167 µm, n = 4) and eliciting relatively slow EPSPs (20-80% rise time = 5.4 ± 1.7 ms and latency = 3.7 ± 0.8 ms, n = 4). Other EPSP characteristics (i.e., amplitude, PPR, failure rate and CV) were not significantly different for the three types of L4-L6A connections. There was a tight correlation between the EPSP rise time, latency, and the synapse-to-soma distance. The synaptic location could not completely predicted solely on the basis of the axo-dendritic overlap suggesting that Peter’s rule of synaptic connectivity was not completely correct here. Using pharmacological treatment and neuronal modeling, we found that the occurrence of ‘slow’ and ‘fast’ EPSPs was not due to different receptor components in the postsynaptic densities but mainly due to the dendritic filtering effect during the EPSP propagation from synaptic location to soma. In addition, the cell-type specific selection of postsynaptic target region was a pre- but not postsynaptic phenomenon. As a comparison, we also performed some paired recordings in layer 4 and 6A and studied the characteristics of excitatory connections in layer 4 and 6A, respectively. For intralaminar monosynaptic L4-L4 and L6A-L6A excitatory connections, we found homogeneous dynamical properties of EPSPs, i.e., fast rise time (20-80% rise time = 1.59 ± 0.49 ms (n = 10) for L4-L4 and 1.39 ± 0.59 ms (n = 5) for L6A-L6A connections) and short latency (latency = 1.17 ± 0.41 ms (n = 10) for L4-L4 and 1.69 ± 0.65 ms (n = 5) for L6A-L6A connections), implying that, for both connections, synaptic inputs to postsynaptic neurons were electrotonically close to somata. The synaptic efficacy of L4-L4 connections were widely distributed from very weak connections (0.30 mV) to very strong ones (4.71 mV) with an average EPSP amplitude of 1.02 ± 1.33 mV (n = 10) compared with L6A-L6A connections that had a substantially lower average EPSP amplitude (0.58 ± 0.50 mV, n = 5), a relatively higher failure rate (17.5 ± 15.0%, n = 5) and a little higher CV (0.53 ± 0.23, n = 5)