Does layer 5 of the cortex project to the thalamic reticular nucleus? Implications for core and matrix thalamocortical circuits and sleep spindles

Abstract

Two distinct thalamocortical (TC) circuits with reciprocal components can be identified in mammals: The core TC circuit, prevalent in sensory thalamus, drives activity focally in the middle cortical layers. In turn, these core thalamic neurons are innervated by small ‘modulatory’ cortical axon terminals from pyramidal neurons in layer 6(L6). The matrix TC circuit, prevalent in high-order thalamus, has a complementary organization: large axon terminals from cortical layer 5(L5) pyramidal neurons drive activity of matrix thalamic neurons that, in turn, innervate broadly and modulate the superficial cortical layers. Situated strategically between the thalamus and cortex, the inhibitory thalamic reticular nucleus (TRN) intercepts all TC communication. Projections from sensory or motor cortices to TRN terminate exclusively as small boutons and originate from L6, akin to core TC circuits. No studies have shown direct projections to TRN from cortical neurons in L5 that participate in matrix circuits. However, in comparison with other cortices, prefrontal cortices issue substantial projections to the thalamus from L5 and send similar driver-like projections to TRN, which terminate as large boutons and could potentially originate from L5. These large prefrontal axon terminals are similar to cortical boutons in the caudate nucleus and the amygdala, which originate mainly from L5. Based on this indirect evidence we tested the hypothesis that cortical L5 neurons project to TRN in matrix networks, by constructing a computational TC circuit that included core and matrix components with an optional cortical L5 to TRN projection (L5-TRN ON/OFF). Based on the features of TC circuits, our model was able to simulate relay and filtering of signals, and could initiate and propagate spindle oscillations. Activation of TRN neurons with L5-TRN ON in our model initiated spindle generation with different powers, depending on the level of cortical feedback and involvement of model core vs. matrix. Our preliminary findings are in agreement with hypotheses that spindles can be classified in core-generated, matrix-generated or mixed types, depending on the pathways involved in their generation, but only if L5-TRN is ON. Simulation results indicate a more diffuse nature of spindles in matrix compared to core, with the mix type showing intermediate properties, suggesting that shifts in the engagement of distinct TRN, core, and matrix circuits may underlie typical sleep spindle generation and states of vigilance. Disruption of TC-TRN circuit balance may underlie seizures, atypical sensory reactivity, and deficits in sleep and attentional gating seen in autism and schizophrenia.